Supported zirconium propoxide - A versatile heterogeneous catalyst for the Meerwein-Ponndorf-Verley reduction

Article abstract of DOI:10.1016/S0021-9517(03)00160-X

Grafting of zirconium 1-propoxide on SBA-15 resulted in highly active catalysts for the MPV reduction. The activity increased with zirconium loading up to a monolayer coverage. In most cases, there were no side products other than the desired alcohol. Electron-donating groups adjacent to the carbonyl group in the substrate facilitate the reaction. The grafted zirconium catalysts did not lose their activity in the presence of moisture or on exposure to ambient atmosphere, making them easy to handle and reuse. No leaching of the grafted zirconium 1-propoxide into the reaction mixture was observed. The addition of pyridine and water to the reaction medium had only a small effect on its activity while benzoic acid led to severe deactivation. The deactivation is attributed to strong adsorption of benzoic acid at the Zr metal centres which could be reversed on removal of the poison. Aluminum 2-propoxide grafted on SBA-15 resulted in a less active catalyst than the zirconium catalysts. The good resistance to hydrolysis of the zirconium catalysts makes them superior to the aluminum 2-propoxide catalysts.

Full text of DOI:10.1016/S0021-9517(03)00160-X

Journal of Catalysis 218 (2003) 396–404
www.elsevier.com/locate/jcat
Supported zirconium propoxide—a versatile heterogeneous catalyst
for the Meerwein–Ponndorf–Verley reduction
Yongzhong Zhu, S. Jaenicke, and G.K. Chuah ∗
Department of Chemistry, National University of Singapore, Kent Ridge, Singapore 119260, Singapore
Received 13 December 2002; revised 2 April 2003; accepted 2 April 2003
Abstract
Grafting of zirconium 1-propoxide on SBA-15 resulted in highly active catalysts for the MPV reduction. The activity increased with
zirconium loading up to a monolayer coverage. In most cases, there were no side products other than the desired alcohol. Electron-donating
groups adjacent to the carbonyl group in the substrate facilitate the reaction. The grafted zirconium catalysts did not lose their activity in the
presence of moisture or on exposure to ambient atmosphere, making them easy to handle and reuse. No leaching of the grafted zirconium
1-propoxide into the reaction mixture was observed. The addition of pyridine and water to the reaction medium had only a small effect on
its activity while benzoic acid led to severe deactivation. The deactivation is attributed to strong adsorption of benzoic acid at the Zr metal
centres which could be reversed on removal of the poison. Aluminum 2-propoxide grafted on SBA-15 resulted in a less active catalyst than
the zirconium catalysts. The good resistance to hydrolysis of the zirconium catalysts makes them superior to the aluminum 2-propoxide
catalysts.
2003 Elsevier Inc. All rights reserved.
Keywords: Zirconium 1-propoxide; Aluminum 2-propoxide; Grafted catalysts; MPV reduction; 4-tert-Butylcyclohexanone; Poisoning; Water resistance
1. Introduction
hydride, has also been reported to be active in the MPV
reaction [6].
In order to hetereogenise the complexes used in MPV
reductions, grafting onto inorganic supports has been car-
ried out. Heterogeneous catalysts offer the advantage of
ease of separation and reusability. Anwander et al. [7]
used MCM-41 as a support material for grafting aluminum
propoxide. The hybrid system was found to be very active in
the MPV reduction of 4-tert-butylcyclohexanone. However,
the solvent 2-propanol must be carefully dried to observe
good catalytic activity. The enhanced activity was ascribed
to the formation of surface complexes involving four- or
five-coordinated geometrically distorted aluminum. Leyrit
et al. [8] reported that tetra-neopentylzirconium anchors to
the surface of silica via a covalent bond, giving mononu-
clear and well-dispersed (siloxy)tris(neopentyl)zirconium.
The heterogeneous complex was found to be an effective
catalyst for the MPVO reactions of a number of ketones and
alcohols.
The Meerwein–Ponndorf–Verley (MPV) reduction of
aldehydes and ketones provides a convenient access to the
corresponding alcohols under mild conditions. The hydride
donor is usually a secondary alcohol such as 2-propanol.
The reaction is chemoselective: under the conditions of
the MPV reduction, unsaturated aldehydes and ketones
are not reduced to the saturated alcohols [1]. Hence, the
MPV reduction offers a convenient synthesis route to α, β-
unsaturated alcohols, many of which are important starting
materials for the production of fine chemicals [2]. Tradition-
ally, the reaction is catalyzed homogeneously by aluminum
alkoxides such as aluminum 2-propoxide. Recently, a num-
ber of other aluminum complexes have been reported to
be highly active for the MPV reduction of carbonyl sub-
strates [3,4]. Precatalysts such as dimethylaluminum chlo-
ride or trimethylaluminum were shown to give superior
yields in the reaction [5]. Besides aluminum complexes,
the zirconium complex, bis(cyclopentadienyl)zirconium di-
We have found that suitably pretreated hydrous zirconia
is an active catalyst in the MPV reduction such as the for-
mation of cinnamyl alcohol from cinnamaldehyde [9]. The
calcination temperature of the hydrous zirconia was found
to be an important parameter as it affects the density of
*
Corresponding author.
