Remarkable particle size effect in Rh-catalyzed enantioselective hydrogenations

Article abstract of DOI:10.1016/j.jcat.2008.12.002

A series of 0.5-4.3 wt% Rh/Al₂O₃ catalysts were prepared by flame synthesis. STEM indicated relatively narrow particle size distributions for all catalysts and the mean particle size increased almost linearly with the Rh content in the range 0.96-1.65 nm. A DRIFTS study of CO adsorption on as prepared Rh/Al₂O₃ and after heat treatment in hydrogen at 400 °C revealed that there was no Rh oxide present at the catalyst surface after the high temperature reduction, which procedure is commonly used prior to enantioselective hydrogenation. In the hydrogenation of ethyl pyruvate and ethyl 3-methyl-2-oxobutyrate the cinchona-modified 4.3 wt% Rh/Al₂O₃ gave considerably higher ee than those achieved with the best known Rh catalyst. A decrease of the metal loading and thus the mean Rh particle size, led to a loss of ee to (R)-lactate by a factor of up to seven at 1 bar and up to two at 10-100 bar. Our interpretation is that the performance of Rh/Al₂O₃ is strongly distorted at atmospheric pressure by catalyst deactivation due to the Al₂O₃-catalyzed aldol condensation of the substrate. During the fast reactions at 100 bar the contribution of strongly adsorbed impurities is small and the variation of ee is mainly due to an intrinsic particle size effect. The structure sensitivity observed under optimal conditions, at high surface hydrogen concentration, is mainly due to steric effects: a small, ca. 1 nm Rh particle cannot accommodate the enantiodifferentiating diastereomeric substrate-modifier complex and the hydrogenation on its surface leads to racemic product. A practical conclusion is that there is no advantage of using small nanoparticles and low metal loading in the enantioselective hydrogenation of α-ketoesters.

Full text of DOI:10.1016/j.jcat.2008.12.002

Journal of Catalysis 261 (2009) 224–231
Contents lists available at ScienceDirect
Journal of Catalysis
www.elsevier.com/locate/jcat
Remarkable particle size effect in Rh-catalyzed enantioselective hydrogenations
∗
Fatos Hoxha, Niels van Vegten, Atsushi Urakawa, Frank Krumeich, Tamas Mallat, Alfons Baiker
Institute for Chemical and Bioengineering, Department of Chemistry and Applied Biosciences, ETH Zurich, Hönggerberg, HCI, CH-8093 Zurich, Switzerland
a r t i c l e i n f o
a b s t r a c t
Article history:
A series of 0.5–4.3 wt% Rh/Al2O3 catalysts were prepared by flame synthesis. STEM indicated relatively
narrow particle size distributions for all catalysts and the mean particle size increased almost linearly
with the Rh content in the range 0.96–1.65 nm. A DRIFTS study of CO adsorption on as prepared
Received 11 November 2008
Revised 2 December 2008
Accepted 3 December 2008
Available online 20 December 2008
◦
Rh/Al2O3 and after heat treatment in hydrogen at 400 C revealed that there was no Rh oxide present
at the catalyst surface after the high temperature reduction, which procedure is commonly used prior to
enantioselective hydrogenation. In the hydrogenation of ethyl pyruvate and ethyl 3-methyl-2-oxobutyrate
the cinchona-modified 4.3 wt% Rh/Al2O3 gave considerably higher ee than those achieved with the best
known Rh catalyst. A decrease of the metal loading and thus the mean Rh particle size, led to a loss of
ee to (R)-lactate by a factor of up to seven at 1 bar and up to two at 10–100 bar. Our interpretation is
that the performance of Rh/Al2O3 is strongly distorted at atmospheric pressure by catalyst deactivation
due to the Al2O3-catalyzed aldol condensation of the substrate. During the fast reactions at 100 bar the
contribution of strongly adsorbed impurities is small and the variation of ee is mainly due to an intrinsic
particle size effect. The structure sensitivity observed under optimal conditions, at high surface hydrogen
concentration, is mainly due to steric effects: a small, ca. 1 nm Rh particle cannot accommodate the
enantiodifferentiating diastereomeric substrate–modifier complex and the hydrogenation on its surface
leads to racemic product. A practical conclusion is that there is no advantage of using small nanoparticles
and low metal loading in the enantioselective hydrogenation of α-ketoesters.
Keywords:
Rhodium
Ethyl pyruvate
Ethyl 3-methyl-2-oxobutyrate
Cinchonidine
Quinine
Asymmetric hydrogenation
Particle size effect
Structure sensitivity
© 2008 Elsevier Inc. All rights reserved.
1. Introduction
It is well established that the catalytic properties may be
strongly influenced by the size of metal particles [16–18]. The clas-
sical calculations from Boronin and Poltorak [19] and van Hard-
In the enantioselective hydrogenation of activated ketones, chi-
rally modified Rh is characterized by moderate chemo- and enan-
tioselectivities [1–7], not comparable to that of the Pt–cinchona
system [8–11]. An exception is the hydrogenation of hydroxyke-
tones on cinchonidine (CD)-modified Rh/Al2O3 with up to 80% ee
[12]. Even in this reaction, however, the reaction rate was low and
high modifier/substrate molar ratio was necessary, mainly due to
the competing hydrogenation and destruction [13,14] of the mod-
ifier itself. Several attempts have been made to improve the per-
formance of Rh and to broaden the scope of suitable substrates.
