Mechanistic study on aerobic oxidation of amine over intermetallic Pd3Pb: Concerted promotion effects by Pb and support basicity

Article abstract of DOI:10.1021/cs501695m

A mechanistic study on aerobic oxidation of amine to imine over Pd₃Pb/MO_x (MO_x = SiO₂, TiO₂, Al₂O₃, and MgO) intermetallic catalysts was performed to clarify the role of Pb and the support in enhancing catalytic activity. Results from X-ray absorption and photoelectron spectroscopies revealed that formation of the Pd₃Pb phase made Pd electron-rich compared to pure Pd, whereas the electronic states of Pd in Pd₃Pb/MO_x were identical and independent of the nature of the support. Kinetic studies indicated that desorption of imine was promoted by Pb and that adsorption of amine was accelerated by basic sites on the support. Infrared temperature programmed desorption (IR-TPD) experiments demonstrated that desorption of imine was indeed promoted on Pd₃Pb compared to Pd. The support effect only appears on Pd₃Pb catalysts and not on pure Pd due to change in the rate-determining step from imine desorption to amine adsorption. The combination of Pb and the support basicity provides a unique and highly efficient bifunctional catalysis. (Chemical Equation Presented).

Full text of DOI:10.1021/cs501695m

Research Article
pubs.acs.org/acscatalysis
Mechanistic Study on Aerobic Oxidation of Amine over Intermetallic
Pd3Pb: Concerted Promotion Effects by Pb and Support Basicity
Shinya Furukawa,*,† Akifusa Suga,‡ and Takayuki Komatsu*,‡
‡
†Department of Chemistry and Department of Chemistry and Materials Science, Tokyo Institute of Technology, 2-12-1-E1-10
Ookayama, Meguro-ku, Tokyo 152-8551, Japan
S
* Supporting Information
ABSTRACT: A mechanistic study on aerobic oxidation of amine to imine
over Pd3Pb/MOx (MOx = SiO2, TiO2, Al2O3, and MgO) intermetallic catalysts
was performed to clarify the role of Pb and the support in enhancing catalytic
activity. Results from X-ray absorption and photoelectron spectroscopies
revealed that formation of the Pd3Pb phase made Pd electron-rich compared to
pure Pd, whereas the electronic states of Pd in Pd3Pb/MOx were identical and
independent of the nature of the support. Kinetic studies indicated that
desorption of imine was promoted by Pb and that adsorption of amine was
accelerated by basic sites on the support. Infrared temperature programmed
desorption (IR-TPD) experiments demonstrated that desorption of imine was indeed promoted on Pd3Pb compared to Pd. The
support effect only appears on Pd3Pb catalysts and not on pure Pd due to change in the rate-determining step from imine
desorption to amine adsorption. The combination of Pb and the support basicity provides a unique and highly efficient
bifunctional catalysis.
KEYWORDS: Pd3Pb, amine oxidation, intermetallic compound, concerted catalysis, mechanistic study
Very recently, we reported that supported Pd-based
intermetallic compounds, such as Pd3Pb and Pd3Bi, showed
much greater catalytic performances in aerobic oxidation of
amine to imine than monometallic Pd.17 A kinetic study
implied that desorption of the product imine was promoted by
formation of the intermetallic phase, enhancing the catalytic
performance. This effect differs from those in the previously
reported systems as mentioned above. Therefore, it would be a
great impact in Pd-based oxidation chemistry to demonstrate
and rationalize the unique effect of the second metal, Pb or Bi.
However, we have obtained no obvious support including
spectroscopic evidence, remaining the role of the second metal
and the detailed reaction mechanism unclear. In addition to the
effect of Pb, we also observed that basicity of the support
strongly enhanced the catalytic activity of Pd3Pb. In this
context, a deep understanding of the reaction mechanism will
also discover a unique concerted catalysis assisted by the
second metal and the support.
INTRODUCTION
■
Pd- or Pt-mediated oxidation is a very important chemical
transformation in industrial chemistry, organic synthesis, and
pharmaceuticals. To date, many effective catalytic oxidation
systems have been reported. The principle one is aerobic
oxidation of organic molecules such as alcohols,1,2 amines,3 and
hydrocarbons.4,5 In this field, development of Pd- or Pt-based
bimetallic systems that are more effective than monometallic
catalysts has been widely studied.1,2 Among a series of second
metal elements coupled with the noble metals, typical heavy
elements such as Pb, Bi, Tl, and Te have been found to be most
effective.1,2 A number of rationales for the enhanced catalytic
activity of these bimetallic systems have been reported: (1) so-
called ensemble effects suppressing a side reaction yielding
poisoning species (i.e., C−C cleavage),6,7 (2) supply of
adsorption sites for oxygen,8−10 (3) complex formation
between the second metal and the substrate,11 (4) prevention
of Pd corrosion,2,12 (5) synergetic effects by the noble and
second metals,2,13 and (6) suppression of hydrogen sorption.14
Recently, Baiker and co-workers employed in situ X-ray
absorption fine structure (XAFS) and attenuated total
reflection infrared (ATR-IR) techniques in aerobic oxidation
of alcohols over a Pd−Bi catalyst, revealing that Bi inhibited
side reactions and supplied oxygen to the catalyst surface.15
They also used a similar technique to a Pt−Bi system and
reported that Bi protected oxidation of Pt and blocked
adsorption sites for solvent−Pt interaction.16 Thus, informative
insight has been provided by spectroscopic methods. However,
the nature of this chemistry remains under discussion and is an
attractive field for further investigation.
In this study, detailed characterization of supported Pd3Pb
catalysts combined with kinetic studies in amine oxidation was
performed to elucidate the reaction mechanism and the role of
Pd3Pb. A series of supported Pd3Pb, Pd3Pb/MOx (MOx = SiO2,
TiO2, Al2O3, and MgO), were used also to clarify the effect of
support on the catalytic performance. Herein, we propose not
only a unique effect of Pb in Pd-based oxidation chemistry but
Received: October 31, 2014
Revised: December 19, 2014
Published: January 15, 2015
© 2015 American Chemical Society
1214
DOI: 10.1021/cs501695m
ACS Catal. 2015, 5, 1214−1222

ACS Catalysis
Research Article
also a unique concerted catalytic system of Pd3Pb and the
support basicity.
flow (50 mL min−1) at a predetermined temperature for 1 h
(Pd/TiO2, 400 °C; Pd3Pb/TiO2, 450 °C; Pd/MgO, 500 °C;
Pd/Al2O3, Pd3Pb/Al2O3, and Pd3Pb/MgO, 600 °C) to obtain
nanoparticles of similar sizes. Unsupported bulk Pd3Pb was
prepared by arc melting of Pd and Pb beads with a 3:1 atomic
ratio under an Ar atmosphere. The resulting ingot was crushed
and ground into a fine powder.
