Hydrogenation of fructose on Ru/C catalysts

Article abstract of DOI:10.1016/S0008-6215(00)00146-4

The hydrogenation of D-fructose on Ru/C catalysts was studied. Under the conditions applied (1 bar H₂, 72 °C), the furanose forms of D-fructose react, while the pyranose forms do not. However, all anomers adsorb with comparable strength on the surface. The reaction rate is controlled by product inhibition. The selectivity to D-mannitol can be increased from 47 to 63% by promotion of Pd/C and Pt/C catalysts with Sn. (C) 2000 Elsevier Science Ltd.

Full text of DOI:10.1016/S0008-6215(00)00146-4

www.elsevier.nl/locate/carres
Carbohydrate Research 328 (2000) 449–457
Hydrogenation of fructose on Ru/C catalysts
Annemieke W. Heinen, Joop A. Peters *, Herman van Bekkum
Laboratory of Applied Organic Chemistry, Delft Uni6ersity of Technology, Julianalaan 136,
NL-2628 BL Delft, The Netherlands
Received 14 April 2000; accepted 4 May 2000
Abstract
The hydrogenation of
the furanose forms of
comparable strength on the surface. The reaction rate is controlled by product inhibition. The selectivity to
-mannitol can be increased from 47 to 63% by promotion of Pd/C and Pt/C catalysts with Sn. © 2000 Elsevier
D-fructose on Ru/C catalysts was studied. Under the conditions applied (1 bar H , 72 °C),
2
D
-fructose react, while the pyranose forms do not. However, all anomers adsorb with
D
Science Ltd. All rights reserved.
Keywords: Activated carbon; Tin, promoter;
D-Mannitol; D-Glucitol
1
. Introduction
mannitol can be isolated by crystallisation. By
contrast, -glucitol is prepared by hydrogena-
D
The hydrogenation of
D
-fructose (see
tion of inexpensive
selectivity.
D
-glucose with 100%
Scheme 1) is an interesting reaction in view of
the production of the low-caloric sweetener
There is growing interest in the production
-mannitol from -fructose due to the
D
-mannitol [1].
from the relatively expensive sources
tose and -mannose in yields of about 45 and
00%, respectively. This explains the price dif-
ference: $3.32/pound for -mannitol and
0.75/pound for -glucitol (US market) [2].
D
-Mannitol can be obtained
of
D
D
D-fruc-
increasing production of inulin. Inulin is an
D
oligosaccharide consisting of (2“1)-linked b-
1
D-fructofuranosyl units attached to a
D-glu-
D
cose end group. Inulin is a potentially
attractive feedstock for the production of
$
D
D
-
When applying an isomerization catalyst (glu-
cose isomerase) in combination with a hydro-
genation catalyst (e.g., supported copper)
mannitol due to its high fructose-to-glucose
ratio. This study focuses on the hydrogenation
of fructose.
D
-glucose can also serve as starting compound
for -mannitol [3,4]. For economical reasons
-mannitol is prepared by hydrogenation of a
:1 mixture of -glucose and -fructose, ob-
tained by hydrolysis of sucrose. In this way a
mixture of about 70% -glucitol (sorbitol) and
0% -mannitol is obtained, from which
Noble metals, preferably supported on acti-
vated carbon, are frequently used catalysts for
carbohydrate conversions [5]. Activated car-
bon is cheap, has a high surface area and is
resistant to both strongly acidic and strongly
basic media. Another advantage is the easy
recovery of the precious metal from the cata-
lyst by burning off the carbon support. How-
ever, due to the fact that carbon originates
from natural products, like for instance peat
or wood, every batch of activated carbon can
D
D
1
D
D
D
3
D
D-
*
Corresponding author. Tel.: +31-15-2785892; fax: +31-
152-784289.
E-mail address: j.a.peters@tnw.tudelft.nl (J.A. Peters).
0
008-6215/00/$ - see front matter © 2000 Elsevier Science Ltd. All rights reserved.
PII: S0008-6215(00)00146-4

4
50
A.W. Heinen et al. / Carbohydrate Research 328 (2000) 449–457
of the amount of carboxylic groups. For the
untreated carbon, the relative intensity of the
−
1
1
700 cm band was 0, whereas for the HNO3
treated carbon this value amounted to 0.38. In
addition to the increased acidity, the carbon
had a more hydrophilic surface by this treat-
ment, as was shown previously by means of
competitive adsorption of water and toluene
Scheme 1. Hydrogenation of
-mannitol. -Fructose is presented in its open form, which
represents just 3% of the mutarotation equilibrium in water at
0 °C.
