Synthesis of palladium phosphides for aqueous phase hydrodechlorination: Kinetic study and deactivation resistance

Article abstract of DOI:10.1016/j.jcat.2018.07.040

We report the synthesis of noble metal phosphides by a temperature-programmed reduction method for aqueous phase hydrodechlorination (HDC) of 4-chlorophenol (4-CP). Pd₃P_0.95/SiO₂, Rh₂P/SiO₂ and PtP₂/SiO₂ catalysts were prepared and showed higher 4-CP HDC rates than their corresponding noble metal catalysts. Pd₃P_0.95/SiO₂ sample exhibited a higher HDC rate than the other phosphide catalysts and Pd/SiO₂ catalysts. The XPS spectra show the surface interaction reactions of the Pd and P species that occur via the electron transfer from Pd to P. Pd₃P_0.95/SiO₂ was synthesized with different phosphide particle size (1.1–13.5 nm based on hydrogen chemisorptions) via changing the metal loading from 1 to 10 wt%. The 4-CP HDC reaction is sensitive to the size of the Pd and Pd₃P_0.95 particles, and the optimum sizes for Pd/SiO₂ and Pd₃P_0.95/SiO₂ were ～8 and ～5 nm, respectively. The effects of the initial 4-CP and Cl^? ion concentration and the S to Pd_surf (surface active Pd metal sites determined by hydrogen chemisorptions) ratio on the 4-CP HDC activity were examined for both catalysts. The Langmuir-Hinshelwood kinetic model that was developed considering the surface reaction as the rate-determining step suggests P has an enhancing effect on sorption in the 4-CP HDC reaction and an inhibiting effect on chloride sorption for Pd₃P_0.95/SiO₂. Moreover, Pd₃P_0.95/SiO₂ possesses better sulfide resistance than Pd/SiO₂ because P can prevent the conversion of Pd metallic sites into inactive Pd₄S compounds with a sulfide treatment. As a result, Pd₃P_0.95/SiO₂ shows better HDC stability than those of Pd/SiO₂ catalysts.

Full text of DOI:10.1016/j.jcat.2018.07.040

Journal of Catalysis 366 (2018) 80–90
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Synthesis of palladium phosphides for aqueous phase
hydrodechlorination: Kinetic study and deactivation resistance
Zhijie Wu a, , Tao Pan , Yan Chai , Shaohui Ge , Yana Ju , Tianshu Li , Kai Liu , Ling Lan ,
⇑
a
b
c
c
c
c
c
d,
⇑
b,
⇑
Alex C.K. Yip , Minghui Zhang
a
State Key Laboratory of Heavy Oil Processing and the Key Laboratory of Catalysis of CNPC, China University of Petroleum, Beijing 102249, China
Key Laboratory of Advanced Energy Materials Chemistry (Ministry of Education), College of Chemistry, Nankai University, Tianjin 300071, China
Petrochemical Research Institute, PetroChina Company Limited, E318, A42 Block, China Petroleum Innovation Base, Changping, Beijing 100195, China
Department of Chemical and Process Engineering, University of Canterbury, Christchurch, New Zealand
b
c
d
a r t i c l e i n f o
a b s t r a c t
Article history:
We report the synthesis of noble metal phosphides by a temperature-programmed reduction method for
Received 7 February 2018
Revised 29 June 2018
Accepted 31 July 2018
aqueous phase hydrodechlorination (HDC) of 4-chlorophenol (4-CP). Pd
PtP /SiO catalysts were prepared and showed higher 4-CP HDC rates than their corresponding noble
metal catalysts. Pd 0.95/SiO sample exhibited a higher HDC rate than the other phosphide catalysts
and Pd/SiO catalysts. The XPS spectra show the surface interaction reactions of the Pd and P species that
occur via the electron transfer from Pd to P. Pd 0.95/SiO was synthesized with different phosphide
particle size (1.1–13.5 nm based on hydrogen chemisorptions) via changing the metal loading from 1
3 2 2 2
P0.95/SiO , Rh P/SiO and
2
2
3
P
2
2
3
P
2
Keywords:
Hydrodechlorination
Metal phosphide
to 10 wt%. The 4-CP HDC reaction is sensitive to the size of the Pd and Pd
sizes for Pd/SiO and Pd 0.95/SiO
ion concentration and the S to Pdsurf (surface active Pd metal sites determined by hydrogen chemisorp-
tions) ratio on the 4-CP HDC activity were examined for both catalysts. The Langmuir-Hinshelwood
kinetic model that was developed considering the surface reaction as the rate-determining step suggests
P has an enhancing effect on sorption in the 4-CP HDC reaction and an inhibiting effect on chloride sorp-
3
P
0.95 particles, and the optimum
ꢁ
2
3
P
2
were ꢀ8 and ꢀ5 nm, respectively. The effects of the initial 4-CP and Cl
4
-Chlorophenol
Noble metal
Catalyst
tion for Pd
can prevent the conversion of Pd metallic sites into inactive Pd
a result, Pd 0.95/SiO shows better HDC stability than those of Pd/SiO
Ó 2018 Elsevier Inc. All rights reserved.
3
P
0.95/SiO
2
. Moreover, Pd
3
P0.95/SiO
2
possesses better sulfide resistance than Pd/SiO
S compounds with a sulfide treatment. As
catalysts.
2
because P
4
3
P
2
2
1
. Introduction
The removal of organochlorinated pollutants is a pressing prob-
kinetic analysis of HDC in aqueous solution has been mainly lim-
ited to 4-CP (4-chlorophenol) since it is relatively soluble (ca.
ꢁ1
ꢁ1
27 g L (0.21 mol L ) in water at 293 K) compared with its iso-
mers [5,7–10]. In order to develop a more effective catalyst and
to build a kinetic database for industrial process, it is important
to advance the fundamental understanding of catalytic HDC of
chlorophenols.
Beside Raney Ni [11] catalysts, noble metal catalysts, especially
Pd [12–25], Pt [6,26] and Rh [27–29] metal catalysts and Pd-based
bimetallic catalysts [30–33], have been reported to be good cata-
lysts for the aqueous phase CP HDC reaction under mild conditions.
For example, Yuan et al. [14] and Roy et al. [16] achieved the HDC
of CPs over Pd-based catalysts at ambient temperatures (273–303
K) under atmospheric pressure. The main products from the HDC
reaction of monochlorophenol over Pd-based catalysts are typically
phenol (e.g., >90% selectivity), cyclohexanone, and trace amount of
cyclohexanol (<0.05% selectivity) [14,34]. The most commonly
lem in water treatment because of their toxicity and tightened
environmental policies [1,2]. Among all chlorinated organic com-
ꢁ1
pounds, chlorophenols (CPs, typically 50–500 mg L in wastewa-
ter that contains phenols) require effective remediation due to
their high toxicity, persistency, low biodegradability [3–5] and
their present in both surface and ground water. Catalytic
hydrodechlorination (HDC) reaction has been investigated in the
past decade and is found to be very effective in eliminating aque-
ous chlorinated pollutants, such as chlorophenols, using hydrogen
over a wide range of concentrations at ambient or near-ambient
temperatures [6,7]. Because of the low solubility of CPs in water,
⇑
Corresponding authors.
E-mail addresses: zhijiewu@cup.edu.cn (Z. Wu), alex.yip@canterbury.ac.nz
(
A.C.K. Yip), zhangmh@nankai.edu.cn (M. Zhang).
https://doi.org/10.1016/j.jcat.2018.07.040
021-9517/Ó 2018 Elsevier Inc. All rights reserved.
0

Z. Wu et al. / Journal of Catalysis 366 (2018) 80–90
81
used support materials are alumina and activated carbon (AC)
14,16]. While the turnover frequency (TOF) of HDC of 4-CP over
2. Experimental section
[
ꢁ1
ꢁ1
Pd/C catalyst ranging from 0.12 to 0.17 mol4-CP molactive-Pd
s
2.1. Chemicals
[
0
19,34–36], it can be further increased to between 0.43 and
ꢁ1
ꢁ1
.7 mol4-CP molactive-Pd
s
by modifying the carbon support itself
NH H PO (ꢂ99%), NH ꢃH O (25–28%), NaCl (ꢂ99.5%), NaOH
4
2
4
3
2
or introducing polymers to the catalyst matrix.
