Platinum catalysts modified by adsorbed amines: A new method of enhancing rate and selectivity of L-sorbose oxidation

Article abstract of DOI:10.1006/jcat.1996.0234

The Pt-catalyzed oxidation at C-1 of unprotected L-sorbose to 2-keto-L-gulonic acid, an intermediate in vitamin C synthesis, has been studied using molecular oxygen and neutral aqueous solution. The performance of Pt/alumina and Pt/C catalysts could be improved by addition of small quantities of tetraalkylammonium hydroxides, trialkyl-, triaryl-, and cycloaliphatic amines. The optimum modifier : Pt_s molar ratio is around 0.1, which corresponds to an amine : sorbose molar ratio of 1 : 1700. Rate enhancement (by a factor of up to 4.6) can be obtained when the pK_a is around 10 or higher, and this correlation depends very little on the chemical structure of the amine. Rate acceleration is proposed to be connected with the hydration of the intermediate aldehyde (bifunctional catalysis). The influence of supported N-bases on the selectivity of Pt is a function of their chemical structure. The best results, 95% selectivity at 30% conversion and about 40% selectivity enhancement in the whole investigated conversion range, was obtained with Pt/C and hexamethylenetetramine. Electrochemical model studies revealed that hexamethylenetetramine is adsorbed on Pt and not oxidized during reaction. Molecular modeling suggests that the preferential oxidation at C-1 is due to complex formation by H-bonding between hexamethylenetetramine and sorbose. The adsorption of this complex on Pt results in a tilted position of the reactant in which only C-1 is exposed to oxidative dehydrogenation.

Full text of DOI:10.1006/jcat.1996.0234

JOURNAL OF CATALYSIS 161, 720–729 (1996)
ARTICLE NO. 0234
Platinum Catalysts Modified by Adsorbed Amines: A New Method of
Enhancing Rate and Selectivity of L-Sorbose Oxidation
C. Bro¨nnimann, Z. Bodnar, R. Aeschimann, T. Mallat, and A. Baiker
Department of Chemical Engineering and Industrial Chemistry, Swiss Federal Institute of Technology, ETH-Zentrum, CH-8092, Zu¨rich, Switzerland
Received November 6, 1995; revised February 9, 1996; accepted February 29, 1996
The reaction is based on the unique properties of Pt met-
als: they can activate molecular oxygen and alcohols un-
der very mild conditions. Excellent selectivities have been
obtained in the partial oxidation of several thermolabile
reactants, such as carbohydrates and steroids, under very
mild conditions (for reviews see Refs. 4–9). The reactivity
of the hydroxyl groups of polyols can be tuned by promot-
ing Pt or Pd with Bi, Pb, or some other metals, metal oxides,
or metal phosphates (9–13). A disadvantage of this aerobic
oxidationmethodisthefrequentlyobservedcatalystdeacti-
vation which necessitates high catalyst loadings, sometimes
even after promotion with heavy metals (14–17).
Unfortunately, not only the reaction rate but the selec-
tivity is also moderate in the direct oxidation of L-sorbose
to 2-KLG (3, 17–19). After screening a broad range of sup-
ported Pt and Pd catalysts, a maximum of 67% selectivity at
58% conversion has been obtained with a 5 wt% Pt/alumina
(19). Promotion with Bi or Pb improves the initial rate but
has a detrimental effect on the selectivity. It has been shown
that the primary reason for poor catalyst performance is the
formation of strongly adsorbed by-products (self-poisoning
of Pt, (19, 20)). The increasing contamination of active sites
results in overoxidation (too high oxygen coverage of Pt).
Partial corrosion of Pt and promoter metals in the neu-
tral aqueous solution in the presence of the good chelating
agent 2-KLG has also been proved.
The Pt-catalyzed oxidation at C-1 of unprotected L-sorbose to
2-keto-L-gulonic acid, an intermediate in vitamin C synthesis, has
been studied using molecular oxygen and neutral aqueous solu-
tion. The performance of Pt/alumina and Pt/C catalysts could be
improved by addition of small quantities of tetraalkylammonium
hydroxides, trialkyl-, triaryl-, and cycloaliphatic amines. The op-
timum modifier : Pts molar ratio is around 0.1, which corresponds
to an amine : sorbose molar ratio of 1 : 1700. Rate enhancement (by
a factor of up to 4.6) can be obtained when the pKa is around 10
or higher, and this correlation depends very little on the chemical
structureoftheamine. Rateaccelerationisproposedtobeconnected
with the hydration of the intermediate aldehyde (bifunctional cata-
lysis). The influence of supported N-bases on the selectivity of Pt is a
functionoftheirchemicalstructure. Thebestresults, 95%selectivity
at 30% conversion and about 40% selectivity enhancement in the
whole investigated conversion range, was obtained with Pt/C and
hexamethylenetetramine. Electrochemical model studies revealed
that hexamethylenetetramine is adsorbed on Pt and not oxidized
during reaction. Molecular modeling suggests that the preferential
oxidation at C-1 is due to complex formation by H-bonding between
hexamethylenetetramine and sorbose. The adsorption of this com-
plex on Pt results in a tilted position of the reactant in which only
c
C-1 is exposed to oxidative dehydrogenation.
ꢀ 1996 Academic Press, Inc.
INTRODUCTION
It has been shown recently (21) that the pH of the aque-
ous medium is the most important parameter in the direct
The industrial production of vitamin C is still based on
the Reichstein–Gruessner synthesis (1, 2) developed in
1934. In the conventional route the oxidation of the C-
1 hydroxyl group of L-sorbose to the corresponding car-
boxylic acid proceeds in three steps. The method requires
the protection (and subsequent hydrolytic deprotection) of
the other four reactive hydroxyl groups to achieve good
selectivity.