EXPERIMENTAL SECTION
■
Materials. All chemical reagents except deuterated dibenzyl-
amines were purchased or supplied: (NH4)2PdCl4 (Aldrich,
97%), Pb(NO3)2 (Kanto, 99%), SiO2 (Cariact G-6, Fuji Silysia
Co., SBET = 673 m2 g−1), TiO2 (JRC-TIO-7, SBET = 270 m2 g−1,
anatase), Al2O3 (JRC-ALO-8, SBET = 148 m2 g−1, γ + θ phase),
MgO (JRC-MGO-4, SBET = 28−38 m2 g−1), H2 (Taiyo Nippon
Sanso, 99.9999%), He (Taiyo Nippon Sanso, 99.9%), 5% O2/
Ar (Taiyo Nippon Sanso), dibenzylamine (TCI, 97%), N-
benzylidenebenzylamine (Wako, 96%), p-xylene (TCI, 99%),
toluene (Kanto 99.5%), benzylamine-α,α-d2 (CDN Isotopes,
99.3 atom %D), benzyl bromide-α,α-d2 (CDN Isotopes, 98
atom %D), D2O (Aldrich, 99.9 atom %D), dimethylformamide
(TCI, 99.5%), Cs2CO3 (Kanto, 98%), Na2SO4 (Wako, 99%),
BN (Wako, 99%).
Characterization. Dispersion of Pd was estimated by CO
chemisorption under a dynamic condition using an assumption
of 1:1 chemisorption to Pd atom. A certain amount of the
catalyst (typically 100 mg) was placed in a quartz tube and
reduced under flowing H2 at 400 °C for 0.5 h. After the
pretreatment, He was introduced at the same temperature for
0.5 h to remove hydrogen, followed by cooling to room
temperature. A pulse of 5% CO/He was introduced into the
reactor and the passed CO was quantified by a thermal
conductivity detector downstream. This was repeated until the
adsorption reached saturation. Powder X-ray diffraction (XRD)
patterns of the prepared materials were recorded by a Rigaku
RINT2400 using a Cu Kα X-ray source. The phase
composition was calculated based on the intensity ratio
Pd3Pb 111 and Pd 111 diffractions and the atomic scattering
factors19 of Pd and Pb (see the Supporting Information for a
detailed derivation). X-ray absorption experiments were carried
out on a beamline BL01B1 at SPring-8 (Hyogo, Japan) with a
ring energy of 8 GeV and a stored current from 60 to 100 mA.
Prior to the measurement, the supported catalysts were
pelletized, followed by reduction under a H2 flow at 400 °C
for 0.5 h. After the reduction pretreatment, the pellets were
sealed into polyethylene bags under a dry Ar atmosphere
without exposure to the air. Bulk Pd3Pb powder was diluted
with BN (boron nitride) and pelletized without any pretreat-
ment. Pd−K edge (24.3 keV) and Pb−LIII edge (13.0 keV) X-
ray absorption fine structure (XAFS) spectra were recorded in
transmission mode with quick scan. A fixed exit Si(111) double
crystal monochromator with an energy resolution of ca. 0.3 eV
in the X-ray absorption near edge structure (XANES) region
was used. Higher harmonics were reduced by reflection on the
two mirrors. For all spectra, a metallic Cu reference foil was
used to provide an energy calibration for the monochromator.
The XANES and extended XAFS (EXAFS) analyses were
performed by Athena.20 X-ray photoelectron spectra (XPS) of
the intermetallic compounds were measured with an ULVAC
PHI 5000 VersaProbe spectrometer. The catalyst was pressed
into a pellet and placed into a quartz reactor, where it was
reduced under flowing hydrogen (50 mL min−1) at 400 °C for
0.5 h prior to measurement. Spectra were obtained with an Al
Kα X-ray source, using C 1s as a reference for binding energy.
Fourier-transformed infrared (FT-IR) spectra of the adsorbed
imine were obtained with a JASCO FT/IR-430 spectrometer in
transmission mode. A self-supporting wafer (20 mg cm−2) of
catalyst was placed in a quartz cell with CaF2 windows and
attached to a glass circulation system. The catalyst was reduced
under flowing H2 at 400 °C for 0.5 h, evacuated at the same
temperature for 0.5 h, and cooled to room temperature. After
the pretreatment, a spectrum was recorded as a baseline for the
subsequent measurements. Degassed N-benzylidenebenzyl-
amine (2.0 kPa) was introduced for 0.5 h and then evacuated.
Temperature-programed desorption (TPD) of imine was
subsequently carried out under evacuation at a ramping rate
of 10 °C min−1. A good linearity was obtained between the
heating time and the actual temperature (Figure S1). All spectra
were recorded at a 4 cm−1 resolution.
Synthesis of Deuterated Dibenzylamines. (1) Diben-
zylamine-α,α,α,α-d4 was synthesized by the following proce-
dure. Benzyl bromide-α,α-d2 (5.5 mmol) in dimethylforma-
mide (4 mL) was slowly added dropwise to a vigorously stirred
mixture of benzylamine-α,α-d2 (5.5 mmol), Cs2CO3 (5.5
mmol), and dimethylformamide (6 mL) at 0 °C, which was
subsequently stirred at 0 °C for 2 h. The reaction mixture was
then filtered and the filtrate was concentrated by evaporation of
dimethylformamide. A solution of 1 N NaOH (6 mL) and
toluene (5 mL) was added to the concentrated residue,
followed by extraction using toluene that was repeated four
times. The organic layer was collected, dried by Na2SO4,
concentrated, and distilled by Kugelrohr to afford 337 mg (31%
1
yield) of colorless oil. H NMR: (500 MHz, CD3OD) δ 5.77
(8H, d), 5.70 (2H, m), 3.42 (1H, s).
(2) Dibenzylamine-N-d1 was synthesized by washing
dibenzylamine (2.0 g) with D2O (6 mL) three times, followed
by extraction using toluene, evaporation, and Kugelrohr
distillation.18 A H NMR spectrum indicated >90% deuterium
1
content at the N-position by comparing of the integrated signal
of the amino-proton with that of the benzylic.
