Ubiquitous aluminum alkyls and alkoxides as effective catalysts for glucose to HMF conversion in ionic liquids

Article abstract of DOI:10.1016/j.apcata.2012.05.035

Metal halides (chlorides in particular) are employed almost exclusively as Lewis acid catalysts for the homogeneous conversion of glucose (or cellulose) to HMF (5-hydroxymethylfurfural) in ionic liquids (ILs), with CrCl₂ being arguably the most effective benchmark catalyst. Reported herein is a discovery that ubiquitous aluminum alkyl or alkoxy compounds are very effective Lewis acid catalysts for the glucose-to-HMF conversion in ILs. Under the current reaction conditions (1-ethyl-3-methylimidazolium chloride [EMIM]Cl, 120 °C, 6 h), simple trialkyl and trialkoxy aluminum species such as AlEt₃ and Al(OⁱPr)₃, which are much cheaper than CrCl ₂ (by a factor of 5 for AlEt₃ or 180 for Al(O ⁱPr)₃), are at least as effective as CrCl₂ to catalyze this conversion process. The molecular structure of [EMIM] ⁺[ClAlMe(BHT)₂]^-, formed upon mixing the alkylaryloxy aluminum MeAl(BHT)₂ and the IL [EMIM]Cl, has been determined by X-ray diffraction; the structure simulates that of the metallate [EMIM]⁺[CrCl₃]^-, the proposed active species responsible for the effective glucose to HMF conversion by CrCl₂ in [EMIM]Cl. Another significant finding is that a gradual substitution of the chloride ligand on aluminum by the alkyl ligand brings about a drastic enhancement on the HMF yield, from 1.6% by AlCl₃ to 7.6% by MeAlCl₂ to 17% by Et₂AlCl and to 51% by AlEt₃, thus showing approximately an overall 32-fold HMF yield enhancement going from AlCl₃ to AlEt₃.

Full text of DOI:10.1016/j.apcata.2012.05.035

Applied Catalysis A: General 435–436 (2012) 78–85
Contents lists available at SciVerse ScienceDirect
Applied Catalysis A: General
journal homepage: www.elsevier.com/locate/apcata
Ubiquitous aluminum alkyls and alkoxides as effective catalysts for glucose
to HMF conversion in ionic liquids
Dajiang (DJ) Liua,b, Eugene Y.-X. Chena,∗
a
Department of Chemistry, Colorado State University, Fort Collins, CO 80523, USA
Department of Mechanical Engineering, Colorado State University, Fort Collins, CO 80523, USA
b
a r t i c l e i n f o
a b s t r a c t
Article history:
Metal halides (chlorides in particular) are employed almost exclusively as Lewis acid catalysts for the
homogeneous conversion of glucose (or cellulose) to HMF (5-hydroxymethylfurfural) in ionic liquids
Received 5 March 2012
Received in revised form 21 May 2012
Accepted 25 May 2012
(ILs), with CrCl2 being arguably the most effective benchmark catalyst. Reported herein is a discovery that
ubiquitous aluminum alkyl or alkoxy compounds are very effective Lewis acid catalysts for the glucose-
to-HMF conversion in ILs. Under the current reaction conditions (1-ethyl-3-methylimidazolium chloride
Available online 4 June 2012
◦
i
[
EMIM]Cl, 120 C, 6 h), simple trialkyl and trialkoxy aluminum species such as AlEt3 and Al(O Pr)3, which
Keywords:
Biomass
Glucose
Ionic liquids
HMF
Aluminum catalysts
i
are much cheaper than CrCl2 (by a factor of 5 for AlEt3 or 180 for Al(O Pr)3), are at least as effective as
CrCl2 to catalyze this conversion process. The molecular structure of [EMIM] [ClAlMe(BHT)2] , formed
upon mixing the alkylaryloxy aluminum MeAl(BHT)2 and the IL [EMIM]Cl, has been determined by X-ray
diffraction; the structure simulates that of the metallate [EMIM] [CrCl3] , the proposed active species
responsible for the effective glucose to HMF conversion by CrCl2 in [EMIM]Cl. Another significant finding
is that a gradual substitution of the chloride ligand on aluminum by the alkyl ligand brings about a drastic
enhancement on the HMF yield, from 1.6% by AlCl3 to 7.6% by MeAlCl2 to 17% by Et2AlCl and to 51% by
AlEt3, thus showing approximately an overall 32-fold HMF yield enhancement going from AlCl3 to AlEt3.
+
−
+
−
©
2012 Elsevier B.V. All rights reserved.
1
. Introduction
These advances were made possible by the discovery of Rogers and
co-workers [7] that showed a class of water-stable and -miscible
ILs, 1-alkyl (R)-3-methylimidazolium chloride salts [8], [RMIM]Cl,
can solubilize cellulose in appreciable wt.% by disrupting the exten-
sive H-bonding network present in cellulose through H-bonding of
the anion of ILs with the hydroxyl groups of cellulose [9]. Excitingly,
IL solvents enabled homogenous hydrolysis of cellulose to water-
soluble reducing sugars in high to quantitative conversion, either
catalyzed by mineral or organic acids [10], or even in the absence
Research directed at developing effective conversion of non-
food plant biomass into fuels and/or chemicals has intensified in
recent years [1], as this process, once becoming technologically
and economically competitive as compared to oil refinery, can
provide humanity with a sustainable source of fuels and chemi-
cals. The majority (60–90 wt.%) of plant biomass is the biopolymer
carbohydrates (sugars) stored in the form of cellulose and hemi-
celluloses. The biomass-derived sugars can be converted into fuels
and value-added chemicals by liquid-phase catalytic processing [2].
Alternatively, cellulosic materials can be directly converted into
the biomass platform chemical 5-hydroxymethylfurfural (HMF)
of any additional catalyst (i.e., with IL–H O mixtures) [11].
2
Through acid-catalyzed dehydration, fructose can be readily
converted to HMF typically in high yields [12]. However, glu-
cose, a more desirable feedstock derived from non-food, cellulosic
biomass, has been showed to be resistant to its conversion into
HMF, thus achieving typically low yields (∼10%) by a variety of cat-
[
3], a versatile intermediate for top-value-added chemicals and
fuels (e.g., 2,5-dimethylfuran, a biofuel with a 40% higher energy
density than ethanol [4]). As environmentally benign alternatives
to volatile organic solvents, recyclable ionic liquids (ILs) have
attracted rapidly growing interest [5], particularly in the pursuit of
renewable energy and chemicals from lignocellulosic biomass [6].
