Article

pubs.acs.org/EF

Direct Conversion of Cellulose into Sorbitol over a Magnetic Catalyst

in an Extremely Low Concentration Acid System
Jun Zhang, Shu-bin Wu,* and Ying Liu*
State Key Laboratory of Pulp and Paper Engineering, South China University of Technology, Guangzhou 510640, People’s Republic
of China

ABSTRACT: Cellulose could be efficiently converted into sorbitol in the presence of hydrogen and an extremely low
phosphoric acid concentration over a magnetic catalyst. Effects of various process conditions toward the conversion efficiency
were discussed in greater detail. X-ray powder diffraction analysis indicated the essential properties of prepared catalysts. A
sorbitol yield of 68.07% was obtained at 488 K with a catalyst dosage of 20%. The experiments showed that Ni4.63Cu1Al1.82Fe0.79
sustained better activity after being reused 3 or 4 times.

1. INTRODUCTION was filtered, washed, and dried at 353 K for 24 h. The precursor was
required to be activated at 923 K under a H2 atmosphere for 3 h before
Cellulose is the most abundant biopolymer on earth, and it being used. The used catalyst was reduced again before the next use.
holds great potential as an alternative source of valuable Inductively coupled plasma−atomic emission spectrometry (ICP−
chemicals and fuels, especially because the gradual depletion of AES) analysis was introduced to identify the chemical composition of
fossil fuels is now becoming of worldwide concern. Besides, the precursor, and the resulting catalyst was named
emissions of greenhouse gases associated with the use of fossil Ni4.63Cu1Al1.82Fe0.79. The method described above was applied to
fuels can be improved through using biomass as a carbon and synthesize the Cu/Al/Fe catalyst, and the obtained sample was
energy source. These aspects have aroused a great deal of denoted as Cu1Al1.71Fe0.72.
interesting research into the use of biomass as a renewable 2.2. Catalytic Reaction. The conversion of cellulose (crystallinity
source of fuels and chemicals.1,2 of 60%) to sorbitol was conducted in a high-pressure reactor (PARR
5500). In a typical run, cellulose, catalysts, and 0.08 wt % H3PO4 were
The resistance of cellulose to hydrolysis is well-known. intrduced into the autoclave. Afterward, the reactor was purged with
However, cellulose can be evidently degraded with liquid acids,3 H2 3 times to remove the residual air, pressurized with H2 to 4 MPa,
bases,4 enzymes,5 and hot-compressed water.6 Currently, and then heated to certain temperatures with a stirring rate of 600
catalytic conversion of cellulose into sorbitol has attracted a rpm. After experiments, the furnace was cooled to ambient
lot of attention in recent research.7−9 The valuable polylols can temperature rapidly with an ice−water mixture and then the products
be obtained through the hydrogenolysis of sorbitol over metal were collected by filtration and stored in the refrigerator prior to
catalysts. More importantly, light paraffins and H2 will be analysis.
produced through aqueous-phase reforming of sorbitol.10 For the heating−stirring pretreatment, the cellulose material in
Usually noble metal catalysts show high activity in the acidic solution was heated to 358 K under a high agitation speed
hydrolytic hydrogenation of cellulose, such as Ru/C11 and Pt/ (∼750 rpm) for 3 h. Then, the obtained mixtures were used for
sorbitol production under the catalyzing of magnetic catalysts.
C.12 However, the preparation costs for these catalysts are too Liquid samples were identified from their retention times as
high that will limit the industrial application. Herein, the cheap measured by ion chromatography (IC) analysis on a Dionex ICS-3000
Ni/Cu/Al/Fe hydrotalcite was synthesized and served as a instrument furnished with a CarboPac PA10 column at a column
precursor for preparing the hydrogenation catalyst, which also temperature of 303 K. The eluent was 40 mM NaOH solution with a
can be easily separated from reaction mixtures in a magnetic flow rate of 0.25 mL/min. The quantitative determination of
field. In this paper, the magnetic catalyst was used for cellulose byproducts was detected by high-performance liquid chromatography
conversion in an extremely low concentration acid, which is (HPLC, Waters 1525, Milford, MA) using a Platisil ODS C18 column
environmentally friendly as well as economical. Notably, the and a refractive index detector (RID, Waters 2414, Milford, MA) at
simple heating−stirring pretreatment was advanced to facilitate 298 K.
the conversion of cellulose into sorbitol. Effects of various 2.3. Characterizations of the Catalyst and Pretreated
Samples. Chemical analysis was carried out on a Thermo Elemental
processing conditions on cellulose conversion are investigated ICP−AES spectrometer after dissolution of the sample in a HNO3
to achieve the optimal sorbitol yield. solution. X-ray powder diffraction (XRD) was recorded on a Bruker
D8 Advance diffractometer using Cu Kα radiation, operated at 40 kV
2. EXPERIMENTAL SECTION and 40 mA at a scan rate of 2°/min. The crystallite size of the catalyst
was calculated by XRD line broadening according to the Scherrer
2.1. Catalyst Preparation. The Ni/Cu/Al/Fe hydrotalcite
precursor was prepared using a co-precipitation method.13 A solution
of 4.53 g of Cu(NO3)2·6H2O, 27.26 g of Ni(NO3)2·6H2O, 3.75 g of Special Issue: International Biorefinery Conference
Fe(NO3)3·9H2O, and 10.55 g of Al(NO3)3·9H2O with a cation
concentration of 1 mol/L was added dropwise into 150 mL of NaOH Received: January 6, 2014
(12.0 g) and Na2CO3 (1.99 g) solutions at a constant pH of around Revised: April 16, 2014
9.0. The slurry thus obtained was kept at 333 K for 10 h. The sample Published: April 16, 2014