(ꢂ96%) and Na S (ꢂ98.0%) were purchased from Guanfu Chemical,
2
Factors that influence activity and selectivity, such as the rela-
tionships between the size of the metallic phase clusters and the
HDC performance, have been studied [20,37]. The HDC of 4-CP is
a structural sensitive reaction [38] that the activity-and-size corre-
lation revealed a volcano plot with Rh metal clusters being the
optimum. Among all tested metals, Pd is the least susceptible to
catalytic poisoning due to chloride ions. It is also considered as
the most suitable active phase catalyst for the liquid phase HDC
reactions [7,12–25,39], especially for the system in which carbon
materials are used as the support. The carbon supported catalysts
exhibit high HDC activities due to promotion of reactive spillover
hydrogen [9] or enhancement of the sorption of CPs [40]. A critical
issue associated with the liquid phase HDC is that substantial cat-
alyst deactivation can easily happen due to metal leaching caused
by the HCl by-product. Nevertheless, due to the weak acidity of
CPs, the addition of base can help increase the solubility and limit
the HCl poisoning [8]. In the study, a kinetic analysis based on the
Langmuir–Hinshelwood mechanisms is performed. The effect of
the size of the active phase on the HDC and the applicability of
the kinetic model are carefully assessed.
Ltd. (Tianjin, China). PdCl (60% Pd basis), H PtCl (ꢂ99.8%),
2
2
6
RhCl ꢃxH O (39% Rh), 4-chlorophenol (ꢂ99%), AgNO (99.8%), phe-
3
2
3
nol (ꢂ99%), cyclohexanone (ꢂ99.5%) and cyclohexanol (ꢂ98%)
were obtained from Sinopharm Chemical Reagent Co., Ltd.
(Beijing). All the reagents were used as purchased without any fur-
ther purification.
Silica gel (SiO , 99%) was obtained from the China National
2
Offshore Oil Corp. (CNOOC) Tianjin Chemical Research and Design
Institute (Tianjin, China). For the catalyst preparation, SiO2 was
dried at 393 K in an oven for 6 h and then calcined in air at
823 K for 4 h.
2
.2. Catalyst synthesis
2 2 3
The SiO -supported noble metal phosphides (PtP , Pd P0.95, and
Rh
2
P) were successfully prepared by a temperature-programmed
reduction (TPR) method [44]. Ammonium dihydrogen phosphate
NH PO ) was used as the phosphorus source. PdCl and
RhCl O were dissolved in an ammonia solution to create a
ꢃxH
homogenous clear solution. A typical TPR method procedure is
described as follows. First, PdCl , RhCl O and H PtCl solutions
(
4
H
2
4
2
3
2
The rise of sulfur utilization in various chemical industries in
the last decade not only means that the sulfur present in water
and wastewater is becoming more significant, but the form of
2
3
ꢃxH
2
2
6
with various concentrations were prepared to obtain supported
catalysts with 1, 2, 3, 5 and 10 wt% noble metal loadings. Then,
ꢁ
2ꢁ
2ꢁ
2ꢁ
3
sulfur also varies from HS , S , SO
4
, SO
and to other sulfur-
4 2 4
the noble metal salt solution was mixed with the NH H PO solu-
containing anions. On the other hand, the widespread water
contamination with chlorinated solvents, pesticides and sulfur
has spurred intense effort to develop efficient and cost-effective
treatment methods for chlorinated compounds, especially with
co-contamination of sulfur. The HDC reactions over metal catalysts
are regarded as a potential route to remove chlorinated com-
pounds. However, the metal catalysts are often suffered from deac-
tion. To obtain pure phase of phosphide, various ratios of P to noble
metal in raw materials were screened and an atomic ratio of P to
noble metal of 2 (for Pd), 5 (for Pt) and 4/3 (for Rh) has been
achieved. SiO
solution weight ratio of 1/20, and the mixture was stirred
400 rpm) for 4 h and subsequently treated at 318 K overnight
2
was then added to the solutions with at a solid to
(
ꢁ
ꢁ
under rotation in a rotary evaporator to remove the water. The
solid samples were treated in ambient air at 373 K for 6 h and then
tivation due to chloride (Cl ) and hydrosulfide (SH or ‘‘sulfide” in
short) ions [30,41,42]. Therefore, the strong poisoning effects asso-
ciated with traces of sulfur compounds need to be addressed in
order to develop noble metal catalysts with long-term durability.
For example, Wong et al. developed Pd/Au nanoparticles for HDC
reaction of trichloroethene in aqueous phase by promoting the
effect of gold on the catalyst resistance to sulfur poisoning at room
temperature [30].
ꢁ
1
heated in air at 673 K (0.067 K s ) for 3 h. To obtain the phos-
phides, the calcined samples were treated in pure hydrogen gas
3
ꢁ1 ꢁ1
ꢁ1
(
1 cm g
s ) from room temperature to 523 K (0.067 K s ),
3
ꢁ1 ꢁ1
and then, they were treated in pure hydrogen gas (1 cm g
at 773 K (0.033 K s ) for 5 h. The samples were passivated under
s
)
ꢁ1
3
ꢁ1 ꢁ1
0
.5 vol%
exposure.
The SiO
procedures similar to those of the metal phosphides without the
introduction of NH PO . To obtain the metallic active sites, the
samples were directly reduced in pure hydrogen gas (1 cm g
at 673 K (0.033 K s
under 0.5 vol% O
exposure.
O
2
/Ar (0.5 cm g
s ) for 1 h at 300 K before air
In light of the excellent performance of metal phosphides in
hydrotreating reactions, such as hydrodesulfurization (HDS)
2
-supported noble metal catalysts were prepared with
[
43–49], metal phosphides should be suitable catalysts for the
4
H
2
4
HDC reaction and have a sulfur resistance. In the past 10 years,
transition-metal phosphides have been used in the gas phase
HDC reaction at high temperatures because their activities are
lower than those of noble metal catalysts at low reaction temper-
ature [50–52]. Little work has been devoted to the catalytic perfor-
mance of noble metal phosphide catalysts in the aqueous phase
HDC reaction. Here, we firstly screened noble metal phosphides
3
ꢁ1 ꢁ1
s
)
ꢁ
1
)
for 4 h. The samples were passivated
3
ꢁ1 ꢁ1
2
/Ar (0.5 cm g
s ) for 1 h at 300 K before air
To obtain supported noble metal catalyst samples with different
metal particle sizes, the precursors of the metal catalysts were
3
ꢁ1 ꢁ1
treated in flowing dry air (1.0 cm g
s
) by increasing the tem-
perature to 673–873 K at 0.05 K s and holding the samples there
h to prepare samples with a broad range of metal fraction disper-
sion (0.07–0.97, Table 1 for Pd/SiO samples with 5 wt% theory
chemisorptions as described below.
(
platinum, rhodium and palladium phosphides) for the aqueous
ꢁ
1
phase HDC reaction of 4-CP. We studied the effects of sulfide and
chloride ions on the 4-CP HDC reaction rates of the palladium
phosphide catalyst and observed that chloride and sulfide had
slight effects that were markedly less than those observed with
the Pd catalyst. To clarify the chloride and sulfide resistances of
the noble metal phosphide in the aqueous phase HDC reaction,
reaction kinetic models based on the Langmuir-Hinshelwood
5
2
metal loading) measured by H
2
2.3. Catalyst characterization
(
L-H) model have been extensively applied to investigate the
The powder X-ray diffraction (XRD) patterns of the obtained
samples were collected on a Bruke AXS D8 Advance X-ray
catalytic process.

8
2
Z. Wu et al. / Journal of Catalysis 366 (2018) 80–90
Table 1
Metal loading, dispersion, and initial HDC turnover rates of SiO
2
-supported metal and metal phosphide catalysts.
Samples
Entry
Metal loading (wt%)
H
2
chemisorptions of metal
Initial HDC turnover rate
ꢁ
4
ꢁ1
ꢁ1
)c
Actuala
Particle size (nm)b
(10
mol molsurf-metal
s
Theoretical
Dispersion
Pd/SiO
2
1
2
3
4
5
6
7
8
9
1
1
1
1
14
1
16
1
18
1
20
2
1
2
3
5
5
5
5
1
2
3
5
7
10
3
5
3
5
3
5
3
5
1.1
1.9
1.9
5.1
5.1
5.1
5.1
0.9
2.1
2.9
4.9
7.2
10.2
3.0
5.1
2.9
5.1
3.1
5.0
2.8
5.3
0.74
0.38
0.23
0.19
0.15
0.11
0.07
0.96
0.46
0.26
0.21
0.13
0.08
0.52
0.18
0.45
0.19
0.49
0.18
0.50
0.13
1.5
2.9
4.8
5.8
7.5
9.8
15.1
1.1
2.4
4.2
5.3
8.5
13.5
2.1
6.2
2.4
5.6
2.3
5.9
2.2
7.8
0.56
0.89
1.21
1.25
1.41
1.42
1.21
0.75
1.02
1.22
1.36
1.32
1.21
0.46
0.57
0.51
0.97
0.55
0.78
0.51
0.72
3 2
Pd P0.95/SiO
0
1
2
3
Rh/SiO
Rh P/SiO
Pt/SiO
PtP /SiO
2
5
2
2
7
2
9
2
2
1
a
b
c
As determined by ICP.
