Monoalkylations with alcohols by a cascade reaction on bifunctional solid catalysts: Reaction kinetics and mechanism

Article abstract of DOI:10.1016/j.jcat.2011.01.029

A bifunctional catalytic system formed by Pd on MgO catalyzes the cascade process between benzyl alcohol and phenylacetonitrile, diethylmalonate and nitromethane, to give the respective α-monoalkylated products without external supply of hydrogen. The process involves a series of three cascade reactions occurring on different catalytic sites. The alcohol undergoes oxidation to the corresponding aldehyde with the simultaneous formation of a metal hydride; then, the aldehyde reacts with a nucleophile formed "in situ" to give an alkene, and finally, the hydrogen from the hydride is transferred to the alkene to give a new C-C bond. A kinetic study on the α-monoalkylation reaction of benzylacetonitrile with benzyl alcohol reveals that the rate-controlling step for the one-pot reaction sequence is the hydrogen transfer reaction from the surface hydrides to the olefin, and consequently, the global reaction rate is improved when decreasing the size of the Pd metal particle.

Full text of DOI:10.1016/j.jcat.2011.01.029

Journal of Catalysis 279 (2011) 319–327
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Monoalkylations with alcohols by a cascade reaction on bifunctional solid
catalysts: Reaction kinetics and mechanism
⇑
⇑
Avelino Corma , Tania Ródenas, María J. Sabater
Instituto de Tecnología Química, Universidad Politécnica de Valencia–Consejo Superior de Investigaciones Científicas, Avenida Los Naranjos s/n, 46022 Valencia, Spain
a r t i c l e i n f o
a b s t r a c t
Article history:
A bifunctional catalytic system formed by Pd on MgO catalyzes the cascade process between benzyl alco-
Received 23 November 2010
Revised 26 January 2011
Accepted 27 January 2011
Available online 5 March 2011
hol and phenylacetonitrile, diethylmalonate and nitromethane, to give the respective
a-monoalkylated
products without external supply of hydrogen. The process involves a series of three cascade reactions
occurring on different catalytic sites. The alcohol undergoes oxidation to the corresponding aldehyde
with the simultaneous formation of a metal hydride; then, the aldehyde reacts with a nucleophile formed
‘‘in situ’’ to give an alkene, and finally, the hydrogen from the hydride is transferred to the alkene to give a
new C–C bond.
Keywords:
Bifunctional catalyst
Pd-MgO
Cascade reaction
Hydrogen transfer
Metal nanoparticle
Palladium dihydride
Basic supports
A kinetic study on the a-monoalkylation reaction of benzylacetonitrile with benzyl alcohol reveals that
the rate-controlling step for the one-pot reaction sequence is the hydrogen transfer reaction from the sur-
face hydrides to the olefin, and consequently, the global reaction rate is improved when decreasing the
size of the Pd metal particle.
Ó 2011 Elsevier Inc. All rights reserved.
1. Introduction
the activated methylenic group of the second reactant to give the
corresponding carbanion. Both the carbanion and the carbonyl
Reactions directing to the formation of C–C bonds are of much
interest in synthetic organic chemistry. Among them, the Knoeve-
compound will react to give an olefinic condensation product
and, finally, the metal hydride will transfer the hydrogen to the
double bond to give the final saturated product (see Scheme 1).
This sequential process is based on hydrogen transfer from an
alcohol that will act as the hydrogen source [2–9] and may also
have interesting and useful applications when the intermediate
carbonyl compound is unstable, since it will react in situ, and its
isolation and purification will not be necessary. Moreover, the pro-
cess described above can be a convenient procedure especially
when a chemoselective hydrogenation of the double bond is re-
quired [9b,c].
Typical homogeneous catalysts such as Ru-, Rh-, and Ir-based
complexes in d6 and d8 electronic configuration are recognized to
be most efficient catalysts for transfer hydrogenation while, with
the exception of Os, other second or third row elements seem to
be much less suited for this catalysis [10–12]. In general, transfer
hydrogenation with the above metal complexes involves the use
of an alcohol that will serve as hydrogen source with the aid of a
metal catalyst, and an alkaline agent that is necessary to activate
the C–H bond through formation of the alkoxide ion.
nagel condensation affords
a–b conjugated enones through the
nucleophilic addition of active methylene compounds to carbonyl
groups, followed by elimination of a molecule of water. For the
Knoevenagel reaction, basic catalysts abstract a proton from the
activated methylenic group and form the intermediate carbanion
(nucleophile) that will react with the carbonyl group to give the
olefin. The double bond can be further hydrogenated in a subse-
quent step to give a saturated compound.
The two reaction steps previously described (Knoevenagel con-
densation and hydrogenation) can be carried out in one-pot reac-
tion by means of bifunctional metal–base solid catalysts, while
working in the presence of hydrogen [1]. However, there is another
way to achieve the final hydrogenated product without the need of
added hydrogen. This will consist of a one-pot reaction between an
alcohol and a molecule having an activated methylenic group in
the presence of a bifunctional metal/base solid catalyst.
In this case, the global process involves in the first step an alco-
hol which is dehydrogenated on the metal function to yield the
corresponding aldehyde (or ketone) and a metal hydride. Mean-
while, the basic function of the catalyst will abstract a proton from
Here, we report the formation of C–C bonds starting from an
alcohol, by
a-monoalkylation reactions using a Pd–MgO bifunc-
tional catalyst. In this case, the metal will perform a dehydrogena-
tion to give the intermediate carbonyl compound and a palladium
dihydride [13]. Then, the reaction of the carbonyl with a variety of
⇑
Corresponding authors. Fax: +34 96 3877809.
E-mail address: acorma@itq.upv.es (A. Corma).