Catalyst Preparation. In this study, Pd content was
adjusted to 3.0 wt % for all Pd-containing catalysts. Pd/SiO2
and Pd3Pb/SiO2 were prepared by a pore-filling impregnation.
For Pd/SiO2, aqueous solution of (NH4)2PdCl4 was added to
dried SiO2 with the solution just filling the pores of the silica gel
(1.64 mL ion-exchanged water per g-SiO2). A mixed aqueous
solution of (NH4)2PdCl4 and Pb(NO3)2 with 3:1 molar ratio
was used to obtain Pd3Pb/SiO2. The mixture was kept
overnight at room temperature and then dried on a hot plate
with stirring. The sample was reduced in a quartz tube with
hydrogen flow (50 mL min−1) at 400 °C for 1 h. Pb/SiO2 (Pb:
3 wt %) was prepared by a similar procedure using Pb(NO3)2.
Other supported Pd or Pd3Pb catalysts (Pd/MOx, Pd3Pb/MOx:
MOx = TiO2, Al2O3, and MgO) were prepared by a simple
impregnation. It is difficult to perform pore-filling impregnation
for these metal oxide supports due to the formation of a highly
viscous slurry. A specific amount of aqueous solutions (at least
for dissolving of salts) containing (NH4)2PdCl4 alone or both
(NH4)2PdCl4 and Pb(NO3)2 with a 3:1 molar ratio were added
dropwise to a vigorously stirred aqueous slurry of the oxide
support (15 mL ion-exchanged water per g-support). After a
stirring for 15 min in an air atmosphere, the slurry was
completely dried on a hot plate and ground to a fine powder.
The sample was then reduced in a quartz tube with hydrogen
1215
DOI: 10.1021/cs501695m
ACS Catal. 2015, 5, 1214−1222

ACS Catalysis
Research Article
a
Catalytic Reaction. Catalyst (50 mg) was placed into a 50
mL three necked round-bottom flask equipped with a silicone
ruber septum, a reflux condenser and a gas storage balloon (2
L) and was pretreated under H2 stream (50 mL min−1) at 400
°C for 0.5 h using a mantle heater. After pretreatment, dry Ar
was passed into the flask to replace residual H2 and the flask
was cooled to room temperature. A reaction mixture containing
solvent (p-xylene, 5 mL), amine (0.50 mmol), and an internal
standard (biphenyl) was added into the flask through the
septum. The atmosphere was then replaced with O2(5%)/Ar.
Using the gas balloon provides an excess amount of O2 toward
amine (80 > O2/amine molar ratio). The catalytic reaction was
initiated by immersing the reaction apparatus into a preheated
oil bath. The temperature of the oil bath was controlled to
maintain the actual temperature of the reaction mixture at 110
°C. Products were quantified by flame-ionization detection gas
chromatograph (FID-GC, Shimadzu GC14B equipped with a
TC-17 capillary column) by using biphenyl as an internal
standard. The turnover frequency was determined as the
consumption rate of amine (mmol h−1) per amount of exposed
Pd (mmol) estimated by CO chemisorption. The initial
consumption rates were typically obtained below 30%
conversion. For the kinetic study, the O2 content was adjusted
by confluence of O2 and Ar flows.
Table 1. Analysis for Supported Pd3Pb Catalysts
support
SiO2
400
18
2.3
TiO2
Al2O3
MgO
T
red/°C
XRD/nm
450
14
600
19
600
20
D
DPCO (%)
2.0
2.7
4.0
Intensity Ratio (XRD)
Pd3Pb 111
Pd 111
1.00
0.00
0.95
0.05
0.83
0.17
0.95
0.05
Composition Fraction (XRD)
Pd3Pb
Pd
1.00
0.00
0.97
0.03
0.90
0.10
0.97
0.03
Composition Fraction (EXAFS)
Pd3Pb
Pd
r2 (k < 12)
0.97
0.03
0.97
0.97
0.03
0.98
0.87
0.13
0.92
0.94
0.06
0.98
a
Tred, reduction temperature; DXRD, crystallite size estimated by
Scherrer’s equation; DPCO, dispersion of Pd calculated by amount of
chemisorbed CO assuming 1:1 adsorption.
a single phase. A similar result was also obtained for Pd3Pb/
TiO2 or Pd3Pb/MgO with only a tiny appearance of
monometallic Pd. In the case of Pd3Pb/Al2O3, small diffractions
of monometallic Pd were observed in addition to the main
peaks of the Pd3Pb phase. The composition fractions of the
Pd3Pb and Pd phases were calculated based on the peak
intensity (height) of the 111 diffractions (Table 1) and atomic
scattering factors19 of Pd and Pb (see the Supporting
Information for the detailed derivation). The phase purity of
Pd3Pb on Al2O3 was estimated as 0.90. These results indicate
that a small fraction of Pd did not participate in the formation
of the intermetallic compound and remains in the mono-
metallic phase. In general, the formation of an intermetallic
compound tends to be inhibited by a support that strongly
captures metal species, such as γ-Al2O3. Higher temperatures
are typically required for successful formation of the
intermetallic phase on such supports to increase atom mobility.
Indeed, in the case of Pd3Pb/Al2O3, an increase in the
reduction temperature up to 800 °C enabled the formation of
almost phase-pure Pd3Pb (Figure S2). However, it should be
noted that the increase in the reduction temperature also
caused sintering of the intermetallic nanoparticles. We
performed a similar XRD analysis also for the monometallic
Pd catalysts. Diffractions corresponding to metallic Pd were
observed (Figure S3), of which the estimated crystallite sizes
were 20−27 nm (Table S1). The XRD analysis indicated that
the desired intermetallic Pd3Pb or monometallic Pd nano-
particles were successfully formed on several oxide supports.
However, since small metallic clusters are XRD-unobservable
(typically <2 nm), if any, their contribution to the phase purity
cannot be evaluated by XRD analysis.
