Thermochemical saccharification of cellulose: The benefit of adding a scavenger

Article abstract of DOI:10.1016/j.apcata.2014.01.052

The solid acid-catalysed saccharification of cellulose was studied under elevated temperatures. Pretreatment of cellulose was necessary to obtain high glucose selectivity at high yields. Dissolution-regeneration from 1-butyl-3-methylimidazolium yielded more reactive cellulose compared to ball-milling. The highest glucose productivity was obtained from regenerated cellulose in the presence of Norit CAP Super (NCS), 685.7 g h^-1 kg^-1 catalyst at a glucose yield of 61.2% and a selectivity of 73.8%. The carbon catalyst, however, is not stable and leaches acidic species. The observed activity is higher than may be expected from the leached species alone. In addition, NCS is able to steer the reaction selectively towards glucose, especially when brought into close contact with the substrate. The highest glucose selectivity observed was 94.5% at a glucose yield of 38.9% and was obtained after co-milling cellulose with NCS. The effects are thought to be related to the ability of NCS to trap by-products, thus preventing the formation of humins and reducing glucose losses due to condensation reactions.

Full text of DOI:10.1016/j.apcata.2014.01.052

Applied Catalysis A: General 475 (2014) 438–445
Contents lists available at ScienceDirect
Applied Catalysis A: General
journal homepage: www.elsevier.com/locate/apcata
Thermochemical saccharification of cellulose:
The benefit of adding a scavenger
∗
Ruud J.H. Grisel , Arjan T. Smit
Energy Research Centre of the Netherlands (ECN), Biomass & Energy Efficiency, PO Box 1, 1755 ZG Petten, The Netherlands
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 21 August 2013
Received in revised form 14 January 2014
Accepted 29 January 2014
Available online 5 February 2014
The solid acid-catalysed saccharification of cellulose was studied under elevated temperatures. Pretreat-
ment of cellulose was necessary to obtain high glucose selectivity at high yields. Dissolution-regeneration
from 1-butyl-3-methylimidazolium yielded more reactive cellulose compared to ball-milling. The high-
est glucose productivity was obtained from regenerated cellulose in the presence of Norit CAP Super
−1 −1
(NCS), 685.7 g h kg catalyst at a glucose yield of 61.2% and a selectivity of 73.8%. The carbon catalyst,
however, is not stable and leaches acidic species. The observed activity is higher than may be expected
from the leached species alone. In addition, NCS is able to steer the reaction selectively towards glucose,
especially when brought into close contact with the substrate. The highest glucose selectivity observed
was 94.5% at a glucose yield of 38.9% and was obtained after co-milling cellulose with NCS. The effects are
thought to be related to the ability of NCS to trap by-products, thus preventing the formation of humins
and reducing glucose losses due to condensation reactions.
Keywords:
Cellulose
Hydrolysis
Ionic liquid
Pretreatment
Solid acid
©
2014 Elsevier B.V. All rights reserved.
1
. Introduction
To overcome these problems, the use of solid acid catalysts
was suggested [13–16]. Solid acids would simplify catalyst reuse
and therefore minimise costs, reduce the amount of waste, and
facilitate continuous flow operation. Conversely, due to the insol-
ubility of cellulose in common solvents, problems with diffusion
limitations may arise when using solid acid catalysts. Indeed,
previous studies showed the necessity of large catalyst loadings
and prolonged reaction times in order to produce glucose from
microcrystalline cellulose, resulting in low glucose productivities
[13–18]. To a certain extent, the productivity can be increased by
applying more severe reaction conditions at the cost of selectivity
due to simultaneously accelerated glucose degradation reactions
[19].
Glucose is considered a green precursor for chemicals and fuels
[
1–6]. Currently, most glucose comes from starch and is in com-
petition with food supply. Alternatively, cellulose can be used as a
source for glucose. The parallel alignment of linear anhydroglucose
chains in cellulose enables a vast network hydrogen bonding, con-
tributing to the overall stability of the structure [7–9]. As a result,
cellulose is recalcitrant and insoluble in water and common organic
solvents. Glucose is thus locked in the robust, crystalline structure
and controlled depolymerisation to d-glucose is challenging.
One way to liberate glucose is by enzymatic hydrolysis. How-
ever, enzymatic hydrolysis is expensive, time-consuming and
suffers from disadvantages, such as enzyme loss through non-
productive binding, shear deactivation of enzyme, and product
inhibition [10,11]. To improve the economics a faster and cheaper
route for the production of glucose is desired. Key in the valorisation
of cheap feedstock such as lignocellulose is to improve the produc-
tivity rather than to pursue an incremental increase in yield and/or
selectivity. Much faster, but intrinsically less selective, is hydroly-
sis catalysed by mineral acids [4,11,12]. The use of mineral acids,
however, is associated with potential corrosion problems, hazard
issues, catalyst loss and significant saline waste streams.
Cellulose may be pretreated in various ways to enhance its reac-
tivity [8,20–22]. Onda et al. [14,15] reported enhanced cellulose
hydrolysis activity over sulfonated active carbon after ball-milling.
They achieved a glucose yield of 41% at a maximum glucose
−
1
−1
productivity of 18.8 g h kg catalyst. Pang et al. [17] opti-
mised the sulfonation procedure, yielding sulfonated activated
carbons with an increased acid density and a superior perfor-
mance. The authors reported maximum 94.4% cellulose conversion
◦
after 24 h reaction at 150 C, yielding 74.5% glucose at a pro-
−
1
−1
ductivity of 34.5 g h kg catalyst. At nearly similar conditions,
Van de Vyver et al. [18] reported 50% glucose yield over sul-
−
1
−1
fonated silica/carbon at a glucose productivity of 23.2 g h kg
catalyst. The energy intensive character of ball-milling, however,
makes it a less attractive method for large scale cellulose pretreat-
ment.
∗ Corresponding author. Tel.: +31 224 564 216; fax: +31 224 568 487.
E-mail address: grisel@ecn.nl (R.J.H. Grisel).