Mean cluster diameter estimated from the metal dispersion.
mol of 4-CP)/(mol of surface metal per second), extrapolate to zero conversion (298 K, 3.1 mmol L 4-CP).
ꢁ1
(
diffractometer with Cu K
a
radiation (k=1.5406 Å, 40 kV, 40 mA) at
2.4. Catalytic reactions
a scanning rate of 4°/min in the range of 10–80° (2h). The size and
morphology of the as-synthesized samples were observed on a
Tecnai G2 F20 transmission electron microscope (TEM) operated
at 200 kV. The TEM specimens were prepared by dispersing the
samples in ethanol by ultrasonication for 3 min and depositing
the dispersed samples onto on carbon-coated Cu grid. Metal cluster
size distributions were determined by counting >400 crystallites.
The 4-CP HDC experiments were performed in a 500 mL batch
mode three-necked flask under continuous stirring at room tem-
perature and atmospheric pressure. The effect of increasing the
2
H flow rate and stirring speed on the hydrodechlorination perfor-
3
ꢁ1
mance was investigated, and a hydrogen flow rate of 1 cm s and
a stirring speed of 600 rpm were set to exclude the limit of H dif-
fusion. In a typical reaction, 0.1–2.0g of catalyst was transferred
2
The surface area-weighted cluster diameters, dTEM, were calculated
P
3
2
3 ꢁ1
using dTEM
=
n d /n
i
i
d [53].
into the reactor and treated by a hydrogen flow (1 cm s ) at
353 K for 1 h to reduce the surface noble metal oxides. Then, the
temperature of the reactor was cooled to room temperature
(298 K), and deionized water with a trace amount of NaOH were
X-ray photoelectron spectroscopy (XPS) with a Fluorolog-Tau-3
ISA-USA) was used to examine the electronic properties of the cat-
(
alysts with non-monochromatized Mg K
a radiation as the excita-
3
ꢁ1
tion source, and the binding energies were calibrated using a C 1
s binding energy of 284.8 eV. The surface of the fresh samples
was etched with Ar+ ions for 15 min to remove the oxided layer.
The peaks were fitted by a nonlinear least square fitting program
using a properly weighted sum of the Lorentzian and Gaussian
component curves after the background subtraction, according to
Shirley and Sherwood.
added under 600 rpm stirring with a 1 cm s
hydrogen flow.
The initial NaOH concentration ([NaOH]) in the reaction mixture
ꢁ1
was in the range of 0.1 and 10 mmol L , such that the [NaOH]/
[4-CP] ratio lies at 1. After stirring for 10 min, a certain amount
ꢁ1
of 4-chlorophenol was injected to create a 0.1–10 mmol L solu-
tion, and the reaction was immediately timed. To evaluate the con-
version rate of the reaction, samples were collected with a glass
The metal content was measured by inductively coupled
plasma (ICP) analysis using an IRIS Advantage spectrometer.
The metal dispersions on Pt, Pd and Rh samples were measured
syringe at 2 min intervals and passed through a 0.45
brane filter.
lm mem-
For the 4-CP HDC reaction, a series of blank experiments were
conducted in the absence of a catalyst with a preliminary 4-CP con-
by H
instrument [53,54]. For the H
first treated in pure H at 623 K (0.067 K s ) for 1 h and then in
a dynamic vacuum at 623 K for 1 h. Hydrogen adsorption iso-
therms were measured at 313 K and 5.0–50 kPa of H for Pt
2
and CO chemisorptions using a Micromeritics ASAP 2020
ꢁ
1
2
chemisorptions, the samples were
centration (C
0
) of 0.5–10 mmol L and a hydrogen flow rate of 1
ꢁ
1
3 ꢁ1
2
cm s . The loss of 4-CP content in the H flow was negligible
2
(<0.1%) in 2 h. The impacts of the hydrogen flow rate, temperature,
stirring speed and initial pH on the reaction were also studied, and
the reaction conditions were selected to minimize the transport
limitations observed in our previous work [55].
2
and Rh. To avoid formation of the b-hydride phase in the Pd
samples, the isotherms were measured at 343 K and 0.4–1.5
kPa of H
2
.
To test the chloride effect, different amounts of NaCl were
For CO chemisorption, the catalysts was reduced under pure H
2
added to the initial HDC solution to obtain chloride concentrations
ꢁ1
ꢁ1
atmosphere at 623 K (0.067 K s ) for 1 h, followed by a dynamic
vacuum at the same temperature for 3 h. The catalyst was then
brought to 313 K under vacuum. After introducing CO to the sam-
ple, adsorption isotherms were measured at 313 K and 5.0–50 kPa.
Metal dispersions were calculated using adsorption stoichiometry
ranging from 0 to 0.8 mol L . To test the sulfide effect, Na
2
S was
also added to the HDC solution at different ratios of sulfur to sur-
face active Pd, which was similar to the work of Wong et al. [30].
The concentration of 4-CP and phenol after the reaction were
measured by an HPLC with a diode-array detector using a C18 col-
umn. Other products (i.e., cyclohexanone) were analyzed by GC
s s s s s
of H/Pt = 1, H/Pd = 1, H/Rh = 1, CO/Pt = 1and CO/Pd = 1.

Z. Wu et al. / Journal of Catalysis 366 (2018) 80–90
83
(
(
GC-7860) using a 30 m length and 0.25 mm i.d. capillary column
DB-Wax, Agilent) connected to a flame ionization detector.
In many cases, the reactions were tested for repeatability under
identical conditions, and the reaction reproducibility was better
than ±5%. The carbon balance was calculated by comparing the
amount of carbon entering per unit time and the number of prod-
ucts exiting per unit time, and the carbon mass balance was within
ꢁ
discrepancy of ±1%. A chlorine (in the form of Cl ions in product)
mass balance was performed by potentiometric titration with
ꢁ
1
0
.01–0.05 mol L
3
AgNO solution via a Metrohm Model 905
Autotitrator. The discrepancy of Cl mass balance was less than ±5%.
The turnover rate (TOR) of HDC reaction was reported as the
molar conversion rate of 4-CP per surface metal atom. Selectivities
were calculated based on the percentage of the converted 4-CP in
ꢁ1
the product. The residence time (mol surface metal s mol reac-
tant) was defined as the reciprocal of the space velocity, which
was controlled by varying the reactant molar rates while keeping
the respective pressures constant [56]. In this study, the 4-CP-to-
catalyst ratio was varied (at the same concentration of 4-CP and
hydrogen) to study the effect of conversion on the TOR. The initial
TOR was obtained by extrapolation to zero conversion.
Fig. 1. XRD patterns of the SiO
samples.
2
-supported palladium and palladium phosphide
3
. Results and discussion
3.1. Catalyst properties
2 2
and PtP compounds (JCPDS No. 1-071-2233) on the SiO -
Table 1 and S1 shows the metal loading and metal dispersion of
supported rhodium, rhodium phosphide, platinum and platinum
phosphide samples, respectively. These data indicate that the crys-
talline noble metals (Pd, Rh and Pt) and their corresponding phos-
the SiO
2
-supported metals and metal phosphides. With the
supports, the metal dis-
increase in the metal loading on the SiO
2
persion decreases, suggesting the growth of the mean cluster size
of the metal and metal phosphide particles. The mean cluster size
of the metal and that of the metal phosphide particles were mea-
sured via hydrogen and CO chemisorptions, and TEM measure-
ment, respectively. The trends of size variation with metal
loading for metal and phosphide particles are similar on all three
measurement techniques. In particular, the sizes of Pd and Pt par-
ticles are similar based on hydrogen and CO chemisorptions, and a
bit smaller than that from TEM results. However, the size of phos-
phide particle obtained from the TEM images is larger than the cal-
culated values from chemisorption technique. TEM imaging is
difficult to distinguish small clusters. Moreover, the mean cluster
size from chemisorptions does not consider increase of crystal cell
parameter caused by the introduction of P into the metal lattice.