Recently, in the hydrogenation of ketopantolactone Rh/Al2O3 in the
presence of β-isocinchonine gave 68% ee and in this facile reac-
tion only a low modifier/substrate ratio was necessary [15]. In the
above studies the research aimed at finding the appropriate com-
bination of catalyst, modifier, and the reaction conditions, without
a profound analysis on the role of catalyst structure, metal particle
size, and surface morphology.
eveld and Hartog [20] showed that the biggest changes of pro-
portions between facets, edges, corners, and micro-defects at the
surface occur between 1 and 5 nm. Above 5 nm the metal–support
interaction tend to decrease rapidly, making the catalyst of lit-
tle value for studying the structural effects [21]. Depending on
whether or not the turnover frequency (TOF) is affected by the
structure of the particle surface, Boudart [22] coined the terms
structure-sensitive or structure-insensitive reaction.
Hydrogenation of unsaturated hydrocarbons on Pd is an early
and thoroughly investigated case of structure sensitive reac-
tions [23,24]. Hydrogenation of alkenes is barely sensitive to the
metal particle size, while strong dependence of rate and selectiv-
ity was observed in the partial hydrogenation of alkadienes and
alkynes.
There are numerous studies [25–28] indicating some structure
sensitivity of asymmetric hydrogenation reactions on Pt, Pd, Ni, Ru,
and Ir, but systematic studies with properly designed catalysts are
rare and the presence of unidentified surface impurities may blur
the real correlation [29,30]. According to an early report on the hy-
drogenation of ethyl pyruvate on CD-modified Pt/Al2O3 [31], both
initial rate and enantioselectivity increased with increasing Pt par-
Corresponding author.
*
E-mail address: baiker@chem.ethz.ch (A. Baiker).
0021-9517/$ – see front matter © 2008 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2008.12.002

F. Hoxha et al. / Journal of Catalysis 261 (2009) 224–231
225
Table 1
Nominal and measured Rh content of Rh/Al2O3 catalysts (ICP-AES technique after
dissolution in aqua regia).
Nominal Rh content (wt%)
Measured Rh content (wt%)
0.5
1
2
3
4
0.5
0.8
1.4
2.5
3.5
4.3
5
the nominal value, with a deviation of up to 32% (rel.). The water
content of the catalysts did not exceed 10 wt%.
Scheme 1. Hydrogenation of ethyl pyruvate and ethyl 3-methyl-2-oxobutyrate on
Rh/Al2O3 in the presence of quinine (QN) and cinchonidine (CD), respectively.
2.2. Catalytic hydrogenation
ticle size up to 3–4 nm. However, due to the different preparation
conditions (Pt precursor, reducing agent, etc.) a significant effect of
surface impurities on the catalyst performance cannot be excluded.
In another study of pyruvate hydrogenation [32], the metal particle
size of Pt/graphite was increased by sintering at elevated tempera-
ture in hydrogen. As a result, the ee increased to some extent but
then dropped when the sintering temperature exceeded 700 K, due
to contamination of Pt with graphite.
Beside the variation of surface impurities, a too broad metal
particle size distribution may also hinder to draw unambiguous
conclusions on the particle size effect [32,33]. In some other cases
the enantioselectivities are low, far from the optimum, and it is
questionable whether the observed correlation between metal par-
ticle size and enantioselectivity is valid also at close to the opti-
mum conditions of the reaction [34,35].
Here we report for the first time the effect of particle size
on the enantioselectivity of chirally modified Rh. The flame made
Rh/Al2O3 catalyst series, with a particle size distribution from 0.6
to 2.4 nm, has already been used for studying the chemoselec-
tivity in the hydrogenation of an aromatic ketone [36]. In the
present study, hydrogenation of the α-ketoesters ethyl lactate and
ethyl 3-methyl-2-oxobutyrate was chosen as test reactions and
the catalysts were modified with CD or quinine (QN), respectively
(Scheme 1).
The catalyst was always reduced at elevated temperature pri-
ore to use. According to the standard procedure, the catalyst was
heated under flowing nitrogen up to 400 C in 30 min, followed by
a reduction in flowing hydrogen for 60 min at the same tempera-
ture, and finally cooling down in hydrogen in 30 min.
Hydrogenations at atmospheric pressure were carried out in a
100-ml glass reactor equipped with magnetic stirrer, septum for
sample collection or substrate addition, and an in and outlet line
for gas flowing. To avoid the contact of the freshly reduced catalyst
with air, the reductive heat treatment and the hydrogenation reac-
tion were made in the same reactor. After introducing the solvent
and the modifier into the reactor containing the freshly reduced
catalyst under hydrogen at room temperature, the mixture was
stirred for 1 min. The reaction was started by introducing the sub-
strate and immediately switching on the stirring.
◦
The same reactor was used to study the Al2O3-catalyzed degra-
dation of ethyl pyruvate. The conditions were identical to those
of catalytic hydrogenation but the Rh/Al2O3 catalyst was replaced
with the appropriate amount of Al2O3. The flame made Al2O3 (spe-
cific surface area 267 m2/g) was heated at 400 C under nitrogen
◦
for 1 h and finally cooled down to room temperature. To the acti-
vated Al2O3 a solution of ethyl pyruvate and tetradecane (internal
standard) in toluene were introduced and the reaction started by
switching on the stirring. The nitrogen atmosphere was maintained
throughout the experiment.
2. Experimental
2.1. Materials
The hydrogenations at elevated pressure were carried out in a
25 ml stainless steel Parr autoclave equipped with a 16 ml glass
liner and a PTFE cover, and a magnetic stirrer. The autoclave was
equipped also with a valve for sample collection or substrate ad-
dition. The treatment of the catalyst was carried out in a fix-bed
reactor following the standard procedure. At the end, the freshly
reduced catalyst was purged with nitrogen for 60 min at room
temperature to remove the excess of hydrogen and then trans-
ferred immediately to the autoclave.
