Selective hydrogenation of chloronitrobenzene to chloroaniline in supercritical carbon dioxide over Ni/TiO2: Significance of molecular interactions

Article abstract of DOI:10.1016/j.jcat.2009.10.024

The hydrogenation of chloronitrobenzene to chloroaniline was investigated over Ni/TiO₂ at 35 °C in supercritical CO₂ (scCO₂), ethanol, and n-hexane. The reaction rate followed the order of scCO₂ > n-hexane > eth

Full text of DOI:10.1016/j.jcat.2009.10.024

Journal of Catalysis 269 (2010) 131–139
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Selective hydrogenation of chloronitrobenzene to chloroaniline in supercritical
carbon dioxide over Ni/TiO2: Significance of molecular interactions
Xiangchun Meng a,b,c, Haiyang Cheng b, Shin-ichiro Fujita a, Yufen Hao b, Yanjiao Shang b, Yancun Yu b,
a,
Shuxia Cai b, Fengyu Zhao b, , Masahiko Arai
*
*
a Division of Chemical Process Engineering, Graduate School of Engineering, Hokkaido University, Sapporo 060-8628, Japan
b State Key Laboratory of Electroanalytical Chemistry, Changchun Institute of Applied Chemistry, Chinese Academy of Sciences, Changchun 130022, China
c School of Chemical Engineering, Changchun University of Technology, Changchun 130012, China
a r t i c l e i n f o
a b s t r a c t
Article history:
The hydrogenation of chloronitrobenzene to chloroaniline was investigated over Ni/TiO2 at 35 °C in
supercritical CO2 (scCO2), ethanol, and n-hexane. The reaction rate followed the order of scCO2 > n-hex-
ane > ethanol. In scCO2, the selectivity to chloroaniline and to aniline over Ni/TiO2 were 97–99.5% and
<1%, respectively, in the conversion range of 9–100%. The high chemoselectivity to chloroaniline cannot
be achieved over Ni/TiO2 in ethanol and n-hexane. In situ high-pressure Fourier transform infrared mea-
surements were made to study the molecular interactions of CO2 with the following reactant and reaction
intermediates: chloronitrobenzene, chloronitrosobenzene, and N-chlorophenylhydroxylamine. The
molecular interaction modifies the reactivity of each species and accordingly the reaction rate and the
selectivity. The influence of Cl substituent on the interaction modes of CO2 with these reacting species
is discussed. Possible reaction pathways for the hydrogenation of chloronitrobenzene in scCO2 over Ni/
TiO2 are also proposed.
Received 25 August 2009
Revised 24 October 2009
Accepted 27 October 2009
Available online 3 December 2009
Keywords:
Chloronitrobenzene
Chloroaniline
Hydrogenation
Carbon dioxide
Intermolecular interaction
Ó 2009 Elsevier Inc. All rights reserved.
1. Introduction
diates formed [6]. The addition of small amounts of vanadium pro-
moters could reduce the accumulation of arylhydroxylamine
The catalytic hydrogenation of chloronitrobenzene (CNB) is a
common method to manufacture chloroaniline (CAN) isomers
(o-, m-, and p-CAN), which are important intermediates in the syn-
thesis of dyes, pharmaceuticals, and agricultural chemicals [1]. The
hydrogenation of CNB, depending on the catalyst used, yields CAN
through several possible pathways along with some reaction
intermediates such as chloronitrosobenzene (CNSB), N-chloro-
phenylhydroxylamine (CPHA), dichloroazoxybenzene (CAOB),
dichloroazobenzene (CAB), and dichlorohydrazobenzene (CHAB)
(Scheme 1) [2–5]. The accumulation of these intermediates, espe-
cially CPHA, is critical when the hydrogenation of CNB is carried
out in batch reactors at low or medium temperatures [3]. Because
these compounds are highly toxic and problematic, their accumu-
lation during the reaction should be avoided for the safe and green
production of CAN [3–5]. To avoid the accumulation of CPHA, it is
of advantage to carry out the hydrogenation at temperatures
>80 °C over well-designed catalysts [4]. Chemoselective hydroge-
nation of aromatic nitro compounds was achieved over Au/TiO2
and Au/Fe2O3 at 100–130 °C, with only trace amounts of interme-
during the hydrogenation (120 °C) of aromatic nitro compounds
over commercial Pd and Pt catalysts [3]. For Raney Ni catalysts
used for the hydrogenation (30–35 °C) in organic solvents, how-
ever, the selection of effective promoters is difficult because mod-
ifiers that reduce the accumulation usually lower the reaction rate
[7]. Mahata et al. communicated that when the hydrogenation of
CNB was performed at 120 °C over filamentous carbon-stabilized
Ni catalyst in methanol, the specific activity was ꢀ50 mmol hꢁ1 gꢁ1
Ni (2.9 mmol hꢁ1 mmolꢁ1 Ni) and there was no intermediate
detected at any level of CNB conversion [8]. Their catalyst was pre-
pared by methane decomposition over Raney Ni, which is expen-
sive for industrial catalyst preparation. Keane and co-workers
reported that selective production of CAN could be achieved in
the gas-phase hydrogenation of CNB over Au/Al2O3 (120–250 °C),
Pd-promoted Au/Al2O3 (Au/Pd mol/mol P 20) (120 °C), and sup-
ported Ni catalysts (120 °C) [5,9].
Dechlorination of CAN to aniline (AN) or of CNB to nitrobenzene
(steps VIII, IX in Scheme 1) is another side reaction that decreases
the selectivity to CAN and results in the formation of hydrogen
chloride. Addition of inhibitors can minimize dechlorination, but
usually lowers the reaction rate. It has recently been reported that
the dechlorination process can almost completely be suppressed
over several novel catalysts such as Ag/SiO2 [10], Au/SiO2 [11],
* Corresponding authors. Fax: +86 431 8526 2410 (F. Zhao), +81 11 706 6594
(M. Arai).
E-mail addresses: zhaofy@ciac.jl.cn (F. Zhao), marai@eng.hokudai.ac.jp (M. Arai).
0021-9517/$ - see front matter Ó 2009 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2009.10.024

132
X. Meng et al. / Journal of Catalysis 269 (2010) 131–139
higher than that in ethanol (84%), but it is difficult to suppress
NO2
NO2
NB
NH2
AN
IX
the dechlorination reaction completely [35]. In our attempts to
use supported Au catalysts such as Au/TiO2 and Au/SiO2 in scCO2
for the CNB hydrogenation, the activity of Au catalysts was lowered
monotonously with the addition of CO2. The reduction of Au activ-
ity in dense-phase CO2 was possibly due to the interaction of CO2
Cl
CNB
IV
I
II
O
NO
Cl
Cl
Cl
N
Cl
Cl
CAOB
CNSB
with Au [36]. As to Pt/c-Fe2O3 and Au/Fe(OH)x, the acid formed
VI
N
N
in situ from CO2 and a hydrogenation byproduct of H2O are likely
to impose negative effects on the support Fe2O3 or Fe(OH)x.
