Pt promotional effects on Pd-Pt alloy catalysts for hydrogen peroxide synthesis directly from hydrogen and oxygen

Article abstract of DOI:10.1016/j.jcat.2011.09.017

H₂O₂ synthesis directly from H₂ and O ₂ over supported Pd-Pt alloy catalysts was carried out using a semibatch reactor under ambient conditions. As compared to pure Pd, the performance of Pd-Pt catalysts was enhanced significantly. The promotional role of Pt was studied systematically by using in situ diffuse reflectance infrared Fourier transform spectroscopy of CO adsorption (DRIFTS), quantitative powder X-ray diffraction (XRD), X-rays photoelectron spectroscopy (XPS), and temperature-programmed desorption of H₂/O₂ (H ₂/O₂-TPD). The spectra of DRIFT, XPS, and XRD demonstrate the formation of Pd-Pt alloy particles, which surfaces are enriched by Pt accompanying with possible electron transfer from Pd to Pt. The addition of Pt into Pd phase was proposed to impact on reactants adsorption, stabilization of intermediates such as OOH and OH radicals, and the formation and decomposition of H₂O₂.

Full text of DOI:10.1016/j.jcat.2011.09.017

Journal of Catalysis 285 (2012) 74–82
Contents lists available at SciVerse ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Pt promotional effects on Pd–Pt alloy catalysts for hydrogen peroxide
synthesis directly from hydrogen and oxygen
⇑
Jing Xu, Like Ouyang, Guo-Jin Da, Qian-Qian Song, Xue-Jing Yang, Yi-Fan Han
State Key Laboratory of Chemical Engineering, East China University of Science and Technology, Shanghai 200237, People’s Republic of China
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 13 June 2011
Revised 13 September 2011
Accepted 14 September 2011
Available online 22 October 2011
H O synthesis directly from H2 and O2 over supported Pd–Pt alloy catalysts was carried out using a
2
2
semibatch reactor under ambient conditions. As compared to pure Pd, the performance of Pd–Pt catalysts
was enhanced significantly. The promotional role of Pt was studied systematically by using in situ diffuse
reflectance infrared Fourier transform spectroscopy of CO adsorption (DRIFTS), quantitative powder X-ray
diffraction (XRD), X-rays photoelectron spectroscopy (XPS), and temperature-programmed desorption of
2 2 2 2
H /O (H /O -TPD). The spectra of DRIFT, XPS, and XRD demonstrate the formation of Pd–Pt alloy particles,
Keywords:
Pd–Pt alloy
Hydrogen peroxide synthesis
XPS
which surfaces are enriched by Pt accompanying with possible electron transfer from Pd to Pt. The addition
of Pt into Pd phase was proposed to impact on reactants adsorption, stabilization of intermediates such as
ꢀ
ꢀ
OOH and OH radicals, and the formation and decomposition of H
2 2
O .
Ó 2011 Elsevier Inc. All rights reserved.
XRD
In situ DRIFT
O -TPD
2
H
2
-TPD
1
. Introduction
[16–18]. One of our studies has also demonstrated the promotional
effects of Au on SiO supported Pd–Au alloy catalysts using a con-
2
The demand for hydrogen peroxide (H
during the last years [1,2]. H O , as a ‘‘green oxidant,’’ is produced
2 2
2
O
2
) has been increasing
tinuous reactor system under ambient conditions [13]. However,
up to now, the promotional effects of Pt on Pd–Pt alloy catalysts
have rarely been reported for this reaction, even though the perfor-
mances of those catalysts are close to that of Pd–Au alloy catalysts.
The deep insights into the mechanism are still unavailable because
the whole reaction involves three phases: solid catalysts, gaseous
reagents, and liquid medium [4,19–23]. In addition, Pd–Pt alloys,
as an interesting catalytic system, have also shown excellent per-
formance for other reactions such as oxygen reduction, hydrogena-
tion of aromatic hydrocarbons in fuel, and methane combustion
[10,24–29].
In this study, supported Pd–Pt alloy catalysts with different Pd/
Pt ratios were prepared by conventional incipient wetness impreg-
nation. Fumed silica was used as the support, which is chemically
inert and does not interact electronically with Pd and Pt, thus will
simplify the study by ruling out the potential strong metal–support
mainly by a circuitous process involving the redox of alkylanthro-
quinone and hydroquinone intermediates [3]. However, there are
several drawbacks in the current process, including slow degrada-
tion of expensive anthraquinone, gradually deactivation toward
hydrogenation, and high capital costs [3]. For environmental and
economic concerns, there is a renewed interest for replacing this
process with a direct process, by which H
rectly on suitable catalysts.
Up to date, the direct process operated under ambient condi-
tions has been studied for several decades [2,4]. For practical appli-
cation, it is imperative to develop novel catalysts that can enhance
2 2
and O are reacted di-
2 2
both the yield and the selectivity toward H O [5–7]. Unfortu-
nately, most of the reported catalysts fail to meet this criterion
owing to the complexity of this reaction system, except for sup-
ported Pd or Pd-alloyed catalysts with a second metal such as
Au, Ag, and Pt. In particular, supported Pd–Au alloy catalysts have
proved excellent performance for this reaction [8–15]. The Hutch-
ings’ group has reported the performance of Pd–Au nanoalloys sup-
interactions. H
semibatch reactor under ambient conditions, which, in principle,
will greatly reduce the risk of explosion that is caused by H –O
2 2
O synthesis was performed by using a tri-phase
2
2
mixtures in the explosive regime. The catalyst structure was char-
acterized using multi-techniques such as quantitative powder
X-ray diffraction (XRD), X-ray photoelectron spectroscopy (XPS),
in situ diffuse reflectance infrared Fourier transform spectroscopy
of CO adsorption (DRIFTS), and temperature-programmed desorp-
ported on different metal oxides such as TiO
acidified carbon using a batch reactor system (275 K, 4.0–9.0 MPa)
2 2 3 2
, Fe O , and ZrO , and
⇑
Corresponding author. Fax: +86 21 64251928.
E-mail address: yifanhan@ecust.edu.cn (Y.-F. Han).
2 2 2 2
tion of H and O (H - and O -TPD).
