Catalytic effect of aluminium chloride on the example of the conversion of sugar model compounds

Article abstract of DOI:10.1016/j.molcata.2015.03.018

Abstract In this work, the catalytic effect of the Bronsted acid hydrochloric acid, the Bronsted base sodium hydroxide and the Lewis acid AlCl₃ on the conversion of biomass derived carbohydrates is investigated. On the example of the glycolaldehyde conversion, it is shown that the Lewis acid catalyses the ketol-endiol-tautomerism, the dehydration, the retro-aldol-reaction and the benzilic-acid-rearrangement. The main products are C₄- and C₆-carbohydrates as well as their secondary products 2-hydroxybut-3-enoic acid 1 and several furans. Under the same reaction conditions hydrochloric acid catalyzes mainly the dehydration and sodium hydroxide the tautomerism and subsequent aldolization.

Full text of DOI:10.1016/j.molcata.2015.03.018

Journal of Molecular Catalysis A: Chemical 402 (2015) 64–70
Contents lists available at ScienceDirect
Journal of Molecular Catalysis A: Chemical
journal homepage: www.elsevier.com/locate/molcata
Catalytic effect of aluminium chloride on the example of the
conversion of sugar model compounds
Martin Schwiderskia,∗, Andrea Kruseb,1
a Karlsruhe Institute of Technology, Hermann-von-Helmholtz-Platz 1, 76344 Eggenstein-Leopoldshafen, Germany
b University of Hohenheim, Garbenstrasse 9, 70599 Stuttgart, Germany
a r t i c l e i n f o
a b s t r a c t
Article history:
In this work, the catalytic effect of the Brønsted acid hydrochloric acid, the Brønsted base sodium hydrox-
ide and the Lewis acid AlCl3 on the conversion of biomass derived carbohydrates is investigated. On the
example of the glycolaldehyde conversion, it is shown that the Lewis acid catalyses the ketol-endiol-
tautomerism, the dehydration, the retro-aldol-reaction and the benzilic-acid-rearrangement. The main
products are C4- and C6-carbohydrates as well as their secondary products 2-hydroxybut-3-enoic acid 1
and several furans. Under the same reaction conditions hydrochloric acid catalyzes mainly the dehydra-
tion and sodium hydroxide the tautomerism and subsequent aldolization.
Received 4 November 2014
Received in revised form 18 February 2015
Accepted 21 March 2015
Available online 24 March 2015
Keywords:
Carbohydrates
Lewis acid
© 2015 Elsevier B.V. All rights reserved.
Brønsted acid
Brønsted base
Furans
␣-Hydroxy acids
1. Introduction
glucose or biomass has been established yet. Since, the conversion
of glucose to HMF 3 requires severe reaction conditions glucose
In the hydrothermal conversion of lignocellulosic biomass, the
catalyst is of special importance for the production of chemicals.
Herein‘, Brønsted acids and bases are often used for catalysis. Using
a Brønsted acid as the catalyst, biomass can be direct converted
into platform chemicals: in the 1920’s, the Quaker Oats process
had been patented [1]. Herein, biomass is treated in aqueous media
containing sulphuric acid under elevated temperatures for several
hours. The acid catalyzes mainly two reactions. First, the hydroly-
sis of the hemicellulose to xylose and second the triple dehydration
of xylose to furfural. The yield of furfural in the Quaker Oats pro-
cess is limited to about 50% (mol/mol). In recent years, production
processes underwent several improvements. In the BIOFINE pro-
cess, the yield is increased up to 75% (mol/mol) [2]. Furthermore,
levulinic acid is formed from the biomass containing cellulose.
Furfural and levulinic acid 2 are both important platform chemi-
cals. They are designated under the top ten carbohydrates derived
molecules by Bozell and Petersen [3]. However, no industrial appli-
cation for the production of 5-hydroxymethylfurfural HMF 3 from
undergoes several undesired side reactions, especially the forma-
tion of humins. Also, the formed HMF 3 tends to hydrolyze to
levulinic acid 2 [4].
The conversion of cellulosic biomass using an alkaline cata-
lyst is well investigated in the literature. Often used catalysts are
sodium hydroxide and calcium hydroxide. Starting from cellu-
lose, the main reaction products are organic acids including oxalic,
acetic, formic, lactic and succinic acid. Herein, high catalyst dosages
of about 1 g catalyst per g biomass and elevated temperatures above
160 ◦C are required for the conversion. A good overview of the
direct conversion of cellulose using alkaline catalysts is given by
Knill and Kennedy [5]. At comparable catalyst loadings but lower
temperatures cellulose is not converted, but most of the lignin is
removed. This makes the cellulose more accessible for the enzy-
matic hydrolysis to glucose [6–8]. Following, the glucose solution
can be fermented to platform chemicals [3].
