Oxidant-free dehydrogenation of alcohols heterogeneously catalyzed by cooperation of silver clusters and acid-base sites on alumina

Article abstract of DOI:10.1002/chem.200802222

A γ-alumina-supported silver cluster catalyst - Ag/Al ₂O₃-has been shown to act as an efficient heterogeneous catalyst for oxidant-free alcohol dehydrogenation to carbonyl compounds at 373 K. The catalyst shows higher activity than conventional heterogeneous catalysts based on platinum group metals (PGMs) and can be recycled. A systematic study on the influence of the particle size and oxidation state of silver species, combined with characterization by Ag K-edge XAFS (X-ray absorption fine structure) has established that silver clusters of sizes below 1 nm are responsible for the higher specific rate. The reaction mechanism has been investigated by kinetic studies (Hammett correlation, kinetic isotope effect) and by in situ FTIR (kinetic isotope effect for hydride elimination reaction from surface alkoxide species), and the following mechanism is proposed: 1) reaction between the alcohol and a basic OH group on the alumina to yield alkoxide on alumina and an adsorbed water molecule, 2) CH activation of the alkoxide species by the silver cluster to form a silver hydride species and a carbonyl compound, and 3) H₂ desorption promoted by an acid site in the alumina. The proposed mechanism provides fundamental reasons for the higher activities of silver clusters on acid-base bifunctional support (Al ₂O₃) than on basic (MgO and CeO₂) and acidic to neutral (SiO₂) ones. This example demonstrates that catalysts analogous to those based on of platinum group metals can be designed with use of a less expensive d¹⁰ element - silver - through optimization of metal particle size and the acid-base natures of inorganic supports.

Full text of DOI:10.1002/chem.200802222

FULL PAPER
DOI: 10.1002/chem.200802222
Oxidant-Free Dehydrogenation of Alcohols Heterogeneously Catalyzed by
Cooperation of Silver Clusters and Acid–Base Sites on Alumina
Ken-ichi Shimizu,* Kenji Sugino, Kyoichi Sawabe, and Atsushi Satsuma[a]
Abstract: A g-alumina-supported silver
cluster catalyst—Ag/Al2O3—has been
shown to act as an efficient heteroge-
neous catalyst for oxidant-free alcohol
dehydrogenation to carbonyl com-
pounds at 373 K. The catalyst shows
higher activity than conventional het-
anism has been investigated by kinetic
studies (Hammett correlation, kinetic
isotope effect) and by in situ FTIR (ki-
netic isotope effect for hydride elimina-
tion reaction from surface alkoxide
species), and the following mechanism
is proposed: 1) reaction between the al-
cohol and a basic OH group on the
alumina to yield alkoxide on alumina
dride species and a carbonyl com-
pound, and 3) H2 desorption promoted
by an acid site in the alumina. The pro-
posed mechanism provides fundamen-
tal reasons for the higher activities of
silver clusters on acid–base bifunctional
support (Al2O3) than on basic (MgO
and CeO2) and acidic to neutral (SiO2)
ones. This example demonstrates that
catalysts analogous to those based on
of platinum group metals can be de-
signed with use of a less expensive d10
element—silver—through optimization
of metal particle size and the acid–base
natures of inorganic supports.
ACHTUNGTRENNUNGeroACHTUNGTRENNUNGgeneous catalysts based on platinum
group metals (PGMs) and can be recy-
cled. A systematic study on the influ-
ence of the particle size and oxidation
state of silver species, combined with
characterization by Ag K-edge XAFS
(X-ray absorption fine structure) has
established that silver clusters of sizes
below 1 nm are responsible for the
higher specific rate. The reaction mech-
ꢀ
and an adsorbed water molecule, 2) C
H activation of the alkoxide species by
the silver cluster to form a silver hy-
ꢀ
Keywords: alcohols · C H activa-
tion · green chemistry · heterogene-
ous catalysis · silver
Introduction
Silver clusters appear to be very popular in the research
field of nanoparticle synthesis,[9–13] and several review arti-
cles have been published.[9,10] The number of investigations
that have focused on specific catalysis by silver clusters or
nanoparticles, however, is surprisingly small in comparison
with research into gold nanoparticle catalysis.[14–24]
We have been paying attention to cluster-specific catalysis
by supported silver catalysts for environmental applications
and have reported a series of studies on size- and support-
specific catalysis by silver clusters for the selective catalytic
reduction of NOx with hydrocarbons for the removal of
NOx emissions from diesel engine exhausts.[20–22] It has been
shown that, of various silver catalysts, silver clusters on alu-
mina constitute the best catalyst, with activity much higher
than that of Pt/Al2O3 catalyst. Fundamental studies showed
that both the silver clusters and the acid–base sites on the
support surface are responsible for the high activity of silver
clusters on alumina.[20–22] For this study we attempted an ex-
tension of the acid–base-assisted silver cluster catalysis to
selective oxidation.
Materials with clusters have interesting chemical properties
unusual for bulk solids. Size- and support-specific catalysis
by supported gold clusters is a well known example.[1–8] Vari-
ous supported gold catalysts have been reported to show
catalytic activities similar to or higher than those of plati-
num group metals (PGMs) for oxidation of alcohols to car-
bonyl compounds,[4–8] as well as for CO oxidation.[1–3] How-
ever, fundamental aspects of the size- and support-specific
catalysis of these reactions still remained controversial. One
might expect that silver, as a less expensive group IB metal,
should be usable for the design of PGM-free catalysts
through optimization of metal particle size and support.
[a] Dr. K.-i. Shimizu, K. Sugino, Dr. K. Sawabe, Prof. Dr. A. Satsuma
Department of Molecular Design and Engineering
Graduate School of Engineering, Nagoya University
Nagoya 464-8603 (Japan)
ꢀ
C H activation catalysis, including oxidation of alcohols
Fax : (+81)52-789-3193
E-mail: kshimizu@apchem.nagoya-u.ac.jp
to carbonyl compounds, is one of the most important topics
Chem. Eur. J. 2009, 15, 2341 – 2351
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2341

in the fields of catalysis and organic chemistry. To achieve
environmental and economical acceptability, much effort
has been devoted to the development of transition-metal-
catalyzed oxidation of alcohols with environmentally friend-
ly oxidants (such as oxygen, hydrogen peroxide, or alkenes)
that avoid the use of large excesses of toxic and expensive
stoichiometric metal oxidants.[3–7,25–30] From the viewpoint of
atom efficiency and safety of the reaction, an oxidant-free
catalytic dehydrogenation of alcohols to carbonyl com-
pounds and molecular hydrogen would be ideal. Several
recent reports show oxidant-free alcohol dehydrogenation
with PGM catalysts (ruthenium[31–35] and iridium[36] com-
plexes and heterogeneous ruthenium catalysts[37–39]). Howev-
er, they suffer variously from the need for acid or base addi-
tives, from difficulties in catalyst synthesis, manipulation, or
reusability, from high cost, or from low activity for the pri-
mary alcohol dehydrogenation. Very recently, Mitsudome
et al. have developed Ag- or Cu-based heterogeneous cata-
lysts for the oxidant-free dehydrogenation of alcohols.[23,40]
In particular, Ag/HT (hydrotalcite-supported silver nanopar-
ticles) are attractive as the most active catalyst for this reac-
tion,[23] although the reaction mechanism and the origins of
the size- and support-specific catalysis are not well estab-
lished. Understanding of how silver-based catalysts operate
is imperative for the rational development of improved
PGM-free alcohol dehydrogenation catalysts.
lytic process: one involves kinetic measurements for the
overall reaction, and the other direct spectroscopic observa-
ꢀ
tion of the C H activation step in a catalytic cycle. Combi-
nation of these approaches should give full understanding of
the catalytic process.