0
021-9517/$ - see front matter Ó 2011 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2011.09.017

J. Xu et al. / Journal of Catalysis 285 (2012) 74–82
75
The modifications of the Pd surface upon alloying with Pt and
the structure–reactivity relationship of Pd–Pt alloy catalysts for
ched from the catalyst surface to liquid as analyzed by an Induc-
tively Coupled Plasma (ICP) analyzer.
the formation and decomposition of H
the basis of those results. This work aims at the understanding of
synthesis on Pd alloy catalysts and rational design of Pd–Pt
or other bimetallic or/and alloy catalysts intended for this or other
applications.
2 2
O were rationalized on
2
2
.2. Characterization
2 2
H O
.2.1. XRD
Powder X-ray diffraction patterns were accumulated with a
Bruker D8 diffractometer using Cu K(a) radiation with a step size
of 0.02° two-theta (2h) over the range 30–120°, while intensive
scanning was implemented in the range 30–50° and 80–84°,
2
. Experimental
2
respectively. Approximately 10 wt.% of standard CaF (NIST 640a,
2
.1. Catalyst preparation and reactivity measurements
a = 5.4633 Å) was mixed with catalysts to allow the precise deter-
mination of lattice constants. The whole pattern was analyzed
using the Rietveld method as implemented in TOPAS-V2 with Pd
and Pt [31] employed as crystallographic starting models. The lat-
All catalysts were prepared by conventional incipient wetness
2ꢀ
ꢀ
impregnation using aqueous solution of PdCl4 and PtCl , which
followed a reported method with a minor modification [30]. Palla-
dium and Platinum were introduced into the system as Pd of
6
2
tice constant of the CaF standard was fixed, and the zero shift and
scale factor were optimized. The isothermal vibration parameters
3
.3 wt.% with varying the Pt content in a range 0.32–3.2 wt.%, cor-
2
of all phases were fixed at 1.0 Å .
responding to Pd/Pt molar ratios from 20 to 2. Pd–Pt alloys were
supported on fumed silica having a Brunauer–Emmett–Teller
2
ꢀ1
2.2.2. XPS
(
BET) surface area of 230 m g . Prior to the reaction, the precur-
XPS analysis was performed on a VG ESCALAB 250 spectrome-
sors were pretreated in O at 673 K and then reduced in H at 573 K
with a flow rate of 20 mL min for 30 min. For comparison, refer-
ences such as Pd–Au alloy (Au sources from AuCl , Pd/Au = 4 (mo-
lar ratio)) [13], pure Pd, and pure Pt catalysts were also prepared
following the same procedure.
All reactions were carried out under ambient conditions using a
micro-tri-phase semibatch reactor, which is similar to the appara-
tus given in Ref. [5]. The reagent gases were introduced into the
reactor via a fine glass frit, and the slurry containing the catalyst
was stirred so as to minimize diffusion limitations. The reactor
was connected to a gas chromatograph (Shimadzu GC-2014), so
2
2
ꢀ
1
ter, using Al Ka radiation (1486.6 eV, pass energy 20.0 eV). The
ꢀ
9
ꢀ
base pressure of the instrument is about 1 ꢁ 10 Torr. The back-
ground contribution B (E) (obtained by the Shirley method) caused
by inelastic process was subtracted, while the curve fitting was
performed with a Gaussian–Lorentzian profile by a standard soft-
ware. The binding energies (BEs) over the supported catalysts were
calibrated by using the Si2p peak at 103.5 eV and C1s peak at
4
2
85.0 eV as references. The instrument was also calibrated by
using Au wire (Au 4f7/2 at 84.0 eV). XPS spectra were recorded at
h = 90° of X-ray sources. The Pd/Pt molar ratios were calculated
through Eq. (2) [32]:
that the concentration of H
cally determined. Improved accuracy in H
by using a H /N mixture that contained 10% N
of H was analyzed colorimetrically using a UV–vis spectropho-
tometer (Epp2000, StellarNet Inc.) after complexation with a TiOS-
/H SO reagent. H selectivity, SH2O2 , was determined from the
rate of H formation and the rate of H conversion using Eq. (1)
2
exiting the reactor could be periodi-
analysis was achieved
. The concentration
2
N
N
¼ I
Pd=SPd
Pt=SPt
Pd
ð2Þ
2
2
2
Pt
I
2 2
O
where the IPd and IPt are the time-normalized intensities of the Pd3d
and Pt4f levels, and SPd (4.642 for Pd3d) and SPt (4.674 for Pt4f) are
the atomic sensitivity factors for X-ray sources at 90°.
O
4
2
4
2 2
O
2
O
2
2
Rate of H
2
O
2
formation ðmole=minÞ
2
.2.3. TEM
S
H2O2
¼
ꢁ 100
ð1Þ
Rate of H
2
conversion ðmole=minÞ
/H
0 mL/min was introduced into the reaction system at 283 K if
The measurements were performed on a Tecnai TF 20 S-twin
with Lorentz lens. The catalyst was first ultrasonically suspended
in ethanol, and then, one drop of this slurry was deposited on a car-
bon-coated copper grid. The liquid phase was evaporated before
the grid was loaded into the microscope. The metal particle size
was estimated on the basis of 300 particles.
In the reaction, a 4:1 O
2
2
gas mixture with a flow rate of
5
without any special statement. The liquid phase composed of
0 mL of ethanol or water that had been acidified with 10 mL of
9
aqueous HCl solution to give 100 mL of liquid having the desired
acid normality. Thus, the ‘‘ethanol solution’’ contained 10 mL of
water in addition to the small amount of water that was formed
during the reaction. The solutions were 0.12 M in HCl unless stated
otherwise. Following each experiment, residual Pd that had depos-
ited on the frit was removed by filling the reactor with a 1.0 M HCl
solution and passing pure O
orange color solution, resulting from PdCl , was observed. In addi-
tion, in order to investigate the Pt effects on H
the experiments were implemented under the same reaction con-
ditions by fixing the initial H concentration of 1.0 wt.% with a
2.2.4. In situ DRIFT spectra of CO adsorption
2
CO adsorption on the fresh catalysts (pretreated in O at 673 K
ꢀ1
2
and H at 573 K with a flow rate of 20 mL min for 30 min) was
conducted in a reaction cell (modified Harricks model HV-DR2)
in order to allow gas flowing continuously through the catalyst
bed (ca. 0.1 g) during spectra acquisition. The spectra were re-
2
through it. In most cases, a yellow-
2ꢀ
4
ꢀ1
corded on a PerkinElmer FT-IR spectrum 100 (resolution: 4 cm
)
2 2
O decomposition,
spectrometer at room temperature. All infrared data were evalu-
ated in Kubelka–Munk units, which are linearly related to the
absorber concentration in spectral. Contributions from gas-phase
CO were eliminated by subtracting the corresponding spectra from
the pure support material.