The use of metal salts for the catalytic conversion of biomass
to furans as well as levulinic acid is also investigated in the lit-
erature. Herein, it is shown that multivalent metal ions, especially
chromium(III), copper(II), iron(III) and aluminium(III) salts catalyze
the conversion of hemicellulose and cellulose [9,10]. On the exam-
ple of the Sn-beta-catalyst, it is shown that metal ions also catalyze
the isomerization of aldoses to ketoses [11]. The dehydration of the
formed ketoses to furans is much faster than the dehydration of
∗
Corresponding author. Tel.: +49 721 608 26511; fax: +49 721 608 22244.
E-mail addresses: martin.schwiderski@kit.edu (M. Schwiderski),
Andrea Kruse@uni-hohenheim.de (A. Kruse).
1
Tel.: +49 711 459 24700; fax: +49 711 459 24702.
http://dx.doi.org/10.1016/j.molcata.2015.03.018
1381-1169/© 2015 Elsevier B.V. All rights reserved.

M. Schwiderski, A. Kruse / Journal of Molecular Catalysis A: Chemical 402 (2015) 64–70
65
3+
Al
3+
Al
O
ketol-endiol-tautomerism
HO
OH
H
OH
4
5
OH
aldol-reaction
OH
5
OH OH
OH
O
OH OH
HO
HO
HO
H
H
H
OH
O
OH
O
7
6
8
OH
OH
OH
H
OH
OH
H
4
5
5
OH
O
OH OH
OH OH
O
OH OH
HO
HO
H
HO
HO
H
OH
HO
OH
OH
O
OH OH
O
OH
OH
HO
11
9
13
12
OH OH OH
HO
OH
O
H
HO
OH
OH
O
OH
OH
10
14
Scheme 1. Formation of C4- and C6-carbohydrates starting from glycolaldehyde 4.
the corresponding aldoses. However, tin containing catalysts are
not effective on the dehydration, thus a Brønsted acid needs to be
added. A mixture of the tin-beta-catalyst and HCl leads to quite
good yields of furfural or HMF 3 starting from xylose or glucose,
respectively [12]. Using aluminium chloride in pulping processes
of lignocellulosic biomass furfural is obtained without addition of
any Brønsted acids. At a temperature of 120 ◦C and pure water as the
solvent furfural yields below 5% are achieved [13]. Using a solvent
mixture like THF/water or ethanol/water and higher temperatures,
the furfural yield increases up to about 50% [14,15]. One of the major
byproducts of the conversion of C5- and C6-carbohydrates using
metal ion catalysts is lactic acid [14,16].
fructose, methylglyoxal and glycolaldehyde. Since, AlCl3 decom-
poses in aqueous solution to hydrochloric acid and the hydroxyl
compound, the catalytic effect is compared to those of HCl and
NaOH. Of special interest is the ability to catalyze the ketol-endiol-
tautomerism, the dehydration, the retro-aldol-reaction and the
benzilic-acid-rearrangement. These reactions are the most com-
mon ones in the conversion of carbohydrates.
2. Reactions of carbohydrates – literature overview
2.1. Enolisation and aldolisation of glycoladehyde
The metal salt aluminium chloride plays a special role. After dis-
solving in water, the formed hexahydrate hydrolyzes furthermore.
Many hydrolysis products are suggested in the literature. A good
overview is given by Bottero et al. [17]. The hydrolysis products are
strongly dependent on the pH of the solution [18,19]. However, all
of these hydrolysis reactions lead to similar products. On the one
hand, hydrochloric acid is released, on the other hand, a complex
consisting of one or more aluminium ions (in one special case 13
aluminium ions are coordinated to 4 oxygen ions) coordinated to
aqua- and hydroxyligands is formed. Due to these structural sim-
ilarities, the hydrolysis of aluminium chloride will be discussed
according to the simplified Eq. (1) in the following part of this paper.
The formose reaction starting with formaldehyde is a well inves-
tigated reaction in the literature. Herein, Ca(OH)2 is often used
as the catalyst and the reaction is carried out at room temper-
ature [20–22]. A cascade of aldol-reactions leads to a mixture of
straight- and branched- chain C3–C7-carbohydrates. In the recent
literature, it is shown that also metal ions like the bivalent zinc
or the tetravalent tin are able to catalyze aldol reactions starting
from glycolaldehyde 7 to form higher carbohydrates [23,24]. Due
to their Lewis acid- character, multivalent metal ions coordinate
multiple OH bonds and stabilize the transition state of enols. In
this work, the trivalent aluminium ions are used as the catalyst. In
Scheme 1, the possiblealdol-reactions starting fromglycolaldehyde
are shown. The key step is the enolisation of glycoladehyde 4 to the
endiol 5 and the subsequent aldolization to form aldotetroses 6.
The aldotetroses 6 can isomerize to the corresponding enols 7 and
to tetruloses 8. Thus, there is a variety of aldol- reactions possible.
AlCl3 + 6H2O ꢀ [Al(H2O)6]Cl3 ꢀ Al(OH)3 + 3H3O⊕ + 3Clꢀ
(1)
In this work, the catalytic effect of AlCl3 is investigated on
the example of the conversion of the sugar model compounds

66
M. Schwiderski, A. Kruse / Journal of Molecular Catalysis A: Chemical 402 (2015) 64–70
OH OH
O
HO
H
OH OH
9
OH OH
HO
OH
OH
HO
H
O
+ H
- H2O
H
HO
O
O
OH
H
H
OH
OH
H
HO
H
HO
OH
O
OH
- H
10
16
HO
O
O
O
HO
O
HO
- H2O
+ 2 H2O
- H2O
O
O
OH
H
H
H
H
H
O
OH
HO
HO
2
3
Scheme 2. Reaction mechanism of the formation of HMF 3 starting from an aldohexose 9.