Results and Discussion
Characterization of Al2O3-supported Ag clusters: The X-ray
diffraction (XRD) pattern of Ag/Al2O3-x (x=1–10 wt%; see
Experimental Section for terminology) was essentially the
same as that of the g-Al2O3 support, and no lines due to Ag
metal or Ag2O were observed, indicating no formation of
large Ag metal or Ag2O particles during the catalyst prepa-
ration. In this study, Ag/Al2O3 catalysts were reduced with
H2 at 573 K before each catalytic reaction. To investigate
the structure of Ag species after the H2 reduction, in situ
Ag K-edge XAFS measurement of Ag/Al2O3 and Ag refer-
ence compounds was carried at 573 K. Figure 1A shows
Ag K-edge X-ray absorption near-edge structures (XANES)
spectra, which are known to be sensitive to the oxidation
states of X-ray absorbing atoms. The XANES feature of Ag/
Al2O3-x (x=1–10 wt%) after the H2 reduction is clearly dif-
ferent from that of Ag2SO4 but rather similar to that of Ag
powder, indicating that silver species in these samples are in
a reduced state. On the other hand, the XANES features of
Ag/Al2O3-5 before the reduction and Ag/Al2O3-0.5 after the
reduction differ from that of Ag powder but resemble that
of Ag2SO4 as a reference compound for ionic Ag+ species
surrounded by oxygen atoms. This suggests that Ag+ is the
dominant silver species in the H2-reduced Ag/Al2O3-0.5 and
unreduced Ag/Al2O3-5 samples.
In the research area of organometallic catalysis, attention
has been focused on the cooperation of acid or basic ligands
in organic reactions catalyzed by transition metal com-
plexes.[28,35,36,41–44] One of the successful examples is Noyoriꢁs
ruthenium catalyst for the hydrogenation of polar bonds, in
which cooperation between an acidic hydrogen on an amido
ligand and ruthenium hydride plays an important role.[41]
For the catalytic dehydrogenative oxidation of alcohols,
Fujita et al. developed an iridium complex containing a 2-
hydroxypyridine ligand. The hydride ion on iridium, formed
ꢀ
by C H cleavage of an iridium alkoxide intermediate, can
be effectively removed through its reaction with the protic
hydrogen on the ligand, which facilitates the release of dihy-
drogen.[36] A ruthenium catalyst developed by Milsteinꢁs
group possesses a cooperative basic site in a b position to
the metal, which facilitates binding of the alcohol substrate
to give an alkoxide intermediate.[35] Knowing that metal
oxides, such as g-alumina, have acid–base sites on their sur-
faces, one might expect that the cooperative effect of the or-
ganic ligand might be replaceable by that of an inorganic
support material.
Here we demonstrate a well characterized g-alumina-sup-
ported silver cluster as an effective heterogeneous catalyst
for the oxidant-free catalytic dehydrogenation of alcohols to
carbonyl compounds. We report detailed mechanistic and
structural studies that address the influence of the size and
charge state of silver and the nature of the support to estab-
lish a detailed reaction mechanism that should be useful for
the design of practical heterogeneous catalysts without use
of organic ligands and PGM elements. We adopted two
complementary ways of studying the mechanism of a cata-
Figure 1. Ag K-edge A) XANES spectra and B) EXAFS Fourier trans-
forms recorded in situ after flowing of various gas mixtures for 0.5 h at
573 K: a) Ag2SO4 in He, b) Ag/Al2O3-5 in O2 (10%), c) Ag/Al2O3-0.5 in
H2 (1%), d) Ag/Al2O3-1 in H2 (1%), e) Ag/Al2O3-5 in H2 (1%), f) Ag/
Al2O3-10 in H2 (1%), and g) Ag powder in He.
Figure 1B shows the Fourier transform (FT) of the k3-
weighted Ag K-edge extended X-ray absorption fine struc-
ture (EXAFS) of Ag/Al2O3 and Ag powder. The structural
parameters derived from curve-fitting analysis are listed in
2342
www.chemeurj.org
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 2341 – 2351

Heterogeneous Catalysis
FULL PAPER
Table 1. Curve-fitting analysis of Ag K-edge in situ EXAFS at 573 K in
H2 (1%) for Ag/Al2O3.
Table 1. The EXAFS of the fresh Ag/Al2O3-5 (spectrum a)
ꢀ
showed large Ag O contribution, and the Ag–Ag coordina-
Catalyst[a]
Shell
CN[b]
R [ꢂ][c]
(s2)[d] [ꢂ2]
Rf [%][e]
tion number was negligibly small (Table 1). When the
A
0.5
O
Ag
O
Ag
O
Ag
O
Ag
O
Ag
O
2.3
0.2
0.5
5.0
0.5
5.4
0.5
5.9
1.2
0.3
0.6
7.8
2.23
2.76
2.24
2.85
2.26
2.84
2.25
2.85
2.17
2.72
2.26
2.85
0.100
0.003
0.015
0.068
0.015
0.079
0.010
0.076
0.070
0.028
0.005
0.081
8.1
spectral feature changed. After H2 reduction at 573 K for
30 min (spectrum e), the Ag Ag shell with coordination
number of 5.9 and bond length of 2.85 ꢂ was observed. The
Ag O coordination number decreased from 1.2 to 0.5 and
the Ag Ag coordination number increased from 0.3 to 5.9
after the reduction. For the Ag/Al2O3 samples with Ag load-
ings of 1–10 wt%, the Ag Ag contribution with the bond
length of 2.84–2.85 ꢂ was also observed. This distance is
1
2.3
1.9
1.8
4.7
2.9
3
5
5 (O2)[f]
10
shorter than that in bulk Ag and is close to the Ag Ag dis-
Ag
tance reported for silver clusters in Ag-exchanged zeolite X
[a] Ag loading (wt%) on Ag/Al2O3. [b] Coordination numbers. [c] Bond
length. [d] Debye–Waller factor. [e] Residual factor. [f] EXAFS data for
the Ag/Al2O3-5 sample before H2 reduction recorded under O2 (10%) at
573 K.
formed by H2 reduction at 573 K.[13] The Ag Ag coordina-
tion numbers (5.0–7.8) were lower than that of bulk Ag(12).
These results indicate that the H2 reduction treatment re-
sults in the reduction and aggregation of Ag+ ions to metal-
lic silver clusters. In the EXAFS for the lowest loading
ꢀ
sample—Ag/Al2O3-0.5—the Ag Ag contribution was negli-
gibly small even after 30 min of the reduction treatment
(spectrum c), indicating that reductive agglomeration of
Ag+ to silver cluster does not occur in this sample. In sum-
mary, the EXAFS and XANES results show that small
silver clusters are the predominant silver species on the Ag/
Al2O3-x (x=1–10 wt%), whereas on Ag/Al2O3-0.5 most of
The dehydrogenation of 4-methylbenzyl alcohol was com-
pletely stopped by the removal of Ag/Al2O3-5 from the re-
action mixture: the catalyst was filtered off after the reac-
tion mixture had been stirred for 2 h (yield 37%), and the
reaction then did not proceed further when the filtrate was
heated at 373 K for 22 h. This result rules out any possible
contribution of homogeneous catalysis by leached silver spe-
cies.
We also studied reuse of the Ag/Al2O3-5 catalyst, with the
dehydrogenation of 4-methylbenzyl alcohol as test reaction.
The catalyst can be easily separated from the reaction mix-
ture by simple filtration. Although the filtered catalyst
showed a decrease in activity (64% yield), its activity was
once more restored to that observed for the first run
(entry 1) after the filtered catalyst had been calcined in air
at 873 K for 10 min, followed by the reduction with H2 at
573 K for 5 min (entries 2, 3 and 4).
the silver species are highly dispersed as Ag+ ions. The Ag
ꢀ
O EXAFS contributions with coordination numbers of 0.5–
0.6 in Ag/Al2O3-x (x=1–10 wt%) samples suggest that
silver clusters are not fully reduced and may exist as cationic
clusters interacting with oxygen atoms at the metal surface
or the metal–support interface.