2 2
O
ꢀ1
H
2
flow rate of 15 mL min
Since the 4:1 O /H gas mixture is still in the explosive regime,
.
2
2
care must be taken to avoid contact of the gas mixture with a dry
catalyst. This was achieved first by mixing 50 mg of the catalyst
with 10 mL of the solution, and then, the slurry was added back
into the solution that remained in the reactor. It should be noted
that with ethanol as the liquid phase, the catalyst remained wet,
even in the upper regions of the reactor. Expect for a trace amount
of Pd (3 ppm in solution), almost no Pt and Au elements were lea-
2 2
2.2.5. H - and O -TPD
The TPD experiments were performed with a micro-fixed-bed
reactor (quartz reactor with 20 cm long and 0.4 cm diameter) con-
nected to a GC–QMS (HPR-20, Hiden Analytical Ltd.), where masses
2 4 2 2 2 2
of m/e 2(H ), 15(CH ), 18(H O), 28(N ), 32(O ), 44(CO ) were

7
6
J. Xu et al. / Journal of Catalysis 285 (2012) 74–82
monitored. Prior to adsorption, the catalyst was pretreated in O
2
at
High-resolution TEM images (Fig. 2a–c) show that the size dis-
tributions are almost unchanged for fresh pure Pd, Pd20Pt , and
Pd16Pt , an average diameter of 3.5 ± 0.2 nm can be estimated. A
typical size distribution curve for Pd16Pt was outlined in Fig. 2c.
However, the rapid growth in particle size could be observed with
further increasing the Pt content (Table 1), the average diameter of
ꢀ
1
6
73 K and reduced in H
2
at 573 K with a flow rate of 20 mL min
. Adsorbents of H or O
1
for 30 min, then cooling down in pure N
2
2
2
1
ꢀ
1
(
20 mL min , 30 min) were introduced into the system at 283 K.
1
ꢀ
1
Then, the system was purged with N
2
(50 mL min , 10 min). The
ꢀ1
temperature ramped from 283 to 853 K with a rate of 20 K min
ꢀ
1
in N
2
(50 mL min ).
5.0 ± 0.3 nm for Pd
and 30.0 ± 0.5 nm for Pd
a high-resolution TEM image for Pd16Pt
8
Pt
1
(Fig. 2d), 10.0 ± 0.2 nm for Pd
Pt (Fig. 2f) were determined. In addition,
(Fig. 2c) showed one-
4 1
Pt (Fig. 2e),
2
1
1
3
. Results
The crystalline phases of Pd–Pt alloys were analyzed by XRD
dimensional lattice fringes of the Pd(111) lattice plane. Mean-
while, spherical particles were found for the samples from pure
Pd to Pd
Pd Pt and Pd
The chemical states of the surface Pd and Pt atoms were ana-
Pt
8 1
, but irregularly shaped particles were observed for
(
(
4
Fig. 1). For all fresh samples, reflections attributable to h111i
4
1
2 1
Pt .
2h = 40.1° (Pd) and 39.7° (Pt)) and h200i (2h = 46.6° (Pd) and
6.2° (Pt)) face-centered cubic structures were presented with
lyzed using XPS. The core levels of Pd 3d5/2 at 335. 1 eV (Fig. 3A)
and Pt 4f7/2 at 71.0 eV (Fig. 3B) attributing to metallic states re-
mained unchanged upon alloying. It seems difficult to determine
the electronic interaction between Pd and Pt even though charge
transfer of 0.1 e from Pd to Pt in Pd–Pt alloy due to the relative high
electron affinity of Pt was predicted by theoretic studies [33–35].
In addition, the Pd/Pt atomic ratios determined by XPS (Table 1)
indicated that all alloy surfaces were enriched by Pt atoms because
of Pt relative large electronegativity compared to Pd (the Pauling
electronegativities of Pt and Pd are 2.28 and 2.20, respectively)
their positions adjusted according to composition (Fig. 1A). The
formation of single phase alloys was identified from Pd20Pt to
Pd Pt 0.1°. With continu-
while the peak h111i shifted down by
ously increasing the Pt content to Pd Pt , a two-phase alloy was
1
8
1
D
4
1
produced as evidenced by the h111i plane at 39.7° (Pt-rich alloy)
and 40.1° (Pd-rich alloy) that were really differential by lattice con-
stants (Table 1). The alloy structures were further demonstrated by
the downshift of the reflection h311i from 82.2° (pure Pd) to 82.0°
(
Pd20Pt
1
) (Fig. 1B). On the other hand, the reflection Pt h311i was
observed clearly for Pd
Pt
4 1
, Pd
2
Pt
1
, and pure Pt, indicating the for-
[
33].
Fig. 4 shows in situ DRIFT spectra recorded with the adsorption
of CO on pure metals and Pt–Pd alloy catalysts. Linear and bridging
mation of a separated Pt-rich alloy in those samples.
ꢀ1
species with maxima at 2096, 1994, and 1966 cm were observed
for pure Pd, attributing to CO adsorbed on metallic Pd atoms
A
0
(
Pd –CO) [36,37]. Upon alloying with Pt,
a
linear band at
Pd (111)
ꢀ
1
2
111 cm appeared for Pd20Pt
1
1
and Pd16Pt , probably correspond-
ing to linear Pd–CO complexes that lead to a remarkable loss in d-
orbital electron density due to charge transfer from Pd to Pt [38]
and became the dominant band for Pd Pt , Pd Pt , and Pd Pt .
