A multiphase reaction medium including pressurized carbon dioxide and water for selective hydrogenation of benzonitrile with a Pd/Al2O 3 catalyst

Article abstract of DOI:10.1016/j.apcata.2013.03.004

The hydrogenation of benzonitrile was studied with a Pd/Al ₂O₃ catalyst in the presence of pressurized CO₂ and water. This multiphase reaction system was effective for the selective production of the desired primary amine

Full text of DOI:10.1016/j.apcata.2013.03.004

Applied Catalysis A: General 456 (2013) 215–222
Contents lists available at SciVerse ScienceDirect
Applied Catalysis A: General
journal homepage: www.elsevier.com/locate/apcata
A multiphase reaction medium including pressurized carbon dioxide
and water for selective hydrogenation of benzonitrile with a Pd/Al O
2
3
catalyst
Hiroshi Yoshida, Yu Wang, Satomi Narisawa, Shin-ichiro Fujita, Ruixia Liu,
∗
Masahiko Arai
Division of Chemical Process Engineering, Faculty of Engineering, Hokkaido University, Sapporo 060-8628, Japan
a r t i c l e i n f o
a b s t r a c t
Article history:
2 3
The hydrogenation of benzonitrile was studied with a Pd/Al O catalyst in the presence of pressurized
Received 18 January 2013
Received in revised form 22 February 2013
Accepted 3 March 2013
Available online xxx
CO2 and water. This multiphase reaction system was effective for the selective production of the desired
primary amine of benzylamine with no deactivation of the catalyst. The enhanced selectivity to the
primary amine resulted from its transfer into water phase, which prevented the primary amine from
reacting with an intermediate of imine into the secondary amine in the organic phase (benzonitrile). In
addition, interactions of CO2 molecules with the primary amine could also increase its selectivity. In the
absence of water, the catalyst was deactivated by the accumulation of a carbamate salt on the surface of
catalyst. But this deactivation was avoided by the addition of water because the salt was soluble in water
and it did not deposit on the catalyst existing at interfacial layer between the water and organic phases.
© 2013 Elsevier B.V. All rights reserved.
Keywords:
Benzonitrile
Selective hydrogenation
Carbon dioxide
Water
Multiphase reaction
1
. Introduction
Multiphase reaction mixtures including pressurized CO and/or
hydrogenation but increase the reactivity of these intermediates.
As a result, the desired aniline can be produced at any conversion
level. The pressurization with CO2 can also promote the rate of
nitrobenzene hydrogenation and the extent of promotion depends
on its pressure used. Xie et al. studied hydrogenation of nitrile and
2
water are an interesting medium for organic synthetic reaction
and separation processes [1–3]. When an organic liquid phase
(
substrate, solvent) is pressurized by CO , a large amount of CO2
imine in CO -dissolved expanded organic solvent phases [10]. In
2
2
is dissolved in the liquid. This causes its volume to expand and
heterogeneous hydrogenation of benzonitrile (BN) and phenylace-
tonitrile with NiCl /NaBH4 in CO -expanded ethanol, the yield of
such a liquid is referred to as CO -dissolved expanded liquid
2
2
2
phase [4,5]. The dissolution of CO2 may facilitate the dissolution
of other coexisting gases such as H , O , and CO and promote
the primary amines was improved. The authors mentioned that the
primary amines were protected by CO2 and so their further hydro-
genation to the secondary amines was effectively suppressed. Note,
2
2
the chemical reactions involving these gaseous reactants, as men-
tioned in recent reviews [6–8]. In addition to this physical effect,
CO2 molecules may have a chemical impact on reacting species
and change the product selectivity as well as the overall reac-
tion rate. For example, the complete selective hydrogenation of
nitrobenzene to aniline can be achieved in a whole range of con-
version up to 100% with supported Ni catalysts in the presence
of pressurized CO2 [9]. In situ high-pressure Fourier transform
infrared (FTIR) measurements indicate molecular interactions of
CO2 with the substrate of nitrobenzene and the intermediates of
nitrosobenzene and N-phenylhydroxylamine; the interactions of
CO2 with the nitro group of the substrate decrease its reactivity to
in addition, that the pressurization with CO is also effective for pro-
2
moting the rate of Heck coupling reactions and modifying the yield
and/or the selectivity in Diels–Alder reactions in which no gaseous
reactants are involved [11–13].
Recently the authors found out that the addition of water caused
the promotion effect of CO2 pressurization to appear at a lower
pressure of about 1 MPa for the hydrogenation of nitrobenzene [14].
Hiyoshi et al. communicated that the coexistence of CO and water
2
enhanced the hydrogenation of acetophenone with a Pd/C cata-
lyst [15]. They discussed possible reasons for the rate enhancement
observed and pointed out the importance of pH value in the aque-
ous phase, which was decreased by the dissolution of CO . Those
2
results show that an interesting and effective multiphase reaction
medium may be designed using green components of pressurized
CO2 and water. In the present work, the features of such a reaction
∗ Corresponding author. Tel.: +81 11 706 6594; fax: +81 11 706 6594.
E-mail address: marai@eng.hokudai.ac.jp (M. Arai).
0
926-860X/$ – see front matter © 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.apcata.2013.03.004

2
16
H. Yoshida et al. / Applied Catalysis A: General 456 (2013) 215–222
system have been studied in detail for the selective hydrogenation
of BN to the primary amine of benzylamine (BA) with a Pd/Al O
measured by transmission electron microscopy (JEOL JEM-2000ES)
2
−1
and it had a BET surface area of 98.0 m g and an average pore
diameter of 9.1 nm measured by nitrogen adsorption at 77 K (Quan-
tachrome NOVA 1000). The hydrogenation of BN (Aldrich) was
conducted in a 50 cm3 stainless steel autoclave. The reactor was
charged with 9.7 mmol (1.0 g) BN, 5 cm3 water (Wako), and 20 mg
Pd/Al O , flushed with H for three times to remove the air, and
2
3
catalyst. The hydrogenation runs were conducted under different
conditions and molecular interactions among the substrate, CO2,
water, and the catalyst were measured by in situ high-pressure FTIR
methods [16,17] in order to elucidate the roles of CO2 and water.