0021-9517/$ - see front matter Ó 2011 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2011.01.029

320
A. Corma et al. / Journal of Catalysis 279 (2011) 319–327
2.2. Catalytic tests
R1
M
2.2.1.
a-Monoalkylation of phenylacetonitrile with benzyl alcohol on
R1CH2OH
metal–MgO as catalysts
EWG
R1
R2
A
mixture of benzyl alcohol (1 mmol), phenylacetonitrile
(3 mmol), 0.0998 g of metal–MgO (Pd: 0.0075 mmol; Au:
0.0075 mmol; Pt: 0.0075 mmol), trifluorotoluene (1 ml) and n-
dodecane (20 ll) as internal standard were placed into an auto-
clave. The resulting mixtures were vigorously stirred at 180 °C
under nitrogen, being monitored by GC.
R1CHO
EWG
R2
M-H
B
2.2.2. Catalyzed
alcohol
a-monoalkylation of malonate diester with benzyl
A
mixture of benzyl alcohol (1 mmol), diethylmalonate
(3 mmol), 0.0998 g of Pd–MgO (Pd: 0.0075 mmol), trifluorotoluene
(1 ml), and n-dodecane (20 l) as internal standard were placed
R2CH2EWG
Scheme 1. One-pot
a-monoalkylation reaction of methylene compounds with
l
alcohols in the presence of a bifunctional metal–base catalyst (EWG = electron
withdrawing group).
into an autoclave. The resulting mixture was vigorously stirred at
180 °C (or 100 °C) under nitrogen. The reaction was monitored
by GC.
methylene compounds with different pKs will produce the olefinic
condensed products. Finally, the olefins will be hydrogenated to
provide the corresponding C–C bond. In this process, which can
be considered an extension of the Wittig reaction (well-known
method for the conversion of aldehydes into alkenes), an alcohol
will serve dually as hydrogen donor and as precursor to the car-
bonyl electrophile, avoiding the use of homogeneous toxic alkylat-
ing agents as well as the formation of dialkylated products.
2.2.3. Catalyzed
alcohol
a-monoalkylation of nitromethane with benzyl
Given that oxidation of benzyl alcohol did not occur in the pres-
ence of nitromethane (probably due to a strong competitive
adsorption of the last molecule on the metal active sites), in this
case, the one-pot synthesis was performed in such a way that
nitromethane was incorporated when practically all benzyl alcohol
was dehydrogenated as follows:
A mixture of benzyl alcohol (1 mmol), 0.0998 g of Pd–MgO (Pd:
0.0075 mmol), trifluorotoluene (1 ml), and n-dodecane (20 ll) as
2. Experimental
internal standard were placed into an autoclave. The resulting mix-
ture was vigorously stirred at 180 °C under nitrogen being moni-
tored by GC. After completing the alcohol transformation to
benzaldehyde, nitromethane (3 mmol) was added, and the result-
ing mixture was vigorously stirred at 180 °C.
Hydroxyapatite (HAP) and hydrotalcite (HT) were prepared fol-
lowing reported procedures [14,15]. A MgO sample with a surface
area of 670 m2/g was purchased from NanoScale Materials. Inor-
ganic salts Pd(acac)2, NaAuCl4, KAuCN4, and AuPPh3Cl were pur-
chased from Aldrich and Pt(acac)2 was from Acros. The reactants
were used without further purification.
3. Results and discussion
2.1. Preparation of catalysts
The reactivity of Pd on different basic supports such as MgO, Al–
Mg hydrotalcite (HT), and hydroxyapatite (HAP) was studied for
2.1.1. Preparation of metal (M)–MgO (M = Pd, Pt, Au) bifunctional
catalysts
Pd–MgO (0.8 wt.% palladium) was prepared following the pro-
cedure reported in [13].
the a-monoalkylation of phenylacetonitrile with benzyl alcohol
as model reaction. 2,3-Diphenylacrylonitrile (1) and 2,3-diphenyl-
propionitrile (2), together with small amounts of benzene and tol-
uene, were obtained as reaction products. Better yields of the
desired final product (2) were obtained with Pd on MgO than on
HT or HAP (see entries 1–3, Table 1).
In fact when Pd was supported on Al–Mg hydrotalcite (HT), the
resultant catalyst was active but the yield and selectivity to 2 and
even 1 + 2 were lower than with MgO. The situation was even
worst with HAP, since in this case the basicity was too weak and
the catalyst was inactive (see entries 1–3, Table 1). Therefore, on
the bases of the results obtained, MgO was selected as the basic
component of the bifunctional catalyst.
Fig. 1 shows the evolution with time of the different products
obtained when benzyl alcohol and phenylacetonitrile were reacted
in the presence of the bifunctional Pd–MgO catalyst.
A priori, formation of the alkene 2,3-diphenylacrylonitrile 1 and
2,3-diphenylpropionitrile 2 can be explained via oxidative removal
of hydrogen from benzyl alcohol to afford benzaldehyde (not de-
tected by gas chromatography in the presence of the arylnitrile)
Pt–MgO (1 wt.% metal loading) was obtained by adding 1 g of
MgO (670 m2/g) to 30 ml anhydrous dichloromethane solution
containing Pt(acac)2 (24.01 mg, mmol) under stirring for 12 h.
After evaporation of the solvent at reduced pressure, the resultant
solid was dried overnight at 353 K in vacuum and then calcined in
nitrogen flow at 823 K (heating rate: 5 °C/min) for 3.5 h. The sam-
ple was activated before reaction by heating the solid at 723 K un-
der air atmosphere for 5 h and then for 5 h under nitrogen. Metal
reduction was performed by heating the solid at 523 K in a flow
of H2/N2 (90/10) for 2 h.