A considerable effort has been expended in the past
in order to substitute the conventional stoichiometric oxi-
dation (with NaOCl or KMnO4) by a catalytic method
and perform the transformation of L-sorbose to 2-keto-L-
gulonic acid (2-KLG) in one step. Heyns (3, 4) reported first
the Pt-catalyzed aqueous phase oxidation of unprotected
L-sorbose with molecular oxygen according to Scheme I.
SCHEME I
720
0021-9517/96 $18.00
c
Copyright ꢀ 1996 by Academic Press, Inc.
All rights of reproduction in any form reserved.

OXIDATION OF L-SORBOSE
721
oxidation of L-sorbose. At low pH the selectivity is good
TABLE 2
but catalyst deactivation prevents high conversion. With
increasing pH (above 7.5) the rate is enhanced and the se-
lectivity drops, mainly due to the poor stability of 2-KLG
even in the absence of catalyst and oxygen. To overcome
this limitation we used an analogy with asymmetric cata-
lysis. The most successful technique in heterogeneously
catalyzed enantioselective hydrogenations is the modifica-
tion of the active metal (Ni or Pt) by traces of strongly ad-
sorbing chiral compounds (22, 23). We assumed that the
performance of Pt in sorbose oxidation can also be im-
proved by strongly adsorbing molecules, without changing
the bulk characteristics of the medium. Promising candi-
dates for modifiers are soft bases, such as tertiary amines,
due to the soft acidity of Pt.
In a preliminary report we presented the first success
of this technique (21). The aim of the present work is to
study the correlation between the chemical structure and
basicity of various N-bases and their influence on the rate
andselectivityofsorboseoxidation. Electrochemicalmodel
experiments and molecular modeling studies were used to
elucidate the nature of Pt–reactant–modifier interactions
during the oxidation process.
Influence of the Molar Ratio of Modifier to Surface Pt Atoms
on the Average Reaction Rate and Selectivity to 2-KLG; 5 wt%
Pt/A12O3 (A), Standard Conditions
Modifier/Pts
(mol/mol)
Sa
%
Rateb
Modifier
mmol (gcat hꢂ1
)
—
—
—
89
94
2.64
2.54
6
0.1
6
6
0.4
1.0
62
—
1.55
<0.1
11
11
11
0.1
0.5
1.0
92
76
43
5.2
3.41
2.8
a Selectivity at 10% conversion.
b Average rate between 0 and 30% conversion.
bles 2 and 3. Tetramethylammonium hydroxide (1, 25%
in water, Fluka), tetrabutylammonium hydroxide (2, 40%
in water, Fluka), triethylamine (3, >99.5%, Fluka), trib-
utylamine (4, >99%, Fluka), triphenylamine (5, >98%,
Fluka), pyridine (6, >99%, J. T. Baker), quinoline (7,
>95%, Fluka), acridine (8, >95%, Fluka), pyrazine (9,
>98%, Fluka), quinoxaline (10, 97%, Fluka), quinucli-
dine (11, >98%, Fluka), diazabicyclooctane (12, 97%,
Fluka) hexamethylenetetramine (13, >99.5%, Fluka),
2-propanol (>99.5%, Riedel–de Hae¨n), and L-sorbose
(ꢃ99%, Fluka) were used as delivered. Distilled water (af-
ter ion exchange) was used for the experiments.
EXPERIMENTAL
Materials
Data and codes of supported Pt catalysts used for the
oxidation reaction are collected in Table 1.
A Pt powder catalyst, used for electrochemical model
experiments, was prepared by hydrogen reduction of the
corresponding hydroxide at 30ꢁC. The hydroxide was pre-
cipitated from an aqueous H2PtCl6 solution with a dilute
NaHCO3 solution at 90ꢁC. The catalyst was filtered off,
washed with water to remove residual Clꢂ ions, and dried
at 105ꢁC. A metal dispersion of 0.052 was determined from Catalyst Modification
the hydrogen region of an electrochemical polarization
curve (24).
In situ catalyst modification was applied with water-
soluble amines and performed in the same glass reactor
used for sorbose oxidation (see below). A quantity of 0.6 g
catalyst was suspended in 40 ml water at ambient tempera-
ture and the system was flushed extensively with nitrogen.
After 20 min reduction of the catalyst with hydrogen at at-
mospheric pressure, hydrogen was substituted by nitrogen,
the proper amount of modifier, dissolved in 2 ml water, was
added and the slurry was mixed for at least 90 min.
In the case of water-insoluble amines, the modification
was performed ex situ. The procedure was similar to that
described above, but 0.8 g catalyst was suspended in 50 ml
isopropyl alcohol and the modifier was dissolved in 2 ml
isopropyl alcohol. After 120 min mixing, the solvent was
evaporated and the catalyst was dried in vacuum at 50ꢁC
for 20 min. A weight of 0.6 g of dried catalyst was used for
the oxidation reaction.
Compounds used are designated as numbers and the
corresponding structural formulae are presented in Ta-
TABLE 1
Catalysts Used for L-Sorbose Oxidation
Dispersion
by
by H2-
Catalystsa Composition
Origin
TEM chemisorption
A
B
C
D
Pt/alumina Engelhard Escat 24 0.30
0.24
0.22
0.29
0.31
Pt/alumina Engelhard No. 4759
—
—
—
Pt/C
Pt/C
Engelhard Escat 21
Engelhard Escat 23
a Pt loading of all catalysts is 5 wt%.

¨
BRONNIMANN ET AL.
722
M Na2CO3. Samples of 1 ml of the reaction mixture were
TABLE 3
periodically taken, filtered, and stored at 5ꢁC. Standard
conditions (pH 7.3, T = 50ꢁC, P(O2) = 1 bar, n = 1100 rpm;
catalyst, L-sorbose weight ratio of 0.27; modifier, PtS = 0.1
molar ratio) were applied if not otherwise stated. Under
these conditions the reactor worked in a transport limited
regime.