As can be seen in Scheme 1, possible reaction products are straight-
chain aldohexoses 9, ketohexoses 10 and 3-keto-hexoses 11. Also
branched- chain C6-carbohydrates 12–14 can be formed, however,
they are not reported in the literature yet. The formation of octoses
is also possible but it is not part of the investigation in this work.
3. Materials and methods
Glycolaldehyde-dimer (98%), aqueous methylglyoxal solution
(∼40%), d-fructose (p.a.), water free aluminium chloride (p.a.), HMF
(p.a.) were purchased from “Sigma–Aldrich“. Aqueous hydrochlo-
ric acid (0.1 M), aqueous sodium hydroxide (0.25 M), both with a
purity of p.a. were purchased from “Merck”.
2.2. Dehydration of C6- carbohydrates – formation of furans
There are several mechanisms of the formation of HMF 3 from
aldohexoses 9 or one of their tautomers postulated [25]. However,
the isomerisation of the aldose 9 to the ketose 10 and then triple
dehydration seems to be the dominant reaction path (Scheme 2)
[26]. One widely accepted mechanism is the dehydration start-
ing from a ketohexose 10. The first step is the isomerisation to
a furanose 16. The next step is Brønsted acid catalyzed. Protona-
tion and dehydration of the hemiacetal hydroxyl group at C2 lead
to a tertiary carbenium ion. Subsequent deprotonation and dou-
ble dehydration lead to HMF 3. As already mentioned HMF 3 is an
intermediate and it tends to hydrolyze to levulinic acid 2.
Considering these facts, the open chained carbohydrates 11–14
first isomerize to their corresponding furanoses 17–20. Herein,
there are two possibilities to form a five membered ring for the
carbohydrate 13 leading to the furanoses 18 and 19. All the fura-
noses dehydrate in an analogue mechanism as the furanose 16.
The initial step is the dehydration of the highest substituted alco-
hol to form a tertiary carbenium ion followed by deprotonation
and two consecutive dehydrations in the ring to form 1-(furan-
2-yl)-2-hydroxyethanone 21, 1-(furan-3-yl)-2-hydroxyethanone
22, 2-(hydroxymethyl)-3-furaldehyde 23, 3-(hydroxymethyl)-2-
furaldehyde 24 (Scheme 3).
3.1. General procedure
In separate volumetric flasks 2.4 g (20 mmol) glycolaldehyde-
dimer, 6.15 ml (40 mmol) methylglyoxal-solution, 7.2 g (40 mmol)
d-fructose and 0.27 g (2 mmol) aluminium chloride were dissolved
in 100 ml water. The resulting concentrations of the sugar com-
pounds were 0.4 M and of aluminium chloride 0.02 M. Before use,
the solutions were stirred over night. As the reactor steel autoclaves
with 10 ml volume were used. Then 3 ml of the sugar compound
solution and 3 ml of the AlCl3- solution or 1.8 ml 0.1 M (0.18 mmol)
HCl and 1.2 ml water or 0.24 ml 0.25 M (0.06 mmol) NaOH and
2.76 ml water were added. The initial concentrations for all exper-
iments were 0.2 M glycolaldehyde-monomer or methylglyoxal or
d-fructose and 0.01 M AlCl3 or 0.01 M NaOH or 0.03 M HCl. Then, the
reactor was placed in a GC-oven and heated up for different heat-
ing rates. When the oven reached the desired temperature, the time
was set to be zero. Therefore, only the time at this temperature was
called “reaction time” here. During the reaction, the autoclave was
mechanically turned with 20 rpm. After the desired reaction time,
the oven was cooled down and the autoclave was left in the oven.
After cooling down the product, solution was poured into a 10 ml
volumetric flask and the reactor was washed twice with 2 ml water.
2.3. Benzilic acid rearrangement
3.2. Isolation of the products
The mechanism of the benzilic acid rearrangement starting from
glyceraldehyde is postulated in the literature. Herein, zinc sulphate
is used as the catalyst and the main reaction product is lactic acid
[16]. Applying this mechanism on the conversion of tetroses and
aluminium chloride as the catalyst, the acid 1 is formed (Scheme 4).
The first step is the ketol-enol-tautomerism of aldotetroses 6 fol-
lowed by two dehydration steps of the enols 7 to form vinylglyoxal
25. Aluminium chloride decomposes in aqueous solutions into alu-
minium hydroxide Al(OH)3 and hydrochloric acid HCl. In the next
step, these OH groups function as nucleophiles and attack the
aldehyde carbon. Subsequent hydrid shift and hydrolysis lead to
the acid 1.