Thus, it remains to be a challenging task to achieve a high che-
moselectivity to CAN for the hydrogenation of CNB in green sol-
vents over non-noble metal catalysts. Recently we have shown
that the selective hydrogenation of nitrobenzene can be achieved
Cl
NHOH
N
CAB
CPHA
NH2
CAN
NH2
AN
V
VII
III
Cl
H
N
Cl
CHAB
over Ni/c-Al2O3 in dense-phase CO2 with 99% selectivity to AN over
N
H
the whole conversion range of 0–100%, and molecular interactions
of CO2 with the reaction species are significant for the improve-
ment in the selectivity [37]. The hydrogenation of CNB has a stron-
ger tendency to give arylhydroxylamine in large amounts during
the reaction [3]. Therefore, it is interesting and significant to see
whether the dechlorination reaction and the accumulation of inter-
mediates can be suppressed simultaneously during the hydrogena-
tion of CNB with the Ni–scCO2 system. The molecular interaction of
CO2 with CNB, CNSB, and CPHA is also a question of common inter-
est because there is an electron-withdrawing substituent, –Cl, at-
tached to the benzene ring of CNB, CNSB, and CPHA. In this work,
we have studied the hydrogenation of o-, m-, and p-CNB over Ni/
Cl
VIII
Scheme 1. Possible reaction pathways for the hydrogenation of CNB to CAN [2–5].
CNB: chloronitrobenzene, CNSB: chloronitrosobenzene, CPHA: N-chloro-
phenylhydroxylamine, CAN: chloroaniline, CAOB: dichloroazoxybenzene, CAB:
dichloroazobenzene, CHAB: dichlorohydrazobenzene, AN: aniline.
Au/ZrO2 [12], AuPd/Al2O3 [9], Au/Fe(OH)x [13], Pt/
Pt/ -ZrP [16], decorated Pt/TiO2 [17,18], Ru/SnO2 [1], and Ni/TiO2
c-Fe2O3 [14,15],
c
[19]. The hydrogenation was usually carried out at 90–150 °C with
Au and Ag [9–13], at 75–120 °C with Ni [8,19], and at 30–60 °C over
Pt catalysts [14–18]. With these catalysts, the selectivity to the
undesired AN was <1% at complete conversion of CNB. However,
the use of these noble metal catalysts (Au, Ag, Pd, Ru, and Pt) in
large-scale production should be practiced after a prudential con-
sideration of the cost. Additionally, it seems necessary to improve
the activity of Au and Ag catalysts for practical applications [15].
c-Al2O3 and Ni/TiO2 in scCO2 at 35 °C. For comparison, the hydro-
genation was also examined in ethanol and n-hexane under similar
reaction conditions. In situ high-pressure Fourier transform infra-
red (FTIR) measurements were made to study the molecular inter-
actions of dense-phase CO2 with CNB, CNSB, and CPHA molecules.
2. Experimental
With Pt/Fe2O3, Pt/c-ZrP, and Ni/TiO2, harmful reaction intermedi-
ates such as CNSB, CPHA, CAOB, and CAB were formed in organic
2.1. Catalyst preparation and characterization
solvents during the reaction [14–16,19].
Ni/TiO2 was prepared by incipient wetness impregnation using
anatase TiO2 (specific surface area 120 m2 gꢁ1, Nanjing High Tech-
nology Nano Material Co., Ltd., China) and Ni(NO3)2ꢂ6H2O. Ni/Al2O3
was prepared by co-precipitation using Ni(NO3)2ꢂ6H2O and
Al(NO3)3 with a Ni/Al atomic ratio of 1/1 [37]. After drying at
120 °C, the two catalysts were calcined at 450 °C for 5 h. Before
the hydrogenation run, the calcined samples were reduced under
H2 flow at 450 and 610 °C for Ni/TiO2 and Ni/Al2O3, respectively.
The Ni loading in the calcined samples of Ni/Al2O3 and Ni/TiO2
was measured by inductively coupled plasma optical emission
spectrometry. The structural properties of Ni catalysts were exam-
ined by X-ray diffraction (XRD, Philips PW1710 BASED) and trans-
mission electron microscopy (TEM, JEM-2000EX).
The melting point of o-, m-, and p-CNB is 32.1, 44.4, and 82 °C,
respectively. To dissolve the solid CNB substrates (at room temper-
ature), organic solvents such as methanol, ethanol, 1-butanol, hex-
ane, cyclohexane, ethyl acetate, diethyl ether, and tetrahydrofuran
are frequently used in both the liquid and the conventional gas-
phase hydrogenation [1–19]. These organic solvents are volatile,
toxic, and associated with serious health hazards and environmen-
tal problems. The use of traditional organic solvents is considered a
major source of waste in the chemical industry. Thus, removal of
the toxic solvents is significant to achieve the green production
of CAN. Ionic liquid was used as a medium for the hydrogenation
of o-CNB over ionic-liquid-like copolymer-stabilized Pt nanocata-
lysts, and the selectivity to CAN was >96% [20]. Because toxic or-
ganic solvents are generally used in the synthesis of ionic liquids,
supercritical carbon dioxide (scCO2) might be a greener alternative
to conventional organic solvents. ScCO2 has several merits such as
non-flammability, relative inertness, complete miscibility with
gases, and easy separation from liquid/solid products after reac-
tions [21–27]. Furthermore, scCO2 can show interesting effects
on reaction rate and product selectivity [28–32]. Ichikawa et al. re-
ported that the rate of hydrogenation of CNB over Pt/C was mark-
edly enhanced and the dechlorination reaction was significantly
suppressed in scCO2. The selectivity to CAN was >99% at 100% con-
version; however, they did not mention whether there were inter-
mediates accumulated during the course of reaction to the
complete conversion [32]. The present authors previously showed
that undesired intermediates were formed and accumulated dur-
ing the hydrogenation of nitrobenzene on Pt/C in scCO2 [33,34].