Au–MgO (1 wt.% metal loading) was prepared as follows [16]:
1 g of MgO (670 m2/g) were added to 30 ml of ethanol solution
containing Au(CH3)2(acac) (17.963 mg, 0.055 mmol) under stirring
for 12 h. The solvent was evaporated at reduced pressure, and the
solid was dried overnight at 353 K under vacuum and then cal-
cined in nitrogen flow at 823 K (ramp rate: 5 °C/min) for 3.5 h.
The sample was activated before reaction by heating the solid at
723 K under air for 5 h and then for 5 h under nitrogen. Metal
reduction was performed by heating the solid at 523 K in a flow
of H2/N2 (90/10) for 2 h.
and
a characteristic palladium dihydride intermediate (see
Scheme 2) [13,17]. Then, in the presence of the solid base MgO,
phenylacetonitrile is activated to give a nucleophile which will
rapidly condense with benzaldehyde to give the condensation
product 1, which is hydrogenated by the metal hydride yielding
Pd–HAP and Pd–HT were prepared according to previous re-
ported procedures [14,15].
the a-monoalkylated product 2 according to Scheme 2.

A. Corma et al. / Journal of Catalysis 279 (2011) 319–327
321
Table 1
Pd
CH22OH
CN
CN
Results on a-monoalkylation reaction of phenylacetonitrile with benzyl alcohol using
diverse bifunctional solid catalysts.a
Ph
Ph
2
Pd-MgO
+
+
PhCH CN
2
PhCH OH
2
CN
CN
CHO
N
2
Ph
Ph
2
1
.
Pd-H2
MgO
1
Entry Catalyst
Cb (%) Yieldc (%)
Time (h) TONd
1
2
PhH
PhCH3
1
2
3
4
5
6
7
Pd–MgO
Pd–HT
Pd–HAP
Pd–C
99
100
0
43
66
60
83
3
6
0
94
79
0
2
8
0
1
7
0
15
3
24
21
0
24
48
459
137
0
CH2CN
0
0
11
2
78
Au–MgO
20
60
0
0
120
110
120
Au1+–MgOe
Traces Traces 66
Au1+
–
74 11
0
Traces 120
Scheme 2. Schematic representation of
tonitrile with benzyl alcohol catalyzed by Pd–MgO.
a-monoalkylation reaction of phenylace-
MgOe,f
8
Pt–MgO
95
85 10
0
0
116
207
a
Reaction conditions: benzyl alcohol (1 mmol), benzylnitrile (3 mmol), n-dode-
cane (0.1 mmol), Pd–MgO (0.0075 mmol Pd), 1 ml trifluorotoluene, T = 180 °C.
b
The influence of substituents at the aromatic alcohol was also
studied, and the results showed that both the position as well as
the nature of the substituent had a significant effect on the alkyl-
ation reaction (the composition and yield of products are collected
in Table 2).
Conversions were determined by GC on the basis of benzyl alcohol
consumption.
c
Determined by GC.
mmol of substrate converted/mmol catalyst.
AuPPh3Cl deposited on MgO.
Reaction carried out at 100 °C.
d
e
f
Thus, the introduction of one nitro group at the para position of
benzyl alcohol led to high yields of monoalkylated product P2 (en-
try 2, Table 2) after prolonged reaction time; whereas the same
electron withdrawing group at the meta position gives lower con-
version and yield of P2 (entry 8, Table 2). With electron donor
groups at the para position (amino and methoxy groups), similar
high yields of saturated product P2 could be obtained in both cases
(entries 3, 4 in Table 2).
The observation that p-substituents having electron withdraw-
ing or electron donating character both reduce the apparent reac-
tivity of benzyl alcohol is not evident. Probably, this should be
attributed to a preferential adsorption of these groups over the hy-
droxy group, a fact that may slow down the transformation of the
alcohol into the corresponding aldehyde.
The introduction of two fluorine atoms at the ortho position re-
sulted in a net electron withdrawing effect and an increase of the
Fig. 1. Plot showing the yields of reaction products versus time in the
a-
steric hindrance for
a-monoalkylation. This fact accounted for
monoalkylation reaction of phenylacetonitrile with benzyl alcohol to give the
saturated product: (a) 2,3-diphenylpropionitrile 2, (b) the alkene 2,3-diphenylacr-
ylonitrile 1, (c) toluene, and d) benzene (entry 1, Table 1). The inset shows the
evolution of PHCH2OH and reaction products at short reaction times: (a) 2,3-
diphenylpropionitrile 2, (b) 2,3-diphenylacrylonitrile 1, and (e) PhCH2OH.
the longer reaction times required to achieve from low to moderate
yields of P2 (see entries 5 and 7 in Table 2).
Besides phenylacetonitrile, other methylene compounds such
as nitromethane and diethyl malonate were also studied by
applying the same methodology. In this case, the proton abstrac-
tion at the
a-C–H position of the above molecules generates two
It is evident from results presented in Fig. 1 that product 1 is a
transient product since its yield rises and then falls as expected for
a sequential cascade as envisaged. Besides this, there is an induc-
tion period in the appearance of 2 as 1 is formed (see inset
Fig. 1) that can be observed when the plot is obtained at short reac-
tion times. Nonetheless, benzaldehyde was not detected by gas
chromatography in the presence of the arylnitrile, even at short
reaction times.
In this case, the formation of very minor amounts of toluene and
benzene can be explained by hydrogenolysis of benzyl alcohol and
decarbonylation of the intermediate benzaldehyde product,
respectively. On the other hand, control reactions showed that
either MgO or palladium separately did not catalyze the one-pot
reaction albeit, in the absence of base, palladium (supported on
carbon) still catalyzes at lower rates the formation of benzalde-
hyde, producing meanwhile higher yields of toluene and benzene
than Pd–MgO (see entry 4, Table 1).
different nucleophiles that immediately add to the aldehyde
formed by alcohol dehydrogenation, giving the unsaturated
compounds 3 and 5, respectively (see Scheme 3 and entries 3–5
in Table 3).