RESULTS AND DISCUSSION
■
Characterization of the Prepared Catalyst. The
prepared Pd−Pb catalysts were characterized by XRD as
shown in Figure 1. For all supports, apparent peaks
Figure 1. XRD patterns of supported Pd3Pb catalysts. Numbers in
parentheses indicate crystallite sizes estimated by Scherrer’s equation.
corresponding to 111 and 200 diffractions of the Pd3Pb
phase were observed at 38.6° and 44.9°, respectively (PDF #04-
002-9725). A small peak assigned to 110 diffraction of Pd3Pb,
which does not appear with an fcc structure due to distinction
rule, was also observed at 31.3°. This indicates that the formed
nanoparticles are truly the Pd3Pb intermetallic phase (space
group: Pm3m) but not Pd0.75Pb0.25 alloy (space group: Fm3m).
Therefore, we subsequently performed structural analysis of
the supported catalysts using the XAFS technique. Figure 2a
shows Pd−K edge EXAFS spectra of the supported Pb3Pb
catalysts and the reference samples. Authentic bulk Pd3Pb
showed an oscillation feature quite different from that of Pd
foil, with a shorter period and smaller amplitude. According to
the formula for EXAFS oscillation (eq 1), the shorter period is
due to the longer Pd−Pd(Pb) distance (285 pm) of Pd3Pb than
that of Pd metal (275 pm).
̅
̅
The crystallite sizes of the supported Pd3Pb nanoparticles were
estimated by applying Scherrer’s equation to the most intense
111 diffractions, which revealed similar sizes (14−20 nm, Table
1). The Pd3Pb/SiO2 catalyst showed no diffraction of
monometallic Pd phase, suggesting the formation of Pd3Pb in
1216
DOI: 10.1021/cs501695m
ACS Catal. 2015, 5, 1214−1222

ACS Catalysis
Research Article
oscillations. The EXAFS oscillation of Pd3Pb/Al2O3 was
subsequently fitted by a linear combination of those of Pd3Pb
and Pd foils using a least-square method. A combination of 0.87
Pd3Pb and 0.13 Pd gave the best fit with a 0.92 R2 factor.
Similar fitting was also performed for other supported Pd3Pb
catalysts (Table 1). The estimated values for Pd3Pb fractions
were consistent with the composition from XRD analysis. Pb−
LIII edge EXAFS spectra of the samples were also recorded as
shown in Figure S4. The EXAFS oscillations of the supported
Pd3Pb catalysts were almost identical to that of bulk Pd3Pb and
quite different from the Pb foil. Thus, the information obtained
by the EXAFS study agreed well with the results of XRD
analysis and supported formation of the desired intermetallic
particles with high phase purities.
Figure 3 shows the crystal structure of Pd3Pb (a) in a single
unit cell or (b) with a cuboctahedron shape and (c) the atomic
Figure 2. Pd−K edge EXAFS spectra of supported Pd3Pb and
reference compounds (a) with and (b) without offsets and (c) their
Fourier transforms.
Figure 3. (a) Crystal structure of Pd3Pb with single unit cell (PDF:
#04-002-9725), (b) Pd3Pb nanoparticle with cuboctahedron structure,
and (c) its {100} and {111} facets.
NF(k)exp(−2k2σ 2)
arrangements on {100} and {111} facets. Pd3Pb has an L12
type structure, i.e., the corner atoms in the unit cell of fcc Pd
are replaced by Pb. Assuming a cuboctahedron structure, {100}
and {111} facets are exposed in a 1:1.15 atomic ratio. When the
particle size is 20 nm, based on this model the dispersion of Pd
can be roughly estimated as 4.2%. This value agreed well with
the dispersion measured by CO chemisorption for Pd3Pb/MgO
(4.0%) having a 20 nm crystallite size of Pd3Pb (Table 1).
Pd3Pb/TiO2 exhibited the lowest dispersion despite the size
smaller than other samples. This may be due to so-called strong
metal−support interaction (SMSI), which often results in
coverage of metal particles with TiO2 layer.23,24 CO
chemisorption was also carried out for the monometallic Pd
catalysts. As summarized in Table S1, the obtained dispersions
(3.9−7.9%) were close to those estimated by the cuboctahe-
dron model (4.6−6.0%). Thus, for both Pd3Pb and
monometallic Pd catalysts, good agreements were obtained
between the measured and estimated dispersions. This indicates
the validity of the cuboctahedron model and the absence of
small metal clusters, which will significantly increase the
dispersion of Pd.
j
j
j
2
χ(k) = S0
sin(2krj + φ(k))
j
∑
krj2
j
(1)
where, χ(k) is EXAFS oscillation, Nj is the number of scattering
atoms, rj is the distance to scattering atoms, Fk(k) is the back
scattering factor, σj is the Debye−Waller factor, φj(k) is the
2
phase shift, and S0 is the many body effect. The smaller
amplitude can be attributed to a much larger intrinsic Debye−
Waller factor (thermal vibration) of Pb21 than of Pd.22 The
difference in Pd−Pd(Pb) distance is represented more clearly
in their Fourier transforms (Figure 2c). The EXASF oscillations
and the Fourier transforms of the supported Pd3Pb, except
Pd3Pb/Al2O3, were almost identical to that of bulk Pd3Pb,
which indicates two important points. One is the formation of
Pd3Pb intermetallic compound with high phase purity and the
other is the absence of small Pd3Pb or Pd clusters, because the
presence of small clusters decreases the average coordination
number and hence decreases the oscillation amplitude.
According to the XRD analysis, a small fraction of
monometallic Pd was contained in Pd3Pb/Al2O3. In the
corresponding EXAFS spectrum, only a small contribution of
the monometallic Pd species was observed in the high k region
(k > 8). The contribution of monometallic Pd was also
confirmed in the Fourier transforms; this was indicated by a
slightly shorter Pd−Pd(Pb) distance and smaller amplitude
than those of other supported Pd3Pb. The smaller amplitude
can be attributed to a phase-cancellation by the two EXAFS
We subsequently investigated the electronic states of the
prepared catalysts by XANES and XPS. Figure 4 shows Pd−K
edge XANES spectra of the supported Pd3Pb catalysts and
reference samples. For bulk Pd3Pb, the energy of absorption
edge (E0) and the height of the white line (Hw) corresponding
to the 1s → 5p transition were lower than those of Pd foil,
indicating the electron-richness of Pd atoms in Pd3Pb
compared to those in monometallic Pd. The supported
1217
DOI: 10.1021/cs501695m
ACS Catal. 2015, 5, 1214−1222

ACS Catalysis
Research Article
similar to those of Pd3Pb/SiO2. In the case of Pd3Pb/MgO,
intense peaks owing to Mg KLL Auger processes26 were also
observed with an appearance of the Pd 3d emissions. Peak
deconvolution revealed that the peak positions for the Pd 3d
emissions were almost identical to those of other supported
Pd3Pb samples. An XPS analysis of Pb 4f region was also
carried out. All the supported Pd3Pb samples showed Pb 4f
emission peaks at around 136.6 (4f7/2) and 141.4 (4f5/2) eV,
corresponding to metallic Pb (Figure S5).27−29 The presence of
27−30
cationic lead assignable to PbO2
was also detected,
particularly for Pd3Pb/Al2O3. This is probably due to the
presence of residual Pb, which did not participate in the
intermetallic formation as mentioned above, being in the oxide
phase. Minimal differences in the peak positions of PbO2 were
observed among the samples. On the basis of the electronic
analyses by XANES and XPS, we concluded that the electronic
states of the supported Pd3Pb particles were almost identical
and independent of the nature of the support.