3
+
3+
alyst systems, such as lanthanide halides LnCl3 (Ln = La –Lu ) in
water or organic solvents [13]; the use of AlCl3 in water or organic
solvents assisted by microwave radiation improves the HMF yield
[14]. Seminar work of Zhang et al. revealed that glucose can also
be converted into HMF in good yields when using CrCl2 as cata-
lyst in ILs such as [EMIM]Cl [15]. Thus, the CrCl -catalyzed process
2
◦
in [EMIM]Cl at 100 C for 3 h achieved a HMF yield of 68–70%; the
∗ Corresponding author at: Department of Chemistry, Colorado State University,
Fort Collins, CO 80523 1872, USA. Tel.: +1 970 491 5609; fax: +1 970 491 1801.
E-mail address: eugene.chen@colostate.edu (E.Y.-X. Chen).
process was proposed to proceed via in situ glucose-to-fructose
isomerization catalyzed by the anion CrCl3 in the resulting
−
0
926-860X/$ – see front matter © 2012 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.apcata.2012.05.035

D. Liu, E.Y.-X. Chen / Applied Catalysis A: General 435–436 (2012) 78–85
79
+
−
metallate [EMIM] CrCl3 formed upon mixing CrCl and [EMIM]Cl,
in ILs, with CrCl being one of the most effective catalysts. We report
2
2
followed by dehydration of fructose to HMF [15]. A subsequent
herein that ubiquitous aluminum alkyl or alkoxy compounds,
which are widely used in the olefin polymerization industry and
also in general catalysis [35], are effective catalysts for the glucose-
to-HMF conversion in ILs. Significantly, aluminum alkyls (e.g.,
study reported a lower HMF yield of 62% under the same condi-
◦
tions ([EMIM]Cl, 6 mol% CrCl , 100 C, 3 h), but this study provided
2
both experimental and theoretical evidence to support the pro-
posed reaction sequence which involves initial isomerization of
glucose to fructose followed by subsequent dehydration of fruc-
tose to HMF [16]. Interestingly, the HMF yields for catalyst systems
that did not contain CrClx, including a large number of metal (main-
group, transition-metal, and rare-earth) halides, were only 10% or
less [15]. Additionally, replacing the CrClx catalyst with a BrØnsted
i
AlEt ) and alkoxides (e.g., Al(O Pr) ) are far more effective catalysts
3
3
than aluminum halides (by 32-fold than AlCl ), and exhibit compa-
3
rable or higher HMF yields with the benchmark catalyst CrCl under
2
i
the same conditions, but AlEt and Al(O Pr) are ∼5 and ∼180 times
3
3
cheaper than CrCl , respectively.
2
acid, either H SO [17] or acidic IL [EMIM][HSO ] [18], failed to con-
2
4
4
vert glucose to HMF, although such a system is still very effective in
converting fructose to HMF, the results of which seemed to point
out that aldohexoses such as glucose do not have a direct dehydra-
tion pathway to HMF and thus require an isomerization catalyst
2. Experimental
2.1. Materials, reagents, and methods
such as CrCl . On the other hand, Chidambaram and Bell recently
All syntheses and manipulations of air- and moisture-sensitive
materials were carried out in flamed Schlenk-type glassware on a
dual-manifold Schlenk line or in an argon or nitrogen-filled glove-
box. HPLC-grade organic solvents were sparged extensively with
nitrogen during filling of the solvent reservoir and then dried by
passage through activated alumina (for Et O, THF, and CH Cl )
followed by passage through Q-5-supported copper catalyst (for
toluene and hexanes) stainless steel columns. HPLC-grade DMF was
degassed, dried over CaH2 overnight, followed by vacuum transfer
2
showed that a combination of a heteropolyacid, H PMo12O40, with
3
[
EMIM]Cl and co-solvent acetonitrile converted glucose to HMF in
high to quantitative conversion [19].
Since the important discovery of the effective CrCl /IL sys-
2
tem [15], many other metal halides were reported to catalyze the
glucose-to-HMF conversion in the IL media. Ying and co-workers
reported that addition of the in situ generated N-heterocyclic car-
2
2
2
bene (NHC) ligand to the CrCl /IL system enhanced the conversion
2
◦
with a HMF yield up to 81% ([BMIM]Cl, 9 mol% NHC/CrCl , 100 C,
(not by distillation). NMR solvents CDCl3 and CD Cl2 were dried
3
2
6
h) and concluded that the NHC/CrClx complex is the catalyst
over activated Davison 4 A˚ molecular sieves, and NMR spectra were
recorded on a Varian Inova 300 (FT 300 MHz, 1H; 75 MHz, C)
13
responsible for the conversion [20]. Han and co-workers found
that SnCl4 to be also effective for the conversion of glucose into
or a Varian Inova 400 MHz spectrometer. Chemical shifts for 1
H
HMF in [EMIM]BF , achieving up to 62% yield [21], while Zhao
spectra were referenced to internal solvent resonances and are
reported as parts per million relative to tetramethylsilane. The
HMF-containing products were analyzed by Agilent 1260 Infinity
HPLC system equipped with an Agilent Eclipse Plus C18 Column
4
et al. revealed that GeCl4 in ILs converted glucose to HMF in 38%
or 48% (with addition of molecular sieves) yield [22]. Lanthanide
salts YbCl3 and Yb(OTf)3 afforded 12% and 24% HMF yields from
glucose conversion in [BMIM]Cl, which were double of those yields
achieved in [EMIM]Cl [23]. Binder and Raines reported that cellu-
lose, in a solvent mixture containing N,N-dimethylacetamide, LiCl
◦
(100 mm × 4.6 mm; 80/20 water/methanol, 0.6 mL/min, 30 C) and
a UV detector (284 nm). Sugar contents of the products were mea-
sured by Agilent 1260 Infinity HPLC system equipped with a Biorad
Aminex HPX-87H Column (300 mm × 7.8 mm; water, 0.6 mL/min,
and [EMIM]Cl, can also be directly converted into HMF in 54% yield
◦
◦
◦
at 140 C for 2 h, using CrCl (25 mol%) and HCl (6 mol%) as catalysts
45 C) and an ELSD (65 C, 3.5 bar, gain 6); under such conditions
possible sugars (e.g., glucose and fructose) in the reaction mixture
can be well separated and quantified (Fig. 1a).