Subsequent reduction of these two products by the intermedi-
ate palladium hydride complex formed during the dehydrogena-
tion step leads to formation of the corresponding hydrogenated
products 4 and 6 (see Scheme 3 and entries 3–5 in Table 3).
It is interesting to notice that the reaction with nitromethane
was the slowest one, partly because this compound was not
incorporated during the initial stages of reaction, but after benzyl
alcohol was completely transformed into benzaldehyde. Effec-
tively, given that oxidation of benzyl alcohol did not occur in
the presence of nitromethane (probably due to a strong compet-
itive adsorption of the last molecule on the metal active sites), in
this case, the one-pot synthesis was performed in such a way
that nitromethane was incorporated when practically all benzyl

322
A. Corma et al. / Journal of Catalysis 279 (2011) 319–327
Table 2
a
-Monoalkylation reaction of phenylacetonitrile with benzyl alcohol derivatives over bifunctional Pd–MgO catalyst.a
R1 CN
R1 CN
Pd-MgO
N2
R CH CN
R CH OH
+
2
2
1
2
+
R2
P2
R2
P1
Entry
Substrate
Alcohol
Time (h)
Cb (%)
Yieldc (%)
TONd
459
P1
3
P2
1
2
7
99
94
CN
CN
OH
16
100
11
13
86
73
123
OH
O N
2
3
16
84
119
CN
OH
H N
2
4
5
28
32
93
70
12
12
76
54
107
82
CN
CN
OH
O
F
OH
F
6
7
24
24
73
65
58
10
15
10
98
77
OH
CN
CN
OH
I
8
24
17
4
9
22
CN
OH
NO2
a
b
c
Reaction conditions: alcohol (1 mmol), phenylacetonitrile (3 mmol), n-dodecane (0.1 mmol), Pd–MgO (0.0075 mmol Pd), 1 ml trifluorotoluene, T = 180 °C under N2.
Conversions were determined by GC on the basis of alcohol consumption.
Determined by GC.
mmol of substrate converted/mmol catalyst.
d
alcohol was dehydrogenated (Fig. 2 shows the evolution of con-
version and yields of final hydrogenated products with time).
It is also worth mentioning that the conversion of benzyl alco-
hol was much more rapid in the case of diethyl malonate; however,
contrarily to our initial expectations, the respective hydrogenated
product (product 6) evolved much more rapidly in the case the
phenylacetonitrile (product 2) (see Fig. 2). The reason relies in
the fact that diethyl malonate was immediately transformed into
a mixed diester before being converted into the desired monoalky-
lated product during the early stages of the reaction.
the rate-controlling step of this complex mechanism, since this
should give information on the function of the catalyst that should
be improved. The rate-controlling step was determined by means
of a kinetic study of the
with benzyl alcohol.
a-monoalkylation of phenylacetonitrile
3.1. Optimization of the bifunctional catalyst
3.1.1. Elucidation of the rate-controlling step in the
of phenylacetonitrile with benzyl alcohol
a-monoalkylation
Effectively, in the case of b-diester, the monotransesterification
The three elementary steps of the overall reaction are given in
Eqs. (1)–(3), where ka, kb, and kc are the corresponding rate kinetic
constants:
reaction efficiently competed with
a-monobenzylation to give a
mixed diester derived from malonic acid (benzyl ethyl malonate),
especially at low conversions and low temperatures. In accordance
with this, the mixed diester was isolated as main product when the
reaction was carried out at 100 °C (see entry 5, Table 3). At higher
temperatures (180 °C), the mixed diester (in equilibrium with
k
a
Pd(0)
ð1Þ
ð2Þ
+
PdH2
CN
PhCH2OH
PhCHO
PhCHO
Ph
+
kb
H2O
+
PhCH2CN
+
Ph
diethyl malonate) was detected only at low conversions, since
a-
P1
monoalkylation shifted progressively the equilibrium toward the
alkylated product 6.
Since the slowest step in a multistep reaction will determine
the rate of the entire sequence, we have attempted to identify
Ph
Ph
CN
CN
Ph
k
c
PdH2
+
+
Pd(0)
ð3Þ
Ph
P2
P1

A. Corma et al. / Journal of Catalysis 279 (2011) 319–327
323
CH2OH
In principle, three kinetic rate expressions can be derived by
considering that the dehydrogenation of benzyl alcohol to give
benzaldehyde and metal hydride (Eq. (1)), or the condensation
reaction (Eq. (2)), or the hydrogenation (Eq. (3)) is the overall
rate-controlling step.
Nonetheless, a priori, the possibility that the first step, which is
the dehydrogenation of the alcohol, would be controlling was re-
jected based on the observation that the initial reaction rate r0 is
first order with respect to alcohol ([PhCH2OH]) and first order with
respect to phenylacetonitrile ([PhCH2CN]) (see Fig. 3 and Supple-
mentary Material).
Pd
Pd-H2
CHO
CH2CN
CN
CH3NO2
NO2
3
1
2
O
O
Pd-H2
Pd-H2
In accordance with these results, it may be that either the con-
densation reaction or the hydrogen transfer reaction (Eqs. (2) and
(3)) would be the reaction-controlling step during the one-pot
reaction. Thus, we carried out a series of additional experiments,
in which the dehydrogenation of the alcohol (to give benzaldehyde
and palladium dihydride) was studied in two different experi-
ments separately
EtO
OEt
Pd
Pd
O
O
CN
EtO
OEt
NO2
4
5
Pd-H2
In this case, an initial amount of benzyl alcohol and the catalyst
(Pd–MgO) was incorporated into two separate equal reactors, lead-
ing to formation of PhCHO and Pd–H2 in both cases. At this point,
equimolar amounts of PhCH2CN and the alkene 1 (P1) were incor-
porated into the first and second reactors, respectively, and the ini-
tial reaction rates (r0) for the condensation and hydrogenation
reactions were calculated (see Fig. 4).