The substrates ethyl pyruvate (Acros, 98%, 0.02% ethyl lactate
after distillation) and ethyl 3-methyl-2-oxobutyrate (Aldrich, 97%)
were carefully distilled in vacuum before use. Other chemicals
were used as received: cinchonidine (CD) (Fluka, ꢀ98% alkaloid),
quinine (QN) (Fluka, ca. 99% alkaloid), toluene (Aldrich, anhydrous
99.8%), rhodium(III) acetylacetonate (Acros, 97%), aluminum(III)
acetylacetonate (ABCR, 99%), acetic acid (Fluka, analytical grade),
methanol (Fluka, analytical grade).
In all experiments, the Rh/modifier/substrate molar ratios were
kept constant at 1/3.7/189. In the hydrogenations at elevated pres-
sure, the proper amount of Rh/Al2O3 containing 2.43 μmol Rh,
9 μmol modifier, 0.46 mmol substrate, and 5 ml solvent were
stirred magnetically (1000 rpm) at room temperature for 60 min.
The pressure was held at a constant value with a constant pressure
regulator valve. In the hydrogenations at atmospheric pressure,
9.72 μmol Rh catalyst, 36 μmol modifier, 1.84 mmol substrate, and
15 ml solvent were stirred magnetically (1000 rpm) at room tem-
perature for 120 min.
The conversion and enantioselectivity (ee) were determined by
GC analysis, using an HP 6890 gas chromatograph and a Chirasil-
DEX CB (Chrompack 7502 25 m × 0.25 mm × 0.25 μm) capillary
column. All experiments were carried out at least twice and the
average values are shown in the figures. In all experiment the
The preparation of Rh/Al2O3 catalysts with the flame spray py-
rolysis technique is described in previous reports [36,37]. Shortly,
the appropriate amounts of Rh and Al precursors were dissolved
in an acetic acid/methanol 1/1 (volume ratio) mixture. The Al con-
centration was always 0.39 M. The solution was pumped through
a capillary at a rate of 5 mL/min and nebulized with 5 Ln/min
O2, and the resulting spray was ignited by a circular supporting
methane/oxygen flame (1.0/0.6 Ln/min), resulting in an approx-
imately 6 cm long flame. Particles were collected on a cooled
Whatman GF/D filter (257 mm diameter). A Busch SV 1040C vac-
uum pump aided in particle recovery.
The metal content of the catalysts was determined with in-
ductively coupled plasma atomic emission spectroscopy (ICP-AES)
technique (by Alab AG, Switzerland) after dissolution in aqua re-
gia. As shown in Table 1, the Rh contents were mostly lower than

226
F. Hoxha et al. / Journal of Catalysis 261 (2009) 224–231
(R)-enantiomer was formed in excess without any detectable side
product. The estimated standard deviation of ee is about 0.5%.
2.3. Characterization of the catalysts
The scanning transmission electron microscopic (STEM) analy-
sis was performed on a Tecnai F30 microscope (FEI, field emis-
sion cathode, operated at 300 kV). The material was dispersed in
ethanol and some drops were deposited onto a perforated carbon
foil supported on a copper grid. STEM images were recorded with
a high-angle annular dark field (HAADF) detector, using almost ex-
clusively incoherently scattered electrons (Rutherford scattering) to
obtain images with atomic number (Z) contrast [38]. The mean Rh
particle size dVS was calculated based on a minimum of 200 parti-
cles using the following equation:
ꢀ
ꢀ
ꢁ
dVS
=
nid3
nid2,
i
i
i
i
were ni is the number of particles with diameter di.
Diffuse reflectance infrared Fourier transform spectroscopy
(DRIFTS) was used to study the CO adsorption on Rh/Al2O3. The
analysis was carried out at 297 K with an EQUINOX 55 spec-
trometer (Bruker Optics) equipped with a liquid nitrogen-cooled
HgCdTe detector. Multiple samples (max. 4), separated by quartz
wool, were placed without dilution in a plug–flow DRIFTS cell [39],
allowing identical experimental conditions for all samples. The off
gas of the cell was analyzed by a mass spectrometer. Spectra were
−1
collected by averaging 200 scans at 4 cm
resolution. The stan-
dard treatment procedure mentioned above was used to reduce
the catalyst (flowing H2, 20 ml/min), in this case with argon (Ar)
as inert gas. CO adsorption was monitored over 60 min in flowing
10 vol% CO/Ar (5 ml/min), followed by flowing Ar (20 ml/min) for
30 min.
◦
Fig. 1. Particle size distribution of Rh/Al2O3 catalysts prereduced at 400 C in H2.
The dashed vertical lines represent the mean diameter dVS
3. Results and discussion
.
3.1. Electron microscopy
The particle size distribution of the six Rh/Al2O3 catalysts was
determined by scanning transmission electron microscopy (STEM).
In all cases Rh was highly dispersed with a relatively narrow par-
ticle size distribution, particularly for the three lowest metal load-
ings (Fig. 1). The mean particle size increased monotonously in
the range 0.96–1.65 nm with increasing Rh loading. The straight-
forward correlation between the mean particle size and the Rh
content is shown in Fig. 2. The catalysts were pretreated according
to the standard procedure at elevated temperature prior to STEM
investigation; hence these data can be directly related to the hy-
drogenation rates and selectivities. A comparison of the samples
◦
before and after pretreatment at 400 C revealed no major struc-
tural changes. For example, the mean Rh particle size of the freshly
prepared 2.5 wt% Rh/Al2O3 catalyst increased by less than 10% dur-
ing the reductive pretreatment.
3.2. IR study of CO adsorption
The DRIFT spectra of CO adsorbed on the flame made 4.3 wt%
Rh/Al2O3 catalyst (both as-received and after reduction at 400 C)
◦
Fig. 2. Influence of Rh loading on the mean size of metallic crystallites. Samples
were prereduced at 400 C in H2.