With Pd/C, the selectivity to o-CAN in scCO2 (92–94%) was also
2.2. CNB hydrogenation
The hydrogenation of CNB was examined over both catalysts at
35 °C. The reaction runs in scCO2 were conducted in a 50 cm3 auto-
clave. The reactor was charged with CNB 1.5 g (9.52 mmol) and a
catalyst sample (0.15 g Ni/TiO2 or 0.1 g Ni/Al2O3), flushed with
N2, and placed into a water bath preset to 35 °C for 20 min. After
the introduction of 4 MPa H2, liquid CO2 was introduced into the
reactor with a high-pressure liquid pump (Jasco SCF-Get) to the de-
sired pressure. The reaction was conducted while the reaction mix-
ture was being stirred with a magnetic stirrer. After the reaction,
the reactor was cooled with an ice-water bath for 20 min, depres-
surized carefully, and the reaction mixture was analyzed with a gas
chromatograph (Shimadzu GC-2010, Rtx-5 capillary column) using
a flame ionization detector. Hydrogenation reactions in ethanol

X. Meng et al. / Journal of Catalysis 269 (2010) 131–139
133
and n-hexane were conducted in the same reactor using similar
3.2. Features of Ni catalysts for selective hydrogenation of CNB
procedures. To ensure that the amount of hydrogen in the reactor
is sufficient enough for a complete conversion of CNB, all reactions
were performed at an initial H2 pressure of 4 MPa; the hydrogena-
tion could proceed smoothly at lower H2 pressures, as shown in
our previous work [37].
Table 1 shows the results of o-CNB hydrogenation over Ni/Al2O3
and Ni/TiO2 in scCO2 at 35 °C. The selectivity to the desired o-CAN
at ca. 80% conversion was >98% and 99% over Ni/Al2O3 and Ni/TiO2,
respectively. Moreover, the turnover frequency (TOF) value for the
reaction over Ni/Al2O3 was much less than that over Ni/TiO2 (en-
tries 1 and 2). These results indicate that chemoselective hydroge-
nation of o-CNB can be achieved in scCO2 over both Ni catalysts,
especially over Ni/TiO2.
Table 1 also compares the results of o-CNB hydrogenation over
Ni/TiO2 in scCO2 with that in polar ethanol and apolar n-hexane at
35 °C. The rate of o-CNB conversion was the highest in scCO2 and
the lowest in ethanol. The TOF value in scCO2 was ꢀ3.1 and ꢀ1.4
times higher than that in ethanol and n-hexane, respectively. In
ethanol, the dechlorination reaction was negligible but the inter-
mediates including CNSB, CAOB, CAB, and CHAB were formed in
a significant amount (entries 3, 4). Because CPHA decomposes to
CNSB and CAN during gas chromatograph analysis [42], the forma-
tion of CPHA during the hydrogenation cannot be excluded. Chen
and co-workers, who performed the hydrogenation of CNB at
90 °C over Ni/TiO2 in ethanol, also observed the formation of signif-
icant amounts of intermediates and negligible amount of AN [19].
In contrast, there was no accumulation of intermediates in n-hex-
ane, but a slight dechlorination occurred (entry 5). In scCO2, both
the dechlorination and the accumulation of intermediates were
suppressed satisfactorily; the selectivity to o-CAN was 99% at
82% conversion (entry 2).
2.3. Phase behavior and FTIR measurements
The solubility of CNB in scCO2 in the presence of 4 MPa H2 was
measured at 35 °C by the naked eye through the transparent sap-
phire windows attached to an 85 cm3 high-pressure reactor. Differ-
ent amounts of CNB substrate were added into the reactor and the
CO2 pressures corresponding to phase transition were carefully
determined. The details of experimental procedures were de-
scribed elsewhere [31]. The phase behavior of the reactant mixture
(i.e., H2 and CNB) in dense-phase CO2 was also examined in the
same reactor; the observation was made with a similar mass/vol-
ume ratio of CNB to the reactor volume as used in the hydrogena-
tion runs.
The high-pressure FTIR was used to examine the molecular
interactions of CO2 with the reacting species. The FTIR spectra of
p-CNB, p-CNSB, and p-CPHA in dense-phase CO2 were collected
with the same spectrometer in similar fashions as used in the pre-
vious work [31]. The measurements were made at 35 °C and in the
presence of 0–20 MPa CO2 and 4 MPa H2. The weight of the sam-
ples used was changed depending on its solubility in CO2. A higher
temperature of 90 °C was also used for p-CNB and p-CNSB at CO2
pressures <4 MPa and <6 MPa, respectively. For those measure-
ments, p-CNB (Wako) was used as received; commercially unavail-
able p-CNSB and p-CPHA, two of the intermediates (Scheme 1),
were synthesized by the oxidation of p-CAN (Wako) with H2O2
catalyzed with ammonium molybdate [38], and by the reduction
of p-CNB with zinc dust [39], respectively. The formation of these
compounds was confirmed by FTIR and NMR (JEOL A200II) [38,40].
Fig. 2 shows the evolution of product species with time during
the hydrogenation of o-CNB over Ni/TiO2 in scCO2. The yield of
o-CAN is similar to the conversion of o-CNB over the whole conver-
sion range from 9% to ꢀ100%; the yield of all byproducts including
AN and reaction intermediates decreased from <3.5% in 5 min to
<1% after 30 min. Thus, Ni/TiO2 is very effective for the chemose-
lective hydrogenation of o-CNB in scCO2. The combination of scCO2
and Ni/TiO2 is also effective for the chemoselective hydrogenation
of m- and p-CNB isomers (Table 2).
3. Results and discussion
The suppression of formation of hydroxylamine is a topic of
industrial importance because aryl hydroxylamines are generally
accumulated in large amounts during the hydrogenation of nitro-
aromatics with electron-withdrawing substituents (e.g., halogen
and sulfonamide) [3,7]. Studer et al. found that the accumulation
of hydroxylamine in methanol at 30–35 °C could be reduced from
70%–80% over unmodified Raney Ni to 11% by careful selection of
vanadium promoters; simultaneously, the reaction rate was down
by a factor of 0.1 [7]. Mahata et al. communicated that there was
no intermediate detected in methanol during the hydrogenation
of o- and p-CNB over filamentous carbon-stabilized Ni catalyst,
3.1. Properties of Ni catalysts
Ni/Al2O3 was the same as used previously for the hydrogenation
of nitrobenzene; the average Ni particle size was ca. 8.5 nm [37].
The Ni loading of Ni/Al2O3 and Ni/TiO2 was 41% and 16%, respec-
tively. The XRD patterns and TEM images of Ni/TiO2 are presented
in Fig. 1. After calcination, the existence of NiO and anatase TiO2
was identified. After reduction, Ni particles in size mostly between
5 and 25 nm were formed. The average particle size determined
from the XRD line broadening was ca. 14 nm.
NiO
Ni
(1)
(2)
TiO
2
b
a
10
20
30
40
50
60
70
80
90
2 theta (degree)
Fig. 1. (1) XRD patterns of Ni/TiO2: (a) calcined under air at 450 °C, (b) reduced with H2 at 450 °C; (2) TEM image of Ni/TiO2 after reduction with H2 at 450 °C.

134
X. Meng et al. / Journal of Catalysis 269 (2010) 131–139
Table 1
Hydrogenation of o-CNB over 41 wt% Ni/Al2O3 in scCO2 and over 16 wt% Ni/TiO2 in different reaction media.