Pd
O
O
EtO
OEt
6
Under the above experimental conditions, it was observed that
the hydrogenation of the alkene P1 by the surface hydrides was
slower than the condensation reaction with the activated
Scheme 3. Schematic representation of the
acetonitrile, diethylmalonate, and nitromethane with benzyl alcohol over Pd–MgO.
a-monoalkylation reaction of phenyl-
Table 3
a
-Monoalkylation reaction of phenylacetonitrile, diethyl malonate, and nitromethane with benzyl alcohol over bifunctional Pd–MgO as catalyst.a
Entry
Substrate
Time (h)
Cb (%)
Yieldc (%)
TONd
1
2
3
4
6
PhH
PhCH3
1
PhCH2CN
PhCH2CN
CH3NO2
7
99
43
84
100
3
0
94
0
1
11
0
Traces
15
0
459
78
66
2e
3
24
40
2
6
78
4
99
7
0
0
126
O
O
O
EtO
O
OEt
OEt
5f,g
76
1.5
1.5
94
EtO
a
b
c
d
e
f
Reaction conditions: benzyl alcohol (1 mmol), substrate (3 mmol), n-dodecane (0.1 mmol), Pd–MgO (0.075 mmol Pd), 1 ml trifluorotoluene, T = 180 °C under N2.
Conversions were determined by GC on the basis of benzyl alcohol consumption.
Determined by GC.
mmol of substrate converted/mmol catalyst.
Pd/C (Norit).
Reaction carried out at 100 °C.
The mixed ester benzyl ethyl malonate was obtained as major product (69% yield).
g
100
100
(B)
(A)
80
60
40
20
0
80
60
40
20
0
0
0,25
0,5
0,75
1
0
0,5
1
Time (h)
Time (h)
Fig. 2. Comparative plots showing the (A) conversion of benzyl alcohol against time in the reactions with diethyl malonate (ꢀ), phenylacetonitrile (N), and nitromethane (j),
and (B) yields of compounds 2, 4, and 6 derived from reactions of benzyl alcohol with diethyl malonate (ꢀ), phenylacetonitrile (N), and nitromethane, respectively (j) versus
time in the presence of Pd–MgO as catalyst (0.8%; 0.0075 mmol Pd).

324
A. Corma et al. / Journal of Catalysis 279 (2011) 319–327
(A)
(B)
1,6
1,4
1,2
1
1,6
y = 0,9458x
R2 = 0,9973
1,4
1,2
1
y = 0,463x
R2 = 0,9904
0,8
0,6
0,4
0,2
0
0,8
0,6
0,4
0,2
0
0
0,5
1
1,5
2
0
1
2
3
[PhCH2OH]
[PhCH2CN]
Fig. 3. Graphic representations obtained when plotting (A) r0 versus [PhCH2OH] and (B) r0 versus [PhCH2CN].
Ph
PhCH2CN
H2O
+
k
Ph CN
PhCH2OH
Pd(0)
PhCHO
Pd-H2
-1
(r = 23.4 x 10-3 mmmol x min
0
)
H
H
Pd
Kdes
Kads
Ph
Ph
MgO
Ph CN
Ph CN
Fig. 5. Graphic representation of the hydrogenation of P1 at the surface of Pd–MgO.
-1
(r = 1.9 x 10-3 mmol x min
)
0
not involved in the hydrogen transfer, they can compete with P1
for adsorbing on the metal active sites.
Model 1
Fig. 4. Schematic representation of kinetic experiments to calculate the conden-
sation and hydrogenation reaction rates.
The adsorption of P1 is the rate-controlling step:
methylenic compound. Even though the situation is not the same
from the global reaction point of view, the result supports the con-
clusion that the catalytic function that will determine the rate of
the overall process corresponds to the metal hydride function on
the metal surface (Pd–H) and not the basic function of the catalyst,
i.e., MgO. This is an important conclusion since it indicates which
one of the components of the bifunctional catalyst has to be
improved.
Once it was established that the step corresponding to the
transfer of the metal hydride to the olefin was the controlling step,
we have still to find out if the adsorption of reactants, desorption of
products or the surface reaction was the final controlling step dur-
ing the hydrogenation process. Then, in order to address this ques-
tion, a new kinetic study was undertaken.
k½P1ꢀ
r0
r0
r0
¼
1 þ KdesðP2Þ½P2ꢀ þ Ki½Iꢀ
Model 2
The hydrogenation reaction of P1 is the rate-controlling step:
k½P1ꢀ
¼
1 þ KadsðP1Þ½P1ꢀ þ KdesðP2Þ½P2ꢀ þ Ki½Iꢀ
Model 3
Desorption of P2 is the rate-controlling step.
k½P1ꢀ
¼
1 þ KadsðP1Þ½P1ꢀ þ Ki½Iꢀ
where k is the kinetic rate constant, Kads is the adsorption equilib-
rium constant, Kdes is the desorption equilibrium constant, and Ki
is the adsorption equilibrium constant for other molecules.
A set of 33 kinetic experiments was designed, and the numerical
results were fitted by least squares using the tool Solver on the Ex-
cel 3.0 program (see Table 1S in Supplementary Material) [18]. The
correlation coefficient R2 (or cross-correlation coefficient) was used
as criterium to discern among the best least squares fitting for the
three possible models (see Tables 2S–4S in Supplementary Mate-
rial), and Table 4 collects the calculated parameters.