© 2014 American Chemical Society 4242 dx.doi.org/10.1021/ef500031w | Energy Fuels 2014, 28, 4242−4246
Energy & Fuels Article

equation.14 The structures of pretreated samples were analyzed on a

scanning electron microscope (SEM) S-3700N operating at 15 kV.

3. RESULTS AND DISCUSSION

3.1. XRD Test of the Catalyst. The XRD pattern of as-
synthesized hydrotalcite-like material was shown in the inset of
Figure 1. The XRD pattern of the precursor revealed the peaks

Figure 1. XRD patterns of prepared catalysts: (a) fresh, (b) used once,
and (c) used twice. (Inset) XRD pattern of the catalyst precursor.

characteristic of layered double hydrotalcite (LDH) phases with

the presence of sharp and symmetrical lines for (003), (006),
and (018) planes and broad and asymmetric lines for (012),
(015), and (113) planes.15 The (003), (006), (012), (015),
(018), and (113) diffractions of the hydrotalcite-like compound
(HTlc) were ascribed to the 2θ of 11.7, 23.3, 35.0, 39.0, 47.2,
and 61.6° respectively (Figure 1). The obtained results
demonstrated that Ni and Cu centers were homogeneously
merged into the matrixes of Ni4.63Cu1Al1.82Fe0.79 HTlc, as
compared with previous work,16−19 because no obvious peaks Figure 2. SEM analysis of cellulose samples under various conditions:
representing individual Ni, Cu, Al, and Fe hydroxides were (A) untreated, (B) 0.1% H3PO4, (C) 0.08% H3PO4, (D) 0.06%
found. H3PO4, (E) 0.1% H2SO4, (F) 0.08% H2SO4, and (G) 0.06% H2SO4.
The XRD diagram of the reduced sample is given in Figure 1.
It was found that all of the hydrotalcite structure was destroyed found in the pretreated samples, as shown in panels B−G of
and some new crystal phases of active metals were formed after Figure 2. When the samples was treated in low acid
reduction. According to the Joint Committee on Powder concentrations (0.06% H3PO4 and 0.06% H2SO4), some large
Diffraction System 4-836 (JCPDS 4-836), peaks corresponding particles were found to be apparently formed in the resulting
to planes (200) at 2θ of 51.7° indicated the presence of Cu0 in mixtures. Higher acid concentrations led to the formation of
the face-center-cubic structure. Furthermore, peaks at 2θ of fine granules, which might be caused by the deep hydrolysis of
44.2° and 75.8° revealed the formation of a Ni-rich alloy cellulose materials. For the untreated sample (Figure 2A), most
(NiCu), which were assigned to the planes (111) and (220), likely, these larger particles may correspond to agglomerations
respectively. These results are in agreement with that described of the primary cellulose crystals because the crystallinity index
by Rao et al.20 and Wu et al.,21,22 in relation to the peak was as high as around 60%. The results demonstrated that the
positions in the XRD diagram of the Cu−Ni alloy. Meanwhile, cellulose structure had been evidently destroyed to some
the crystallite sizes of catalysts increased apparently during the extent. Therefore, a change in the morphology of the cellulose
recycling experiments (as given in the table in Figure 1), which particles occurred during the pretreatment.
might be caused by particle sintering at a high reduction 3.3. Catalytic Performance for the Conversion of
temperature. It was also worth noting that no obvious phase Cellulose. 3.3.1. Effect of the Reaction Temperature. The
change was observed in the recycled catalysts. influence of the reaction temperature on the conversion of
3.2. SEM Analysis of Pretreated Cellulose Samples. To cellulose into sorbitol was discussed, and the experiments were
obtain information on a possible morphology evolution, we carried out at 478, 488, and 498 K within a given time. As seen
performed SEM analysis of the samples before and after the in Figure 3, the reaction temperature played a major role in the
heating−stirring pretreatment. The pictures in Figure 2 indicate conversion efficiency. First, the sorbitol yield grew significantly
that the average particle size decreased significantly and the with increasing the temperature from 478 to 488 K in the same
cellulose particles were widely dispersed after 3 h of reaction time. However, the obtained sorbitol would be further
pretreatment. Furthermore, many particles below 10 μm were degraded into lower alcohols at a higher temperature in a
4243 dx.doi.org/10.1021/ef500031w | Energy Fuels 2014, 28, 4242−4246
Energy & Fuels Article