2
[
24]. We showed that the reducing sugar mixture resulted from the
cellulose hydrolysis with the IL–H O mixture can be effectively con-
2
verted into HMF in the presence of CrCl ; this conversion to HMF
d-Glucose (Granular powder, Fisher Chemical), CrCl2 (Alfa
2
t
can also be carried out in a one-pot fashion starting directly from
cellulose [11]. Even without adding additional catalysts or reagents
besides IL [EMIM]Cl and water, Wu and co-workers showed that,
under optimized conditions, cellulose can be converted to HMF
Aesar), and aluminum species (AlCl , MeAlCl , Et AlCl, Al(O Bu) ,
3
2
2
3
i
i
Al(O Pr) , AlMe , AlEt , and Al Bu , Strem Chemicals) were used
3
3
3
3
as received. Ionic liquids (Fluka) were further purified as follows:
1-ethyl-3-methylimidazolium acetate [EMIM]OAc, dried under
◦
in up to 21% yield [25]. Zhang et al. utilized the CrCl /CuCl2 pair
vacuum at 100 C for 24 h; 1-ethyl-3-methylimidazolium chloride
2
for direct conversion of cellulose to HMF in [EMIM]Cl achieving
[EMIM]Cl and 1-butyl-3-methylimidazolium chloride [BMIM]Cl),
dried under vacuum at 100 C for 24 h, followed by repeated recrys-
◦
∼
55% yield [26], while a somewhat higher HMF yield (∼60%) was
reported later by Cho et al. using the CrCl /RuCl pair [27]. Zhao
tallization from CH Cl2 and hexanes at room temperature; and
2
3
2
et al. reported high HMF yields of 91% and 61% by CrCl in [BMIM]Cl
1-H-3-methylimidazolium chloride [HMIM]Cl, purified by subli-
mation at 120 C/500 mTorr for 12 h. The purified ionic liquids were
stored in an argon-filled glovebox.
Solution alkyl aluminoxanes (MAO in toluene, Sigma–Aldrich;
MMAO in heptanes, Akzo Nobel) were used as received, while the
corresponding solid alkyl aluminoxanes were prepared by remov-
ing all volatiles of the solutions under vacuum. Butylated hydroxy-
toluene (BHT-H, 2,6-di-tert-butyl-4-methylphenol, Sigma–Aldrich)
was recrystallized from hexanes prior to use, and MeAl(BHT)2
was prepared by the reaction of AlMe3 with BHT–H according lit-
3
◦
under microwave irradiation from glucose and cellulose, respec-
tively [28], but a subsequent study by Qi et al. reported a HMF yield
of 67% from glucose by CrCl3 in [BMIM]Cl also under microwave
irradiation [29]. A two-step process involving gradual addition of
water in the acid-catalyzed hydrolysis of cellulose step in [EMIM]Cl,
which was shown to achieve a high glucose yield of nearly 90% by
Binder and Raines [30], followed by addition of the CrCl catalyst in
3
the subsequent step afforded a higher HMF yield of 73% based on
cellulose [31]. A lower cellulose-to-HMF yield of 34% was observed
using FeCl as catalyst for the cellulose conversion in a protic IL [32].
erature procedures [36]. Tris(pentafluorophenyl)borane, B(C F5)3
2
6
Other metal halides or a combination them, such as CrCl /LaCl [33]
was obtained as research gifts from Boulder Scientific Co and
further purified by recrystallization from hexanes at –30 C;
3
3
◦
and MnCl [34], have also been employed for the cellulose-to-HMF
2
conversion in ILs.
tris(pentafluorophenyl)alane, Al(C F5) , as a 0.5 toluene adduct
6
3
It can be readily seen from the above overview, metal halides
chloride salts in particular) are employed nearly exclusively as
Al(C F5) ·(C7H )
(for vacuum-dried samples), was prepared by
the reaction of B(C F5) and AlMe3 in a 1:3 toluene/hexanes
6 3
6
3
8
0.5
(
Lewis acid catalysts for the glucose (or cellulose)-to-HMF conversion
solvent mixture in quantitative yield (extra caution should be

8
0
D. Liu, E.Y.-X. Chen / Applied Catalysis A: General 435–436 (2012) 78–85
supernatant was passed through the cation and anion exchange
columns to discharge the ionic liquid. A total of 5 mL eluent was
collected for HPLC (ELSD detector) analysis (vide supra). Glucose
recovery after passing through the ions-exchange column was
also performed, which showed 96.2% of recovery. For both AlEt3
and MeAl(BHT)2 catalyzed conversion systems analyzed, no sug-
ars (glucose and fructose) were detected, thus showing glucose
was quantitatively converted into HMF (major product) and other
side products. Analysis of other minor products by HPLC [Agilent
1
8
260 Infinity HPLC system equipped with a Biorad Aminex HPX-
◦
7H Column (300 mm × 7.8 mm; 0.005 M H SO , 0.6 mL/min, 45 C)
2
4
and a UV detector (268 nm)] showed no levulinic acid but contained
other common HMF degradation products such as formic and acetic
acids.
2
[
.3. Formation and X-ray crystallographic analysis of
+
−
EMIM] [ClAlMe(BHT) ]
2
[
EMIM]Cl (15.0 mg, 0.102 mmol) and MeAl(BHT)2 (50.0 mg,
0
.104 mmol) were mixed in the argon-filled glovebox and heated
◦
at 120 C for 6 h. After being cooled to room temperature, the solid
product was dissolved in warm chloroform (40 C). The resulting
clear solution was stored in a freezer inside the glovebox at–30 C
◦
◦
for 3 d, affording colorless single crystals suitable for X-ray diffrac-
1
tion analysis. H NMR of the product is consistent with the structure
1
of the title imidazolium aluminate compound. H NMR (CD Cl ,
2
2
◦
+
−
2
3 C) for [EMIM] [ClAlMe(BHT) ] : ı 8.30 (s,1H, NCHN), 7.15, 7.11
2
(
d, 2H, N(CH) N), 6.94 (s, 4H, Ar), 4.08 (q, 2H, CH CH ), 3.75 (s,
2
2
3
3
H, N–Me), 2.20 (s, 6H, Ar–Me), 1.49 (t, 3H, CH CH ), 1.44 (s, 36H,
2 3
Ar–CMe ),–0.55 (s, 3H, Al–Me). The molecular structure of this com-
3
pound was confirmed by X-ray diffraction analysis.