Entry
Medium
Catalyst
C0 (mmol cmꢁ3
a
)
Conv. (%)
Selectivity (%)
TOFb (hꢁ1
)
o-CAN
AN
Others
1
2
3
4
5
scCO2
scCO2
Ni/Al2O3
Ni/TiO2
Ni/TiO2
Ni/TiO2
Ni/TiO2
0.19
0.19
0.19
0.95
0.95
78c
82
26
36
59
>98
99
70
70
97
<1.5
<1
<0.5
<1
90
318
101
140
229
Ethanold
Ethanol
n-Hexane
<1
>29e
30e
1
f
–
2
Reaction conditions: H2 4 MPa, o-CNB 1.5 g (9.52 mmol), Ni/Al2O3 0.1 g or Ni/TiO2 0.15 g, organic solvent 10 cm3 or CO2 10.5 MPa, reactor 50 cm3, 35 °C, 50 min.
a
Initial concentration of o-CNB.
b
Turnover frequency of hydrogenation of o-CNB. The number of active sites was calculated using the average nickel diameter (d) determined by XRD line broadening via
the following equation: Dispersion = 6(vm/am)/d, where vm and am are equal to 10.95 Å3 and 6.51 Å2, respectively, for Ni [41].
c
After 60 min of reaction.
d
Ethanol 50 cm3, reactor 100 cm3.
e
Other products in ethanol include CNSB, CAOB, CAB, and CHAB.
Not detected.
f
intermediates can be extended to a more challenging task of
chemoselective hydrogenation of CNB in the absence of organic
solvents. To our knowledge, such a chemoselective hydrogenation
of CNB to CAN over Ni catalysts at mild temperature has not been
reported so far.
100
80
60
40
20
0
3.3. Influence of CO2 pressure and phase behavior
The present hydrogenation occurs in a multi-phase system, in
which the reaction rate and/or the product distribution may de-
pend on the phase behavior. The hydrogenation of CNB was stud-
ied with the most effective Ni/TiO2 catalyst at different CO2
pressures while making the other conditions unchanged. Before
the reaction results, the phase behavior observations are given
first. Fig. 3 shows the changes in the phase state of CNB isomers
in scCO2 at different CO2 pressures and at 35 °C in the presence
of 4 MPa H2. At 35 °C, m- and p-CNB are solid in the absence of
CO2; m-CNB turned into liquid at ca. 3.6 MPa CO2 indicating that
its melting point could be lowered to <35 °C under dense-phase
CO2. The CNB substrates were soluble to some extent in scCO2 at
35 °C and their quantities in scCO2 increased with CO2 pressure.
The solubility was influenced somewhat by the substituent posi-
tion of the CNB isomers and followed the order of m- > o- and p-
CNB. With the same ratio of CNB weight to reactor volume as used
in the hydrogenation runs, the phase transition from two phases to
a single phase occurred at CO2 pressures of 12.3, 10.2, and
12.8 MPa for o-, m-, and p-CNB, respectively. Upon careful depres-
surization, the reverse transition was observed at 11.8, 9.6, and
11.6 MPa for o-, m-, and p-CNB, respectively.
0
20
40
60
80
100
120
Reaction time (min)
Fig. 2. Evolution of species with time during the hydrogenation of o-CNB over
16 wt% Ni/TiO2 in scCO2 at 35 °C. Reaction conditions: Ni/TiO2 0.15 g, CO2 11 MPa.
(N) o-CNB conversion, (s) o-CAN yield, (5) total yield of all byproducts.
and the specific activity was ꢀ50 mmol hꢁ1 gꢁ1 Ni (2.9 mmol
h
ꢁ1 mmolꢁ1 Ni) at 120 °C under 1.5 MPa H2 [8]. Recently, we have
shown that the accumulation of intermediates during the hydroge-
nation of nitrobenzene can be completely avoided over Ni/Al2O3 in
dense-phase CO2 [37]. The present hydrogenation of o-CNB over
Ni/TiO2 in scCO2 gave
a
TOF value of 318 hꢁ1 (22.9 mmol
h
ꢁ1 mmolꢁ1 Ni) at 35 °C under an initial H2 pressure of 4 MPa;
the TOF value was much higher than that in ethanol and n-hexane
(Table 1). Additionally, it is difficult to achieve a high chemoselec-
tivity in both ethanol and n-hexane because either significant accu-
mulation of intermediates or slight dechlorination takes place. In
the context, the present results are important because the reduc-
tion of intermediates can be achieved along with the increased
reaction rate and the negligible dechlorination. In other words,
the effectiveness of scCO2 and Ni catalysts for the reduction of
The influence of CO2 pressure was examined with a selected
substrate of o-CNB. Fig. 4 shows the changes in the conversion of
o-CNB and the selectivity to o-CAN with CO2 pressure. The total
conversion of o-CNB increased first and then decreased with
increasing CO2 pressure. The conversion was maximal at ca. 9–
12.5 MPa CO2. The pressure for the maximum conversion is close
Table 2
Results of hydrogenation of m- and p-CNB in scCO2 over 16 wt% Ni/TiO2.
Entry
Reactant
CO2 (MPa)
Time (min)
Conv. (%)
Selectivity (%)
CAN
AN
Others
1
2
m-CNB
p-CNB
11
10
30
80
64
100
96
>99.5
<0.1
<0.5
4
<0.1
a
3
4
30
90
51
100
99
>97
–
1
<2.5
<0.5
Reaction conditions: CNB 1.5 g (9.52 mmol), H2 4 MPa, 16 wt% Ni/TiO2 0.15 g, 35 °C.
a
Not detected.

X. Meng et al. / Journal of Catalysis 269 (2010) 131–139
135
0.30
0.25
0.20
0.15
0.10
0.05
0.00
v
(NO )
2
v (NO )
as
s
2
In ethanol
0.5
20.2 MPa
17.2
14.2
11.0
8.1
7.1
5.5
3.5
0 MPa
8
9
10
11
12
13
14
15
1200
1300
1400
1500
1600
1700
-1
Wavenumber (cm )
CO Pressure (MPa)
2
Fig. 5. FTIR spectra of nitro group stretching vibrations for p-CNB in the presence of
4 MPa H2 and dense-phase CO2 at different pressures given. The spectra at CO2
pressures <4 MPa and >5 MPa were collected at 90 °C and 35 °C, respectively. Also
included are the same group vibrations of p-CNB in ethanol (p-CNB/ethanol = 0.05,
molar ratio) at room temperature.