As the reactor a non stirred steel autoclave with 500 ml vol-
ume was used. The reaction solution was prepared by adding
100 ml 0.4 M aqueous glycolaldehyde to 100 ml 0.02 M aqueous
AlCl3- solution. The mixture was stirred with a spatula for 1 min.
Then the reactor was sealed and placed in a GC- oven. The oven
was heated to 170 ◦C and the reaction time was 60 min. After
cooling down the product, solution was three times extracted
with 200 ml dichloromethane. The organic layer was dried over
magnesium sulphate, concentrated and purified over a column
using mesoporous silica as the stationary phase and ethylac-
etate/cyclohexane (1:1 = v/v) as the solvent. Following furans could
be isolated: 1-(furan-3-yl)-2-hydroxyethanone22 (RF = 0.48): 1H

M. Schwiderski, A. Kruse / Journal of Molecular Catalysis A: Chemical 402 (2015) 64–70
67
OH
O
OH
O
OH OH
HO
HO
OH
OH
O
HO
H
H
HO
OH
HO
OH
OH
14
OH
OH
O
OH
OH
OH OH
HO
13
12
11
OH
OH
OH
OH
O
OH
OH
O
O
HO
O
OH
OH
OH
HO
OH
OH
HO
HO
OH
HO
HO
OH
20
19
18
17
- 3 H2O
- 3 H2O
- 3 H2O
- 3 H2O
- 3 H2O
OH
O
O
O
O
O
H
O
OH
H
OH
OH
O
O
22
24
21
23
Scheme 3. Formation of furans from open chained hexoses via furanose intermediates.
4J1,2 < 2 Hz, 4J1,3 < 2 Hz. 13C NMR (250 MHz, C3D6O): ı = 71 (1-C), 118
(2,3-C), 133 (4-C), 176 (q-C) ppm. The 1H NMR-spectra of the iden-
tified compounds are shown in supplement Fig. S1–S5. The 13C-
signals were assigned with their corresponding 2D-NMR-spectra
(HMQC, HMBC).
NMR (250 MHz, CDCl3): ı = 8.11 (m, 1-H1), 7.52 (m, 2-H1), 6.81
(m, 3-H1), 4.65 (s, 4-H2) ppm. 13C NMR (250 MHz, CDCl3):
ı = 147 (1-C), 145 (2-C), 108 (3-C), 66 (4-C), 194 (q-C), 124 (q-C)
ppm. 2-(Hydroxymethyl)-3-furaldehyde23 (RF = 0.42): 1H NMR
(250 MHz, CDCl3): ı = 10.00 (s, 1-H1), 7.41 (d, 2-H1), 6.80 (d. 3-
3
H1), 4.88 (s, 4-H2) ppm. J2,3 = 2.0 Hz. 13C NMR (250 MHz, CDCl3):
ı = 187 (1-C), 142 (2-C), 110 (3-C), 57 (4-C), 162 (q-C), 123 (q-C)
ppm. 3-(Hydroxymethyl)-2-furaldehyde24 (RF = 0.32): 1H NMR
(250 MHz, CDCl3): ı = 9.83 (s, 1-H1), 7.64 (d, 2-H1), 6.62 (d, 3-
3.3. Analytical methods
Glycolaldehyde 4, methylglyoxal, levulinic acid
2 and 2-
3
H1), 4.82 (s, 4-H2) ppm. J2,3 = 1.6 Hz. 13C NMR (250 MHz, CDCl3):
hydroxybut-3-enoic acid 1 were measured using HPLC on a Merck
Hitachi device. The column is Aminex® HPX-87H. The eluent is
aqueous 0.004 M sulphuric acid, the flow rate is 0.5 ml min−1 and
the temperature 60 ◦C. Glycolaldehyde 7 and methylglyoxal were
quantified using an RI- detector, the acid 1 using an UV–vis-detector
at a wavelength of 210 nm and levulinic acid 2 at a wavelength of
280 nm. Furans were also measured using HPLC on a Merck/Hitachi-
device. It is a “Hypersil ODS” column. The mobile phase is a 9:1
(v:v)- mixture of acetonitrile and water with a flow rate of 0.5
ml min−1, the temperature is 30 ◦C. The samples are detected with
an UV–vis- detector at a wavelength of 290 nm. Before measuring
on HPLC basic samples were adjusted to a pH of 2 and acidic samples
were measured directly.
