Coupling metal halides with a co-solvent to produce furfural and 5-HMF at high yields directly from lignocellulosic biomass as an integrated biofuels strategy

Article abstract of DOI:10.1039/c4gc00747f

Metal halides are selective catalysts suitable for production of the fuel precursors furfural and 5-HMF from sugars derived from lignocellulosic biomass. However, they do not perform nearly as well when applied to biomass even in combination with immiscible extracting solvents or expensive ionic co-solvents. Here, we couple metal halides with a highly tunable co-solvent system employing renewable tetrahydrofuran (THF) to significantly enhance co-production of furfural and 5-HMF from biomass in a single phase reaction strategy capable of integrating biomass deconstruction with catalytic dehydration of sugars. Screening of several promising metal halide species at 170 °C in pH-controlled reactions with sugar solutions and larger 1 L reactions with maple wood and corn stover revealed how the interplay between relative Bronsted and Lewis acidities was responsible for enhancing catalytic performance in THF co-solvent. Combining FeCl₃ with THF co-solvent was particularly effective, achieving one of the highest reported simultaneous yields of furfural (95%) and 5-HMF (51%) directly from biomass with minimal levulinic acid formation (6%). Furthermore, over 90% of the lignin from biomass was extracted by THF and recovered as a fine lignin powder. Tuning the volume ratio of THF to water from 4 : 1 to 1 : 1 preserved 10% to 31% of the reacted biomass as a glucan-rich solid suitable for further catalytic reaction, enzymatic digestion, or possible pulp and paper production. This journal is the Partner Organisations 2014.

Full text of DOI:10.1039/c4gc00747f

Green Chemistry
View Article Online
View Journal | View Issue
PAPER
Coupling metal halides with a co-solvent to
produce furfural and 5-HMF at high yields directly
from lignocellulosic biomass as an integrated
biofuels strategy†
Cite this: Green Chem., 2014, 16,
3819
Charles M. Cai,a,b Nikhil Nagane,a,b Rajeev Kumara and Charles E. Wyman*a,b
Metal halides are selective catalysts suitable for production of the fuel precursors furfural and 5-HMF from
sugars derived from lignocellulosic biomass. However, they do not perform nearly as well when applied
to biomass even in combination with immiscible extracting solvents or expensive ionic co-solvents. Here,
we couple metal halides with a highly tunable co-solvent system employing renewable tetrahydrofuran
(THF) to significantly enhance co-production of furfural and 5-HMF from biomass in a single phase reac-
tion strategy capable of integrating biomass deconstruction with catalytic dehydration of sugars. Screen-
ing of several promising metal halide species at 170 °C in pH-controlled reactions with sugar solutions
and larger 1 L reactions with maple wood and corn stover revealed how the interplay between relative
Brønsted and Lewis acidities was responsible for enhancing catalytic performance in THF co-solvent.
Combining FeCl3 with THF co-solvent was particularly effective, achieving one of the highest reported
simultaneous yields of furfural (95%) and 5-HMF (51%) directly from biomass with minimal levulinic acid
formation (6%). Furthermore, over 90% of the lignin from biomass was extracted by THF and recovered as
a fine lignin powder. Tuning the volume ratio of THF to water from 4 : 1 to 1 : 1 preserved 10% to 31% of
the reacted biomass as a glucan-rich solid suitable for further catalytic reaction, enzymatic digestion, or
possible pulp and paper production.
Received 25th April 2014,
Accepted 2nd June 2014
DOI: 10.1039/c4gc00747f
www.rsc.org/greenchem
petitiveness issues.4,5 What we term as fuel precursors (FPs)
must generally be derived as intermediate platform com-
1. Introduction
Lignocellulosic biomass in such forms as agricultural and for- pounds6 from lignocellulosic biomass before conversion to
estry residues and herbaceous and woody energy crops is the “drop-in” liquid transportation fuels and other fuel products is
only sufficiently prevalent sustainable resource for impactful possible. However, the major challenge to realizing this pathway
conversion into renewable liquid transportation fuels.1,2 has been to produce primary fuel precursors, such as mono-
Furthermore, because lignocellulosic biomass sold at $60 meric sugars, and secondary fuel precursors, such as furfural,
ton−1 is theoretically equivalent in energy cost to oil at about 5-hydroxymethylfurfural (5-HMF) and levulinic acid (LA), directly
$20 per barrel, it provides the most promising near-term from biomass at the high yields (>80% recovery of C5 and C6
option for achieving low enough costs to alleviate our depen- products) essential to low unit costs (<$5 gal−1)7 without compli-
dence on fossil resources.3 The conversion of cellulosics into cated processes, expensive catalysts, and/or high energy demand.
compatible transportation fuels has enormous benefits for
Fig. 1 outlines a reaction network for the production of
addressing global climate change, energy security, rural econ- ethanol and promising gasoline, jet, and diesel range fuel pro-
omic growth and employment, trade deficits, and global com- ducts from primary and secondary FPs. As shown, xylose from
hemicellulose and glucose from cellulose can be fermented to
ethanol or dehydrated with acid catalysts to produce furfural
and 5-HMF. Further 5-HMF hydrolysis results in equimolar for-
aCenter for Environmental Research and Technology (CE-CERT), Bourns College of
Engineering, University of California, Riverside, 1084 Columbia Avenue, Riverside,
mation of more stable products LA and formic acid (not shown).
LA can also be synthesized from furfural by a furfuryl alcohol
intermediate. These secondary FPs can be catalytically upgraded
into potential fuel products by selective hydrogenation over
metal-based solid catalysts.8 As shown, catalytic hydrogenation
of furfural and 5-HMF produces promising gasoline blending
California 92507, USA
bDepartment of Chemical and Environmental Engineering, Bourns College of
Engineering, University of California, Riverside, 446 Winston Chung Hall,
900 University Ave., Riverside, CA 92507, USA. E-mail: cewyman@engr.ucr.edu;
Fax: +1 (951) 781-5790; Tel: +1 (951) 781-5703
†Electronic supplementary information (ESI) available. See DOI:
10.1039/c4gc00747f
This journal is © The Royal Society of Chemistry 2014
Green Chem., 2014, 16, 3819–3829 | 3819

View Article Online
Paper
Green Chemistry
steam stripping to remove furfural, use of corrosive mineral
acids, and/or biphasic reactions. Alternatively, co-production
of furfural and 5-HMF would appear more desirable as both
products could be recovered together by a suitable extracting
solvent and simultaneously converted into “drop-in” fuels
such as MF and DMF by a single catalyst.21
Recently, we demonstrated that tetrahydrofuran (THF) is an
exceptionally effective single phase co-solvent for integrated
biomass reactions that enhance fuel precursor yields during
biomass deconstruction, as well as delignification.22 Using
just dilute sulfuric acid in a miscible solution of THF and
water, we achieved higher overall yields of furfural, 5-HMF,
and LA from maple wood than previously reported in a single
phase reaction.22 However, because sulfuric acid favored fur-
fural and LA production, it became apparent that tuning of
this co-solvent system with different catalysts could improve
yields for co-production of furfural and 5-HMF. Because
aqueous monophasic reactions with dilute mineral acids typi-
cally suffer from low 5-HMF yields (<5%)9,22 as it readily hydro-
lyzes to form LA and formic acid, methyl isobutyl ketone
(MIBK) was employed as an extracting solvent in a biphasic
reaction,23 but solvent recovery was an issue and the high
energy requirements for heating and stirring and limited
effective solids loading of a biphasic reaction for large scale
fuel production from solid biomass hinder its commercial
appeal.22 Thus, a single phase reaction is beneficial if a more
selective acid catalyst could be used to improve selectivity of
biomass glucan to 5-HMF instead of LA.