5
N, 4), and p-CNB (., ) in CO2 at 35 °C
Fig. 3. The phase behavior of o- (j, h), m- (
in the presence of 4 MPa H2. A certain amount of o-CNB was added to the view cell,
H2 was introduced up to 4 MPa, and then CO2 was gradually introduced until the
liquid o-CNB disappeared at a CO2 pressure P1. Upon carefully depressurization, the
first liquid drop of o-CNB appeared at a pressure P2. These two pressures P1 and P2
were carefully determined at several amounts (concentrations in the cell) of o-CNB,
and are given with solid and open marks, respectively. The data for m- and p-CNB
were collected using the same procedures.
stretching vibrations of the nitro group, vs(NO2) and vas(NO2) [2].
The peaks of vs(NO2) and vas(NO2) were shifted with CO2 pressure,
and these shifts are depicted in Fig. 6. The peak position of vs(NO2)
was blue shifted from 1343 cmꢁ1 at 0 MPa CO2 to 1350 cmꢁ1 at
7 MPa CO2 and remained unchanged at 7–20 MPa CO2. The peak
position of vas(NO2) was also blue shifted from 1521 to
1538 cmꢁ1 with CO2 pressure up to ca. 5.5 MPa; between 5.5 and
20 MPa CO2, the peak position was slightly shifted to a lower wave-
number and then changed little. The absorption bands at 20 MPa
CO2 were located at larger wavenumbers than those at 0 MPa
CO2 and in ethanol. These FTIR results indicate that the N–O bond
of the nitro group becomes stronger in the presence of dense-phase
CO2 compared to the ambient gas and liquid conditions. Addition-
ally, it was observed that an absorption band appeared at
1109 cmꢁ1 at 0 MPa CO2 and was blue shifted to 1112 cmꢁ1 with
CO2 pressure up to 7 MPa, above which it did not change further
(not shown). This band is due to the C–N stretch [43].
100
80
60
40
6
8
10
12
14
16
18
The above changes of vs(NO2) and vas(NO2) for p-CNB with CO2
pressures were similar to those of nitrobenzene in dense-phase
CO2 [37]. We have proposed that the overall IR result of nitroben-
zene in dense-phase CO2 is controlled by the following modes of
interactions between CO2 molecules and the nitro group: (1) one
CO pressure (MPa)
2
Fig. 4. Influence of CO2 pressure on the conversion of o-CNB (d) and the selectivity
to o-CAN (s) in the hydrogenation of o-CNB over 16 wt% Ni/TiO2 at 35 °C and at H2
4 MPa for 50 min.
to that of the reversible phase transition of o-CNB (11.8–12.3 MPa
CO2), indicating that the phase behavior is an important factor in
determining the rate of CNB hydrogenation. However, the high
selectivity to o-CAN remained unchanged when the reaction mix-
ture changed from the gas–liquid–solid system to the gas–solid
system.
1540
3.9 cm-1
1530
12.5 cm-1
1520
1350
3.4. FTIR measurements: molecular interactions with CO2
High-pressure FTIR was used to examine molecular interactions
of dense-phase CO2 with such selected reacting species as p-CNB,
p-CNSB, and p-CPHA, which would affect the reaction pathways
of selective hydrogenation of CNB and change the reaction rate
and the product selectivity.
6.3 cm-1
1345
1340
0
5
10
15
20
25
CO pressure (MPa)
2
3.4.1. p-CNB
Fig. 5 shows the FTIR spectra of nitro group vibrations in the
presence of 4 MPa H2 and 0–20 MPa CO2. The result of p-CNB in
ethanol is also shown for comparison. The bands at ca. 1343 and
1521 cmꢁ1 were assigned to the symmetric and asymmetric
Fig. 6. Influence of CO2 pressure on the absorption peak position of vs(NO2) and
vas(NO2) of p-CNB. (s) Symmetric and (d) asymmetric stretching vibrations in
dense-phase CO2; (4) symmetric and (N) asymmetric stretch in ethanol (p-CNB/
ethanol = 0.05, molar ratio) at room temperature.

136
X. Meng et al. / Journal of Catalysis 269 (2010) 131–139
of the oxygen atoms of CO2 interacts with the positive nitrogen
of –NO2 group and (2) the carbon atom of CO2 interacts with the
oxygen of –NO2 group. The former interaction might strengthen
both the C–N and N–O bonds, while the latter might weaken the
strength of the N–O bond. The former interaction has a dominant
effect on the stretch vibration of N–O bond [37]. It seems likely
that the nitro group of p-CNB interacts with CO2 in the same modes
as in the case of nitrobenzene.
At similar CO2 pressures, the bands of vs(NO2) and vas(NO2) of p-
CNB appeared at lower wavenumbers than those of the corre-
sponding peaks of nitrobenzene. For example, the wavenumbers
of vs(NO2) and vas(NO2) for p-CNB at 20 MPa CO2 were 1350 and
1534 cmꢁ1; the corresponding values for nitrobenzene were 1354
and 1538 cmꢁ1, respectively [37]. The difference in wavenumbers
of the nitro group vibrations between p-CNB and nitrobenzene
might be due to the Cl substituent, an electron-withdrawing group
attached to the benzene ring of p-CNB.
1520
1515
1510
1505
1500
10.9 cm-1
In dense phase CO2
4.8 cm-1
25
In ethanol
5
0
10
15
20
CO2 pressure (MPa)
Fig. 8. Influence of CO2 pressure on the peak position of the nitroso group (O-
interacted) vibration for p-CNSB at 35 °C. The spectrum in the absence of CO2
(0 MPa) was collected at 90 °C. Also shown is the same group vibration in ethanol
(p-CNSB/ethanol = 0.03, molar ratio) at room temperature.
3.4.2. p-CNSB
Fig. 7 shows the FTIR spectra of the nitroso (N@O) moiety of
p-CNSB in the presence of 4 MPa H2 and 0–20 MPa CO2. The spec-
trum of p-CNSB in ethanol at room temperature was recorded for
comparison. The absorption bands at 1109 and 1517 cmꢁ1 before
the introduction of CO2 are due to the C–N stretch of the CNSB
monomer and the stretching vibration of N@O, respectively [43].
The peak position of v(C–N) of CNSB monomer was slightly blue
shifted from 1109 to 1116 cmꢁ1 with the increase of CO2 pressure
up to ca. 15 MPa and then remained unchanged. We have shown
that the N@O group of nitrosobenzene can interact with CO2
molecules through its oxygen (O-interacted) and nitrogen
(N-interacted) atoms; the stretching vibration of N@O gives a band
around 1520 cmꢁ1 in the former case and a band at 1485 cmꢁ1 in
the latter case [37]. Thus, the band at ca. 1510 cmꢁ1 in the spectra
of p-CNSB would arise from the stretching vibration v(N@O)
(O-interacted). The peak position of v(N@O)(O-interacted) changed
with CO2 pressure and these shifts are shown in Fig. 8. The stretch-
ing vibration v(N@O)(O-interacted) in CO2 was monotonously red
shifted with CO2 pressure up to 13 MPa and then remained un-
changed. Additionally, there was no peak at 1485 cmꢁ1 in the spec-
tra of p-CNSB at all pressures examined. Therefore, the N@O of
p-CNSB interacts with the carbon of CO2 through its oxygen but
not through its nitrogen. The N@O group functions as a Lewis base
(LB) and CO2 plays a role of a Lewis acid (LA). This type of intermo-
lecular interaction becomes stronger with the increase in CO2 pres-
sure and weakens the N@O bond of p-CNSB. Simultaneously, the
C–N stretch band was blue shifted, in agreement with the previous
results for nitroso compounds that v(C–N) increases when v(N@O)
decreases [37,43].