8 1 4 1 2 1
Pd (200)
Pt(200)
Pt(111)
Meanwhile, the bridging bands decreased drastically and disap-
peared completely for the last three samples. The changes in vibra-
tion indicated that the Pd surface structure was significantly
modified by Pt atoms. The same phenomenon was observed for a
bimetallic Pt–Pd/NaY zeolite catalyst [38]. Moreover, the CO–Pt
a
b
c
d
ꢀ1
linear band (2062 cm ) [39] observed for pure Pt disappeared
after alloying with Pd.
e
2
The H -TPD spectra (Fig. 5) show two peaks centered at 392 and
6
81 K for pure Pd and Pd–Pt alloy samples. The first peak is close to
the one at 340 K observed for a polycrystalline Pd, associating with
the multiple states of hydrogen on the Pd surface [40]. The second
peak may result from the decomposition of the surface OH groups,
38
40
42
44
46
48
50
2
θ/degree
which was also identified for a Pt/
desorption amount of H decreased from 106 (pure Pd) to
mol g (Table 1), and the second peak shifted down
2 3
c-Al O catalyst [37]. The
B
2
Pd(311)
ꢀ
1
4ðPd
2
Pt
1
Þ
l
cat
Pt (311)
to 651 K; meanwhile, the first peak decreased gradually with
increasing the Pt content and disappeared completely for Pd
The desorption amount of O
2
Pt
1
.
a
b
c
2
decreased from 60 (pure Pd) to
ꢀ1
2
l
mol g (Pd
2 1
Pt ) (Table 1), this trend is similar to that for H
2
-
cat
TPD. Three peaks at 394, 496, and 653 K were detected for pure
Pd but the last one diminished upon alloying with Pt (Fig. 6). The
first two peaks may be attributed to the weakly adsorbed oxygen
at different Pd sites, and the last peak probably originated from
the decomposition of 2D palladium oxide, which has proven a pre-
cursor in the formation of PdO [41]. Presumably, the Pd–O bond
strength was weakened by neighboring Pt atoms, thus led to the
disappearance of the peak at 653 K.
d
e
As listed in Table 1, the activity order for all catalysts can be dis-
8
0.0
80.5
81.0
81.5
82.0
82.5
83.0
83.5
84.0
played as Pd16Pt
Pure Pd showed a middle activity, while pure Pt has no activity
toward H synthesis but complete oxidization of H to H O. This
fact is evident that the formation of H occurs primarily on Pd
1
> Pd20Pt
1
> Pd
8
Pt
1
> Pd > Pd
Pt
4 1
> Pd
2
Pt
1
ꢂ Pt.
2
θ/degree
2
O
2
2
2
Fig. 1. (A) XRD diffraction patterns of Pd–Pt alloy catalysts in the 2h range 35–50°;
B) in the 2h range 80–84°. (a) Pd20Pt , (b) Pd16Pt , (c) Pd Pt , (d) Pd Pt , (e) Pd Pt
(
1
1
8
1
4
1
2
1
.
2 2
O

J. Xu et al. / Journal of Catalysis 285 (2012) 74–82
77
Table 1
The structural information and the performance of fresh SiO
2
supported Pd and Pd–Pt/Au alloy catalysts in the H
2
O
2
synthesis directly from H
2
and O
.a
2
Catalystsb Particle
sizec (nm)
Surface composition
(mole ratio Pd/Pt)
Cell constant ae (Å) Desorption amount of
Desorption amount of
mol/gcat
Reaction rate
H
(wt.%)
O
Selectivity
(%)
2
2
H
2
(
lmol/gcat
)
O
2
(l
)
(mol/h gPd
)
Pd-rich
phase
Pt-rich
phase
Pure Pd
3.8
3.5
3.6
–
3.8896
3.8901
3.8917
3.8935
3.8948
3.8966
3.9811
–
106
42
22
7
6
4
60
21
19
11
9
2
–
–
0.99
1.64
1.77
1.62
0.61
0.12
ꢄ0
0.34
0.56
0.60
0.55
0.21
0.06
–
12
70
63
52
31
19
–
Pd20Pt
Pd16Pt
1
16/1
12/1
6/1
2.5/1
1/1
–
–
–
1
Pd
8
Pd
4
Pd
2
Pt
1
Pt
1
Pt
1
5.0
3.9035
3.9187
3.9211
10.0
30.0
8.0
Pure Pt
–
–
d
Pd
4
Au
1
6.0
3.6/1
–
1.70
0.59
52
a
All experiments were carried out in the solution of HCl (0.12 M)-ethanol by a tri-phase semibatch reactor at 10 °C, atmospheric pressure, and 5-h reaction. A trace amount
of Pd and no Pt were leached from the silica surface to liquid as analyzed by ICP.
b
The 3.3 wt.% of Pd was fixed for Pd alone and Pd–Pt/Au alloy catalysts, the subscripts are the molar ratios of Pd/Pt/Au in the bulk.
The average particle size was obtained by TEM.
Data taken from Ref. [13].
Cell constants were obtained from XRD results analyzed using the Rietveld method.
c
d
e
sites. Comparing to the pure Pd catalyst, the catalyst performance
was significantly improved upon alloying with the appropriate
atoms should be randomly mixed because of its relatively low en-
ꢀ1
thalpy, for instance, only ꢀ4 kJ mol for the formation of Pd0.5Pt0.5
[42]. However, theoretical studies [43–45] have evidenced that a
strong segregation of Pd atoms may take place easily in the
(100) and (111) surfaces of Pd–Pt alloys because of the lower sur-
face energy of Pd and the greater cohesive energy of Pt. It has been
further pointed out that for non-stoichiometric Pt–Pd clusters,
even doping a single Pt atom into a Pd cluster (or vice versa) is suf-
ficient to change the geometrical structure of the cluster [43]. The
geometric structure and disorder degree for Pd–Pt nanoalloys were
found to vary with the compositions of alloys [44].
amount of Pt. In particular, the rate for H
2
O
2
formation was
ꢀ1
ꢀ1
ꢀ1 ꢀ1
0
.99 mol h
for Pd Pt
dropped down to 0:12 mol h
g
for pure Pd, and increased to 1:62 mol h
g
Pd
Pd
ꢀ
1
ꢀ1
8
1
surpassed the maxima 1:77 mol h
g
Pd
1
for Pd16Pt , then
ꢀ
1
ꢀ1
g
2 1
for Pd Pt . At the same time, the
Pd
H
2
O
2
selectivity increased from 12% for pure Pd to 70% for Pd20Pt
and then went down to 19% for Pd Pt . Nevertheless, the selectivity
of 70% for Pd20Pt is slightly lower than that (80%) obtained for a
Pd/C catalyst under the similar reaction conditions as recently re-
1
2
1
1
ported by Liu et al. [4]. It is noted that the performance for Pd20Pt
is also superior to Pd Au
catalyst we have ever prepared.