Figure 4. Pd−K edge XANES spectra of supported Pd3Pb and
reference compounds.
Aerobic oxidation of dibenzylamine into N-benzylideneben-
zylamine was carried out over the supported Pd3Pb and Pd
catalysts. Figure 6 shows the time course of amine conversion
Pd3Pb catalysts showed spectral features similar to that of bulk
Pd3Pb. A small deviation of E0 was observed between the bulk
and supported samples, which may be attributed to the
difference in the style, that is, the bulk or supported particles.
The spectral features of the supported catalysts were almost
identical with each other. Only a tiny difference in E0 (closer to
that of Pd foil) was observed for the Al2O3-supported catalyst.
Based on the structural analyses mentioned above, this could be
due to the presence of a small amount of monometallic Pd.
These results suggest that the electronic state of Pd in Pd3Pb is
minimally dependent on the nature of the support. Figure 5
Figure 6. Time course of dibenzylamine conversion and N-
benzylidenebenzylamine selectivity in aerobic oxidation of dibenzyl-
amine over Pd3Pb/MgO and Pd/MgO.
and imine selectivity when Pd3Pb/MgO or Pd/MgO was used
as a catalyst. Pd/MgO showed a very low conversion (8% at 2
h), whereas Pd3Pb/MgO exhibited remarkably high catalytic
activity (100% at 1.5 h). For Pd/MgO, small amounts of
benzaldehyde and benzylamine were formed as fragmented
byproducts, which resulted in 89 C% imine selectivity. The
formation of these byproducts was inhibited over Pd3Pb/MgO,
affording an excellent selectivity (98 C% at 100% conversion).
The turnover number (TON) and turnover frequency (TOF/
h−1) were calculated as the yield (mmol) and formation rate
(mmol h−1) of imine per number of exposed Pd atoms (mmol),
respectively. Pd/MgO gave 16 h−1 of initial TOF and 32 of
TON after 2 h of reaction, indicating that the reaction
catalytically proceeded in spite of the very low conversion.
Much higher initial TOF (2990 h−1) and TON (892, 1.5 h)
were obtained using Pd3Pb/MgO than Pd/MgO. In a similar
fashion, the initial TOF values were estimated for other Pd3Pb
and Pd catalysts. For Pd3Pb catalysts, the order of the obtained
TOF values was 470 (Pd3Pb/SiO2) < 1710 (Pd3Pb/TiO2) <
2120 (Pd3Pb/Al2O3) < 2990 (Pd3Pb/MgO). Since the
electronic states of Pd in these catalysts are almost identical
as mentioned above, it is likely that the nature of the support
itself contributed to the reaction outcome. Figure 7 shows the
Figure 5. Pd 3d XPS of Pd/SiO2 and supported Pd3Pb catalysts.
shows Pd 3d XPS of Pd/SiO2 and the supported Pd3Pb
catalysts. Pd/SiO2 exhibited 3d emission peaks at 335.3 (3d5/2
)
and 340.6 (3d3/2) eV that were consistent with the reported
values of metallic Pd supported on SiO2.25 A small fraction of
the cationic species was also observed. The peak positions of
metallic Pd for Pd3Pb/SiO2 were lower in energy than those of
Pd/SiO2. This result indicates that Pd atoms at the surface of
Pd3Pb supported on SiO2 are electron-rich than those of Pd/
SiO2, which is consistent with the result of XANES. Pd3Pb/
TiO2 and Pd3Pb/Al2O3 showed emission peaks with positions
1218
DOI: 10.1021/cs501695m
ACS Catal. 2015, 5, 1214−1222

ACS Catalysis
Research Article
Figure 7. Relation between TOF in dibenzylamine oxidation and pH
of isoelectric point of catalyst support.
relationship between the obtained TOF and the pH of
support’s isoelectric point, as a scale of its basicity. An obvious
positive correlation with support basicity was observed for
Pd3Pb catalysts. In the previous study, we observed a similar
relationship; however, the catalytic activity increased exponen-
tially with increase in the support basicity, where the Pd
dispersion had not been considered.17 The true relationship
was thus obtained by taking the dispersion into account. For
monometallic Pd catalysts, in contrast, all TOF values were
similar, remaining small and independent of the support
basicity: 21 (Pd/SiO2), 32 (Pd/TiO2), 24 (Pd/Al2O3), and 16
(Pd/MgO). Thus, the positive effect of support basicity was
observed only with Pd3Pb catalysts and not monometallic Pd
catalysts. Emphasis should be placed on the fact that the TOF
was ca. 190 times higher using Pd3Pb/MgO compared to Pd/
MgO. Even for SiO2-supported catalysts, TOF was ca. 20 times
higher for Pd3Pb/SiO2 than for Pd/SiO2.
Figure 8. Reaction rates (r) over Pd3Pb/SiO2 and Pd/SiO2 as
functions of (a) amine concentration ([S]) and (b) oxygen pressure
(PO2). The standard conditions are designated as dotted lines.