Single crystals of this complex were quickly covered with a
◦
−6
layer of Paratone-N oil (Exxon, dried and degassed at 120 C/10
Torr for 24 h) after decanting the mother liquor. A crystal was
then mounted on a thin glass fiber and transferred into the cold
nitrogen stream of a Bruker SMART CCD diffractometer. The struc-
ture was solved by direct methods and refined using the Bruker
SHELXTL program library [38]. The structure was refined by full-
Fig. 1. (a) Example of sugar analysis of a mixture containing glucose and fructose
by HPLC with an ELSD detector and (b) example of HMF analysis of the product,
◦
derived from Et3Al-catalyzed glucose conversion to HMF at 120 C for 6 h, by HPLC
2
matrix least-squares on F for all reflections. All atoms were located
with a UV detector (284 nm).
by differenceFourier synthesisand refined anisotropically, whereas
hydrogen atoms were included in the structure factor calculations
at idealized positions, except for the C(2)–H on the imidazolium
ring which was located by the difference Fourier synthesis and
refined. There is a distorted CHCl3 molecule (crystallization sol-
vent) in the lattice. CCDC-845942 contains the supplementary
crystallographic data for this paper. These data can be obtained
free of charge from The Cambridge Crystallographic Data Centre
via www.ccdc.cam.ac.uk/data request/cif.
exercised when handling this material, especially the unsolvated
form, because of its thermal and shock sensitivity) [37].
2.2. Conversion of glucose to HMF
In a typical experiment, glucose (40.0 mg, 0.220 mmol) was
premixed with [EMIM]Cl (200 mg, 1.36 mmol) in a 5 mL vial in
the argon-filled glove box, followed by further loading of an
aluminum catalyst (0.022 mmol, 10 mol% relative to glucose). Co-
solvent (1 mL), when applicable, was added into the mixture. The
+
−
X-ray crystal structural data for [EMIM] [ClAlMe(BHT) ] ·
2
CHCl : C
H
AlCl N O , Mr = 746.67, T = 120(2) K, ꢀ = 0.71073 A˚ ,
3
38 61 4 2 2
crystal dimensions 0.26 mm × 0.19 mm × 0.086 mm, mono-
clinic, C2/c, a = 38.6843 (10) Å, b = 10.8212(3) A˚ , c = 20.4111(5) A˚ ,
sealed vial was placed in a temperature-controlled orbit shaker
◦
3
3
ˇ = 102.708(3) , V = 8335.0(4) Å , Z = 8, ꢁ
= 1.190 Mg/m , ꢂ
calcd
◦
(
120 C, 300 RPM) and heated at this temperature for 6 h. The reac-
◦
range for data collection = 2.05–24.72 , 64684 reflections collected,
tion was quenched with ice-water and then diluted with a known
amount of deionized water. HMF was quantified with calibra-
tion curves generated from the commercially available standard
in water. A typical HPLC chromatogram of the reaction product for
HMF analysis is shown in Fig. 1b.
7
097 unique (R = 0.0494), zero restraints, 99.9% complete-
int
◦
2
ness to ꢂ = 24.72 , goodness-of-fit on F = 1.013, final R = 0.0617
1
and wR = 0.1591 with I > 2␴(I), and residual electron density
2
−
3
extremes = 0.711 and−0.685 eA˚
.
Experiments were also performed to analyze possible residue
sugars such as glucose and fructose. [EMIM]Cl (1.0 g) and glucose
3. Results and discussion
(
0.2 g) were charged into a 5 mL vial in a glovebox, followed by fur-
ther loading of AlEt3 or MeAl(BHT)2 (10 mol% relative to glucose).
The mixture was heated at 120 C for 6 h, under which condition
3.1. Comparative studies of aluminum-based catalysts
◦
the highest HMF yield was achieved. The resulting mixture was
diluted to 25 mL after quenching by ice-water, and 0.5 mL of the
A total of 12 different aluminum species were investigated for
their effectiveness as catalyst for the glucose-to-HMF conversion

D. Liu, E.Y.-X. Chen / Applied Catalysis A: General 435–436 (2012) 78–85
81
Fig. 2. Plot of the HMF yield vs. aluminum species under identical conditions:
◦
EMIM]Cl, 10 mol% Al (relative to glucose), 120 C, 300 RPM (shaker), and 6 h.
[
under identical reaction conditions: [EMIM]Cl (purified, see Sec-
◦
tion 2), 10 mol% Al relative to glucose, 120 C, and 6 h, which are
optimal conditions for these aluminum-based catalysts (vide infra).
These aluminum species can be grouped into 5 different classes:
(
a) aluminum halides or alkyl halides (AlCl , MeAlCl , Et AlCl); (b)
3 2 2
alkyl aluminoxanes (MAO and MMAO, both in solution and solid
state); (c) triaryl aluminum (Al(C F5) ), (d) trialkoxy or alkylary-
6
3
i
t
loxy aluminum (Al(O Pr) , Al(O Bu) , MeAl(BHT) ), and (e) trialkyl
aluminum (AlMe , AlEt , and Al Bu ). It is clear from the results
3
3
2
i
3
3
3
summarized in Fig. 2 that AlCl3 is least effective, with a low HMF
yield of only 1.6%, while AlEt3 is most effective, achieving a much
higher HMF yield of 51%. Analysis of sugars in the reaction prod-
ucts produced by the two representative catalysts, MeAl(BHT) and
2
AlEt , showed absence of glucose and fructose, indicating quanti-
3
Fig. 4. Plot of HMF yield as a function of reaction time (a) and catalyst loading (b)
6 h for the glucose-to-HMF conversion by MMAO (10 mol%) in [EMIM]Cl at 120 C.
◦
tative conversion of glucose to HMF (the major product) and other
side products resulted from degradation of HMF (formic and acetic
acids, see Experimental). In our hand, the benchmark catalyst CrCl2
under conditions identical to those employed for the current alu-
◦
minum catalysts (i.e., [EMIM]Cl, 10 mol% catalyst, 120 C) gave a
HMF yield in the range of 45–50%, depending on reaction time.