The medium of gas (H , CO )–liquid (water)–liquid (BN)–solid
2
2
2
3
2
(
catalyst) has been confirmed to be effective for the selective pro-
heated to a reaction temperature of 323 K in a water bath. After
duction of BA by BN hydrogenation with no catalyst deactivation.
It is important to avoid the deactivation of supported noble metal
catalysts in the presence of both H2 and CO2 at high pressures. For
the introduction of 2 MPa H , liquid CO2 (99.99%) was introduced
2
into the reactor to the desired pressure with a high pressure liquid
pump (JASCO SCF-Get). The reaction was conducted while stirring
the reaction mixture with a magnetic stirrer. After the reaction, the
reactor was cooled in an ice-water bath and depressurized care-
fully. The reaction mixture was analyzed by a gas chromatograph
(GL Sciences GC-4000) using a capillary column (Phenomenex
ZB-5) and a flame ionization detector. The total conversion was
determined from the initial and final amounts of BN. Under the con-
ditions used, BA, dibenzylamine (DBA), and toluene were detected
to form and the product selectivity was determined by the follow-
ing equation: ˛ × (moles of a product)/(moles of BA + 2 x moles of
DBA + moles of toluene), in which ˛ = 1 for BA and toluene and ˛ = 2
for DBA. For the present reaction runs, the stirring speed was set in
a range where it little affected the overall rate of reaction, which
was confirmed by a few preliminary runs conducted at different
stirring speeds. Note, generally speaking, that the use of magnetic
stirring was unable to completely take off its influence on the rate of
chemical reactions in liquid phases in the presence of solid catalysts
and/or gaseous reactants.
the hydrogenation with supported metal catalysts in CO -dissolved
2
expanded liquids, the deactivation of catalysts occurs in some cases
due to the formation and adsorption of CO from H + CO2 on the
2
surface of metal catalysts [18,19]. For example, in hydrogenation
of phenol with Rh catalysts supported on C and Al O , the Rh/C
2
3
does not lose its activity during the reaction but the Rh/Al O is
2
3
easily deactivated at an early stage of reaction [17]. This difference
is caused by weak and strong adsorption of CO (originating from
CO ) on the former and latter catalysts, respectively, as proved by
2
in situ FTIR measurements.
The selective hydrogenation of nitriles to primary amines is
one of industrially important chemical reactions [20–22]. The
primary amines can find various applications as dyes, pharmaceu-
ticals, explosives, agrochemicals, and others. The hydrogenation
of nitrile compounds was previously studied by several authors
using homogeneous [21] and heterogeneous [22] metal catalysts.
For homogeneous catalysts, the selectivity to the primary amine is
high >95% but the difficulty of catalyst separation after reactions
is a main problem. Therefore, the use of heterogeneous catalysts
is desirable but the low selectivity to primary amines due to the
formation of secondary amines should be improved. Application
of NH3 with heterogeneous catalysts can improve the selectivity
to primary amines [23], but the amount of NH3 required is large.
Another way is the addition of acids such as HCl and acetic acid over
Pd/C and RANEY Ni catalysts [24–26]. A biphasic medium including
2.2. In situ high-pressure FTIR measurement
Molecular interactions of pressurized CO with reacting species
2
are important for determining the rate of reaction and the prod-
uct selectivity [30]. In situ high-pressure FTIR was used to examine
the interactions between CO2 and BN molecules in the absence
and presence of water using a JASCO FTIR-620 spectrometer with a
−
1
acidic water (NaH PO ) and dichloromethane is effective for a high
triglycine sulfate detector at a frequency resolution of 2 cm . The
FTIR measurements were made with a home-designed high pres-
sure cell of 1.18 cm3 attached with a ZnSe rod in attenuated total
reflection (ATR) mode [31]. A certain volume of BN was introduced
into the cell and it was purged with CO2 several times. After heat-
ing to 323 K by resistivity heating with coils embedded in the cell,
compressed CO2 was introduced into the cell to the desired pres-
sure while stirring the mixture. The FTIR spectra were collected
for a mixture of CO2 dissolved in liquid BN at different pressures.
Similar measurements were also made for a mixture of bulk water
dissolving CO2 and BN.
2
4
selectivity of 95% to BA in hydrogenation of BN with supported Pd
catalysts [27]. From an environmental point of view, however, it is
desirable to avoid the use of those harmful base and acid additives.
Recently, Bakker et al. reported an extensive work on heteroge-
neous hydrogenation of BN [28]. They compare the performance of
various supported metal catalysts reported so far in the literature. It
is reported that a high selectivity of 95% to the primary amine can
be achieved by supported Pd catalysts at a H2 pressure of 1 MPa
[
28]. The use of dense phase CO2 in hydrogenation of nitriles was
recently investigated by Chatterjee et al. [29]. They show that the
selectivity to the primary and secondary amines can be tuned by
CO2 pressure and BA is produced with a Pd/MCM-41 catalyst in a
high selectivity of about 90% at a CO2 pressure of about 10 MPa.
Those previous results and interesting impacts of dense phase CO2
and water have set the present authors to work on a multiphase
medium including those two green components in the selective
hydrogenation of nitriles, using BA as a model substrate.