In addition, H decomposition, as a reverse reaction of H
formation, was investigated with a solution of 1.0 wt.% H
flow. As shown in Fig. 7, the activity order was displayed as
1
4
1
(Table 1) that was the best Pd-Au alloy
In this study, the formation of alloys in the broad range of Pd/Pt
ratios was demonstrated by the XRD patterns (Fig. 1). However, the
Pd and Pt atoms in the bulks of alloys were not mixed homoge-
nously because Pd-rich and Pt-rich mixed phases were detected
2
O
2
2
O
2
2 2
O in a
H
2
4 1 2 1
for both Pd Pt and Pd Pt samples; meanwhile, the lattice con-
Pt ꢂ Pd
2
Pt
1
> Pd
4
Pt
1
> Pd
8
Pt
1
> Pd16Pt
1
ꢃ Pd20Pt
1
ꢃ Pd, being con-
stants (Table 1) varying between 3.9811 Å (pure Pt) and 3.8896 Å
(pure Pd) reflect a contraction of Pt–Pt bond and an increase in
the Pd–Pd distance in alloys [35]. On the other hand, the alloy sur-
faces were enriched by Pt as evidenced by XPS analysis (Fig. 3 and
Table 1), this is contrary to the results reported by other groups
trary to that for H
2
O
2
synthesis.
4
. Discussion
[
43,44]. A possible reason is that the effect of d-band filling may
In line with previous studies [1,2], three competitive reactions
2] depicted as a triangular network (Scheme 1) were proposed
to be mainly responsible for the production of H (reaction I)
and H O (reaction II, III) in the direct reaction system. In addition,
this reaction has been reported to be affected by several factors,
such as halide ions, acid concentration, reaction medium, and cat-
alyst property [1,2,5]. In order not to diverse the focus, herein, we
mainly discuss the structure–activity relationship of this reaction
system, the role of Pt, and the elementary steps involving the pro-
drive a segregation of the more electronegative Pt toward the sur-
faces [33].
The DRIFT spectra of CO adsorption (Fig. 4) showed an upshift of
the linear CO–Pd band for Pd–Pt alloys, which is probably due to
electron transfer from Pd to Pt. More interestingly, upon alloying
[
2 2
O
2
ꢀ1
with Pd, the linear CO–Pt band (2062 cm ) that could be observed
for pure Pt disappeared completely accompanying with a signifi-
cant decrease in the intensity of the bridge CO–Pd bands. In com-
bination of all results, herein, two structure models for Pd–Pt
nanoalloys [46] envisaged were depicted in Scheme 2. One is a
Pdshell–Ptcore model ((a) in Scheme 2), in which most of Pt atoms
are located inside the bulk of nanoparticles. It may explain well
that no linear CO–Pt band can be observed for alloy samples; how-
ever, it is contradictory with the fact of the Pt enrichment in alloy
surfaces as detected by XPS. The other is a random model ((b) in
Scheme 2), in which Pd and Pt are randomly distributed, and both
atoms in the alloy surfaces are separated because of strong interac-
tions. Likewise, it is understandable that the intensities for the
bridge CO–Pd bands (Fig. 4) decrease drastically upon alloying with
Pt, because the Pd atoms in alloy surfaces are highly diluted in
comparison with that for pure Pd. The same reason may be respon-
sible for the disappearance of the linear CO–Pt band on alloys, even
2 2 2
duction of H O and H O.
4.1. Structure of Pd–Pt alloys
The relationships of structure–activity of Pd–Pt alloy catalysts
have been studied for several decades because of its important
applications in electrochemistry, hydrogenation, and oxygen
reduction reactions [10,24–29]. Generally, upon alloying with Pt,
the Pd surface structure is modified through two ways: (i) the Pd
ensembles in the surface are finely tuned by the change of the
Pd–Pd bond length and (ii) the electronic density in the Pd valence
band is changed by neighboring Pt atoms. In principle, Pd–Pt alloys
are composed of continuously solid solutions, in which Pd and Pt

7
8
J. Xu et al. / Journal of Catalysis 285 (2012) 74–82
(a)
(b)
2
0 nm
20 nm
(c)
(d)
0.23nm
2
0 nm
2
.5 3.0 3.5 4.0 4.5
D/nm
(e)
(f)
2
0 nm
1 1 8 1 4 1 2 1
Fig. 2. TEM images of the fresh reduced catalysts. (a) Pure Pd, (b) Pd20Pt , (c) Pd16Pt , (d) Pd Pt , (e) Pd Pt , (f) Pd Pt .
though the Pt concentrations in the alloy surfaces are proved to be
higher than those in the bulks. Owing to structure changes in nano-
on Pd–Pt alloy catalysts altered significantly compared to that on
pure Pd. Elementary steps of this reaction were profoundly im-
alloys, the capability of H
2
and O
2
adsorption on alloys was de-
pacted during reaction, including the adsorption of H
2 2
and O ,
creased drastically with increase in Pt concentration (Figs. 5 and 6).
the formation and desorption of H , and the decomposition of
2 2
O
2 2
H O .
4.2. Plausible mechanism
In the present reaction system, as illustrated in Scheme 1, side
reactions such as H over-oxidation and H decomposition occur
2
2 2
O
On the basis of the presented results, together with the previous
studies [1,2,10], we assume that the mechanism for H O synthesis
2 2
simultaneously. Several elementary steps have been proposed for
those reactions [10], as expressed by Eqs. (3)–(13):

J. Xu et al. / Journal of Catalysis 285 (2012) 74–82
79
0
0
.08
.07
(A)
340.2
335.1
a
b
0.06
a
b
c
d
e
0
.05
f
0.04
C
d
350
345
340
335
330
0.03
Binding Energy (eV)
(
B)
0.02
2056
74.1
e
b
71.0
0.01
f
g
c
2200
2100
2000
1900
-1
Wavenumber (cm )
Fig. 4. In situ DRIFTS of CO (2.0% in He) adsorption. (a) Pure Pd, (b) Pd20Pt
1
, (c)
1
Pd16Pt , (d) Pd
Pt
8 1
, (e) Pd Pt
4 1
, (f) Pd
2
1
Pt , (g) pure Pt.
d
e
681
f
3
92
80
78
76
74
72
70
68
Binding Energy (eV)
Fig. 3. (A) XPS Pd3d spectroscopy and (B) XPS Pt4f spectroscopy over the fresh
a
b
c
reduced catalysts. (a) Pure Pd, (b) Pd20Pt
Pd Pt
1
, (c) Pd16Pt
1
, (d) Pd
8
1 4
Pt , (e) Pd Pt
1
, (f)
2
1
.