E-mail address: chmcgk@nus.edu.sg (G.K. Chuah).
0021-9517/03/$ – see front matter
2003 Elsevier Inc. All rights reserved.
doi:10.1016/S0021-9517(03)00160-X

Y. Zhu et al. / Journal of Catalysis 218 (2003) 396–404
397
hydroxyl groups, necessary for ligand exchange with the
reductant, 2-propanol. In this work, we extend our studies
to the use of zirconium 1-propoxide grafted on a number
of high surface area supports with a range of well-defined
pores. Zirconium propoxide is less moisture sensitive than
aluminum 2-propoxide and this property makes it easier to
handle. Grafting on high surface area supports offers good
dispersion of zirconium propoxide. For the present study,
we evaluated various mesoporous silicas as support, namely
SBA-15, MCM41, MCM-48, and a commercial silica gel.
SBA-15 has hexagonal mesopores of 50–90 Å, depending
on the preparation conditions [10] while MCM-41 has meso-
pores of 30–40 Å [11]. MCM-48 has a cubic structure of
interconnected channels with average pore diameter of 20–
30 Å and a very high surface area of 1600 m2/g [12,13]. The
commercial silica gel 60 has a wider pore size distribution
centered around 60 Å. The grafted catalysts were tested to-
ward the chemoselective MPV reduction of a number of ke-
tones and aldehydes. Some of the substrates and the reduced
products are used in the fragrance industry [14]. For exam-
ple, citral and its reduction products, geraniol, citronellol,
and 3,7-dimethyloctan-1-olare used as fragrance and flavour
chemicals. Upon reduction, 4-tert-butylcyclohexanone is re-
duced to cis- and trans-4-tert-butylcylcohexanol which can
be further acetylated to yield 4-tert-butylcyclohexyl acetate,
used in soap perfumes.
(Aldrich) was added to 75 cm3 of 1.6 M HCl solution. The
mixture was stirred at 40 ◦C for 3 h until all P123 had dis-
solved. Next, 4.25 g of TEOS was added to the solution and
the mixture stirred for another 24 h. The composition of the
mixture was 1.0 TEOS:5.71 HCl:158 H2O. The resulting gel
was placed in a Teflon-lined autoclave and heated at 100 ◦C
for 24 h. The solid product was filtered, washed, and cal-
cined as above.
2.2. Preparation of catalysts
The grafting of zirconium 1-propoxide (Alfa) onto the
supports was carried out following Ref. [15]. Prior to graft-
ing, the support (1.0 g) was dried at 300 ◦C for 4 h before
cooling down to 100 ◦C. It was then added to a solution of
zirconium 1-propoxide in dry hexane (30 mL). The suspen-
sion was refluxed at 69 ◦C for 12 h. The concentration of
zirconium 1-propoxide in hexane was varied to give 5, 7, 11,
and 14 wt% Zr in the grafted sample. The product was re-
covered by filtration and washed with hexane to remove any
unreacted precursor. It was transferred to a vacuum desicca-
tor and dried under vacuum for 6 h. Aluminum 2-propoxide
(Fluka) was similarly grafted onto SBA-15 to give 3.74 wt%
Al. This Al loading should result in the same molar con-
tent of active sites as found in the 11 wt% Zr-grafted sample
(1.4 mmol/g support). The samples are referred to as wM-S
where w is the wt% Zr or Al (M) and S is the support SBA-
15, MCM-41, MCM-48, or silica gel 60 (Merck).
2. Experimental
2.3. Characterisation
2.1. Synthesis of supports
The surface area, pore size distribution, and pore vol-
ume were determined by nitrogen adsorption using a Quan-
tachrome NOVA 2000. The crystalline phase of the samples
was determined by powder X-ray diffraction. A Siemens
D5005 diffractometer (Cu anode operated at 40 kV and
40 mA) equipped with variable slits was used for wide-
angle measurements of all materials except SBA-15. The
crystallinity of SBA-15 was determined with a Bruker D8
GADDS at a sample-to-detector distance of 30 cm using Cu
radiation. FT-IR spectra were made to ascertain that zirco-
nium 1-propoxide had been grafted onto the support. SBA-
15 and the grafted samples were pressed into self-supporting
wafers of ∼ 8–10 mg. The wafer was mounted in a Pyrex IR
cell with NaCl windows. The sample was dried by evacu-
ating under vacuum (10−3 mbar) for 2 h at 200 ◦C. Subse-
quently, the sample was cooled to room temperature before
making an IR measurement. A Bruker Equinox 55 spectrom-
eter was used with a resolution of 2 cm−1. The elemental
compositions of the grafted catalysts were determined by
AES-ICP after dissolution of the solid in HF.
Silica MCM-41 was synthesised by dissolving 2.1 g of
tetraethylammonium hydroxide (TEAOH, 40%, Fluka) and
2.95 g cetyltrimethylammonium bromide (CTMAB, Merck)
in 12.2 g of deionised water with heating and stirring un-
til a clear solution was obtained. Fumed silica (1.8 g) was
added to the solution with stirring for another 2 h. The gel
was aged for 24 h at room temperature. The molar composi-
tion of the final gel mixture was 1.0 SiO2:0.19 TEAOH:0.27
CTMAB:40 H2O. The mixture was placed in a Teflon-lined
stainless-steel autoclave and kept at 150 ◦C for 48 h under
autogeneous pressure. The product was filtered, washed with
deionised water, dried at 100 ◦C, and calcined at 550 ◦C for
12 h.