For instance, the cubic Pd metal crystal can be converted to
phides (Pd
the temperature-programmed reduction with hydrogen without
the introduction of other phosphating agents (e.g., PH ).
3 2 2
P0.95, Rh P and PtP ) were successfully synthesized by
3
Supplementary data associated with this article can be found, in
the online version, at https://doi.org/10.1016/j.jcat.2018.07.040.
The morphology and size of palladium phosphide particles with
2
different Pd loadings (1, 5 and 10 wt%) on SiO were inspected by
TEM, as shown in Figs. 2 and S2. The palladium phosphides have
spherical particles and the particle size increased as the metal
loading increased from 1 to 10 wt%. Notably, the phosphide parti-
cles on the supported palladium phosphide sample with 1 wt% Pd
loading (Fig. S2) were difficult to distinguish because of the small
particle size and high dispersion of the particles (0.96 Pd dispersion
according to the hydrogen chemisorption characterization, entry 8
in Table 1). Approximately 7.8 nm palladium phosphide particles
(TEM characterization, entry 11 in Table S1) were observed on
the sample with 5 wt% Pd loading, which is a bit larger than the
hydrogen chemisorption results (5.3 nm mean cluster size calcu-
lated based on the 0.21 palladium dispersion, entry 11 in Table 1).
However, a broad size distribution (5–20 nm, 16.1 nm mean clus-
ter size from TEM characterization, entry 13 in Table S1) of palla-
dium phosphide particles was found for the sample with a 10 wt
% metal loading, which suggested a low metal dispersion (0.08,
entry 13 in Table 1). These data agreed with the increase in the
3
orthorhombic Pd P0.95 crystal. In order to obtain a reasonable reac-
tion rate based on the active surface sites of metal, the dispersion
from hydrogen chemisorption was adopted and its corresponding
mean size of metal and phosphide particles are used to elucidate
the structural sensitivity of the HDC reaction.
Figs. 1 and S1 show the XRD patterns of the SiO
noble metals (Pd, Rh and Pt) and their corresponding phosphides
Pd P and PtP ). The XRD pattern (Fig. 1a) of the Pd/SiO
2
2
-supported
(
3
P0.95, Rh
2
2
sample has three reflection peaks at 2h = 40.4°, 46.8° and 68.4°,
which correspond to the Pd (1 1 1), (2 0 0) and (2 0 0) lattice
planes, respectively; these results indicate the formation of metal-
lic palladium compounds (JCPDS No. 1-1201). The XRD pattern
3 0.95
intensity of the diffraction peaks corresponding to the Pd P
compounds in the XRD patterns of the supported palladium phos-
phide samples as the Pd loading increased from 5 to 10 wt% (Fig. 1).
To further investigate the surface structural and electronic
(
Fig. 1a) of the palladium phosphide with 10 wt% Pd loading only
contains peaks that are characteristic of Pd 0.95 compounds
JCPDS No. 1-089-3046), whereas the sample with a lower Pd load-
P
2 3
properties of the SiO -supported Pd and Pd P0.95 samples, XPS
3
(
measurements were conducted, and the results are illustrated in
Figs. 3 and 4 and Tables 2 and 3. Tables 2 and 3 include the XPS
data of the Pd binding energies and surface atomic ratios for the
ing (5 wt%) contains broad peaks at approximately 2h = 35–45°
because of the small size and homogeneous distribution of the
Pd
chemisorptions results (see discussion below). Fig. S1 shows
the formation of metallic rhodium (JCPDS No. 1-1214), Rh P com-
pounds (JCPDS No. 2-1299), metallic platinum (JCPDS No. 1-1194)
3
P
0.95 particles, which was confirmed by the TEM and hydrogen
catalysts before and after hydrodechlorination. Fresh Pd/SiO
2
0
(Fig. 3) displays peaks at 341.0 eV that correspond to Pd 3d3/2
0
and bands at 335.7 eV that correspond to Pd 3d5/2. In comparison
2
2 3 2
to the Pd/SiO peaks, both the peaks in Pd P0.95/SiO slightly

8
4
Z. Wu et al. / Journal of Catalysis 366 (2018) 80–90
2
Fig. 2. The TEM images of the SiO -supported palladium phosphides with 5 and 10
wt% Pd loading.
increase to 341.2 and 335.9 eV. This may be attributed to the trans-
fer of electrons from the Pd species to P species, which has been
reported for amorphous metal-phosphor nanoparticles [57]. It
should be noted that the small binding energy shift (<0.2 eV) could
be attributed to analytical error. In order to show the interaction
between Pd and P species on the fresh catalyst, the FT-IR spectra
of CO adsorption was obtained (Fig. S3). Two absorption bands at
ꢁ
1
ꢁ1
1
(
964 cm (CO mainly on the Pd bridging sites) and 2089 cm
CO occupying the Pd atop sites) are given by the Pd/SiO . On the
other hand, the bands associated with the CO adsorption on the
2
ꢁ
1
Pd atop sites in Pd
3
P
0.95/SiO
2
were shifted from 2089 cm and
ꢁ1
ꢁ1
ꢁ1
1
964 cm to 2072 cm and 1952 cm , respectively, indicating
an obvious influence from the ligand through changing the chem-
ical environment around the Pd atoms.
After the introduction of P into Pd metal, the binding energy of
the P species in Pd
3
P
0.95/SiO
2
is 133.1 eV, which corresponds to the
groups. Moreover, a high atomic ratio of P to Pd at
approximately 12 was determined from the surface composition
Table 3). This suggests that the surface of Pd is rich in
3
ꢁ
energy of PO
4
(
3 2
P0.95/SiO
phosphates which agreed with the report that the surface of the
phosphides synthesized via the temperature-programmed reduc-
4 2 4
tion method with excess phosphate (i.e., NH H PO ) is always cov-
ered by phosphate [58].
After the HDC reaction, the binding energies of the Pd species in
Pd/SiO (Fig. 3) slightly increase from 340.9 and 335.6 eV to 341.4
and 336.2 eV, respectively. Fig. 4 and Tables 2 and 3 show the for-
mation of surface Cl species on the Pd/SiO . These results indicate
the interaction of Cl and Pd metal via the transfer of electrons from
Pd to Cl species. Unlike Pd/SiO , Pd possesses slight
2
2
3 0.95
Fig. 3. XPS spectra of the Pd species on SiO -supported Pd and Pd P samples
with 5 wt% Pd loading.
2
2
3 2
P0.95/SiO
ꢁ
2
ꢁ2
growth (ꢀ0.1 eV) in the binding energies of the Pd species. How-
the surface concentration decreases (7.01 ꢄ 10 to 0.21 ꢄ 10
atomic ratio of P/O, Table 3). The sharp decrease in the surface con-
centration of P species should be due to the dissolution of surface
ever, the binding energy of P species (Fig. 4 and Table 2) changes
from 133.1 eV (for P species) to 131.1 eV (for P0 species), and
5
+

Z. Wu et al. / Journal of Catalysis 366 (2018) 80–90
85
reaction. Accordingly, the introduction of P to the Pd lattice to form
ꢁ
Pd
3
P
0.95 inhibits the sorption of Cl ions. In addition, the phosphide
catalyst could be the stable catalyst in the HDC reaction even the
2ꢁ
presence of poison species (i.e., S for metal catalysts). For the
Pd/SiO samples treated in Na S solution, a new binding energy
2
2
2
+
at 337.1 eV is observed (attributed to Pd in the palladium sulfide
compounds) (Fig. 3), suggesting the surface reaction of S2 and the
metallic Pd sites. The XPS spectra of S species (Fig. 4 and Table 3)
ꢁ
6
+
confirm the surface is rich in S (corresponding to the peak at
2
ꢁ
2ꢁ
3
1
68.1 eV) species, which form via the oxidation of S to S
cies during the drying process of the treated samples in air prior to
XPS characterization. In contrast, the Pd sample still
2
O
spe-
3 2
P0.95/SiO
shows a shift in the binding energy (ꢀ0.2 eV) compared to the
fresh catalyst. An obvious change in the shift of the binding energy
(
from 133.1 to 133.8 eV) for the P species is observed and may be
due to the formation of surface phosphorus-sulfur compounds.