H
2
ðgasÞ ! H2ad
ðgasÞ ! O2ad
2ad ! 2H
ð3Þ
ð4Þ
ð5Þ
ð6Þ
ð7Þ
ð8Þ
ð9Þ
ð10Þ
ð11Þ
ð12Þ
ð13Þ
O
H
O
O
2
d
e
2ad ! 2O
2ad þ H ! OOHad
f
O þ H ! OHad
OOHad þ H ! H
2 2
O
300
400
500
600
700
OOHad þ H
2
! H
2
O
2
þ H
Temperature (K)
OHad þ H ! H
OHad þ H ! H
! H ðgasÞ
2
O
2
2
O þ H
Fig. 5. H
2 1 1 8 1 4 1
-TPD spectra. (a) Pure Pd, (b) Pd20Pt , (c) Pd16Pt , (d) Pd Pt , (e) Pd Pt , (f)
H
2
O
2
2
O
2
Pd Pt
2
1
.

8
0
J. Xu et al. / Journal of Catalysis 285 (2012) 74–82
Pd
Pt
6
53
Pd
Pd
4
96
Pd
Pt
Pd
Pt
Pd
3
94
Pd
Pt
Pd
Pd
Pt
Pd
Pd
Pd
Pd
Pt
Pt
Pd
Pd
a
Pt
Pd
Pt
Pd
Pt
Pt
Pt
Pd
Pd
Pd
Pd
Pt
Pd
Pt
Pd
Pd
b
(
a)
(b)
c
Scheme 2. Cross-section of the Pd–Pt nanoparticle models: (a) Pdshell–Ptcore; (b)
random model [44].
d
e
Gaseous O
Eqs. (3) and (4)) and then dissociated into atomic O and H imme-
diately (Eqs. (5) and (6)). It has been demonstrated that molecular
and H can dissociate readily at 120 K [47] and 40 K [48] on the
Pd(111) surface. Furthermore, parts of O molecules that were
2 2
and H molecules were first adsorbed on active sites
(
O
2
2
2
probably weakly bonded to the Pd atoms were reacted with H to
produce OOHad (Eq. (7)), while the O and H atoms reacted with
each other to form OHad (Eq. (8)). OOHad and OHꢀ are believed
ad
to be the primary intermediates involved into reactions for the
2 2 2 2 2
production of H O and H O, respectively. In particular, H O was
probably generated through the interaction between OOHad and
atomic (Eq. (9)) or molecular hydrogen on the Pd surface (Eq.
300
400
500
600
700
800
900
Temperature (K)
(
10)). However, we argue the reaction of Eq. (10) unlikely occurs
because of the low diffusion rate of H from gas phase to the cata-
lyst surface [2]. On the other hand, OHad also reacted with atomic
Eq. (11)) and molecular (Eq. (12)) hydrogen to generate H O. Fi-
nally, H was desorbed from metal surfaces (Eq. (13)). It is wor-
thy to note that other pathways for H
formation such as OHꢀ
radical disproportionation suggested by Olivera et al. [10] are not
displayed here because those reactions are kinetically unfavorable.
The Lunsford’s group has proposed that the colloidal palladium
Fig. 6. O
2
1
-TPD spectra. (a) Pure Pd, (b) Pd20Pt , (c) Pd16Pt
1
, (d) Pd
8 1
Pt , (e) Pd
2 1
Pt .
2
(
2
2 2
O
1
0
0
0
0
0
.0
.8
.6
.4
.2
.0
2 2
O
derived either from Pd/SiO
2
or from PdCl
2
via the reduction in
2ꢀ
PdCl ions in aqueous solutions acidified by HCl may work as ac-
4
tive sites for H
2
O
2
synthesis [2,4,7,21,22]. However, in the present
system, the Pd concentration in the working solution was below
ppm after reaction of 5 h, suggesting that only a trace amount of
2
+
3
Pd was leached during reaction. Therefore, we assume that H
produced mainly on the solid catalyst surface.
2 2
O is
4.3. The Pt role
0
20
40
60
80
100
120
140
160
Up to date, little experimental evidence has been found for the
Reaction time (min)
proposed reaction mechanisms because of technical limitations.
ꢀ
Fig. 7. Decomposition of H
N), Pd Pt Pt (ꢁ), Pd
.0 wt.% H
5 mL/min.
2
O
2
over all catalysts, pure Pd (j), Pd20Pt
1
(d), Pd16Pt
ðJÞ, pure Pt ðIÞ. Reaction conditions: initial
flow rate of
1
OOH radical is believed to be the primary intermediate species
(
1
1
8
1
ð.Þ, Pd
4
1
2
Pt
1
in the generation of H
2 2
O (Eqs. (9) and (10)) [2,49]. Dissanayake
2
O
2
in ethanol (90 mL) + water (10 mL), 0.12 N HCl, 283 K, H
2
and Lunsford have investigated the dissociative adsorption of oxy-
1
6
gen on a Pd/SiO
and
2
catalyst during reaction using a mixture of
O
2
1
8
16 18
2 2
O [21,22]. H O O, as a major product, should be detected
whether a dissociated form of oxygen were involved. However, as
H
O
2 2
16
evidenced by the Raman spectra, only H
2 2
O with a peak at
ꢀ
1
18
ꢀ1
I
8
79 cm and H
2 2
O with a peak at 830 cm were present as
products. Meanwhile, no significant peak was detected at about
852 cm , which corresponds to H O O. Clearly, H O is derived
2 2 2
only from a diatomic form of oxygen that is presumably adsorbed
on the Pd sites.
The addition of a small amount of Pt into Pd phase (from Pd to
1 2 2
Pd16Pt ) could significantly enhance H O synthesis (Table 1). It
ꢀ
1
16 18
H
2
+ O
2
III
II
H
2
O
can be explained that the dissociative adsorption of oxygen on
the alloy surfaces (Eq. (6)) may be suppressed by the weakening
Scheme 1. Reaction network in the synthesis of H
2
O
2
: (I) formation of H
2 2
O ; (II)
formation of H O by H oxidation; (III) formation of H
2
2
2
O by H decomposition.
2 2
O

J. Xu et al. / Journal of Catalysis 285 (2012) 74–82
81
of Pd–O bonds, which will lead to an increase in the concentration
decline of the adsorption capability to H
by the spectra of O -TPD and H -TPD.