The samples were analyzed by HPLC using a Waters
liquid chromatograph (pump, 600 E; autosampler, 717;
UV detector, 486; RI detector, 410; and integrator, 746).
The separation was carried out at 30ꢁC on a 250 ꢅ 4-mm
stainless-steel column slurry-packed with Nucleosil-5 NH2
resin (Macherey Nagel). The eluent consisted of 4.5 g
KH2PO4/330 ml water/670 ml acetonitrile with a flow rate
of 0.8 ml minꢂ1 and a column pressure of 90 bar. Peak de-
tection was performed with a refractive index (RI) detec-
tor thermostated at 35ꢁC. Conversion and selectivity were
obtained by using the external standard method. Concen-
trations were corrected for dilution during pH-controlling
and for the effect of sampling.
Average Reaction Rate and Selectivity Using 5 wt% Pt/Al2O3 (A)
Modified with Amines; Standard Conditions, Modifier: Pts = 0.1
Sa
(%)
Rateb
Modifier
pKa
mmol (gcat hꢂ1
)
—
1
2
3
4
—
—
71
28
42
58
62
44
2.64
12.2
9.7
6.64
5.8
Me4N+ OHꢂ ca. 14
Bu4N+ OHꢂ ca. 14
Et3N
Bu3N
Ph3N
10.7c
9.9c
5
4.8d
6.81
6
7
5.3c
72
71
75
73
64
2.54
1.69
1.49
1.53
1.64
4.9c
8
5.6c
9
0.65e
0.56e
10
11
12
13
10.9c
8.7f
62
69
81
5.2
Electrochemical Polarization
All potentials in the paper are referred to the reversible
hydrogen electrode emerging in the same supporting elec-
trolyte which was in the main compartment of the cell. Ni-
trogen (99.995%) was bubbled continuously through the
cell to deoxygenate the solution.
The electrochemical cell, the modified carbon paste elec-
trode and the experimental setup used for cyclic voltam-
metric measurements have been described previously (24,
28). Pt catalyst powder (2 mg) was dispersed on the sur-
face of the flat carbon paste and polarized with a sweep
rate of 2 mV sꢂ1. The speed of the rotating carbon paste
electrode was 500 minꢂ1. A Pt rod counter electrode and
an Ag/AgCl/KClsat (E = 197 mV) reference electrode were
2.27
1.23
8.87g
a Selectivity at 30% conversion.
b Average rate between 0 and 30% conversion.
c See Ref. (33)
d See Ref (37).
e See Ref. (34).
f See Ref. (35).
g See Ref. (36).
Oxidation of L-Sorbose
The oxidation reactions were performed in a 200-ml flat- used. The latter was separated from the main compartment
bottomed glass batch reactor (diameter 55 mm) equipped of the cell by two diaphragms to avoid the pollution of the
with mechanical stirrer (diameter 28 mm), gas distribu- electrolyte by Clꢂ ions.
tor, condenser, combined temperature and pH electrode
Polarizations were carried out in a nitrogen atmosphere
(Metrohm, 6.02181.010), and combined reference and mea- at 25ꢁC in 40 ml 1 wt% aqueous Na-acetate supporting
suring electrode (Metrohm, 6.0415.100). The reproducibil- electrolyte (pH adjusted to 7.3). At first, the catalyst pow-
ity of the potential measurement was ꢄ10 mV. More details der was reduced and oxidized electrochemically between
of the method can be found elsewhere (25–27). The tem- 25 and 1025 mV. This first cycle is not shown in the figures.
perature was controlled with a water bath within ꢄ1ꢁC. The second cycle provides the reference (unmodified Pt).
The oxygen flow rate (5 ml minꢂ1) was measured by a During the third cathodic polarization 1 ml supporting elec-
rotameter.
trolyte containing the proper amount of amine was injected
A weight of 0.6 g of ex situ modified catalyst was sus- before hydrogen adsorption started, as illustrated in Fig. 1.
pended in 40 ml water under nitrogen and heated to 50ꢁC. (At this point both oxygen and hydrogen coverage on Pt
In case of in situ modification the slurry was heated under is negligible.) For the sake of simplicity, only the anodic
nitrogen to 50ꢁC directly after catalyst preparation. Then sweeps of the second and third cyclic voltammograms are
2.2 g L-sorbose was added and the pH adjusted to 7.3 by compared in Figs. 6 and 7. A new portion of catalyst was
dosing 0.4 M aqueous Na2CO3. The oxidation reaction was used for each series of experiments.
started by substituting nitrogen with oxygen. During reac-
The degree of coverage of Pt by amines (ꢀ) has been
tion the pH was controlled by automatic titration with 0.4 determined from the hydrogen region of the anodic sweeps

OXIDATION OF L-SORBOSE
723
13) and L-sorbose were used as starting coordinates for the
minimizations.
RESULTS
Influence of Modifier: Pts Ratio
In preliminary experiments the catalytic performance of
amine-modified catalysts was studied as a function of the
molar ratio of amine to surface Pt atoms (Pts). Two exam-
ples, the influence of an aromatic (6, pyridine) and a cy-
cloaliphatic (11, quinuclidine) amine are shown in Table 2.
In general, high modifier loadings in the range of 0.1–1.0
result in a catalytic behavior similar to that of a strongly de-
activated Pt/alumina catalyst, likely to be due to a too-high
coverage of active sites by the modifier. (The adsorption of
amines on Pt will be discussed later.) The best selectivities
and reaction rates were obtained using rather low amounts
of amine. Accordingly, the influence of the chemical struc-
ture of various amines on the performance of Pt/alumina
was investigated with a molar ratio of modifier to Pts of
0.1–0.2.