Catalytic properties: With the Ag/Al2O3-5 as a standard cat-
alyst, we examined its catalytic activity for the oxidant-free
oxidation of alcohols. When 4-methylbenzyl alcohol was
treated in the presence of Ag/Al2O3-5 (2.0 mol% of Ag) at
373 K in toluene for 24 h, 4-
methylACHTUNGTRENNUNGbenzACHTUNGTRENNUNGaldehyde was pro-
Table 2. Dehydrogenation of various alcohols.[a]
duced in 93% yield and with
93% selectivity at a level of al-
cohol conversion of 99%
(Table 2, entry 1). After 12 h re-
action time, the yields of gas-
phase H2 (75%) and 4-methyl-
Entry
Substrate
Product
Conv. [%]
Yield [%]
1
4-methylbenzyl alcohol
4-methylbenzaldehyde
99
99
99
99
100
100
79
99
53
82
72
100
27
93
90
94
91
82
86
70
93
50
80
70
94
16
2[b]
3[c]
4[d]
5
benzaldehyde
(76%)
were
nearly identical to the level of
conversion of the alcohol
(77%), indicating that H2 was
generated quantitatively during
the dehydrogenation. Because
H2 escapes from the reaction
mixture into the gas phase, the
overall endothermic dehydro-
genation reaction is driven to
the product side.
benzyl alcohol
4-methoxybenzyl alcohol
4-chlorobenzyl alcohol
1-phenylethanol
octan-2-ol
propan-2-ol
cyclohexanol
cyclodecanol
octan-1-ol
benzaldehyde
4-methoxybenzaldehyde
4-chlorobenzaldehyde
acetophenone
octan-2-one
6
7[e]
8
9[e]
10[f]
11[e]
12[e]
13[e]
acetone
cyclohexanone
cyclodecanone
1-octanone
[a] Substrate (1.0 mmol), toluene (3 mL), Ag/Al2O3-5 catalyst (2.0 mol%), T=373 K, t=24 h. Levels of con-
version of the alcohols and yields of carbonyl compounds were determined by GC. [b] Reuse 1. [c] Reuse 2.
[d] Reuse 3. [e] t=48 h. [f] T=353 K.
Chem. Eur. J. 2009, 15, 2341 – 2351
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
2343

K.-i. Shimizu et al.
The total turnover number (TON) based on total Ag con-
tent was 184, which is comparable with those reported for
oxidant-free secondary alcohol dehydrogenation with some
homogeneous Ru catalysts (TONs: 160 to 186).[31–35] The
commercially available PGM-based heterogeneous catalysts
Ru/C, Ru/Al2O3, Pd/C, and Pd/Al2O3 (Ru or Pd=5 wt%)
were tested for the oxidant-free dehydrogenation of 4-meth-
ylbenzyl alcohol under conditions similar to those in Table 2
(entry 1). After treatment at 373 K in toluene for 60 h, 4-
methylbenzaldehyde yields with Ru/C, Ru/Al2O3, Pd/C, and
Pd/Al2O3 were 14, 81, 20, and 56%, respectively, which were
less than the yield obtained with Ag/Al2O3-5 (93% after
24 h). This result clearly demonstrates higher catalytic effi-
ciency of Ag/Al2O3-5 for this reaction.
The scope of alcohol dehydrogenation by Ag/Al2O3-5 was
studied (Table 2). Oxidation of benzyl alcohol proceeded
with moderate selectivity (82%); small amounts of benzyl
benzoate (1%) and dibenzyl ether (2%) were obtained as
side products. The oxidation of p-substituted benzyl alcohols
proceeded with good yields. It should be noted that most of
the previously reported homogeneous transition metal com-
plexes are unable to catalyze the oxidant-free dehydration
of primary alcohols.[31–36] Secondary alcohols, including the
less reactive aliphatic alcohols (cyclohexanol, cyclodecanol,
propan-2-ol, and octan-2-ol), were also tolerated with good
to moderate yields. Treatment of octan-1-ol produced
*
Figure 2. Effect of Ag loading of Ag/Al2O3 on ) the initial rate, and
*
) TOF based on the number of surface Ag atoms for dehydrogenation
of benzyl alcohol at 373 K. ) CN for Ag Ag shell of EXAFS (from
Table 1), and ) average particle size of Ag from the EXAFS analysis.
~
ꢀ
~
A
are present mainly as metal clusters, TOF per surface Ag
atoms decreases with increase in Ag loading. The TOF per
surface Ag sites versus average particle size is shown in
Figure 3. Clearly, the intrinsic activity per external silver
octyl octanoate was negligible.
Structure–activity relationships: The reaction rates for the
dehydrogenation of benzyl alcohol were measured under
conditions where levels of conversion were below 30% for a
series of Ag/Al2O3 catalysts with different Ag loadings. Fig-
ure 2A plots the benzaldehyde formation rate per gram of
Ag/Al2O3 as a function of silver loading. The catalyst with
lowest loading (0.5 wt%) is nearly inactive. The rate in-
creased with the silver loading up to 5 wt% and then de-
creased with further increases in the loading.
To investigate the relationship between structure and cat-
alytic activity, we also examined the effect of silver loading
on the particle size of metallic silver species in Ag/Al2O3.
ꢀ
By using the coordination number of Ag Ag contribution
in EXAFS (Table 1), the average metallic silver cluster size
was determined by the model of Jentys.[45] The average
silver powder particle size (a plot for Ag loading of 100%
in Figure 2B) was estimated from the XRD line broadening
*
Figure 3. ) TOF based on the number of surface Ag atoms for dehydro-
genation of benzyl alcohol at 373 K as a function of average particle size
of Ag in Ag/Al2O3-x (x=1–100 wr%) catalysts (from Figure 1).
of the AgACHTUNGTRENNUNG(111) reflection at 2q=38.18 due to Ag metal by
use of the Scherrer equation. As shown in Figure 2B, the
average silver cluster size increased with silver loading. By
using the mean diameter of silver clusters and the atomic di-
ameter of silver (0.289 nm) and by assuming that the sup-
ported silver clusters can be modeled as a fcc crystal lattice,
the number of surface silver atoms for each catalyst was
statistically determined according to the methods of Mori
et al.[26] and Abad et al.[6] This value was then used to esti-
mate turnover frequency (TOF) per surface Ag site as
shown in Figure 2C. Above 1 wt%, at which silver species
atom decreases with the particle size. From these quantita-
tive studies, the following conclusions relating to the struc-
ture–activity relationship can be drawn. The silver cluster
acts as a more efficient catalytic site for alcohol dehydroge-
AHCTUNGTRENNUNG
nation than monomeric Ag+ ion and bulk Ag particle, and
the silver clusters with smaller particle sizes give higher in-
trinsic activity.
2344
www.chemeurj.org
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 2341 – 2351

Heterogeneous Catalysis
FULL PAPER
To examine the role of the support in silver-catalyzed de-
hydrogenation of alcohols, we prepared a series of support-
ed silver catalysts with the same silver content (5 wt%) and
of similar particle size (0.8–3.0 nm), but with a variety of
supports. EXAFS and XANES analysis of these catalysts
showed that silver clusters with mean particle size of 0.8 to
2.9 nm make up the main silver species on these catalysts
(results not shown). The catalytic activities and selectivities
in the dehydrogenation of 4-methylbenzyl alcohol with vari-
ous silver catalysts were measured, and the aldehyde yields
were plotted as a function of the electronegativities of the
metal species in the support materials, which have been
used as a parameter for the acidity of metal oxides
(Figure 4). There is a volcano-type relationship between the
Figure 5. IR spectra after exposure to propan-2-ol (0.5 mmol per g cata-
lyst), followed by purging with He for 60 s at 313 K: a) SiO2, b) MgO,
c) Al2O3, and d) Ag/Al2O3-5. Spectra e and f are results for [D2]propan-2-
ol adsorption at 313 K on Ag/Al2O3-5 and Al2O3, respectively.
cates the absence of non-dissociatively adsorbed propan-2-ol
on the catalyst.