Silica MCM-48 was synthesised following Ref. [12]. In a
typical synthesis, 10 cm3 of tetraethoxysilane (TEOS, Fluka)
was added to an aqueous solution containing 8.8 g of a
10% CTMAB solution and 10 cm3 of 2 M NaOH. The mo-
lar composition of the final gel mixture was 1.0 SiO2:0.46
NaOH:0.55 CTMAB:112 H2O. The mixture was stirred for
30 min before it was left for crystallisation in a Teflon-lined
stainless-steel autoclave at 100 ◦C for 72 h. The product was
washed and calcined as above.
2.4. Catalytic activity
Silica SBA-15 was synthesised according to the proce-
dure described in Ref. [10]. Typically, 2 g of Pluronic P123
The reaction mixture containing 1.3 mmol of the alde-
hyde or ketone and 5 g (83 mmol) of 2-propanol was placed

398
Y. Zhu et al. / Journal of Catalysis 218 (2003) 396–404
in a round-bottomed flask equipped with a septum port, re-
flux condensor, and a guard tube. To this mixture was added
100 mg of freshly vacuum-dried catalyst. The MPV reduc-
tion was carried out under reflux at 82 ◦C. Aliquots were
removed at different reaction times and the products were
analyzed by gas chromatography(HP5, 0.25 mm, 30 m). The
identity of the products was verified by comparing the reten-
tion times and GCMS spectra with authentic samples.
To test for any leaching, about 100 mg of the grafted cat-
alyst was refluxed in 5 g of 2-propanol at 82 ◦C for 6 h.
The solution was filtered and the filtrate was tested for ac-
tivity in the MPV reduction of 4-tert-butylcyclohexanone.
The reuse of catalyst was also tested by washing the used
catalyst with 2-propanol and testing it in further rounds of re-
action. Poisoning of the grafted catalyst was investigated by
adding 0.1098 mmol (equivalent to the Zr content in 100 mg
of 11Zr-SBA-15) of water, pyridine, or benzoic acid to the
reaction medium. The “poisoned” catalyst was filtered after
the reaction, washed with 2-propanol, and tested for any re-
covery of activity in a subsequent MPV reaction.
The moisture sensitivity of the grafted zirconium 1-prop-
oxide and aluminum 2-propoxidewas investigated by expos-
ing the samples to ambient environment (humidity ∼ 80%)
for 48 h before testing for their activity in the MPV reduc-
tion of 4-tert-butylcyclohexanone. Homogeneous catalysis
by zirconium 1-propoxide and aluminum 2-propoxide was
carried out to compare the activity with the grafted cata-
lysts. Forty-nine microliters of 70% zirconium 1-propoxide
(containing 0.1098 mmol Zr) or 0.0238 g of aluminum
2-propoxide (0.1386 mmol) was added to the reaction
medium of 4-tert-butylcyclohexanone and 2-propanol. At
different reaction times, an aliquot was removed, water was
added to precipitate the catalyst, and the filtrate was analysed
by gas chromatography.
Table 1
Chemical and textural properties of the catalysts used in the study
Sample
Zr or Al [%]
a
Surface
Pore
Mean
Expected
Measured
area
2
volume
3
pore Ø
(Å)
(m /g) (cm /g)
SBA-15
–
5
7
11
14
4
–
3.8
5.0
10.0
11.4
3.75
–
816
764
630
606
570
545
1136
784
1630
1342
307
1.20
1.13
0.92
0.82
0.70
0.90
1.01
0.63
1.24
0.71
0.77
0.62
73
70
65
60
58
65
33
31
26
22
64
57
5Zr-SBA-15
7Zr-SBA-15
11Zr-SBA-15
14Zr-SBA-15
3.7Al-SBA-15
MCM-41
11Zr-MCM-41
MCM-48
11Zr-MCM-48
–
11
–
11.0
–
11
–
10.5
–
SiO
2
11Zr-SiO
11
5.5
295
2
a
From alkoxide used in grafting.
to 0.70 cm3/g in 14Zr-SBA-15. Silica gel has a wide pore
diameter of 64 Å (Fig. 2). After grafting, the mean pore di-
mension in 11Zr-SiO2 was 57 Å, and the total pore volume
decreased from 0.77 to 0.62 cm3/g.
In contrast to SBA-15 and silica gel supports, the reduc-
tion in the pore volume of zirconium 1-propoxide grafted
on MCM-41 and MCM-48 was bigger, ∼ 40%. Based on
a trimeric structure for zirconium 1-propoxide, a molec-
ular mechanics (MM+; Hyperchem) calculation gives a
kinetic diameter of about 17.9 Å. The size of the zirco-
nium 1-propoxide complex could hinder its penetration into
smaller pores; at the same time, bigger pores could be com-
pletely filled up or considerably reduced in size after graft-
ing. Hence, the loss in pore volume is more drastic than for
bigger pore material, such as SBA-15 or silica gel.