0
926-860X/$ – see front matter © 2014 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.apcata.2014.01.052

R.J.H. Grisel, A.T. Smit / Applied Catalysis A: General 475 (2014) 438–445
439
Another method to enhance the reactivity of cellulose is through
2. Experimental
the use of ionic liquids. Swatloski et al. [23] reported that cer-
tain ionic liquids, such as 1-butyl-3-methylimidazolium chloride,
2.1. Materials
[
bmim]Cl, were capable of dissolving cellulose. The dissolution pro-
cess disrupts the cellulosic fibres, leaving the hydroxyl groups and
the ␤-glycosidic linkages accessible, thus improving the reactiv-
ity of the cellulose. Indeed, dissolved in ionic liquid, cellulose can
easily be hydrolysed to glucose in high yields by mineral acids
The cellulose used was Avicel PH 101 (Fluka, particle size
∼50 ␮m, DPw 200–240 AGU), the ionic liquid used was 1-butyl-3-
methylimidazolium chloride (Basionic ST 70, purity ≥95%). Both
were purchased from Sigma–Aldrich. Beta zeolite and mordenite
used were obtained from Zeolyst. Sulphated and tungstated zir-
[
24,25]. Herein, the ionic liquid can be acidic, thus serving as both
TM
solvent and catalyst, making addition of a mineral acid redun-
dant [26–28]. However, efficient methods for extracting glucose
from such solutions remain challenging. Only recently, Feng et al.
conia were kindly provided by Saint-Gobain NorPro, Amberlyst
TM
15Dry and Amberlyst
70 by Rohm and Haas, Y Zeolite by Albe-
marle, and phosphoric acid activated carbon CAP Super by Norit.
The Si/Al atomic ratios of the zeolites are indicated by a number
in parentheses, for example H-beta (75) is the proton-form of beta
zeolite with a Si/Al atomic ratio of 75.
[
29] demonstrated the use of alumina column chromatography
for the separation of glucose from N-methyl-N-methylimidazolium
di-methyl phosphate and Caes et al. [30] reported on the recov-
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ery of 3-methyl-1-(2 -2 -3 -3 -3 -pentafluoropropyl)-imidazolium
chloride after cellulose hydrolysis.
Apart from their use as a (catalytic) medium for cellulose
hydrolysis, ionic liquids have also been used for cellulose pre-
treatment prior to enzymatic hydrolysis. Precipitation of cellulose
from an ionic liquid solution by addition of an anti-solvent, such
as water or methanol, results in a cellulose, termed regenerated
cellulose, with a less crystalline and more accessible structure
and is significantly more susceptible to enzymatic hydrolysis rel-
ative to untreated cellulose [31–34]. Still, due to the restricted
maximum temperature that can be applied when using enzymes,
the glucose productivity is limited. The incubation time, tem-
perature and nature of the anti-solvent had little effect on the
improved enzymatic digestibility of the resultant regenerated cel-
luloses [32,33].
2.2. Cellulose pretreatment
A portion of the Avicel cellulose (25 g) was dry milled in a zir-
conia pot (2 L), for one third filled with zirconia balls (diameter
1.5 cm). The vessel was closed, placed on a roller bank and the
cellulose was milled at approximately 60 rpm for 48 h. The milled
cellulose was recovered and stored at RT in a closed vessel, in the
absence of moist. The recovered material is referred to as ball-
milled cellulose. In one experiment cellulose was milled in the
presence of Norit CAP Super in a 1:1 weight ratio.
For cellulose pretreatment in ionic liquid, 1-butyl-3-
methylimidazolium chloride, [bmim]Cl, was heated above its
melting point before cellulose was added. In some experiments
◦
Studies on the use of both ionic liquids and solid acid catalysts
[bmim]Cl was pre-dried at 150 C for 2 h under a nitrogen flow
remain sparse. Particularly Rinaldi et al. [35,36] reported on the
before adding the cellulose. In other experiments extra water
TM
TM
hydrolysis of ␣-cellulose in [bmim]Cl using Amberlyst
15Dry.
and Amberlyst
15Dry was added. After the predetermined
Rather than complete saccharification they stopped the reaction
at the cellooligomer stage, allowing recycling of the ionic liq-
uid while obtaining solid, regenerated cellulose with a decreased
degree of polymerisation. The maximum cellulose recovery was
incubation period the heating was removed and the cellulose was
precipitated by adding 5 parts (v/v) of hot demineralised water
(>80 C). During pretreatment, precipitation and initial cooling, the
◦
liquid [bmim]Cl solutions were stirred at all times. The solutions
◦
9
1% after 1 h of reaction and decreased with prolonged reaction
were stored at 4 C overnight to allow the precipitate to settle. The
TM
+
time. Amberlyst
was not stable in [bmim]Cl and H O was
clear solution was decanted and the precipitates were separated
by centrifugation (Thermo Scientific SL 40R, 10 min at 4000 rpm)
3
released into the liquid. The leached species were responsible for
the observed catalytic activity.
◦
and washed three times with hot (>80 C) and once with cold
Kim et al. [37] proposed complete disentanglement of the
ionic liquid pretreatment and the solid acid mediated hydrolysis
reaction. Microcrystalline cellulose was first regenerated from
demineralised water. The moisture content of the freshly prepared
sample was determined (Mettler Toledo HR83 Halogen). The
regenerated cellulose samples were diluted with water yielding
a colloidal aqueous suspension with 5.0% (w/w) regenerated
cellulose and stored at 4 C until tested. Prior to testing the sample
was homogenised and the moisture content was determined
again.
[
bmim]Cl and subsequently hydrolysed over Nafion NR50 for 4 h
◦
◦
at 160 C. Although solid acid catalysed hydrolysis of cellulose
was facilitated by the pretreatment, the maximum glucose yield
was limited to 16% based on the initial amount of cellulose,
considerably lower compared to those reported on ball-milled
−
1
−1
cellulose [14–18]. The glucose productivity, 88.9 g h kg cat-
2.3. Cellulose hydrolysis experiments
alyst, however, was higher as a result of the relatively short
hydrolysis time and lower substrate to catalyst ratio. The over-
all low yields were accredited to solid–solid mass transfer
limitations.
Hydrolysis experiments were performed in six parallel 125 mL
batch reactors (acid digestion bomb type 4748, SS 316 with Teflon
liner, Parr Instrument Company, Moline, IL). An exact amount of
substrate was mixed with demineralised water. In catalysed exper-
iments a liquid or insoluble catalyst was added. The closed reactor
vessels were placed in a heating block (adapted RS600, Thermo
Fisher Scientific, Rochford, UK) and the block temperature was
Previously, we have demonstrated the effectiveness of a
[
bmim]Cl pretreatment for the solid acid mediated hydrolysis
of Avicel cellulose as monitored by in situ ATR-IR spectroscopy
[
2
19]. The glucose productivity from regenerated cellulose was
−
1
−1
◦
−1
−1 ◦
catalyst at 180 C.