In the FTIR spectrum of the adsorbed species formed by
the reaction between Ag/Al2O3-5 and [D2]propan-2-ol (a-
deuteriated propan-2-ol), bands at 2970 cmꢀ1(methyl asym-
ꢀ1
ꢀ
ꢀ
metric C H stretching), 2940 cm (methyl asymmetric C H
Figure 4. Yield of p-methylbenzaldehyde (from Table 2) as a function of
ꢀ
*
*
the electronegativity of support cation under N2 ( ) or O2 ( ).
stretching), 2870 cmꢀ1 (methyl symmetric C H stretching),
2100 cmꢀ1 (a-C D stretching), 1465 cm (methyl asymmet-
ꢀ1
ꢀ
ric deformation), 1376 cmꢀ1 (methyl symmetric deforma-
ꢀ1
ꢀ
ꢀ
ꢀ
catalytic activity and the electronegativity. The silver clus-
ters supported on alumina, which is known to have both
acidic and basic surface sites,[46,47] had the highest catalytic
activity. The supports with low electronegativity (strong
basic character) gave lower yields. The silver clusters on
SiO2, which can be regarded as an acidic or neutral oxide,[48]
had lowest activity. These results suggest that both acidic
and basic surface sites are necessary for the efficient catalyt-
ic dehydrogenation of alcohols.
tion), 1328 cmꢀ1 (a-C H deformation), 1190 cm (C O/C
C coupled stretchings) were assigned to a-deuteriated 2-
propoxide groups. A weak band at 2140 cmꢀ1 may be as-
signed to a different type of 2-propoxide species. It is note-
worthy that in both propan-2-ol and [D2]propan-2-ol adsorp-
tion experiments the IR spectra for Al2O3 (spectra c, f) were
basically the same spectra as seen for Ag/Al2O3-5 (spectra d,
e), suggesting that 2-propoxide is adsorbed not on the silver
site, but on the surface of the Al2O3.
Immediately after introduction of propan-2-ol onto the
catalyst (t=60 s), an intense negative peak at about
3740 cmꢀ1 was observable in the OH stretching region; this
has been attributed to the surface OH-m2-Al group.[47] Broad
overlapping bands centered around 3300 cmꢀ1 in the OH
stretching region and a weak feature at 1640 cmꢀ1 were also
observed. These are characteristic of the OH stretching vi-
bration and the HOH bending vibration of adsorbed
water.[52] It is established that hydrogen bonding causes the
downward shift and broadening of the OH stretching bands
of the surface hydroxy groups.[48] The broad feature due to
OH stretching vibration suggests that the water molecules
are strongly hydrogen bonded. A similar IR spectrum with a
negative peak at about 3750 cmꢀ1 and bands due to ad-
sorbed water was observed for h-alumina after HCl adsorp-
Reaction mechanism: The adsorption complexes formed
from the adsorption of propan-2-ol (0.5 mmol per g catalyst)
on various reference oxides and Ag/Al2O3-5 at 313 K were
studied in situ by FTIR spectroscopy (Figure 5). The result
for Ag/Al2O3-5 (spectrum d) shows the predominance of
ꢀ1
ꢀ
very strong bands at 2970 cm (methyl asymmetric C H
stretching), 2940 cmꢀ1 (methyl asymmetric C H stretching),
ꢀ
2874 cmꢀ1 (coupling of a-C H stretching and methyl sym-
ꢀ
ꢀ1
ꢀ
metric C H stretching), 1465 cm (methyl asymmetric de-
formation), 1376 cmꢀ1 (methyl symmetric deformation),
1328 cmꢀ1 (a-C H deformation), 1166 and 1130 cm (C O/
ꢀ1
ꢀ
ꢀ
[49,50]
ꢀ
C C coupled stretchings) typical of 2-propoxide groups.
A very weak peak at 2640 cmꢀ1 may correspond to a combi-
ꢀ1
ꢀ
ꢀ
nation of the C O/C C coupled stretchings (1166 cm ) and
the methyl deformation (1465 cmꢀ1) of the 2-propoxide, as
previously reported for FTIR study of methanol adsorption
on alumina.[51] A band at 1280 cmꢀ1 due to OH deformation
of the propan-2-ol molecule is hardly observable, which indi-
tion, which was interpreted as arising from the reaction be-
[53]
ꢀ
tween Al OH with HCl to yield AlCl and H2O. This indi-
cates that the OH-m2-Al group acts as a Brønsted base, as
has also been supported by a recent quantum chemical
Chem. Eur. J. 2009, 15, 2341 – 2351
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
2345

K.-i. Shimizu et al.
study.[54] From these observations the following surface reac-
tion [Eq. (1)] is established: the reaction between propan-2-
ol and a OH-m2-Al group (Brønsted base site) results in the
formation of 2-propoxide coordinated to the unsaturated Al
site (Lewis acid site) and a hydrogen-bonded water mole-
cule on the alumina surface.
curred with much lower reaction rate than 2-propoxide for-
mation. This result indicates that step (1) is faster than
steps (2) and (3).
R2CHꢀOꢀAlþAgn ! ðCH3Þ2CdþꢀOꢀAlþAgnꢀHdꢀ
ðCH3Þ2CdþꢀOꢀAlþH2O ! ðCH3Þ2C¼OþH2O ꢁ ꢁ ꢁ Al
ð2Þ
ð3Þ
ðCH3Þ2CHOHþAlꢀOH ! ðCH3Þ2CHꢀOꢀAlþH2O
ð1Þ
The intensities of the negative band at 3740 cmꢀ1 and the
band due to hydrogen-bonded water (3300 cmꢀ1) decreased
with increasing reaction time, indicating the consumption of
the hydrogen-bonded water to reproduce the OH-m2-Al
group on the alumina surface. This can reasonably be inter-
preted as follows: the adsorbed water molecule reacts with
the neighboring hydride ion to form dihydrogen, which re-
sults in a regeneration of a basic OH group of alumina
[Eq. (4)].
To investigate the reaction behavior of 2-propoxide species,
the reaction between propan-2-ol and Ag/Al2O3-5 at 373 K
was studied by in situ FTIR (Figure 6A). After rapid forma-
tion of the 2-propoxide species (t=60 s), the intensities of
the bands due to the 2-propoxide species gradually de-
creased and a new band at 1690 cmꢀ1, due to C=O stretching
of acetone adsorbed on alumina,[55] appeared and its intensi-
ty increased with time. This suggests that 2-propoxide under-
ꢀ
AgnꢀHdꢀþH2O ꢁ ꢁ ꢁ Al ! H2þAlꢀOHþAgn
goes C H cleavage, resulting in the formation of acetone.
Note that 2-propoxide species on silver-free alumina did not
convert to acetone in a temperature range of 313–373 K.
ð4Þ
Kinetic analysis of the conversion of the 2-propoxide groups
can be studied by the time-resolved IR method, and the
result at 313 K is shown in Figure 7 as an example. The rela-
tive amounts of (CH3)2CHOꢀ and (CH3)2CDOꢀ groups were
estimated from the changes in the areas of the IR bands in
the 1080–1210 cmꢀ1 or 1100–1250 cmꢀ1 ranges, respectively,
because these bands had the least overlap with other bands
and, therefore, were the most suitable for the quantitative
analysis. The first-order plots of 2-propoxide during its con-
version into acetone gave fairly good straight lines. The
first-order rate constants of 2-propoxide consumption at
313 K were estimated to be 3.2ꢃ10ꢀ4 ꢂ7.4ꢃ10ꢀ6 sꢀ1 and
1.4ꢃ10ꢀ4 ꢂ5.1ꢃ10ꢀ6 sꢀ1 for (CH3)2CHOꢀ and (CH3)2CDOꢀ
groups, respectively. Figure 8 shows the rate constants for 2-
propoxide consumption at a range of temperatures (313–
373 K). Arrhenius plots for 2-propoxide consumption gave
good straight lines, and the apparent activation energies de-
termined from the slope were 27.3ꢂ2.3 and 33.8ꢂ
1.3 kJmolꢀ1 for (CH3)2CHOꢀ and (CH3)2CDOꢀ, respective-
ly. The difference in these activation energies (EHꢀED =
6.3ꢂ3.6 kJmolꢀ1) is consistent within experimental error
with the zero-point energy difference (4.6 kJmolꢀ1) between
ꢀ
This indicates that the silver clusters are necessary in the C
H activation step. We tentatively assume that the acetone
formation [Eq. (3)] proceeds via silver hydride and carbocat-
ion intermediates [Eq. (2)], which was confirmed by liquid-
phase kinetic studies. It is important to note that the
propan-2-ol molecule introduced onto the catalyst surface
was nearly completely converted into the 2-propoxide
within 60 s, whereas 2-propoxide conversion to acetone oc-
ꢀ1
ꢀ1
ꢀ
ꢀ
a-C H (2874 cm ) and a-C D (2100 cm ) estimated from
the stretching vibration energies of 2-propoxide species. This
indicates that the rate-determining step of 2-propoxide con-
ꢀ
version to acetone is the cleavage of the C H bond on the
a-carbon atom.