◦
have already been published [36]. Here we extended the investi-
gation to the whole catalyst series in order to gain insight into
the oxidation state of rhodium species after the standard reductive
treatment. Since only the metallic Rh surface sites are active in the
hydrogenation reactions, determination of the oxidation state of
Rh is inevitable to understand the performance of these catalysts.
The standard catalyst treatment procedure prior to hydrogenations
was used here also: the samples were reduced in flowing hydrogen
at 400 C and the exhaust gas was analyzed using mass spectrome-
try. The stabilized reduced state of the catalyst was reached within
40 min (the concentration of the exhaust gas mixture was constant
showing low water content).
◦

F. Hoxha et al. / Journal of Catalysis 261 (2009) 224–231
227
Fig. 4. Typical DRIFT spectra of CO adsorption on Rh/Al2O3 samples, prereduced
at 400 C in H2 (top) and as received (bottom).
◦
◦
Fig. 3. IR spectrum of the 0.5 wt% Rh/Al2O3, prereduced at 400 C in H2, and the
deconvoluted components of the CO bands.
The DRIFT spectrum of CO adsorbed on the reduced 0.5 wt%
Rh/Al2O3 catalyst is shown in Fig. 3. Three kinds of CO bands were
−1
identified in the IR spectra. The broad band (1832–1835 cm
)
is attributed to the multi-bonded CO species Rhn0 (–CO, n ꢀ 2).
−1
The bands between 2000 and 2100 cm
are associated with
−1
two species: a doublet corresponds to the symmetric (2086 cm
)
and asymmetric (2015 cm−1) stretch of gem-dicarbonyl Rh –(CO)2
+
−1
dimer vibrations and the band at 2053–2063 cm
is assigned to
linear Rh0–CO species [40–48]. Figs. 4 and 5 illustrate that with in-
creasing Rh content of the catalysts both bands corresponding to
linear and bridged CO species on Rhn0 sites are enhanced in relative
intensity and shifted to higher wavenumbers. The latter correlation
indicates increasing metal particle size [48–51], in line with the
STEM measurements (Fig. 2) that revealed a positive correlation
between Rh loading and the mean metal particle size. The bands
of the carbonyl species have a constant wavenumber, independent
of the Rh loading [46,50], and thus the band positions (2086 and
2015 cm−1) were set constant during the spectral deconvolution
procedure.
In the infrared spectra of the unreduced samples the bands cor-
responding to the gem-dicarbonyl species Rh –(CO)2 (2015 and
+
Fig. 5. Shift of the linear Rh0–CO band with increasing Rh loading. Data are ex-
tracted from the deconvoluted IR spectra of CO adsorption on reduced Rh/Al2O3
catalysts.
−1
2086 cm−1) were dominant. The shoulder at ca. 2110 cm
cor-
responds to linearly bonded CO on Rh2+, typical for oxidized
rhodium [49,52]. Importantly, this band was not detectable on the
reduced samples.
Rh/Al2O3 catalysts treated at elevated temperature under an oxy-
gen atmosphere [58–60]. Formation of Rh(AlO2)y has been con-
sidered the origin of the deactivation of Rh/Al2O3 in oxidation
reactions. Their activity may partially be restored by an appropri-
+
The presence of Rh species in the IR spectra is commonly at-
tributed to interaction with gaseous CO. The Rh –(CO)2 species
+
can be formed during CO adsorption by the oxidation of Rh0
sites involving the hydroxyls groups on the surface of the sup-
port [53–55]. Van’t Blik et al. [56] studied the coordination of Rh
atoms on a 0.57 wt% Rh/Al2O3 catalyst with EXAFS technique. They
proved that the adsorption of CO causes a significant disruption of
the local coordination environment of the Rh0x crystallites, lead-
◦
ate reductive treatment at temperatures above 600 C [59,61–64],
but the conditions in the reductive treatment used in the present
study are too mild for this type of regeneration.
+
It was shown that the fraction of Rh –(CO)2 species created
during CO adsorption decreased with increasing rhodium particle
size [40,48,62], and a similar correlation may be expected for Rh
aluminate formation. In our study based on the deconvoluted spec-
+
ing to the formation of isolated, atomically dispersed Rh sites
localized and stabilized on the Al2O3 surface [53,54]. The presence
+
tra, the ratio of Rh0x to cationic Rh species is 2–2.5 at low Rh
+
of the Rh –(CO)2 bands in the IR spectra of the unreduced cata-
lyst may be explained by reduction of the oxidized species Rhn+
(n ꢀ 2) with the involvement of the OH groups on the Al2O3 sur-
face [57] or by the adsorbant CO itself [54].
content (0.5–1.4 wt% Rh/Al2O3) and 9–18 at higher Rh loadings
(Fig. 6).
Despite of the well known chemical interaction of CO and
Rh/Al2O3 that complicates the adsorption study, the DRIFT mea-
surements prove that (i) the Rh0x particle size increases monoto-
nously with the Rh loading, (ii) there is no rhodium oxide present
Another feasible explanation for the presence of cationic species
is the formation of Rh aluminate (Rh(AlO2)y [58]) during the flame
spray catalyst synthesis. The presence of Rh(AlO2)y is typical for

228
F. Hoxha et al. / Journal of Catalysis 261 (2009) 224–231
(a)
+
Fig. 6. Ratio of metallic (Rh0x ) to cationic (Rh ) rhodium species in Rh/Al2O3 cata-
lysts, as determined from the DRIFT spectra of CO adsorption.
◦
after the standard reductive pretreatment procedure at 400 C, and
(iii) the existence of ionic Rh species as Rh aluminate in the flame
made catalysts cannot be excluded.