3.2. Determination of adsorption (Kads), desorption (Kdes), and
reactivity constants (kr) during the hydrogenation transfer step from
Pd–H to P1 [18]
It was assumed in this study that the surface of the solid was
uniform, and a number of adsorption sites were available to the
reactants. On these sites, the condensation product P1 would ad-
sorb on active sites following a Langmuir type isotherm, with Kads
as adsorption constant (Fig. 5).
Then, the adsorbed P1 molecules will react with the metal hy-
drides Pd–H (k), previously formed at the metal surface to give
the hydrogenated product P2, which would be desorbed at the
end (Kdes) (see Fig. 5).
On these premises, three possible rate expressions or models
derived from the Langmuir–Hinshelwood methodology were postu-
lated assuming that either adsorption, desorption, or the surface
reaction were the controlling step in the hydrogenation process
[18]. In this case, the concentrations of PhCHO, PhCH2CN, and P2
were also included in the equations since, even though they are
According to these calculations, the model that best fits the
experimental data corresponds to the surface reaction (model 2).
So again this fact corroborates that the catalytic function that will
determine the rate of overall process with the Pd–MgO catalyst is
indeed the metal hydride function on the surface (Pd–H) of the cat-
alyst and no the basic function of the catalyst, i.e., MgO.
At this point, and in order to increase the rate of hydrogenation,
we have on one hand changed the hydrogenation function (Pd, Pt,
and Au), and on the other hand, the crystallite size of the best oper-
ating metal has also been varied.

A. Corma et al. / Journal of Catalysis 279 (2011) 319–327
325
Table 4
Constant values obtained for the different models in the presence of inert molecules (P2, PhCHO and PhCH2CN) based on least squares adjustment using the tool Solver on the
Excel 3.0 program.
P
Model
k
Kads(P1)
Kdes(P2)
Kads(PhCH2CN)
Kads(PhCHO)
[r0(exp) ꢁ r0(mod)]2
0.024
R2
1
0.803
0.91
0.89
0.001
0.23
0.22
0.592
0.71
0.63
0.083
0.10
0.14
0.058
0.09
0.05
0.96
0.96
0.99
0.95
2
0.020
0.006
0.029
2a
3
0.0867
0.198
0.001
0.0093
0.0093
a
Constant values obtained with [PhCH2CN] = 3 mmol/ml.
3.3. Influence of the nature of the metal
200
a
160
120
80
40
0
Au–MgO was found to be inactive for catalyzing the global reac-
tion, though benzyl alcohol afforded reasonable yields of benzalde-
hyde when a gold (I) salt was deposited on MgO in two additional
experiments either at 100 °C or at 180 °C (see entries 5–7 in
Table 1).
With platinum supported onto MgO, the alcohol was almost
completely transformed into benzaldehyde that condensates to
give 1, but the yields of 2 are much lower than those obtained with
palladium (compare entries 1 and 8 in Table 1). The clear superior-
ity of palladium over gold and platinum for catalyzing this reaction
can be explained by taking into account that the stability of gold
and platinum hydrides are either too low or too high, respectively,
and consequently, they are less reactive than palladium hydrides
[19–22].
This explanation is in line with the previous observation that
the controlling step of the reaction is the hydrogenation of the dou-
ble bond through hydrogen transfer from the metal hydride, and
consequently, the surface hydride concentration and its ability to
release the hydrogen will determine the activity and selectivity
of the catalyst.
b
c
d
d
e
2.0 2.4 2.8 3.2 3.6 4.0 4.4 4.8 5.2
d(Pd0)nm
Fig. 6. Plots showing the initial reaction rates (r0) per surface atom as a function of
metal particle size for: (a) the hydrogenation transfer reaction to afford 2; (b)
dehydrogenation of benzyl alcohol to afford benzaldehyde; (c) condensation
reaction to give 1; (d) hydrogenolysis reaction to give toluene, and (e) decarbony-
lation reaction to afford benzene.
The spectrum of the catalyst recovered after completing the
benzylic alcohol transformation to benzaldehyde (in the absence
of the arylnitrile) was very similar to that of Pd(acac)2 (Fig. 7),
hence suggesting an initial oxidative addition process from Pd0
to afford the intermediate Pd2+ dihydride with simultaneous for-
mation of benzaldehyde.
If Pd was the metal of choice, we could still improve its reactiv-
ity by optimizing the crystallite metal size.
After adding phenylacetonitrile, formation of the Knoevenagel
condensation product 1 occurs being smoothly converted into
the target product 2. At this point, the DR–UV–vis spectrum of
the solid was again recorded, showing that the initial broad band
at 315 nm was again restored. These spectral changes confirm that
the catalysis by palladium involves the transition between the Pd0
and Pd2+ oxidation states [13].
The formation of palladium dihydride species over Pd–MgO has
been indirectly confirmed by the identification of the correspond-
ing deuterium labeled species using NMR spectroscopy [13]. This,
together with the fact that the controlling step of the reaction
3.4. Influence of Pd crystallite size
The concentration of surface hydrides on Pd could be varied by
changing the metal crystallite size. Then, different Pd–MgO cata-
lysts were prepared with average Pd metal particle size ranging
from 2.2 to 5.3 nm, and the number of Pd surface atoms in each
sample was calculated taking into account the total metal content
and crystal size distribution, following the calculation procedure
given in [23a–d].
TOFs were calculated by dividing the initial reaction rate by the
number of surface metal atoms, and the results were plotted versus
the average particle size for the different catalysts (see Fig. 6).
As can be deduced from Fig. 6, it is evident that TOF increases
when decreasing the Pd crystallite size, especially for the hydride
transfer reaction (see curve a) in Fig. 6), while remains practically
constant for those reactions giving the subproducts toluene and
benzene. Then, it appears that the reaction considered here is a
structure sensitive reaction [23e] that requires highly dispersed
Pd on MgO.