Metal halides are inexpensive acid catalysts that are well
studied for selectively promoting alternate reaction mecha-
nisms of xylose to furfural and glucose to 5-HMF compared to
traditional mineral acids.24–26 In analogous pathways, aldose-
to-ketose isomerization of glucose to fructose and xylose to
xylulose was observed in the presence of certain bi- and tri-
valent metal cations that can more easily undergo acid-cata-
lyzed dehydration.26 Metal cations such as Al3+, Fe3+, and Cr3+
also form oxides with water molecules that increase the acidity
of the solution as a Brønsted acid.27 When bound to halide
anions such as I–, Br–, and Cl–, the Lewis acid/base pair can
further catalyze production of furfural and 5-HMF from aldose
sugars through consecutive dehydration reactions that first
proceed by enolization to a 1,2-enediol intermediate.25 However,
evidence also suggests that the strong Lewis acid character of
metal halides accelerates several competing loss reactions that
could potentially decrease product yields. When used in bipha-
sic28 and expensive ionic-liquid (IL) co-solvent29 systems, metal
halides demonstrated good performance with sugar solutions
but poor performance (19% to 26% 5-HMF, 51% to 66% fur-
fural for biphasic, <30% for IL) on cellulose and biomass,30
necessitating additional biomass pretreatment.
Fig. 1 Reaction network illustrating primary and secondary fuel precur-
sors for production of ethanol and gasoline, jet, and diesel range aro-
matic and hydrocarbon fuels. 5-HMF: 5-hydroxymethylfurfural; MF:
2-methylfuran; DMF: 2,5-dimethylfuran; MTHF: 2-methyltetrahydro-
furan; Fur-Alc: furfuryl alcohol.
products 2-methylfuran (MF, 131 Research Octane Number
RON)9 and 2,5-dimethylfuran (DMF, 119 RON),10 respectively.
2-Methyltetrahydrofuran (MTHF, 86 RON)11 can be produced
from hydrogenation of LA or MF and ethanol can be produced
from sugars by yeast and/or bacteria fermentation,12 both of
which are primary components in P-series biofuels. Ethanolysis
of LA produces ethyl levulinate,13 a diesel blendstock, whereas
aldol-addition using acetone and hydrodeoxygenation of second-
ary fuel precursors with hydrogen can produce longer-chained
hydrocarbon fuels of up to 16 carbon lengths for jet and diesel
applications.8,14
Ongoing advances in catalysis have improved the selective
conversion of secondary fuel precursors to so-called drop-in
fuel products that are compatible with the existing fuel
infrastructure,15–18 but obtaining high overall fuel precursor
yields directly from lignocellulosic biomass has been a long-
standing barrier to achieving reasonable product costs of
<$5 gal−1 (ref. 19 and 20). Thus, there is a pressing need to
develop effective strategies that integrate catalytic conversion
with biomass deconstruction to co-produce FPs from both C5
and C6 sugars in order for biomass drop-in fuels to have
impact.19,20 Achieving high overall product yields from the
major biomass fractions hemicellulose, cellulose, and lignin
in an integrated process has the highest potential to enable
future biomass-to-fuel technologies.19 Various acid catalyzed
co-production schemes from biomass are possible including
co-producing furfural with LA, furfural with 5-HMF, furfural
with cellulose, and LA from both furfural and 5-HMF, but
many suffer from low yields due to the complex heterogeneous
nature of biomass.9 For example, furfural and 5-HMF pro-
duced early in biomass deconstruction are rapidly degraded
before sufficient LA yields from C6 sugars can be achieved.
Consequently, LA production and recovery would have to
follow furfural removal, thereby necessitating multi-stage reac-
tions with independent product recovery steps, expensive
In this paper, we demonstrate that metal halide acid cata-
lysts in combination with THF as a novel miscible co-solvent
can significantly improve yields for co-production of both fur-
fural and 5-HMF from lignocellulosic biomass such as maple
wood and corn stover from that possible before. In this way,
biomass pretreatment and catalytic dehydration of soluble
3820 | Green Chem., 2014, 16, 3819–3829
This journal is © The Royal Society of Chemistry 2014

View Article Online
Green Chemistry
Paper
sugars can be performed in a one-pot reaction. We first investi- tities and specific quantification of these minor sugars was
gated five promising metal salt acid catalysts AlCl3, CuCl2, difficult via HPLC, we elected to treat all quantified biomass
CrCl3, FeCl3, and ZrOCl2 for their sugar conversion and selecti- pentosans as xylan and all hexosans as glucan.
vity to furfural, 5-HMF, and LA production by applying our co-
solvent system to sugar solutions. We then optimized reaction
severity and solvent loadings in 1 L scale biomass reactions
with maple wood and corn stover catalyzed by these metal
2.2 THF co-solvent sugar reactions
halides to achieve the highest furfural and 5-HMF yields Sugar solutions were prepared in 1 : 1 THF–water co-solvent
reported from these feedstocks by a single phase reaction strat- mixture containing 20 g L−1 glucose or 10 g L−1 xylose and
egy. The results reveal how different Brønsted and Lewis activi- 0.1 M (anhydrous) loading of the metal halide catalyst based
ties of metal halide acid catalysts can play a key role in on total liquid volume. Due to the different acidities of each
harmonizing the dehydration kinetics of both C5 and C6 catalyst, all solutions were normalized to 1.6 pH by titrating
sugars and degradation reactions of the final products to maxi- with 72 wt% concentrated sulfuric acid. An acidity of 1.6 pH
mize overall yields of furanic products for a biorefinery was selected because it was below the Brønsted acidity of the
process.
most acidic 0.1 M ZrOCl2-containing mixture (pH 1.65). Pure
sugar solutions containing only sulfuric acid were also pre-
pared and titrated to 1.6 pH to directly compare with metal
halide acid catalysts as an acid control.