FIG. 1. Third cyclic voltammogram of Pt during which 12 mg pyridine
(6) is injected into the solution; 1 wt% aqueous Na-acetate (pH adjusted
to 7.3); sweep rate, 2 mV sꢂ1
.
of the second and third cycles,
ꢀ = (NH,0 ꢂ NH,A)/NH,0
Rate Enhancement by Modification with Amines
,
The reaction rate has been determined as the average
rate of L-sorbose consumption between 0 and 30% conver-
sion. This value is considered only as a qualitative measure
of the activity of the catalyst. The reactor worked in a trans-
port limited regime; i.e., both the rate of the surface oxida-
tion reaction and the rate of oxygen transport from the gas
phase to the catalyst surface had an influence on the rate
of sorbose consumption. These conditions were necessary
to avoid rapid catalyst deactivation due to overoxidation of
active sites, in agreement with earlier observations (12, 19,
31, 32).
where NH,0 and NH,A are the amounts of adsorbed hydro-
gen in the absence and presence of an amine, respectively
(29, 30).
For the determination of the rate of electrocatalytic oxi-
dation of L-sorbose, 1 ml supporting electrolyte containing
0.5 wt% sorbose was added at the end of the second po-
larization cycle. The reaction rate was calculated from the
difference between the anodic sweeps of the second and
third cycles in the potential range of 450–900 mV.
Hydrogen Chemisorption
The influence of various tetraalkylammonium hydrox-
ides and tertiary amines as modifiers on the rate of sorbose
consumption and selectivity to 2-KLG is summarized in
Table 3. (Primary and secondary amines were not used to
avoid any possible interaction between the amine and the
carbonyl compound reactant or product.) Unfortunately,
there is a considerable deviation among the pKa values
of certain amines reported in the literature (33–37), which
weakens any conclusion concerning the correlation of reac-
tion rate with pKa. In addition, no pKa values were found for
tetraalkylammonium hydroxides and therefore they were
estimated as being approximately 14.
Hydrogen chemisorption was measured in an all-glass
system equipped with an oil diffusion pump. The hydro-
gen used was 99.999% purity. About 0.5 g of sample was
first evacuated at 150ꢁC for 2 h and then reduced at 200ꢁC
with hydrogen for 3 h. Afterward the samples were de-
gassed at 200ꢁC overnight. Adsorption was carried out at
25 ꢄ 1ꢁC starting with an initial pressure of ca. 5 ꢅ l03 Pa,
and the system was allowed to equilibrate for 2 h. After the
first isotherm was measured, the samples were evacuated
to l0ꢂ3 Pa for 2 h and the physisorption isotherm was then
measured, also at 25 ꢄ 1ꢁC.
Modification of alumina-supported Pt with strongly basic
tetraalkylammonium hydroxide provides significant rate
Molecular Modeling
Biosym programs Insight II (version 2.3.5) and Discover enhancement. For example, the use of 1 or 2 as modifiers re-
(version 2.96) were used in the molecular modeling study. sults in a 3.7- to 4.6-fold higher rate of sorbose consumption,
Molecular mechanics minimum energy calculations were compared to the rate obtained with unmodified Pt/alumina
performed using the CVFF force field and the default val- under otherwise identical conditions. Similar rate accel-
ues as supplied by the software. X-ray data from the Cam- eration, but to a less extent, was observed using trialkyl-
bridge Structural Database (CSD) for the modifiers (11, 12, and triarylamines 3–5 (pKa between 5 and 11). Aromatic

¨
BRONNIMANN ET AL.
724
The oxidation of L-sorbose on Pt/alumina (A, Table 1)
modified with aromatic amines is characterized by a 5–10%
increase in selectivity to 2-KLG up to 25–30% conversion,
compared to the unmodified reaction. The change of se-
lectivity with conversion in the presence of pyridine (6),
quinoline (7), and acridine (8) as modifiers is illustrated in
Fig. 3. The decrease of reaction rate with increasing number
of benzene rings of the aromatic amines (6–8 in Table 3) in-
dicates an increasing “geometric blocking” of the Pt surface
by the modifier. Unfortunately, this site blocking has only a
minor positive effect on the selectivity. A similar but much
more significant influence of the decrease of the size of ac-
tivesiteensemblesbyBiadatomshasalreadybeenreported
for alcohol oxidation (25, 38). A likely explanation for the
minor blocking effect in the case of aromatic amines is that
they can “move” relatively easily on the Pt surface and can-
not efficiently decrease the size of active site ensembles.
The heteroaromatic amines studied have a planar
two-dimensional structure. The possible influence of a
three-dimensional-like structure was investigated using cy-
cloaliphatic amines. Figure 4 shows the catalytic perfor-
mance of Pt/alumina modified with quinuclidine (11), diaz-
abicyclooctane (12), and hexamethylenetetramine (13). A
significant improvement of selectivity over the whole inves-
tigated conversion range was achieved only with 13. High
selectivity of 97% at 20% conversion, compared to 81% for
the unmodified reaction, was obtained. The decrease of se-
lectivity above 20% conversion is attributed to consecutive
side reactions (product degradation) which are not affected
by the modifier. Another possible explanation is the partial
protonation of 13 by the product 2-KLG or by acidic by-
products, which weakens the adsorption of the amine on
FIG. 2. Influence of (᭹) bulk pH and (᭺) basicity of tri- and tetraalky-
lamine modifiers on the average reaction rate and selectivity to 2-KLG in
the oxidation of L-sorbose; 5 wt% Pt/Al2O3 (A), modifier to Pts molar ratio
of 0.2, standard conditions.
amines 7–10 (pKa between 0.5 and 5.5) have a negative
effect on the reaction rate, except when pyridine (6) was
used as modifier. Interesting results were obtained with
cycloaliphatic amines 11–13. Quinuclidine (11) provided a
2-fold rate acceleration, whereas hexamethylenetetramine
(13; pKa lower by 2) decreased the average rate by the
same degree. As a first approximation we propose that
rate enhancement can be expected only when the pKa value
of the amine is around 10 or higher (except for the triary-
lamine, 5).