byproducts, such as glycol, glycerol, and 1,2-propylene glycol,

were formed.
3.3.3. Catalyst Recycling. Long-term reusability of the
heterogeneous catalyst is an extremely key point for future
industrial application to lower practical cost. After reaction, the
catalysts separated from the liquid products were reactivated at
923 K for 3 h in a hydrogen atmosphere before next use. As
seen from Figure 5, the sorbitol yield decreased from 42.91 to

Figure 3. Effect of the reaction temperature on the yield of sorbitol

with the reaction time. Reaction conditions: 0.5 g of cellulose, 50 mL
of 0.08 wt % H3PO4, 0.1 g of catalyst, and 4 MPa H2.

hydrogen atmosphere, thus leading to an obvious decrease in

the yield of sorbitol. At the temperatures of 488 and 498 K,
much sorbitol was produced in the initial phase of the reaction
and then the sorbitol yield decreased obviously with increasing
the reaction time because of the hydrogenolysis of produced
sorbitol. The glycerol yield was raised up to 20.04% with Figure 5. Sorbitol yield as a function of the recycling times of the
increasing time to 3 h at 478 K. The optimum yield was catalyst. Reaction conditions: 0.5 g of cellulose, 50 mL of 0.08 wt %
attained at the reaction time of 3 h. At 488 K, the yield of H3PO4, 0.1 g of catalyst, 488 K, 4 MPa H2, and 3 h.
sorbitol increased smoothly with the prolonging of the reaction
time, which may need a longer reaction time to reach the 40.96% in the second run. In the following two runs, the
equilibrium conversion. catalytic activity of the catalyst had not been completely
3.3.2. Effect of the Catalyst Dosage. Figure 4 shows the recovered after reduction and the sorbitol yield was around
influence of the catalyst dosage on the yield of sorbitol, and the 35%, which reached about 81% of that obtained for the fresh
catalyst. This may be caused by partial loss of active sites in the
catalyst during catalysis. It can be seen from Table 1 that active

Table 1. Amount of Leached Cu and Ni in the Fresh Catalyst

recycling time (h) Cu content (wt %) Ni content (wt %)
1 0 0.043
2 0 0.067
3 0 0.095