3. Results and discussion
3.1. Textural properties
The experimentally determined zirconium loading on
the SBA-15-grafted samples was between 75 and 92% of
the intended values (Table 1). The slightly lower than ex-
pected values can be due to limited accessibility of some
of the surface hydroxyl groups during the grafting reaction.
Fig. 1 shows the nitrogen adsorption-desorption isotherms
for SBA-grafted samples with different loading of zirconium
1-propoxide. The isotherms were all of IUPAC type IV. The
H1 hysteresis loop is indicative for agglomerates or com-
pacts of spheroidal particles of fairly uniform size and ar-
ray [16]. The mean pore diameter of SBA-15 was 73 Å,
while that of the grafted SBA samples was around 58–70 Å.
The reduced pore diameter together with the retention of the
uniform pore distribution shows that there is an even dis-
persion of zirconium 1-propoxide on the pore surface. The
total pore volume decreased from 1.20 cm3/g in SBA-15
Fig. 1. Nitrogen adsorption/desorption isotherms for (a) SBA-15, (b) 5Zr-
SBA-14, (c) 7Zr-SBA-15, and (d) 11Zr-SBA-15.

Y. Zhu et al. / Journal of Catalysis 218 (2003) 396–404
399
Fig. 2. Pore-size distribution for (!) support and (") 11 wt% Zr-grafted catalysts on (a) MCM-41, (b) MCM-48, (c) SBA-15, and (d) SiO .
2
X-ray powder diffraction was used to confirm the crys-
talline phase of the different supports. The X-ray powder
diffraction patterns for pure silica SBA-15 support and the
grafted samples are shown in Fig. 3. SBA-15 shows one
strong primary peak at 2θ ∼ 0.85◦ corresponding to the
(100) reflection, and two smaller higher order peaks indexed
as the (110) and (200) reflections at 1.5 and 1.75◦. The pres-
ence of these reflections indicates a high degree of order
in the arrangement of the pore network. The X-ray diffrac-
tograms of the grafted samples were nearly identical to that
of the support, verifying the retention of the pore structure
even after grafting.
The IR spectra of the pure silica SBA-15 and MCM-
41 have a broad absorption band in the OH stretching
range from 3200 to 3800 cm−1 (Fig. 4). The sharp peak at
3741 cm−1 originates from terminal silanol groups, and the
Fig. 3. XRD diffractograms of (a) SBA-15, (b) 5Zr-SBA-15, (c) 7Zr-
SBA-15, (d) 11Zr-SBA-15, and (e) 14Zr-SBA-15.
Fig. 4. IR spectra of (a) SBA-15, (b) 7Zr-SBA-15, (c) MCM-41, and
(d) 11Zr-MCM-41.

400
Y. Zhu et al. / Journal of Catalysis 218 (2003) 396–404
broad shoulder at the low frequency side is assigned to OH
groups located at defect sites [17]. Upon grafting of zirco-
nium 1-propoxide onto the support, the vibration band due to
silanols at 3741 cm−1 was reduced in intensity. This clearly
points to the role of surface silanols in reacting with the zir-
conium 1-propoxide. The grafted samples show additional
peaks at 2963 and 2905 cm−1, which can be assigned to C–H
stretching of the propoxide.
3.2. Catalytic activity
Unmodified SBA-15 was not active in the MPV reduc-
tion of 4-tert-butylcylcohexanone.However, once zirconium
1-propoxide was grafted onto SBA-15, the heterogenised
material showed good activity (Fig. 5). Almost all 4-tert-
butylcyclohexanone was reduced within 100–250 min, de-
pending on the catalyst. The rate of reaction increased
with zirconium propoxide loading; however, the most highly
loaded samples, 11Zr-SBA and 14Zr-SBA, had very similar
activity.
4-tert-butylcyclohexanone was reduced to the corre-
sponding trans- and cis-alcohol. The trans:cis ratio was
84:16. No other reduction products were detected, show-
ing the high selectivity of the reaction. 1-Propanol was also
detected in the reaction mixture immediately after the start
of the reaction, indicating that the 1-propoxide ligand was
readily exchanged with excess 2-propanol. Based on the
Zr loading (from ICP results) and the 1-propanol detected
(by GC), a Zr:1-propanol ratio of 1:2 was obtained. Hence,
zirconium 1-propoxide was immobilised to the surface via
reaction with two OH groups, leaving two propoxide groups
at the metal centre.
Scheme 1. MPV reduction of citral (a and b) to geraniol and nerol.
11Zr-SBA was 15.8 and 80.1%, respectively. From the tem-
perature dependence, the activation energy was calculated
to be 59.5 kJ/mol. This compares well with the value of
51.7 kJ/mol reported by Shibagaki et al. for cyclohexanone
reduction [18].
For comparison, the activity of the homogeneously cata-
lysed reaction was tested using an equivalent amount of
zirconium 1-propoxide as found in 11Zr-SBA-15. Here,
the conversion of 4-tert-butylcyclohexanone was 32% after
120 min. The rate of reaction over the supported catalyst
was faster: the conversion reached 96% in the same time.
The selectivity to the trans-alcohol in the homogeneously-
catalysed reaction was very high, 91% as compared to 84%
in the grafted catalyst.