In situ high-pressure FTIR was also used to examine the for-
mation and adsorption of CO from compressed H2 and CO2 over
Pd/Al O3 using a 1.5 cm3 high pressure cell. The Pd/Al O3 sample
2
2
was ground in a mortar and the powder was pressurized at 50 MPa
for 1 h to prepare a sample pellet. The pelletized sample was placed
into the cell and it was purged with H2 three times. The cell was
heated to 343 K and then H was introduced to 4 MPa to reduce the
2
catalyst sample for 1 h. Then, the cell was purged with N , cooled
2
to 323 K, and the spectrum was collected. The cell was loaded with
2 MPa H2 followed by the introduction of CO2 to the desired pres-
sures. The FTIR spectra were measured at different CO2 pressures
in the absence of BN and water vapor. When BN co-existed, as men-
tioned later, a certain carbamate species was produced from CO2
and BA, the product of BN hydrogenation, which accumulated on
and covered the surface of catalyst. So, it was impossible to examine
the formation and adsorption of CO from H2 and CO2 in the pres-
ence of BN. The influence of BN and/or water on the adsorption of
CO was examined using CO instead of CO2 and H2 using the same
FTIR cell. The pelletized sample was treated in the same manner
2
. Experimental
2.1. Hydrogenation
A commercially available 5 wt% Pd/Al O3 (Wako) was used as a
2
catalyst for the hydrogenation of BN. Prior to the reaction, the cat-
alyst sample was reduced in a glass tube with a pure H (99.99%)
stream of 60 cm min at room temperature for 30 min and then
2
3
−1
−1
heated at 5 K min to 473 K, at which it was further reduced for
h. The reduced sample had an average Pd particle size of 3.6 nm
1

H. Yoshida et al. / Applied Catalysis A: General 456 (2013) 215–222
217
Scheme 1. Possible reaction pathways in the hydrogenation of benzonitrile (BN).
as described above and a certain amount of water and/or BN was
introduced into the cell with a syringe through a small hole where
N2 was flowing out. Then CO (1% in He) was introduced into the
cell at ambient pressure for 15 min, and the FTIR spectra were col-
lected. The FTIR spectra were further measured after purging with
N2 to remove water and/or BN vapor from the atmosphere.
conversion (in a longer reaction time) (entry 5). These results show
that the co-existence of pressurized CO2 and H O is indispensable
2
for the selective hydrogenation of BN to the primary amine of BA.
To understand the features of this multiphase reaction medium, the
influence of CO2 pressure was further examined.
3.2. Hydrogenation in the presence of CO2 and water
2.3. Phase behavior observation
Fig. 1a gives the changes of total conversion and product selec-
There are a few different phases in the present reaction medium
tivity with CO2 pressure in the absence of H O. The conversion
2
and so it is important to examine the phase behavior under high
pressure reaction conditions. A 10 cm3 sapphire-windowed view
cell was used for this purpose [32]. The observation was made with
a similar volumetric ratio of BN to the reactor as used in the hydro-
genation reactions. A certain amount of BN was added into the
cell and it was purged with H2 three times. The cell was heated
up to 323 K (reaction temperature) by circulation of preheated
water outside the cell, followed by the introduction of 2 MPa H2.
Then CO2 was slowly introduced into the cell to the desired pres-
sure while stirring the mixture by a magnetic stirrer. The phase
behavior was visually examined by naked eyes at different CO2
slightly decreased but the selectivity did not change with CO pres-
2
sure <10 MPa. At higher pressures, however, the conversion and
the BA selectivity significantly decreased while the DBA selectivity
increased. The phase behavior observations of the reaction mixture
indicated that the volume of liquid BN simply decreased due to its
dissolution into the CO2 phase with increasing pressure and dis-
appeared at about 10.6 MPa. The significant decreases in the total
conversion and the BA selectivity occurred in the homogeneous
CO2 phase. Hence, in the absence of H O, pressurized CO2 has no
2
positive impact on the selective hydrogenation of BN to BA under
both gas–liquid (BN)–solid (catalyst) and gas–solid conditions. The
pressures. The mixture was observed to change from three-phase
state (gas–liquid–solid) to two-phase state (gas–solid) at a certain
results obtained in the presence of H O are given in Fig. 1b, indi-
2
cating quite different impacts of pressurized CO . The presence of
2
pressure. In addition, the dispersion of granules of Pd/Al O cata-
2
3
1 MPa CO2 increased the total conversion and the BA selectivity
up to >95% but decreased the DBA selectivity. When CO2 pressure
was further raised up to 14 MPa, the conversion and the selectiv-
ity did not change so much. The phase behavior of the reaction
lyst in the present multiphase reaction medium was examined at
atmospheric pressure.
3
. Results and discussion
mixture was gas–liquid (BN)–liquid (H O)–solid (catalyst) at CO2
2
pressures < 11 MPa and gas–liquid (H O)–solid at higher pressures.
2
3.1. Hydrogenation in different media
Thus, the reaction medium including both pressurized CO and H O
2
2
is effective for the selective hydrogenation of BN to BA irrespective
of the phase behavior.
The hydrogenation of BN was examined with a Pd/Al O catalyst
2
3
in different multiphase reaction media at 323 K and a H2 pressure
of 2 MPa. Generally, several products may be formed, including BA,
DBA, and toluene, as depicted in Scheme 1 [20–22,27,28]; benzyli-
deneimine is an intermediate and its condensation with BA results
in the formation of DBA. The tertiary amine may also be formed
in a similar way from DBA and benzylideneimine. In the present
hydrogenation of BN, the products of BA, DBA, and toluene were
detected to form without any other ones. The product distribu-
tion was observed to depend on the reaction media and conditions
employed.