At 20 MPa CO2, the peak position of v(N@O)(O-interacted) was
at 1506 cmꢁ1
, which was still higher than that in ethanol
(1501 cmꢁ1). This difference is similar to that of nitrosobenzene,
indicating that the LA–LB interaction of CO2 with the N@O group
of p-CNSB is also less strong compared with the hydrogen bonding
in ethanol [37].
In the spectra of p-CNSB, the band of v(N@O)(O-interacted) was
shifted from 1517 cmꢁ1 at 0 MPa CO2 to 1506 cmꢁ1 at 20 MPa CO2;
in the case of nitrosobenzene, the same vibration gave a band at
1521 and 1513 cmꢁ1 at the corresponding pressures [37]. That is,
the stretching vibration of N@O of p-CNSB is less strong than that
on nitrosobenzene at similar CO2 pressures. The difference might
be due to the electron-withdrawing effect of the Cl substituent of
p-CNSB.
3.4.3. p-CPHA
The FTIR spectra of p-CPHA in dense-phase CO2 and in ethanol
are given in Fig. 9. The absorption bands at ca. 920, 1494–1500,
v(N=O)
v(C-N)
v(N-O)
Ring stret.
0.5
0.5
In ethanol
In ethanol
20.4 MPa
20.3 MPa
17.5
13.0
11.2
9.5
17.1
11.1
9.7
8.6
8.2
7.8
7.0
6.8
0.2
5.5
5.3
0 MPa
1550
0 MPa
800
900
1450
1500
1550
1600
1050
1100
1450
1500
-1
-1
Wavenumber (cm )
Wavenumber (cm )
Fig. 7. FTIR spectra of p-CNSB in the presence of 4 MPa H2 and dense-phase CO2 at
pressures given. The spectra at CO2 pressures <6 MPa and >6 MPa were collected at
90 °C and 35 °C, respectively. Also shown is the spectrum of p-CNSB in ethanol (p-
CNSB/ethanol = 0.03, molar ratio) at room temperature.
Fig. 9. FTIR spectra for the ring stretching vibration of p-CPHA in the presence of
4 MPa H2 and dense-phase CO2 at pressures given (35 °C). The spectrum of p-CPHA
in ethanol (p-CPHA/ethanol = 0.15, molar ratio) collected at room temperature is
also shown.

X. Meng et al. / Journal of Catalysis 269 (2010) 131–139
137
and 3200 cmꢁ1 are related to the N–O stretch, a ring stretching
mode, and the O–H stretching vibration of phenylhydroxylamine
derivatives, respectively [44,45]. The O–H stretch band of p-CPHA
in dense-phase CO2 was not distinct at all pressures examined,
possibly because scCO2 has a combination band in the region
where the stretching vibration of O–H occurs [46]. The N–O stretch
of p-CPHA could only be identified at 915 cmꢁ1 above 8 MPa CO2,
and its peak position did not show any shift with CO2 pressures
from 8 to 20 MPa. Here, our interest is focused on the absorption
band at 1494 cmꢁ1, which is related to the ring stretch of p-CPHA,
for the discussion on interactions between p-CPHA and CO2
molecules.
The position of the ring absorption band at 1494 cmꢁ1 scarcely
changed between ca. 0 and 5 MPa CO2, and was blue shifted to
1498 cmꢁ1 with further increasing CO2 pressure up to ca. 8 MPa.
At 9.7–20.4 MPa CO2, the band split into two peaks at 1498 and
1505 cmꢁ1, and their positions did not change with CO2 pressures.
The splitting of the ring absorption band was also observed for N-
phenylhydroxylamine at above 8 MPa CO2; however, the band was
red shifted at 4–11 MPa CO2 and its position remained unchanged
from 11 to 20 MPa CO2 [37].
3.4.4. CO2
Interactions of CO2 with LB, polymer, or ionic liquids cause
some changes in the IR spectra corresponding to the CO2 v2-bend-
ing (610–680 cmꢁ1
) and v3-antisymmetric stretching (2100–
2500 cmꢁ1) mode regions [47–49]. It was impossible to distinguish
the v2 mode in our IR results. Fig. 10 gives the FTIR spectra of CO2 in
the v3 region for the CO2–p-CNB, CO2–p-CNSB, and CO2–p-CPHA
systems at different CO2 pressures. The results in Fig. 10 were ob-
tained by using the substrate-absent system at the corresponding
pressures as background. For the CO2–p-CNB system, there was
one main peak at 2347 cmꢁ1 at CO2 pressures >8 MPa and the peak
position scarcely changed with CO2 pressure. For the CO2–p-CNSB
system, there was only one peak at 2352 cmꢁ1 at above 9.5 MPa,
and its position did not change with CO2 pressure. The fact that
only one band appears in the v3 region suggests that there is only
one type of the site within the p-CNB or p-CNSB molecule to inter-
act with CO2 molecules [48]. Additionally, the wavenumber of the
C@O stretching vibration of CO2 in the CO2–p-CNB system was
lower than that in the CO2–p-CNSB system. This indicates that
the C@O bond strength of CO2 is weaker in the CO2–p-CNB system
than that in the CO2–p-CNSB system. The IR results of p-CNB have
shown that the interaction of CO2 through its oxygen with the
nitrogen of –NO2 group had a dominant effect on the stretching
vibration of N–O bond; both the C–N and N–O bonds were
strengthened. It seems likely that the interaction between CO2
and p-CNB induces partial electron transfer from the C@O bond
of CO2 to the C–N and N–O bonds of p-CNB, and this weakens
the C@O bond of CO2. In contrast, the N@O of p-CNSB interacted
with the carbon of CO2 through its oxygen (not nitrogen), and
the N@O bond was weakened with CO2 pressures. The interaction
of CO2 with p-CNSB is likely to induce partial electron transfer from
the N@O bond of p-CNSB to the C@O bond of CO2, strengthening
the C@O bond. Thus, the IR spectra of CO2 provide additional evi-
dence for the interaction mode of CO2 with p-CNB and p-CNSB.