1.6 g h kg
at 150 C and 150.3 g h kg
◦
typically set at temperature ranging from 150 to 180 C. The sus-
The glucose selectivity was 64% in both cases. In this work, we
report on the progress made on the solid acid mediated hydrol-
ysis of cellulose to glucose. Herein, we will discuss the ionic
liquid pretreatment conditions in more detail. The hydrolysis
of ball-milled and regenerated cellulose is studied over various
solid and mineral acids and the glucose productivity is opti-
mised.
pensions were stirred by a magnetic rod at the bottom of the
vessels (1000 rpm). Visual tests with model slurries showed that
the mixing in the reactor was adequate, for typical experiments.
All experiments were carried out at autogenous pressure. After the
pre-set incubation time, the heating block was switched off and the
reactors in the block were allowed to cool passively.

4
40
R.J.H. Grisel, A.T. Smit / Applied Catalysis A: General 475 (2014) 438–445
Table 1
Cellulose recovery from [bmim]Cl. Reaction conditions: Avicel cellulose (as supplied,
4
% moist w/w), 5.00 g, [bmim]Cl (as supplied), 100 g.
Entry
Temperature
Incubation
(min)
Precipitate (%,
w/w)
Glucose yield
(%, w/w)
◦
(
C)
a
1
2
3
4
100
100
100
100
100
130
150
150
15
30
45
60
30
30
30
30
85
75
59
38
1.4
3.6
7.1
a
a
a
10.9
c
5
6
7
>100 (gel)
>100 (gel)
99
–
–
c
0.0
0.0
b
8
96
a
Added: AmberlystTM 15Dry (1.0 g), water (2.0 g).
Prior to cellulose dissolution, [bmim]Cl dried for 2 h at 150 C under dry N2.
Not measured.
b
c
◦
2.4. Liquor composition analysis and definitions
◦
After reaction the samples were separated by centrifugation
Thermo Scientific SL 40R, 10 min at 4000 rpm). The super-
Fig. 1. Cellulose hydrolysis over various solid acid catalysts at 150 C. Reaction
conditions: milled cellulose 1.0 g, catalyst 1.0 g, demineralised water 50 mL, 24 h
(
(
WSOC = water soluble organic compounds).
natants were analysed for glucose, fructose, 5-(hydroxymethyl)-2-
furaldehyde (HMF), furan-2-carbaldehyde (furfural), acetic, formic
and levulinic acid (HPLC, Agilent 1100 series, equipped with a RI
analyser and UV-detector and a Biorad AMINEX HPX-87H column).
Compounds were identified by elution times and quantified by cal-
ibration standards. The solid residue was washed three times with
the catalyst loading was decreased, and the pretreatment temper-
ature lowered. Both shortening the pretreatment and lowering the
catalyst loading led to the formation of gels after addition of the
anti-solvent. Precipitation with hot water under vigorous stirring
reduced the tendency to form gels, but only to a certain extent.
◦
water to remove water soluble species and dried (50 C, vacuum)
until no further mass changes were observed.
TM
Upon formation, the gels encapsulated most of the Amberlyst
Products yields are expressed in C-mol% and are calculated
as follows: product yield = (moles of product × number of carbon
atoms in the product)/(moles of carbon in initially present in the
substrate). The glucose productivity is expressed as the yield of
glucose (g) per hour of hydrolysis per kg catalyst (dry weight). The
degree of cellulose solubilisation is calculated from the dry mass
loss after reaction, corrected for the dry weight of the catalyst. For
further calculations, it was assumed that solid humins formation is
negligible and the catalyst weight is constant. Then, cellulose sol-
ubilisation in weight percent is approximately the solubilisation
in C-mol% and the total yield of water soluble organic compounds
not being glucose (WSOC) can be estimated from the difference
between the cellulose solubilisation and the glucose yield. The glu-
cose selectivity is calculated by dividing the glucose yield by the
estimated yield of all solubilised organic species including glucose.
and part of the ionic liquid and were difficult to purify. Pretreatment
temperatures below 100 C led to mixtures that were too viscous to
be mixed sufficiently and slow cellulose dissolution rates. Addition
of a mixing agent, such as DMSO, DMAc or little water, lowered the
viscosity, but also limited the cellulose dissolution capacity of the
liquid mixture.
◦
A second pretreatment series was carried out in the absence
◦
of a catalyst between 100 and 150 C (Table 1). Without catalyst,
◦
30 min incubation up to temperatures of 130 C led to the forma-
tion of gels, immediately upon addition of the anti-solvent. Within
the range studied, the tendency to form gels was not affected by
the duration of the incubation, as it was by increasing the pretreat-
ment temperature. In addition, a higher temperature lowered the
viscosity of the mixture and benefited the cellulose solubility (rate).
◦
After 30 min of incubation at 150 C over 95% (w/w) of the cellulose
could be recovered as a colloid. The liquid phase of the colloidal
suspension was slightly discoloured, indicative of some cellulose
degradation. Shorter incubation (15 min) led to a clear, predomi-
nantly colloidal system, however, with some residual gel. Longer
incubations resulted in progressive discolouration and lower cel-
lulose recoveries. Drying the ionic liquid prior to pretreatment had
certainly a positive effect on the dissolution rate of cellulose, but
not on the cellulose recovery.
3
. Results and discussion
3.1. Cellulose pretreatment in ionic liquid
Microcrystalline Avicel cellulose (DPw 200–240 AGU) was
regenerated by dissolution in [bmim]Cl followed by precipitation
with demineralised water. A first pretreatment series was carried
◦
out in the presence of AmberlystTM 15Dry as is described by Rinaldi
For Avicel, 30 min incubation in dry [bmim]Cl at 150 C followed
by precipitation with hot water yielded the best regenerated cel-
lulose in terms of recovery and reactivity. These conditions were
chosen to prepare the regenerated cellulose used in this study.
