Selective dehydration of glucose to hydroxymethylfurfural and a one-pot synthesis of a 4-acetylbutyrolactone from glucose and trioxane in solutions of aluminium salts

Article abstract of DOI:10.1016/S0008-6215(99)00037-3

Saturated water solutions of Al₂(SO₄)₃ and AlCl₃ were applied as solvent/matrices for dehydration of Glc to hydroxymethylfurfural (HMF). Addition of oxygen ligands: methanol, ethanol, THF, furan, dibutyl ether, ethyl orthoformate and trioxane influenced the yield and selectivity, the best being observed with ethanol. When Glc and trioxane were present together in reacting solution, formation of a 4-acetylbutyrolactone was observed. Copyright (C) 1999 Elsevier Science Ltd.

Full text of DOI:10.1016/S0008-6215(99)00037-3

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
Carbohydrate Research 315 (1999) 268–272
Selective dehydration of glucose to
hydroxymethylfurfural and a one-pot synthesis of a
4-acetylbutyrolactone from glucose and trioxane in
solutions of aluminium salts
a,
a
b
b
*
Stanislaw K. Tyrlik *, Dorota Szerszen´ , Marian Olejnik , Witold Danikiewicz
a Institute of Chemistry, Agriculture and Teachers Uni6ersity, 3go Maja 54, PL-08-110 Siedlce, Poland
b Institute of Organic Chemistry, Polish Academy of Sciences, Kasprzaka 44/52, PL-01-224 Warszawa, Poland
Received 3 August 1998; accepted 24 January 1999
Abstract
Saturated water solutions of Al2(SO4)3 and AlCl3 were applied as solvent/matrices for dehydration of Glc to
hydroxymethylfurfural (HMF). Addition of oxygen ligands: methanol, ethanol, THF, furan, dibutyl ether, ethyl
orthoformate and trioxane influenced the yield and selectivity, the best being observed with ethanol. When Glc and
trioxane were present together in reacting solution, formation of a 4-acetylbutyrolactone was observed. © 1999
Elsevier Science Ltd. All rights reserved.
Keywords: Glucose; Dehydration; Aluminium salts; Oxygen ligands
elimination, which in turn originates from an
1. Introduction
‘anions guiding rail’ effect [2]. The sulfates
and chlorides thereby constitute two classes of
salts exerting different effects on the reactivity
of glucose.
The dehydration reaction of glucose in satu-
rated aqueous solutions of MgCl2 and MgSO4
affords various amounts of hydroxymethylfur-
fural (HMF), other furan derivatives, the 2-
acetyl propionic acid (I, levulinic acid) and
humins [1]. The dehydration to HMF in solu-
tions of sulfate probably proceeds with the
participation of aquo and hydroxy complexes
by intermediacy of an ‘extensively hydrogen-
bonded species’ [1]. This is a selective reaction
but proceeds with low conversion. In MgCl2
solution conversion is much greater than in
MgSO4. The formation of five carbon furan
derivatives in water solutions of MgCl2 is
probably caused by an exocyclic –CH2OH
A high-yield, selective and experimentally
simple, method for dehydration of glucose
(partial or complete) has not so far been de-
scribed. To achieve these goals there has to be
a rational selection of the metallic centre and
associated ligands. We limited ourselves to an
application of ligands, L, which are able to
offer an oxygen atom to the metal. In reaction
solution one can therefore expect formation of
{MLx}Ap according to the reaction:
Metal salt MAp+xL“{MLx}Ap
(1)
(A, anion; p, number of associated anions).
The oxygen ligands tested were the CH3OH,
EtOH, trioxane, di-n-butyl ether, furan, THF
* Corresponding author. Fax: +48-25-44-20-45.
0008-6215/99/$ - see front matter © 1999 Elsevier Science Ltd. All rights reserved.
PII: S0008-6215(99)00037-3