3.3. Enantioselective hydrogenation
Based on former observations in the Rh-catalyzed hydrogena-
tion of activated ketones [15], the best substrate-modifier-solvent
systems were chosen here to test the performance of the flame
made Rh/Al2O3 catalysts. In order to avoid the re-oxidation of the
metallic Rh particles during transfer of the catalyst from the fixed
bed reactor commonly used for catalyst prereduction to the auto-
clave, here we used a special glass reactor for both steps, the cata-
◦
lyst pretreatment at 400 C and the enantioselective hydrogenation
of α-ketoesters at room temperature and atmospheric pressure.
Variation of enantioselectivity and reaction rate (expressed as
TOF) in the hydrogenation of ethyl pyruvate on quinine-modified
Rh/Al2O3 are plotted in Fig. 7 as a function of mean metal particle
size. The TOF values were calculated based on the STEM-derived
Rh dispersions [46,65]; i.e. the reaction rate is related to the frac-
tion of surface Rh atoms. The ees determined at a relatively low
(9 2%) conversion show that the increase of the Rh mean parti-
cle size from 0.96 to 1.65 nm induced the increase of ee from 7
to 47%. Similar trend was observed after 2 h reaction time, where
the ee increased from 11 to 53% and the yield to ethyl lactate grew
from 15 to 56% (Fig. 7a). The increase of the mean particle size
from 0.96 to 1.65 nm led to a nine fold enhancement of the re-
action rate at 9 2% conversion and up to five fold enhancement
after 2 h reaction time (Fig. 7b).
Interpretation of the surprisingly big changes in rate and enan-
tioselectivity needs caution. It is well known that hydrogenation
of α-ketoesters is accompanied by extensive side reactions [66].
A dominant side reaction is the aldol reaction of the activated
ketone leading to strongly adsorbing dimers and oligomers. This
side reaction is catalyzed by the basic quinuclidine N of the mod-
ifier, the basic sites on the Al2O3 support [67,68], and the metal
surface [69,70]. A simple way to suppress the base-catalyzed al-
dol reaction is to use an acidic solvent (acetic acid), but in our
case it is not a good solution due to the poor selectivity of Rh
in this medium [15]. In all experiments, the Rh/modifier/substrate
molar ratios were kept constant in order to obtain comparable re-
sults. An obvious consequence of this approach is that variation
(b)
Fig. 7. Influence of Rh mean particle size on the enantioselectivity and reaction rate
in the hydrogenation of ethyl pyruvate to (R)-ethyl lactate over quinine-modified
Rh/Al2O3 catalysts; standard conditions, toluene, 1 bar, 24 C; filled symbols: data
◦
at 9 2% conversion, open symbols: data after 2 h reaction time.
of the Rh loading of the catalysts leads to changes in the ethyl
pyruvate/Al2O3 molar ratio, in our case from 0.94 to 9.4 for the
lowest and highest Rh loading, respectively. To prove the signifi-
cance of the Al2O3-catalyzed aldol reaction under the conditions
applied here, we repeated the reaction in Fig. 7 with the 0.5 wt%
Rh/Al2O3 catalyst but replaced the catalyst with the appropriate
amount of flame made Al2O3 and the hydrogen with nitrogen. The
disappearance of ethyl pyruvate in dry toluene as solvent is shown
in Fig. 8. The rapid drop of pyruvate concentration in solution in
the first few minutes is attributed to adsorption on the high sur-
face area support. The slower but still considerable conversion of
ethyl pyruvate is mainly due to the aldol condensation, as shown
previously by ATR-IR spectroscopy [67,68]. No (volatile) products
were detectable by GC analysis. Oligomers that are formed by aldol
condensation may strongly adsorb on the catalyst surface including
the metal particles and diminish the reaction rate and enantiose-

F. Hoxha et al. / Journal of Catalysis 261 (2009) 224–231
229
Fig. 8. Ethyl pyruvate consumption in the presence of Al2O3, standard conditions,
toluene, 1 bar, 24 C, N2 atmosphere.
Fig. 9. Influence of mean Rh particle size on the enantioselectivity in the hydrogena-
tion of ethyl pyruvate to (R)-ethyl lactate with quinine-modified Rh/Al2O3 catalysts;
standard conditions, toluene, 24 C; filled symbols: data at 10 bar, open symbols:
◦
◦
data at 100 bar.
lectivity. It has been shown for the hydrogenation of ethyl pyruvate
on Pt/Al2O3 in toluene that non-volatile impurities in the substrate
diminish the reaction rate by a factor of up to 11 and the enan-
tioselectivity by 1–8% [71]. Assuming that Pt/Al2O3 and Rh/Al2O3
behave similarly in this respect, we can conclude that the striking
variation in reaction rates in Fig. 7a cannot simply be considered
as a particle size effect but it is mainly due to the different extent
of catalyst deactivation. On the other hand, the changes in enan-
tioselectivity are probably less affected by the impurities.
To prove this assumption we repeated the experiments shown
in Fig. 7 at higher pressures. High surface hydrogen concentra-
tion suppresses the initial step of aldol reaction, the abstraction
of an α-H atom on the metal surface. In addition, it accelerates
the hydrogenation reaction while the rate of the base-catalyzed
aldol reaction remains the same. The hydrogenation experiments
carried out at 10 and 100 bar (Fig. 9) reveal that variation of
the ee with Rh particle size is smaller but still considerable. Hy-
drogenations at 100 bar give 4–8% higher ees than the values at
10 bar, and full conversion was achieved with all catalysts in 1 h,
while at 10 bar the conversions were in the range 70–100%. In
the best case 73% ee was obtained with the highest Rh loading (at
1.65 nm Rh mean particle size). It has been shown that high pres-
sure (high surface hydrogen concentration) improves the enantios-
electivity in the hydrogenation of activated ketones on supported
Rh [5,15].