The recovered cellulose was centrifuged, washed, and diluted with
water yielding a colloidal aqueous suspension with 5% (w/w) regen-
erated cellulose. The suspension was stable for weeks when stored
et al. [35]. The amount of recovered regenerated cellulose was rel-
atively low (Table 1). Rinaldi et al. reported yields of 91% after 1 h
of incubation starting from ␣-cellulose (DPw > 2000 AGU). In our
experiments no induction period was observed; the solid mass loss
and glucose production appeared instantaneous. The dissimilarities
may be explained by the differences in degree of polymerisation of
the starting material, since an equal amount of scissions in Avicel
cellulose would lead to higher fractions of soluble sugars than in
◦
at 4 C.
␣
-cellulose.
3.2. Catalyst screening
To ensure optimum utilisation of the raw cellulose material,
the amount of recovered precipitate should be maximised while
the duration of the pretreatment should be kept to a minimum to
obtain high glucose productivities in the overall envisaged process.
In an attempt to increase yield the incubation time was shortened,
Screening experiments were carried out with ball-milled Avi-
cel cellulose at 150 C (Fig. 1). The cellulose was milled for 48 h,
similarly to the work of Onda et al. [14,15]. The results obtained
for the none-catalysed reaction are comparable with those of Onda
◦

R.J.H. Grisel, A.T. Smit / Applied Catalysis A: General 475 (2014) 438–445
441
et al., whereas hydrolysis in the presence of comparable solid acids
all resulted in lower glucose yields and typically a lower degree
of cellulose solubilisation, except for H-mordenite (10) and H-beta
(
75). In any case, the degree of cellulose solubilisation seems to be
limited to about 30%. The reasons for the observed differences are
not known, but may be related to a less effective cellulose milling
pretreatment or the different scale of reaction.
Similarities were observed as well, such as the relatively low glu-
cose selectivity over sulphated zirconia and H-beta (12.5). Herein,
circa 80% of the converted carbon was recovered in water solu-
ble organic compounds (WSOC) other than glucose, such as soluble
cellooligomers, humins, 5-hydroxymethylfurfural (HMF), levulinic
acid, formic acid, and small quantities of furfural and acetic acid.
Levels of cellobiose and fructose were below the detection limit.
The H-form zeolites with lower Si/Al molar ratio such as H-beta
(
75), H-mordenite (45) and H-Y (27.5) are less acidic in nature and
resulted in higher glucose selectivities as compared to their corre-
sponding, more acidic zeolite relatives (Fig. 1). An explanation may
be that glucose, once it is formed, is less likely to be converted into
its degradation products, once trapped in the micropores of zeolites
with lower Si/Al ratios. This does not hold for the macroreticu-
TM
lar Amberlyst
Amberlyst
ion exchange resins, in which the more acidic
TM
+
15Dry (≥4.7 mmol H /g) resulted in both a higher
degree of cellulose solubilisation and a higher glucose selectivity
TM
+
as compared to Amberlyst
open structure of these materials also allow cellooligomers to enter
70 (≥2.55 mmol H /g). Probably, the
and get hydrolysed yielding relatively more glucose with a higher
TM
density of acid sites. However, Amberlyst
15Dry is hydrother-
mally unstable under the applied conditions [14,35] and leached
acidic species may be responsible for the observed higher cellulose
solubilisation and glucose yield. Still, the higher glucose selectivity
in comparison with the also leaching sulphated zirconia is remark-
able.
The best performing catalyst in terms of glucose yield and
selectivity were obtained with non-neutralised Norit CAP Super
(
NCS), a chemically activated carbon using the phosphoric acid
process. Nonetheless, the glucose yield (21.1%) and productivity
−
1
−1
◦
(
2
9.8 g h kg catalyst) obtained after 24 h at 150 C are a factor
–4 lower in comparison with similar studies [14–18]. Nonetheless,
Fig. 2. Effect of pretreatment on the hydrolysis of cellulose over H-mordenite (45)
NCS was chosen to evaluate the effect of substrate pretreatment
on the saccharification of cellulose. The results are compared to
H-mordenite (45), the best performing zeolite in the screening
experiments, in terms of glucose yield.
and Norit CAP super (NCS): cellulose 1.0 g, catalyst 1.0 g, demineralised water 25 mL.
Cellulose solubilisation and glucose yield after (a) 24 h at 150 C, and (b) 4 h at 180 C.
◦
◦
It was noted that the hydrolysate, typically coloured light-
yellow to dark-brown due to the presence of HMF derived
breakdown products and condensates, was colourless in the pres-
ence of NCS. In the hydrolysate with NCS no HMF and only little
levulinic acid was found. It is known from literature that active car-
bons can selectively adsorb HMF in the acid catalysed dehydration
of fructose and thus slow down consecutive reactions towards lev-
ulinic and formic acid [38]. The adsorption hypothesis was tested
qualitatively at RT by addition of NCS to samples of coloured
hydrolysates obtained from the screening experiments. Addition
of NCS indeed led to discolouration of the hydrolysates. Similar
behaviour was observed by addition of other activated carbons,
such as Vulcan XC72, Norit ZN2, and Norit PAC200.
(w/w), whether or not a catalyst was present. Ball-milling led to an
increase in solubilisation and a slight increase in glucose selectivity,
however, even in the presence of a catalyst the solubilisation was
limited to just below 30% (w/w). Interestingly, large differences in
glucose selectivity were obtained, in which NCS showed superior
performance. Regenerated cellulose was found most susceptible
to solubilisation, especially in the presence of NCS. The solid sub-
strate mass loss in the presence of NCS was 76.5% (w/w) versus
39.1% (w/w) in the case of mordenite (45). Also here, the presence
of NCS resulted in the higher glucose yield and selectivity. Still, the
−
1
−1
productivity was only 25.1 g h kg catalyst, mainly due to the
long processing time.
◦
A similar comparison was performed for a 4 h process at 180 C
(Fig. 2b). Overall, similar trends were found, but both the degree
3.3. Effect of cellulose pretreatment
of substrate solubilisation and the glucose yield were increased
as compared to the 24 h process at 150 C. The yield increase
◦
The effect of the optimum cellulose [bmim]Cl treatment on the
was largest in the case of mordenite (45) and smallest for NCS.
The highest glucose yield was 54.3 C-mol% in the case of regen-
erated cellulose with NCS, corresponding to a productivity of
◦
saccharification over NCS and H-mordenite (45) at 150 C were
investigated and the results were compared to experiments start-
ing from untreated and ball-milled Avicel (Fig. 2a). In all cases
the presence of a solid acid led to an increase in glucose yield
and for pretreated cellulose also to improved substrate solubili-
sation. The solubility of untreated cellulose was limited at 5–6%
−
1
−1
153.7 g h kg catalyst. The increase in productivity originates
mainly from the shorter reaction time. Remarkably, the higher pro-
cess temperature did not lower the selectivity towards glucose
much.