Scheme 1. Kinetic Isotope Effects for (a) Amino and (b)
Benzylic Positions of Dibenzylamine
Kinetic studies were subsequently performed to elucidate the
reaction mechanism. We have observed above two prominent
contributions to the catalytic activity, i.e., the Pd3Pb phase and
the support basicity. It is better to consider these contributions
individually. To clarify the role of the Pd3Pb phase without any
involvement of the support basicity, we initially chose silica-
supported catalysts (Pd3Pb/SiO2 and Pd/SiO2) as the targets in
the kinetic study. Figure 8 shows the dependence of the
reaction rates on (a) amine concentration, [S], and (b) the
partial pressure of O2, PO2. Pd3Pb/SiO2 displayed a first-order
relationship with [S] and a zero-order dependence on PO2,
suggesting that adsorption of amine is the rate-determining
step. In contrast, for Pd/SiO2, the reaction orders with [S] and
when Pd3Pb/MgO was used as a catalyst (data not shown),
indicating that dissociative adsorption of amine remained rate-
determining even for Pd3Pb/MgO. In the case of Pd/SiO2, we
focused on C−H activation at the α position of dibenzylamine,
which is mandatory for imine formation. For this purpose,
dibenzylamine-α,α,α,α-d4, of which all α hydrogen atoms were
deuterized, was used as a substrate. However, no KIE was
observed (kH/kD = 1.0, Scheme 1b), revealing that the C−H
activation step is similarly not rate-determining for Pd/SiO2.
Pd/MgO as a catalyst gave a similar result (kH/kD = 0.9),
suggesting no kinetic difference between Pd/MgO and Pd/
SiO2. A similar kinetic study using dibenzylamine-N-d1 was also
performed for Pd/SiO2, revealing a KIE value close to unity
(kH/kD = 1.3). This result indicates that dissociative adsorption
of dibenzylamine is not the rate-determining step for Pd/SiO2,
which is consistent with the zero-order dependence on amine
concentration mentioned above.
On the basis of the obtained results, we propose a plausible
reaction mechanism shown in Scheme 2a. First, dissociative
adsorption of amine occurs over Pd atoms. This process results
in the formation of Pd−H and Pd−amide species (1). C−H
activation at the α position against N, corresponding to the β
position against Pd, takes place to form imine (2), which is so-
called β-H elimination. The formed imine is then desorbed as a
P
O2 were both close to zero, indicating that adsorption of amine
and the O2-involved step are not rate-determining. With the
increase in PO2 over the standard condition, the reaction order
became negative in the case of Pd3Pb. This result suggests that
competitive adsorption of amine and O2 occurs in this pressure
region. In this context, saturation coverage of O2 around the
standard condition is not likely despite the zero-order
dependence. To further understand the reaction mechanism,
we subsequently performed a kinetic isotope study using
deuterized amines as described in Scheme 1. For Pd3Pb/SiO2,
deuteration at the amino position of dibenzylamine (dibenzyl-
amine-N-d1) showed a positive kinetic isotope effect (KIE, kH/
kD = 2.2, Scheme 1a). Combined with the first-order
dependence on [S], it appears that dissociative adsorption of
dibenzylamine (N−H scission) is the rate-determining step for
Pd3Pb/SiO2. A similar KIE value (kH/kD = 2.3) was obtained
1219
DOI: 10.1021/cs501695m
ACS Catal. 2015, 5, 1214−1222

ACS Catalysis
Research Article
contribution of π back-bonding interaction, in addition to
coordination with the lone pair. In the literature, a similar
situation can be observed in alcohol oxidation using Pd
catalysts; the presence of heteroatoms, such as N and S, in
alcoholic substrates fatally deactivates the catalysis probably by
poisoning Pd sites.31 In contrast, for Pd3Pb/SiO2, the situation
is drastically altered; the rate-determining step is shifted to
adsorption (1). Considering the enhanced catalytic activity, this
shift can be attributed to an acceleration of imine desorption.
Thus, the role of the Pd3Pb phase or Pb itself was suggested as
a desorption promoter. In addition to the positive role of the
intermetallic phase, we observed another effect by involving the
support basicity. For Pd3Pb catalysts, it is likely that the basicity
contributes to adsorption of amine as this step is rate-
determining. A possible interpretation is that deprotonation
from amine by neighboring basic sites on the support promotes
the formation of amide species (Scheme 2b). The stronger the
basicity, the more promoted the deprotonation. Indeed, some
of us recently reported that the dissociative adsorption of amine
took place on a metal oxide surface to form metal-amide and a
hydroxyl group.32 This has been demonstrated by an FT-IR
study using benzylamine-N-d2, which reported the formation of
amide species and surface O−D bonds. In the case of TiO2-,
Al2O3-, and MgO-supported Pd3Pb catalysts, in this context,
metal−support interfaces seem to be the most effective reaction
sites. In contrast, for monometallic Pd catalysts, as the reaction
rate is limited by desorption, such a promotion effect in
adsorption was not reflected in the catalytic activity. Thus, in
this catalytic system, a combination of desorption and
adsorption promoters afforded a remarkable enhancement in
the catalytic activity.
Scheme 2. (a) Reaction Mechanism of Amine Oxidation
over Pd3Pb/SiO2 and (b) Deprotonation by a Basic Site on
TiO2, Al2O3, or MgO17
product (3). Finally, the eliminated hydrogen atoms are
consumed by O2 to form H2O (4). In the case of Pd/SiO2,
steps 1, 2, and 4 are not rate-determining as indicated above.
Therefore, it can be said that the desorption of imine (3) is the
rate-determining step for Pd/SiO2. This implies a strong
adsorption capability of imine to Pd, most likely due to a
It is also important to consider the contribution of oxygen to
the oxidation state of surface Pd and Pb. When we performed
the amine oxidation with Pd3Pb/SiO2 without H2 reduction
Figure 9. (a) FT-IR spectra of N-benzylidenebenzylamine adsorpbed on Pd3Pb/SiO2, Pb/SiO2, and Pd/SiO2. (b) Change in FT-IR spectra of imine
adsorbed on Pd3Pb/SiO2 during heating at 10 °C min−1 under evacuation. TPD profiles of imine at (c) 1652 and (d) 1686 cm−1.
1220
DOI: 10.1021/cs501695m
ACS Catal. 2015, 5, 1214−1222

ACS Catalysis
Research Article
pretreatment, the catalytic activity drastically dropped com-
pared to that with the treatment (data not shown). Therefore,
it seems to be important for the catalysis that Pd and Pb are in
metallic state. A recent in situ XAFS study for alcohol oxidation
over Pd−Bi catalysts revealed that both Pd and Bi kept metallic
states during the reaction in the presence of 1 atm O2 and that
after the complete consumption of alcohol Bi was oxidized into
Bi3+ state.15 It appears that alcohol prevents the oxidation of
metallic species probably because O2 reacts preferentially with
the eliminated hydrogen. A similar effect may work in our
system, hence prevent the oxidation of Pd and Pb.