The reactions were carried out in non-stirred 14.3 mL
Hastelloy tube reactors (Hastelloy C-276, O.D. of 0.0127 m or
0.5 in.) with a wall thickness of 0.0008890 m (0.035 in.) and
2. Materials and methods
2.1 Experimental materials
Reagent-grade THF (>99% purity, Fisher Scientific, Pittsburgh, length of 0.1524 m (6 in.) to give a working reaction liquid
PA) was used in all THF co-solvent reactions. The THF co- volume of 10 mL. The tube reactors were loaded into a heavy-
solvent solution was prepared on a volume basis of increasing duty custom steam chamber made of readily available steam
the amount of THF additions to realize 1 : 1 (THF 50% v/v) to rated (to 1 MPa steam pressure) 316 stainless steel 0.102 m (4
7 : 1 (THF 87.5% v/v) THF-to-water ratios. Hydrated metal in.) internal diameter fittings (McMaster, Santa Fe Springs,
halide catalysts and ≥99% pure xylose and glucose sugars were CA). A high-pressure steam boiler (FB-075-L, Fulton Compa-
purchased from Sigma Aldrich (St. Louis, MO, US). Due to the nies, Pulaski, NY) provided steam for rapid and stable heating
high purity of the purchased sugars from Sigma, we use the of triplicate tube reactors. Temperature was monitored by both
term pure sugars to designate solutions prepared using these in-line pressure gauges and two K-type thermoprobes (Omega
sugars. The hydrate form of each metal halide catalyst Engineering Co., Stamford, CT) and controlled by a PID con-
(AlCl3·6H2O, CuCl2·2H2O, CrCl3·6H2O, FeCl3·6H2O, and troller via steam pressure. Due to the lengthwise construction
ZrOCl2·8H2O) was used but were loaded based on their equi- of the tube reactors and the application of steam for heating
valent anhydrous mass to achieve 0.1 M or 1 wt% catalyst and cold water for quenching, heat transfer was relatively
loading. Concentrated sulfuric acid (72 wt% H2SO4) was pur- rapid (<30 s) even for shorter reaction times (<10 min).32
chased from Ricca Chemical Company (Arlington, TX) and Initial time was defined when a reaction temperature of 170 °C
used to make dilute sulfuric acid solutions.
was reached. At the end of the reaction, the steam supply was
Maple wood chips obtained in upper New York State were shut off and the steam chamber was flooded with tap water to
provided by Mascoma Corporation (Lebanon, NH), and air- stop the reaction.
dried Kramer corn stover was provided by the National Renew-
The liquid content of each reaction tube was transferred
able Energy Laboratory (NREL, Golden, CO, Lot #33A14). The into 2 mL glass vials. These samples were centrifuged
relatively dry biomass (10–15% moisture) was knife milled to (2500 rpm for 20 min) and the supernatant was transferred
pass through a 1 mm particle size interior sieve using a labora- into glass HPLC vials for HPLC analysis by an Agilent 1200
tory mill (Model 4, Arthur H. Thomas Company, Philadelphia, system equipped with a Bio-Rad Aminex® HPX-87H column
PA). Biomass composition was determined according to the and RI detector with an eluent (5 mM sulfuric acid) flow rate
established National Renewable Energy Laboratory procedure of 0.6 mL min−1. The calculations for conversion of sugars and
(TP-510-42618, ver. 8-03-2012) in triplicates with a resulting selectivity of secondary fuel precursor products are shown
composition of 40.9 0.3 wt% glucan, 15.5 0.2% xylan, 2.1
below where Ω is the molar equivalence ratio from the starting
0.1% mannan, 24.4 0.3% K-lignin, and 17.1% other material sugar:
for maple wood and 32.7 0.4 wt% glucan, 20.7 0.2% xylan,
2.6 0.1% arabinan, 16.0 0.1% K-lignin, and 28.0% other
material for corn stover. Other materials needed for the
biomass composition to total 100% were not characterized in
this study but were expected to include minor saccharides,
ash, sugar acids, acetate, and protein.31 Because arabinan,
galactan, and mannan were not present in significant quan-
ꢀ
ꢁ
Sugar concentrationfinalðg Lꢀ1
Þ
% Conversion ¼ 1 ꢀ
ꢁ 100%
ð1Þ
Sugar concentrationinitialðg Lꢀ1
Þ
% Selectivity ¼ FP concentrationfinal ðg Lꢀ1
ꢁ Ω=% conversion
Þ
ð2Þ
This journal is © The Royal Society of Chemistry 2014
Green Chem., 2014, 16, 3819–3829 | 3821

View Article Online
Paper
Green Chemistry
1:563
% FP yield ¼ θFP
Ωfurfural
¼
ð3Þ
Concentration of xyloseinitial
FPfinalðg Lꢀ1Þ ꢁ mass of liquorfinalðgÞ
ꢁ
total biomassinitial ðgÞ ꢁ density of liquorfinalðg Lꢀ1
Þ
ꢁ100%
1:552
ΩLA
¼
ð4Þ
ð5Þ
Concentration of glucoseinitial
ð6Þ
ð7Þ
ð8Þ
ð9Þ
1:428
1:375
Ω5-HMF
¼
Concentration of glucoseinitial
θfurfural
¼
fraction of total xylaninitial
1:396
θLA
¼
fraction of total glucaninitial
1:286
2.3 THF co-solvent maple wood and corn stover biomass
reactions
θ5-HMF
¼
fraction of total glucaninitial
For the recovery of extracted lignin, the reaction liquor was
transferred to a glass bottle with a screw-on cap that was
tapped with a 0.25 in. metal hose barb fitting. The fitting
was connected by a flexible hose to a vacuum pump to perform
vacuum distillation of THF. The liquor was agitated by a mag-
netic stir bar on a stir plate as THF was boiled off at room
temperature under vacuum. Once the THF was removed from
the aqueous liquor, the extracted lignin precipitates from solu-
tion as a black resinous solid. The solid lignin residue was
then separated from the liquor and crushed to a fine powder
by a ceramic mortar and pestle. The powder was then rinsed
with water, air-dried, and then rinsed with diethyl ether.
Corn stover or maple wood solids loadings were calculated
based on the total mass of the reaction (800 g) so that each
reaction contained 5 wt% solids (40 g dry basis) and 1 wt%
acid (7.6 g by anhydrous weight) based on THF–water mixture
weight (760 g). Biomass mixtures were then allowed to pre-
soak overnight at 4 °C to insure an even distribution of acid
catalyst within the biomass pores. Contents were then left in
the laboratory for an hour for the temperature to reach about
room temperature prior to reaction.
The whole biomass slurry was then transferred to a high-
pressure continuously stirred 1 L Parr reactor (Parr Instrument
Company, Moline, IL) heated by a 4 kW fluidized sand bath
(Model SBL-2D, Techne, Princeton, NJ). Mixing was performed
by twin 6-blade impellers operating at 200 rpm by a top
mounted electric motor, and the reactor temperature was
directly measured by an in-line thermocouple (Omega, K-type).
At the conclusion of a run, the reactor was cooled by quickly
lowering it into a large room temperature water bath. All liquid
containing receptacles were made of glass to prevent the loss
of furfural and THF that was observed when plastics were
used. The solids were then separated from the reaction liquor
by vacuum filtration at room temperature through glass fiber
filter paper (Fisher Scientific, Pittsburgh, PA). Mass and
density of the liquid fraction were measured to complete accu-
rate yield calculations. Due to the difference in density
between the co-solvent mixtures and pure water, final densities
were determined by weighing 25 mL of the reacted liquid in a
volumetric flask after each reaction.