The influence of reaction time was then examined for BN hydro-
genation in the presence of CO2 and/or H O. Fig. 2a shows that the
2
Table 1
Results of hydrogenation of BN in the absence and presence of pressurized CO2
a
and/or H2O with Pd/Al2O3 catalyst.
Entry CO2 (MPa) H2O (cm3) Phases
present
Conv. (%) Selectivityc (%)
b
BA
DBA
TOL
Table 1 shows the results obtained in different reaction media
1
2
3
4
5
0
9
0
7
7
0
0
5
5
5
G-L1
G-L1
G-L1-L2
G-L1-L2
G-L1-L2
43.6
21.6
10.7
13.4
99.6
84.6
82.7
83.0
95.9
98.4
14.8
16.1
12.1
0.9
0.6
1.2
4.9
3.2
0.3
in the absence and presence of CO2 and/or H O. In neat BN (entry
2
1
), a conversion of 44% was obtained with a BA selectivity of 85%.
The addition of either CO2 (9 MPa) or H O decreased the total
2
1.3
conversion but did not change the selectivity so much (entries
a
Reaction conditions: BN 1.0 g (9.7 mmol), catalyst 20 mg, H2 2 MPa, Temperature
2
, 3). When both CO2 and H O were added, the conversion was
2
3
23 K, Reaction time 4 h (24 h for entry 5).
also lowered to 13% but, interestingly, the BA selectivity was
significantly enhanced to 96% (entry 4). This high BA selectivity
remained unchanged during the reaction up to an almost complete
b
Phases present in the reaction mixture except for the solid catalyst. G: H2 and/or
CO , L : BN, L : water.
2
1
2
c
BA: benzylamine, DBA: dibenzylamine, TOL: toluene.

2
18
H. Yoshida et al. / Applied Catalysis A: General 456 (2013) 215–222
Fig. 1. Influence of CO2 pressure on total conversion and product selectivity in hydrogenation of BN in the absence (a) and presence (b) of H2O using Pd/Al2O3 at 323 K and
at a H2 pressure of 2 MPa for 4 h.
3
Fig. 2. (a) Time profiles of total conversion in hydrogenation of BN in the absence (ꢀ, ♦, ꢀ) and presence (᭹) of H2O (5 cm ) and at CO2 pressures given. (b) The relationship
3
between BA selectivity and total conversion in the absence (♦) and presence (᭹, ᭹) of H2O (5 cm ) and at CO2 pressures given.
reaction went smoothly in the absence of CO and was almost com-
these surface modified Pd/Al O catalysts. The product distribution
2 3
2
pleted in < 15 h under the conditions used. In the presence of 1 MPa
was little affected by the surface treatments but the total conver-
sion was lowered and improved by the treatments with hot water
and 1-heptanol, respectively. The hydrophobic nature is better for
the catalyst to give a larger total conversion in this multiphase reac-
tion medium. It was indicated by X-ray photoelectron spectroscopy
that the surface properties of supported Pd particles were not dif-
ferent between the hot water and 1-heptanol treated samples (Fig.
S3 in SI).
CO , the rate of reaction became lowered and leveled off with time
2
and the reaction stopped in 12 h. Similar but larger negative effects
of CO were observed at 14 MPa. The catalyst (Pd/Al O ) was deac-
2
2
3
tivated by pressurized CO2 in the absence of H O. The reason for
2
this catalyst deactivation will be discussed in detail later. When
H O co-existed, in contrast, the reaction went smoothly up to 100%
2
BN conversion even in the presence of pressurized CO ; that is, no
2
catalyst deactivation occurred in this multiphase reaction medium.
The selectivity to BA is plotted against total BN conversion in Fig. 2b.
It is known that the pH of H O phase decreases on CO2 pressur-
ization [33] and this pH change could affect the rate of reaction and
the product distribution in the present BN hydrogenation. When
2
The addition of H O alone is effective for enhancing the selectiv-
2
ity but more significant enhancement is caused by the presence of
H O phase is pressurized by CO2 at pressures of 1–8 MPa, the pH
2
both CO2 and H O, giving a BA selectivity >95% during the whole
value decreases to about 3 according to the literature [34]. So,
2
course of reaction up to 100% BN conversion.
an acetic acid solution of pH 3 was used instead of pure H O for
2
The state of the multiphase reaction mixture was visually exam-
ined. The catalyst granules were observed to exist at the interface
between the organic (BN) and aqueous phases (Fig. S1 in SI). The
authors attempted to modify the surface properties of the catalyst
and examine its activity after the surface modification. The cata-
lyst was treated by immersing in either hot water or 1-heptanol at
the reaction. Table 3 indicates that acidic medium may be slightly
Table 2
Influence of surface modification with either hot water or 1-heptanol on the activity
a
of Pd/Al2O3 catalyst in BN hydrogenation
.