Therefore, the specifically high catalytic activities of Pd3Pb,
Pd13Pb9, and Pd3Bi cannot be explained by the electronic effect.
A possible interpretation alternative to the electronic effect is a
geometric one. Since Pb and Bi atoms have considerably larger
metallic radii than Pd (Pb, 1.75 Å; Bi, 1.70 Å; Pd, 1.37 Å),39
steric repulsion at the Pd site should be greater than that of
monometallic Pd due to the presence of the neighboring Pb or
Bi atom. This situation can be observed with Pd3Pb (111) and
Pd3Pb (100) surfaces as shown in Figure 3. Such a steric
hindrance seems to significantly weaken adsorption of imine,
promoting desorption. Furthermore, this interpretation is
consistent with the result of IR-TPD measurement, which
showed the milder imine-capturing capability of Pd.
To obtain evidence of the promoted desorption on Pd3Pb,
we performed TPD of imine monitored by FT-IR. Figure 9a
shows the FT-IR spectra of N-benzylidenebenzylamine, the
reaction product, chemisorbed on Pd3Pb/SiO2, Pb/SiO2, and
Pd/SiO2. Peaks assigned to CN stretching vibration
appeared intensely in the region of 1650−1700 cm−1. A slight
difference in the peak position was observed between Pd/SiO2
(1652, 1686, and 1695 cm−1) and Pb/SiO2 (1659, 1695, and
1702 cm−1), whereas Pd3Pb/SiO2 exhibited both features.
These assignments were summarized in Table S2. Figure 9b
shows the spectral changes for Pd3Pb/SiO2 during heating at 10
°C min−1 under vacuum. A decrease in the peak intensities was
observed with an increase in temperature, corresponding to
desorption of imine. Similar results were also obtained for Pd/
SiO2 and Pb/SiO2 (Figure S6). To construct a TPD profile for
Pd sites, the changes in peak intensity at 1652 cm−1 (Figure 9c)
and 1686 cm−1 (Figure 9d) were plotted with respect to
temperature for Pd3Pb/SiO2 and Pd/SiO2. In the case of Pd/
SiO2, desorption began at ca. 80 °C. In contrast, for Pd3Pb/
SiO2, desorption occurred near room temperature, providing
strong evidence for promoted desorption on Pd sites of Pd3Pb/
SiO2. In the TPD profiles of Pd3Pb/SiO2, desorption seems to
proceed via a stepwise fashion, that is, 25−80 and 80−160 °C.
This may be due to overlapping of the desorption from Pb sites
as a shoulder. It is clearly evident in Figure 9b that the
significant decrease at 1704 or 1694 cm−1 occurred at a much
higher temperature than that at 1686 cm−1. Therefore, the first
decrease (25−80 °C) in the TPD profile can be assigned to
desorption from Pd sites and the second (80−160 °C) from Pb
sites. A similar situation is observed in Figure S6; desorption
from Pd/SiO2 completed at 130 °C, whereas that from Pb/
SiO2 at 150 °C, indicating that desorption from the Pb sites is
more endothermic than from Pd sites. These results suggest
that Pb atoms themselves do not act as effective desorption
sites but diminish the imine-holding capability of Pd sites.
Finally, we discuss the essential role of Pb, regarding why
desorption is promoted. As mentioned above, the electronic
state of Pd was modified to be electron-rich by the formation of
the Pd3Pb phase. Because imine is coordinated to Pd with a
lone pair as a base, an electron-enriched Pd site seems to
facilitate desorption of imine. In this context, such an electronic
effect looks plausible to explain the enhanced catalytic activity.
However, we should also consider our previous results of a
series of Pd-based intermetallic compounds in terms of the
electronic effect. For example, Pd13Pb9 and Pd3Bi also showed
good catalytic performances comparable to Pd3Pb.17 However,
we previously reported that Pd atoms consisting of Pd13Pb9
were electron-deficient compared to monometallic Pd.33
Moreover, although it has been reported by some researchers
that Pd atoms in PdGa34,35 and PdZn36−38 were negatively
charged, these compounds exhibited much lower catalytic
performances than Pd3Pb (comparable to monometallic Pd).17
CONCLUSION
■
In this study, a mechanistic study based on kinetic and
spectroscopic approaches revealed the detailed catalysis of
aerobic amine oxidation over intermetallic Pd3Pb catalysts and
the role of Pb in enhancing the catalytic activity. Oxidative
dehydrogenation of amine to imine occurs via N−H scission, α-
C−H activation, desorption of imine, and removal of hydrogen
by O2. For a monometallic Pd catalyst, strong coordination of
imine to Pd causes imine desorption to be rate-determining. In
contrast, for Pd3Pb, desorption of imine is promoted by Pb,
which results in the shift of the rate-determining step to
adsorption of amine. Adsorption of amine is accelerated by
deprotonation by the basic site on the support, which further
enhances the catalytic activity. However, in the case of
monometallic Pd, because the reaction rate is limited by
desorption of imine, such a promotion effect of the basicity is
not reflected in the catalytic activity. Thus, in the oxidation of
amine over supported Pd-based intermetallic catalysts, a
combination of the effects of the second metal and the support
basicity constitutes a well-concerted bifunctional catalysis. The
obtained insights in the present study have significant
implications for Pd-based oxidation chemistry.
ASSOCIATED CONTENT
* Supporting Information
The following file is available free of charge on the ACS
■
S
Publications website at DOI: 10.1021/cs501695m.
Derivation of phase composition from peak intensity in
XRD pattern, Tables S1 and S2, and Figures S1−S6
(PDF)
AUTHOR INFORMATION
Corresponding Authors
*E-mail: furukawa.s.af@m.titech.ac.jp. Tel.: +81-3-5734-2602.
Fax: +81-3-5734-2758 (S.F.).
*E-mail: komatsu.t.ad@m.titech.ac.jp. Tel.: +81-3-5734-3532.
Fax: +81-3-5734-2758 (T.K.).
■
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
This work was supported by JSPS KAKENHI Grant No.
26820350. The author appreciates Dr. Hiroyuki Asakura of
Nagoya University and Prof. Dr. Tsunehiro Tanaka of Kyoto
University for the aid of XAFS measurement. The author
thanks Dr. Kosuke Ono of the Tokyo Institute of Technology
for the help of NMR measurement.