Liquid samples were analyzed by an Agilent 1200 HPLC
system equipped with a Bio-Rad Aminex® HPX-87H column
and RI detector with an eluent (5 mM sulfuric acid) flow rate
of 0.6 mL min−1. Since the HPX-87H column cannot dis-
tinguish between xylose, mannose, and galactose sugar peaks,
we also equipped our HPLC with an Aminex® HPX-87P
column to differentiate xylose from the other C6 sugars for
yield calculations. Since the HPX-87P column is incompatible
with acids, we elected not to use this column for fuel precursor
analysis. Calculation of the fuel precursor yields is given by
eqn (6) where the molar equivalent number (θ) of furfural
(eqn (7)), LA (eqn (8)), and 5-HMF (eqn (9)) are individually
calculated and divided by the fraction of the total glucan or
xylan in the raw material.
3. Results and discussion
3.1 Assessment of catalyst performance in sugar reactions
Little is currently known about the application of THF as a
water-miscible co-solvent to enhance the production of FPs
from biomass sugars. Although strong mineral acids such as
sulfuric acid have been used successfully to improve FP yields
with THF co-solvent,22 metal halide acid catalysts are promis-
ing alternatives that are less corrosive, recyclable, and more
selective to furanics than strong mineral acids.33 Table 1
shows the measured pH for the sugar co-solvent solutions con-
taining 0.1 M of each metal halide before titration to 1.6 pH. It
is known that metal oxide species form when the metal
halides are hydrolyzed by water at elevated temperatures and
the formation of OH ligands (as electron pair donors) during
hydrolysis of the metal cations increases their acidity.34 The
resulting pH of the solution is related to the first hydrolysis
Table 1 pH of metal halide catalysts in co-solvent solution containing
1 : 1 THF–watera
Catalyst
pH
CrCl3·6H2O
AlCl3·6H2O
CuCl2·2H2O
FeCl3·6H2O
ZrOCl2·8H2O
3.13 0.01
2.88 0.02
2.78 0.01
1.90 0.01
1.65 0.01
a 0.1 M catalyst loading based on each catalyst’s anhydrous mass.
3822 | Green Chem., 2014, 16, 3819–3829
This journal is © The Royal Society of Chemistry 2014

View Article Online
Green Chemistry
Paper
constant of the cationic species,34 where we found Zr and Fe sugar concentrations from real biomass reactions at 5 wt%
cations to be the strongest. Metal chlorides are also known to solids loading performed in Section 3.2. Each metal halide was
form stable adducts with THF which can influence their ioniz- added based on their anhydrous catalyst mass to a concen-
ability and catalytic activity. As Brønsted acidity typically domi- tration of 0.1 M for each reaction. The sugar solutions were
nates sugar dehydration kinetics, we elected to normalize the then loaded into Hastelloy tube reactors (10 mL working
pH of all the sugar solutions to 1.6 using sulfuric acid (close to volume) and heated to 170 °C by a custom designed stainless
that of the most acidic metal halide) prior to each reaction. steel steam chamber. The reaction proceeded until the steam
Doing so allowed us to better understand how the relative was turned off and the chamber was flooded with cold tap
Lewis strength of each catalyst influenced their selectivity to water to quench the reaction. As shown in Fig. 2A and 2B, the
secondary fuel precursors, the propensity for degradation of conversion of both xylose and glucose was significantly
final products, and the tunability of the catalysts for optimiz- improved by all metal acid catalysts beyond what was possible
ing co-production of furfural and 5-HMF from biomass.
for sulfuric acid alone in the THF co-solvent system. The rela-
In order to characterize catalyst performance with this co- tive performances of the metal halides were also very consist-
solvent system, we reacted pure glucose and xylose in 1 : 1 ent for glucose and xylose. The most active metals were Cr, Zr,
(v : v) THF–water solutions using different metal halides to and Al due to their high Lewis acid strength, achieving nearly
compare sugar conversion and selectivity toward furfural, complete conversion of xylose by 5 min and glucose by
5-HMF, and LA. Metal halides AlCl3·6H2O, CuCl2·2H2O, 10 min. Cu- and Fe-based catalysts were notably slower in
CrCl3·6H2O, FeCl3·6H2O, and ZrOCl2·8H2O were selected for sugar conversion, but still achieved near complete conversion
this comparison as they have demonstrated high selectivity after 20 min.
towards furfural and 5-HMF in previous literature
In all of the THF co-solvent sugar reactions, we observed
reports.24,26,33,35,36 The sugar co-solvent solutions contained accumulation of secondary sugar species whose retention
either 20 g L−1 glucose or 10 g L−1 xylose to simulate likely times on the HPLC matched that of fructose and xylulose.
Fig. 2 Conversions and selectivities for pure sugar reactions with metal halide acid catalysts in THF co-solvent mixture plotted against reaction
time. (A) xylose and (B) glucose conversions and (C) furfural selectivity from xylose and (D) 5-HMF and (E) LA selectivity from glucose. Reaction con-
ditions: 170 °C, 20 g L−1 glucose or 10 g L−1 xylose, 1 : 1 THF–water ratio, 0.1 M catalyst loading, and normalization of all solutions to pH 1.6 using
72% sulfuric acid. Black squares represent sulfuric acid control also titrated to pH 1.6. Error bars represent one standard deviation.
This journal is © The Royal Society of Chemistry 2014
Green Chem., 2014, 16, 3819–3829 | 3823

View Article Online
Paper
Green Chemistry
Their concentrations also tracked that of glucose and xylose The more rapid sugar conversions observed with Cr-, Zr-, and
disappearance over the course of the reaction suggesting that Al-based halides compared to the slower Fe- and Cu-based
aldose-to-ketose isomerization occurred at a faster rate than halides are important differences among these catalysts that
sugar dehydration (concentrations shown in ESI Fig. S1 and can help explain their performance in reacting actual biomass
S2†). Although it is known that metal halides catalyze an open- reported in the next section. Quantifiable parameters in the
chain reaction mechanism that produces ketose sugars with a sugar reactions such as sugar conversion, fuel precursor
lower energy barrier for dehydration,26 we also found ketose selectivity, and acidity of these metal halide catalysts will
sugars present in the sulfuric acid reactions suggesting that impact biomass conversion to achieve high combined furfural
THF co-solvent promoted a similar mechanism involving and 5-HMF yields.
sugar isomerization. This observation supports our earlier
3.2 Co-production of furfural and 5-HMF from maple wood
and corn stover
findings that THF appeared to co-catalyze C5 and C6 sugar
dehydration22 by promoting a more kinetically favorable open-
chain dehydration pathway.