4
42
R.J.H. Grisel, A.T. Smit / Applied Catalysis A: General 475 (2014) 438–445
The observed behaviour is consistent with the hypothesis that
completely but the formation of acids can be reduced by operat-
ing at shorter contact times. In addition, shorter reaction times
are interesting from the point of view of productivity, as it is
defined as yield per hour. An extensive study has been carried out
to optimise the glucose productivity from [bmim]Cl regenerated
cellulose in which the reaction time, temperature, substrate and
catalyst concentration were varied. The main results are listed in
Table 2.
the first step in saccharification of cellulose is the formation of
water soluble cellooligomers, and that subsequent hydrolysis of
the solubilised fractions to glucose is accelerated in the presence
of a solid catalyst. Herein, NCS is more effective as compared to H-
mordenite (45). The amount of substrate that is solubilised depends
on the process time and temperature, but mostly on the substrate
pretreatment. The enhanced cellulose conversion after pretreat-
ment is explained by an increased accessibility and the lower
crystallinity of the substrate, shifting the solubilisation limits.
◦
Productivity numbers for saccharification of cellulose at 150 C
were limited by the long process times required for substantial
conversions, typically 24 h (Table 2, Entry 1–4). Cellulose pretreat-
ment in [bmim]Cl led to some improved productivity due to higher
conversion rates, especially with NCS (Table 2, Entry 4). The pro-
Various hydrolysates were subjected to a post-hydrolysis step
◦
(
2 h, 100 C, 1 M H SO ) in order to determine the amount of
2
4
soluble cellooligomers originally present in the hydrolysate from
the increase of glucose concentration after post-hydrolysis. An
increase in concentration was indeed found, however, quantitative
analysis was not possible due to simultaneous glucose degradation
reactions. Besides glucose and cellooligomers, also other water-
soluble organic compounds (WSOC) were identified, such as HMF,
furfural, formic and levulinic acid, as well as traces acetic acid. After
−
1
−1
ductivity was increased to maximum 34.9 g h kg catalyst by
shortening the process from 24 to 16 h, however, at the cost of con-
version (Table 2, Entry 5). Further decreasing the process time does
not seem functional without simultaneously increasing the tem-
perature to compensate for the observed loss of activity. At a 4 h
◦
process a temperature raise to 180 C more than makes up for the
◦
processing at 150 C the combined yield of these compounds was
shorter reaction time in terms of conversion, while the selectivity
to glucose increased for H-mordenite (45) (Table 2, Entry 7–8). The
selectivity for the NCS catalysed reaction decreased to some extent,
however still overall resulting in a 6–8 times higher productivity
(Table 2, Entry 10–11).
1
–2 C-mol% for all catalysts and all substrates, indicative of limited
glucose degradation under these conditions. After processing
◦
regenerated cellulose at 180 C, an increased HMF yield was found
for the none-catalysed reaction, viz. 1.1 C-mol%. For the reaction
catalysed by H-mordenite (45) increased levels of formic and lev-
ulinic acid were observed, respectively 0.9 and 2.1 C-mol%. Despite
of the considerably higher activity, the yields of HMF (0.1 C-mol%),
formic acid (0.5 C-mol%) and levulinic acid (0.3 C-mol%) were
significantly lower when NCS was used as the catalyst. The lower
yields are thought to be related to the adsorption capability of NCS.
Reducing the substrate concentration at fixed substrate to cat-
alyst ratio led to a slight improvement in the glucose productivity
−
1
−1
over NCS from 153.7 to 172.9 g h kg catalyst (Table 2, Entry
11–14). The improvement may be correlated to slower glucose
degradation reactions as a result of the lower glucose concentra-
tion. This is in agreement with the higher selectivity found for
the lower concentrations. Quite the opposite was found for H-
mordenite (45), in which higher substrate ratios were preferred
3
.4. Optimising glucose productivity
(Table 2, Entry 8, 15–17). The improved productivity is directly
related to an enhanced selectivity. The degree of cellulose solu-
bilisation is similar for all cases. Using H-mordenite (45) at low
In situ spectroscopic results [19] suggest that preventing degra-
dation of glucose to products, such as HMF, cannot be avoided
Table 2
Cellulose saccharification over solid acids. Reaction conditions: catalyst 50% (w/w, based on total dry weight), demineralised water 25 mL.
◦
Entry
Catalyst
Temperature ( C)
Time (h)
Substrate
Solubilisation
(%, w/w)
Yield
(C-mol%)
Selectivity
(C-mol%)
Productivity
(g h kg
a
b
−1
−1
)
(
g)
1c
2
3
H-mordenite (45)
H-mordenite (45)
Norit CAP Super
Norit CAP Super
Norit CAP Super
H-mordenite (45)
H-mordenite (45)
H-mordenite (45)
Norit CAP Super
Norit CAP Super
Norit CAP Super
Norit CAP Super
Norit CAP Super
Norit CAP Super
H-mordenite (45)
H-mordenite (45)
H-mordenite (45)
Norit CAP Super
Norit CAP Super
Norit CAP Super
Norit CAP Super
Norit CAP Super
Norit CAP Super
Norit CAP Super
150
150
150
150
150
180
180
180
180
180
180
180
180
180
180
180
180
180
180
180
180
170
170
170
24
24
24
24
16
4
4
4
4
4
4
4
4
4
4
4
4
4
0.96
0.25
1.00
0.50
0.50
1.00
1.00
1.00
1.00
1.00
1.02
0.75
0.50
0.25
0.25
0.50
0.75
1.02
1.00
1.00
1.00
1.01
1.00
1.00
29.9
39.1
27.1
76.5
66.6
10.6
40.3
37.0
22.7
43.1
83.2
87.0
88.1
83.9
40.9
37.8
36.3
82.7
83.0
68.5
88.7
64.6
76.9
85.1
9.0
9.5
21.1
53.9
50.1
5.4
22.0
26.3
8.2
28.4
54.3
55.8
61.1
62.2
18.8
22.2
24.1
61.3
61.2
45.4
63.0
43.0
54.3
60.8
30.1
24.2
77.8
70.5
75.2
50.4
54.6
71.1
36.2
65.9
65.3
64.2
69.4
74.2
45.8
58.9
66.2
74.0
73.8
66.3
71.0
66.6
70.5
71.5
4.0
4.4
9.7
25.1
34.9
15.0
61.1
72.5
23.1
c
4
5
6
d
c
7
8
9
d
c
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
78.8
153.7
155.5
169.4
172.9
52.3
61.4
67.2
e
f
f
f
f
f
f
346.8
685.7
674.4
468.5
481.9
390.5
340.3
4
3
6
4
6
8
a
b
c
d
e
f
Dry weight.