Influence of Amines on the Selectivity
Figure 2 demonstrates the advantage of using supported
N-bases instead of increasing the pH of the aqueous so-
lution. Here the rate of sorbose oxidation and the selecti-
vity to 2-KLG is shown as a function of (i) the basicity of
some tri- and tetraalkylamines (working at a bulk pH of
7.3) and (ii) the pH of the aqueous solution using unmodi-
fied Pt/alumina. The rate of sorbose consumption can be
increased by increasing the bulk pH, but this results in a
drop in selectivity to 2-KLG. Simple tetraalkylammonium
hydroxides and trialkylamines provide rather poor selectiv-
ities, compared to those measured in the presence of aro-
maticandcycloaliphaticN-compounds(Table3). Neverthe-
less, significantly higher selectivities were obtained at the
same reaction rate when using catalytic amounts of trialky-
lamine or tetraalkylammonium hydroxide modifiers, com-
pared to the case when the bulk pH was increased.
In the following part the discussion will be focused on
the use of aromatic and cycloaliphatic amines, as our main
purpose was to achieve the highest possible selectivity even
at the expense of moderate oxidation rate.
FIG. 3. Selectivity to 2-KLG as a function of sorbose conversion us-
ing ( ) 5 wt% Pt/Al2O3 (A) without modification or modified with (
)
pyridine (6), (᭺) quinoline (7), and (᭡) acridine (8); modifier to Pts molar
ratio of 0.1, standard conditions.

OXIDATION OF L-SORBOSE
725
FIG. 5. Selectivity to 2-KLG as a function of sorbose conversion using
(᭹) unmodified and (᭺) hexamethylenetetramine (13)-modified 5 wt%
Pt/C (C); modifier to Pts molar ratio of 0.1, standard conditions.
FIG. 4. Selectivity to 2-KLG as a function of sorbose conversion us-
ing (᭺) 5 wt% Pt/Al2O3 without modification (A) or modified with (᭹)
hexamethylenetetramine (13), (᭜) quinuclidine (11), and (᭡) diazabicy-
clooctane (12); modifier to Pts molar ratio of 0.1, standard conditions.
tivity changed significantly. The best result, almost quan-
titative conversion of L-sorbose to 2-KLG at low conver-
sion, was obtained with a Pt/C catalyst (C, Table 1). This
catalyst provided rather poor selectivity without modifier,
as illustrated in Fig. 5. Modification with 13 resulted in an
average selectivity enhancement by 40% in the whole in-
vestigated conversion range. Interestingly, the outstanding
performance of the 5 wt% Pt/C catalyst C is limited to mod-
ification with hexamethylenetetramine (13). When using
modifiers 7, 8, 9, 11, or 12, the selectivities at 30% conver-
sion were significantly below those values obtained with the
similarly modified 5 wt% Pt/alumina catalyst A.
Pt. At conversions higher than 30% the reaction became
rather slow due to catalyst deactivation.
Comparison of Alumina- and Carbon-Supported
Pt Catalysts
An attempt was made to improve selectivity and avoid
catalyst deactivation by using other types of Pt catalysts.
Typical results obtained with some alumina- and carbon-
supportedcatalystsmodifiedwithhexamethylenetetramine
(13) are shown in Table 4. The molar ratio of 13 to
L-sorbose is 1 : 1700. In general, carbon-supported catalysts
possess significantly higheractivitythanalumina-supported
Pt. Pt/alumina (A) was used in previous studies as this cata-
lyst afforded the highest selectivity with acceptable rate of
sorbose oxidation (19). Unexpectedly, after modification
with 13 the order of the catalysts with respect to their selec-
Electrochemical Model Experiments
The real nature of the Pt–amine–sorbose interaction is
not clear from the above results. Is the action of amines lim-
ited to base catalysis or do they also take part in the surface
oxidation process? Electrochemical model experiments us-
ingpolycrystallinePtwereperformedtorevealsome details
of the role of amines. The use of unsupported Pt was nec-
essary due to the conductivity requirements of the method.
An aqueous solution with a pH of 7.3 was used, similarly
to the catalytic measurements. Sodium acetate was applied
as conductive salt to mimic the salts of 2-KLG (and other
carboxylic acid by-products formed by chain-breaking (19)
which are present during the oxidation reaction).
TABLE 4
Oxidation of L-Sorbose to 2-KLG over Various Supported Pt
Catalysts Modified with Hexamethylenetetramine (13); Standard
Conditions
Ratec
Catalyst
Compositiona
Sb (%)
mmol (gcat hꢂ1
)
Four amines, tetramethylammonium hydroxide (1), tri-
ethylamine (3), pyridine (6), and hexamethylenetetramine
(13) were chosen to represent the structurally different
types of amines used for catalyst modification (Table 3).
Figure 6 illustrates the influence of 6 on the hydrogen
and oxygen sorption (as OHad in the neutral solution)
characteristics of Pt. The shoulder and maximum on the
A
B
C
D
Pt/alumina
Pt/alumina
Pt/C
81
29
95
47
1.23
0.82
4.56
5.95
Pt/C
a Pt loading of all catalysts 5 wt%.
b Selectivity at 30% conversion.
c Rate of L-sorbose consumption between 0 and 30% conversion.

¨
BRONNIMANN ET AL.
726
FIG. 6. Positive sweeps of the cyclic voltammograms of Pt in the ab-
FIG. 7. Positive sweeps of the cyclic voltammograms of Pt (a) in the
absence and (b) in the presence of triethylamine (3); 1 wt% aqueous Na-
acetate (pH adjusted to 7.3), 2 mg Pt, 2 mV sꢂ1 sweep rate.
sence(a)andinthepresenceof(b)0.12mgor(c)1.2mgpyridine(6);1wt%
aqueous Na-acetate (pH adjusted to 7.3), 2 mg Pt, 2 mV sꢂ1 sweep rate.
amine modifier is adsorbed on the Pt surface and changes
the adsorption characteristics of the active metal, and (ii)
triethylamine is oxidized under reaction conditions but the
other three types of amines can act only as bases during
sorbose oxidation.