Next, we studied kinetic experiments for the overall
liquid-phase reaction with Ag/Al2O3-5 catalyst. Firstly, the
dependence of the initial reaction rate on the initial benzyl
alcohol concentration (0.09 to 0.32m) showed a nonlinear
dependence (Figure 9A). Good linear correlation is ob-
served in a Lineweaver–Burk plot (Figure 9B), indicating
that the reaction follows Michaelis–Menten-type kinetics:
first-order dependence on alcohol at the low concentration
and zero-order dependence on alcohol at higher concentra-
tions.
Figure 6. IR spectra taken at 373 K after exposure of A) Ag/Al2O3-5 or
B) Ag/MgO-5 to propan-2-ol (0.5 mmol per g catalyst), followed by purg-
ing with He.
2346
www.chemeurj.org
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 2341 – 2351

Heterogeneous Catalysis
FULL PAPER
Figure 7. First-order plots for the decomposition of the surface 2-propox-
ide species on Ag/Al2O3-5 at 313 K. Rate constants estimated from the
changes in the area of IR band in the 1080–1210 cmꢀ1 range for
ꢀ1
*
*
)
(CH3)2CHOH ( ) and the 1100–1250 cm range for (CH3)2CDOH (
were 3.2ꢃ10ꢀ4 ꢂ7.4 sꢀ1 and 1.4ꢃ10ꢀ4 ꢂ5.1 sꢀ1, respectively.
Figure 9. A) Rate dependence on the benzyl alcohol concentration, and
B) Lineweaver–Burk plot. Conditions: benzyl alcohol (0.03–1.28m), Ag/
Al2O3-5 (2.0 mol%), toluene (3 mL), 373 K.
Figure 8. Arrhenius plots for the decomposition of surface 2-propoxide
species on Ag/Al2O3-5. Apparent activation energies determined from
the plots are 27.5ꢂ2.3 and 33.8ꢂ1.3 kJmolꢀ1 for (CH3)2CHOH and
(CH3)2CDOH, respectively.
The dependence of the initial reaction rate on the amount
of the catalyst is shown in Figure 10. The result shows that
the rate has a linear relationship with the silver content in
the system, indicating a first-order dependence of the rate
on the silver concentration in the reaction system. These re-
sults suggest that both catalyst and alcohol are involved in a
rate-limiting step and therefore rule out the decomposition
of silver hydride to regenerate the silver site as rate-deter-
mining. Because the observed saturation kinetics in [alco-
hol] point toward a pre-equilibrium between catalyst and
substrate, it is unlikely that alcohol binding is rate-determin-
ing.
Kinetic isotope effects in the liquid-phase propan-2-ol de-
hydrogenation under N2 at 353 K were studied with
(CH3)2CHOH or (CH3)2CDOH as reactants and Ag/Al2O3-5
as catalyst (Figure 11). In the initial reaction period, the
concentration of acetone increased linearly with time. From
the slope of the lines, the kinetic isotope effect (kH/kD) at
353 K was estimated to be 2.6. Ando et al. reported a kH/kD
Figure 10. Effect of the amount of Ag/Al2O3-5 catalyst on the reaction
rate for benzyl alcohol dehydrogenation. Conditions: Ag/Al2O3-5 (0.5–
9.3 mol%), benzyl alcohol (1.0 mmol), toluene (3 mL), 373 K.
value of 1.57 for the liquid-phase dehydrogenation of
(CH3)2CHOH and (CH3)2CDOH to yield acetone and mo-
lecular hydrogen under reflux conditions (at 355.6 K) with a
carbon-supported ruthenium catalyst, and they concluded
ꢀ
that the cleavage of the a-C H bond is facile and that the
highest activation energy is required for the formation of
molecular hydrogen.[37] The primary kinetic isotope effect of
2.6 observed in our case indicates that the cleavage of the a-
ꢀ
C H bond is the rate-determining step in the overall reac-
tion in the presence of Ag/Al2O3-5. The validity of the
Chem. Eur. J. 2009, 15, 2341 – 2351
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
2347

K.-i. Shimizu et al.
ꢀ
mining step in the overall reaction is the C H activation of
alkoxide species by silver clusters to form silver hydride spe-
cies and carbonyl compound.
A reaction mechanism based on the mechanistic and ki-
netic experiments for the dehydrogenation of alcohols by
Ag/Al2O3-5 is presented in Scheme 1 in the case of propan-
*
)
Figure 11. Isotopic effects in the dehydrogenation of (CH3)2CHOH (
*
and (CH3)2CDOH ( ) with Ag/Al2O3-5 as catalyst. Reaction conditions:
alcohol (1.0 mmol), Ag/Al2O3-5 (2.0 mol%), toluene (3 mL), 353 K.
mechanism was also confirmed by a comparison of the reac-
tion rate of a possible intermediate—2-propoxide—and the
overall reaction rate as follows. The rate of the liquid-phase
propan-2-ol dehydrogenation at 353 K (0.83 mmolgꢀ1 hꢀ1)
was the same order of magnitude as the rate (ca.
1.98 mmolgꢀ1 hꢀ1) estimated from the rate constant for the
transient reaction of the 2-propoxide conversion to acetone
at 353 K (Figure 8) and the initial surface coverage of the 2-
propoxide.
Scheme 1. Proposed mechanism for the alcohol dehydrogenation by Ag/
Al2O3.
2-ol dehydrogenation as an example. The first step of the re-
action is proton abstraction from the alcohol by a basic alu-
mina OH-m2-Al group to yield an alkoxide group on an
AlCUS site (Lewis acid site) and a hydrogen-bonded water
molecule on an acid site on the alumina surface. Alkoxide
species at the interface between a silver cluster and alumina
undergo hydride abstraction to yield hydride species on the
silver cluster and acetone via a transition state with a posi-
Figure 12 shows the relationship between logACTHNUTRGNE(UNG kX/kH) and
the Brown–Okamoto (s+) or Hammett (s) parameters for
the competitive dehydrogenation of benzyl alcohol and p-
substituted benzyl alcohols. Dehydrogenation of a series of
different benzylic alcohols was catalyzed by Ag/Al2O3-5,
and the order of reactivity found for substituted benzyl alco-
hols was p-CH3O > p-CH3 > p-H > p-Cl. There is fairly
ꢀ
tive charge at the a-carbon atom. The C H cleavage step is
good linearity between log
N
rate-determining, as evidenced by the observation of the pri-
mary isotopic effects on 2-propoxide conversion to acetone
(Figure 8) and overall propan-2-ol dehydrogenation
(Figure 11). Next, a hydride ion on the silver cluster reacts
with a water molecule adsorbed on a neighboring Lewis
acid site to yield dihydrogen, accompanied by regeneration
of a basic alumina OH group. It is reasonable to assume
that Brønsted acidities of hydrogen atoms of adsorbed water
are increased under the influence of alumina Lewis acid
sites. This acid–base mechanism should facilitate the release
of dihydrogen. The resulting hydride-free silver cluster and
a basic alumina OH-m2-Al group are available in the next
catalytic cycle.