ı = 181 (1-C), 148 (2-C), 113 (3-C), 57 (4-C), 150 (q-C), 138 (q-
C) ppm. 5-(Hydroxymethyl)-2-furaldehyde HMF 3 (RF = 0.26): 1
H
NMR (250 MHz, CDCl3): ı = 9.61 (s, 1-H1), 7.24 (d, 2-H1), 6.54 (d,
3
3-H1), 4.74 (s, 4-H2) ppm. J2,3 = 3.5 Hz. The remaining extracted
aqueous phase was adjusted with 0.25 M NaOH- solution to a pH
of 10 and three times extracted with 200 ml diethylether. The
diethylether layers were discarded. Then the aqueous phase was
adjusted with 1 M HCl- solution to a pH of 2 and again three times
extracted with 200 ml diethylether. The combined organic phases
were dried over magnesium sulphate and the solvent was removed
under reduced pressure. The resulting yellow solid was identi-
fied as 2-hydroxybut-3-enoic acid 1. 1H NMR (250 MHz, C3D6O):
ı = 4.69 (ddd, 1-H1), 5.21 (ddd, 2-H1), 5.48 (ddd, 3-H1), 6.04 (ddd,
4-H1) ppm. 2J2,3 < 2.0 Hz, 3J1,4 = 4. 8 Hz, 3J2,4 = 10.4 Hz, 3J3,4 = 17.2 Hz,
Calibration of the furans (except HMF 3) and the acid 1
was done as following described: a small portion of the puri-
OH
H
OH
H
H
- H2O
HO
HO
H
HO
O
O
O
HO
HO
O
H
Al(OH)3
Al(OH)3
Al(OH)3
6
7
- H2O
H
H
O
OH
O
OH
-
H2O
H
-
~ H
-
O
OH
O
O
O
O
25
OH
Al
Al
Al
1
HO
HO
OH
HO
OH
+
OH
OH
Al(OH)3
Scheme 4. AlCl3 catalyzed reaction mechanism of the formation of 2-hydroxybut-3-enoic acid 1.

68
M. Schwiderski, A. Kruse / Journal of Molecular Catalysis A: Chemical 402 (2015) 64–70
A
100
80
B
40
30
20
10
0
60
40
130
140
150
160
[°C]
170
180
130
140
150
160
[°C]
170
180
T
T
Fig. 1. (A) Conversion of glycolaldehyde 7 and (B) product yields: ꢁ C4-carbohydrates, ꢀ C6-carbohydrates, ᭹ 2-hydroxybut-3-enoic acid 1, X Furans. Reaction conditions:
c0,glycolaldehyde = 0.2 M, c0,AlCl3 = 0.01 M, VH2O = 6 ml, T = 170 ◦C, t = 20 min.
fied compounds was dissolved in 0.2 ml D2O and 0.2 ml of
3-(trimethylsilyl)-2,2ꢁ,3,3ꢁ-tetradeuteropropionic acid TMSP-d4 in
D2O (cTMSP = 6.34 g l−1) was added. A 1H NMR-spectrum was mea-
sured. Then three dilutions with water were made and measured on
the HPLC- devices. HMF 3 was calibrated using commercial avail-
able substance.
temperature their yield is also increasing. At temperatures above
170 ◦C, the acid 1 is the main product. The yield of the furans is with
below 6% low. Their distribution is shown in Fig. 2.
With the results shown on Fig. 2, the reaction schemes 1, 3 and 4
will be discussed. The yield of HMF 3 compared to the other furans
is at every temperature low and furan 21 is not found at all. Except
the aldohexose 9, all carbohydrates are able to form a furanose ring
which is suitable for the dehydration to furans. Thus an isomeri-
sation of the aldose 9 to the ketose 10 is required to form HMF 3.
The low yield of HMF 3 and that exclusively straight- chain carbo-
hydrates are found lead to the assumption that the isomerisaton of
C6-carbohydrates is the rate limiting step in the formation of furans.
The isomerisation of 10 to 11 is more unlikely and thus no furan 21
is formed. Another possibility to produce carbohydrate 11 and thus
furan 21 is the aldolisation of the C4-enol 7 and glycolaldehyde 4.
However, in a competitive reaction the enol 7 dehydrates to form
vinylglyoxal 25. Probably, the dehydration is much faster and the
sugar 11 and thus furan 21 are not present. The carbohydrate 13 is
able to dehydrate without further isomerisation to any tautomers.
That’s why in the whole temperature range the secondary products
22 and 23 are the most present furans in the product mixture.
Carbohydrates were quantified as their alditol acetates. The
derivatisation procedure is described elsewhere [27]. Basic sam-
ples were derivatised directly and acidic samples were adjusted
to a pH of 10. Xylose was added as an internal standard. Quan-
titative measurements of the carbohydrates were carried out on
an Agilent GC 6890 on a Resteck 13323 column. The carrier gas is
helium with a flow rate of 1.4 ml min−1. The injection temperature
is 35 ◦C and the GC-oven is heated to 280 ◦C within 10 min. The
split is 30.8:1 and the FID temperature 310 ◦C. GC–MS- analysis is
carried out on a Trace Ultra GC coupled with a DSQ II quadrupole
MS detector from the Thermo Scientific. The samples are measured
at 250 ◦C (injector temperature) on a 25-m ULTRA 2 silica col-
umn ((5% phenyl)-methylpolysiloxane) from Agilent Technologies.
A constant gas flow rate of 1 ml min−1 with the following temper-
ature program are applied: initial temperature 40 ◦C, increased at
6 ◦C min−1 to 200 ◦C, further increased with 8 ◦C min−1 to a final
300 ◦C, kept for 5 min. The MS detector is operated in positive ion-
ization mode at 70 eV with an ion source temperature of 200 ◦C. A
GC-chromatogram and the mass spectra of the carbohydrates are
shown in supplement Fig. S6–S14.