Citral occurs as cis- and trans-isomers, known as neral
and geranial, respectively. Scheme 1 outlines the reduction
of citral to the corresponding alcohols using 2-propanol
as reductant. Although the rate of reaction was slower in
the reduction of citral, a similar trend to that of 4-tert-
butylcyclohexanone was observed (Fig. 6). The rate in-
creased with Zr loading up to a limit with both 11Zr-SBA-15
The reaction was also carried out at lower tempera-
tures. At 25 and 60 ◦C, the conversion after 300 min over
Fig. 5. Conversion of 4-tert-butylcyclohexanone over (") 5Zr-SBA-15,
(F) 7Zr-SBA-15, (2) 11Zr-SBA-15, and (Q) 14Zr-SBA-15.
Fig. 6. Conversion of citral over (") 5Zr-SBA-15, (F) 7Zr-SBA-15,
(2) 11Zr-SBA-15, and (Q) 14Zr-SBA-15.

Y. Zhu et al. / Journal of Catalysis 218 (2003) 396–404
401
and 14Zr-SBA-15 having a similar reaction rate. Further in-
Table 2
Substrates used for the MPV reduction with 2-propanol over 11Zr-SBA-15
crease in Zr loading does not lead to significant increase in
the rate. Over the catalysts, the selectivity to nerol and geran-
iol was more than 90%. Some isomerisation of the formed
alcohols occurred, leading to the by-products, 3,7-dimethyl
1,6-octadien-3-ol and 2,7-dimethyl 2,6-octadien-1-ol, as de-
tected by GCMS.
The observation that the activity of the catalysts increases
with zirconium loading to a limiting value suggests that
the surface coverage of zirconium 1-propoxide approached
a monolayer at 10–11.4 wt% Zr. Based on the Zr loading
and the BET surface area (816 m2/g), the surface coverage
works out to be 1.11–0.96 nm2/Zr propoxide. From the cal-
culated kinetic diameter for trimeric zirconium 1-propoxide
of 1.79 nm, a value of 0.84 nm2/Zr propoxide is obtained.
The good fit of the experimental and calculated values
strongly supports the assumption that a monolayer is at-
tained at these Zr loadings. Once a monolayer is formed
in a grafting process, the loading can only be increased if
the sample is calcined to regenerate hydroxyl groups for the
grafting reaction [19].
Entry
Substrate
Time
(h)
Conversion
(%)
Selectivity
(%)
1
5
16.9
100
Cyclopentanone
Cyclohexanone
2
3
5
5
98.1
100
100
100
4-tert-Butylcyclohexanone
4
5
18.3
100
Acetophenone
5
6
5
23.8
< 1
100
–
4-Chloroacetophenone
Benzyl methyl ketone
24
The results from the MPV reduction of a number of
aldehydes and ketones over 11Zr-SBA catalyst are given in
Table 2. Cyclopentanone was less easily reduced than cy-
clohexanone. Cyclohexanone was reduced at the same rate
as 4-tert-butylcyclohexanone despite the bulky tertiary butyl
group on the latter. The pore size of the supported catalyst
did obviously not restrict site accessibility for the 4-tert-
butylcyclohexanone. However, both adamantone and benzyl
methyl ketone were not reduced to the corresponding alco-
hols under the reaction conditions. Steric hindrance due to
the polycyclic rings of adamantone could contribute to its
low activity. The low activity of benzyl methyl ketone as
compared to acetophenone and benzaldehyde may indicate
that the benzene ring is involved in binding to the active site.
If the benzene ring binds planar to the surface, the C=O
group in benzaldehyde and acetophenone will be aligned
parallel to the surface. However, this is not possible in ben-
zyl methyl ketone. The conversion of acetophenone (18.3%
after 5 h) is slightly lower than that of 4-chloroacetophenone
(23.8%). The para-substituted chlorine is too far away to in-
fluence the electron density at the keto group as indicated by
the 13C NMR shift. Hence, the slightly higher activity for
chloroacetophenone may be due to a resonance effect.
Aldehydes were efficiently reduced by 2-propanol with
very high selectivity. Benzaldehyde was almost completely
reduced within 1 h, while a conversion of 55.6% was mea-
sured for cinnamaldehyde after 5 h. Cinnamyl alcohol was
the only product in the MPV reduction of cinnamaldehyde
over the grafted catalysts. The conversion of heptanal and
octanal reached more than 95% within 5 h.
7
24
< 1
–
Adamantone
8
9
1
5
99
100
100
Benzaldehyde
55.6
Cinnamaldehyde
Heptanal
10
11
5
5
95.8
94.4
100
100
Octanal
12
1
53.0
> 90
Citral
1.3 mmol of substrate, 5 g of 2-propanol, 100 mg catalyst.
ever, the addition of benzoic acid drastically decreased the
rate to about one-ninth. The poisoning effect can be reversed
to some extent. Removal of the catalyst from the reaction
mixture and washing with 2-propanol caused a recovery of
the activity to about one-third that of the fresh catalyst. From
the results, it can be inferred that the benzoic acid did not
cause leaching of zirconium 1-propoxide from the support.