giving a negative slope (1=ꢀ1.78, r2 =0.97). The correlation
with the s+ parameters had better fits than that with the s
parameters. These observations indicate that a transition
state in the rate-determining step of the alcohol dehydration
involves a positive charge at the a-carbon atom adjacent to
the phenyl ring, which is stabilized by electron-donating sub-
stituents. In combination with the results of in situ IR and
kinetic isotope effects, it was concluded that the rate-deter-
If the surface of alumina is highly dehydrated, the above
catalytic cycle can be written as follows [Eq. (5)–(7)], where
dꢀ
ꢀ
AlCUS and Al O represent the surface Lewis acid and
Lewis base sites, respectively.:
ðCH3Þ2CHOHþAlCUSþAlꢀOdꢀ
ð5Þ
! ðCH3Þ2CHꢀOꢀAlþAlꢀOHdþ
R2CHꢀOꢀAlþAgn ! ðCH3Þ2C¼OþAlCUSþAgnꢀHdꢀ
ð6Þ
Figure 12. Hammett plots for competitive dehydrogenation of benzyl al-
cohol and p-substituted benzyl alcohols with Ag/Al2O3-5. logACTHUNGTRENNNUG
versus s+
*
( ) and logACHTUNGTRENNUNG
*
alcohol (1.0 mmol), p-substituted benzyl alcohol (1.0 mmol), Ag/Al2O3-5
AgnꢀHdꢀþAlꢀOHdþ ! H2þAlꢀOdꢀþAgn
(2.0 mol%), toluene (3 mL). Slope (1+)=ꢀ 1.78 (r2 =0.97).
ð7Þ
2348
www.chemeurj.org
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 2341 – 2351

Heterogeneous Catalysis
FULL PAPER
Although this tentative mechanism may not be operative
under the conditions of the catalytic experiments in this
study, under which the catalyst cannot be highly dehydrated,
the mechanism also indicates the importance of cooperation
of acid–base sites with silver sites. The base sites are needed
for proton abstraction from alcohol, the silver site is indis-
face basic site. The surface of silica does not have such sites
and hence cannot form 2-propoxide groups. The low catalyt-
ic activity of Ag/SiO2 for alcohol dehydration can therefore
be explained as follows: the basicity of the surface OH
groups on silica is too weak to abstract proton from alcohol.
IR spectra for propan-2-ol adsorption on Ag/MgO-5 are
shown in Figure 6B. The spectrum taken immediately after
propan-2-ol adsorption (t=60 s) showed bands due to 2-
propoxide groups, and the spectral feature is basically iden-
tical to those for Ag/Al2O3-5 and MgO (Figure 5). There are
two remarkable differences between the spectra of Ag/
Al2O3-5 and Ag/MgO-5. The band around 1640 cmꢀ1 due to
the HOH bending vibration of adsorbed water was not ob-
served in the case of Ag/MgO-5. In the OH stretching
region the band around 3300 cmꢀ1 due to hydrogen-bonded
water is negligible, whereas a band at 3627 cmꢀ1 is observed.
The band at 3627 cmꢀ1 has been assigned to a hydrogen-
ꢀ
pensable in C H dissociation step, and the acid sites pro-
mote the release of dihydrogen.
Origin of size- and support-specific silver catalysis: In this
section we discuss mechanistic reasons for the strong effects
of the particle size and oxidation state of the silver species
(Figures 2, 3) and the oxide support (Figure 4) on the basis
of the in situ FTIR result (Figures 5, 6). To check the ability
of ionic Ag+ species on alumina to activate the a-C H
ꢀ
bond of 2-propoxide, we performed in situ IR experiments
with Ag/Al2O3-0.5 under the same conditions as in Figure 6.
Note that most of the silver species in Ag/Al2O3-0.5 are
present as Ag+ ions dispersed on the alumina surface
(Figure 1). It was found that 2-propoxide species on this cat-
alyst did not convert to acetone at 373 K (result not shown).
ꢀ
bonded Mg OH···O species, where O represents an adjacent
surface oxygen atom.[58] At t=300 s, a band due to adsorbed
ꢀ1
ꢀ
acetone was observed at 1700 cm . The band due to Mg
OH was shifted slightly to 3560 cmꢀ1 but remained after
1200 s. In contrast, the hydrogen-bonded water on Ag/Al2O3
was practically absent after 1200 s. From these results, it is
suggested that silver clusters on Ag/MgO-5 catalyze hydride
elimination from the 2-propoxide group on MgO to produce
acetone and water, which may react with a coordinatively
ꢀ
This indicates that the silver cluster is necessary in the a-C
H activation step. It is well known that the numbers of cor-
ners, edges, and flat surfaces on metal nanoparticles change
with particle size. Statistic calculations for Au, which pos-
sesses
a similar atomic diameter (0.288 nm) to Ag
(0.289 nm), were reported by Janssens et al.,[2] who assumed
Au clusters shaped as the top half of a truncated octahe-
dron. It has been shown that that the fraction of Au corner
atoms decreases from ca. 64% to ca. 16% when the average
size increases from 0.8 nm to 1.6 nm, whereas the fractions
of edge and plane Au atoms increase slightly. Thus, below
2 nm, the fractions of Ag atoms at corner sites decease dra-
matically with Ag particle size. A sharp drop in TOF per
surface Ag atom in a size below 2 nm as shown in Figure 3
suggests that the alcohol dehydrogenation by Ag/Al2O3 cata-
lyst is a structure-sensitive reaction and that the corner Ag
atoms might be the active species. In view of the role of the
silver cluster proposed in the above section, it is suggested
unsaturated Mg site to produce Mg OH. The protonic
(electrophilic) nature of the surface OH groups for MgO is
ꢀ
ꢀ
weaker than that in alumina, and so Mg OH groups are less
reactive with nucleophilic hydride species on silver clusters
to produce hydrogen molecules. When the oxidation of 4-
methylbenzyl alcohol was performed under oxygen as the
hydrogen accepter, silver clusters on the support with low
electronegativity (basic character) showed higher activity
than under oxygen-free conditions (Figure 4). This supports
the idea that under the oxygen-free conditions protonic
(electrophilic) OH groups play an important role in the re-
moval of hydride species from the silver sites. The acid-as-
sisted decomposition of silver hydride is supported by the
experimental evidence of a similar system reported by Baba
et al.[14–16] By in situ 1H MAS NMR, they studied the thermal
stability of silver hydride species in zeolite, a porous alumi-
nosilicate with Brønsted acid sites, and made the direct ob-
servation that hydride species on Ag clusters reacted with
ꢀ
that corner Ag atoms play an important role in the a-C H
activation step.
With regard to the support effect, the result shown in
Figure 4 demonstrates that both supports with low electro-
negativity (basic oxides) and oxides with high electronega-
tivity (an acidic oxide, SiO2) gave low yields, whereas the
support with a moderate acid and basic nature—alumina—
gave the highest yield. As mentioned above, the spectrum
for propan-2-ol adsorption on MgO is basically identical to
that for Al2O3 (Figure 5), indicating the formation of 2-prop-
oxide groups on the surface of both these oxides. When
propan-2-ol was introduced onto SiO2, on the other hand,
the bands due to 2-propoxide groups were hardly observed
(Figure 5).
+
acidic protons to form H2 and Ag3 clusters at 353 K.[14–16]
From the above discussions, the volcano-shaped correla-
tion between the electronegativity and alcohol dehydration
activity (Figure 4) can be explained as follows. The basic
(nucleophilic) OH groups present on the oxide surface are
responsible for the formation of alkoxide intermediates, but
their strongly basic natures will reduce the protonic nature
ꢀ
of water (or M OH species). As a consequence, the rate of
alcohol dehydrogenation is reduced by slow removal of hy-
dride species. On the other hand, with acidic or neutral
metal oxide (SiO2), dissociative adsorption of propan-2-ol is
diminished by the weak nucleophilic character of the surface
OH groups.
The chemisorption and reaction of propan-2-ol on some
metal oxides have been extensively studied.[49,50,56,57] It is es-
tablished that the initial step of the dissociative adsorption
of alcohols on metal oxides is proton abstraction by a sur-
Chem. Eur. J. 2009, 15, 2341 – 2351
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
2349

K.-i. Shimizu et al.
nitrate 2-hydrate in distilled water by gradual addition of an aqueous
NH4OH solution (1.0 moldmꢀ3), filtration of precipitate, washing three
times with distilled water, and drying at 373 K for 24 h in air. Ag/MOx
catalysts (Ag=5 wt%, MOx =CeO2, MgO, ZrO2, SiO2) were prepared by
impregnation of an oxide support with an aqueous solution of silver ni-
trate, followed by evaporation to dryness at 393 K and by calcination in
air at 773 K for 3 h.