4.2. Conversion of glycolaldehyde using different catalysts
Aluminium chloride AlCl3 decomposes in aqueous solutions to
aluminium hydroxide Al(OH)3 and hydrochloric acid HCl. The cat-
4. Results and discussion
6
5
4
3
2
1
0
4.1. AlCl3 catalyzed conversion of glycolaldehyde at various
temperatures
In this section, the conversion of glycolaldehyde 4 using AlCl3 as
the catalyst is discussed. Therefore, the simplified reaction scheme
including the main reactions aldolisation, dehydration and the ben-
zilic acid rearrangement is shown in Scheme 5. In Fig. 1A, the
conversion of glycolaldehyde 4 and in Fig. 1B, the product yields
at various temperatures are shown. At 130 ◦C, the conversion of
glycolaldehyde 7 is 41% and at 170 ◦C and higher temperatures it is
completely converted. At 130 ◦C besides glycolaldehyde 4 only C4-
and C6-carbohydrates are found in the product mixture. The C4-
carbohydrates achieve a maximum yield of 40% at a temperature
of 150 ◦C. The C6-carbohydrates are in the whole temperature range
from 130 ◦C to 180 ◦C with maximum 3% less concentrated. Herein,
no branched- chain carbohydrates can be detected. At 180 ◦C, no
C6-sugars are found at all. The secondary products of the carbo-
hydrates are present at 140 ◦C for the first time. With increasing
140
150
160
[°C]
170
180
T
Fig. 2. Furan yields in dependency of the reaction temperature. Reaction conditions:
c0,glycolaldehyde = 0.2 M, c0,AlCl3 = 0.01 M, VH2O = 6 ml, t = 20 min.
furan 24, HMF 3.
furan 22,
furan 23,

M. Schwiderski, A. Kruse / Journal of Molecular Catalysis A: Chemical 402 (2015) 64–70
69
O
Aldolisation
Aldolisation
O
2
C4- carbohydrates
OH
C6- carbohydrates
Dehydration
H
4
OH
H
1. Dehydration
2. Benzilic-acid-
rearrangement
4
O
O
O
HO
O
O
O
OH
OH
O
H
O
H
OH
O
22
21
1
2-hydroxybut-3-enoic acid
3
OH
O
O
H
H
OH
O
23
24
Furans
Scheme 5. Simplified reaction scheme of the AlCl3 catalyzed conversion of glycolaldehyde 4.
Table 1
Conversion of glycolaldehyde 7 and product yields using different catalysts. Reaction conditions: c0,glycolaldehyde = 0.2 M, VH2O = 6 ml, T = 170 ◦C, t = 20 min.
Entry
Catalyst
Conversion
glycolaldehyde 4
[mol-%]
Yield
C4-carbohydrates
[mol-%]
Yield acid 1 [mol-%]
Yield
C6-carbohydrates
[mol-%]
Yield furans
[mol-%]
1
2
3
4
none
34%
51%
100%
100%
8%
0
0
0
23%
Traces
Traces
27%
0
0
0.03 M HCl
0.01 M NaOH
0.01 M AlCl3
12%
20%
20%
Traces
5%
2%
alytic effect of aluminium ions compared to the effect of a Brønsted
acid HCl and a Brønsted base sodium hydroxide NaOH is examined.
Therefore the conversion of glycolaldehyde 4 using these three cat-
alysts is investigated (Table 1).
effect of the enolisation of glycoladehyde 4. Furthermore, AlCl3 cat-
alyzes the dehydration of the formed carbohydrates, whereas in the
NaOH catalyzed reaction just traces of the dehydration products are
formed.
The non catalyzed (entry 1) and the HCl catalyzed (entry 2)
reaction show a similar product distribution. The conversion of gly-
colaldehyde 4 and the yield of C4- carbohydrates in entry 2 are just
slightly higher than in entry 1. In both cases C6- carbohydrates are
formed just in traces and no acid 1 and no furans are formed. In
the 1H NMR-spectrum of the extracted product solution of entry 2
(Supplement Fig. S15) just glycolaldehyde 4 in its monomeric form
can be determined.
4.3. Conversion of fructose
On the example of the fructose conversion, the ability of HCl,
NaOH and AlCl3 to catalyze the dehydration, the retro-aldol-
reaction and the benzilic acid rearrangement is investigated. The
results are given in Table 2.
Using NaOH (entry 3) as the catalyst glycolaldehyde 4 is fully
converted. The main products are C4- and C6-carbohydrates. The
distribution of the C6-carbohydrates is about 75% straight- chain
and 25% branched- chain (Supplement Fig. S6 and Fig. S7). In entry
3 no acid 1 is present and just HMF 3 is formed in traces.
Also, in the AlCl3 catalyzed reaction (entry 4) glycolaldehyde 4
is fully converted. The main products using AlCl3 as the catalyst
are tetroses and the acid 1. The C6-carbohydrates are the minor
products, the main part of them is further converted to furans.