Entry
Modification
Conv. (%)
Selectivityb (%)
3
53 K for 6 h and dried under ambient conditions. The hot water
and 1-heptanol treatments were expected to make the catalyst
more hydrophilic and more hydrophobic, respectively. This was
confirmed by FTIR measurements that the broad absorption band
due to surface hydroxyl groups at 3700–3000 cm became larger
by the hot water treatment, while the new absorption bands due to
the surface C7H13O- group appeared by the 1-heptanol treatment
BA
DBA
TOL
1
2
3
–
H2O
1-Heptanol
13.4
11.5
20.4
95.9
94.5
95.3
0.9
0.9
3.1
3.2
4.6
1.6
−
1
a
Reaction conditions: BN 1.0 g (9.7 mmol), catalyst 20 mg, H2 2 MPa, Temperature
323 K, Reaction time 4 h
b
(
Fig. S2 in SI). Table 2 shows the results of BN hydrogenation with
BA: benzylamine, DBA: dibenzylamine, TOL: toluene

H. Yoshida et al. / Applied Catalysis A: General 456 (2013) 215–222
219
Fig. 3. In situ high-pressure FTIR–ATR spectra collected at 323 K for liquid BN (a) and for H2O dissolving BN (b) compressed by CO2 at pressures given.
better for the total conversion and the BA selectivity (entry 2), as
blue-shifted with increasing CO2 pressure and the extent of the
compared to the reaction with pure H O alone (entry 1). Although
blue-shift was the same irrespective of the absence and presence
2
the conversion levels were very similar between the reaction with
of H O. So, the interactions between BN and CO2 molecules are
2
pressurized CO2 and H O (entry 3) and that in the acetic acid solu-
unlikely to change by the presence of H O. In other words, a BN
2
2
tion, the BA selectivity was larger in the former medium than in the
latter.
molecule should be preferentially surrounded by CO2 molecules.
Note again, thus, that the reactivity of C-N bond of BN is lowered
by the presence of CO2 and this does not depend on the presence
of H O.
3.3. In situ high-pressure FTIR measurements
2
3.3.1. Molecular interactions of CO2 with BN
3.3.2. Formation and adsorption of CO
Carbon dioxide molecules are dissolved in organic liquids at high
pressures and may interact with chemical species in these phases
9,30,31,35]. For example, CO2 molecules interact with carbonyl
group of cinnamaldehyde [35] and nitro group of nitrobenzene [9],
as evidenced by in situ high-pressure FTIR measurements for the
There is a possibility of the formation and adsorption of CO
through reverse water gas shift reaction (CO2 + H2 → CO + H2O)
or other pathways when hydrogenation reactions are conducted
with supported metal catalysts in the presence of pressurized
H2 and CO2. A few authors investigated this phenomenon using
in situ high-pressure FTIR techniques and stated that the forma-
tion of CO was responsible for the catalyst deactivation observed
in several hydrogenation reactions [18,19]. The present authors
also observed the formation and adsorption of CO on pelletized
Pd/Al2O3 catalyst in the presence of H2 and CO2 at 323 K using
in situ high-pressure FTIR technique (Fig. S4 in SI). The adsorbed
CO species still remained under ambient conditions. To examine
the competitive adsorption of CO and BN, CO was adsorbed on
the catalyst in the presence of H2O and/or BN vapor. Fig. 4 shows
that the absorption bands assignable to bridged CO species are
[
organic compounds dissolved in pressurized CO . The interactions
2
with CO increase the reactivity of the carbonyl group and decrease
2
that of the nitro group, which results in the improved selectivity to
cinnamyl alcohol and to aniline in hydrogenation of cinnamalde-
hyde and nitrobenzene, respectively, in the presence of compressed
CO . Similar FTIR measurements were made for liquid BN dissolv-
2
ing CO2 at high pressures by ATR mode [31]. Fig. 3a presents the
−
1
FTIR-ATR spectra obtained in the range of 2260–2200 cm at dif-
ferent CO pressures. The absorption band appeared at around
2
1
−
2
228 cm , assignable to the stretching vibration of C-N bond of
BN. The peak position was blue-shifted with increasing CO2 pres-
sure, indicating that the CN bond became stronger and pressurized
CO2 should decrease its reactivity. Similar FTIR spectra were col-
lected for BN dissolved in H O compressed by CO in the same
2
2
ATR mode. Fig. 3b shows the spectra collected, indicating that the
absorption band of C-N bond became weaker with increasing pres-
sure due to the transfer of BN from H O to CO phase. It was seen,
2
2
however, that the peak position of ␯(CN) absorption band was
Table 3
Results of BN hydrogenation in different aqueous media in the absence and presence
of pressurized CO2 and acetic acid.a
Entry
CO2 (MPa)
Aqueous media
Conv. (%)
Selectivityb (%)
BA
DBA
TOL
1
2
3
0
0
7
H2O
10.7
13.2
13.4
83.0
89.7
95.9
12.1
5.9
0.9
4.9
4.4
3.2
Acetic acidc
H2O
a
Reaction conditions: BN 1.0 g (9.7 mmol), catalyst 20 mg, Aqueous media 5 cm3,
H2 2 MPa, Temperature 323 K, Reaction time 4 h
Fig. 4. In situ high-pressure FTIR spectra of CO adsorbed on Pd/Al2O3 after reduction
(a), after exposure to atmospheric CO for 10 min (b), and after purging with N2 (c)
in the absence (1) and presence of H2O (2), BN (3), and H2O + BN (4).
b
BA: benzylamine, DBA: dibenzylamine, TOL: toluene
c
0
.1 mol% of acetic acid solution

2
20
H. Yoshida et al. / Applied Catalysis A: General 456 (2013) 215–222
Fig. 5. Illustration of features of BN hydrogenation reactions in four different multiphase media in the absence and presence of CO2 and/or H2O. The solid catalyst phase is
not displayed herein.