Fig. 5. Plot of HMF yield as a function of ionic liquid structure for the glucose-to-HMF
◦
conversion by MMAO (10 mol%) at 120 C for 6 h.
Fig. 3. Plot of HMF yield as a function of temperature for the glucose-to-HMF con-
version by MMAO (10 mol%) in [EMIM]Cl for 6 h.

8
2
D. Liu, E.Y.-X. Chen / Applied Catalysis A: General 435–436 (2012) 78–85
Fig. 6. Plot of HMF yield as a function of reaction time (a) and catalyst loading (b), 6 h
for the glucose-to-HMF conversion by MeAl(BHT)2 (10 mol%) in [EMIM]Cl at 120 C.
Fig. 7. Plot of HMF yield as a function of reaction time (a) and catalyst loading (b,
◦
◦
6
h) for the glucose-to-HMF conversion by AlEt3 (10 mol%) in [EMIM]Cl at 120 C.
Noteworthy are several interesting trends. First, a gradual sub-
stitution of the chloride ligand on Al by the alkyl ligand brought
(solution), MeAl(BHT) , and AlEt , in more detail through varia-
2 3
about a drastic enhancement on the HMF yield: AlCl , 1.6%;
tions in reaction temperature and time, catalyst loading, and IL
structure. These studies were attempted in order to optimize the
catalyst performances.
3
MeAlCl , 7.6%; Et AlCl, 17%; and AlEt , 51%. Second, alkyl alumi-
2
2
3
noxanes, having a typical structural formula of [ (R)Al
O ]n [35],
are good catalysts for the glucose-to-HMF conversion, with MMAO
In the case of MMAO (10 mol%), the glucose-to-HMF conversion
was first carried out in [EMIM]Cl for 6 h at four different temper-
i
derived from mixed trialkyl aluminum (AlMe3 and Al Bu ) [35]
3
◦
◦
◦
◦
being considerably more effective than the AlMe -derived MAO.
atures of 80 C, 100 C, 120 C, and 130 C, achieving HMF yields of
2.0%, 13%, 45%, and 44%, respectively (Fig. 3). Next, the temperature
3
For both MAO and MMAO, the catalyst state (solution vs solid) gave
only modest variations in the HMF yield: 30% (solution) and 27%
◦
was fixed at 120 C, but the reaction time was varied from 1 to 24 h
(
solid) for MAO; 45% (solution) and 46% (solid) for MMAO. Consis-
(Fig. 4a). This time profile clearly showed the HMF yield increased
initially as time increased, from 24% (1 h) to 36% (3 h) to 45% (6 h),
but a further increase in time caused the HMF yield to drop to 34%
(12 h) and only 9.8% (24 h), as a result of side reactions associated
with the gradual HMF degradation at high temperature with time
i
tent with this finding, Al Bu3 was noticeably more effective (48%)
than AlMe3 (39%). Third, Lewis acidity of these aluminum catalysts
is not an important factor for achieving highly HMF yield. For exam-
ple, highly Lewis acidic Al(C F5)3 and AlCl3 achieved only modest
3
with lower Lewis acidity, such as Al(O Pr) , Al(O Bu) , MeAl(BHT) ,
and AlEt , afforded much higher HMF yields of 49%, 49%, 50%,
and 51%, respectively. It is worth noting that the freshly vacuum-
distilled Al(O Pr)3 performed identically to that as received.
6
◦
1% and low 1.6% yields, respectively, whereas aluminum catalysts
[23]. Using the conditions with the optimized temperature (120 C)
i
t
3
3
2
and time (6 h), we varied the MMAO catalyst loading from 0 mol%
to 30 mol% (Fig. 4b). Under these conditions, the control run (no
catalyst) produced HMF in 5.5% yield; with a catalyst loading of 1,
2, 5, 10, and 20 mol%, the HMF yield steadily increased to 14, 28,
3
i
3
9, 45, and 48%, respectively. A further increase in catalyst loading
3
.2. Effects of temperature, time, catalyst loading, IL structure,
to 30 mol% actually decreased the HMF yield to 37%, presumably
due to the Lewis acid-assisted degradation of the product. Lastly,
four different types of ILs were investigated for their relative per-
formances in the glucose-to-HMF conversion catalyzed by MMAO
and co-solvent
We selectively describe herein our investigations into the
three representative, more effective systems, including MMAO
◦
under fixed conditions: 10 mol% catalyst, 120 C, and 6 h. The results

D. Liu, E.Y.-X. Chen / Applied Catalysis A: General 435–436 (2012) 78–85
83
Fig. 8. Overlay plot of 1H NMR (CD2Cl2, 23 C) spectra of [EMIM] [ClAlMe(BHT)2] (top), MeAl(BHT)2 (middle), and [EMIM]Cl (bottom).
◦
+
−
of this study, summarized in Fig. 5, clearly showed the best per-
forming IL is [EMIM]Cl (46%), while the worst performing IL is the
basic IL with the acetate anion, [EMIM]OAc (0%), which is known to
rapidly degrade HMF even at 100 C [23]. Amongst the ILs with the
same chloride anion, the one with the smallest N-substituents (H,
The alkylaryloxy aluminum catalyst, MeAl(BHT) , was also
2
investigated for its performance in the glucose-to-HMF conversion,
◦
carried out in [EMIM]Cl at 120 C for 6 h, as a function of time and
◦
catalyst loading (Fig. 6a). It is clear from Fig. 6 that the reaction
reached a peak yield of 50% between 3 h and 6 h, but a longer reac-
tion time of 12 h lowered the yield to 46%. On catalyst loading, the
reactions with 5 and 10 mol% catalyst loadings gave a same HMF
yield of 50%, while the yield was reduced to 42% or 21% when cata-
lyst loading was either increased to 20 mol% or lowered to 2.5 mol%
(Fig. 6b).
Me), [HMIM]Cl, which is also a protic IL, gave the lowest HMF yield
(
26%); between the two 1-alkyl-3-methylimidazolium chloride ILs,
n
the Bu substituted IL, [BMIM]Cl, gave a lower HMF yield (38%)
than the Et substituted IL, [EMIM]Cl (45%). Overall, the optimized
conditions for the glucose-to-HMF conversion catalyzed by MMAO
◦
are in [EMIM]Cl at 120 C for 6 h, with a 10 mol% catalyst loading.