We speculate that the tuning of Pd electronic structure by the
addition of a small amount of Pt may stabilize dioxygen on the
2
and O
2
, as evidenced
ꢀ
of OOH radicals (Eq. (7)) and the reaction rate of Eq. (9). In short,
2
2
ꢀ
the formation of OH radicals, which is the only intermediate
2
responsible for H O synthesis, was greatly reduced. Obviously,
ꢀ
the Pt role in Pd–Pt alloy catalysts is different from the Au role in
Pd–Au alloy catalysts, in which synergetic effects of Au and Pd
on the adsorption and transfer of oxygen were suggested [13].
The excess Pt atoms in alloys (Pd/Pt < 8) led to a drastic de-
Pd sites, which is the precursor for the formation of OOH radicals
that react with atomic and molecular hydrogen to form H
2
O .
2
ꢀ
However, excess Pt may destabilize OOH radicals and decompose
2 2 2
H O into H O.
2 2
crease in the production of H O (Table 1). This fact may be ex-
plained with the following reasons: (i) the high coverage of Pt on
the Pd surface blocks the adsorption of reactants on the Pd sites,
Acknowledgments
as evidenced by the O
the kinetics of oxygen dissociation may be favored on the higher
Pt content catalysts, and theoretical studies [10,36] revealed the
bond strength of oxygenated species such as OOH radical on the
Pt sites was stronger than that on the Pd sites, while O and OH
2
/H
2
-TPD experiments (Figs. 5 and 6); (ii)
The authors are grateful to the support from the Chinese Educa-
tion Ministry 111 Project (B08021), the National Science Founda-
tion (21176071, 21106041), Shanghai PuJiang Talent Program
(2010/10PJ1402500, 2011/11PJ1402400), Innovation Program of
Shanghai Municipal Education Commission (11ZZ52), Shanghai
Natural Science Foundation (11ZR1408400), Opening Project
Program of State Key Laboratory of Chemical Engineering (SKL-
ChE-10C05), Creative Team Development Project of Ministry of
Education (IRT0721), and Fundamental Research Funds for the
Central Universities.
ꢀ
ꢀ
could be reduced to H
ciate on the Pt sites. Actually, H
catalysts during reaction (Fig. 7) as depicted in Eqs. (14) and (15)
2]. The rate for H decomposition was almost unaffected by Pt
from pure Pd to Pd16Pt , but enhanced significantly with continu-
ous increase in the Pt content. The excess Pt atoms may facilitate
O formation because of reasons (ii) and (iii). It is also under-
2
O on the Pt sites; (iii) H
2 2
O tends to disso-
2 2
O
is prone to decompose on all
[
2 2
O
1
H
2
References
standable that pure Pt is incapable of generating H
2
O
2
but H
2
O.
[
1] J.M. Campos-Martin, G. Blanco-Brieva, J.L.G. Fierro, Angew. Chem. Int. Ed. 45
2006) 6962.
(
H
H
2
O
O
2
! H
þ H
2
O þ 1=2O
! 2H
2
ð14Þ
ð15Þ
[
[
2] J.H. Lunsford, J. Catal. 216 (2003) 455.
3] W.T. Hess, in: J.I. Kroschwitz, M. Howe-Grant (Eds.), Kirk–Othmer
Encyclopedia of Chemical Technology, fourth ed., vol. 13, Wiley, New York,
2
2
2
2
O
1995, p. 961.
A common point is that the rate for H
dent on the surface hydrogen coverage, which is affected by H
fusion from liquid to solid catalyst surfaces [2]. It seems to be
contradictory to the phenomenon observed from the H -TPD spec-
tra, which has proven the adsorption amount of H is irrelevant
with H formation. As listed in Table 1, an increase in Pt concen-
tration in alloys caused a remarkable decrease in the capability for
adsorption but enhancing the activity. For example, the desorp-
tion amount of H dropped drastically from 106 (pure Pd) to 7
Pd Pt mol/gcat corresponding to the rate of 0.99 and
.62 mol/h gPd, respectively. This fact suggests that the surface
hydrogen coverage is not a decisive factor for H formation.
2
O
2
formation is depen-
[
[
4] Q. Liu, J.C. Bauer, R.E. Schaak, J.H. Lunsford, Angew. Chem. Int. Ed. 47 (2008) 1.
5] Y.-F. Han, J.H. Lunsford, J. Catal. 230 (2005) 313.
2
dif-
[6] V.R. Choudhary, P. Jana, J. Catal. 246 (2007) 434.
[
[
7] Y.-F. Han, J.H. Lunsford, Catal. Lett. 99 (2005) 13.
8] R.E. Schaak, A.K. Sra, B.M. Leonard, R.E. Cable, J.C. Bauer, Y.-F. Han, J. Means, W.
Teizer, Y. Vasquez, E.S. Funck, J. Am. Chem. Soc. 127 (2005) 3506.
2
2
2
O
2
[9] Q. Chen, E.J. Beckman, Green Chem. 9 (2007) 802.
[
[
10] P.P. Olivera, E.M. Patrito, H. Sellers, Surf. Sci. 313 (1994) 25.
11] S. Abate, G. Centi, S. Perathoner, S. Melada, F. Pinna, G. Strukul, Topics Catal. 38
H
2
(
2006) 181.
2
[12] S. Melada, F. Pinna, G. Strukul, S. Perathoner, G. Centi, J. Catal. 237 (2006) 213.
[13] Y.-F. Han, Z. Zhong, K. Ramesh, F. Chen, L. Chen, T. White, Q. Tay, S.N. Yaakub, Z.
Wang, J. Phys. Chem. C 111 (2007) 8410.
(
8
1
) l
1
[
14] L.W. Gosser, Jo-Ann T. Schwartz, US Patent 4832938, to E. I. Du Pont de
Nemours and Company, 1989.
2 2
O
Furthermore, the surface structures of the catalysts may be
modified further during reaction because of adsorption and reac-
tion-induced reconstructions. The reasons for the Pt promotional
effects could be more complex than we have speculated here,
and several questions about the mechanism have still no clear-
cut answers. For instance, how is the surface local structure chan-
ged owing to the addition of Pt? Whether the charge transfer be-
tween Pd and Pt impacts on the adsorption and stabilization of
[15] Q. Liu, J.C. Bauer, R.E. Schaak, J.H. Lunsford, Appl. Catal. A: Gen. 339 (2008) 130.