Strangely, the sulfur compound is difficult to detect. These data
ꢁ
show that Pd
S
3
P
0.95/SiO
2
possesses good resistance to Cl and
2
ꢁ
ions in aqueous solutions and is a potential stable catalyst for
the aqueous phase HDC reaction.
3.2. HDC activity of the noble metal and phosphide catalysts
The pseudo-first-order kinetics is selected for the aqueous
phase HDC of 4-CP over the noble metal catalysts [5–25]:
dc
dt
¼
ꢁkobs
C
ð1Þ
where C is the 4-CP concentration (mol Lꢁ1) in the solution phase
ꢁ1
and kobs (min ) is the observed pseudo-first-order reaction rate
constant. The rate constants were obtained from a linear regression
of the natural log of the solution phase concentration versus time
[
5,55]. Figs. S4 and S5 and Table S2 Supporting information show
that the palladium and rhodium phosphide catalysts also follow
pseudo-first-order kinetics.
Fig. 5 shows the product distribution of the 4-CP HDC reaction
for different reaction conversions. Only phenol and cyclohexanone
organic compounds were detected in our experiments, which sug-
gested cyclohexanone was not hydrogenated to cyclohexanol over
the Pd-based catalysts [6]. The selectivity of phenol decreases with
the conversion while the selectivity of cyclohexanone increases.
Interestingly, a higher selectivity of cyclohexanone was observed
on palladium than the palladium phosphide catalyst, which indi-
cates the better hydrogenation capability of the aromatic rings
via the metallic surface rather than the phosphide surface. How-
ever, the selectivity of phenol is still much higher than that of
cyclohexanone. This represents a difficulty in the hydrogenation
of aromatic rings.
The initial turnover rate was estimated based on the common
rate analysis technique using the extrapolation of the conversion
to zero residence time (the reciprocal of the space velocity) [56].
Given the turnover rate or the conversion is a function of residence
time, the hypothetical rate at zero conversion can be taken as the
initial turnover rate. For the liquid HDC reaction, we used different
ratios of 4-CP to catalyst (at the same concentration of 4-CP and
hydrogen flow rate) to study the effect on the turnover rate. This
is equivalent to the effect of space velocity on conversion in gas-
phase heterogeneous catalytic reactions. We used the plot of turn-
over rate as a function of conversion to approximate the initial
turnover rate of HDC via extrapolation to zero conversion. The ini-
tial turnover rates (TOR) obtained in Fig. 5b and Table 1 were used
to discuss the activity and reaction pathways of the 4-CP HDC
reactions.
2 3
Fig. 4. XPS spectra of P, Cl and S species on SiO -supported Pd and Pd P0.95 samples
with 5 wt% Pd loading.
phosphate in water, and the surface atomic ratio of Pd/P becomes
.2, which is similar to the bulk composition of Pd 0.95. Interest-
ingly, few Cl species were observed on Pd 0.95/SiO after HDC
Among the Pd, Rh and Pt nanoparticles, which have similar
3
3
P
particle size of 5–8 nm, on SiO
2 2
(Fig. 6 and Table 1), the Pd/SiO
3
P
2
and Rh/SiO samples gave the highest and lowest initial TOR of
2

8
6
Z. Wu et al. / Journal of Catalysis 366 (2018) 80–90
Table 2
Binding energy of Pd, P, S and Cl species based on XPS data for palladium catalysts with 5 wt% Pd loading before and after reaction.
Sample
Palladium species
Binding energy (eV)
Pd 3d5/2
Pd 3d3/2
P 2p1/2
Cl 2p1/2
Cl 2p3/2
S 2p1/2
S 2p3/2
Pd/SiO
Pd 0.95/SiO
Pd/SiO
Pd
Pd/SiO
2
Pd0
335.7
335.9
336.2
335.7
336.1
337.1
335.8
341.0
341.2
341.4
341.1
341.4
–
–
–
–
–
–
–
–
–
–
–
–
–
–
0
5+)
3
P
2
Pd
Pd
Pd
Pd
133.1 (P
–
131.1(P )
0
ꢁ
ꢁ
2
after HDC reaction
200.1 (Cl )
198.3(Cl )
0
0
3
P
0.95/SiO
2
after HDC reaction
–
–
–
–
–
–
–
–
0
2ꢁ
2ꢁ
2
with Na
2
S
–
–
161.8 (S
)
166.2 (S )
2
0
+
168.1 (S+6)
Pd
Pd
Pd
3
P
0.95/SiO
2
with Na
2
S
341.1
133.8 (P5+)
Table 3
Surface compositions of Pd, P, S and Cl species based on XPS data for palladium
catalysts with 5 wt% Pd loading before and after reaction.
Sample
Atomic ratio (ꢄ10ꢁ2)
P/O
Pd/O
Cl/O
S/O
Pd/SiO
Pd 0.95/SiO
Pd/SiO
Pd
Pd/SiO
Pd 0.95/SiO
2
–
7.01
–
0.21
–
0.28
0.69
0.59
0.56
0.67
0.77
0.75
–
–
0.99
0
–
–
–
–
0
5.01
0
3
P
2
2
after HDC reaction
3
P
0.95/SiO
with Na
with Na
2
after HDC reaction
2
2
S
3
P
2
2
S
–
ꢁ
4
ꢁ1
ꢁ1
ꢁ4
ꢁ1
ꢁ1
1
.25 ꢄ 10 mol molsurf-Pd
s
and 0.57 ꢄ 10 mol molsurf-Rh
s ,
respectively. This result shows that Pd metal is a more suitable
choice among other noble metal catalysts, which is in agreement
with the work reported by Diaz et al. [59]. Unlike a higher activity
given by Rh/Al
2 3 2 3
O over Pt/Al O [59], it should be noted that the
ꢁ4
TOR of our synthesized Rh/SiO
2
catalysts (0.46–0.57 ꢄ 10 mol
ꢁ
1
ꢁ1
molsurf-Rh
s ) is slightly lower than that of our synthesized
ꢁ4
ꢁ1
ꢁ1
Pt/SiO
2
catalysts (0.55–0.78 ꢄ 10 mol molsurf-Pt
s ). Molina
et al. reported an activity order for pillared clays-supported Pt,
Pd and Rh catalysts: Rh > Pd > Pt [21]. It is possible that the HDC
activity on Rh and Pd catalysts were improved by the metal-
support interactions in pillared clay and Al
ferent from our system using SiO as a support. For the phosphide
nanoparticles with a size of 5–8 nm, Pd gave a TOR of
2 3
O [21,59], which is dif-
2
3 2
P0.95/SiO
, which is higher than that of Rh
ꢁ
4
ꢁ1
ꢁ1
ꢁ1
1
SiO
.36 ꢄ 10 mol molsurf-Pd
s
2
P/
ꢁ4
ꢁ1
ꢁ4
2
(0.97 ꢄ 10 mol molsurf-Rh
s
) and PtP
2
/SiO
2
(0.72 ꢄ 10
ꢁ
1
ꢁ1
mol molsurf-Pt
s
). Interestingly, Bussell et al. reported a higher
hydrodesulfurization turnover frequency on Rh P/SiO over
Pd /SiO [45,47]. In this work, palladium phosphide (Pd
SiO ) is found to be more reactive than the other metal phosphide
2
2
5
P
2
2
3 0.95
P /
2
catalysts in the aqueous phase HDC reaction of 4-CP.
Fig. 6 shows the effect of the particle size of the metal and phos-
phide nanoparticles on the initial TOR. Both Pd/SiO
2 3 0.95
and Pd P /
SiO show similar trends. The initial TOR increases with the parti-
2
cle size for particles smaller than ꢀ8 nm and decreases slightly
with enlarging particles. These data show that the structural sensi-
tivity could be a determining factor to the HDC reactivity, which is
consistent with the results reported by Diaz et al. [10,20,27,37]. It
is proposed that the rate-limiting step of the HDC reaction is the C–
Cl hydrogenolysis of the adsorbed molecule [60,61], which is cat-
alyzed by the surface palladium atoms. In HDC of 4-CP, the hydro-
xyl group of 4-CP increases the electron density within the ring via
an inductive effect and, hence, activates the aromatic ring for elec-
trophilic attack. 4-CP possesses multiple bonds that can be chemi-
Fig. 5. (a) Product distribution of the 4-CP HDC reaction over the palladium and
palladium phosphide catalysts and (b) the effect of the reaction conversion rate on
the turnover rate (TOR) of the HDC reaction over the palladium and palladium
ꢁ
1
phosphide catalysts (Initial TOR, extrapolated to zero conversion (298 K, 3.1 mmol L
-CP)).