Solid mass loss not including catalyst weight.
Regenerated cellulose substituted with ball-milled cellulose.
Regenerated cellulose substituted with Avicel cellulose.
Catalyst loading 33% (w/w) based on total dry weight.
Catalyst loading 25% (w/w) based on total dry weight.

R.J.H. Grisel, A.T. Smit / Applied Catalysis A: General 475 (2014) 438–445
443
substrate loadings, it is thought that after reaction still a signifi-
cant amount of cellulose resides in the hydrolysate as incompletely
hydrolysed, soluble cellooligomers. At higher substrate loadings
it may be expected that the concentration of those soluble cel-
looligomers in the hydrolysate is higher and, hence, there will be
a higher probability for soluble cellooligomers to (successfully)
interact with the catalyst yielding glucose in more concentrated
suspensions [16,39]. If so, regarding the unselective nature of this
catalyst, also higher yields of glucose degradation products may be
expected. Indeed, an increased levulinic acid yield was observed
when increasing from 0.8 to 2.1 C-mol% in the more concentrated
suspensions. The yields of HMF and formic acid, were stable around
With subsequent heat treatments, the activity of the NCS
catalyst reduced significantly (Table 3). In addition, the liquor
originating from the first hydrothermal treatment was active in cel-
lulose hydrolysis, indicative of leaching of catalytic active species
into the liquid phase upon hydrothermal treatment. Indeed, PO43
−
was found in the hydrolysates and the leaching of acidity was
confirmed by the acidic character of the liquors (pH 2.2). The phos-
phate content was quantified by standard Ion Chromatography
(IC) analysis with conductivity detection (Dionex DS3 conductiv-
ity analyser). The error in measured concentration is <0.1 mmol/L.
As expected, the amount of phosphate detected in the hydrolysate
became smaller with subsequent hydrothermal treatments.
0
.1 and 0.9 C-mol%, respectively. The highest glucose productivity
Remarkable is the relative high glucose selectivity when NCS
is present, irrespective of the activity. Apparently, close contact
is beneficial for high glucose selectivity. The highest selectivity,
94.5 C-mol%, was obtained after co-milling cellulose with (fresh)
NCS in a ball-mill in a 1:1 weight ratio. The higher observed selectiv-
ity is expected to originate from adsorption of glucose degradation
products, such as HMF and derivatives immediately upon their for-
mation. Since the selectivity is expressed as the quotient of the
amount of carbon in glucose over the amount of carbon in all solu-
bilised species, adsorption of non-glucose species leads to a higher
apparent glucose selectivity. Nonetheless, the corresponding glu-
−
1
−1
obtained with H-mordenite was 72.5 g h kg catalyst (Table 2,
Entry 8), roughly half the amount obtained with NCS at similar
conditions. Further optimisation was limited to NCS catalysed reac-
tions, only.
Reducing the NCS catalyst loading to 33% and 25% (w/w, dry
weight) resulted in similar conversions at an improved glucose
selectivity (Table 2, 18–19) as compared to a similar experiment
with 50% catalyst loading (Table 2, Entry 11). Apparently, the cata-
lyst concentration is not limiting at loadings above 25% (w/w, dry
weight). As a result of the improved performance and the lower
amount of catalyst required, the glucose productivity added up
−
1
−1
cose productivity was 108.1 g h kg catalyst.
−
1
−1
to 685.7 g h kg catalyst at a glucose selectivity of 73.8 C-mol%.
Subsequent alterations in time (Table 2, Entry 20–21) and tempera-
ture (Table 2, Entry 22–24) did not result in further improvements,
due to lower glucose yields per hour of reaction.
3.6. Sorption enhanced saccharification of cellulose
To evaluate the true effects of NCS as a solid catalyst, hydrolysis
of regenerated cellulose was performed with H PO in the absence
3
4
3.5. Catalyst stability
and presence of NCS. The H PO4 concentration chosen was based
3
on the concentration of phosphate found in the hydrolysate after
hydrolysis with fresh NCS. Special attention was given to the detec-
tion of HMF and levulinic acid, the key by-products in acid catalysed
hydrolysis of cellulose.
To test the robustness of the catalyst, NCS was heated in dem-
◦
ineralised water for 4 h at 200 C and autogenous pressure. The
sample was centrifuged and the supernatant was separated by
decantation and stored. The solid was washed and dried in vacuum
Under the applied conditions, dilute H PO4 was catalytically
3
◦
at 50 C. This procedure was repeated on a portion of the resulting
active yielding glucose with 66.3% selectivity. The major by-product
identified was HMF. Also small quantities of levulinic acid, formic
acid, and 2-furaldehyde and traces acetic acid were found. At a
nearly similar yield, H-mordenite (45) performed slightly better
with 71.9% selectivity towards glucose. The amount of HMF found,
treated material. Hydrolysis of ball-milled cellulose was performed
on once and twice treated NCS, as well as in the supernatant
resulting from the first treatment. The results were compared to
hydrolysis using fresh NCS.
Table 3
◦
Hydrolysis of ball-milled cellulose. Reaction conditions unless noted otherwise: 1.00 g cellulose (dry weight), 1.00 g NCS, 25 mL H2O, T = 180 C at autogenous pressure and
t = 4 h.
−
Catalyst
–
Cellulose dissolved (%, w/w)
Glucose yield (C-mol%)
Glucose selectivity (C-mol%)
PO43 (mmol/L)
14.0
32.1
43.1
32.1
22.0
41.1
5.2
12.6
28.4
16.6
14.5
38.9
37.5
39.3
65.9
51.6
65.9
94.5
–e
a
–
14.6
10.6
1.8
0.3
–e
NCS
NCS
NCS
NCS
b
c
d
a
Water replaced by liquor resulting from first hydrothermal treatment.