located at about 1970 and 1910 cm−1 [36] and a weak absorp-
3.4. Features of the reaction medium including both CO2 and
water
−
1
tion band of linear CO species at 2080 cm [37,38] in the absence
of both H O and BN. When H O co-exists, the absorption band
2
2
of linear CO disappears while those of bridged CO still remain
From those results obtained, the authors will discuss the pos-
sible roles of pressurized CO2 and H2O in hydrogenation of BN
with respect to the total conversion, the BA selectivity, and the
catalyst life in the following. The multiphase reaction media stud-
ied are categorized as illustrated in Fig. 5. The media (b) and (d)
were examined at a H2 pressure of 2 MPa, similar to those (a)
and (c), and at different CO2 pressures. The media of (a) and (c)
initially included two (gas–organic liquid) and three (gas–organic
liquid–water) phases, respectively, except for the solid catalyst. The
phases existing in the initial mixtures of (b) and (d) were differ-
ent depending on CO2 pressure used. The organic liquid phase of
BN decreased in its volume with increasing CO2 pressure and dis-
appeared at 10.6 MPa, as illustrated in SI (Figs. S4, S5), while the
volume of water did not change so much at CO2 pressures exam-
ined. When the reaction proceeds, BN transforms mainly into BA, a
water-soluble compound. The solubility of H2O into the gas phase
(H2 and/or CO2) is negligible under the conditions used. The phase
behavior changes during the reaction in different manners: for the
systems of (a) and (b) of Fig. 5 including no H2O, no change but the
[
39]. In the presence of BN, in contrast, the absorption bands of
−
1
the bridged and linear CO species appear at 1920 and 2010 cm
,
respectively. The spectrum obtained in the presence of both H O
2
and BN is similar to that with BN alone. The adsorbed CO species
can exist in the presence of H O and some CO species may be
2
replaced by BN molecules. In addition to those FTIR measure-
ments, a reaction run was conducted with a catalyst sample on
which CO was pre-adsorbed by exposure to 1% CO gas at 0.1 MPa
for 10 min. The reaction was observed to occur with this catalyst.
The pre-adsorption of CO on the catalyst does not cause its deac-
tivation because the adsorbed CO species may easily be replaced
with the substrate BN. Those results imply that CO is also formed
and adsorbed on the surface of Pd/Al O catalyst in the pres-
2
3
ence of pressurized H2 and CO . However, this is not responsible
2
for the catalyst deactivation observed in the presence of pressur-
ized CO2 alone (without H O). We should consider other possible
2
reasons for the catalyst deactivation, which will be discussed
later.

H. Yoshida et al. / Applied Catalysis A: General 456 (2013) 215–222
221
Scheme 2. Formation of a carbamate salt from BA and CO2.
component of the liquid phase changes from BN to BA; for the sys-
[29]. The total conversion and the BA selectivity were reported to
increase with CO2 pressure up to 10 MPa, which are different from
tems of (c) and (d) including H O, the liquid phase (BN) decreases
2
and disappears due to the dissolution of the product of BA into
the CO pressure effects observed in the present work (Fig. 1a). With
2
H O.
2
the Pd/MCM-41 catalyst, the BA selectivity was maximal (about
9
0%) at a conversion of 90% at a CO2 pressure of around 10 MPa, at
3
.4.1. Medium (a)
In the absence of CO2 and H O, the hydrogenation of BN goes
which the reaction mixture was in a single phase (except for the
solid catalyst). It is noted that the formation of carbamate from BA
2
smoothly to 100% conversion; however, the BA selectivity is about
0% at an initial stage of reaction and tends to decrease with con-
and CO , which may depend on the phase behavior, is an important
factor for the high selectivity to BA.
2
8
version to about 60% at a conversion of 60% (Fig. 2). No catalyst
deactivation occurs in this reaction medium. The selective hydro-
genation of BN to BA can be achieved with an alumina-supported Pd
catalyst under ordinary conditions; Bakket et al. reported a high BA
selectivity of about 95% in a solvent of 2-propanol at a H2 pressure
of 1 MPa and at 353 K [28].
3
.4.3. Medium (c)
When H O is added to medium (a), the BA selectivity is enhanced
2
(Fig. 2). This may be explained by the nature of BA, which is sol-
uble in H O phase and so it becomes unlikely to react with the
2
imine intermediate producing the secondary amine in the organic
BN phase. The initial BA selectivity is about 90% but, unfortunately,
the BA selectivity decreases with conversion, similar to the case
of medium (a). Some quantity of BA is likely to be soluble in the
organic BN phase and react with the imine intermediate. A larger
amount of BA is formed at a larger conversion and so the change
of BA to the secondary amine should be more significant in a later
stage of reaction. The total conversion is lowered by the addition
3.4.2. Medium (b)
When 1 MPa CO2 is added to medium (a), the conversion is
slightly lowered. It does not change with increasing CO2 pressure
up to about 10 MPa but decreases at higher pressures (Fig. 1a).
The large decrease in the conversion at pressures >10 MPa may
be ascribed to the dilution effect; the reacting species, except for
the solid catalyst, are completely soluble in the CO2 gas phase
at 10.6 MPa and become more diluted with increasing pressure.
A few possible factors would be related to the influence of pres-
surized CO2 at pressures <10 MPa, in which the reaction mixture
of H O (Table 1). The catalyst granules may be dispersed in organic
BN phase in medium (a) but these exist at the H O–BN interface
2
2
in medium (c). The contact between the catalyst and BN phase in
the latter should be smaller than that in the former, which would
be a reason for the smaller conversion in medium (c). The impor-
tance of the catalyst–BN contact is supported by the differences in
the activity among untreated and surface-treated, hydrophilic and
includes gas (CO , H ), liquid (BN), and solid (catalyst). The molec-
2
2
ular interactions with CO decrease the reactivity of C-N bond of the
2
substrate, decreasing the rate of BN hydrogenation. The conversion
levels are almost the same at different CO2 pressures in the range
of 1–10 MPa. So, the decreased reactivity of BN could be offset by an
hydrophobic Pd/Al O3 catalysts in the hydrogenation of BN in the
presence of H O (Table 2)
2
2
increase in the concentration of H , which is assisted by the dissolu-
2
tion of CO . In medium (b), the rate of hydrogenation stopped after
2
a certain period of time; namely, the catalyst deactivation occurs
3.4.4. Medium (d)
(
Fig. 2). The in situ FTIR measurements show that CO is formed from
The BA selectivity is more significantly enhanced by the pres-
H2 and CO2 at high pressures and adsorbed on the catalyst (Fig. S6
in SI) but this is not responsible for the catalyst deactivation. When
CO2 was introduced to liquid BA, we shortly observed the forma-
tion of a white solid material. It should be a carbamate salt [10,29],
as depicted in Scheme 2. It is likely to accumulate on and cover
the surface of catalyst during the reaction, resulting in the catalyst
deactivation.