4
sorbed on the metal surface through
p-bonding, involving
donation of -electron density from the adsorbate towards the
p
metal surface which requires large ensemble. The probability of
forming such an ensemble increases with increasing nanoparticle
size, which leads to an increase in the initial TOR. The number of
atoms surrounding each surface metal atom decreases with
decreasing particle size; affecting the nature and strength of the
interactions between the surface of metal particles and 4-CP mole-
cules [10,20]. Therefore, the change in the coordination associated
with the nanoparticle size has a strong effect on the adsorption

Z. Wu et al. / Journal of Catalysis 366 (2018) 80–90
87
Fig. 7. Proposed reaction pathways for the 4-CP HDC Pd-based catalysts (M is a
metal surface site, denotes a quasi-equilibrated step, and k and K are the
kinetic and equilibrium constants, respectively, for the individual elementary steps;
-CP is 4-chlorophenol, Ph is phenol, and one is cyclohexanone.)
x
x
Fig. 6. Effect of the metal and phosphide cluster size on the initial TOR for the 4-CP
HDC reaction (298 K, 3.1 mmol L 4-CP).
ꢁ
1
4
energy. In general, smaller particles contain more surface atoms on
the edges of the crystallographic planes or on the edge junctions.
They tend to increase the number of active sites that are accessible
to the 4-CP molecules [62]. These factors indicate that an optimum
particle size is expected for the HDC reaction and that the particles
must possess a sufficient ensemble to adsorb and activate 4-CP.
Metal particles with a large particle size and a high coordination
number attends to give low reactivity. For particles larger than
cally relevant molecular or ion-adsorption events. However, the
hydrogen chemisorbed by the Pd metal is in the lattice of the sur-
face metal crystal. Here, all the palladium-based catalysts were
reduced by hydrogen prior to the HDC reaction, which suggested
a hydrogen-rich surface and the formation of surface PdHx com-
pounds [6-25]. Thus, the equilibrium adsorption of 4-CP and
hydrogen should not be competitive.
The surface coverages (h , i = 4ꢁCP, phenol, cyclohexanone, HCl,
i
ꢁ
1
0 nm, however, Pd
3
P0.95/SiO
2
exhibits a slightly lower initial
and Cl ) of the adsorbed species for the steps (1.1–1.9) are given by
TOR than the Pd/SiO
2
catalyst. Here, geometric or ensemble effects
the Langmuir adsorption isotherms. Because NaOH was present in
the initial solution to maintain a [NaOH]/[4-CP] ratio at 1, HCl com-
pounds should be consumed as soon as they are formed. Moreover,
a much lower selectivity for cyclohexanone than phenol, as shown
in Fig. 5, suggests the low hydrogenation activity of the aromatic
ring on the Pd catalysts. Thus, we assumed the sorption of HCl
and cyclohexanone could be ignored compared to the sorption of
other species.
emerge because of the dilution of the surface active metal (Pd) by
an inactive one (P). However, the presence of P changes the valence
of the surface Pd species and inhibits the interaction between Pd
ꢁ
and Cl ions, as shown in the XPS results. As a result, both elec-
tronic and geometric effects should be considered when screening
the particle size of Pd
concluded that a size of ꢀ5 nm and ꢀ8 nm is suitable for
Pd and Pd/SiO , respectively.
3 2
P0.95/SiO . Based on the results, it is
3
P0.95/SiO
2
2
K
CP½4 ꢁ CPꢅ
hMCP ¼ 1
ꢁ
ð2Þ
þ KCP½4 ꢁ CPꢅ þ KCl½Cl ꢅ þ KPh½Phꢅ
3.3. Kinetic modeling and the effect of chloride on the HDC reaction
The HDC turnover rate can be written as:
rate
k
1
K
CP½4 ꢁ CPꢅ
ꢁ
r ¼ k
1
½MCPꢅ½MHꢅ ¼
ꢄ ½MHꢅ
ꢁ
In the XPS results, different contents of Cl species are observed
1 þ KCP½4 ꢁ CPꢅ þ KCl½Cl ꢅ þ KPh½Phꢅ
on Pd
3
P
0.95/SiO
2
and Pd/SiO
2
after the HDC reaction (Fig. 4 and
Table 3), and thus, the effect of the sorption of Cl species on Pd/
SiO should be more important than that on Pd P0.95/SiO for the
2 3 2
kK ½4 ꢁ CPꢅ
CP
ꢁ
¼ 1
ꢁ
ð3Þ
þ KCP½4 ꢁ CPꢅ þ KCl½Cl ꢅ þ KPh½Phꢅ
where k = k
1
ꢄ [MH].
HDC reaction. Here, we attempted to explore the kinetic model
of the HDC of 4-CP based on the Langmuir-Hinshelwood (L-H)
mechanism to clarify the difference. For the L-H mechanism, the
rate-determining step of the HDC reaction is regarded to be the
C–Cl hydrogenolysis of the 4-CP reaction with H atoms on metal
surface [6, 43, and 44]. Here, Fig. 7 presents the elementary steps
of the 4-CP HDC reaction over the palladium and palladium phos-
phide catalysts.
As shown in Fig. 7, 4-CP and hydrogen adsorb onto the catalyst
surface to form MꢁCP and MꢁH, which are the most abundant
reactive intermediates, via quasi-equilibrium reactions. The
quasi-equilibrium assumption requires fast 4-CP and
adsorption-desorption processes on the time scale of the kineti-
At the beginning of the reaction, the kinetic model can be
expressed as:
kKCP½4 ꢁ CPꢅ
0
r
0
¼ 1
ð4Þ
þ KCP½4 ꢁ CPꢅ
0
ꢁ
1
ꢁ1
where r
[4-CP]
intrinsic rate constant.
0
(mol molsurf-Pd
s
) is the initial TOR, [4-CP]
0
is the initial
ꢁ1
ꢁ1
ꢁ1
0
concentration (mol L ), and k (mol molsurf-Pd
s
) is the
To test the kinetic model and investigate the relative contribu-
tion of the 4-CP adsorption and surface reaction to the HDC reac-
tion, the influence of the 4-CP initial concentration was evaluated
(Fig. 8a). The initial reaction rate was measured for various initial
H

8
8
Z. Wu et al. / Journal of Catalysis 366 (2018) 80–90
and palladium phosphide catalysts in the order of Pd
Pd/SiO over the entire range of the tested concentrations
Fig. 8a). The L-H model (Equation (4)) well fits the initial TOR ver-
3 2
P0.95/SiO >
2
(
2
sus the initial concentration (R > 0.99, Table 4). A higher intrinsic
rate constant k and equilibrium adsorption constant, KCP, were
obtained for Pd
3
P0.95/SiO
2
2
than the Pd/SiO catalysts. This agrees
with the electronic properties of the Pd
3
P0.95 particles determined
by the XPS characterization; i.e., Pd electron transfer to P and 4-CP
is preferable to Pd site adsorption.
3 2 2
The reactivity of the Pd P0.95/SiO and Pd/SiO catalysts were
further tested in aqueous solutions with varying concentrations
ꢁ1
of NaCl from 0 to 0.8 mol L at a [NaOH]/[4-CP] ratio of 1. The
kinetic model can be expressed as:
kKCP½4 ꢁ CPꢅ
0
r
0
¼ 1
ꢁ
ð5Þ
þ KCP½4 ꢁ CPꢅ þ KCl½Cl ꢅ
0
0
The results show a decrease in the rate constant of Pd/SiO
pared with that of Pd
the initial TOR, Fig. 8b) when the NaCl concentration increases
2
com-
3 2
P0.95/SiO (21% decrease vs. 10% decrease in
ꢁ1
from 0 to 0.8 mol L . Both Pd
3 2 2
P0.95/SiO and Pd/SiO show a sim-
ilar trend for the change in the initial TOR; i.e., a faster decrease
ꢁ1
in the rate is observed from 0 to 0.1 mol L followed by a marginal
change upon further increases in the NaCl concentration. The L-H
model (Equation (5)) well fits the initial TOR versus the initial con-
2
centration of Cl ions (R > 0.99, Table 4). A much lower equilibrium
adsorption constant, KCl, was obtained for Pd
SiO catalysts, which confirms the result that few Cl elements
could be detected on the surface of Pd after the HDC
3 2
P0.95/SiO than Pd/
2
3 2
P0.95/SiO
reaction. The parity plots in Fig. S6 compares the calculated and
the measured initial turnover rates for the HDC at 303 K over Pd/
SiO
that Equation (5) describes the measured initial turnover rates
accurately for the Pd/SiO and Pd . These data show that
2 3 2
and Pd P0.95/SiO . The plots show good correlation, indicating
2
3 2
P0.95/SiO
the 4-CP HDC mechanism followed the typical Langmuir-
Hinshelwood mechanism as reported by previous works
[
5,63,64]. That is, both 4-CP and hydrogen are first adsorbed on
the metal surface, followed by their activation and reaction that
produces phenol and Cl adsorbed on the metal surface.