S.K. Tyrlik et al. / Carbohydrate Research 315 (1999) 268–272
Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
269
and triethyl orthoformate (TEOF) systems. In
this paper we describe mainly the application
of Al2(SO4)3 and AlCl3; in some cases MgCl2
was used.
scribed in Ref. [1]. The retention times span
up to 1.7 min because the experiments were
done over a period of about 2 years, which
caused changes in the column properties. The
identity of chromatographic peaks was always
established by complementary MS data.
Isolation and purification of a methyl ether
of HMF was by column chromatography as
follows: crude reaction products (1.7 g) were
chromatographed on silica gel (E. Merck, 30
g) with 1:4 acetone–CHCl3. The chromatogra-
phy was repeated on the intermediate fraction
(the one which was least contaminated as
assessed by TLC) with toluene on silica gel to
give 0.007 g of V. NMR data: l 4.45 (s. 3H,
–OCH3); 4.50 (s, 2H, OCH3); 6.53 (d, 1H,
ꢀCH); 7.22 (d, 1H, ꢀCH); 9.63 (s, 1H,
–CHO). MS: MW 140; m/z (relative inten-
sity):140 (32), 139 (6), 112 (6), 111 (100), 109
(40), 81 (18), 80 (6), 68 (6), 55 (11), 53 (45),
52(14), 51(12), 50(8), 45(6), 39(14). The molec-
ular weight was confirmed by an LSIMS tech-
nique, using m-nitrobenzyl alcohol as a
matrix. IR: no –OH band; 1690 cm−1 vs.
(CꢀO). HMF ethyl ether (III) was identified
by MS spectrum: MW 154; m/z (relative in-
tensity) 154 (27), 125(86), 111 (13), 109(100),
97 (93), 81 (44), 69(24), 53(48), 52(29), 51(17),
50 (10), 41 (25), 38(20) and by measurement of
the molecular weight by LSIMS. Separation
of 2-acetylbutyrolactone (VI) from crude reac-
tion products was performed as follows: the
crude product (0.09 g ) was chromatographed
on silica gel (E. Merck, ꢀ2.0 g) with CHCl3.
NMR data of combined TLC pure fractions:
l 2.26 (s, 3H, CH3CO), 2.60–2.87 (m, 2H),
3.48–3.62 (m, 1H), 4.33–4.55 (m, 2H). Our
NMR data agree with those of Muller [3]. The
NMR data of other papers [4,5] show minor
differences. MS: MW 128; m/z (relative inten-
sity: 128(6), 100(5), 86(30), 71(9), 55(22),
43(100), 42(11), 38(8.); IR (film): 1780, 1718
cm−1. The reference conditions for synthesis
of VI were: 10 cm3 of saturated water solution
of AlCl3 (ꢀ29.6 mmol), 0.5 g of Glc (2.7
mmol), 1.0 g of trioxane (11.1 mmol), and 2
cm3 of TEOF (12 mmol).The pH of the react-
ing systems was always assessed by litmus
paper, to within pH 3–6 without addition of
an acid, so that reactions were always per-
formed in slightly acidic solutions except when
alkaline reagents had been added.
2. Experimental
General.—For GC–MS analysis a Hewlett-
Packard 5890 Series II gas chromatograph
combined with a model 5972 mass-selective
detector was used equipped with HP-5 and
Innovax columns. The molecular weight was
measured using LSIMS on an AMD-604 dou-
ble focusing mass spectrometer (MAGD In-
1
tectra GmbH, Harpsted, Germany). H NMR
spectra were recorded in CDCl3 at 200 MHz
with a Varian Gemini instrument. IR spectra
were recorded on a Perkin–Elmer 1670 FT
unit.
Analytical grade Glc, MeOH, EtOH Al and
Mg salts were all purchased from Polskie Od-
czynniki Chemiczne (POCh, Gliwice, Poland).
Remaining metal salts, (n-Bu)2O, furane,
THF, trimethyl and triethylorthoformates,
trimethyl and triethyl orthoacetates, were
commercial products of Aldrich or E. Merck.
MeOH, EtOH, water and triethyl orthoacetate
were distilled but other reactants were used as
received. Reactions were performed under
gentle reflux with agitation by a magnetic
stirrer.
Methods.—The composition of the reacting
solution defined as a reference reaction was as
follows: 9 cm3 of saturated water solution of
an aluminium sulfate (containing 8.33 mmol
of Al2(SO2)4, 8.33 mmol of Glc and 324 mmol
of EtOH). This reference reaction mixture was
then modified: (a) at a constant concentration
of Glc; the molar ratio Glc/Al2(SO4)3 varied
from 9 to 0.67; the molar ratio EtOH/Glc
286–6.1; in such a range of conditions the
highest yield of HMF ꢀ7% was observed for
Glc/Al2(SO4)3 0.67 and EtOH/Glc 6.1 (data
concerning yields 91.5%); and (b) at a con-
stant EtOH/Al2(SO4)3 ratio of 40; the Glc/
Al2(SO4)3 varied from 9 to 0.67; the highest
yield of ꢀ8% was observed for the range
1.3–0.67. Reactions performed only in water
gave no products. Isolation of crude products
and their GC analysis were performed as de-