D-fructose to D-glucitol and
[9].
D
D
Carbon monoxide adsorption measure-
8
ments, X-ray fluorescence spectroscopy (XRF)
and transmission electron microscopy (TEM)
analyses were performed on the Ru/C cata-
lysts. The XRF and CO data are displayed in
Table 1. These data show that the Ru content
have different properties. These differences
can have large influences on the performance
of the catalyst.
Ruthenium is the most active catalyst in
carbohydrate hydrogenations, both in homo-
geneous [6,7] and in heterogeneous catalysis
of SX1G, impregnated with aqueous RuCl , is
3
somewhat lower than that of the other cata-
lysts, probably due to the smaller amount of
anchoring groups on the surface and the lim-
ited wettability of the surface by water. The
CO adsorption of these catalysts is, however,
comparable with the other ones. An Ru dis-
persion of 77%, assuming 100% linear CO
adsorption, is the highest of all catalysts.
A TEM study was performed on these cata-
lysts. On Ru/SX1G impregnated with aqueous
RuCl and reduced in 10% H /N , metal parti-
[
8]. This paper describes a study on the hydro-
genation of -fructose using Ru/C catalysts.
D
Besides the kinetics of this reaction and an
investigation of the reacting tautomeric forms,
the effect of different carbon supports and
impregnation solvents on the activity and se-
lectivity of these catalysts is presented. Fur-
thermore, the effect of tin promotion on the
selectivity to
D-mannitol is investigated.
3
2
2
cles could not be clearly discerned, although
Ru was clearly detected using EDX elemental
analysis. Therefore, the metal must be present
as very small (B0.5 nm) particles. The same
observations were made after use of this cata-
lyst in a hydrogenation experiment. Since the
atomic radius of Ru is 0.134 nm [10], this
means that the metal is present as clusters of
only a few atoms.
2
. Results and discussion
Catalyst characterization.—The amount of
acidic oxygen sites (mainly carboxylic groups)
on the carbon was increased by treatment with
nitric acid. Titration of the oxidized carbon
SX1GN65 (see experimental) with sodium hy-
−
1
droxide showed the presence of 1.0 meq g
Reduction of the same catalyst-precursor in
−
1
of acid sites, compared with 0.1 meq g
for
100% H gave a catalyst with mainly small
2
the untreated carbon. The intensity of the
particles of around 0.5 nm, besides a few of a
somewhat larger size (1–1.5 nm). The catalyst
−
1
1
700 cm
band in the IR spectrum, relative
−1
to the band at 1580 cm is also an indication
impregnated with ethanolic RuCl showed
3
Table 1
CO adsorption results of the Ru/C catalysts prepared on different carbons and with different solvents
−
1
Catalyst
Ru content (wt%)
CO adsorbed (mL g
)
CO/Ru
Ru/SX1G (acetone) 10% H2
Ru/SX1G (water) 10% H2
Ru/SX1G (water) 100% H2
Ru/Darco (water) 10% H2
3.50 (90.11)
2.82 (90.09)
2.82 (90.09)
3.47 (90.11)
3.35 (90.11)
5.0
5.1
5.2
4.0
5.7
0.59
0.75
0.77
0.47
0.69
‘
5 wt% Ru/C’, Acros

A.W. Heinen et al. / Carbohydrate Research 328 (2000) 449–457
451
ditions for
D
-glucose (batch) hydrogenation
are 70–140 bar and 120–160 °C [11]. In con-
tinuous hydrogenation, higher pressures (up
to 180 bar) are applied. The effect of tempera-
ture and pressure on the conversion of -fruc-
D
tose over the commercial 5 wt% Ru/C is
shown in Fig. 1. The order in H is about zero
2
and even at 1 bar a high activity is obtained.
Therefore, the reaction was further studied
under 1 bar H pressure and at 72 °C.
2
The reaction was 100% selective to the two
alditols, i.e., no side product was formed. The
Fig. 1. Conversion of
genation of -fructose over 0.05 g Ru/C (Acros) at 100 °C
and 100 bar H (×); 100 °C, 30 bar (ꢀ); 60 °C, 30 bar (ꢁ);
D-fructose against time in the hydro-
D
selectivity to
D-mannitol over this catalyst
2
varied from 40% (100 °C, 100 bar) to 43%
72 °C, 1 bar (ꢂ); 38 °C, 1 bar ("); 1 g D-fructose, 80 mL
(
72 °C, 1 bar) in the reactions over Ru de-
water.
scribed here.