Fig. 10c shows the FTIR difference spectra of CO2 in the v3 region
for the CO2–p-CPHA system at different pressures. At above
8.6 MPa CO2, there was one or two peaks at around 2351 cmꢁ1
with no other peaks in the v3 region. In contrast, an additional peak
appears at 2332 cmꢁ1along with the bands at 2348–2355 cmꢁ1 for
the CO2–N-phenylhydroxylamine system [37]. This indicates that
p-CPHA does interact with CO2; however, the available sites within
the p-CPHA molecule for CO2 are less than those within the N-
phenylhydroxylamine molecule. Presumably, CO2 interacts with
p-CPHA through its carbon.
2347
a
20.2 MPa
0.2
17.2
14.2
11
8.1 MPa
2450
2250
2300
2350
2400
-1
Wavenumber (cm )
2352
b
20.3 MPa
17.5
13.0
11.2
9.5
0.2
8.2 MPa
2250
2300
2350
2400
2450
-1
Wavenumber (cm )
c
2351
0.2
20.4 MPa
17.1
14.1
11.1
9.7
3.5. Improvement of the selectivity to CAN
8.6
7.8 MPa
The dechlorination reaction was suppressed over Ni/TiO2 in
either ethanol or scCO2, which might be mainly related to the
properties of Ni/TiO2. The formation of harmful intermediates
occurring in ethanol was almost completely retarded in scCO2,
indicating the significance of medium effects. The FTIR results
show that the strength of the N–O bond of CNB in dense-phase
CO2 was stronger than that in the ambient gas phase and it in-
creased with CO2 pressure up to ca. 6 MPa, above which it slightly
decreased (Fig. 6). However, the strength of the N@O bond of CNSB
2250
2300
2350
2400
2450
-1
Wavenumber (cm )
Fig. 10. FTIR difference spectra in the CO2 v3-antisymmetric-stretching mode
region for the CO2-p-CNB (a), CO2-p-CNSB (b), and CO2-CPHA (c) system at CO2
pressures given. All spectra were collected at 35 °C in the presence of 4 MPa H2. The
difference spectra were obtained by using the substrate-absent system at the
corresponding pressures as background.

138
X. Meng et al. / Journal of Catalysis 269 (2010) 131–139
was monotonously reduced with CO2 pressure (Fig. 8). Namely, the
reactivity of the –NO2 group may decrease with CO2 pressure up to
6 MPa and will not change so much at higher pressures. In contrast,
the reactivity of the N@O group may increase with CO2 pressure. It
is likely, therefore, that the hydrogenation of CNSB is more signif-
icantly accelerated than that of CNB when dense-phase CO2 is
used. Moreover, the FTIR results indicate that interactions occurred
between CPHA and dense-phase CO2 molecules. These interactions
should promote the transformation of CPHA to CAN (step III in
Scheme 1), contributing to the improved selectivity to CAN. As a re-
sult, CAN was formed with almost 100% selectivity in/under dense-
phase CO2. Therefore, the molecular interactions of CO2 with the
reacting species should be an important factor in determining
the product selectivity over Ni/TiO2.
to CAN is likely promoted. Probably, the hydrogenation of CNB
mainly occurs through the direct hydrogenation route, i.e.,
CNB ? CNSB ? CPHA ? CAN. The transformation of CNB to CNSB
might be the rate-determining step.
The strength of the N–O bond of CNB, the N@O bond of CNSB,
and the N–O bond of CPHA is weaker compared with the corre-
sponding bonds of nitrobenzene, nitrosobenzene, and N-phen-
ylhydroxylamine, due to the electron-withdrawing effect of Cl
substituent. There is no interaction between CO2 and the nitrogen
of N@O group of CNSB, which is different from the modes of CO2
interaction with nitrosobenzene. The available sites to interact
with CO2 are also less within the CPHA molecule than within the
N-phenylhydroxylamine one.
Acknowledgments
3.6. Reaction pathways for the hydrogenation of CNB in scCO2 over Ni/
TiO2
The authors gratefully acknowledge the financial support from
the One Hundred Talent Program of CAS, NSFC 20873139, and
KJCX2, YW.H16. This work was also supported by the Japan Society
for the Promotion of Science with Grant-in-Aid for Scientific Re-
search (B) 18360378 and by the CAS-JSPS International Joint Pro-
ject GJHZ05. X. Meng thanks the Global COE (Center of Excellence)
fellowship of Hokkaido University.
The path IV in Scheme 1 occurs on Au/TiO2, and the step V be-
comes significant over Pd/Pt/Ni catalysts in the presence of vana-
dium promoters [2,3,7]. Herein, we discuss the reaction
pathways over Ni/TiO2 in dense-phase CO2 without the consider-
ation of pathways IV and V. The total yield of all byproducts was
<3.5% over the whole conversion range of 9–100% (Fig. 2 and Table
2). Accordingly, the condensation route (steps VI and VII) in
Scheme 1 should be negligible, and the hydrogenation of CNB to
CAN proceeds likely through the consecutive steps I ? II ? III.
The rate of CAN formation is determined by the slowest reaction
rate among steps I, II, and III. The intermediates, detected by GC
analysis, included CNSB and trace amounts of CAOB, CAB, and
CHAB. These intermediates may be present in the reaction mix-
ture; or they are absent in the reaction mixture but come from
the decomposition of CPHA during the GC analysis [42]. Anyway,
if the reaction rate of step II or III is the slowest, the amount of
these intermediates should increase first and then decrease with
the conversion of CNB. Fig. 2 shows that the yield of all byproducts
decreased from 3.5% at 9% conversion to <1% at conversion >70%.
The yield of CAN was similar to the conversion of CNB over the
conversion range of 9–100%. In other words, the rate of CAN forma-
tion was almost equal to that of CNB conversion, i.e., the rate of
step I. Therefore, in the present case of CNB hydrogenation in
dense-phase CO2 over Ni/TiO2, the transformation of CNB to CNSB
(step I in Scheme 1) seems to be the relatively slow step that deter-
mines the rate of CAN formation. The proposed reaction pathways
may be related to the molecular interactions of CO2 with CNB,
CNSB, and CPHA. The relative reactivity of these reacting species
was changed through interactions with CO2 as discussed above,
and consequently, the relative hydrogenation rates of steps I, II,
and III are altered in dense-phase CO2.
References
[1] X. Wang, M. Liang, J. Zhang, Y. Wang, Curr. Org. Chem. 11 (2007) 299.
[2] A. Corma, P. Concepcion, P. Serna, Angew. Chem., Int. Ed. 46 (2007) 7266.
[3] P. Baumeister, H.U. Blaser, M. Studer, Catal. Lett. 49 (1997) 219.
[4] H.U. Blaser, A. Schnyder, H. Steiner, F. Rossler, P. Baumeister, in: G. Ertl, H.
Knözinger, F. Schüth, J. Weitkamp (Eds.), Handbook of Heterogeneous
Catalysis, Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, Germany, 2008,
p. 3291.