The primary fractions of lignocellulosic biomass of interest for
Selectivity to the secondary FPs was then calculated based catalytic conversion to platform chemicals are cellulose, hemi-
on their concentrations after each reaction. In terms of furfural cellulose, and lignin. Xylan contained within amorphous
selectivity from xylose (Fig. 2C), sulfuric acid achieved a hemicellulose presents the most readily available source of
maximum of about 70% selectivity at 10 min that bested all sugars as it can be completely hydrolyzed at mild to moderate
other metal salt catalysts. This result suggested that although severity reaction.38 Cellulose, on the other hand, is composed
the Lewis acid character of the metal halides accelerated con- of crystalline polymeric glucose chains that are a primary
version of sugars, it also promoted competing loss reactions source of C6, but remains the most recalcitrant sugar fraction
that diminished furfural accumulation in solution attributed to acid hydrolysis and is usually treated with cellulase enzymes
to condensation reactions between intermediate sugar species after pretreatment to achieve high yields of glucose monomers
and furanic products to form humins.26 Due to rapid sugar in solution.39 Because the hemicellulose fraction of ligno-
conversion, the Cr catalyst achieved the highest furfural selecti- cellulosic biomass is far more acid-labile than crystalline cellu-
vity of about 65% earliest at 5 min, whereas FeCl3 was slower lose, furfural is produced much sooner than 5-HMF and LA.9
reacting and required 20 min to reach a maximum furfural Thus, an integrated conversion strategy to co-produce furfural
selectivity of also about 65%. ZrOCl2 had the lowest selectivity and 5-HMF directly from biomass must be tunable to mini-
to furfural despite its ability to rapidly degrade xylose, likely mize competing side reactions of the least stable products to
due to the high formation of loss products.
maximize product yields. For this reason, optimization for
In the case of 5-HMF selectivity from glucose (Fig. 2D), all high furfural yields is a primary concern as the glucan remain-
metal salts except CuCl2 achieved higher selectivities (∼40%) ing in the slower solubilizing cellulose fraction can be recov-
than sulfuric acid (∼22%) in the co-solvent system, with Al, Cr, ered as a solid product for further biological conversion into
and Fe metals performing best. Again, the Fe-based catalyst glucose or thermochemical conversion into glucose, 5-HMF,
required the longest reaction time and continually increased and/or LA if not all of it was converted into 5-HMF or LA.
5-HMF selectivity over the entire time, with it reaching 38%
In addition to manipulation of the temperature, time, and
after 20 min. Interestingly, for all catalysts except CuCl2, the acid loadings (combined reaction severity) to optimize yields
maximum selectivity for both furfural and 5-HMF occurred at from a biomass reaction, the THF co-solvent strategy allowed
approximately the same time with the best co-production of additional tuning by increasing THF concentrations in water
the furfurals demonstrated by Al, Cr, and Fe metals. Overall, to achieve greater selectivity to 5-HMF and increased biomass
however, 5-HMF selectivity was significantly lower than fur- solubilization.22 In this study, we compared the performance
fural selectivity owing to significant loss reactions to both con- of the metal halide catalysts against sulfuric acid in 1 L THF
densation products and LA.37 This was apparent by the more co-solvent reactions with 5 wt% loading of maple wood or corn
drastic drop in 5-HMF selectivity than observed with furfural stover. Table 2 lists optimized secondary FP yields achieved at
over the course of the reaction for all the metal halides the reaction conditions found to maximize production of total
except FeCl3.
furfural and 5-HMF by each catalyst. The THF–water ratio was
In the case of LA selectivity from glucose (Fig. 2E), all metal also varied from 1 : 1 to 7 : 1 (by volume) to determine the
halides demonstrated lower selectivity to LA formation than extent of improved product yields and the limit of the single
sulfuric acid, in line with the goal of this study. As LA is pro- phase regime. The catalysts were loaded on a mass basis
duced from the hydrolysis of 5-HMF in this system, CuCl2 and similar to commercial operation at a dilute 1 wt% in terms of
sulfuric acid achieved the highest LA selectivity as their selec- the total liquid mass contained within the reaction. The
tivity towards 5-HMF was lowest. By extrapolation of the heating temperature profile of the reactor shown in the ESI
increasing trend of LA selectivity over longer reaction times, we (Fig. S3†) demonstrated that total heating time to a stable
believe furfural and LA are not compatible co-products on a 170 °C reaction temperature could be achieved in about 5 min.
basis of their formation kinetics. Instead, furfural and 5-HMF
As shown in Table 2 (runs 1–9), with the exception of CrCl3
can be produced together, whereas LA production would be and ZrOCl2, metal halide catalysts demonstrated very consist-
most effectively targeted in a reaction independent of furfural. ent performance on both maple wood and corn stover, achiev-
3824 | Green Chem., 2014, 16, 3819–3829
This journal is © The Royal Society of Chemistry 2014

View Article Online
Green Chemistry
Paper
Table 2 Acid-catalyzed co-production of furfural, 5-HMF, and LA from maple wood and corn stover in batch reactions with THF co-solventa
Yields (of theoretical)f
Run #
THF–waterc
Substrateb
Acid catalystd
Time (min)
Solids remaining (%)
Furfural (%)
5-HMF (%)
LA (%)
1e
2
3
4
5
6
7
8
9
1 : 1
1 : 1
1 : 1
1 : 1
1 : 1
1 : 1
1 : 1
1 : 1
1 : 1
3 : 1
3 : 1
3 : 1
3 : 1
3 : 1
3 : 1
3 : 1
4 : 1
4 : 1
7 : 1
Maple wood
Maple wood
Maple wood
Maple wood
Maple wood
Maple wood
Corn stover
Corn stover
Corn stover
Maple wood
Maple wood
Maple wood
Maple wood
Corn stover
Corn stover
Corn stover
Maple wood
Corn stover
Maple wood
H2SO4
FeCl3
CuCl2
AlCl3
CrCl3
ZrOCl2
H2SO4
FeCl3
ZrOCl2
H2SO4
FeCl3
CuCl2
AlCl3
FeCl3
CuCl2
AlCl3
FeCl3
FeCl3
FeCl3
40
40
30
40
40
40
40
40
40
60
60
60
60
80
60
60
60
80
60
21
25
29
30
30
39
19
31
43
1
11
16
16
14
20
22
10
15
21
87
85
83
58
43
44
84
85
38
86
97
81
75
97
89
76
95
95
83
13
16
14
18
15
14
16
12
14
21
41
22
33
42
22
36
51
45
43
11
4.7
6.3
9.3
5.9
11
11
4.0
12
40
13
21
8.8
12
14
17
6
10e
11
12
13
14
15
16
17
18
19
7
3
a All reactions were performed using a 1 L Parr reactor at 170 °C reaction temperature. b 5 wt% total solids loading. c By volume ratio. d All
catalysts were loaded at 1 wt% anhydrous content. e Data from these runs are reported previously.22 f Furfural yield calculated from raw xylan
content, 5-HMF and LA yield calculated from raw glucan content.
ing maximum furfural yields close to that of sulfuric acid for a
1 : 1 THF–water mixture. 5-HMF yields were more comparable
among catalysts, but metal halides produced lower LA yields
compared to sulfuric acid owing to their increased selectivity
to 5-HMF as found for the sugar reactions. Surprisingly, the
high furfural (65%) and 5-HMF (40%) selectivities achieved by
CrCl3 from the model sugar solutions were not translated to
better furfural yields (43%) from biomass. The results of the
sugar reactions shed some light on the poor performance
CrCl3 and ZrOCl2 with biomass. In solution, the lower acidity
of CrCl3 (pH 3.13) compared to the other metal halides
(Table 1) may have limited its ability to hydrolyze sugar poly-
mers effectively, particularly without the help of sulfuric acid.