Table 2
The system was checked for external trans-
port limitation by varying the catalyst concen-
tration and the impeller speed. A linear
relationship between the initial rate constant
and the amount of catalyst was observed,
which indicates that the reaction is first order
in catalyst concentration. This, together with
the fact that there is no influence of the stir-
ring speed (at]1000 rpm) on the activity,
shows that external diffusion is not rate con-
trolling. With the Ru dispersion displayed in
Table 1 (assuming 100% linear CO adsorp-
tion, i.e., CO:Ruexposed =1:1), the turnover
number (TON) amounts to 483.
Reaction rate of D-fructose hydrogenation at different D-fruc-
tose starting concentrations; 0.05 g Ru/C (Acros), 80 mL
water, 72 °C, 1 bar H2
−
1
−1 −1
[
F] (mol L
)
Fructose (g)
r0 (mol L
h
)
0
0
.034
0.5
1.0
0.047
0.033
.069 (standard
experiment)
.139
.347
.656
0
0
1
2.0
5.0
24.0
0.033
0.035
0.030
small particles of 0.5–1 nm, just visible with
this technique. Impregnation with an acetone
solution resulted in a high Ru signal with
EDX analysis, although Ru particles were not
observed. This indicates the presence of Ru
particles with a size below the detection limit.
The catalysts based on Darco and SX1GN65
showed somewhat larger particles. The parti-
cles on Ru/Darco range from 1 to 1.5 nm,
with, according to EDX, smaller particles in
between; on SX1GN65 small particles (0.5–2
nm) were observed, besides a few clusters up
to 20 nm. The same observations were made
for the commercial Ru/C catalyst.
Comparing the TEM results with the CO
adsorption data of Table 1, we can conclude
that the trends observed with TEM are in
agreement with the results of the adsorption
data.
Kinetics of fructose hydrogenation.—Hydro-
genations of carbohydrates are usually carried
out under high pressure. Typical reaction con-
A plot of ln [
straight line (with a correlation coefficient of
0.9980), suggesting first order kinetics in
fructose. However, upon increasing or de-
creasing the initial -fructose concentration,
D
-fructose] against time gives a
D
-
D
the initial reaction rate does not change ac-
cordingly (Table 2). The selectivity, however,
was independent of the
D
-fructose starting
concentration. Even at a concentration as
high as 30 wt% of -fructose, the selectivity to
D
mannitol amounted to 43%.
The reaction was checked for product inhi-
bition by addition of
D
-glucitol and/or -man-
D
nitol to the reaction mixture (Fig. 2). This
Figure indicates a competition between
D
-
fructose and the reaction products for adsorp-
tion on the catalyst.
The results displayed in Fig. 2 can be ex-
plained using Langmuir–Hinshelwood kinet-
ics [12,13]. According to the following
reaction (where * presents an empty reaction
site):

4
52
A.W. Heinen et al. / Carbohydrate Research 328 (2000) 449–457
b
f
fructose (aq)+ꢀ X fructose (adsorbed)
C can be introduced, which reduces the rate
expression to:
k
“
mannitol+glucitol
the reaction rate will be proportional to the
-fructose surface coverage qf:
r=kb c /(b c +b c +C)
f f
f f
p p
This model explains the gradually decreasing
rate that was observed during the hydrogena-
tion experiments (see Fig. 2). As Table 2
shows, the reaction rate is independent of the
D
r= −d[F]/dt=kqf
The adsorption coefficients b an b
are
f
H2
D-fructose starting concentration. The model
defined as:
is only in line with this observation if the
constant C can be neglected. Furthermore, the
pseudo first-order kinetics that were observed
b =q /c q
ꢀ
f
f
f
2
2
b =q /(p q )
H2
H2
H2
ꢀ
during the experiments indicate that b and bf
p
During the hydrogenation, the surface cover-
age of -fructose is given by:
must be of the same order of magnitude. With
curve fitting, a good correlation was obtained
D
−
1
q =b c /(b c +b c +ꢀ(b p )+b c +1)
when b /b :2 and k=0.049 h
.
f
f f
f f
p p
H2 H2
s s
f
p
Re-use of the catalyst.—Upon re-use of the
catalyst after washing with cold water, the
activity dropped considerably (Fig. 3). How-
ever, after refluxing in water for a night, the
filtered and dried catalyst regained its original
activity. This indicates the presence of
strongly adsorbing species on the catalyst sur-
In this equation, the subscripts f, H2, s and p
denote -fructose, hydrogen, solvent and
D
products, respectively. Neglecting the adsorp-
tion of the solvent and upon working under a
constant hydrogen pressure, a constant value
1
face after reaction. The H NMR spectrum of
the filtrate showed that, besides
D
-fructose,
D
-mannitol and -glucitol, a mixture of other
D
polyhydroxy compounds was present. With
this method, we were not able to identify these
species. The concentrations were too low to be
1
3
analyzed with C NMR.