ence of both CO2 and H O. A high BA selectivity of >95% can be
obtained at a low CO2 pressure of 1 MPa (Fig. 1b), which is even
lower compared to 10 MPa needed for the high BA selectivity with
2
Pd/MCM-41 in the presence of CO alone [29]. Fig. 2b indicates that
2
this high BA selectivity can remain unchanged during the whole
course of reaction, being different from the case of medium (c). The
BA molecules produced are likely to react with CO2 and move as
the carbamate salt from the organic phase to the aqueous phase.
This may reduce the chance for BA to react with the intermediate
of benzylideneimine yielding the secondary amine. The carbamate
salt is insoluble in the BN organic phase, resulting in that the high
BA selectivity can remain unchanged even at larger conversion lev-
els. A similar effect was previously reported by Xie et al., who
Fig. 1a also shows that the BA selectivity decreases with CO2
pressure in a similar way as does the total conversion. The homo-
geneous CO gas phase (except for the solid catalyst) facilitates the
2
formation of the secondary amine DBA, which is produced via con-
densation between the primary amine of BA and the intermediate
of benzylideneimine (Scheme 1). The imine intermediate might be
concentrated on the catalyst in the CO gas phase at high pressures
investigated the hydrogenation of BN using NiCl /NaBH4 in CO2-
2
2
due to dilution of H2 and/or interactions with CO2 molecules. As
mentioned above, when CO2 was added to liquid BA, a white solid
material was shortly produced; so, unfortunately, the authors failed
to examine the interactions between the BA and CO2 molecules by
the in situ high-pressure FTIR method.
The hydrogenation of BN to BA in the presence of pressurized
CO2 was recently studied by Chatterjee et al. using Pd catalysts on
MCM-41 and C supports at a H2 pressure of 2 MPa and at 323 K
expanded ethanol at 303 K [10]. They indicated the formation of
carbamate salt in the ethanol phase and observed an enhanced BA
selectivity of 98%. The formation of carbamate salt from BA and
CO2 was also reported by Chatterjee et al., who noted its impor-
tance for the high BA selectivity [29]. The carbamate salt is unstable
and so the corresponding primary amine can easily be obtained
by heating [40]. Compared to medium (a), this medium (d) gives
a smaller conversion (Table 1). This may also be explained by a

2
22
H. Yoshida et al. / Applied Catalysis A: General 456 (2013) 215–222
difference in the catalyst–BN contact between the media (a) and
d).
The multiphase reaction medium (d) including both CO2 and
H O is significant for not only the selective production of BA but
References
(
[
[
1] W. Leitner, Acc. Chem. Res. 35 (2002) 746–756.
2] C.M. Rayner, Org. Process Res. Dev. 11 (2007) 121–132.
2
[3] T. Maegawa, A. Akashi, K. Yaguchi, Y. Iwasaki, M. Shigetsura, Y. Monguchi, H.
Sajiki, Chem. Eur. J. 15 (2009) 6953–6963.
also no catalyst deactivation in hydrogenation of BN with Pd/Al O .
2
3
[
4] B. Subramaniam, C.J. Lyon, V. Arunajatesan, Appl. Catal. B: Environ. 37 (2002)
79–292.
5] P.G. Jessop, B. Subramaniam, Chem. Rev. 107 (2007) 2666–2694.
Carbon monoxide is formed even in the presence of H O but does
2
2
not cause the catalyst deactivation. As mentioned above, the carba-
mate salt is formed from BA and CO and it is soluble in H O. So, the
[
2
2
[6] A. Kruse, H. Vogel, Chem. Eng. Technol. 31 (2008) 23–32.
[7] G.R. Akien, M. Poliakoff, Green Chem. 11 (2009) 1083–1100.
[8] M. Arai, S. Fujita, M. Shirai, J. Supercrit. Fluids 47 (2009) 351–356.
[9] X. Meng, H. Cheng, Y. Akiyama, Y. Hao, W. Qiao, Y. Yu, F. Zhao, S. Fujita, M. Arai,
J. Catal. 264 (2009) 1–10.
salt does not cover the surface of Pd/Al O catalyst and it cannot
cause the deactivation during the reaction.
2
3
[
[
10] X. Xie, C.L. Liotta, C.A. Eckert, Ind. Eng. Chem. Res. 43 (2004) 7907–7911.
11] R.D. Weinstein, A.R. Renslo, R.L. Danheiser, J.W. Tester, J. Phys. Chem. B 103
4
. Conclusion
(
1999) 2878–2887.
The multiphase reaction medium including both pressurized
[12] S. Fujita, T. Tanaka, Y. Akiyama, K. Asai, J. Hao, F. Zhao, M. Arai, Adv. Synth. Catal.
350 (2008) 1615–1625.
CO2 and H O is promising for the selective hydrogenation of BN to
2
[
[
[
13] J.W. Ford, J. Lu, C.L. Liotta, C.A. Eckert, Ind. Eng. Chem. Res. 47 (2008)
32–637.
the primary amine BA with a Pd/Al O3 catalyst. The BA selectivity
2
6
is >95% at any conversion level and the catalyst is not deactivated
during the reaction even though CO is formed from H2 and CO2.