2.5
2.0
e)
1.5
d)
3.5. Reaction mechanism
1.0
c)
a)
In order to better define the metal active species and their
mechanistic cycle for the overall process, we have followed the
reaction by DR–UV visible. The results given in Fig. 7 show that a
freshly prepared Pd–MgO catalyst gives a spectrum with a broad
and structureless band at 300 nm that has been assigned to Pd(0)
[24]. On the other hand, the formation of different palladium spe-
cies during reaction was detected through the analysis of the DR–
UV–vis spectra of the solid at different reaction times (Fig. 7).
b)
0.5
250
300
350
400
Wavelength (nm)
Fig. 7. DR–UV–vis spectra of (a) MgO, (b) Pd(acac)2, (c) Pd–MgO recovered after
completing the benzyl alcohol transformation to benzaldehyde (before the arylnit-
rile addition), (d) Pd–MgO recovered at the end of reaction (after the arylnitrile
addition), and (e) fresh prepared Pd–MgO sample.

326
A. Corma et al. / Journal of Catalysis 279 (2011) 319–327
the olefin by the surface metal hydrides being highly dispersed
R
R'
Pd a very adequate metal function.
A molecular mechanistic cycle has been presented which was
supported by kinetic, isotopic, and DR–UV–vis spectroscopic
studies.
The utility of the present catalytic system has to do with the
avoidance of toxic classic alkylating agents and the concurrent for-
mation of undesirable waste salts as well as the formation of dial-
kylated byproducts as side reaction.
Ph
2 (4, 6)
Pd(0)
PhCH OH
2
H
R
R'
Pd
Acknowledgments
H
PhCH O
Pd
2
8
H
Ph
Financial support by Ministerio de Educación y Ciencia e Inno-
vación (Project MIYCIN, CSD2009-00050; PROGRAMA CONSOLID-
ER.INGENIO 2009) and Generalidad Valenciana (GV PROMETEO/
2008/130) is gratefully acknowledged. T.R. thanks to Consejo Supe-
rior de Investigaciones Científicas for an I3-P fellowships.
Pd
H
H
7
R
R'
R
R'
Appendix A. Supplementary data
R'CH R
2
+
H O
PhCHO
2
MgO
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.jcat.2011.01.029.
Ph
Ph
1 (3, 5)
1 ( 3, 5)
References
Scheme 4. Proposed reaction mechanism for the
a-monoalkylation of phenylace-
tonitrile, diethyl malonate, and nitromethane with benzyl alcohol catalyzed by Pd–
MgO under inert atmosphere.
[1] (a) B. Breit, S.K. Zahn, Angew. Chem. Int. Ed. 40 (10) (2001) 1910;
(b) D. Tichit, B. Coq, Cat. Chem. 7 (6) (2003) 206;
(c) D.B. Ramachary, M. Kishor, J. Org. Chem. 72 (14) (2007) 5056;
(d) S. Iborra, A. Corma, J. Catal. 2 (2009) 500. 2;
(e) M.J. Climent, A. Corma, S. Iborra, Chem. Sus. Chem. 2 (2009) 500. 2.
[2] R.L. Augustine, Adv. Catal. 25 (1976) 56.
[3] P. Rylander, Catalytic Hydrogenation in Organic Synthesis, Academic Press,
New York, 1979.
[4] M. Hudlicky, Reductions in Organic Chemistry, Wiley, New York, 1984.
[5] S. Siegel, in: B.M. Trost, I. Fleming (Eds.), Comprehensive Organic Synthesis,
vol. 8, Pergamon Press, Oxford, 1991 (Chapter 3.1).
[6] H. Takaya, R. Noyori, in: B.M. Trost, I. Fleming (Eds.), Comprehensive Organic
Synthesis, vol. 8, Pergamon Press, Oxford, 1991 (Chapter 3.2).
[7] E. Keinan, N. Greenspoon, in: B.M. Trost, I. Fleming (Eds.), Comprehensive
Organic Synthesis, vol. 8, Pergamon Press, Oxford, 1991 (Chapter 3.5).
[8] A.J. Birch, D.H. Williamson, Org. React. 24 (1976) 1.
[9] (a) M. Hudlicky, Reductions in Organic Chemistry, second ed., American
Chemical Society, Washington, DC, 1996 (Chapter 13);
(b) N. Uematsu, A. Fujii, S. Hashiguchi, T. Ikariya, R. Noyori, J. Am. Chem. Soc.
118 (1996) 4916;
corresponds to the transfer of hydride to the alkene 1, led us to
propose the following tentative reaction mechanism for the mono-
alkylation of phenylacetonitrile, and by extension to nitromethane
and malonic ester, on Pd–MgO as catalysts (see Scheme 4).
In this mechanism, palladium dihydride species Pd–H2 (7) are
formed upon direct interaction with the alcohol (and/or through
formation of a palladium alkoxylate intermediate 8). The dihydride
metal species 7 react with the resulting unsaturated compounds 1,
3, and 5 formed after condensation reaction of benzaldehyde with
the respective nucleophile (formed in situ on the basic sites of
MgO) to give the hydrogenated compounds 2, 4, and 6.
3.6. Catalyst recycle
(c) K. Fujita, C. Kitatsuji, S. Furukawa, R. Yamaguchi, Tetrahedron Lett. 45
(2004) 3215.