The catalyst was checked for leaching of
ruthenium. AAS measurements indicated
0
.575 mg Ru to be present in the reaction
solution (80 mL), which corresponds with a
0.03% loss of Ru from the catalyst surface.
This small amount of Ru was checked for its
catalytic activity by re-use of the reaction
Fig. 2. Hydrogenation of
different starting mixtures; 1 g
tose+0.5 g -mannitol+0.5 g
tose+1 g
D-fructose over Ru/C (Acros) with
D
-fructose (ꢀ), 1 g
D-gluctitol (×); 1 g
D
-fruc-
-fruc-
D
D
D
-glucitol ("); 0.050 g 5 wt% Ru/C (Acros), 80 mL
aqueous solutions, 72 °C, 1 bar H2.
mixture filtrate. No conversion of
D
-fructose
occurred in the absence of Ru/C.
Reacti6ity of the tautomeric forms.—In so-
lution, -fructose is present in the mutarota-
tion forms presented in Scheme 2. At 80 °C, a
-fructose solution contains 53% b-fructopy-
D
D
ranose, 2% a-fructopyranose, 32% b-fructo-
furanose, 10% a-fructofuranose and 3% open
keto-form [14]. The rate limiting step in the
mutarotation is the conversion of b-fructopy-
ranose into the keto form. The interconver-
sion between the furanose species is much
faster. Flood et al. [15] have reported that the
Fig. 3. First (ꢀ) use of Ru/C (Acros) and re-use after
washing with cold water (") and refluxing water (2); 1 g
-fructose, 0.050 g catalyst, 80 mL water, 72 °C, 1 bar H2.
D

A.W. Heinen et al. / Carbohydrate Research 328 (2000) 449–457
453
6
2% b-pyranose).
D
-Glucose does not react
under the mild conditions applied (70 °C, 1
bar), but apparently the
forms adsorb stronger on the catalyst than the
-fructose forms. Strong -glucose adsorption
and lower reactivity was also observed in the
hydrogenation of -glucose/ -fructose mix-
tures over a SiO supported copper catalyst
D-glucose tautomeric
D
D
D
D
2
[
8].
Fig. 5(a) compares the hydrogenation rates
of isomaltulose, 6-O-(a- -glucopyranosyl)-
fructose, and -fructose. The -fructose units
in isomaltulose (Scheme 3) react with the same
D
D
-
D
D
rate as
M substrate.
D-fructose, both with 0.034 and 0.069
Scheme 2.
D-Fructose mutarotation.
Isomaltulose can adsorb via the
D
-fructose
and via the glucoside unit. The adsorption
strength of the a-glucopyranoside unit can be
determined from competition experiments
with both
D
-fructose and methyl a- -glucopy-
D
ranoside. Fig. 5(b) indicates inhibition of
D
-
fructose hydrogenation by methyl glucoside,
resulting in a decrease in rate by a factor of
Fig. 4. Effect of
tose; 1 g -fructose (ꢀ), 1 g
.050 g Ru/C (Acros), 80 mL aqueous solution, 72 °C, 1 bar
H2.
D
-glucose on the hydrogenation of
D-fruc-
D
D
-fructose+1 g -glucose (ꢃ);
D
0
mutarotation rate constant in water at 70 °C
−
1
amounts to around 1.3 min . In our stan-
−
1
dard experiment with 0.069 mol L
D
-fruc-
tose, the mutarotation rate will, therefore, be
−
1
−1
around 2.4 mol L
than the rate of hydrogenation (see Table 2).
Interconversion between the different -fruc-
h
, which is much faster
D
tose species will therefore not be rate limiting.
These tautomeric forms will all have differ-
ent adsorptivities on the catalyst surface and
their own characteristic rates of hydrogena-
tion. Ruddlesden et al. [16] and Makkee et al.
[
8] showed that hydrogenation over Cu and
Ni catalysts occurs by hydrogenation of the
ring forms and not by hydrogenation of the
acyclic form. The furanose forms were found
to be more reactive than the pyranose forms.