Ni species of magnetic catalysts were clearly leached into the

reaction solution during the recycling experiments, which might
be due to the chelation between the hydroxyl group from sugar
alcohol compounds and active Ni.
Meanwhile, the deposited humins on the surface of the used
Figure 4. Effect of the catalyst dosage on the yield of sorbitol with the catalyst could not be easily removed during reduction
reaction time. Reaction conditions: 0.5 g of cellulose, 50 mL of 0.08 wt
treatment; therefore, the activity of the recycled catalyst could
% H3PO4, 498 K, and 4 MPa H2.
be partly improved after reduction. Accordingly, how to
improve the long-term stability of the reduced Cu/Ni/Al/Fe
specific experiments were carried out at three different catalyst hydrotalcite-like catalyst remained as one of the topics for
loadings (20, 30 and 40%) as a function of the reaction time. future exploration. On the other hand, the leaching of active Ni
With increasing the amount of magnetic catalyst, the sorbitol was a big problem for the practical application, because the final
yield decreased significantly in the initial stage at a temperature solutions would be contaminated to a certain extent. The
of 498 K, especially for the catalyst dosage of 40%. It was production cost was also raised significantly for removing
inferred that much of the produced sorbitol was converted into residual Ni.
lower alcohols under such a high temperature at hydrogen 3.3.4. Effect of the Heating−Stirring Pretreatment on
atmosphere. For a catalyst dosage of 20%, the sorbitol yield Cellulose Conversion. Table 2 shows the influences of different
began to increase with the prolonging of the reaction time that pretreatment conditions on the conversion of cellulose into
reached the highest level of 40.32% after a reaction of 1.5 h. sorbitol. It can be seen that much produced sorbitol and
Further increasing the reaction time was unfavorable for the mannitol would be further converted into lower molecular
direct conversion of cellulose into sorbitol, and many weight alcohols and other unidentified byproducts in the acid
4244 dx.doi.org/10.1021/ef500031w | Energy Fuels 2014, 28, 4242−4246
Energy & Fuels Article

Table 2. Effect of Pretreatment Conditions on the Conversion of Cellulosea

yield (%)
conditions sorbitol mannitol glycerol glycol propanediol
untreated 42.91 ± 0.39 9.68 ± 0.19 10.98 ± 0.15 2.41 ± 0.07 1.47 ± 0.11
0.1% H2SO4 11.72 ± 0.23 3.61 ± 0.08 15.30 ± 0.21 4.49 ± 0.09 5.79 ± 0.05
0.08% H2SO4 14.15 ± 0.49 3.99 ± 0.05 14.15 ± 0.08 4.32 ± 0.12 4.24 ± 0.11
0.06% H2SO4 35.31 ± 0.22 8.87 ± 0.10 18.41 ± 0.17 5.04 ± 0.07 2.41 ± 0.09
0.1% H3PO4 21.20 ± 0.38 3.79 ± 0.06 9.16 ± 0.31 4.72 ± 0.04 1.96 ± 0.04
0.08% H3PO4 42.07 ± 0.24 7.00 ± 0.04 1.20 ± 0.04 4.03 ± 0.03 0.34 ± 0.06
0.06% H3PO4 68.07 ± 0.41 11.96 ± 0.34 3.25 ± 0.03 4.21 ± 0.06 0.87 ± 0.03
0.06% H3PO4b 28.69 ± 0.32 5.37 ± 0.09 2.31 ± 0.05 3.35 ± 0.11 0.82 ± 0.07
a
Reaction conditions: 0.5 g of cellulose, 0.1 g of catalyst (Ni4.63Cu1Al1.82Fe0.79), 488 K, 4 MPa H2, and 3 h. bA total of 0.1 g of catalyst
(Cu1Al1.71Fe0.72).