Figure 9 demonstrates that the unsupported Pt powder
catalyst, dispersed on the rotating carbon paste electrode in
the electrochemical cell, is really a good model for the be-
havior of Pt/C in the catalytic reactor. In the reactor the
catalyst potential was measured with a combined electrode
(Pt sheet + Ag/AgCl/KCl) during reaction. The catalyst po-
tential is an in situ measure of the oxidation state of Pt: the
curve of unmodified Pt at 250–350 mV corresponds to
the oxidation of adsorbed hydrogen. The presence of 6,
a strongly adsorbing aromatic amine, suppresses both hy-
drogen and oxygen sorption (below and above ca. 450–500
mV, respectively) on Pt. The higher the amount of 6 in-
jected into the cell during cathodic polarization, the lower
is the amount of oxidized (preadsorbed) hydrogen and
chemisorbed oxygen (the areas under the anodic sweeps
of the voltammograms).
The oxidation of amine modifier itself on the Pt surface
could be proved only in the case of triethylamine (3). In the
presence of 3 a considerable increase in current density was
observed in the oxygen region, compared to the reference
curve of unmodified Pt (Fig. 7). The oxidation current be-
tween 0.45 and 1.0 V increased with increasing amount of
3 in the investigated low concentration range. For compa-
rison, the reaction in the catalytic reactor runs mainly at a
catalyst potential of around 0.7 V, as will be discussed be-
low. It is not yet clear what is the extent of oxidation of 3 in
the presence of sorbose.
All four types of amine decreased the amount of hydro-
gen adsorbed on Pt. The correlation between the amine
concentration in solution and the coverage of Pt by amine
hasalsobeen determined, basedonthechangeinthe hydro-
gen sorption characteristics of Pt (29, 30). Figure 8 shows
that pyridine (6) adsorbs most strongly, as expected, but
the other three types of amines also occupy a considerable
fraction of active sites. The direct application of this quan-
titative correlation to Pt/C in the catalytic reactor is not
possible as the amine : Pts : sorbose molar ratios are substan-
tially different in the catalytic and electrocatalytic systems.
Nevertheless, the electrochemical model experiments have
FIG. 8. Coverage of Pt by tetramethylammonium hydroxide (1), tri-
ethylamine (3), pyridine (6), and hexamethylenetetramine (13) as a func-
tion of amine concentration; 1 wt% aqueous Na-acetate (pH adjusted to
provided unambiguous evidence that (i) a fraction of the 7.3), 2 mg Pt.

OXIDATION OF L-SORBOSE
727
on condition that the pKa value is around 10 or higher. This
is a clear indication of a base-catalyzed reaction (bifunc-
tional catalysis).
There is further evidence concerning the nature of Pt–
amine–sorbose interaction during the oxidation reaction.
Electrochemical model experiments proved that all four
types of amines (tetraalkylammonium hydroxides, trialkyl-,
triaryl, and cycloaliphatic amines) are partially adsorbed on
Pt under reaction conditions. The cyclic voltammetric study
also showed that triethylamine (3) can be oxidized on Pt,
but the other types of amines can influence the sorbose
oxidation only as bases. It is not yet clear whether there is
any interaction between the oxidation of sorbose and 3.
The effect of supported N-bases is more advantageous
than increasing the pH of the aqueous medium. In the lat-
ter case even a moderate gain in reaction rate is accompa-
nied by a striking loss of selectivity. Using a tertiary amine
FIG. 9. Relative rate of sorbose oxidation as a function of the po- or quaternary ammonium hydroxide modifier in catalytic
tential of ( ) unsupported Pt, (᭹) Pt/C, and (᭡) Pt/C modified with hex-
amethylenetetramine (13); modifier to Pts molar ratio of 0.1, standard
conditions.
amounts, the selectivity is always higher (at the same rate)
than that obtained with unmodified catalyst at higher bulk
pH (Fig. 2).
It is known (8, 9) that bases can accelerate the hydration
of the aldehyde intermediate of the alcohol → carboxylic
higher the potential above 0.45–0.5 V (in this solution), the
acid transformation, as shown schematically in Scheme II.
The oxygen insertion to an aldehyde requires higher oxy-
gen coverage of Pt than the oxidative dehydrogenation of
an alcohol or geminal diol (“hydrate”), which explains why
the former route is considerably slower and why nonhy-
drating aldehydes can be prepared in good yield (38, 39). A
strong support to this conception is the frequently reported
absence of pH dependence of the alcohol → carbonyl com-
pound type transformations (39–42). Note that there are
other possible interpretations of the effect of bases on the
reaction rate, such as the base-catalyzed deprotonation of
the hydroxyl group as the first step of the oxidation reac-
tion (31), or the change with pH of surface coverage of Pt
by oxygen and reactant (16).
higher is the oxygen-coverage of Pt (19). The correlation
between the relative reaction rate and catalyst potential is
very similar in the electrochemical cell and in the reactor,
as shown in Fig. 9. The reaction rate is negligible at 0.5 V or
below due to the self-poisoning of Pt by L-sorbose (19, 20).
After the oxidative removal of by-products, the rate rapidly
increases to its maximum at 0.7–0.75 V. A further increase
of oxygen coverage (catalyst potential) suppresses the rate
of sorbose oxidation due to the decrease of the proportion
of active sites (Pt0). The potential shift of ca. 30–50 mV be-
tween the curves of Pt and Pt/C is negligible compared to
the different structures (dispersion, support effect, etc.) of
the two catalysts.