Fig. 5. (a) Hydrogenation of
D-fructose (ꢀ/ꢃ) and isomal-
Upon addition of
D
-glucose to the reaction
tulose ("/2), solid symbols: 0.069 M substrate; open sym-
bols: 0.034 M substrate; (b) hydrogenation of 0.069 M
0
0
mixture (c =c ), we observed strong inhibi-
tion (Fig. 4). At 31 °C,
9
g
f
D
-fructose with (ꢀ) and without (ꢃ) 0.069 M methyl a-D-
D
-glucose is more than
glucopyranoside present. 0.050 g Ru/C (Acros), 80 mL
9% present in the pyranose forms (38% a-,
aqueous solution, 72 °C, 1 bar H2.

4
54
A.W. Heinen et al. / Carbohydrate Research 328 (2000) 449–457
impregnated with acetone, and Ru/SX1GN65
impregnated with water is twice as high than
on the other catalysts. Impregnation of SX1G
with ethanol also leads to some increase in the
activity, which indicates an effect of the sol-
vent–support interaction.
The results are in line with the observations
of Machek et al. [17]. These authors studied
the effect of the support-solvent interaction on
the Pt distribution and particle size. On acti-
vated carbon, better metal distributions and
smaller Pt particles were obtained by using
more hydrophobic solvents. In fact, the Pt
particle size on carbon decreased in the order
water\ethanol\acetone. For hydrophilic
supports, like alumina, polar solvents resulted
in higher dispersions. Using a solvent with a
low affinity for the support, the metal particles
will be non-uniformly distributed through the
pore system.
Scheme 3. The two forms of isomaltulose.
Table 3
Initial reaction rate (r ) of
D
-fructose hydrogenation on differ-
0
ent Ru/C catalysts reduced for 2 h at 400 °C in 20% H /N ; 1
D-fructose, 0.05 g Ru/C, 80 mL water, 72 °C, 1 bar H2
2
2
g
−
1
−1
Catalyst
r0 (mol L
h
)
Ru/Darco (water)
Ru/SX1G (water)
Ru/SX1G (ethanol)
Ru/SX1G (acetone)
Ru/SX1GN65 (water)
Ru/C (Acros)
0.031
0.033
0.038
0.067
0.067
0.033
The data of Table 3 suggest that low sol-
vent–support interactions lead to less active
catalysts. The selectivity did not change. Com-
parison of these results with the CO adsorp-
tion data indicates that part of the Ru of these
catalysts is not accessible to -fructose. There-
D
two. Therefore, the adsorption strength of the
glucoside unit of isomaltulose is comparable
with that of the fructofuranoside unit, which
means that only half of the available Ru sur-
face is used for adsorption of reactive species.
The equal hydrogenation rates of isomal-
fore, the performance of the catalyst can be
influenced by the catalyst preparation method.
The type of carbon support, however, does
not directly influence the hydrogenation rate.
Sn promotion.—A favourable effect of Sn
on the activity of Ru catalysts in the hydro-
genation of carbonyl bonds is frequently ob-
served and explained by Sn mediated
activation of the CꢀO bond, facilitating the
hydrogen transfer from adjacent Ru sites.
Upon hydrogenation of unsaturated carbonyl
compounds, this results in a selectivity change
towards the unsaturated alcohol [18–21].
tulose and
that on Ru/C one or both furanose forms are
the reactive species of -fructose. The pyra-
D
-fructose are a strong indication
D
nose form (about 50% of the total concentra-
tion) adsorbs with equal strength, without
reacting. To maintain the furanose/pyranose
ratio around 1, the mutarotation rate of the
-fructose must be higher than the hydro-
genation rate.
D
Although
D-fructose is not hydrogenated
Effect of catalyst preparation method.—
Table 3 shows the effect of the carbon sup-
port, the impregnation solvent and the
reduction method on the hydrogenation activ-
ity. Since the rate of the reaction increases
linearly with the catalyst concentration, the
reaction rate will depend on the Ru surface
available for adsorption.
via the open keto form, the effect of Sn pro-
motion on the selectivity and activity in this
reaction was studied (Table 4). The effect of
promotion with Bi is also displayed. Sn was
able to increase the selectivity to
D
-mannitol.