Consequently, the excellent xylose conversion performance of
CrCl3 (100% in 5 min, Fig. 2A) may have negatively impacted
furfural yields from biomass as furfural losses quickly
exceeded furfural production due to much slower release of
Fig. 3 Composition of raw maple wood and distribution of major com-
xylose from biomass. As seen by the rapid drop in furfural
selectivity for xylose reactions with CrCl3 (Fig. 2C), the poten-
tial for high furfural yields from biomass suffered from the
longer reaction times needed by the biomass reactions. We
also suspect that the strong Lewis acid character of CrCl3 was
responsible for significant furfural losses. For ZrOCl2, poor
furfural selectivity from xylose (37%) resulted in low furfural
yields (44%) despite having the highest Brønsted acidity
(pH 1.65) of the group. Conversely, FeCl3 was the best perfor-
mer due to its higher Brønsted acidity, slower xylose conver-
ponents to the solids remaining after reaction with 1 : 1 THF co-solvent
and various acid catalysts based on 100 g of initial maple wood fed to
the systems. Metal halides improved upon sulfuric acid performance by
delivering greater amounts of glucan-rich solids for enzymatic conver-
sion to glucose or thermochemical reaction to glucose, 5-HMF, and/or
LA. Reaction conditions: 5 wt% maple wood, 0.1 M acid catalyst concen-
tration, 1 : 1 THF–water, 170 °C, 30 min batch reactions. Numerical data
and calculated error is shown in Table S1.†
In order to investigate catalyst performance on solid
sion rates, and higher furfural selectivity at longer reaction biomass and compare the extent of biomass solubilization by
times. AlCl3 and CuCl2 were middle performers owing to their each catalyst, 5 wt% maple wood was reacted for 30 min in an
more moderate Brønsted character, with trade-offs between 1 : 1 THF co-solvent mixture and 0.1 M equivalent catalyst
higher 5-HMF yields or high furfural yields, respectively, con- loading in the 1 L Parr reactor. The results are shown in Fig. 3
sistent with their selectivity with the sugar reactions.
by the composition of raw maple wood solids and the distri-
This journal is © The Royal Society of Chemistry 2014
Green Chem., 2014, 16, 3819–3829 | 3825

View Article Online
Paper
Green Chemistry
bution of the major components in the solids remaining after higher solvent ratio was also beneficial to further reducing
reaction on the mass basis of 100 g of raw maple wood feed. yields of LA (6–7% at 4 : 1 ratio), the most difficult product to
As shown, the extent of biomass solubilization with metal recover due to its high boiling point (245 °C). Thus, FeCl3
halides was reduced compared to sulfuric acid. Also shown, proved to be the best metal halide for catalyzing co-production
over 90% of the maple wood K-lignin was removed during all of furfural and 5-HMF in a biomass process using THF as a
metal halide reactions, except with ZrOCl2, leaving behind a miscible co-solvent. Its strong acid strength allowed for reason-
substantial amount of glucan-rich solids that contain no hemi- ably fast deconstruction of both maple wood and corn stover,
cellulose and minute amounts of other components. In the and the close reaction time for optimal furfural and 5-HMF
THF reactions, maximizing lignin free glucan rich solids recov- production was beneficial to achieving good yields of both in
ery from the co-solvent reaction is crucial to enhance the econ- one pot reaction. As shown in Fig. 4, the 10% solids remaining
omics of this process as the cleanly fractionated solids are after 4 : 1 co-solvent reaction of maple wood were very rich in
suitable as a direct feed to produce fermentable glucose or glucan and could be recovered for efficient conversion to
used to make additional 5-HMF or LA. In the case of ZrOCl2, glucose by enzymes or further thermochemical reaction to
reduced delignification and the presence of a large non-sugar 5-HMF and/or LA. A mass balance is shown for this run in the
fraction (labeled ‘Other’ in Fig. 3) provide an interesting oppor- ESI (Fig. S4†) insuring accountability for 80% of the C6 pro-
tunity for future study to help explain why its performance for ducts and 95% of the C5 products in the soluble and insoluble
producing FPs from real biomass was significantly poorer than products after reaction.
by other catalysts. The non-sugar fraction may have resulted
We found that at a 7 : 1 THF-to-solvent ratio (Table 2, run
from accumulation of polymeric degradation products on the 19) or beyond, the behavior of the reaction suddenly shifted to
solids as the actual glucan remaining in the solids was much resemble a two-phase regime. Although we were unable to
lower than from sulfuric acid catalyzed reactions. Composition qualitatively determine this yet (such as by a sight glass in the
of this non-sugar fraction still needs to be determined to reactor), the reaction kinetics and composition of the resulting
assess the value of the byproduct.
solid material strongly suggested that the system could have
Greater biomass solubilization at the higher solvent ratio of become biphasic at such high THF ratios over the course of
3 : 1 (as shown in Table 2, runs 10–16) can support a reaction the reaction. In such a biphasic reaction, the dehydration kine-
strategy that is more focused on furfural and 5-HMF pro- tics of the aqueous phase are largely unaffected by the pres-
duction with less recoverable solids. At a 3 : 1 THF-to-water ence of the extracting solvent.40 Thus, THF would no longer be
volume ratio, the top three performing (Al, Cu, Fe) metal able to accelerate biomass solubilization as evident by the
halide catalysts greatly enhanced co-production of furfural and much higher solids fraction that remained after reaction (21%
5-HMF from maple wood and corn stover compared to sulfuric for run 19 vs. 10% for run 17 in Table 2). As shown in Fig. 4,
acid. In these reactions, FeCl3 outperformed CuCl2 and AlCl3 compositional analysis of this solid fraction also revealed that
in both furfural (97% yield for maple wood and corn stover)
and 5-HMF production (41% yield for maple wood and 42%
for corn stover) and biomass solubilization (11% solids
remaining). CuCl2 was unable to solubilize biomass as quickly
as FeCl3, and its lower 5-HMF yields from both maple wood
and corn stover reactions could be explained by its lower
5-HMF selectivity (Fig. 2D) from glucose in the sugar reactions.
For AlCl3, tuning the reaction to achieve high furfural and
5-HMF co-production was difficult as the optimal reaction
time for furfural was 20 min shorter than for 5-HMF. Thus,
higher furfural yields (81%) were achieved with CuCl2 at the
expense of greater 5-HMF losses (22% yield) at 60 min and
higher 5-HMF yields (36%) were achieved with AlCl3 at the
expense of lower furfural yields (76%) at 60 min. Overall, the
consistency in performance between corn stover and maple
wood in all reactions indicated that the THF co-solvent system
may be largely feedstock agnostic and capable of achieving
high yields in heterogeneous or mixed feedstock streams,
Fig. 4 Composition of raw maple wood and distribution of major com-
appealing to commercial feasibility.