[
[
[
[
16] J.K. Edwards, B.E. Solsona, P. Landon, A.F. Carley, A. Herzing, C.J. Kiely, G.J.
Hutchings, J. Catal. 236 (2005) 69.
17] P. Landon, P.J. Collier, A.F. Carley, D. Chadwick, A.J. Papworth, A. Burrows, C.J.
Kiely, G.J. Hutchings, Phys. Chem. Chem. Phys. 5 (2003) 1917.
18] J.K. Edwards, B. Solsona, E. Ntainjua N, A.F. Carley, A.A. Herzing, C.J. Kiely, G.J.
Hutchings, Science 323 (2009) 1037.
19] V.V. Krishnan, A.G. Dokoutchaev, M.E. Thompson, J. Catal. 196 (2000) 366.
[20] T.A. Pospelova, N.I. Kobozev, E.N. Ermin, Rus. J. Phys. Chem. (Trans) 35 (1961)
43.
1
[
[
21] D.P. Dissanayake, J.H. Lunsford, J. Catal. 206 (2002) 173.
22] D.P. Dissanayake, J.H. Lunsford, J. Catal. 214 (2003) 113.
the adsorbed molecular O
2
and its subsequent reactions with
hydrogen. The relevant study is still ongoing in our laboratory.
[23] C. Burato, S. Campestrini, Y.-F. Han, P. Canton, P. Centomo, P. Canu, B. Corain,
Appl. Catal. A. 358 (2009) 224.
[
[
24] K. Persson, K. Jansson, S.G. Järås, J. Catal. 245 (2007) 401.
25] R. Strobel, J.-D. Grunwaldt, A. Camenzind, S.E. Pratsinis, A. Baiker, Catal. Lett.
104 (2005) 9.
26] P.B. Balbuena, S.R. Calvo, E.J. Lamas, P.F. Salazar, J.M. Seminario, J. Phys. Chem.
B 110 (2006) 17452.
5
. Conclusions
synthesis could be improved greatly over the Pd catalysts
[
[
2 2
H O
27] J.R. Chang, S.L. Chang, T.B. Lin, J. Catal. 169 (1997) 338.
upon alloying with Pt. Among all catalysts, Pd16Pt
performance with a rate of 1:77 mol h
6
cess Pt (Pd/Pt < 8) in alloys could remarkably deteriorate the cata-
lytic performance.
As evidenced by the XRD results, alloys were created in the
whole Pd/Pt ratio range, and two separated alloy phases (Pd-rich
1
showed the best
and a selectivity of
[28] A. Stanislaus, B.H. Cooper, Catal. Rev. Sci. Eng. 36 (1994) 75.
29] R. Ferrando, J. Jellinek, R.L. Johnston, Chem. Rev. 108 (2008) 845.
[30] Y.-F. Han, D. Kumar, D.W. Goodman, J. Catal. 230 (2005) 353.
31] H.W. King, F.D. Manchester, J. Phys. F 8 (1978) 15.
ꢀ1
ꢀ1
[
g
Pd
ꢀ1
ꢀ1
0%, while 0:99 mol h
g
and only 12% observed for pure Pd. Ex-
Pd
[
[32] J.F. Moulder, W.F. Stickle, P.E. Sobol, K.D. Bomben, in: J. Chastain, C. King Jr.
(Eds.), Handbook of X-ray Photoelectron Spectroscopy, Physical Electronics,
Inc., 1995. ISBN: 0-9648124-1-X.
[
33] A. Dianat, J. Zimmermann, N. Seriani, M. Bobeth, W. Pompe, L.C. Ciacchi, Surf.
Sci. 602 (2008) 876.
and Pt-rich Pd–Pt alloys) were observed for Pd
4
Pt
1
and Pd
2
Pt
1
,
[34] R.R. Squires, J. Am. Chem. Soc. 107 (1985) 4385.
[
[
35] S.R. Calvo, P.B. Balbuena, Surf. Sci. 601 (2007) 4786.
36] M. Lepage, T. Visser, Ad M.J. van der Eerden, F. Soulimani, B.M. Weckhuysen,
Vib. Spectrosc. 48 (2008) 92.
respectively. In situ DRIFTS of CO adsorption and XPS results have
proved that the Pd–Pt alloy surfaces were enriched by Pt. The
changes in the surface structures may be responsible for the
[37] O. Alexeev, D.-W. Kim, G.W. Graham, M. Shelef, B.C. Gates, J. Catal. 185 (1999) 170.

8
2
J. Xu et al. / Journal of Catalysis 285 (2012) 74–82
[
38] T. Rades, V.Yu. Borovkov, V.B. Kazansky, M. Polisset-Thfoin, J. Fraissard, J. Phys.
Chem. 100 (1996) 16238.
39] Y. Wang, N. Toshima, J. Phys. Chem. B 101 (1997) 5301.
40] N.V. Hieu, J.H. Craig Jr., Surf. Sci. 150 (1985) L93.
41] H.H. Kan, R.B. Shumbera, J.F. Weaver, Surf. Sci. 602 (2008) 1337.
42] F.R. De Boer, R. Boom, W.C.M. Mattens, A.R. Miedama, A.K. Niessen, Cohesion in
Metals: Transition Metal Alloys, Elsevier, Amsterdam, 1988.
43] C. Massen, T.V. Mortimer-Jones, R.L. Johnston, J. Chem. Soc., Dalton Trans. 23
[44] L.D. Lloyd, R.L. Johnston, S. Salhi, N.T. Wilson, J. Mater. Chem. 14 (2004) 1691.
[45] D.J. Watson, G.A. Attard, Surf. Sci. 515 (2002) 87.
[46] N. Toshima, M. Harada, T. Yonezawa, K. Kushihashi, K. Asakura, J. Phys. Chem.
95 (1991) 7448.
[47] M.K. Rose, A. Borg, J.C. Dunphy, T. Mitsui, D.F. Ogletree, M. Salmeron, Surf. Sci.
561 (2004) 69.
[
[
[
[
[48] H. Conrad, G. Ertl, E.E. Latta, Surf. Sci. 41 (1974) 435.
[49] T.A. Pospelova, N.I. Kobozev, Rus. J. Phys. Chem. (Trans.) 35 (1961) 262.
[
(
2002) 4375.