concentration of 0.1%. The structures of cellulose crystals were

greatly destroyed under higher concentrations, which would
■ AUTHOR INFORMATION
Corresponding Authors
facilitate the hydrolytic hydrogenation and hydrogenolysis of
*Telephone/Fax: +86-20-22236808. E-mail: shubinwu@scut.
pretreated samples. As a comparison, samples treated by H2SO4
edu.cn.
solutions could be deeper degraded than those treated by
*Telephone/Fax: +86-20-22236808. E-mail: [email protected].
H3PO4 solutions, thus leading to an obvious decrease in the
cn.
yield of sorbitol. Because the acid strength of H2SO4 was much
higher than that of the same concentration of H3PO4, the Notes
glucose obtained from cellulose hydrolysis might be further The authors declare no competing financial interest.
converted into other chemicals under the catalyzing of
inorganic acids, such as 5-hydroxymethylfurfural and levulinic
acid.
■ ACKNOWLEDGMENTS
This work was supported by the National Key Basic Research
In comparison to untreated cellulose, the 0.08% H3PO4- Program of China (2013CB228101) and the National Natural
pretreated sample did not show better activity toward sorbitol Science Foundation of China (31270635). Finally, the authors
production, which might be caused by the excess degradation of are grateful to the kind support from the Committee of the 4th
glucose during the process of the heating−stirring pretreat- International Conference on Biorefinerytowards Bioenergy
ment. For H3PO4 pretreatment, a lower acid concentration (ICBB 2013) in Xiamen, China.

■
favored the formation of sorbitol and mannitol, with the yields
reaching up to 68.07 and 11.96%, respectively. Therefore, a REFERENCES
proper heating−stirring pretreatment of cellulose samples in
0.06% H3PO4 could facilitate the proceeding of hydrolysis− (1) Mariefel, B. V.; Christopher, W. J.; Pradeep, K. A. Biorefineries:
hydrogenation reactions. Current status, challenges, and future direction. Energy Fuels 2006, 20,
1744−1752.
To verify the specific active sites, the prepared Cu1Al1.71Fe0.72 (2) Zhang, J.; Wu, S. B.; Liu, Y. Advances in the catalytic production
was used for the conversion of 0.06% H3PO4-pretreated and utilization of sorbitol. Ind. Eng. Chem. Res. 2013, 52, 11799−
cellulose. Under the catalysis of Cu1Al1.71Fe0.72, the sorbitol 11815.
yield was around 29%. However, a sorbitol yield of 68.07% was (3) Mok, W. S.; Antal, M. J.; Varhergyi, G. Productive and parasitic
attained when catalyzed by Ni4.63Cu1Al1.82Fe0.79 under the same pathways in dilute acid-catalyzed hydrolysis of cellulose. Ind. Eng.
conditions. The catalyst activity had been greatly improved Chem. Res. 1992, 31, 94−100.
after the modification of the Ni element. Therefore, Cu and Ni/ (4) Ishida, M.; Otsuka, K.; Takenaka, S.; Yamanaka, I. One-step
Cu were the active sites for the hydrogenation reaction production of CO- and CO2-free hydrogen from biomass. J. Chem.
Technol. Biotechnol. 2005, 80, 281−284.
combined with XRD analysis, and the modification of Ni
(5) Zhang, Y. H. P.; Lynd, L. R. Toward an aggregated understanding
accelerated the proceeding of the catalytic reaction. There of enzymatic hydrolysis of cellulose: Non-complexed cellulase systems.
might be a synergistic effect between active Cu and Ni/Cu Biotechnol. Bioeng. 2004, 88, 797−824.
during the reaction. (6) Yu, Y.; Wu, H. W. Significant differences in the hydrolysis
behavior of amorphous and crystalline portions within microcrystalline
4. CONCLUSION cellulose in hot-compressed water. Ind. Eng. Chem. Res. 2010, 49,
3902−3909.
This work presented a feasible method for efficient conversion (7) Deng, W. P.; Tan, X. S.; Fang, W. H.; Zhang, Q. H.; Wang, Y.
of cellulose into sorbitol with the presence of a magnetic Conversion of cellulose into sorbitol over carbon nanotube-supported
catalyst and an extremely low phosphoric acid concentration. A ruthenium catalyst. Catal. Lett. 2009, 133, 167−174.
simple and efficient preprocessing method had been developed. (8) Xi, J. X.; Zhang, Y.; Xia, Q. N.; Liu, X. H.; Ren, J. W.; Lu, G. Z.;
The structures of cellulose samples could be obviously Wang, Y. Q. Direct conversion of cellulose into sorbitol with high yield
destroyed on the basis of the SEM analysis. The desired by a novel mesoporous niobium phosphate supported ruthenium
bifunctional catalyst. Appl. Catal., A 2013, 459, 52−58.
sorbitol yield could reach 68.07% at 488 K for 3 h under 4 MPa (9) Liu, M.; Deng, W. P.; Zhang, Q. H.; Wang, Y. L.; Wang, Y.
H2. The XRD test of used catalysts indicated that a high Polyoxometalate-supported ruthenium nanoparticles as bifunctional
crystallite size of the catalyst was not conducive to the heterogeneous catalysts for the conversions of cellobiose and cellulose
formation of sorbitol. Ni4.63Cu1Al1.82Fe0.79 could sustain better into sorbitol under mild conditions. Chem. Commun. 2011, 47, 9717−
catalytic activity during catalyst recycling. 9719.