3.4. The effect of sulfide on the HDC reaction rate
Fig. 8c shows the results for the sulfide poisoning experiments
on the SiO
Pd atom ratios were considered. In the absence of sulfide, the
Pd and Pd/SiO catalysts had initial TOR of 1.36 and
.25 mol molsurf-Pd , respectively. The 4-CP HDC activities of
the Pd 0.95/SiO and Pd/SiO catalysts monotonically decreased
with the addition of sulfide, but the addition of sulfide had little
effect on the Pd and Pd/SiO catalysts until S:Pdsurf ratios
2 3
-supported Pd and Pd P0.95, and the sulfide-to-surface
3
P
0.95/SiO
2
2
ꢁ
1
ꢁ1
1
s
3
P
2
2
3
P0.95/SiO
2
2
of 0.8 and 1.0, respectively. As a result, the addition of sulfide at a
S:Pdsurf ratio >1.0 decreased the initial HDC TOR to ꢀ0.7 of its value
3 2
prior to the sulfide addition on Pd P0.95/SiO and fully suppressed
the rate on Pd/SiO
2
(ꢀ0.1 of its original value). This small decrease
in the HDC ratio on the phosphide catalysts reflects either the com-
petitive sorption between 4-CP and sulfide on the phosphide sur-
face or irreversible sulfide poisoning of the metal sites with low
coordination numbers [30]. The introduction of P apparently
enhances the sulfide poisoning resistance of Pd for the 4-CP HDC
reaction. This beneficial presence of phosphide was observed in
gas-phase HDS reactions using phosphides [43,58]. The phosphides
show a good sulfur or sulfide resistance which prevents bulk for-
ꢁ
Fig. 8. Influence of (a) the initial 4-CP concentration, (b) the initial Cl ion
2ꢁ
concentration and (c) the initial S ion concentration on the initial TOR of the HDC
reaction (298 K, 3.1 mmol L 4-CP, Pd-based catalysts with 5 wt% metal loading as
shown in entries 4 and 11 in Table 1).
ꢁ
1
mation of catalytically inactive Pd
Fig. S7. Hayes et al. proposed that bonding in noble metal phos-
phide (i.e., Rh P) is dominated by metal-P (i.e., Rh-P) interactions,
as indicated by the metal-P distance (i.e., Rh-P, 0.2381 nm) that
4
S crystallites, as shown in
4
-CP concentrations from 0.05 to 10 mmol Lꢁ1. r
the increase in the initial 4-CP concentration for the palladium
0
increases with
2

Z. Wu et al. / Journal of Catalysis 366 (2018) 80–90
89
Table 4
Fitting parameters of the L-H model for HDC of 4-CP at pH = 7 with 25 mg of 5 wt% Pd catalyst loading.
ꢁ
4
ꢁ1
ꢁ1
ꢁ4
ꢁ1
ꢁ4
ꢁ1
R2
Samples
Pd/SiO
Pd 0.95/SiO
2
k (10 mol molsurf-Pd
s
)
K
CP (10 L mol
)
K
Cl (10 L mol
)
2
4.4 ± 0.09
4.7 ± 0.15
0.15 ± 0.02
0.16 ± 0.02
0.04 ± 0.002
0.01 ± 0.001
0.991
0.987
3
P
2
phosphide catalyst. The Pd/SiO catalysts show an obvious
decrease in the HDC rate in the presence of chloride because of
Cl chemisorption onto Pd metal sites. The introduction of P into
the metallic lattice of the Pd/SiO catalysts changes the valence
2
of the Pd atoms. The surface interactions between Pd and P species
promote the adsorption of 4-CP but inhibit the sorption of Cl ions
after the HDC reaction, which was shown in the XPS characteriza-
tion and kinetic study results. The Pd/SiO
inactive (90% decrease in the TOR) at a S:Pdsurf ratio of 1, while
the Pd 0.95/SiO catalyst possesses a better sulfide poisoning resis-
2
catalysts were almost
3
P
2
tance (30% decrease in the TOR). This improved sulfide resistance
was related to the protection of the Pd active site from the forma-
tion of the Pd
structure. The HDC of 4-CP over Pd
4
S compounds via the formation of a phosphide
catalysts is a
3 2
P0.95/SiO
structure-sensitive reaction, and the electronic and ensemble
effects of the palladium phosphide particles determine the initial
TOR. Pd P0.95/SiO catalysts can be the basis for highly active and
3 2
highly stable catalysts for water treatment applications.
Acknowledgements
The present work was supported by the Natural Science of
Foundation China (U1662131, 21206192 and 21576140), 111 pro-
ject B12015, the Science Foundation of China University of
Petroleum-Beijing (C201603), and the Fundamental Research
Funds for the Central Universities.
Fig. 9. The initial TOR of the HDC reaction of 4-CP over Pd-based catalysts with
wt% metal loading (entry 4 an 11 shown in Table 1) during the recycle runs.
5
is shorter than the metal-metal distance (Rh-Rh, 0.2749 nm) [45].
Thus, the sulfur/sulfide resistance in the noble metal phosphide
catalyst is likely attributed to the P in the phosphide particles. They
inhibit the irreversible adsorption of S at the particle surface and
avoid possible incorporation of S into the bulk phase (i.e. prevent
formation of noble metal sulfide) [45–47].
References
[
1] M.A. Keane, ChemCatChem 3 (2011) 800–821.
[
2] F. Alonso, I.P. Beletskaya, M. Yus, Chem. Rev. 102 (2002) 4009–4091.
[3] I. Ali, V.K. Saini, Environ. Sci. Technol. 38 (2004) 4012–4018.
[
4] M.A. Keane, J. Chem. Technol. Biotechnol. 80 (2005) 1211–1222.
[
5] N. Jadbabaei, T. Ye, D. Shuai, H. Zhang, Appl. Catal. B 205 (2017) 576–586.
3.5. Catalyst stability
[6] C.B. Molina, A.H. Pizarro, J.A. Casas, J.J. Rodriguez, Appl. Catal. B 148–149
2014) 330–338.
(
[
[
7] M. Munoz, B.J.M. Malte Kaspereit, Etzold, Chem. Eng. J. 285 (2016) 228–235.
8] V. Felis, C.D. Bellefon, P. Fouiloux, D. Schweich, Appl. Catal. B. 20 (1999) 91–
100.
To determine the stability of the Pd-based catalysts, the
Pd P0.95/SiO and Pd/SiO catalysts were recycled for the 4-CP
3 2 2
HDC reaction. For each recycling experiment, the catalysts were
collected and sequentially washed with water and ethanol and
[9] Y. Shindler, Y. Matatov-Meytal, M. Sheintuch, Ind. Eng. Chem. Res. 40 (2001)
301–3308.
10] J.A. Baeza, L. Calvo, D.Y. Murzin, J.J. Rodriguez, M.A. Gilarranz, Catal. Lett. 144
2014) 2080–2085.
[11] X. Ma, Y. Liu, X. Li, J. Xu, G. Gu, C. Xia, Appl. Catal. B 165 (2015) 351–359.
3
[
dried in vacuum prior to use. As shown in Fig. 9, The Pd
SiO catalysts exhibited stability over the first six recycles based
on the initial TOR data (3% decrease), whereas an obvious decrease
11%) in the TOR of the Pd/SiO catalysts was observed. According
3 0.95
P /
(
2
[12] G. Yuan, M.A. Keane, Catal. Commun. 4 (2003) 195–201.
[13] G. Yuan, M.A. Keane, Catal. Today 88 (2003) 27–36.
[14] G. Yuan, M.A. Keane, Appl. Catal. B 52 (2004) 301–314.