With SnPd/C and SnPt/C (2 wt% Sn and 2.5
wt% noble metal) selectivities above 60% were
obtained. The selectivity in this reaction is
independent of the conversion. The increase in
selectivity after promotion of Ru/C was less
than on SnPt/C and SnPd/C. Furthermore,
Although the ruthenium surface available
for CO is equal for most of the catalysts (see
Table 1), the initial reaction rate on Ru/SX1G

A.W. Heinen et al. / Carbohydrate Research 328 (2000) 449–457
455
for Ru and Pd the activity decreased dramati-
cally. Lower Sn loadings (up to 0.2 wt%)
resulted in higher activities, but in loss of
selectivity.
was explained by coordination of the furanose
forms to the surface and attack of a hydride
like species with inversion of configuration. In
this way, hydrogenation of the a form leads to
XRD of the SnPd/C and SnPt/C catalysts
showed only diffraction patterns of Pd and Pt,
respectively; SnRu/C also showed SnO dif-
fraction patterns. To ascertain whether the
action of tin is entirely heterogeneous, the Sn
concentration in the reaction mixture was
measured with AAS. It appeared that 0.8% of
the total amount of Sn present on the SnPt/C
D-glucitol, the b isomer gives
D-mannitol. In
contrast with this Cu catalyst, our Sn pro-
moted catalysts do not affect the selectivity in
the hydrogenation of isomaltulose. Perhaps
with this catalyst and under these conditions,
the pyranose form of
D-fructose is one of the
reacting species which can coordinate to Sn in
such a way that formation of one of the
isomers is preferred. However, more insight
into tin–fructose coordination is required.
In conclusion, we can say that the hydro-
4
+
catalyst was dissolved. The influence of Sn
2
+
and Sn
ions on the selectivity was checked
by addition of 7 mg of SnCl and SnCl ,
4
2
respectively, to the reaction mixture before
hydrogenation with Ru/C. The catalyst read-
ily deactivated, suggesting precipitation of Sn
species on the Ru surface. Only 36% selectiv-
genation of
glucitol can be conducted under very mild
conditions (70 °C, 1 bar H ) using Ru/C cata-
D
-fructose to
D
-mannitol and
D
-
2
lysts. Under these conditions, the furanose
forms are hydrogenated preferentially, fruc-
topyranose is less reactive. The adsorption
constants of fructopyranose, fructofuranose
and the hydrogenation products are of the
same order of magnitude, which results in
inhibition of the hydrogenation by fructopyra-
ity to
D-mannitol was obtained. It is, there-
fore, not likely that the selectivity increase is
due to homogeneous tin ions in solution.
These results suggest that Ru is covered by
Sn species using this preparation technique. A
detailed study of the preparation of Sn-pro-
nose,
D-mannitol and
D
-glucitol.
moted Ru/C catalysts in relation to the
mannitol selectivity is therefore recommended.
Hydrogenation of -sorbose, which leads to
-glucitol and -iditol, results in a similar
selectivity effect upon promotion of the Pt/C
catalyst with tin: the selectivity to -glucitol
D-
Although the carbon surface chemistry does
not directly influence the reaction rate, it influ-
ences the location of the Ru particles. A better
wettability, either by using a less polar im-
pregnation solvent or by using a more hy-
drophilic carbon, results in a Ru surface better
L
D
L
D
increased from 48 to 63%. The selectivity of
the isomaltulose hydrogenation, however, was
not affected by Sn.
accessible to
D
-fructose. The particle size,
however, is not strongly influenced by the
wettability. The Ru particles of the different
catalysts were below or just above the detec-
tion limit (0.5 nm) of the TEM microscope.
Tin appeared to be a good promoter for
In the hydrogenation of
D
-fructose, selectiv-
ities to
D
-mannitol over 60% were already
reported on Cu/SiO catalysts [8]. This effect
2
Table 4
high
D
-mannitol selectivities. SnPt/C and
Effect of promoters (1 wt%) on the conversion of
and selectivity to -mannitol after 20 h of reaction; 1 g
-fructose, 80 mL water, 0.050 g catalyst, 100 °C, 100 bar H2
D-fructose
SnPd/C increased the
D
-mannitol selectivity
D
from 42 to 62%. This promising result de-
serves further study.
D
Catalyst
Conversion
%)
Selectivity D-mannitol (%)
(
Ru/C
Pd/C
Pt/C
BiPt/C
SnPt/C
SnPd/C
SnRu/C
100 (2 h)
100 (7 h)
40
39
47
43
63
61
50
3. Experimental
85
90
80
40
75
Materials.—Two carbons were used: a
steam activated peat-based carbon (SX1G)
and a chemically activated wood-based carbon
(Darco-KBB), both donated by Norit. SX1G
(10 g) was refluxed in 150 mL of 65% HNO3

4
56
A.W. Heinen et al. / Carbohydrate Research 328 (2000) 449–457
for 1 h. After washing and drying at 80 °C,
the carbon denoted as SX1GN65 was
obtained.
days, the carbon was centrifuged, washed and
filtered over a 0.45 mm filter (Chromofil), and
titrated with 0.05 M HCl.