270
S.K. Tyrlik et al. / Carbohydrate Research 315 (1999) 268–272
Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
3. Results
The same was observed in saturated aqueous
solutions of MgCl2 [1]. However, upon addition
of EtOH to saturated aqueous solutions of
magnesium chloride (10 cm3 of MgCl2 solution,
3 cm3 of EtOH and 8.33 mmol of Glc) a clean
reaction mixture without tar and coke was
observed [1]. The presence of MeOH or EtOH
here also made the reaction more selective and
cleaner. Capillary gas chromatograms of reac-
tion products formed in systems containing
water solutions of Al2(SO4)3 and of AlCl3 with
alcohols are shown in Fig. 1.
When Glc reacts in saturated water solution
of Al2(SO4)3, after 3–4 h a dark heterogeneous
mixture with coke-like material is produced.
Consequently, reactions were performed in
the presence of EtOH. There were four main
products in the presence of EtOH in the range
of conditions studied: HMF, 2-acetylpropionic
acid (I), the ethyl ester of 2-acetylpropionic acid
(II) and HMF ethyl ether (III). With MeOH,
2-acetylpropionic acid methyl ester (IV) and
HMF methyl ether (V) formed in place of II and
III.
Mass spectra of HMF ethers are not in the
Wiley Library so the ethers were isolated and
identified by known methods (Section 2). The
remaining products were identified using the
Wiley Library (1993) with a PBM search al-
gorithm and by comparison with authentic
samples. In the reference reaction humins were
not formed. Other compounds detected by GC
were present in minor quantities, so the refer-
ence system produced HMF (together with
compounds of its subsequent reactions) rather
selectively. In a previous paper [1] we described
the existence of a dehydration product of Glc
in the presence of EtOH having a retention time
of 55.8 min and called a ‘154’ product. Contrary
to our preliminary assumptions that this
product was a carbocyclic substance, we found
it to be the ethyl ether of HMF (Section 2). Also
observed was formation of this ether with
aqueous MgCl2. The reference system was than
modified by several procedures: acidification or
making the solution alkaline (Table 1), and by
introducing salts of cations considered as hard
(or borderline [6]) Lewis acids (Table 2). Con-
version of Glc in saturated aqueous solutions
of sulfates of Ag(I), Fe(II), Fe(III), Cu(II),
Zn(II), Cr(III) and of ‘RuSO4’ was negligible
except sulfates of Fe(III) and Cr(III), where the
HMF and its ether were formed with ꢀ8%
yield (with EtOH; concentrations as in the
Fig. 1. Capillary gas chromatograms: (A) products of a
reaction in 9 cm3 of saturated aqueous solution of Al2(SO4)3
(8.33 mmol), 19.3 cm3 of EtOH (324 mmol) and of Glc (8.33
mmol). (B) Products of reaction with 10 ml of a saturated
aqueous solution of AlCl3, 8 cm3 of MeOH (197 mmol) and
8.33 mmol of Glc. In this system, HMF methyl ether and
levulinic acid methyl ester are formed.
Table 1
Influence of acidification and of the presence of alkaline
reagents on yield of products in the reference system
Acid or alka- Molar ratio (Glu:acid, S yield of all
line reagent
Glu:alk)
products (%)
HCl
40:1
8
HBr
H2SO4
HCl
HBr
H2SO4
4:1
6
4
NaOH
CaO
40:1
MgO
NEt3
NaOH
CaO
4:1
2
MgO
NEt3