Reusability experiments have been carried out after separation
of the Pd–MgO catalyst by filtration, followed by an exhaustive
washing with trifluorotoluene and calcination. The resultant sam-
ple was used in a second cycle and showed a similar activity in its
reuse. It was found from ICP-AES that metal leaching from the MgO
surface could not be detected. Moreover, when analyzing by TEM
the solid after recovering, it was observed that the size of Pd nano-
particles was still optimal, while the XRD spectrum of the MgO be-
fore calcination experienced noticeable changes. XRD revealed that
during the process, the support undergoes a phase transformation
to Mg(OH)2 (see Fig. S1 in Supplementary Material) that is trans-
formed back into MgO upon calcination.
[10] S. Gladiali, E. Alberico, Transferhydrogenations, in: M. Beller, C. Bolm (Eds.),
Transition Metals for Organic Synthesis, vol. 2, second ed., Wiley-VCH Verlag
GmbH & Co. KGaA, Weinheim, Germany, 2004.
[11] C. Löfberg, R. Grigg, M.A. Whittaker, A. Keep, A. Derrick, J. Org. Chem. 712
(2006) 8023.
[12] K. Motokura, D. Nishimura, K. Mori, T. Mizugaki, K. Ebitani, K. Kaneda, J. Am.
Chem. Soc. 126 (2004) 5662.
[13] A. Corma, M.J. Sabater, T. Ródenas, Chem. Eur. J. 16 (2010) 254.
[14] (a) Kohsuke Mori, Takayoshi Hara, Tomoo Mizugaki, Kohki Ebitani, Kiyotomi
Kaneda, J. Am. Chem. Soc. 126 (2004) 10657;
(b) A. Abad, C. Almela, A. Corma, H. García, Tetrahedron 62 (2006) 6666;
(c) C. Elliot, Structure and Chemistry of the Apatites and Other Calcium
Orthophosphates, Elsevier, New York, 1994;
(d) C. Elliot, J. Chem. Soc. Faraday Trans. 92 (1996) 293.
[15] F. Cavani, F. Trifirò, A. Vaccari, Catal. Today 11 (1991) 173.
[16] (a) M.J. Climent, A. Corma, S. Iborra, M. Mifsud, J. Catal. 247 (2007) 223;
(b) Y. Hao, M. Mihaylov, E. Ivanova, K. Hadjiivanov, H. Knözinger, B.C. Gates, J.
Catal. 261 (2009) 137.
[17] (a) J.S.M. Samec, J.-E. Bäckvall, P.G. Andersson, P. Brandt, Chem. Soc. Rev. 35
(2006) 237;
(b) O. Pamies, J.-E. Bäckvall, Chem. Eur. J. 7 (2001) 5052;
(c) Y.R.S. Laxmi, J.-E. Bäckvall, Chem. Commun. (2000) 611.
[18] (a) J.M. Smith, Chemical Engineering Kinetics, McGraw-Hill International,
1984;
4. Conclusions
Pd–MgO catalyzes selectively the sequential oxidation of benzyl
alcohol to benzaldehyde and the generation of carbon nucleophiles
by activation of the
a-C–H bond adjacent to different substrates,
i.e., -arylnitrile, b-diester, and nitro compound derivatives. The
a
(b) F. Pukelsheim, J.L. Rosengerger, Experimental designs for model
discrimination, J. Am. Stat. Assoc. 88 (1993) 642;
resulting nucleophiles will rapidly condense to afford a double
(c) P. Zamostny, Z. Belohlav, Identification of kinetic models of
heterogeneously catalyzed reactions, Appl. Catal. A 225 (2002) 291.
[19] (a) J.-P. Deng, W.-C. Shih, C.-Y. Mou, Chem. Phys. Chem. 6 (2005) 2021;
(b) A.S.K. Hashmi, R. Salathe, T.M. Frost, L. Schwarz, J.-H. Choi, Appl. Catal. A
291 (2005) 238.
bond that will be hydrogenated at the end.
A kinetic study for the
a-monoalkylation reaction of benzyl-
acetonitrile with benzyl alcohol shows that the rate-controlling
step for the one-pot reaction sequence is the hydrogenation of

A. Corma et al. / Journal of Catalysis 279 (2011) 319–327
327
[20] (a) M.C. Gimeno, A. Laguna, in: D.E. Fenton (Ed.), Comprehensive Coordination
5d metal atoms than the 3d and 4d metal atoms.;
(c) P. Pyykko, Chem. Rev. 88 (1988) 563.
[23] (a) N. Mahata, V. Vishwanathan, J. Catal. 196 (2000) 262;
Chemistry II, vol. 6, Elsevier, Amsterdam, p. 911 (Chapter 6.7).;
(b) R.J. Puddephatt, in: G. Wilkinson (Ed.), Comprehensive Coordination
Chemistry, vol. 5, Elsevier, Amsterdam, p. 861 (Chapter 55).
[21] G.N. Khairallah, R.A.J. O’Hair, M.I. Bruced, Dalton Trans. (2006) 3699.
[22] (a) E. Suchanek, N. Lange, G. Auffermann, W. Bronger, H.D. Lutz, J. Raman
Spectrosc. 30 (1999) 981;
(b) S. Domínguez-Domínguez, A. Berenguer-Murcia, A. Linares-Solano, D.
Cazorla-Amorós, J.Catal. 257 (2008) 87;
(c) R. Van Hardeveld, F. Hartog, Surf. Sci. 15 (1969) 189;
(d) R.E. Benfield, J. Chem. Soc. Faraday Trans. 88 (1992) 1107;
(e) M. Boudart, Chem. Rev. 95 (3) (1995) 661.
(b) Raman studies which are
a valuable tool in elucidating the relative
strength of metal–hydrogen bonds in transition metal hydrido complexes
confirms that MH stretching modes display much stronger bonds in the case of
[24] (a) A. Corma, H. García, A. Leyva, J. Mol. Catal. A: Chem. 230 (2005) 97;
(b) A. Corma, H. García, A. Leyva, J. Catal. 225 (2004) 350.