A 5 wt% Ru/C, a 5 wt% Pt/C catalyst and
An indication of the amount of carboxylic
acid sites was obtained by infrared spec-
troscopy. Infrared spectra were recorded with
a Perkin–Elmer spectrum 1000 FTIR spec-
trometer. KBr tablets were used containing 2
mg of carbon in 250 mg KBr. The spectra
were obtained by co-adding 20 spectra with a
methyl a- -glucopyranoside were purchased
D
from Acros Chimica N.V. (Geel, Belgium),
-fructose (anhydrous) and -glucose (anhy-
drous) were obtained from E. Merck KGaA
Darmstadt, Germany) and isomaltulose was
a gift from S u¨ dzucker. RuCl ·x H O (40 wt%
D
D
(
3
2
−
1
Ru) was a gift from Johnson Matthey (Hert-
fordshire, UK). Other Ru/C catalysts were
home-made (see below). A 2.5 wt% Pd/C cata-
lyst was a gift from Engelhard De Meern BV
resolution of 4 cm . The original spectra
were corrected for a curved baseline.
The metal loadings on the support were
measured by Mr J. Padmos, using X-ray
fluorescence spectroscopy, XRF (Philips
PW1480). The Ru particles were studied with
TEM using a Philips CM 30 T electron micro-
scope, combined with energy dispersive analy-
sis of X-rays (EDX). CO adsorption
measurements were performed by the pulse
(
De Meern, The Netherlands). This catalyst
was based on Norit SX1G.
Catalyst preparation.—The Ru/C catalysts
were prepared by incipient wetness impregna-
tion of the carbon (0.95 g) with a solution of
1
10 mg RuCl ·2.5 H O in 2.0 mL solvent.
3
2
Three solvents were used: water, EtOH and
acetone. After drying overnight at ambient
conditions and 3 h at 80 °C, the catalysts were
reduced in a 10% H /N flow for 2 h at
method at rt, after reduction in 15% H /Ar at
2
100 °C.
Hydrogenation experiments.—The hydro-
genation experiments at elevated pressure
were performed in a Parr 4842 autoclave
made of Hastelloy C276. A 50 mL aq suspen-
sion of the catalyst was introduced into the
autoclave. The system was flushed with nitro-
gen and hydrogen and heated to the desired
2
2
4
00 °C. Before reduction, water was removed
at 120 °C (1 h) under a nitrogen flow. After
reduction, the catalysts were cooled down to
room temperature (rt) under N atmosphere
2
and then the oxygen concentration was slowly
(1% per 5 min) raised to 20% O /N .
temperature under 10 bar H . After that, a 30
2
2
2
Catalyst promotion.—Tin-promoted cata-
lysts were prepared by deposition of 1% Sn on
mL solution of
D
-fructose in water was intro-
duced in the system and the H pressure was
2
5
wt% Ru/C, and on commercial 2.5% Pd/C
adjusted to the desired value.
and 5% Pt/C catalysts by the following proce-
The reactions at 1 bar H pressure were
2
dure. Under a N atmosphere, 30 mL of an aq
performed in a glass batch reactor of 200 mL,
equipped with a gas-tight stirrer and a ther-
mostatic bath. Prior to each experiment, the
system, including a 50 mL aq suspension of
2
soln of SnCl ·5 H O (138 mg) was slowly (0.8
4
2
−
1
mL min ) added to the catalyst suspension
(3.8 g/50 mL water), pre-reduced under H2.
The pH of the suspension was increased to 8.5
using 1 M NaOH. The suspension obtained
was filtered off, the catalyst was washed until
neutral and dried in air, first one night at rt
and then 3 h at 120 °C. A Bi-promoted Pt/C
catalyst was prepared similarly using
the catalyst, was flushed with N . After that,
2
the system was heated to the desired tempera-
ture and the catalyst was pre-reduced by flush-
ing the system with H for 30 min. The
2
reaction was started by adding
D-fructose.
Unless stated otherwise, the following con-
BiONO [22].
ditions were applied: 72 °C, 1 bar H , 1500
3
2
Catalyst characterization.—The amount of
acid sites on the carbon surface was deter-
mined by selective neutralization with NaOH,
according to the method of Boehm [23]. To
rpm, 0.07 M
D
-fructose (1 g in 80 mL of
water) and 0.05 g of Ru/C catalyst.
