K. Beckerle, J. Okuda / Journal of Molecular Catalysis A: Chemical 356 (2012) 158–164
159
2. Experimental
the water peak can cause problems, considerable deviations can
occur for single points in the conversion curves. Determination of
cellobiose conversion refers to the doublet of the proton on the gly-
cosidic bond at 4.40 ppm. HMF content was determined based on
the average of the doublets of the ring protons at 6.58 and 7.44 ppm.
As there is no well separated signal for any single proton in fructose,
the overall integral of all the protons was used to estimate fructose
conversion.
In order to confirm the formation of HMF, several samples where
further analyzed by GCMS. For GCMS analysis, 2 mL samples were
diluted with water to 10 mL and extracted 3 times with CH2Cl2.
The organic phase was separated and filtered over a glass fibre
filter to remove any humins. The volume of the collected organic
extracts was reduced to 5 mL and the analysis of the solution was
performed on a Shimadzu GCMS-QP 2010 plus with helium as car-
rier gas. Oven temperature was raised from 60 ◦C to 250 ◦C at a
heating rate of 10 K/min. A Supreme-5-MS column (30 m length,
0.25 mm diameter, 0.25 m pores) at a column flow of 2 mL/min.
No soluble products other than HMF could be detected.
2.1. General
DMA (98%), DMF p. A., d-glucose (96%) and cellobiose (98%)
were purchased from Aldrich, fructose (99.5%) was purchased from
Südzucker, YCl3·6 H2O was purchased from Nanosolutions, ScCl3
and LaCl3 were purchased from Strem chemicals and LiCl extra
pure from Riedel-de Haën; all chemicals were used without further
purification.
Experiments at temperatures up to 145 ◦C were performed on
a Chemspeed ASW 1000 synthesizer with up to 36 wells with a
reaction volume of 13 mL connected to argon supply. The hood
containing the reactor blocks with the wells was permanently
flushed with nitrogen. Temperature was controlled by a Huber
Tango Nuevo cryostat. Reactions were vortexed at 600 rpm.
4 mL of a stock solution of the substrate (and LiCl where appro-
priate; see below) in DMA containing DMF as internal standard
were transferred to the reactor wells. Where necessary, 0.5 mL of
DMA were added to adjust the overall volume to 5 mL and the
atmosphere in the reactor wells was exchanged against argon by
repeated cycles of decreasing pressure and flushing with argon. The
reaction zone was heated to the desired temperature. A 0.1 mL sam-
ple of the mixture was injected into 0.5 mL of D2O as a reference for
NMR spectroscopic analysis prior to the transfer of 0.5 mL of stock
solution of the appropriate metal chloride in DMA (1 mL for ScCl3
stock solution), giving reaction mixtures that contained 2.24 mmol
of glucose or fructose (1.12 mmol of cellobiose) and 0.22 mmol of
metal chloride. The sampling procedure was repeated at the respec-
tive times.
3. Results and discussion
The conversion of fructose into HMF appears to be a fairly simple
cellulose as starting materials becomes more complicated as addi-
tional equilibria and side reactions occur. Specifically, hydrolysis of
glucose oligomers has to be addressed when HMF yield needs to be
optimized (Scheme 1). Another crucial step in the transformation
into HMF is the rearrangement of glucose to fructose [8]. If the equi-
librium between these two hexoses is shifted towards fructose, or
if equilibration proceeds quickly, the problem of HMF production
from d-glucose is reduced to the optimization of fructose conver-
sion.
are formed from d-glucose at elevated temperatures in the pres-
ence of rare earth metal chlorides but high yields are hampered by
humin generation. At 145 ◦C and with 10 mol% of yttrium trichlo-
ride, glucose is fully converted after 3 h, while only 20% of HMF are
formed (Fig. 1), with the HMF yield decreasing further at prolonged
reaction times.
Experiments at temperatures above 150 ◦C were performed in
a thick-walled Büchi glass reactor and stirred with a magnetic stir-
ring bar. Stock solutions where transferred manually with a plastic
syringe and the mixture was flushed with argon for 1 min before
the reactor was closed.
The results with scandium were reproduced manually as scan-
dium(III) chloride had to be applied as a suspension in DMA due to
its low solubility.
Experiments with fructose were conducted manually as devia-
tion of reaction times were to high due to the sampling speed of
the Chemspeed ASW 1000.
The following stock solutions were used:
Analysis of the 1H NMR spectra of samples suggests that there
is no substantial build-up of intermediates such as open-chained
saccharides, fructose, or coordination compounds. They should be
visible in the double bond region of the spectra in the case of dehy-
dration and in the region around ı 9–10 ppm for aldehydes that are
present in many of the intermediates discussed in the literature [9].
Since there are no by-products that could be converted into HMF
Yttrium(III) chloride: 673 mg (2.22 mmol) of YCl3·6 H2O in 5 mL of
DMA, giving a 0.444 M solution.
Lanthanum(III) chloride: 544 mg (2.22 mmol) of LaCl3 in 5 mL of
DMA, giving a 0.444 M solution.
Scandium(III)chloride: 168 mg (1.11 mmol) of ScCl3 in 5 mL of
DMA, giving a 0.222 M solution.
Glucose: 10 g (0.056 mol) d-Glucose (and 4.71 g LiCl where appro-
priate) were transferred to a 100 mL volumetric flask. 5 mL of
N,Nꢀ-dimethylformamide (DMF) were added as internal standard
for NMR measurement. Overall volume was adjusted to 100 mL
with DMA.
Cellobiose: 9.5 g of cellobiose and LiCl (4.71 g) were transferred to
a 100 mL volumetric flask. 5 mL of N,Nꢀ-dimethylformamide (DMF)
were added as internal standard for NMR measurement. Overall
volume was adjusted to 100 mL with DMA.
2.2. Analytical
NMR measurements were performed on a Bruker Avance
400 MHz spectrometer. For determination of glucose conversion
the integrals of the hydrogen on the anomeric carbon at 4.52 and
5.11 ppm were used and correlated with the proton of the formic
acid moiety of DMF at 7.83 ppm. As separation from the tailing of
Fig. 1. Conversion of d-glucose into HMF in DMA at 145 ◦C with 10% LnCl3.