There is a negative potential shift by 60–120 mV mea-
sured on Pt/C in the presence of hexamethylenetetramine
(13) and related to the unmodified Pt/C (Fig. 9). Similar po-
tential shifts by 50–100 mV in the negative direction were
observed when using other types of tertiary amine or qua-
ternary ammonium hydroxide modifiers. This shift may be
due, at least in part, to the change of pH on the Pt sur-
face compared to the bulk pH of the solution (see also the
disturbance in Fig. 1 after the addition of pyridine).
In order to use the rate accelerating effect of amines, they
have to be applied in a rather low concentration, typically
in a molar ratio of modifier to Pts of 0.1–0.2. The amines
cover a fraction of active sites (Pt0) and this effect can sur-
pass the positive influence of base catalysis. The geometric
blocking of active sites is clearly indicated by the decreasing
DISCUSSION
Rate Acceleration by Modification with Amines
It has been shown that small amounts of supported
tetraalkylammonium hydroxides and tertiary amines can
significantly enhance the rate of Pt-catalyzed oxidation of
L-sorbose to 2-keto-L-gulonic acid. Rate acceleration can be
obtained with amines of rather different chemical structure,
SCHEME II

¨
BRONNIMANN ET AL.
728
FIG. 10. Side and top views of the models for adsorption on a flat Pt surface of quinuclidine (11), diazabicyclooctane (12), and an optimized
hexamethylenetetramine (13)–L-sorbose (ꢁ-L-sorbopyranose) complex.
rateofsorboseconsumptionwithincreasingnumberofben- amines quinuclidine (11), diazabicyclooctane (12), and hex-
zeneringsofaromaticaminemodifiers6–8, underotherwise amethylenetetramine (13) provides some insight. The min-
identical conditions.
imum energy conformations of these modifiers are shown
Trialkyl- and triarylamines are assumed to adsorb on Pt in Fig. 10. The monoamine 11 is likely to be adsorbed on
with the electron-rich N atoms lying on the surface (“an- a flat Pt surface as illustrated: the molecule is anchored to
choring” atom) and being surrounded by alkyl or aryl moi- the Pt surface (soft acid) by the N-atom (soft base). In this
eties. Theirinteractionwiththeadsorbedaldehydeinterme- most stable form the N atom is not accessible to any inter-
diate resulting in base-catalyzed hydration is unlikely, as (i) action with L-sorbose due to the shielding effect of the alkyl
the basicity of the adsorbed N atom should be significantly chains, which conclusion is in agreement with the observed
smaller after interaction with Pt, and (ii) the N atom is ster- negligible effect on selectivity (Fig. 4). Nevertheless, the un-
ically not accessible to the carbonyl group of the adsorbed adsorbed amine can act as a strong base (pKa = 10.9) which
intermediate. It is assumed that the adsorbed amine is only explains the rate acceleration by a factor of 2 (Table 3), as
a “reservoir”; the hydration of the reaction intermediate is discussed above.
catalyzed by an unadsorbed amine molecule close to the Pt
surface.
The optimized structure of diamine 12, adsorbed on a
flat Pt surface, is rather similar to that of 11, except for the
second N atom at the “top” of the molecule. This basic N
atom is accessible to interaction with L-sorbose but the com-
plex cannot easily adsorb on a flat Pt surface. This amine
provides some moderate (5–6%) increase in selectivity to
High Selectivity with Hexamethylenetetramine (13)
An intriguing question is to enquire “What is the nature
of interaction between Pt, L-sorbose, and hexamethylenete-
tramine (13) which provides the outstanding selectivity to 2-KLG at low conversion, but this positive effect is elimi-
2-keto-L-gulonic acid?” Although a definite answer cannot nated by side reactions above 25% conversion (Fig. 4).
begivenbasedonthepresentedresults, wethinkthatacom-
The tetramine 13 is proposed to be adsorbed by one N
parison of the structure and influence of the cycloaliphatic atom facing to a Pt atom. An “upside-down” adsorption

OXIDATION OF L-SORBOSE
729
mode in which three N atoms could interact with the Pt sur-
face (modeled on an ideal Pt (111) structure) is sterically
hinderedbythesurroundingthreemethylenegroups. When
using this amine as modifier, the accessibility of one of the
N atoms to an interaction with one of the OH functional
groups of L-sorbose is the highest. Several possible com-
plexes between L-sorbose and modifier 13 were optimized.
A common feature of these complexes is that after adsorp-
tion on a flat Pt surface, the pyranose ring of sorbose is in
a tilted, “partially adsorbed” position. Note that the ꢁ-L-
sorbopyranose structure was considered in the calculations
as L-sorbose is present mainly in this form (95%) in aqueous
solutions and at ambient temperature (43). Figure 10 shows
one example of a H-bond stabilized (by 29 kJ molꢂ1) assem-
bly of L-sorbose and hexamethylenetetramine 13. (It is very
5. Fried, J., and Edwards, J. A., “Organic Reactions in Steroid Chem-
istry,” Vol. 1, p. 329. Van Nostrand Reinhold, New York, 1972.
6. Haines, A. H., “Methods for the Oxidation of Organic Compounds,”
p. 158. Academic Press, London, 1988.
7. Ro¨per, H., in “Carbohydrates as Organic Raw Materials” (F. W.
Lichtenthaler, Ed.), p. 267. VCH, Weinheim, 1990.
8. van Bekkum, H., in “Carbohydrates as Organic Raw Materials” (F. W.
Lichtenthaler, Ed.), p. 289. VCH, Weinheim, 1990.
9. Mallat, T., Baiker, A., Catal. Today 19, 247 (1994).
10. Besson, M., Lahmer, F., Gallezot, P., Fuertes, P., andFle`che, G., J. Catal.
152, 116 (1995).
11. Despeyroux, B. M., Deller, K., and Peldszus, E., Stud. Surf. Sci. Catal.
55, 159 (1990).