Samples taken during the reaction were an-
alyzed by high-performance liquid chromatog-
raphy (HPLC) using a Millipore-Waters 590
pump and a 300×7.8 mm cation exchange
1
00 mg of carbon, 10 mL of 0.05 M NaOH
was added. After shaking the suspension for 4

A.W. Heinen et al. / Carbohydrate Research 328 (2000) 449–457
457
2
+
[5] P. Gallezot, M. Besson, Carbohydr. Eur., 13 (1995)
–10.
6] F. Jo o´ , Z. T o´ th, M.T. Beck, Inorg. Chim. Acta, 25
1977) L61–L62.
[7] S. Kolaric, V. S
89–193.
column in the Ca form (Phenomenex), con-
nected to a refractive index detector (Shodex
RI SE-51). The samples were eluted with de-
5
[
(
−
1
gassed water at a flow rate of 0.6 mL min
and a column temperature of 80 °C.
&
unjic, J. Mol. Catal. A, 110 (1996)
1
[8] M. Makkee, A.P.G. Kieboom, H. van Bekkum, Carbo-
hydr. Res., 138 (1985) 225–236.
[
9] A.W. Heinen, J.A. Peters, H. van Bekkum, Appl. Catal.,
194–195 (2000) 193–202.
Acknowledgements
[
[
10] J.A. Dean, Lange’s Handbook of Chemistry, McGraw-
Hill, New York, 1992, pp. 4–16.
11] L.W. Wright, Chemtech, (1974) 42–46.
The investigation was carried out with the
support of the Innovation Oriented Program
on Catalysis (‘IOP katalyse’). Dr P.J. Kooy-
man of the National Centre of High Resolu-
tion Electron Microscopy, Delft University of
Technology is acknowledged for performing
the transmission electron microscopy investi-
gations. J. Padmos is acknowledged for the
XRF measurements. Thanks are also due to
Engelhard de Meern BV, for performing the
CO adsorption measurements, S u¨ dzucker for
the generous gift of isomaltulose and Norit
NV for providing the activated carbon.
[12] F. Kapteijn, J.A. Moulijn, R.A. van Santen, in J.A.
Moulijn, P.W.N.M. van Leeuwen, R.A. van Santen
(
Eds.), Catalysis. An integrated Approach to Homoge-
neous, Heterogeneous and Industrial Catalysis, Elsevier,
Amsterdam, 1993, pp. 69–86.
[13] P.G.J. Koopman, Ph.D. Thesis, Delft University of
Technology, The Netherlands, 1980.
[
14] S.J. Angyal, Ad6. Carbohydr. Chem. Biochem., 42 (1984)
5–68.
[15] A.E. Flood, M.R. Johns, E.T. White, Carbohydr. Res.,
88 (1996) 45–56.
1
2
[
16] J.F. Ruddlesden, A. Steward, D.J. Thompson, R. Whe-
lan, Faraday Discuss. Chem. Soc., 72 (1981) 397–411.
17] V. Machek, J. Hanika, K. Sporka, V. Ruzicha, J. Kunz,
L. Janacek, in G. Poncelet, P. Grange, P.A. Jacobs
[
(
Eds.), Preparation of Catalysts III, Elsevier, Amster-
dam, 1983, pp. 69–80.
[
18] S. Galvagno, A. Donato, G. Neri, R. Pietropaolo, G.
Capannelli, J. Mol. Catal., 78 (1993) 227–236.
19] L. Stievano, S. Calogero, F.E. Wagner, S. Galvagno, C.
Milone, J. Phys. Chem. B, 44 (1999) 9545–9556.
References
[
[
1] For a review on
A.P.G. Kieboom, H. van Bekkum, Starch, 37 (1985)
36–141.
2] Chemical Market Reporter, September 13 (1999) 46–53.
3] M. Makkee, A.P.G. Kieboom, H. van Bekkum, J.A.
Roels, J. Chem. Soc., Chem. Commun., (1980) 930–931.
4] M. Makkee, A.P.G. Kieboom, H. van Bekkum, Starch,
D-mannitol synthesis, see: M. Makkee,
[20] Y. Pouilloux, F. Autin, C. Guimon, J. Barrault, J.
Catal., 176 (1998) 215–224.
[21] V.M. Despehande, K. Ramnarayan, C.S. Narasimhan,
J. Catal., 121 (1990) 174–182.
[22] A.W. Heinen, J.A. Peters, H. van Bekkum, Carbohydr.
Res., 304 (1997) 155–164.
1
[
[
[
37 (1985) 232–241.
[23] H.P. Boehm, Ad6. Catal., 16 (1966) 179–274.
.