1221
DOI: 10.1021/cs501695m
ACS Catal. 2015, 5, 1214−1222

ACS Catalysis
REFERENCES
Research Article
(35) Kovnir, K.; Armbruster, M.; Teschner, D.; Venkov, T. V.;
Jentoft, F. C.; Knop-Gericke, A.; Grin, Y.; Schlogl, R. Sci. Technol. Adv.
Mater. 2007, 8, 420−427.
(36) Armbruster, M.; Behrens, M.; Fottinger, K.; Friedrich, M.;
Gaudry, E.; Matam, S. K.; Sharma, H. R. Catal. Rev. 2013, 55, 289−
367.
(37) Friedrich, M.; Teschner, D.; Knop-Gericke, A.; Armbruster, M.
J. Catal. 2012, 285, 41−47.
(38) Iwasa, N.; Takezawa, N. Top. Catal. 2003, 22, 215−224.
(39) Pauling, L. J. Am. Chem. Soc. 1947, 69, 542−553.
■
(1) Mallat, T.; Baiker, A. Chem. Rev. 2004, 104, 3037−3058.
(2) Mallat, T.; Baiker, A. Catal. Today 1994, 19, 247−283.
(3) Wang, J. R.; Fu, Y.; Zhang, B. B.; Cui, X.; Liu, L.; Guo, Q. X.
Tetrahedron Lett. 2006, 47, 8293−8297.
(4) Choudhary, T. V.; Banerjee, S.; Choudhary, V. R. Appl. Catal. A:
Gen. 2002, 234, 1−23.
(5) Lin, M. R.; Hogan, T.; Sen, A. J. Am. Chem. Soc. 1997, 119,
6048−6053.
(6) Herrero, E.; Fernandezvega, A.; Feliu, J. M.; Aldaz, A. J.
Electroanal. Chem. 1993, 350, 73−88.
(7) Mallat, T.; Baiker, A.; Patscheider, J. Appl. Catal. 1991, 79, 59−
75.
(8) Ocon, P.; Beden, B.; Lamy, C. Electrochim. Acta 1987, 32, 1095−
1101.
(9) Mallat, T.; Bodnar, Z.; Baiker, A.; Greis, O.; Strubig, H.; Reller, A.
J. Catal. 1993, 142, 237−253.
(10) Furuya, N.; Motoo, S. J. Electroanal. Chem. 1979, 98, 189−194.
(11) Smits, P. C. C.; Kuster, B. F. M.; Vanderwiele, K.; Vanderbaan,
H. S. Appl. Catal. 1987, 33, 83−96.
(12) Tsujino, T.; Ohigashi, S.; Sugiyama, S.; Kawashiro, K.; Hayashi,
H. J. Mol. Catal. 1992, 71, 25−35.
(13) Fiege, H.; Wedemeyer, K. Angew. Chem., Int. Ed. 1981, 20, 783−
784.
(14) Parsons, R.; Vandernoot, T. J. Electroanal. Chem. 1988, 257, 9−
45.
(15) Mondelli, C.; Ferri, D.; Grunwaldt, J. D.; Krumeich, F.;
Mangold, S.; Psaro, R.; Baiker, A. J. Catal. 2007, 252, 77−87.
(16) Mondelli, C.; Grunwaldt, J. D.; Ferri, D.; Baiker, A. Phys. Chem.
Chem. Phys. 2010, 12, 5307−5316.
(17) Furukawa, S.; Suga, A.; Komatsu, T. Chem. Commun. 2014, 50,
3277−3280.
(18) Smith, J. K.; Bergbreiter, D. E.; Newcomb, M. J. Org. Chem.
1985, 50, 4549−4553.
(19) Brown, P. J.; Fox, A. G.; Maslen, E. N.; O’Keefec, A.; Willisd, B.
T. M. In International Tables for Crystallography; Prince, E., Ed.; Wiley-
VCH: Weinheim, 2006; Vol. C, p 544−590.
(20) Ravel, B.; Newville, M. J. Synchrotron Radiat. 2005, 12, 537−
541.
(21) Yu, Y. H.; Tyliszczak, T.; Hitchcock, A. P. J. Phys. Chem. Solids
1990, 51, 445−451.
(22) Kochubey, D. I.; Pavlova, S. N.; Novgorodov, B. N.; Kryukova,
G. N.; Sadykov, V. A. J. Catal. 1996, 161, 500−506.
(23) Tauster, S. J.; Fung, S. C.; Baker, R. T. K.; Horsley, J. A. Science
1981, 211, 1121−1125.
(24) Tauster, S. J.; Fung, S. C.; Garten, R. L. J. Am. Chem. Soc. 1978,
100, 170−175.
(25) Huang, L.; Wang, Z.; Ang, T. P.; Tan, J.; Wong, P. K. Catal. Lett.
2006, 112, 219−225.
(26) Claus, P.; Berndt, H.; Mohr, C.; Radnik, J.; Shin, E. J.; Keane, M.
A. J. Catal. 2000, 192, 88−97.
(27) Hirano, T.; Ozawa, Y.; Sekido, T.; Ogino, T.; Miyao, T.; Naito,
S. Appl. Catal. A: Gen. 2007, 320, 91−97.
(28) Hirano, T.; Kazahaya, Y.; Nakamura, A.; Miyao, T.; Naito, S.
Catal. Lett. 2007, 117, 73−78.
(29) Abdel−Samad, H.; Watson, P. R. Appl. Surf. Sci. 1998, 136, 46−
54.
(30) Evarestov, R. A.; Veryazov, V. A. Phys. Stat. Sol. B 1991, 165,
411−418.
(31) Kakiuchi, N.; Maeda, Y.; Nishimura, T.; Uemura, S. J. Org.
Chem. 2001, 66, 6620−6625.
(32) Furukawa, S.; Ohno, Y.; Shishido, T.; Teramura, K.; Tanaka, T.
J. Phys. Chem. C 2013, 117, 442−450.
(33) Furukawa, S.; Yoshida, Y.; Komatsu, T. Acs Catal. 2014, 4,
1441−1450.
(34) Kovnir, K.; Armbruster, M.; Teschner, D.; Venkov, T. V.;
Szentmiklosi, L.; Jentoft, F. C.; Knop-Gericke, A.; Grin, Y.; Schlogl, R.
Surf. Sci. 2009, 603, 1784−1792.
1222
DOI: 10.1021/cs501695m
ACS Catal. 2015, 5, 1214−1222