Hydrothermally treated at 200 C, 4 h.
Twice hydrothermally treated.
Co-milled with cellulose.
Not measured.
b
c
◦
d
e
Table 4
◦
Hydrolysis of regenerated cellulose. Reaction conditions: unless noted otherwise: 1.00 g cellulose (dry weight), 1.00 g catalyst (dry weight), 25 mL H2O, T = 180 C at autogenous
pressure and t = 4 h.
Catalyst
–
Glucose yield (C-mol%)
HMF yield (C-mol%)
Levulinic acid yield (C-mol%)
Formic acid yield (C-mol%)
7.5
25.2
26.3
54.3
54.6
0.9
2.3
0.1
0.2
0.3
0.0
0.6
2.1
0.1
0.2
0.2
0.5
0.9
0.1
0.1
a
–
MOR (45)
NCS
a
NCS
a
Added H3PO4 (10.8 mmol/L).

4
44
R.J.H. Grisel, A.T. Smit / Applied Catalysis A: General 475 (2014) 438–445
Fig. 3. Proposed reaction pathway for cellulose hydrolysis.
Adapted from Ref. [19].
however, was low. The amounts of levulinic and formic acid, on
the other hand, were higher, indicative of enhanced decomposi-
tion of HMF over this catalyst. Using NCS, the glucose selectivity was
4. Conclusions
A limited part of microcrystalline cellulose can be hydrolysed
under relatively mild conditions to form soluble cellooligomers and
glucose, without the formation of large amounts of degradation
products. A catalyst is not required per se, but the presence of an
acid accelerates the depolymerisation and may steer the reaction
further towards the desired product. However, the glucose produc-
tivity is limited.
To improve the productivity, microcrystalline cellulose must
first be pretreated to enhance its reactivity. Ball-milling or disso-
lution/regeneration from an ionic liquid, such as [bmim]Cl results
in more reactive celluloses. In particular, regenerated cellulose can
be converted to glucose at high yields and high selectivity using
(solid) acids. Herein, proper selection of pretreatment conditions is
important to prevent gelling and substrate loses.
Regenerated cellulose can be converted into glucose with high
selectivity over microporous zeolites, in which mildly acidic zeo-
lites with larger pores are preferred. The glucose production over
zeolites is limited by restricted cellulose solubilisation under mild
conditions and by progressive glucose degradation reactions at
more severe conditions. Highest yields were obtained with H-
6
5.3%, but due to the superior solubilisation activity, still higher glu-
cose yields were obtained (Table 4). Despite of the higher activity,
the amount of typical by-products found was limited. The spe-
cific selectivity, as well as the high activity, was maintained upon
addition of H PO to the NCS; in the NCS/H PO mixture, NCS dom-
3
4
3
4
inated the results. It was noted that all NCS containing hydrolysates
were colourless.
Clearly, NCS is not stable under typical reaction conditions,
in a sense that phosphoric acid groups leach into the solution.
Undoubtedly, these species will take part in the hydrolysis of cel-
lulose. However, when comparing the results of NCS and H PO
3
4
catalysed hydrolysis (at equal final PO43 concentration) some dif-
ferences can be observed. First, for both ball-milled and regenerated
cellulose, the carbon catalyst results in a higher degree of solu-
bilisation. Apparently, the leached phosphorous species are not
the only species active in the depolymerisation of cellulose. Sec-
ond, in the presence of NCS the concentrations of by-products,
in particular glucose degradation products, such as HMF, levulinic
acid and formic acid, are much lower. We explain this by assum-
ing that carbons like NCS act as scavengers for HMF (derivatives),
intermediate(s) in the formation of formic and levulinic acid. In
addition, HMF is an important species in the formation of humins
−
−
1
−1
mordenite (Si/Al = 45), producing glucose at a rate of 72.5 g h kg
catalyst with 71.1% selectivity.
Considerably higher glucose productivities are obtained with
Norit CAP Super (NCS), 685.7 g h kg catalyst at 73.8% selec-
−
1
−1
(Fig. 3).
Once formed, organic acids, such as formic and levulinic acid,
tivity. The high productivity results from a high solubilisation
activity at low catalyst loadings and short processing time at ele-
vated temperatures, without losing much of the selectivity. NCS
is not hydrothermally stable and loses activity due to leaching of
phosphate moieties. These leached species aid in solubilising the
substrate. Still, NCS retains some intrinsic activity and outperforms
phosphoric acid in terms of selectivity. We hypothesise that NCS,
and related carbon species, is able to promote higher glucose yield
and selectivity due to in situ scavenging of HMF and derivatives
thereof, thus preventing the formation of organic acids and soluble
humins that otherwise trap and convert glucose.
catalyse their own formation, a process in which glucose is con-
sumed. In situ removal of HMF (D) suppresses the subsequent
formation of levulinic acid (E) and formic acid (F), thus inhibiting
the auto-catalysed glucose degradation. In addition, due to removal
of species reactive in humins formation, the actual loss of glucose
(
B) by condensation will also be reduced. Both features lead to
increased observed glucose selectivity. These effects will be most
pronounced at high product concentrations, and thus, in the case of
highly reactive [bmim]Cl regenerated cellulose. Upon adsorption,
the weight of the carbon catalyst, and thus of the solid residue, must
increase. Indeed, a slight increase in solid-weight was observed by
screening experiments with glucose and NCS.
Acknowledgements
Current study illustrates that adding a scavenger is beneficial
for the (selective) production of glucose from cellulose by (solid)
acid catalysed hydrolysis. Further research should focus on the
development of a more hydrothermally stable catalyst with intrin-
sic selective sorption capacity for glucose degradation products,
or a catalytic system consisting of a (solid) acid catalyst and an
optimised (sacrificial) scavenger.
Ben van Egmond and Karina Vogelpoel-de Wit are acknowl-
edged for their contribution to the experimental work. This
research has been performed within the framework of the Catch-
Bio program. The authors gratefully acknowledge the support of
the Smart Mix Program of the Dutch Ministry of Economic Affairs
and the Dutch Ministry of Education, Culture and Science.

R.J.H. Grisel, A.T. Smit / Applied Catalysis A: General 475 (2014) 438–445
445
References
[20] S.M. Hick, C. Griebel, D.T. Restrepo, J.H. Truitt, E.J. Buker, C. Bylda, R.G. Blair,
Green Chem. 12 (3) (2010) 468–474.