Conclusion
The alumina-supported silver cluster catalyst Ag/Al2O3 acts
as an effective heterogeneous catalyst for oxidant-free alco-
hol dehydrogenation. The activity of the Ag/Al2O3 depends
strongly on the charge and the size of silver species. Silver
clusters, possibly having a slightly cationic natures, show
higher rates per surface Ag site than monomeric Ag+ ions
and bulk Ag particle. Silver clusters with smaller particle
sizes, especially below 1 nm, give higher intrinsic activity, in-
dicating that this reaction is a structure-sensitive reaction
demanding coordinatively unsaturated Ag sites, such as
corner sites. It is proposed that the catalytic reaction pro-
ceeds in the following sequence: 1) reaction between the al-
cohol and a basic alumina OH group to yield the alkoxide
In situ FTIR: In situ IR spectra were recorded with a JASCO FT/IR-620
instrument fitted with a quartz IR cell with CaF2 windows and connected
to a conventional flow reaction system. The sample was pressed into a
self-supporting wafer (50 mg) and mounted in the quartz IR cell. Spectra
were measured with accumulation of 20 scans at a resolution of 4 cmꢀ1
.
A reference spectrum of the catalyst wafer in He taken at the measure-
ment temperature was subtracted from each spectrum. Prior to each ex-
periment the catalyst disk was heated in an O2/He (10%) flow
(100 cm3 minꢀ1) at 823 K for 1 h, followed by cooling to the desired tem-
perature and purging for 30 min in He.
XAFS: Ag K-edge Quick XAFS measurements were performed in trans-
ꢀ
and a water molecule adsorbed on alumina, 2) C H activa-
mission mode at the BL01B1 in the SPring-8. The storage ring was oper-
ated at 8 GeV. A SiACTHUNTGRNEUNG(111) single crystal was used to obtain a monochro-
tion of the alkoxide species by the silver cluster to form a
silver hydride species and a carbonyl compound as the rate-
determining step, and 3) H2 desorption promoted by an alu-
mina acid site. The proposed mechanism provides funda-
mental reasons for the higher activities of silver clusters on
an acid–base bifunctional support (Al2O3) than on basic
(MgO and CeO2) or acidic to neutral (SiO2) ones. Coopera-
tive basic sites at the silver–support interface facilitate bind-
ing of the alcohol substrate to give the alkoxide intermedi-
ate on alumina. Protonic (electrophilic) OH groups at the
interface facilitate the removal of hydride species from the
silver sites to regenerate coordinatively unsaturated sites on
the silver clusters.
matic X-ray beam. For Ag/Al2O3 samples, a self-supported wafer form of
the sample (0.1–1.0 g) of ca. 10 mm diameter was placed in a quartz in
situ cell[59] in an O2 flow (10%) diluted with He (200 cm3 minꢀ1) at atmos-
pheric pressure. For Ag/MOx catalysts (MOx =CeO2, MgO, ZrO2, SiO2),
samples pre-reduced for 10 min under H2 (1 atm) at 373 K were sealed in
cells made of polyethylene under ambient atmosphere, and XAFS spec-
tra were taken at room temperature. Analysis of the extended X-ray ab-
sorption fine structure (EXAFS) was performed with the aid of the REX
version 2.5 program (RIGAKU). The Fourier transformation of the k3-
weighted EXAFS oscillation from k-space to r-space was performed over
the 30–140 nmꢀ1 range to obtain a radial distribution function. The in-
ACHUTNGRENUvNG erseCAHUTNGTRENlNNGU y Fourier filtered data were analyzed with a conventional curve fit-
ting method in the k range of 48–140 nmꢀ1. For the curve-fitting analysis
of Ag/Al2O3 samples, the empirical phase shift and amplitude functions
ꢀ
ꢀ
for Ag Ag and Ag O shells were extracted from the data for Ag foil
and Ag2O, respectively, measured at 573 K. For the curve-fitting analysis
of Ag/MOx samples, the empirically determined phase shift and ampli-
Silver is less expensive than PGM elements, but it has
ꢀ
been regarded as less reactive in C H activation catalysis.
ꢀ
ꢀ
tude functions for Ag Ag and Ag O shells were extracted from the data
for Ag foil and Ag2O, respectively, measured at room temperature.
Fundamental information in this study, demonstrating
design of new alcohol dehydrogenation catalysts by use of a
combination of size-controlled silver and acid–base bifunc-
tional inorganic ligands, should accelerate research in the
Typical procedures for the dehydrogenation of alcohols: Before a reac-
tion, Ag/Al2O3 or Ag/MOx (MOx =CeO2, MgO, ZrO2, SiO2) was treated
in the reaction vessel for 10 min under H2 (1 atm) at 573 or 373 K, re-
spectively, allowed to cool to room temperature, and then exposed to the
ambient atmosphere. Note that XANES and EXAFS results (not shown)
showed that the silver species in Ag/MOx were nearly completely re-
duced. The prereduced Ag/Al2O3-5 (0.05 g, 2.0 mol% Ag) was added to
the mixture of toluene (3.0 mL) and benzyl alcohol (0.11 g, 1 mmol) in a
reaction vessel fitted with a condenser, and the system was placed under
ꢀ
area of C H activation catalysis and allows chemists to
design practical PGM-free catalysts without using organic li-
gands.
N2. The resulting mixture was vigACTHNUGTRENNUGorACHUTGTNRENoUNG usly stirred at 373 K. The reaction
Experimental Section
mixtures were analyzed by GC and GCMS. Levels of conversion of alco-
hols and yields of carbonyl compounds were determined by GC with n-
dodecane as an internal standard. The amounts of evolved H2 was deter-
mined by gas chromatography with a thermal conductivity detector
(TCD), molecular sieve (5 ꢂ) column, and Ar carrier.
General: The GC (Shimadzu GC-17 A) and GCMS (Shimadzu GC-17 A)
analyses were carried out with an Rtx-65 capillary column (Shimadzu)
and nitrogen as the carrier gas. Commercially available organic and inor-
ganic compounds were used without further purification. [D2]propan-2-ol
(98 atom% D) was purchased from Aldrich. The commercially available
PGM-based heterogeneous catalysts Ru/C (Ru=5 wt%, AC4504), Ru/
Al2O3 (Ru=5 wt%, AA-4501), Pd/C (Pd=5 wt%, AC-2501), and Pd/
Al2O3 (Pd=5 wt%, AA-2501) were purchased from N.E. Chemcat Cor-
poration.
[1] M. Haruta, N. Yamada, T. Kobayashi, S. Iijima, J. Catal. 1989, 115,
301.
Catalyst preparation: Ag/Al2O3 catalysts were prepared by impregnating
g-AlOOH (Catapal B alumina purchased from Sasol) with an aqueous
solution of silver nitrate, followed by evaporation to dryness at 393 K.
Before each catalytic or spectroscopic experiment, the precursor was cal-
cined in air at 873 K for 1 h. The Ag/Al2O3 catalysts are designated as
Ag/Al2O3-x, where x is the silver loading (wt%). Ag/Al2O3-5 was used as
a standard catalyst. CeO2 (JRC-CEO-1) and MgO (JRC-MGO-1) were
supplied by the Catalysis Society of Japan. SiO2 (Q-15) was supplied by
[2] T. V. W. Janssens, B. S. Clausen, B. Hvolbak, H. Falsig, C. H. Chris-
tensen, T. Bligaard, J. K. Norskov, Top. Catal. 2007, 44, 15.
[3] T. Ishida, M. Nagaoka, T. Akita, M. Haruta, Chem. Eur. J. 2008, 14,
8456.
[4] H. Tsunoyama, H. Sakurai, Y. Negishi, T. Tsukuda, J. Am. Chem.