12. Smits, P. C. C., Kuster, B. F. M., van der Wiele, K., and van der Baan,
H. S., Appl. Catal. 33, 83 (1987).
13. Abbadi, A., and van Bekkum, H., Appl. Catal. A 124, 409 (1995).
14. Dirkx, J. M. H., and van der Baan, H. S., J. Catal. 67, 1 (1981).
15. Dirkx, J. M. H., and van der Baan, H. S., J. Catal. 67, 14 (1981).
likely that the other two unadsorbed N atoms also interact 16. Schuurman, Y., Kuster, B. F. M., van der Wiele, K., and Marin, G. B.,
Stud. Surf. Sci. Catal. 72, 43 (1992).
17. Hronec, M., Cvengrosova, Z., Stolcova, M., and Klavski, J., React.
Kinet. Catal. Lett. 20, 207 (1982).
18. Kim, V. I. E., Zakharova, V., and Kozhevnikov, I. V., React. Kinet.
with sorbose or water molecules by H-bonding, which is not
shown here.) The steric arrangement of the presented 13–
sorbose complex can explain the outstanding preference to
C-1 oxidation: only the C-1–O–H fragment of the tilted sor-
Catal. Lett. 45, 271 (1991).
bose molecule is close to the Pt surface, liable to oxidative 19. Bro¨nnimann, C., Bodnar, Z., Hug, P., Mallat, T., and Baiker, A., J.
Catal. 150, 199 (1994).
20. Parsons, R., and VanderNoot, T., J. Electroanal. Chem. 257, 9 (1988).
21. Bro¨nnimann, C., Mallat, T., and Baiker, A., J. Chem. Soc. Chem. Com-
dehydrogenation.
CONCLUSIONS
mun. 1377 (1995).
22. Lipkin, D., and Stewart, T. D., J. Am. Chem. Soc. 61, 3295 (1939).
23. Nakamura, Y., Bull. Chem. Soc. Jpn. 16, 367 (1941).
It has been demonstrated that the modification of sup-
ported Pt by traces of tertiary amines and quaternary 24. Mallat, T., Bodnar, Z., Baiker, A., Greis, O., Stru¨big, H., and Reller,
A., J. Catal. 142, 237 (1993).
25. Mallat, T., Bodnar, Z., and Baiker, A., Stud. Surf. Sci. Catal. 78, 377
ammonium hydroxides can significantly enhance the rate
and/or selectivity of sorbose oxidation. The efficiency of
this new method is far better than that of tuning the bulk
pH of the solution or modifying Pt by heavy metal adatoms.
(1993).
26. Sokolskii, D. V., “Hydrogenation in Solution.” Izd. Nauka Kaz. SSR
Alma Ata, 1979. [in Russian]
It is likely that the method can be applied in other Pt- or 27. Mallat, T., and Baiker, A., Catal. Today 24, 143 (1995).
28. Mallat, T., Allmendinger, T., and Baiker, Appl. Surf. Sci. 52, 189 (1991).
29. Goncalves, R. S., Leger, J. M., and Lamy, C., Electrochim. Acta 33,
Pd-catalyzed oxidations of sensitive compounds, such as
polyols.
A model for the Pt–amine–sorbose interaction which
1581 (1988).
30. Breiter, M. W., in “Modern Aspects of Electrochemistry” (J. M.
provides a possible explanation for the rate acceleration
and selectivity enhancement observed has been proposed.
The future refinement of this model requires further evi-
dence concerning the catalytic influence of amines on each
of the consecutive reaction steps of the alcohol → acid
transformation and the effect of oxidation products on the
Pt–amine–sorbose interaction.
Bockris and B. E. Conway, Eds.), Vol. 10, p. 161. Plenum, New York,
1975.
31. Vinke, P., de Wit, D., de Goede, A. T. J. W., and van Bekkum, H., Stud.
Surf. Sci. Catal. 72, 1 (1992).
32. Dirkx, J. M. H., van der Baan, H. S., and van den Broek, J. M. A. J. J.,
Carhohydr. Res. 59, 63 (1977).
33. Perrin, D. D., “Dissociation Constants of Organic Bases in Aqueous
Solution.” Butterworths, London, 1965.
34. Weast, R. C. (Ed.), “CRC Handbook of Chemistry and Physics,” 67th
ed., p. D-159. CRC Press, Boca Raton, FL, 1986.
35. Mayer, B., Spencer, T. A., and Onan, K. D., J. Am. Chem. Soc. 106,
6343 (1984).
36. Chebotarev, A. N., and Kachan, S. V., Russ. J. Phys. Chem. 65, 360
(1991). [English transition]
ACKNOWLEDGMENTS
Financial support of this work by Hoffmann-La Roche AG, Switzer-
land, and the “Kommission zur Fo¨rderung der wissenschaftlichen
Forschung” is gratefully acknowledged.
37. Kolling, M., Inorg. Chem. 9, 408 (1970).
38. Mallat, T., Bodnar, Z., Hug, P., and Baiker, A., J. Catal. 153, 131 (1995).
39. Heyns, K., and Blazejewicz, L., Tetrahedron 9, 67 (1960).
40. Nicoletti, J. W., and Whitesides, G. M., J. Phys. Chem. 93, 759 (1989).
41. Vinke, P., van Dam, H. E., and van Bekkum, H., Stud. Surf. Sci. Catal.
55, 147 (1990).
REFERENCES
1. Reichstein, T., and Gruessner, A., Helv. Chim. Acta 17, 311 (1934).
2. Crawford, T. C., and Crawford, S. A., Adv. Carbohydr. Chem. Bio-
chem. 37, 79 (1980).
42. Tillaart, J. A. A., Kuster, B. F. M., and Marin, G. B., Appl. Catal. A.
120, 127 (1994).
3. Heyns, K., Ann. 558, 177 (1947).
4. Heyns, K., and Paulsen, H., Adv. Carbohydr. Chem. 17, 169 (1962).
43. Que, L., and Gray, G. R., Biochemistry 13, 146 (1974).