(
2
to previous studies [30,41,42], catalyst deactivation mainly occurs
due to the chemisorption of chloride onto active sites and leaching
and sintering of active sites. In our study, the ICP and chemisorp-
tion results exclude the leaching and sintering problems of the
Pd/SiO and Pd P0.95/SiO catalysts. Moreover, the Pd P0.95/SiO
2 3 2 3 2
catalysts exhibited a better capacity to resist poisoning than the
Pd catalysts, which can be attributed to the presence of P species
[15] G. Yuan, M.A. Keane, J. Catal. 225 (2004) 510–522.
[
[
[
[
[
[
16] H.M. Roy, C.M. Wai, T. Yuan, J.K. Kim, W.D. Marshall, Appl. Catal. A 271 (2004)
37–143.
17] L. Calvo, A.F. Mohedano, J.A. Casas, M.A. Gilarranz, J.J. Rodriguez, Carbon 42
(2004) 1377–1381.
1
18] L. Calvo, M.A. Gilarranz, J.A. Casas, A.F. Mohedano, J.J. Rodríguez, Appl. Catal. B
67 (2006) 68–76.
19] C. Xia, Y. Liu, S. Zhou, C. Yang, S. Liu, S. Guo, Q. Liu, J. Yu, J. Chen, Catal.
Commun. 10 (2009) 1443–1445.
20] J.A. Baeza, L. Calvo, M.A. Gilarranz, A.F. Mohedano, J.A. Casas, J.J. Rodriguez, J.
Catal. 293 (2012) 85–93.
(
see Fig. 9).
. Conclusion
Pd
21] Z.M. de Pedro, E. Diaz, A.F. Mohedano, J.A. Casas, J.J. Rodriguez, Appl. Catal. B
103 (2011) 128–135.
4
[
[
22] L. Cheng, Z. Jin, X. Wang, Catal. Commun. 41 (2013) 60–64.
23] J. Zhou, K. Wu, W. Wang, Z. Xu, H. Wan, S. Zheng, Appl. Catal. A 470 (2014)
336–343.
3
P
0.95/SiO
2
, Rh
2
P/SiO
2
2
and PtP /SiO
2
catalysts were prepared
[24] H. Deng, G. Fan, C. Wang, L. Zhang, Catal. Commun. 46 (2014) 219–223.
[25] Z. Dong, C. Dong, Y. Liu, X. Le, Z. Jin, J. Ma, Chem. Eng. J. 270 (2015) 215–222.
[26] A.H. Pizarro, C.B. Molina, J.L.G. Fierro, J.J. Rodriguez, Appl. Catal. B 197 (2016)
236–243.
via a temperature-programmed reduction method and were uti-
lized for aqueous phase HDC of 4-CP under mild reaction condi-
tions (298 K). Pd P0.95/SiO was shown to be the most active
3 2

9
0
Z. Wu et al. / Journal of Catalysis 366 (2018) 80–90
[
[
[
27] E. Diza, A.F. Mohedano, J.A. Casas, L. Calvo, M.A. Gilarranz, J.J. Rodriguez, Appl.
Catal. B 106 (2011) 469–475.
28] M. Munoz, Z.M. De Pedro, J.A. Casas, J.J. Rodriguez, Appl. Catal. A 488 (2014)
[47] G.H.L. Savithra, R.H. Bowker, B.A. Carrillo, M.E. Bussell, S.L. Brock, ACS Appl.
Mater. Interfaces 5 (2013) 5403–5407.
[48] Y. Kanda, K. Kawanishi, T. Tsujino, A.M. Al-otaibi, Y. Uemichi, Catalysts 8
(2018) 160.
[49] Y. Shi, B. Zhang, Chem. Soc. Rev. 45 (2016) 1529–1541.
[50] X. Liu, J. Chen, J. Zhang, Ind. Eng. Chem. Res. 47 (2008) 5362–5368.
[51] J. Chen, S. Zhou, D. Ci, J. Zhang, R. Wang, J. Zhang, Ind. Eng. Chem. Res. 48
(2009) 3812–3819.
[52] J.A. Cecilia, A. Infantes-Molina, E. Rodríguez-Castellón, A. Jiménez-López, J.
Hazard. Mater. 260 (2013) 167–175.
7
8–85.
29] J.A. Baeza, L. Calvo, M.A. Gilarranz, J.J. Rodriguez, Chem. Eng. J. 240 (2014) 271–
80.
2
[
[
[
[
30] K.N. Heck, M.O. Nutt, P. Alvarez, M.S. Wong, J. Catal. 267 (2009) 97–104.
31] P.D. Vaidya, V.V. Mahajani, Appl. Catal. B 51 (2004) 21–31.
32] J.J. Wei, X.H. Xu, Y. Liu, D.H. Wang, Water Res. 40 (2006) 348–354.
33] J.A. Baeza, L. Calvo, J.J. Rodriguez, E. Carbo-Argibay, J. Rivas, M.A. Appl, Catal. B
1
68–169 (2015) 283–292.
[53] Z.J. Wu, G. Sarika, M. Choi, E. Iglesia, J. Catal. 311 (2014) 458–468.
[54] N. Mahata, V. Vishwanathan, J. Catal. 196 (2000) 262–270.
[55] C.X. Sun, Z.J. Wu, Y.Z. Mao, X.Q. Yin, L.Y. Ma, D.H. Wang, M.H. Zhang, Catal. Lett.
141 (2011) 792–798.
[
[
34] C. Xia, Y. Liu, J. Xu, J. Yu, W. Qin, X. Liang, Catal. Commun. 10 (2009) 456–458.
35] C. Ruiz-Garcia, F. Heras, N. Alonso-Morales, L. Calvo, J.J. Rodriguez, M.A.
Gilarranz, Catal. Sci. Technol. 8 (2018) 2598–2605.
[
36] Y. Lan, L. Yang, M. Zhang, W. Zhang, S. Wang, ACS Appl. Mater. Interface 2
[56] H. Wang, E. Iglesia, J. Catal. 273 (2010) 245–256.
[57] H. Li, H.X. Li, W.L. Dai, W.J. Wang, Z.G. Fang, J.F. Deng, Appl. Surf. Sci. 152
(1999) 25–34.
[58] R. Prins, M.E. Bussell, Catal. Lett. 142 (2012) 1413–1436.
[59] E. Diaz, J.A. Casas, A.F. Mohedano, L. Calvo, M.A. Gilarranz, J.J. Rodriguez, Ind.
Eng. Chem. Res. 47 (2008) 3840–3846.
(
2010) 127–133.
[
[
[
[
[
37] M. Munoz, S. Ponce, G.R. Zhang, B.J.M. Etzold, Appl. Catal. B 192 (2016) 1–7.
38] E.J. Shin, M.A. Keane, React. Kinet. Catal. Lett. 69 (2000) 3–8.
39] J.W. da-Silva, R.E. Bruns, A.J.G. Cobo, Chem. Eng. J. 131 (2007) 59–64.
40] M. Munoz, G.R. Zhang, B.J.M. Etzold, Appl. Catal. B 203 (2017) 591–598.
41] D. Angeles-Wedler, K. Mackenzie, F.-D. Kopinke, Environ. Sci. Technol. 42
[60] K. Early, V.I. Kovalchuk, F. Lonyi, S. Deshmukh, J.L. d’Itri, J. Catal. 182 (1999)
219–227.
(
2008) 5734–5739.
[
[
[
42] E. Lopez, S. Ordonez, F.V. Diez, Appl. Catal. B 62 (2005) 57–65.
43] Y.K. Lee, S.T. Oyama, Appl. Catal. A 548 (2017) 103–113.
44] R.H. Bowker, M.C. Smith, B.A. Carrillo, M.E. Bussell, Top. Catal. 55 (2012) 999–
[61] A.L.D. Ramos, M. Schmal, D.A.G. Aranda, G.A. Somorjai, J. Catal. 192 (1999)
423–431.
[62] A.Yu. Stakheev, I.S. Mashkovskii, G.N. Baeva, N.S. Telegina, Russ. J. Gen. Chem.
80 (2010) 618–629.
1
009.
[
45] J.R. Hayes, R.H. Bowker, A.F. Gaudette, M.C. Smith, C.E. Moak, C.Y. Nam, T.K.
Pratum, M.E. Bussell, J. Catal. 276 (2010) 249–258.
[63] B. Coq, G. Ferrat, F. Figueras, J. Catal. 101 (1986) 434–445.
[64] I.F. Cheng, Q. Fernando, N. Korte, Environ. Sci. Technol. 31 (1997) 1074–1078.
[
46] Y. Kanda, Y. Matsukura, A. Sawada, M. Sugioka, Y. Uemichi, Appl. Catal. A 515
(
2016) 25–31.