Lastly, the most effective catalyst of this series, AlEt , was
3
Although the reaction with 20 mol% catalyst gave somewhat higher
yield (by 3%), this condition is not desirable, considering the cost
associated with a doubling of the catalyst loading.
further investigated for its performance in the glucose-to-HMF
conversion at 120 C with different reaction time, catalyst loading,
and IL structure. The reaction time profile, summarized in Fig. 7a,
◦
+
−
,
Fig. 9. X-ray single crystal structure of [EMIM] [ClAlMe(BHT)2]
[
1
with thermal ellipsoids drawn at the 50% probability. Selected bond lengths [Å] and angels
◦
]: Al(1) O(1) = 1.752(2), Al(1) O(2) = 1.757(2), Al(1) C(1) = 2.019(3), Al(1) Cl(1) = 2.203(1), C(2) H(2) = 0.93(4); C(1) Al(1) Cl(1) = 111.81(10), C(1) Al(1) O(1) =
16.97(11),C(1) Al(1) O(2) = 106.24(11), Cl(1) Al(1) O(1) = 100.78(8), Cl(1) Al(1) O(2) = 110.57(8), O(1) Al(1) O(2) = 110.46(10).

8
4
D. Liu, E.Y.-X. Chen / Applied Catalysis A: General 435–436 (2012) 78–85
showed that the peak yield of 51% was achieved at 6 h and that
shorter time (3 h) or longer time (12 h) gave lower yields of 45%
and 46%, respectively. Similarly to MeAl(BHT) , the reactions
2
with 5 and 10 mol% AlEt3 gave a same HMF yield of 51%, while
the yield was reduced to 30% or 37% when catalyst loading was
either increased to 20 mol% or lowered to 2.5 mol% (Fig. 7b).
Investigations into IL structure effects revealed that [BMIM]Cl was
less effective (47%) than [EMIM]Cl (51%), under otherwise identical
◦
conditions (10 mol% catalyst, 120 C, 6 h). Potential organic co-
solvent effects were also examined; addition of 1 mL toluene to the
◦
reaction carried out in [EMIM]Cl (10 mol% catalyst, 120 C, 6 h) did
not significantly alter the HMF yield (47%), while the polar donor
solvent DMF substantially reduced the HMF yield to 36%. A similar
co-solvent effect was also observed for other trialkyl aluminum
i
catalysts such as Al Bu .
3
3.3. Structure of aluminum catalyst in IL
The active species responsible for the effective conversion of
glucose to HMF by CrCl2 in [EMIM]Cl has been proposed to be
+
−
the metallate [EMIM] CrCl3 , formed upon mixing the CrCl2 cat-
alyst with the IL [15]. To examine if an analogous metallate salt
is formed upon mixing the aluminum catalyst with the IL or
not, we carried out the reaction of the aluminum alkyl catalyst
MeAl(BHT) with [EMIM]Cl under the glucose conversion condi-
2
◦
1
tions (120 C, 6 h, see Section 2). The H NMR spectrum (Fig. 8) of
the resulting product is indicative of imidazolium aluminate ion
pair [EMIM] [ClAlMe(BHT) ] , where the chloride is now attached
+
−
2
to the aluminum center forming the aluminate anion. The over-
lay plot (Fig. 8) of 1H NMR spectra of [EMIM] [ClAlMe(BHT) ]
+
−
2
and the starting reagents, aluminum Lewis acid MeAl(BHT)2 and
IL [EMIM]Cl, clearly shows that, upon the aluminate formation, the
Al–Me signal is high-field shifted by 0.22 ppm from −0.33 ppm for
the neutral aluminum Lewis acid to −0.55 ppm for the anionic alu-
minate product, and the C(2)–H signal in the imidazolium cation
is also shifted to a higher field by 2.6 ppm (from 10.9 to 8.30 ppm).
All other signals are also high-field shifted upon the imidazolium
aluminate formation, although to a lesser extent.
+
−
Fig. 10. Molecular packing diagram of [EMIM] [ClAlMe(BHT)2] , highlighting weak
intermolecular (electrostatic) H–bonding between the C(2) on the imida-
The molecular structure of this compound was con-
firmed by X-ray diffraction analysis, featuring the ion pair
H
zolium ring and the Al Cl on the aluminate anion: C(2A) H(2A) = 0.93(4) A˚ ,
◦
consisting of the unassociated imidazolium cation [EMIM]+
H(2A) Cl(1B) = 2.742
A˚ , C(2A) Cl(1B) = 3.416 A˚ ; C(2A) H(2A) Cl(1B) = 130.08 .
−
and the aluminate anion [ClAlMe(BHT)2]
(Fig. 9). How-
+
−
ever, the crystal packing diagram depicted in Fig. 10 clearly
shows close intermolecular contacts between the cation
Overall, the structure [EMIM] [ClAlMe(BHT)2]
simulates
+
−
[EMIM] [CrCl3] , the proposed active metallate species respon-
sible for the effective glucose-to-HMF conversion by CrCl2 in
[EMIM]Cl [15], thereby reasoning why the [EMIM]Cl + AlMe(BHT)2
system is also very effective for this conversion. In comparison,
the [EMIM]Cl + AlCl3 or [EMIM]Cl + MeAlCl2 system is much infe-
rior. Although room-temperature chloroaluminate molten salts can
be readily formed by mixing [EMIM]Cl with AlCl3, the structure
depends on the mole fraction (N) of AlCl3 used in preparing the
melt [42]. The formed metallate salt [EMIM][AlCl4] has been shown
+
−
[
EMIM] of one molecular with the anion [ClAlMe(BHT)2]
of another molecule, via intermolecular H–bonding between
the C(2) H on the imidazolium ring and the Al Cl on the
aluminate anion. Metric parameters involved in this inter-
molecular H-bonding motif consist of C(2A) H(2A) = 0.93(4) A˚ ,
H(2A) Cl(1B) = 2.742 A˚ , C(2A) Cl(1B) = 3.416 A˚ ,
and
◦
C(2A) H(2A) Cl(1B) = 130.08 , pointing to weak (electro-
static) H-bonding [39]. Hence, this structure is significantly
+
−
◦
different from [EMIM]2n [Cr Cl ]n , derived from the reaction
to rapidly decompose HMF at 100 C [23]. Another key difference
2
6
of [EMIM]Cl + CrCl , in which the resulting anion forms one-
dimensional chains of chloride-bridged distorted octahedral
Cr(II) centers [40]. The geometry around the four-coordinate
between the metallate salts derived from aluminum chloride and
aluminum alkyl or alkoxy compounds is their relative hydrolytic
stability; as the conversion process eliminates water, the less
hydrolytically stable chloroaluminate salts from AlCl3 are expected
to undergo rapid catalyst deactivation as soon as the dehydration
reaction takes place.