This was confirmed by FTIR measurements of the used cata-
lyst where the C–H absorption band of the propoxide groups
was still present. The addition of small amounts of water also
did not significantly reduce the activity of the catalyst. The
good resistance of the catalyst to the presence of moisture
was also verified by exposing the catalyst, 11Zr-SBA-15, to
the ambient environment for 48 h prior to catalyst testing.
The conversion of 4-tert-butylcyclohexanone after 300 min
was 97.5% as compared to 100% over the unexposed cata-
lyst (Table 3).
3.3. Poisoning experiments
In the presence of pyridine, the rate of reduction of 4-tert-
butylcyclohexanone was only a little reduced (Fig. 7). How-

402
Y. Zhu et al. / Journal of Catalysis 218 (2003) 396–404
Table 4
Recycling test of 11Zr-SBA-15 in the MPV reduction of 4-tert-butyl-
cyclohexanone
a
Cycle
Conversion (%)
Selectivity (%)
Trans:cis
1
2
3
4
5
6
100
100
100
100
99.5
99.3
100
100
100
100
100
100
84:16
84:16
85:15
84:16
85:15
85:15
a
After 300 min.
Table 5
Conversion of 4-tert-butylcyclohexanone over grafted zirconium 1-propo-
xide on different supports
a
Catalyst
Conversion (%)
trans:cis ratio of product
11Zr-MCM-41
11Zr-MCM-48
11Zr-SBA-15
86.6
74.3
59.8
57.3
80:20
79:21
84:16
89:11
Fig. 7. Conversion of 4-tert-butylcyclohexanone over 11Zr-SBA-15 in (2)
pure reaction medium, after addition of (Q) pyridine, (P) water, (") ben-
zoic acid, and (!) after removal of benzoic acid.
11Zr-SiO
2
a
Reaction time: 30 min.
1-propoxide (1.79 nm), a monolayer on the smaller surface
area silica should be reached at 5.26 wt% Zr. This result
lends support to the role of surface hydroxyl groups in lim-
iting the grafting to no more than a monolayer. We did not
expect that the mesoporous supports would have an effect on
the stereoselectivity of the reaction. However, the trans:cis
ratio was highest for the SiO2-supported catalyst, 89:11.
The trans:cis ratio of 4-tert-butylcyclohexanol decreased to
79:21 for 11Zr-MCM-48 which had the narrowest pore size
of the supports used. These results must to be compared to
the ratio 91:9 observed for the reaction in the homogeneous
phase.
Table 3
Activity of grafted aluminum- and zirconium-propoxide catalysts before
and after exposure to air
a
Catalyst
Conversion (%)
Selectivity (%)
trans:cis
11Zr-SBA-15
11Zr-SBA-15
100
97.5
38.1
100
100
100
100
84:16
84:16
78:22
78:22
b
3.7Al-SBA-15
3.7Al-SBA-15
a
b
c
10.5
After 300 min.
After exposure to air for 48 h.
After 24 h.
b
c
3.4. Reuse of grafted catalysts
3.6. Grafted aluminum 2-propoxide on SBA-15
The filtrate obtained after refluxing the grafted catalyst in
2-propanol was not active for the MPV reduction of 4-tert-
butylcyclohexanone, showing the absence of leaching into
the reaction medium. The grafted catalyst could be reused
with good activity and selectivity. After each round of re-
action, the catalyst was washed with 2-propanol and tested
in another batch reaction. Table 4 shows that in the MPV
reduction of 4-tert-butylcyclohexanone, the activity of the
catalyst, 11Zr-SBA-15, remained high even after six rounds
of reaction. The selectivity to the 4-tert-butylcyclohexanol
was 100% in each round.
Aluminum 2-propoxide is established as homogeneous
catalyst for MPV reactions [2]. For the comparison with
zirconium 1-propoxide, both homogeneous and heteroge-
neous forms of aluminum 2-propoxide were tested for
the reduction of 4-tert-butylcyclohexanone. As a homoge-
neous catalyst, aluminum 2-propoxide (0.1386 mmol) was
less active than zirconium 1-propoxide when compared
on an equivalent molar basis. The conversion of 4-tert-
butylcyclohexanoneafter 120 min was only 7.2%, increasing
to 15.2% after 300 min. The trans:cis selectivity of the al-
cohol was very similar to that over zirconium 1-propoxide,
91:9. After grafting aluminum 2-propoxide onto the SBA-
15 support, the conversion increased to 38.1% after 300
min (Table 3). The trans:cis ratio of the products was 78:22.
However, the activity of the grafted aluminum 2-propoxide
was still well below that of 11Zr-SBA-15 where the substrate
was almost completely converted in the same time.
3.5. Comparison of supports for grafting
The catalytic activity at the same loading, 10–11 wt% Zr,
was compared for MCM-41, MCM-48, and SBA-15 (Ta-
ble 5). All catalysts showed high activity in the reaction.