Fig. 10. Influence of mean Rh particle size on the enantioselectivity in the hy-
drogenation of ethyl 3-methyl-2-oxobutyrate with cinchonidine-modified Rh/Al2O3
◦
catalysts; standard conditions, toluene, 10 bar, 24 C.
Our interpretation of the results in Figs. 7–9 is that the perfor-
mance of Rh/Al2O3 is strongly distorted at atmospheric pressure
by catalyst deactivation, but the similar enantioselectivities at 10
and 100 bar indicate that at high pressure the variation of ee is
mainly due to an intrinsic particle size effect and the contribution
of strongly adsorbed impurities is probably minor.
In the hydrogenation of ethyl 3-methyl-2-oxobutyrate at 10 bar
in the presence of CD the difference between the highest and low-
est ee was only 10% (Fig. 10). The conversion in 1 h reaction time
increased from 70 to 100% with increasing particle size.
The potential of the flame spray catalyst synthesis is illustrated
by the best enantioselectivities achieved in the hydrogenation
of ethyl pyruvate and ethyl 3-methyl-2-oxobutyrate: the 4.3 wt%
Rh/Al2O3 catalyst gave 24 and 17% higher ees, respectively, than
those provided by the best commercial catalyst (Engelhard 4759)
under the same reaction conditions [15].
3.4. Origin of particle size effect
As mentioned in the introduction, there are numerous catalytic
and theoretical studies addressing the particle size effect in het-
erogeneous catalysis [16,17]. Various explanations have been for-
warded, putting the emphasis on the electronic [18] or the ge-
ometric [72] parameters. Regarding the electronic properties, the
critical size above which the band structure appears is about 2 nm.
At a lower size, the d-band becomes narrower with the appear-
ance of discrete levels, leading to electron deficiency of small
particles [17,73], where the small metal particles bind their core
electrons more tightly. Regarding the geometrical features, the ide-
alized topology of the surface shows big changes up to about 5 nm
[20], due to variations in the proportion of facets, edges, steps,
corners, and defects. The metal–support interaction may also play

230
F. Hoxha et al. / Journal of Catalysis 261 (2009) 224–231
synthesis of Rh/Al2O3 catalysts by flame spray pyrolysis at high
temperature. These ionic Rh species at the metal–oxide interface
(boundary or “adlineation” sites [76]) may also contribute to the
striking particle size effect depicted in Fig. 7. Evaluation of this
possibility needs an independent evidence for the existence of Rh
aluminate in the flame made catalysts.
4. Conclusions
The effect of metal particle size on the catalyst performance, or
more general the structure sensitivity in heterogeneous catalysis, is
an extensively studied—and debated—field. It is still a challenging
task to localize the origin of the phenomenon and separate it from
the disturbance caused by, for example, the different impurities in
the catalyst systems compared.
Here we applied the flame spray pyrolysis technique to pre-
pare a series of Rh/Al2O3 catalysts containing 0.5 to 4.3 wt% Rh.
This method provided very small Rh particles with relatively nar-
row particle size distributions and the best catalyst among them
(4.3 wt% Rh) possessed outstanding enantioselectivity in the enan-
tioselective hydrogenation of α-ketoesters. Still, the dramatic drop
of reaction rate and enantioselectivity with decreasing particle size
(Rh content) cannot simply be attributed to a “classical” particle
size effect. Hydrogenation of the substrate in the apolar medium
was accompanied by the base-catalyzed aldol condensation lead-
ing to strongly adsorbing, large molecular weight byproducts and
the extent of this side reaction depends also on the Rh loading.
Working at 100 bar we could accelerate the target reaction and
minimize the contribution of surface impurities. Under these con-
ditions variation of enantioselectivity with the Rh loading (mean
particle size) was still considerable. We assume that the low ef-
ficiency of very small Rh particles is mainly due to a geometric
effect: these particles cannot adopt the bulky cinchona alkaloid-
α-ketoester complex, the formation of which is assumed to be a
necessary prerequisite to achieve enantioselection. It seems no ad-
vantage of using metal nanoparticles with a particle size of about
2 nm or below, in line with the conclusion of some former studies
in enantioselective hydrogenation over Pt [31] and Ir [34].
Fig. 11. Two idealized Rh cuboctahedron clusters containing 55 and 309 atoms, and
the space-filled models of quinine and ethyl pyruvate; all represented with the same
scale.
a crucial role in controlling the structure and properties of small
metal particles [74–76] and these effects cannot easily be separated
from those mentioned previously.
The activity and enantioselectivity of chirally modified metal
catalysts in the hydrogenation of C=O bonds is generally attributed
to the presence of metallic surface sites [26,70,77,78]. There is
less agreement in the real nature of substrate–modifier interac-
tions leading to enantioselection. In most cases, the mechanistic
models assume the adsorption of the chiral modifier and the ke-
tone substrate on neighboring surface sites. Due to the bulkiness
of cinchona alkaloids, the substrate–modifier complex would oc-
cupy an ensemble of 20–25 metal atoms [79,80]. In the present
study the lowest and the highest average metal particle size of
Rh/Al2O3 catalysts are 0.96 and 1.65 nm (Fig. 2), and the presence
Acknowledgment
Financial support by the Swiss National Science Foundation
(Project No.: 200020-113427) is kindly acknowledged.
References
◦
of surface Rh oxide after prereduction at 400 C can be excluded
based on the DRIFT measurements. Fig. 11 shows two idealized Rh
cuboctahedral particles with 55 and 309 atoms, corresponding to
a diameter of ca. 1.0 and 1.6 nm, respectively. The latter crystal-
lite seems to be sufficiently big (25 surface atoms on one face)
to accommodate the quinine–ethyl pyruvate interacting complex,
which molecules are also presented in Fig. 11 in the same scale.