ponents of the solids remaining after reaction with THF co-solvent with
FeCl3·6H2O at 1 : 1, 4 : 1, and 7 : 1 THF–water volume ratios. Solid mass is
based on 100 g of initial maple wood fed to the systems. Suspected
phase separation at 7 : 1 ratio is evident by larger lignin fraction,
decreased solids solubilization, and increased remaining glucan fraction
after 60 min reaction compared to the 4 : 1 ratio case. Reaction con-
ditions: 5 wt% maple wood, 1 wt% FeCl3·6H2O based on anhydrous
mass, 170 °C. Numerical data and calculated error is shown in Table S1.†
At a 4 : 1 solvent ratio (Table 2, runs 17–18), FeCl3 signifi-
cantly outperformed sulfuric acid and the other metal halides
and achieved the highest reported co-production yields of fur-
fural (95% for both maple wood and corn stover) and 5-HMF
(51% for maple wood and 45% for corn stover) from ligno-
cellulosic biomass by a one-pot single phase reaction. The
3826 | Green Chem., 2014, 16, 3819–3829
This journal is © The Royal Society of Chemistry 2014

View Article Online
Green Chemistry
Paper
a larger glucan portion remained unsolubilized and more of leading downstream catalytic upgrading operations. In the
the lignin was still intact and not extracted as would be process concept pictured in Fig. 5, raw biomass and acid cata-
expected at lower concentrations of THF. Reaction pressures lyst are loaded into a reactor along with THF co-solvent solu-
increased from 225 psig to 265 psig and was the highest tion, as depicted in Fig. 6A. Following reaction, high yields of
observed and close to the sum of the saturated vapor pressures both furfural and 5-HMF are achieved, and the reacted slurry
of water and THF. Nevertheless, high yields of both furfural is then collected and filtered to separate the solid residues. As
(83%) and 5-HMF (43%) were achieved with very little LA pro- THF is a low boiling solvent (66 °C) and forms a 95.4% azeo-
duction (3%), but yields were slightly lower than the 4 : 1 case. trope with water, it could be easily flashed off in a biorefinery
In addition, because biomass often has moisture contents of to be recovered and recycled.42 In fact, we found that room
up to 50% by weight for woods, elevated solvent ratios, such as temperature vacuum distillation was sufficient to recover THF
7 : 1 or higher would likely hurt biomass processing economics from the water phase.22
by requiring larger reactor sizing and consume more heat. The
The removal of THF also precipitates extracted biomass
impressive co-production yields achieved by the THF co- lignin as a solid (Fig. 6B) that can be recovered and rinsed
solvent system without needing high solvent concentrations with diethyl ether to produce a very pure lignin powder as
(such as >9 : 1 using GVL) is an important consideration when shown in Fig. 6C (left). This powder can in turn be re-dissolved
comparing other recently developed co-solvent systems.41
in THF or DMSO (Fig. 6C, right) and is suitable for catalytic
Fig. 5 outlines a proposed integrated THF co-solvent strat- upgrading to valuable chemicals.43 Afterwards, an appropriate
egy for application of metal halide catalysts to enhance direct organic solvent (Fig. 5, stream (1)), such as MTHF, can be used
conversion of biomass into furfural and 5-HMF followed by as an immiscible solvent to extract and concentrate furfural
two possibilities for their hydrogenation to MF and DMF, and 5-HMF into the organic layer, leaving most of the trace
respectively. The experimental work in this study was focused sugars and contaminants in the aqueous layer. Alternatively,
on producing high FP yields to be most compatible with the aqueous stream resulting from THF removal could be fed
directly to a catalytic reactor (Fig. 5, stream (2)) if desired,
depending on the catalyst system chosen for upgrading fur-
fural and 5-HMF. The aqueous stream containing the catalyst
could be recycled as FeCl3-containing aqueous streams have
been shown to remain effective over several reactions in other
systems.44 In future studies, we will develop and optimize a
high performance catalyst system to achieve high selectivity to
final fuel products, such as MF and DMF, from the product
stream of the THF co-solvent reaction.
4. Conclusions
Fig. 5 Simplified process diagram of the proposed THF co-solvent
strategy for direct conversion of lignocellulosic biomass to co-produce
Metal halides are non-corrosive and highly selective acid cata-
furfural and 5-HMF for catalytic upgrading to aromatic fuel products.
lysts suitable for co-production of furfural and 5-HMF directly
from lignocellulosic biomass without a separate pretreatment
step. We presented here coupling metal halides with THF as a
novel green co-solvent in a highly effective single phase conver-
sion strategy that achieved one of the highest reported co-
production yields of furfural and 5-HMF directly from
biomass, producing a clean product stream suitable for cata-
lytic hydrogenation to final fuel products. pH-controlled reac-
tions with pure sugar and larger 1 L scale reactions with maple
wood and corn stover demonstrated that key differences in the
catalytic nature of these metal halides affected conversion of
lignocellulosic biomass. Screening of several promising metal
halides AlCl3·6H2O, CuCl2·2H2O, CrCl3·6H2O, FeCl3·6H2O, and
ZrOCl2·8H2O on the basis of sugar conversion and selectivity
to secondary fuel precursors showed FeCl3 to perform best in
the THF co-solvent system owing to its high Brønsted acidity
and moderate sugar conversion rate. Surprisingly, CrCl3 did
not produce high furfural yields, despite having demonstrated
high sugar conversion and selectivity, due to its weaker
Furfural and 5-HMF will be extracted by an organic solvent and hydro-
genated (blue box, right) to produce aromatic fuels such as MF and DMF.
Lignin is precipitated upon recovery of THF. (1) Organic stream contain-
ing furfural and 5-HMF. (2) Aqueous stream containing metal halide
catalyst, furfural and 5-HMF.
Fig. 6 (A) 1 L THF co-solvent solution containing 5 wt% maple wood.
(B) Precipitated lignin residue after co-solvent reaction, recovery of THF,
and water removal. (C) Left, precipitated lignin powder from maple
wood after co-solvent reaction with FeCl3 catalyst. Right, same lignin
powder shown dissolved in a large droplet of dimethylsulfoxide (DMSO).
Metric ruler is shown for reference.
This journal is © The Royal Society of Chemistry 2014
Green Chem., 2014, 16, 3819–3829 | 3827

View Article Online
Paper
Green Chemistry
Brønsted acidity and high Lewis acidity that caused xylose to
be rapidly dehydrated and degraded.
8 G. W. Huber, J. N. Chheda, C. J. Barrett and J. A. Dumesic,
Science, 2005, 308, 1446–1450.
At an optimum 4 : 1 THF–water ratio and 1 wt% FeCl3, the
co-solvent reaction achieved 95% yield of furfural and 51%
9 J.-P. Lange, E. van der Heide, J. van Buijtenen and R. Price,
ChemSusChem, 2012, 5, 150–166.
yield of 5-HMF directly from maple wood and similar yields 10 Y. Roman-Leshkov, C. J. Barrett, Z. Y. Liu and
from corn stover after 60 min reaction at 170 °C. Depending J. A. Dumesic, Nature, 2007, 447, 982–985.
on the concentration of THF in the reaction, we could tune 11 J. J. Bozell, L. Moens, D. C. Elliott, Y. Wang,
biomass solubilization to preserve a glucan-rich solid residue
that is suitable for further catalytic reaction, enzymatic diges-
G. G. Neuenscwander, S. W. Fitzpatrick, R. J. Bilski and
J. L. Jarnefeld, Resour., Conserv. Recycl., 2000, 28, 227–239.
tion, or a potential pulp and paper product. During the co- 12 A. J. Shaw, K. K. Podkaminer, S. G. Desai, J. S. Bardsley,
solvent reaction, THF extracted over 90% of the lignin from
biomass that could be recovered as a fine powder. Due to its
S. R. Rogers, P. G. Thorne, D. A. Hogsett and L. R. Lynd,
Proc. Natl. Acad. Sci. U. S. A., 2008, 105, 13769–13774.
low boiling point, THF was recovered by room temperature 13 J.-P. Lange, W. D. van de Graaf and R. J. Haan, Chem-
vacuum distillation. Furfural and 5-HMF can be concentrated SusChem, 2009, 2, 437–441.
by an immiscible extracting solvent and the catalyst can be 14 R. Xing, A. V. Subrahmanyam, H. Olcay, W. Qi, G. P. van
recycled in the aqueous stream. Future study will confirm the
recyclability of the catalyst in the aqueous phase and inte-
gration with the catalytic conversion of FPs to fuel products.