2
aluminum center is that of
a sum of the O Al O(Cl) angles of 321.8 , as in the case
a distorted tetrahedron with
◦
of another rare example of the MeAl(BHT) -derived anion,
2
+
i
−
Li [Me C C(O Pr)OAlMe(BHT) ] [41]. The Al C bond [2.019(3) A˚ ]
2
2
is somewhat longer than that found in the neutral MeAl(BHT)2
1.927(3) A˚ ] [36], as predicted on changing from a planar to tetra-
hedral geometry in terms of the increased p character in the Al
bond. The Al O (BHT) bonds [1.752(2), 1.757(2) A˚ ] are also longer
than those found in the neutral MeAl(BHT)2 [1.687(2), 1.685(2) A˚ ]
4. Conclusions
[
C
Five classes of aluminum species have been investigated as
Lewis acid catalysts for the glucose-to-HMF conversion in IL
solvents. Amongst several interesting trends observed for these
aluminum-based catalysts, the most significant is the drastic
[
36].

D. Liu, E.Y.-X. Chen / Applied Catalysis A: General 435–436 (2012) 78–85
85
enhancement of the HMF yield upon a gradual substitution of
[5] (a) P. Wasserscheid, T. Welton (Eds.), Ionic Liquids in Synthesis, 2nd ed., Wiley-
VCH, Weinheim, 2008;
the chloride ligand on Al by the alkyl ligand. Hence, AlEt , which
3
(
(
b) V.I. Pârvulescu, C. Hardacre, Chem. Rev. 107 (2007) 2615–2665;
c) A. Stark, K. Seddon, Ionic Liquids in Kirk-Othmer Encyclopedia of Chemical
achieved 51% HMF yield from the reaction carried out in [EMIM]Cl
◦
at 120 C for 6 h, is a much better catalyst than AlCl , which gave
Technology, 26, John Wiley and Sons, New York, 2007, pp. 836–920.
3
[
6] (a) N. Sun, H. Rodriguez, M. Rahman, R.D. Rogers, Chem. Commun. 47 (2011)
only 1.6% HMF yield, under otherwise identical conditions.
Investigations into effects of temperature, time, catalyst load-
ing and IL structure on the HMF yield revealed that: (a) [EMIM]Cl
1
405–1421;
b) A. Pinkert, K.N. Marsh, S. Pang, M.P. Staiger, Chem. Rev. 109 (2009)
6712–6728.
(
is the most effective IL amongst the four ILs investigated herein;
[7] R.P. Swatloski, S.K. Spear, J.D. Holbrey, R.D. Rogers, J. Am. Chem. Soc. 124 (2002)
4974–4975.
◦
(
b) reactions carried out at 120 C for 6 h afford higher yields than
[
8] J.S. Wilkes, M.J. Zaworotko, Chem. Commun. (1992) 965–967.
the reactions under other conditions; and (c) 5 or 10 mol% alu-
minum alkyl catalyst loadings typically give similar yields, which
are higher than those with lower or higher catalyst loadings. Under
[9] R.C. Remsing, R.P. Swatloski, R.D. Rogers, G. Moyna, Chem. Commun. (2006)
1271–1273.
10] (a) J.D. Sean, A.T. Bell, Chem. Sus. Chem. 4 (2011) 1166–1173;
[
[
(
b) L. Vanoye, M. Fanselow, J. Holbrey, M.P. Atkins, K.R. Seddon, Green. Chem.
i
the current reaction conditions, AlEt3 and Al(O Pr)3 are at least as
11 (2009) 390–396;
effective as CrCl —the most effective benchmark metal halide cat-
(c) C. Li, Q. Wang, Z.K. Zhao, Green. Chem. 10 (2008) 177–182;
(d) C. Li, Z.K. Zhao, Adv. Synth. Catal. 349 (2007) 1847–1850.
11] Y. Zhang, H. Du, X. Qian, E.Y.-X. Chen, Energy & Fuels 24 (2010) 2410–2417.
2
i
alyst reported in literature, and yet AlEt3 and Al(O Pr)3 are about
5
and 180 times cheaper than CrCl , respectively, thereby showing
2
[12] For selected examples, see:
cost-saving features of the present alkyl aluminum catalyst system.
(a) L. Lai, Y. Zhang, Chem. Sus. Chem. 3 (2010) 1257–1259;
(b) J.Y.G. Chan, Y. Zhang, Chem. Sus. Chem. 2 (2009) 731–734;
+
−
The molecular structure of [EMIM] [ClAlMe(BHT) ] , formed
2
(
c) X. Qi, M. Watanabe, T.M. Aida, R.L. Smith Jr., Chem. Sus. Chem. 2 (2009)
upon mixing the aluminum Lewis acid MeAl(BHT)2 and the IL
944–946;
[
EMIM]Cl under the glucose conversion conditions, shows no
(d) Y. Román-Leshkov, J.N. Chheda, J.A. Dumesic, Science 312 (2006)
1933–1937;
intramolecular but weak intermolecular H-bonding between the
C(2) H and the Al Cl. This structure simulates that of the met-
allate [EMIM] [CrCl ] , the proposed active species responsible
(
(
e) C. Moreau, A. Finiels, L. Vanoye, J. Mol. Catal. A: Chem. 253 (2006) 165–169;
f) C. Lansalot-Matras, C. Moreau, Catal. Commun. 4 (2003) 517–520.
+
−
3
[13] (a) K.-I. Seri, Y. Inoue, H. Ishida, Bull. Chem. Soc. Jpn. 74 (2001) 1145–1150;
(b) K.-I. Seri, Y. Inoue, H. Ishida, Chem. Lett. (2000) 22–23;
for the effective glucose to HMF conversion by CrCl2 in [EMIM]Cl,
thereby reasoning why the [EMIM]Cl + AlMe(BHT)2 system is also
very effective for this conversion. Significantly, the unexpected
finding that the aluminum alkyl (alkoxy) catalysts are ∼32-fold
more effective than the aluminum halide catalyst should impact
future developments of new and more effective catalysts based on
other metals. The identification of the true active species under
the real biomass conversion conditions (i.e., in a mixture compris-
ing the sugar substrate, the aluminum precatalyst, the IL solvent,
and the water co-product) and the mechanistic aspects of the
aluminum-catalyzed glucose-to-HMF conversion are a subject of
on-going research.