11Zr-SiO2 showed the slowest rate of reaction. This may
be explained by its lower Zr loading, 5.5 wt%, compared
to the other samples. This loading corresponds to a mono-
layer of Zr 1-propoxide. Based on the diameter of zirconium
One explanation for the lower activity observed here
could be hydrolysis of the aluminum 2-propoxide during
handling or under reaction conditions. To verify this, 3.7Al-

Y. Zhu et al. / Journal of Catalysis 218 (2003) 396–404
403
SBA-15 was also exposed to ambient environment for 48 h
before catalytic testing as for 11Zr-SBA-15. The activity of
the exposed 3.7Al-SBA-15 decreased tremendously; after
24 h, the conversion was only 10.5%. In contrast, 11Zr-
SBA-15 remained active at 97.5% conversion. This shows
the advantage of grafted zirconium propoxide for practical
applications.
form the surface propoxide. In the grafted catalysts used in
this study, the chemisorption of a secondary alcohol and the
ketone or aldehyde occurs at the Zr metal centre [1]. The
poisoning experiments with pyridine and benzoic acid indi-
cated that only the latter hindered the reaction. This may be
explained by a displacement of 2-propanol by benzoic acid
at the Zr centre. Lopez et al. [24] reported that the adsorption
coefficient of benzoic acid was 860 times higher than that
of 2-propanol. The strongly adsorbed benzoate will interfere
with the binding of the keto-compound to the metal centre
and the formation of the cyclic transition state required for
the hydride transfer.
As zirconium 1-propoxide was used as reagent for graft-
ing zirconium onto the support, ligand exchange between the
1-propoxide group and 2-propanol has to occur before re-
duction of the substrate becomes possible. For the ketone
or aldehyde, the presence of electron-donating groups ad-
jacent to the C=O bond facilitates the adsorption, as can
be deduced from the increasing rate of conversion in the
series from benzyl methyl ketone, acetophenone to 4-chloro-
acetophenone. Bulky molecules like adamantone and ben-
zyl methyl ketone have very low reactivity. The MPV re-
duction of a number of unsaturated aldehydes and ketones
was highly selective toward the corresponding alcohols. Cin-
namaldehyde was reduced to cinnamyl alcohol.
4. Discussion
The MPV reaction is usually catalysed by metal alkox-
ides, such as aluminum 2-propoxide. In this study, it is
found that for the homongeneously catalysed MPV reduc-
tion of 4-tert-butylcyclohexanone, zirconium 1-propoxide
has a higher activity than aluminum 2-propoxide. Hetero-
genisation of the metal alkoxides leads to an increase in
the catalytic activity when compared on a molar basis with
the homogeneous catalysts. High surface area mesoporous
materials like SBA-15, MCM-41, MCM-48, and silica gel
provide a useful support for the heterogenisation of zirco-
nium 1-propoxide. Surface hydroxyl groups are involved in
the grafting reaction, allowing the zirconium 1-propoxide to
build up to a monolayer.
The immobilised zirconium 1-propoxide is resistant to
leaching under reaction conditions. In addition, grafted zir-
conium 1-propoxide does not lose its catalytic activity fol-
lowing long exposure to ambient environment or water in
the reaction medium. This is in contrast to the grafted alu-
minum 2-propoxide. The high activity of the zirconium
1-propoxide even in the presence of water may be related
to the good activity of zirconium hydroxide. We have pre-
viously reported that zirconium hydroxide is a good catalyst
for the MPV reduction of cinnamaldehyde. We postulated
that the hydroxyl groups on the hydroxide are involved in
ligand exchange with 2-propanol. Indeed, facile exchange
between ligands and the reductant has been demonstrated
to be important in the MPV reaction. The efficiency of the
MPV reaction was found not to depend on the ligands [5].
Indeed, alkylaluminum such as Al(CH3)3 or Al(CH3)2Cl
was found to convert under reaction conditions to aluminum
2-proproxide while in bis(cyclpentadienyl)zirconium dihy-
dride, the cyclopentadienyl groups were found to split off
when reacted with 2-propanol resulting in zirconium 2-prop-
oxide [6]. Campbell et al. [5] attributed the high activity of
the catalysts formed from these alkylaluminum reagents to a
lower aggregation state with fewer bridging alkoxides than
in bulk aluminum 2-propoxide. It was further postulated that
only the nonbridging alkoxy group is involved in the hydride
transfer to the carbonyl group during the MPV reduction.
It has been proposed that the MPV mechanism for het-
erogeneous catalysts involves basic or acidic sites. In zeo-
lites, aluminum Lewis acid sites are believed to be the active
centres [20], while in MgO and hydrotalcites, the activity
seems to be linked to basic sites [21–23]. The reductant,
2-propanol, chemisorbs either at an acidic or a basic site to
Comparing the activity of grafted zirconium 1-propoxide
on supports with smaller pores than SBA-15, it was found
that the pore structure has little effect on the rate of re-
action. Although MCM-48 with its 3-dimensional network
of channels would be expected to allow more facile dif-
fusion of reactants and products, the rate of reaction was
fastest over zirconium 1-propoxide supported on MCM-41,
which has uniform one-dimensional channels. A slight in-
fluence of the nature of the support on the cis:trans ratio of
the 4-tert-butylcyclohexanol was observed. The highest se-
lectivity toward cis-4-tert-butylcyclohexanol was observed
over 11Zr-MCM-48 which has the smallest pore diameter
of the materials studied. This indicates some shape-selective
effect within the pore space and can be rationalised by the
less bulky transition state for the formation of cis- rather
than the trans-alcohol. This had also been observed in the
group of van Bekkum over zeolite beta [20]. In line with
this interpretation is the fact that the trans:cis ratio for
4-tert-butylcyclohexanol was very high when the reaction
was homogeneously catalysed by zirconium 1-propoxide or
aluminum 2-propoxide, but the ratio decreased when the cat-
alyst was immobilized onto a support.