The high BA selectivity results from a synergistic effect of CO2 and
14] X. Meng, H. Cheng, S. Fujita, Y. Yu, F. Zhao, M. Arai, Green Chem. 13 (2011)
570–572.
15] N. Hiyoshi, O. Sato, A. Yamaguchi, M. Shirai, Chem. Commun. 47 (2011)
1
1546–11548.
H O. The hydrogenation product of BA reacts with CO2 giving a
2
[16] Y. Akiyama, S. Fujita, H. Senboku, C.M. Rayner, S.A. Brough, M. Arai, J. Supercrit.
Fluids 46 (2008) 197–205.
carbamate salt and this water-soluble species moves into the aque-
ous phase, preventing the BA from reacting with the intermediate
of benzylideneimine yielding the secondary amine. The carbamate
species accumulates on the surface of catalyst and causes its deac-
tivation in the absence of H O. When H O co-exists, however, the
[
17] S. Fujita, T. Yamada, Y. Akiyama, H. Cheng, F. Zhao, M. Arai, J. Supercrit. Fluids
4 (2010) 190–201.
18] D. Ferri, T. Bürgi, A. Baiker, Phys. Chem. Chem. Phys. 4 (2002) 2667–2672.
5
[
[19] M. Burgener, D. Ferri, J.-D. Grunwaldt, T. Mallat, A. Baiker, Phys. Chem. B 109
2005) 16794–16800.
(
2
2
[
[
20] C. de Bellefon, P. Fouilloux, Catal. Rev. Sci. Eng. 36 (1994) 459–506.
21] S. Das, S. Zhou, D. Addis, S. Enthaler, K. Junge, M. Beller, Top. Catal. 53 (2010)
979–984.
22] S. Gomez, J.A. Peters, T. Maschmeyer, Adv. Synth. Catal. 344 (2002) 1037–1057.
23] C.F. Winnas, J. Am. Chem. Soc. 61 (1939) 3566–3567.
24] W.H. Carothers, G.A. Jones, J. Am. Chem. Soc. 47 (1925) 3051–3057.
carbamate salt is soluble in the aqueous phase and the catalyst is
unlikely to be covered and deactivated by this species. Carbon diox-
ide interacts with the C-N group of the substrate and decreases its
reactivity. When pressurized CO2 is introduced into the reaction
[
[
[
mixture including H O, however, the total conversion is enhanced.
[25] W.H. Hartung, J. Am. Chem. Soc. 50 (1928) 3370–3374.
[26] E. Miller, J.M. Sprague, L.W. Kissinger, L.F. McBurney, J. Am. Chem. Soc. 62 (1940)
2
The CO2 is soluble in the organic and aqueous phases and may
change their solvent properties, which might influence the state of
dispersion of catalyst granules at their interface, the rate of hydro-
2
099–2103.
[
[
27] L. Heged u˝ s, T. Máthé, Appl. Catal. A: Gen. 296 (2005) 209–215.
28] J.J.W. Bakker, A.G. van der Neut, M.T. Kreutzer, J.A. Moulijn, F. Kapteijn, J. Catal.
274 (2010) 176–191.
genation of BN, the concentration of H in the organic liquid phase,
2
[
[
[
[
[
[
[
[
29] M. Chatterjee, H. Kawanami, M. Sato, T. Ishizaka, T. Yokoyama, T. Suzuki, Green
and other factors.
Chem. 12 (2010) 87–93.
30] H. Yoshida, K. Kato, J. Wang, X. Meng, S. Narisawa, S. Fujita, Z. Wu, F. Zhao, M.
Arai, J. Phys. Chem. C 115 (2011) 2257–2267.
Acknowledgements
31] S. Fujita, H. Yoshida, K. Asai, X. Meng, M. Arai, J. Supercrit. Fluids 60 (2011)
1
06–112.
This work was supported by Japan Society for the Promotion
of Science with Grant-in-Aid for Scientific Research (B) 22360327
and for JSPS Fellows 2200038601 and by the Ministry of Education,
Culture, Sports, Science and Technology, Japan, with the Global COE
Program (Project No. B01: Catalysis as the Basis for Innovation in
Materials Science).
32] S. Fujita, S. Akihara, F. Zhao, R. Liu, M. Hasegawa, M. Arai, J. Catal. 236 (2005)
101–111.
33] C. Roosen, M. Schumacher, T. Mang, W. Leitner, L. Greiner, Green Chem. 9 (2007)
4
55–458.
34] J.J. Carroll, J.D. Slupsky, A.E. Mather, J. Phys. Chem. Ref. Data 20 (1991)
1201–1209.
35] F. Zhao, S. Fujita, J. Sun, Y. Ikushima, M. Arai, Chem. Commun. (2004)
2
326–2327.
36] A. Palazov, G. Kandinov, C. Bonev, D. Shopov, J. Catal. 74 (1982) 44–54.
Appendix A. Supplementary data
[37] A. Erdöhelyi, M. Pásztor, F. Solymosi, J. Catal. 98 (1986) 166–177.
[
[
38] A. Wille, P. Nickut, K. Al-Shamery, J. Mol. Struct. 695-696 (2004) 345–352.
39] H. Yoshida, S. Narisawa, S. Fujita, R. Liu, M. Arai, Phys. Chem. Chem. Phys. 14
Supplementary data associated with this article can be
found, in the online version, at http://dx.doi.org/10.1016/j.
apcata.2013.03.004.
(
2012) 4724–4733.
[40] M. George, R.G. Weiss, Langmuir 18 (2002) 7124–7135.