(
c) H. Ishida, K.-I. Seri, J. Mol. Catal. A. Chem. 112 (1996) L163–L165.
[
14] S. De, S. Dutta, B. Saha, Green. Chem. 13 (2011) 2859–2868.
[15] H. Zhao, J.E. Holladay, H. Brown, Z.C. Zhang, Science 316 (2007) 1597–1600.
[16] E. Pidko, V. Degirmenci, R.A. van Santen, E.J.M. Hensen, Angew. Chem. Int. Ed.
49 (2010) 2530–2534.
[
17] C. Sievers, I. Musin, T. Marzialetti, M.B.V. Olarte, P.K. Agrawal, C.W. Jones, Chem.
Sus. Chem. 2 (2009) 665–671.
[18] S. Lima, P. Neves, M.M. Antunes, M. Pillinger, N. Ignatyev, A.A. Velente, Appl.
Catal. A 363 (2009) 93–99.
19] M. Chidambaram, A.T. Bell, Green. Chem. 12 (2010) 1253–1262.
20] G. Yong, Y. Zhang, J.Y. Ying, Angew. Chem. Int. Ed. 47 (2008) 9345–9348.
[
[
[21] S. Hu, Z. Zhang, J. Song, Y. Zhou, B. Han, Green. Chem. 11 (2009) 1746–1749.
[
22] Z. Zhang, Q. Wang, H. Xie, W. Liu, Z.K. Zhao, Chem. Sus. Chem. 4 (2011)
31–138.
1
[
[
[
23] T. Ståhlberg, M.G. Sørensen, A. Riisager, Green. Chem. 12 (2010) 321–325.
24] J.B. Binder, R.T. Raines, J. Am. Chem. Soc. 131 (2009) 1979–1985.
25] W.-H. Hsu, Y.-Y. Lee, W.-H. Peng, K.C.-W. Wu, Catal. Today 174 (2011)
Acknowledgments
65–69.
[
[
[
26] Y. Su, H.M. Brown, X. Huang, X.-D. Zhou, J.E. Amonette, Z.C. Zhang, Appl. Catal.
A 361 (2009) 117–122.
27] B. Kim, J. Jeong, D. Lee, S. Kim, H.-J. Yoon, Y.-S. Lee, J.K. Cho, Green. Chem. 13
This work was supported by US Department of Energy-Office
of Basic Energy Sciences, grant DE-FG02-10ER16193. DJL acknowl-
edges Prof. Susan James as a co-advisor. We thank Dr. Yuetao Zhang
for the help with X-ray diffraction analysis.
(
2011) 1503–1506.
28] C. Li, Z. Zhang, Z.K. Zhao, Tetrahedron Lett. 50 (2009) 5403–5405.
[29] X. Qi, M. Watanabe, T.M. Aida, R.L. Smith Jr., Chem. Sus. Chem. 3 (2010)
071–1077.
1
[
[
[
30] J.B. Binder, R.T. Raines, Proc. Natl. Acad. Sci. 107 (2010) 4516–4521.
31] X. Qi, M. Watanabe, T.M. Aida, R.L. Smith Jr., Cellulose 18 (2011) 1327–1333.
32] F. Tao, H. Song, L. Chou, Chem. Sus. Chem. 3 (2010) 1298–1303.
References
[1] For selected reviews, see:
[33] P. Wang, H. Yu, S. Zhan, S. Wang, Biores. Tech. 102 (2011) 4179–4183.
[34] F. Tao, H. Song, J. Yang, L. Chou, Carbohydrate Polym. 85 (2011) 363–368.
[35] E.Y.-X. Chen, T.J. Marks, Chem. Rev. 100 (2000) 1391–1434.
[36] A.P. Shreve, R. Mulhaupt, W. Fultz, J. Calabrese, W. Robbins, S. Ittel,
Organometallics 7 (1988) 409–416.
[37] S. Feng, G.R. Roof, E.Y.-X. Chen, Organometallics 21 (2002) 832–839.
[38] SHELXTL, Version 6.12; Bruker Analytical X-ray Solutions: Madison, WI, 2001.
[39] G.A. Jeffrey, An Introduction to Hydrogen Bonding, Oxford University Press,
1997.
[40] J.J. Danford, A.M. Arif, L.M. Berreau, Acta Cryst. E 65 (2009) m227.
[41] A. Rodriguez-Delgado, E.Y.-X. Chen, J. Am. Chem. Soc. 127 (2005)
961–974.
[42] K.M. Dieter, C.J. Dymek Jr., N.E. Heimer, J.W. Rovang, J.S. Wilkes, J. Am. Chem.
Soc. 110 (1988) 2722–2726.
(
(
(
a) D.M. Alonso, J.Q. Bond, J.A. Dumesic, Green. Chem. 12 (2010) 1493–1513;
b) M. Stöcker, Angew. Chem. Int. Ed. 47 (2008) 9200–9211;
c) NSF 2008. Breaking the Chemical and Engineering Barriers to Lignocellulosic
Biofuels: Next Generation Hydrocarbon Biorefineries”, G.W. Huber, Ed. report
from the NSF Workshop;
(
(
d) A. Corma, S. Iborra, A. Velty, Chem. Rev. 107 (2007) 2411–2502;
e) G.W. Huber, S. Iborra, A. Corma, Chem. Rev. 106 (2006) 4044–4098.
[
2] (a) J.N. Chheda, G.W. Huber, J.A. Dumesic, Angew. Chem. Int. Ed. 46 (2007)
164–7183;
b) G.W. Huber, J.N. Chheda, C.J. Barrett, J.A. Dumesic, Science 308 (2005)
446–1450.
7
(
1
[3] “Top Value Added Chemicals from Biomass”, T. Werpy, G. Petersen (Eds.), U.S.
Department of Energy (DOE) report: DOE/GO-102004-1992, 2004.
[4] Y. Román-Leshkov, C.J. Barrett, Z.Y. Liu, J.A. Dumesic, Nature 447 (2007)
9
82–986.