A more detailed consideration is not yet possible, since the mech-
anism of the reaction on Rh has not been investigated. Assuming
similar reaction mechanisms on Pt and Rh, it is very probable that
the steric limitation to adsorption of the substrate–modifier com-
plex parallel to the metal surface on small Rh particles plays a
role in the low activity and enantioselectivity of Rh/Al2O3 catalysts
with low Rh loading. According to this assumption, only those Rh
particles would give good enantioselectivity, which are big enough
to accommodate the substrate–modifier complex, while on smaller
particles the enantioselective control of the hydrogenation of the
carbonyl group is less efficient or it is not possible at all (leading
to racemic product).
[1] R. Hess, F. Krumeich, T. Mallat, A. Baiker, J. Mol. Catal. A Chem. 212 (2004) 205.
[2] Y.L. Huang, Y.Z. Li, J.Y. Hu, P.M. Cheng, H. Chen, R.X. Li, X.J. Li, C.W. Yip, A.S.C.
Chan, J. Mol. Catal. A Chem. 189 (2002) 219.
[3] H.X. Ma, H. Chen, Q. Zhang, X.J. Li, J. Mol. Catal. A Chem. 196 (2003) 131.
[4] M. Maris, T. Mallat, A. Baiker, J. Mol. Catal. A Chem. 242 (2005) 151.
[5] W. Xiong, H.X. Ma, Y.Y. Hong, H. Chen, X.J. Li, Tetrahedron: Asymmetry 16
(2005) 1449.
[6] E. Toukoniitty, S. Franceschini, A. Vaccari, D.Y. Murzin, Appl. Catal. A 300 (2006)
147.
[7] M. Maris, D. Ferri, L. Konigsmann, T. Mallat, A. Baiker, J. Catal. 237 (2006) 230.
[8] M. Schürch, N. Künzle, T. Mallat, A. Baiker, J. Catal. 176 (1998) 569.
[9] K. Balàzsik, K. Szöri, K. Felföldi, B. Török, M. Bartók, Chem. Commun. (2000)
555.
[10] K. Szo˝ri, K. Balázsik, K. Felföldi, M. Bartók, J. Catal. 241 (2006) 149.
[11] S. Diezi, S. Reimann, N. Bonalumi, T. Mallat, A. Baiker, J. Catal. 239 (2006) 255.
[12] O.J. Sonderegger, G.M.W. Ho, T. Bürgi, A. Baiker, J. Catal. 230 (2005) 499.
[13] E. Schmidt, D. Ferri, A. Baiker, Langmuir 23 (2007) 8087.
[14] E. Schmidt, D. Ferri, A. Vargas, A. Baiker, J. Phys. Chem. C 112 (2008) 3866.
[15] F. Hoxha, T. Mallat, A. Baiker, J. Catal. 248 (2007) 11.
[16] B. Coq, F. Figueras, Coord. Chem. Rev. 180 (1998) 1753.
[17] M. Che, C.O. Bennett, Advances in Catalysis, vol. 36, Academic Press, San Diego,
1989, p. 55.
+
The Rh species, detected by the DRIFT measurement of CO
adsorption, may indicate the formation of Rh aluminate during the
[18] L. Guczi, G. Petö, A. Beck, Z. Pàszti, Top. Catal. 29 (2004) 129.
[19] O.M. Poltorak, V.S. Boronin, Int. Chem. Ing. 7 (1967) 452.

F. Hoxha et al. / Journal of Catalysis 261 (2009) 224–231
231
[20] R. Van Hardeveld, F. Hartog, Surf. Sci. 15 (1969) 189.
[21] B. Coq, F. Figueras, in: A.S. Wieckowski, E.R. Savinova, C.G. Vayenas (Eds.), Catal-
ysis and Electrocatalysis at Nanoparticle Surfaces, Marcel Dekker, New York,
2003, p. 847.
[22] M. Boudart, in: Advances in Catalysis, Academic Press, San Diego, 1969, p. 20.
[23] G.C. Bond, Appl. Catal. A 149 (1997) 3.
[47] O. Dulaurent, K. Chandes, C. Bouly, D. Bianchi, J. Catal. 192 (2000) 262.
[48] A. Maroto-Valiente, I. Rodriguez-Ramos, A. Guerrero-Ruiz, Catal. Today 93–95
(2004) 567.
[49] S. Trautmann, M. Baerns, J. Catal. 150 (1994) 335.
[50] J.T. Yates, T.M. Duncan, S.D. Worley, R.W. Vaughan, J. Chem. Phys. 70 (1979)
1219.
[24] S. Azad, M. Kaltchev, D. Stacchiola, G. Wu, W.T. Tysoe, J. Phys. Chem. B 104
(2000) 3107.
[51] J. Rasko, J. Bontovics, Catal. Lett. 58 (1999) 27.
[52] J.P. Wey, W.C. Neely, S.D. Worley, J. Catal. 134 (1992) 378.
[53] P. Basu, D. Panayotov, J.T. Yates, J. Am. Chem. Soc. 110 (1988) 2074.
[54] P. Basu, D. Panayotov, J.T. Yates, J. Phys. Chem. 91 (1987) 3133.
[55] F. Solymosi, M. Pàsztor, J. Phys. Chem. 89 (1985) 4789.
[56] H.F.J. Van’t Blik, J. Vanzon, T. Huizinga, J.C. Vis, D.C. Koningsberger, R. Prins,
J. Am. Chem. Soc. 107 (1985) 3139.
[57] S.S.C. Chuang, S. Debnath, J. Mol. Catal. 79 (1993) 323.
[58] C.P. Hwang, C.T. Yeh, Q.M. Zhu, Catal. Today 51 (1999) 93.