5. Conclusion
Grafted zirconium 1-propoxide supported on SBA-15
was particularly active in the selective reduction of a num-
ber of aldehydes and ketones. The selectivity to the desired
alcohols was in most cases 100%. The activity of the cata-
lysts increased with zirconium 1-propoxide loading to reach

404
Y. Zhu et al. / Journal of Catalysis 218 (2003) 396–404
a limit at a monolayer coverage. Further loading did not
lead to a significant increase in activity. Although zirco-
nium propoxide easily undergoes hydrolysis, the grafted zir-
conium 1-propoxide catalysts were not deactivated by the
presence of water and showed good stability despite expo-
sure to ambient environment. In contrast to grafted zirco-
nium 1-propoxide catalysts, grafted aluminum 2-propoxide
showed lower activity and was very sensitive to humidity or
moisture in the reaction medium. The turnover frequency of
the surface-immobilised zirconium 1-propoxide was higher
than that of the free complex, an indication of the importance
of the reduction in dimensionality in a surface-mediated re-
action.
[8] P. Leyrit, C. McGill, F. Quignard, A. Choplin, J. Mol. Catal. A 112
(1996) 395.
[9] S.H. Liu, S. Jaenicke, G.K. Chuah, J. Catal. 206 (2002) 321.
[10] D. Zhao, Q. Huo, J. Feng, B.F. Chmelka, G.D. Stucky, J. Am. Chem.
Soc. 120 (1998) 6024.
[11] C.T. Kresge, M.E. Leonowicz, W.J. Roth, J.C. Vartuli, J.S. Beck, Na-
ture 359 (1992) 710.
[12] J. Xu, Z. Luan, H. He, W. Zhou, L. Kevan, Chem. Mater. 10 (1998)
3690.
[13] K. Schumacher, M. Grün, K.K. Unger, Micropor. Mesopor. Mater. 27
(1999) 201.
[14] K. Bauer, D. Garbe, H. Surburg, in: W. Gerhertz (Ed.), Ullmann’s En-
clyclopedia of Industrial Chemistry, Vol. A11, VCH, Weinheim, 1988,
p. 141.
[15] M.S. Morey, G.D. Stucky, S. Schwarz, M. Fröba, J. Phys. Chem. B 103
(1999) 2037.
[16] S.J. Gregg, K.S.W. Sing, Adsorption, Surface Area and Porosity, Aca-
demic Press, New York, 1982.
[17] A. Jentys, N.H. Pham, H. Vinek, J. Chem. Soc., Faraday Trans. 92
(1996) 3287.
References
[18] M. Shibagaki, K. Takahashi, H. Matsushita, Bull. Chem. Soc. Jpn. 61
(1988) 3283.
[19] J. Ma, G.K. Chuah, S. Jaenicke, R. Gopalakrishnan, K.L. Tan, Ber.
Bunsen.-Gesellschaft Phys. Chem. 99 (1995) 184.
[20] E.J. Creyghton, S.D. Ganeshie, R.S. Downing, H. van Bekkum, J. Mol.
Catal. A 115 (1997) 457.
[21] P.S. Kumbhar, J. Sanchez-Valente, J. Lopez, F. Figueras, Chem. Com-
mun. (1998) 535.
[22] T.M. Jyothi, T. Raja, K. Sreekurma, M.B. Talawar, S. Rao, J. Mol. Ca-
tal. A 157 (2000) 193.
[1] C.F. de Graauw, J.A. Peters, H. van Bekkum, J. Huskens, Synthesis 10
(1994) 1007.
[2] E.J. Creyghton, J. Huskens, J.C. van der Waal, H. van Bekkum, in:
H.U. Blaser, A. Baiker, R. Prins (Eds.), Heterogeneous Catalysis and
Fine Chemicals IV, Elsevier, Amsterdam, 1997, p. 531.
[3] T. Ooi, T. Miura, K. Maruoka, Angew. Chem., Int. Ed. 37 (1998) 2347.
[4] T. Ooi, Y. Itagaki, T. Miura, K. Maruoka, Tetrahedron Lett. 40 (1999)
2137.
[5] E.J. Campbell, H. Zhou, S.T. Nguyen, Organic Lett. 3 (2001) 2391.
[6] Y. Ishii, T. Nakano, A. Inada, Y. Kishigami, K. Sakurai, M. Ogawa,
J. Org. Chem. 51 (1986) 240.
[23] M.A. Aramendía, V. Borau, C. Jiménez, J.M. Marina, J.R. Ruiz, F.J.
Urabano, Appl. Catal. A 206 (2001) 95.
[7] R. Anwander, C. Palm, G. Gerstberger, O. Groeger, G. Engelhardt,
Chem. Commun. (1998) 1811.
[24] J. Lopez, J. Sanchez Valente, J.-M. Clacens, F. Figueras, J. Catal. 208
(2002) 30.