Walsum, H. Pendse and G. W. Huber, Green Chem., 2010,
12, 1933–1946.
15 H. Olcay, A. V. Subrahmanyam, R. Xing, J. Lajoie, J. A. Dumesic
and G. W. Huber, Energy Environ. Sci., 2013, 6, 205–216.
16 S. Dutta, S. De, B. Saha and M. I. Alam, Catal. Sci. Technol.,
2012, 2, 2025–2036.
Acknowledgements
17 C. Barrett, J. Chheda, G. Huber and J. Dumesic, Appl.
Catal., B, 2006, 111–118.
We are grateful to the Sun Grant Initiative and UC Riverside
Research and Economic Development Office of Technology
Commercialization for primary funding of this study and to
the University of California Transportation Center (UCTC) for
additional scholarship support. We also acknowledge the
Center for Environmental Research and Technology (CE-CERT)
of the Bourns College of Engineering at the University of
California Riverside for providing facilities used in this
research and the Ford Motor Company for funding the Chair
in Environmental Engineering that facilitates projects such as
this one. We would also like to thank UCR undergraduates
Priyanka Singh and Phillip Mui for their diligence in assisting
the experimental portion of this study.
18 G. W. Huber and J. A. Dumesic, Catal. Today, 2006, 111,
119–132.
19 C. M. Cai, T. Zhang, R. Kumar and C. E. Wyman, J. Chem.
Technol. Biotechnol., 2014, 89, 2–10.
20 C. E. Wyman, Biological production of chemicals from renew-
able feedstocks, Washington, DC, 1990.
21 S. Nishimura, N. Ikeda and K. Ebitani, Catal. Today, 2014,
232, 89–98.
22 C. M. Cai, T. Zhang, R. Kumar and C. E. Wyman, Green
Chem., 2013, 15, 3140–3145.
23 T. Zhang, R. Kumar and C. E. Wyman, RSC Adv., 2013, 3,
9809–9819.
24 S. Dutta, S. De, M. I. Alam, M. M. Abu-Omar and B. Saha,
J. Catal., 2012, 288, 8–15.
25 G. Marcotullio and W. De Jong, Green Chem., 2010, 12,
1739–1746.
26 B. Danon, G. Marcotullio and W. de Jong, Green Chem.,
2014, 16, 39–54.
27 V. Choudhary, S. H. Mushrif, C. Ho, A. Anderko,
V. Nikolakis, N. S. Marinkovic, A. I. Frenkel, S. I. Sandler
and D. G. Vlachos, J. Am. Chem. Soc., 2013, 135, 3997–4006.
References
1 L. R. Lynd, J. H. Cushman, R. J. Nichols and C. E. Wyman,
Science, 1991, 251, 1318–1323.
2 C. Wyman and N. Hinman, Appl. Biochem. Biotechnol.,
1990, 24–25, 735–753.
3 L. R. Lynd, C. E. Wyman and T. U. Gerngross, Biotechnol. 28 Y. Yang, C. W. Hu and M. M. Abu-Omar, ChemSusChem,
Prog., 1999, 15, 777–793.
2012, 5, 405–410.
4 R. G. Lugar and R. J. Woolsey, Foreign Aff., 1999, 78, 88–94.
5 L. R. Lynd, J. H. Cushman, R. J. Nichols and C. E. Wyman,
Science, 1991, 251, 1318–1323.
29 S. Siankevich, Z. Fei, R. Scopelliti, G. Laurenczy,
S. Katsyuba, N. Yan and P. J. Dyson, ChemSusChem, 2014,
DOI: 10.1002/cssc.201301368.
6 J. J. Bozell and G. R. Petersen, Green Chem., 2010, 12, 30 B. Saha and M. M. Abu-Omar, Green Chem., 2014, 16, 24–38.
539–554.
31 J. B. Sluiter, R. O. Ruiz, C. J. Scarlata, A. D. Sluiter and
7 J. Q. Bond, A. A. Upadhye, H. Olcay, G. A. Tompsett, J. Jae,
D. W. Templeton, J. Agric. Food Chem., 2010, 58, 9043–9053.
R. Xing, D. M. Alonso, D. Wang, T. Zhang, R. Kumar, 32 H. L. Trajano, J. D. DeMartini, M. H. Studer and C. E. Wyman,
A. Foster, S. M. Sen, C. T. Maravelias, R. Malina, Ind. Eng. Chem. Res., 2013, 52, 4932–4938.
S. R. H. Barrett, R. Lobo, C. E. Wyman, J. A. Dumesic and 33 H. Zhao, J. E. Holladay, H. Brown and Z. C. Zhang, Science,
G. W. Huber, Energy Environ. Sci., 2014, 7, 1500–1523.
2007, 316, 1597–1600.
3828 | Green Chem., 2014, 16, 3819–3829
This journal is © The Royal Society of Chemistry 2014

View Article Online
Green Chemistry
Paper
34 M. F. Holovko, Condens. Matter Phys., 1997, 12, 13–25.
35 J. Binder, J. Blank, A. Cefali and R. Raines, ChemSusChem,
2010, 1268–1272.
36 L. Y. Mao, L. Zhang, N. B. Gao and A. M. Li, Bioresour.
Technol., 2012, 123, 324–331.
37 R. Weingarten, J. Cho, R. Xing, W. C. Conner and
G. W. Huber, ChemSusChem, 2012, 5, 1280–1290.
38 J. F. Saeman, Ind. Eng. Chem., 1945, 37, 43–52.
39 R. Kumar and C. E. Wyman, in Aqueous Pretreatment of Plant
40 R. Weingarten, J. Cho, W. Conner and G. Huber, Green
Chem., 2010, 1423–1429.
41 D. M. Alonso, S. G. Wettstein, M. A. Mellmer, E. I. Gurbuz
and J. A. Dumesic, Energy Environ. Sci., 2013, 6, 76–80.
42 I. M. Smallwood, Solvent Recovery Handbook, Blackwell
Science Ltd., Oxford, UK, 2002.
43 G. Henriksson, J. Li, L. Zhang and M. E. Lindström, in Ther-
mochemical Conversion of Biomass to Liquid Fuels and Chemi-
cals, The Royal Society of Chemistry, 2010, ch. 9, pp. 222–262.
Biomass for Biological and Chemical Conversion to Fuels and 44 L. Mao, L. Zhang, N. Gao and A. Li, Green Chem., 2013, 15,
Chemicals, John Wiley & Sons, Ltd, 2013, pp. 281–310.
727–737.
This journal is © The Royal Society of Chemistry 2014
Green Chem., 2014, 16, 3819–3829 | 3829