Evaluating these results, it can be concluded that both AlCl3
and NaOH are efficient catalysts for the enolisation and subsequent
aldolization. In the case of using HCl, there is just a minor catalytic
In the case of using HCl, fructose is completely converted after
60 min reaction time. The yield of the dehydration products (HMF 3
and the secondary product levulinic acid 2) is 74%. No retro-aldol-
products like dihydroxyacetone, glyceraldehyde, methylglyoxal or
lactic acid are formed.
In the AlCl3, catalyzed reaction fructose is also completely con-
verted and the yield of the dehydration products is 42%. In this
reaction a quite high amount of 27% of lactic acid is produced. Lac-
tic acid derives from the retro-aldol-products dihydroxyacetone or
its isomer glycerinaldehyde. Double dehydration leads to methyl-
glyoxal and a benzilic acid rearrangement to the final product lactic
acid. This mechanism is equal to the one shown in Scheme 4.
Table 2
Conversion of fructose and product yields using different catalysts. Reaction conditions: c0,fructose = 0.2 M, VH2O = 6 ml, T = 170 ◦C, t = 60 min.
Entry
Catalyst
Conversion
fructose
[mol-%]
Yield
HMF 3
[mol-%]
Yield
levulinic acid 2
[mol-%]
Yield
lactic acid
[mol-%]
1
2
3
0.03 M HCl
0.01 M NaOH
0.01 M AlCl3
100%
54%
99%
11%
11%
13%
63%
0
29%
0
0
27%

70
M. Schwiderski, A. Kruse / Journal of Molecular Catalysis A: Chemical 402 (2015) 64–70
Table 3
the tautomerism. Aluminium chloride combines the catalytic prop-
erties of a Brønsted acid and a Brønsted base. All the previously
mentioned reactions are catalyzed by the Lewis acid.
Conversion of methylglyoxal and product yields using different catalysts. Reaction
conditions: c0,methylglyoxal = 0.2 M, VH2O = 6 ml, T = 170 ◦C, t = 20 min.
Entry
Catalyst
Conversion
methylglyoxal
[mol-%]
Yield lactic acid
[mol-%]
In view of biorefineries, especially in the production of the plat-
form chemical 5-Hydroxymethylfurfural 3, Brønsted bases are not
suitable because of their low dehydration ability on carbohydrates.
If inulin containing biomass, such as Jerusalem artichoke or chicory
roots, is used as the starting material Brønsted acids are very effec-
tive catalysts for the formation of HMF 3. However, Brønsted acids
fail in the direct conversion of cellulosic biomass to HMF 3 because
of their missing ability to catalyze the tautomerism of glucose to
fructose. Since the Lewis acid AlCl3 catalyzes both, the tautomerism
and the dehydration, it is an effective catalyst for the direct conver-
sion of cellulose to platform chemicals. But AlCl3 also catalyzes the
retro-aldol-reaction and the benzilic-acid-rearrangement. Both are
undesired side reactions. Thus there are extensive kinetic studies
required to optimize the yield of HMF 3.
1
2
3
0.03 M HCl
0.01 M NaOH
0.01 M AlCl3
29%
100%
100%
0
0
87%
If glucose is used as the starting material much more HMF 3 is
produced, if AlCl3 is used as the catalyst [14]. Although HCl is a very
efficient catalyst for dehydration reactions, it fails to enolize and
isomerize glucose to fructose. Aluminium chloride decomposes in
aqueous solutions to triaquatrihydroxyaluminium and hydrochlo-
ric acid. The aluminium ions catalyze the isomerisation to fructose
and the hydrochloric acid the dehydration to HMF 3.
NaOH has the weakest effect on both the conversion and HMF
3 yield. Just 54% of the initial fructose is converted and the yield of
the dehydration products is 11%. Conspicuous is that the secondary
product levulinc acid 2 is not formed. Also, no retro-aldol-products
are detected. In the literature, it is shown that Brønsted bases are
effective catalysts for the retro-aldol-reaction and the benzilic-
acid-rearrangement to form lactic acid from C6-carbohydrates.
Treating an aqueous glucose solution containing Ca(OH)2 at the
boiling point, on the one hand lactic acid is formed with a yield of
26%. On the other hand, cyclic enols and phenols are found in the
product solution [28]. That means an alkaline catalyst is efficient for
the retro-aldol-reaction. Next to the benzilic acid rearrangement
to form lactic acid other reactions such as recombination, aldol-
condensation and cyclisation to form cyclic compounds take place
[29]. In this work, it is shown that at higher temperatures also the
dehydration of C6-carbohydrates to form HMF 3 plays an important
role. But the effect of the dehydration of NaOH is low compared to
HCl and AlCl3.
Acknowledgements
This work was financially supported by Karlsruhe Institute of
Technology. We also thank Ms. Birgit Rolli, Ms. Sonja Habicht, Mr.
Armin Lautenbach and Mr. Klaus Raffelt for analytical support.
Appendix A. Supplementary data
Supplementary data associated with this article can be
found, in the online version, at http://dx.doi.org/10.1016/j.molcata.
2015.03.018.
References
[1] H.J. Brownlee, C.S. Miner, UK Pat., 203691-A, 1923.