[5] F. Cardenas-Lizana, S. Gomez-Quero, M. Keane, Appl. Catal. A 334 (2008) 199.
[6] A. Corma, P. Serna, Science 313 (2006) 332.
[7] M. Studer, S. Neto, H.U. Blaser, Top. Catal. 13 (2000) 205.
[8] N. Mahata, A.F. Cunha, J.J.M. Órfão, J.L. Figueiredo, Catal. Commun. 10 (2009)
1203.
[9] F. Cárdenas-Lizana, S. Gómez-Quero, A. Hugon, L. Delannoy, C. Louis, M. Keane,
J. Catal. 262 (2009) 235.
[10] Y. Chen, C. Wang, H. Liu, J. Qiu, X. Bao, Chem. Commun. (2005) 5298.
[11] Y. Chen, J. Qiu, X. Wang, J. Xiu, J. Catal. 242 (2006) 227.
[12] D. He, H. Shi, Y. Wu, B. Xu, Green Chem. 9 (2007) 849.
[13] L. Liu, B. Qiao, Y. Ma, J. Zhang, Y. Deng, Dalton Trans. (2008) 2542.
[14] X. Wang, M. Liang, H. Liu, Y. Wang, J. Mol. Catal. A 273 (2007) 160.
[15] M. Liang, X. Wang, H. Liu, H. Liu, Y. Wang, J. Catal. 255 (2008) 335.
[16] F. Wang, J. Liu, X. Xu, Chem. Commun. (2008) 2040.
[17] B. Coq, A. Tijani, R. Dutartre, F. Figuéras, J. Mol. Catal. 79 (1993) 253.
[18] A. Corma, P. Serna, P. Concepción, J. Calvino, J. Am. Chem. Soc. 130 (2008) 8748.
[19] J. Xiong, J. Chen, J. Zhang, Catal. Commun. 8 (2007) 345.
[20] C. Xiao, H. Wang, X. Mu, Y. Kou, J. Catal. 250 (2007) 25.
[21] C.A. Eckert, B.L. Knutson, P.G. Debenedetti, Nature 383 (1996) 313.
[22] P.G. Jessop, W. Leitner, in: P.G. Jessop, W. Leitner (Eds.), Chemical Synthesis
Using Supercritical Fluids, Wiley-VCH, Weinheim, Germany, 1999, p. 9.
[23] A. Baiker, Chem. Rev. 99 (1999) 453.
[24] B.M. Bhanage, M. Arai, Catal. Rev. Sci. Eng. 43 (2001) 315.
[25] J.R. Hyde, P. Licence, D. Carter, M. Poliakoff, Appl. Catal. A 222 (2001) 119.
[26] W. Leitner, Acc. Chem. Res. 35 (2002) 746.
[27] E.J. Beckman, J. Supercrit. Fluid 28 (2004) 121.
[28] B. Subramaniam, M.A. McHugh, Ind. Eng. Chem. Process Des. Dev. 25 (1986) 1.
[29] C.M. Rayner, Org. Process Res. Dev. 11 (2007) 121.
4. Conclusions
[30] T. Seki, J.D. Grunwaldt, A. Baiker, Ind. Eng. Chem. Res. 47 (2008) 4561.
[31] F. Zhao, S. Fujita, S. Akihara, M. Arai, J. Phys. Chem. A 109 (2005) 4419.
[32] S. Ichikawa, M. Tada, Y. Iwasawa, T. Ikariya, Chem. Commun. (2005) 924.
[33] F. Zhao, R. Zhang, M. Chatterjee, Y. Ikushima, M. Arai, Adv. Synth. Catal. 346
(2004) 661.
[34] F. Zhao, Y. Ikushima, M. Arai, J. Catal. 224 (2004) 479.
[35] C. Xi, H. Cheng, J. Hao, S. Cai, F. Zhao, J. Mol. Catal. A 282 (2008) 80.
[36] M. Mihaylov, E. Ivanova, Y. Hao, K. Hadjiivanov, H. Knözinger, B.C. Gates, J.
Phys. Chem. C 112 (2008) 18973.
The chemoselective hydrogenation of CNB to CAN was achieved
over Ni/TiO2 in scCO2 at 35 °C. Both the accumulation of harmful
intermediates and the formation of dechlorinated byproduct were
suppressed effectively in scCO2. The selectivity to the desired prod-
uct, CAN, was >97% over the conversion range of 9–100%. One of
the important factors for this improved selectivity to CAN is the
interactions of CO2 with the reacting species, i.e., CNB, CNSB, and
CPHA. The nitro group of CNB interacts with CO2 through its nitro-
gen and oxygen atoms. The former mode has a dominant effect on
the stretch vibration of N–O bond, and the reactivity of CNB is de-
creased. The nitroso group of CNSB interacts through its oxygen
with the carbon of CO2 and the reactivity of CNSB is increased.
CPHA can also interact with scCO2, and the transformation of CPHA
[37] X. Meng, H. Cheng, Y. Akiyama, Y. Hao, W. Qiao, Y. Yu, F. Zhao, S. Fujita, M. Arai,
J. Catal. 264 (2009) 1.
[38] A. Defoin, Synthesis (2004) 706.
[39] O. Kamm, C.S. Marvel, Org. Synth. 4 (1925) 57.
[40] S. Ung, A. Falguières, A. Guy, C. Ferroud, Tetrahedron Lett. 46 (2005) 5913.
[41] G. Bergeret, P. Gallezot, in: G. Ertl, H. Knözinger, F. Schüth, J. Weitkamp (Eds.),
Handbook of Heterogeneous Catalysis, Wiley-VCH Verlag GmbH & Co. KGaA,
Weinheim, Germany, 2008, p. 738.

X. Meng et al. / Journal of Catalysis 269 (2010) 131–139
139
[42] S. Galvagno, A. Donato, G. Neri, R. Pietropaolo, Z. Poltarzewski, J. Mol. Catal. 42
[46] J.D. Grunwaldt, R. Wandeler, A. Baiker, Catal. Rev. Sci. Eng. 45 (2003) 1.
[47] T. Seki, J.D. Grunwaldt, A. Baiker, J. Phys. Chem. B 113 (2009) 114.
(1987) 379.
[43] S. Meijers, V. Ponec, J. Catal. 160 (1996) 1.
[44] Y.K. Agrawal, J. Chem. Eng. Data 22 (1977) 70.
[45] Y. Wu, J. Jiang, Y. Ozaki, J. Phys. Chem. A 106 (2002) 2422.
[48] S.G. Kazarian, M.F. Vincent, F.V. Bright, C.L. Liotta, C.A. Eckert, J. Am. Chem. Soc.
118 (1996) 1729.
[49] J.C. Dobrowolski, M.H. Jamróz, J. Mol. Struct. 275 (1992) 211.