Soc. 2005, 127, 9374.
[5] H. Miyamura, R. Matsubara, Y. Miyazaki, S. Kobayashi, Angew.
Chem. 2007, 119, 4229; Angew. Chem. Int. Ed. 2007, 46, 4151.
[6] A. Abad, A. Corma, H. Garcia, Chem. Eur. J. 2008, 14, 212.
Fuji Silysia Chemical. ZrO2 was prepared by hydroACHTUNTRGENNUGlyACHUTGTNRENsNUG is of zirconium oxy-
2350
www.chemeurj.org
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 2341 – 2351

Heterogeneous Catalysis
FULL PAPER
[7] F.-Z. Su, J. Ni, H. Sun, Y. Cao, H.-Y. He, K.-N. Fan, Chem. Eur. J.
2008, 14, 7131.
[33] J. H. Choi, N. Kim, Y. J. Shin, J. H. Park, J. Park, Tetrahedron Lett.
2004, 45, 4607.
[34] H. Junge, M. Beller, Tetrahedron Lett. 2005, 46, 1031.
[35] J. Zhang, M. Gandelman, L. J. W. Shimon, H. Rozenberg, D. Mil-
stein, Organometallics 2004, 23, 4026.
[8] G. J. Hutchings, Catal. Today 2007, 122, 196.
[9] T. Sun, K. Seff, Chem. Rev. 1994, 94, 857.
[10] A. Henglein, Chem. Rev. 1989, 89, 1861.
[36] K.-I. Fujita, N. Tanino, R. Yamaguchi, Org. Lett. 2007, 9, 109.
[37] Y. Ando, M. Yamashita, Y. Saito, Bull. Chem. Soc. Jpn. 2003, 76,
2045.
[38] M. Yamashita, T. Kawamura, M. Suzuki, Y. Saito, Bull. Chem. Soc.
Jpn. 1991, 64, 272.
[39] W.-H. Kim, I. S. Park, J. Park, Org. Lett. 2006, 8, 2543.
[40] T. Mitsudome, Y. Mikami, K. Ebata, T. Mizugaki, K. Jitsukawa, K.
Kaneda, Chem. Commun. 2008, 4804.
[41] R. Noyori, T. Ohkuma, Angew. Chem. 2001, 113, 40; Angew. Chem.
Int. Ed. 2001, 40, 40.
[42] H. Grꢄtzmacher, Angew. Chem. 2008, 120, 1838; Angew. Chem. Int.
Ed. 2008, 47, 1814.
[43] S. E. Clapham, A. Hadzovic, R. H. Morris, Coord. Chem. Rev. 2004,
248, 2201.
[44] M. Ito, A. Osaku, A. Shiibashi, T. Ikariya, Org. Lett. 2007, 9, 1821.
[45] A. Jentys, Phys. Chem. Chem. Phys. 1999, 1, 4059.
[46] C. Morterra, G. Magnacca, Catal. Today 1996, 27, 497.
[47] M. Digne, P. Sautet, P. Raybaud, P. Euzen, H. Toulhoat, J. Catal.
2004, 226, 54.
[48] G. Busca, Phys. Chem. Chem. Phys. 1999, 1, 723.
[49] P. F. Rossi, G. Busca, V. Lorenzelli, O. Saur, J. C. Lavalley, Langmuir
1987, 3, 52.
[50] M. I. Zaki, M. A. Hasan, L. Pasupulety, Langmuir 2001, 17, 4025.
[51] A. R. McInroy, D. T. Lundie, J. M. Winfield, C. C. Dudman, P. Jones,
D. Lennon, Langmuir 2005, 21, 11092.
[52] H. A. Al-Abadleh, V. H. Grassian, Langmuir 2003, 19, 341.
[53] A. R. McInroy, D. T. Lundie, J. M. Winfield, C. C. Dudman, P. Jones,
S. F. Parker, D. Lennon, Catal. Today 2006, 114, 403.
[54] A. Dyan, P. Cenedese, P. Dubot, J. Phys. Chem. B 2006, 110, 10041.
[55] M. I. Zaki, M. A. Hasan, F. A. Al-Sagheer, L. Pasupulety, Langmuir
2000, 16, 430.
[11] G. A. Ozin, M. D. Baker, J. Godber, J. Phys. Chem. 1984, 88, 643.
[12] H. Beyer, P. A. Jacobs, J. B. Uytterhoeven, 1977, 73, 1111.
[13] Y. Suzuki, N. Matsumoto, T. Ainai, T. Miyanaga, H. Hoshino, Poly-
hedron 2005, 24, 685.
[14] T. Baba, Y. Tohjo, T. Takahashi, H. Sawada, Catal. Today 2001, 66,
81.
[15] T. Baba, H. Sawada, T. Takahashi, M. Abe, Appl. Catal. A 2002,
231, 55.
[16] T. Baba, Y. Iwase, K. Inazu, D. Masih, A. Matsumoto, Microporous
Mesoporous Mater. 2007, 101, 142.
[17] P. Claus, H. Hofmeister, J. Phys. Chem. B 1999, 103, 2766.
[18] Y. Chen, C. Wang, H. Liu, J. Qiu, X. Bao, Chem. Commun. 2005,
5298.
[19] Z. Qu, W. Huang, M. Cheng, X. Bao, J. Phys. Chem. B 2005, 109,
1584.
[20] K. Shimizu, A. Satsuma, Phys. Chem. Chem. Phys. 2006, 8, 2677.
[21] K. Shimizu, M. Tsuzuki, K. Kato, S. Yokota, K. Okumura, A. Satsu-
ma, J. Phys. Chem. C 2007, 111, 950.
[22] K. Shimizu, K. Sugino, K. Kato, S. Yokota, K. Okumura, A. Satsu-
ma, J. Phys. Chem. C 2007, 111, 1683.
[23] T. Mitsudome, Y. Mikami, H. Funai, T. Mizugaki, K. Jitsukawa, K.
Kaneda, Angew. Chem. 2008, 120, 144; Angew. Chem. Int. Ed. 2008,
47, 138.
[24] T. Mitsudome, S. Arita, H. Mori, T. Mizugaki, K. Jitsukawa, K.
Kaneda, Angew. Chem. 2008, 120, 8056; Angew. Chem. Int. Ed.
2008, 47, 7938.
[25] T. Matsumoto, M. Ueno, N. Wang, S. Kobayashi, Chem. Asian J.
2008, 3, 196.
[26] K. Mori, T. Hara, T. Mizugaki, K. Ebitani, K. Kaneda, J. Am. Chem.
Soc. 2004, 126, 10657.
[27] K. Yamaguchi, N. Mizuno, Chem. Eur. J. 2003, 9, 4353.
[28] M. S. Sigman, D. R. Jensen, Acc. Chem. Res. 2006, 39, 221.
[29] T. Mallat, A. Baiker, Chem. Rev. 2004, 104, 3037.
[30] F. Zaccheria, N. Ravasio, R. Psaro, A. Fusi, Chem. Eur. J. 2006, 12,
6426.
[56] D. Kulkarni, I. E. Wachs, Appl. Catal. A 2002, 237, 121.
[57] M. Seman, MJ. N. Kondo, K. Domen, C. Reed, S. T. Oyama, J. Phys.
Chem. B 2004, 108, 3231.
[58] C. Chizallet, G. Costentin, M. Che, F. Delbecq, P. Sautet, J. Am.
Chem. Soc. 2007, 129, 6442.
[31] G. R. A. Adair, J. M. J. Williams, Tetrahedron Lett. 2005, 46, 8233.
[32] G. B. W. L. Ligthart, R. H. Meijer, M. P. J. Donners, J. Meuldijk,
J. A. J. M. Vekemans, L. A. Hulshof, Tetrahedron Lett. 2003, 44,
1507.
[59] K. Okumura, K. Yoshino, K. Kato, M. Niwa, J. Phys. Chem. B 2005,
109, 12380.
Received: October 27, 2008
Published online: January 21, 2009
Chem. Eur. J. 2009, 15, 2341 – 2351
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
2351