4245 dx.doi.org/10.1021/ef500031w | Energy Fuels 2014, 28, 4242−4246

Energy & Fuels Article

(10) Huber, G. W.; Shabaker, J. W.; Dumesic, J. A. Raney Ni−Sn

catalyst for H2 production from biomass- derived hydrocarbons.
Science 2003, 300, 2075−2077.
(11) Geboers, J.; de Vyver, S. V.; Carpentier, K.; Jacobs, P.; Sels, B.
Hydrolytic hydrogenation of cellulose with hydrotreated caesium salts
of heteropoly acids and Ru/C. Green Chem. 2011, 13, 2167−2174.
(12) Fukuoka, A.; Dhepe, P. L. Catalytic conversion of cellulose into
sugar alcohols. Angew. Chem., Int. Ed. 2006, 45, 5161−5163.
(13) Velu, S.; Ramaswamy, V.; Sivasanker, S. New hydrotalcite-like
anionic clays containing Zr4+ in the layers. Chem. Commun. 1997, 21,
2107−2108.
(14) Klug, H. P.; Alexander, L. E. X-ray Diffraction Procedures for
Polycrystalline and Amorphous Materials, 2nd ed.; Wiley: Hoboken, NJ,
1974.
(15) Courty, P.; Durand, D.; Freund, E.; Sugier, A. C1−C6 alcohols
from synthesis gas on copper−cobalt catalysts. J. Mol. Catal. 1982, 17,
241−254.
(16) Chen, Z. B.; Nai, J. W.; Ma, H.; Li, Z. Q. Nickel hydroxide
nanocrystals-modified glassy carbon electrodes for sensitive L-histidine
detection. Electrochim. Acta 2014, 116, 258−262.
(17) Wylie-van Eerd, B. J.; Young, A. G.; Al-Salim, N. I.; Kemmitt, T.;
Strickland, N. M. Ligand-functionalised copper(II) hydroxide for
quantum dot photoluminescence quenching. J. Colloid Interface Sci.
2010, 346, 288−295.
(18) Ogata, F.; Kawasaki, N. Adsorption of Pt(IV) and Pd(II) by
calcined dried aluminum hydroxide gel from aqueous solution system.
J. Environ. Chem. Eng. 2013, 1, 1013−1019.
(19) Song, H.; Jeon, B. H.; Chon, C. M.; Kim, Y.; Nam, I. H.;
Schwartz, F. W.; Cho, D. W. The effect of granular ferric hydroxide
amendment on the reduction of nitrate in groundwater by zero-valent
iron. Chemosphere 2013, 93, 2767−2773.
(20) Rao, G. R.; Mishra, B. G.; Sahu, H. R. Synthesis of CuO, Cu and
CuNi alloy particles by solution combustion using carbohydrazide and
N-tertiarybutoxy-carbonylpiperazine fuels. Mater. Lett. 2004, 58,
3523−3527.
(21) Wu, S. P. Preparation of ultra fine nickel−copper bimetallic
powder for BME-MLCC. Microelectron. J. 2007, 38, 41−47.
(22) Wu, S. P.; Ni, J.; Jiao, L.; Zeng, Z. N. Preparation of ultra-fine
copper−nickel bimetallic powders with hydrothermal-reduction
method. Mater. Chem. Phys. 2007, 105, 71−75.

4246 dx.doi.org/10.1021/ef500031w | Energy Fuels 2014, 28, 4242−4246