[59] C. Wong, R.W. McCabe, J. Catal. 119 (1989) 47.
[60] K. Dohmae, T. Nonaka, Y. Seno, Surf. Interface Anal. 37 (2005) 115.
[61] H.C. Yao, W.G. Rothschild, J. Chem. Phys. 68 (1978) 4774.
[62] D. Duprez, J. Barrault, C. Geron, Appl. Catal. 37 (1988) 105.
[63] D. Duprez, G. Delahay, H. Abderrahim, J. Grimblot, J. Chim. Phys. Phys.-Chim.
Biol. 83 (1986) 465.
[64] J.C. Summers, S.A. Ausen, J. Catal. 58 (1979) 131.
[65] A. Borodzinski, M. Bonarowska, Langmuir 13 (1997) 5613.
[66] M. von Arx, T. Mallat, A. Baiker, Top. Catal. 19 (2002) 75.
[67] D. Ferri, S. Diezi, M. Maciejewski, A. Baiker, Appl. Catal. A 297 (2006) 165.
[68] Z.M. Liu, X.H. Li, P.L. Ying, Z.C. Feng, C. Li, J. Phys. Chem. C 111 (2007) 823.
[69] J.M. Bonello, R.M. Lambert, N. Künzle, A. Baiker, J. Am. Chem. Soc. 122 (2000)
9864.
[25] R.L. Augustine, S.K. Tanielyan, L.K. Doyle, Tetrahedron: Asymmetry
1803.
4 (1993)
[26] D.Y. Murzin, P. Maki-Arvela, E. Toukoniitty, T. Salmi, Catal. Rev.-Sci. Eng. 47
(2005) 175.
[27] D.Y. Murzin, E. Toukoniitty, React. Kinet. Catal. Lett. 90 (2007) 19.
[28] X. Zuo, H. Liu, D. Guo, X. Yang, Tetrahedron 55 (1999) 7787.
[29] Y. Nitta, T. Kubota, Y. Okamoto, Bull. Chem. Soc. Jpn. 74 (2001) 2161.
[30] Y. Nitta, J. Watanabe, T. Okuyama, T. Sugimura, J. Catal. 236 (2005) 164.
[31] J.T. Wehrli, A. Baiker, M. Monti, H.U. Blaser, J. Mol. Catal. 61 (1990) 207.
[32] G.A. Attard, K.G. Griffin, D.J. Jenkins, P. Johnston, P.B. Wells, Catal. Today 114
(2006) 346.
[33] J.L. Margitfalvi, E. Tàlas, L. Yakhyaeva, E. Tfirst, I. Bertóti, L. Tóth, in: D.G. Morrell
(Ed.), Catalysis of Organic Reactions, Dekker, New York, 2003, p. 393.
[34] T. Marzialetti, M. Oportus, D. Ruiz, J.L.G. Fierro, P. Reyes, Catal. Today 133 (2008)
711.
[35] M.A. Keane, Can. J. Chem. 72 (1994) 372.
[36] N. van Vegten, D. Ferri, M. Maciejewski, F. Krumeich, A. Baiker, J. Catal. 249
(2007) 269.
[37] R. Strobel, F. Krumeich, W.J. Stark, S.E. Pratsinis, A. Baiker, J. Catal. 222 (2004)
307.
[38] S.J. Pennycook, Ultramicroscopy 30 (1989) 58.
[70] M. Studer, H.U. Blaser, C. Exner, Adv. Synth. Catal. 345 (2003) 45.
[71] H.U. Blaser, H.P. Jalett, F. Spindler, J. Mol. Catal. A Chem. 107 (1996) 85.
[72] C.O. Bennett, M. Che, J. Catal. 120 (1989) 293.
[39] A. Urakawa, A. Baiker, manuscript in preparation.
[40] P.J. Levy, V. Pitchon, V. Perrichon, M. Primet, M. Chevrier, C. Gauthier,
J. Catal. 178 (1998) 363.
[73] T. Stace, Nature 331 (1988) 116.
[41] A.C. Yang, C.W. Garland, J. Phys. Chem. 61 (1957) 1504.
[42] N. Kaufherr, M. Primet, M. Dufaux, C. Naccache, C. R. Hebdomadaires Des
Seances Acad. Sci. Ser. C 286 (1978) 131.
[43] G.M. Alikina, A.A. Davydov, I.S. Sazanova, V.V. Popovskii, React. Kinet. Catal.
Lett. 27 (1985) 279.
[74] C.R. Henry, Surf. Sci. Rep. 31 (1998) 235.
[75] G. Haller, D.E. Resasco, Adv. Catal. 36 (1989) 173.
[76] K. Hayek, R. Kramer, Z. Paal, Appl. Catal. A 162 (1997) 1.
[77] T. Mallat, E. Orglmeister, A. Baiker, Chem. Rev. 107 (2007) 4863.
[78] M. Bartók, Curr. Org. Chem. 10 (2006) 1533.
[44] C.A. Rice, S.D. Worley, C.W. Curtis, J.A. Guin, A.R. Tarrer, J. Chem. Phys. 74 (1981)
6487.
[45] P.B. Rasband, W.C. Hecker, J. Catal. 139 (1993) 551.
[46] R.R. Cavanagh, J.T. Yates, J. Chem. Phys. 74 (1981) 4150.
[79] A. Vargas, T. Bürgi, A. Baiker, J. Catal. 226 (2004) 69.
[80] K.E. Simons, P.A. Meheux, S.P. Griffiths, I.M. Sutherland, P. Johnston, P.B. Wells,
A.F. Carley, M.K. Rajumon, M.W. Roberts, A. Ibbotson, Recl. Trav. Chim. Pays-
Bas 113 (1994) 465.