S.K. Tyrlik et al. / Carbohydrate Research 315 (1999) 268–272
Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
271
Table 2
Influence of selected metal salts and complexes on yield of products
Salt or complex
System
Molar ratio
Yield (%)
Sulfates of Ag(I), Fe(III), Cu(II), Reference system
Zn(II), Cr(III), ‘RuSO4’ [1]
Glu:sulfate=10 10–20% decrease in yield compar-
ing to the reference system
Chlorides of Al(III), Ti(III),
Fe(III); nitrates of Al(III) and
of Fe(III)
Reference system
Glc:salt=10
10–40% decrease
PdCl2
Saturated aq solution of AlCl3+
MeOH or EtOH (50–300 mmol)
Glc conc as in ref system
Glc:Pd=10–50 HMF and its ethers 3–14, levulinic
acid and its esters 3–22
H2IrCl6
Saturated aq solution of MgCl2+ Glc:Ir=10–50 HMF and its ethers 3, levulinic
MeOH or EtOH; Glc conc as in
ref system
acid and esters 0.5
reference reaction). Upon diluting the reference
system with water the conversions were negligi-
ble; and similarly in a saturated solution of
sodium sulfate and in aluminium sulfate with
different amounts of 30% H2O2 in the presence
of Fe(II) sulfate (Fenton reagent). Saturated
aqueous solutions of MgCl2 with MeOH (50–
300 mmol), produced HMF and V in 4–10%
yield (concentrations as in the reference sys-
tem), whereas saturated aqueous solutions of
AlCl3 with MeOH produced only levulinic acid
(28%) and its ester (2%). HMF and I–V were
sometimes accompanied by other furane
derivatives such as 2-furancarboxaldehyde,
methoxy furan 1-(2-furanyl)ethanone, albeit in
minor quantities. A series of reactions of the
reference system were performed with pro-
longed reaction times. Results are given in
Table 3. Gas chromotograms of extractable
products indicated the existence of other minor
products in comparison to those observed in the
‘24 h’ reaction. Yields reached maximum with
time but than decreased. In addition to MeOH
and EtOH, THF, furane, (n-Bu)2O, TEOF and
trioxane were used as ligands in aqueous solu-
tions of Al salts. The presence of THF com-
pletely inhibited any reaction of Glc; tar and
coke-like materials were also not formed. With
furane and (n-Bu)2O only I was formed with
24–26% yield. Addition of furan produced
considerable amounts of humins. When triox-
ane and TEOF were present together in solu-
tion, it was observed that one compound
appeared as the most abundant component of
a crude extract (retention time 62.4 min, Inno-
vax column). This compound was unknown to
the Wiley Library, and after isolation was
identified as 4-acetylbutyrolactone VI. We
studied the influence of several variables on this
reaction and found that a molar ratio of triox-
ane/glucose=4 was the best. With a higher
trioxane/glucose ratio, reactions become very
non-selective, and with trioxane/glucose=7–8
VI was not formed at all. In 10 mL of saturated
aqueous solution of AlCl3 there are ꢀ29.6
mmol of AlCl3 and ꢀ479 mmol of water, so
the AlCl3/H2O ratio was ꢀ16. A reaction
performed with this ratio gave a negligible yield
of lactone and is non-selective. Increasing the
water/AlCl3 ratio by addition of water gave the
results shown in Table 4. The effect of the
(TEOF/AlCl3) ratio was also studied: when it
was \0.6 formation of VI was inhibited. For
a ratio up to 0.2, formation of VI was negligible
and the reaction was not selective. The best
results were obtained for TEOF/AlCl3 =0.3–
0.4 with other conditions summarised in Table
4. Application of other orthoesters such as
triethyl orthoacetate, trimethyl orthoformate
or acetate did not ameliorate either selectivity
or the yield. These esters were studied in condi-
Table 3
Dependence of product yields upon reaction time using the
ratios of reactants as in the reference system
Reaction
time (h)
Yield of HMF and III
(%)
Yield of I and II
(%)
24
72
124
144
168
7
13
20
22
15
2
4
6
3