[2] S.W. Fitzpatrick, U.S. Pat., 4897497, (1990).
[3] J.J. Bozell, G.R. Petersen, Green Chem. 12 (2010) 539.
[4] B. Girisuta, L.P.B.M. Janssen, H.J. Heeres, Ind. Eng. Chem. Res. 46 (2007) 1696.
[5] C.J. Knill, J.F. Kennedy, Carbohydr. Polym. 51 (2003) 281.
[6] V.S. Chang, M. Nagwani, M.T. Holtzapple, Appl. Biochem. Biotechnol. 74
(1998) 135.
4.4. Conversion of methylglyoxal
[7] W.E. Karr, M.T. Holtzapple, Biomass Bioenergy 18 (2000) 189.
[8] M.J. Payne, Biotechnol. Bioeng. 26 (1984) 426.
[9] L. Peng, L. Lin, J. Zhang, J. Zhuang, B. Zhang, Y. Gong, Molecules 15 (2010)
5228.
[10] Y. Ma, W. Ji, X. Zhu, L. Tian, X. Wan, Biomass Bioenergy 39 (2012) 106.
[11] M. Moliner, Y. Roman-Leshkov, M.E. Davis, PNAS 107 (2010) 6164.
[12] E. Nikolla, Y. Roman-Leshkov, M. Moliner, M.E. Davis, ACS Catal. 1 (2011)
408.
[13] Y. Luo, L. Hu, D. Tong, C. Hu, RSC Adv. 4 (2014) 24194.
[14] Y. Yang, C. Hu, M.M. Abu-Omar, Green Chem. 14 (2012) 509.
[15] M. Schwiderski, A. Kruse, R. Grandl, D. Dockendorf, Green Chem. 16 (2014)
1569.
[16] M. Bicker, S. Endres, L. Ott, H. Vogel, J. Mol. Catal. A 239 (2005) 151.
[17] J.Y. Bottero, J.M. Cases, F. Fiessinger, J.E. Poirier, J. Phys. Chem. 84 (1980)
2933.
[18] Proceedings, Twenty-First Workshop on Geothermal Reservoir Engineering
Stanford University. Stanford, California, January 22-24, 1996, page 201-207.
[19] M. He, H. Fu, B. Su, H. Yang, J. Tang, C. Hu, J. Phys. Chem. B 118 (2014)
13890.
[20] E.H. Ruckert, E. Pfeil, G. Scharf, Chem. Ber. 98 (1965) 2558.
[21] J.D. Sutherland, J.N. Whitfield, Tetrahedron 53 (1997) 11493.
[22] A. Butlerow, Liebigs Ann. Chem. 120 (1861) 295.
[23] J. Kofoed, M. Machuqueiro, J.-L. Reymond, T. Darbre, Chem. Commun. 13
(2004) 1540.
To form the acid 1 or lactic acid, an enolization step of the
C4- or C3-carbohydrates is necessary (Scheme 4). To form lactic
acid from C6-carbohydrates, the retro-aldol-reaction to form C3-
carbohydrates is additionally needed. In the previous results, it is
shown that there is either just a weak or no catalytic effect of HCl
on the enolisation and retro-aldol-reaction. To prove the ability of
HCl, to catalyze the benzilic acid rearrangement additional experi-
ments of the conversion of methylglyoxal are done. The results are
shown in Table 3. The AlCl3 catalyzed conversion is completely and
leads to a high yield of 87% of lactic acid. Also, in the NaOH catalyzed
reaction methylglyoxal is completely converted. However, no lactic
acid is formed. As mentioned above at higher temperatures other
reactions than the benzilic acid rearrangement are favoured. The
effect of HCl on the reaction is weak. Just 29% of methylglyoxal is
converted and no lactic acid is formed.
5. Conclusion
[24] M. Dusselier, P. Van Wouwe, S. De Smet, R. De Clercq, L. Verbelen, P. van
Puyvelde, F.E. Du Prez, B.F. Sels, ACS Catal. 3 (2013) 1786.
[25] R.-J. van Outten, J.C. van der Waal, E. de Jong, C.B. Rasrendra, H.J. Heeres, J.G.
de Vries, Chem. Rev. 113 (2013) 1499.
[26] M.J. Antal, W.S.L. Mok, G.N. Richards, Carbohydr. Res. 199 (1990) 91.
[27] M. Schwiderski, A. Kruse, R. Grandl, D. Dockendorf, RSC Adv 4 (2014)
45118.
In this work, the catalytic effect of NaOH, HCl and AlCl3
on the example of the conversion of glycolaldehyde 4, fructose
and methylglyoxal is investigated. It is shown that the Brønsted
acid catalyzes the dehydration but it has a weak effect on the
ketol-endiol-tautomerism, retro-aldol-reaction and the benzilic-
acid-rearrangement. The catalytic effect of the Brønsted base is
opposite. It is a weak dehydration catalyst but it mostly catalyzes
[28] B.Y. Yang, R. Montgomery, Carbohydr. Res. 280 (1996) 27.
[29] I. Forsskähl, T. Popoff, O. Theander, Carbohydr. Res. 48 (1976) 13.