[
[
[
21] M. Benoit, A. Rodrigues, Q. Zhang, E. Fourré, K. de Oliveira Vigier, J.M. Tatibouët,
F. Jérôme, Angew. Chem. 123 (38) (2011) 9126–9129.
22] F. Guo, Z. Fang, C.C. Xu, R.L. Smith Jr., Prog. Energy Combust. Sci. 38 (5) (2012)
[
1] T. Werpy, G. Peterson, A. Aden, J. Bozell, J. Holladay, J. White, A. Manheim, Top
Value Added Chemicals from Biomass, U.S. Department of Energy, Oak Ridge,
2
004.
6
72–690.
23] R.P. Swatloski, S.K. Spear, J.D. Holbrey, R.D. Rogers, J. Am. Chem. Soc. 124 (18)
2002) 4974–4975.
[2] B. Kamm, P.R. Gruber, M. Kamm, Biorefineries, Industrial Processes and Prod-
ucts, Wiley-VCH, Weinheim, 2006.
(
[
[
[
3] A. Corma, S. Iborra, A. Velty, Chem. Rev. 107 (6) (2007) 2411–2502.
4] R. Rinaldi, F. Schüth, ChemSusChem 2 (12) (2009) 1096–1107.
5] S. van de Vyver, J. Geboers, P.A. Jacobs, B.F. Sels, ChemCatChem 3 (1) (2011)
[
[
[
[
[
24] C. Li, Z.K. Zhao, Adv. Synth. Catal. 349 (11–12) (2007) 1847–1850.
25] S.J. Dee, A.T. Bell, ChemSusChem 4 (8) (2011) 1166–1173.
26] A.S. Amarasekara, O.S. Owereh, Catal. Commun. 11 (13) (2010) 1072–1075.
27] T. Furong, H. Song, L. Chou, ChemSusChem 3 (2010) 1298–1303.
28] Z.-D. Ding, J.-C. Shi, J.-J. Xiao, W.-X. Gu, C.-G. Zheng, H.-J. Wang, Carbohydr.
Polym. 90 (2) (2012) 792–798.
29] D. Feng, L. Li, F. Yang, W. Tan, G. Zhao, H. Zou, M. Xian, Y. Zhang, Appl. Microbiol.
Biotechnol. 91 (2) (2011) 399–405.
30] B.R. Caes, J.B. Binder, J.J. Blank, R.T. Raines, Green Chem. 13 (10) (2011)
8
2–94.
[
[
[
6] H. Kobayashi, A. Fukuoka, Green Chem. 15 (7) (2013) 1740–1763.
7] A.C. O’Sullivan, Cellulose 4 (3) (1997) 173–207.
8] H. Krässig, J. Schurz, R.G. Steadman, K. Schliefer, W. Albrecht, M. Mohring, H.
Schlosser, Cellulose, Ullmann’s Encyclopedia of Industrial Chemistry, Wiley-
VCH, Weinheim, 2004.
[
[
[
[9] R.J. Moon, A. Martini, J. Nairn, J. Simonsen, J. Youngblood, Chem. Soc. Rev. 40
2
719–2722.
31] M.B. Turner, S.K. Spear, J.D. Holbrey, R.D. Rogers, Biomacromolecules 5 (4)
2004) 1379–1384.
(
2011) 3941–3994.
[
[
[
10] Q. Gan, S.J. Allen, G. Taylor, Process Biochem. 38 (7) (2003) 1003–1018.
11] M. Balat, Energy Convers. Manage. 52 (2) (2011) 858–875.
12] B. Girisuta, L.P.B.M. Janssen, H.J. Heeres, Ind. Eng. Chem. Res. 46 (6) (2007)
(
[
[
32] A.P. Dadi, S. Varanasi, C.A. Schall, Biotechnol. Bioeng. 95 (5) (2006) 904–910.
33] A.P. Dadi, S. Varanasi, C.A. Schall, Appl. Biochem. Biotechnol 137–140 (1–12)
1
696–1708.
(
2007) 407–421.
34] H. Zhao, C.L. Jones, G.A. Baker, S. Xia, O. Olubajo, V.N. Person, J. Biotechnol. 139
1) (2009) 47–54.
35] R. Rinaldi, R. Palkovits, F. Schüth, Angew. Chem. Int. Ed. 47 (42) (2008)
047–8050.
36] R. Rinaldi, M. Meine, J. vom Stein, R. Palkovits, F. Schüth, ChemSusChem 3 (2)
2010) 266–276.
[
13] S. Suganuma, K. Nakajima, M. Kitano, D. Yamaguchi, H. Kato, S. Hayashi, M.
[
[
[
[
Hara, J. Am. Chem. Soc. 130 (38) (2008) 12787–12793.
(
[
[
[
14] A. Onda, T. Ochi, K. Yanagisawa, Green Chem. 10 (10) (2008) 1033–1037.
15] A. Onda, T. Ochi, K. Yanagisawa, Top. Catal. 52 (6–7) (2009) 801–807.
16] D. Yamaguchi, M. Kitano, S. Suganuma, K. Nakajima, H. Kato, M. Hara, J. Phys.
Chem. C 113 (8) (2009) 3181–3188.
8
(
[
[
[
17] J. Pang, A. Wang, M. Zheng, T. Zhang, Chem. Commun. 46 (37) (2010)
37] S.-J. Kim, A.A. Dwiatmoko, J.W. Choi, Y.-W. Suh, D.J. Suh, M. Oh, Bioresour.
Technol. 101 (21) (2010) 8273–8279.
38] P. Vinke, H. van Bekkum, Starch 4 (3) (1992) 90–96.
39] B. Op de Beek, J. Geboers, S. van de Vyver, J. van Lishout, J. Snelders, W.J.J. Hui-
jgen, C.M. Courtin, P.A. Jacobs, B.F. Sels, ChemSusChem 6 (1) (2013) 199–208.
6
935–6937.
18] S. van de Vyver, L. Peng, J. Geboers, H. Schepers, F. de Clippel, C.J. Gommes, B.
Goderis, P.A. Jacobs, B.F. Sels, Green Chem. 12 (9) (2010) 1560–1563.
19] J. Zakzeski, R.J.H. Grisel, A.T. Smit, B.M. Weckhuysen, ChemSusChem 5 (2)
[
[
(
2012) 430–437.