Tailoring High-Performance Pd Catalysts for Chemoselective Hydrogenation Reactions via Optimizing the Parameters of the Double-Flame Spray Pyrolysis

Article abstract of DOI:10.1021/acscatal.6b00396

Tuning the chemical composition during the synthesis is a widely used method to control the activity of catalysts. Here, we reported an alternative synthesis strategy to tune the catalytic properties of nanocatalysts without changing their precursors and

Full text of DOI:10.1021/acscatal.6b00396

Research Article
pubs.acs.org/acscatalysis
Tailoring High-Performance Pd Catalysts for Chemoselective
Hydrogenation Reactions via Optimizing the Parameters of the
Double-Flame Spray Pyrolysis
Kyung Duk Kim,†,¶ Suman Pokhrel,‡,¶ Zichun Wang,† Huajuan Ling,† Cuifeng Zhou,† Zongwen Liu,†
Michael Hunger,*,§ Lutz Madler,*,‡ and Jun Huang*,†
̈
†Laboratory for Catalysis Engineering, School of Chemical and Biomolecular Engineering, The University of Sydney, Sydney, New
South Wales 2006, Australia
‡Foundation Institute of Materials Science (IWT), Department of Production Engineering, University of Bremen, 28359 Bremen,
Germany
§Institute of Chemical Technology, University of Stuttgart, 70550 Stuttgart, Germany
S
* Supporting Information
ABSTRACT: Tuning the chemical composition during the synthesis is a widely used method to
control the activity of catalysts. Here, we reported an alternative synthesis strategy to tune the
catalytic properties of nanocatalysts without changing their precursors and compositions. We
synthesized a series of Pd catalysts on the most popular SiO2-, Al2O3-, and silica−alumina supports
using the double-flame spray pyrolysis (FSP) technique. It was observed that various flow rates used
for the synthesis of catalysts with the same composition affected the formation of the catalyst
particles and their structures to further tune the surface acidity due to the correlation between
acidity and structure, but the flow rates did not influence the electronic properties of Pd particles. It
was observed that surface OH groups could associate Pd for the hydrogenation, but Lewis acid sites
could not, as Pd/SA-30 and Pd/SiO2 showed much higher activity than Pd/Al2O3 for the same Pd
size and surface properties. For Pd catalysts with Brønsted acid sites (silica−alumina) or weak/
nonacidic SiOH groups (SiO2), their catalytic performance for the chemoselective hydrogenation of
acetophenone was obviously enhanced by tuning the surface OH groups via changing the flow rates
for the same precursor solution during this ultrafast synthesis.
KEYWORDS: palladium catalysts, silica/alumina supports, double-flame spray pyrolysis, structure−activity correlation,
chemoselective hydrogenation
As reviewed earlier,3 as popular supports, various transition,
rare earth, and basic and acidic metal oxides with special surface
properties have been used as efficient supports for metal
nanocatalysts according to the target reactions. For the
INTRODUCTION
■
Supported metal catalysts are widely employed in various
reactions, including oxidation, catalytic reforming, C−C
coupling, and, especially, the chemoselective hydrogenation1,2
hydrogenation of alcohols and ketones, the surface acidity of
of aldehyde and ketones to unsaturated alcohols. The
the supports was found to be associated with the metal particles
chemoselective hydrogenation of organic compounds including
for the higher reactant conversion and the product selectivity
carbonyl bonds is a demanding task due to the thermodynam-
during the reactions.1,8−10,13 The acid sites on the supports
ical preference of the CC hydrogenation compared with C
were proposed to withdraw the surface electrons from metal
O bonds.3 However, it is an important reaction both in organic
particles and induce an ionic effect from the supports to the
syntheses for industrial chemicals such as fragrances and drugs
metal particles, which change the catalytic properties on the
metal surface and influence the reaction process.12,14−16
and in bio-oils upgrading to fuel for the global sustainable
development.4−7 For enhancing the catalytic activity and
Zeolites and amorphous silica−alumina based on low-cost
selectivity, the metal particle size and shape, chemical
compositions and physical structures, and metal−support
interactions are important for the various synthesis methods
and preparation procedures.2 It was reported recently that the
surface properties of supports not only associate the metal sites
for the multifunctional catalysts but also optimize the size,
shape, dispersity, and electronic properties of the metal
particles.8−12
and naturally abundant silica and alumina resources are widely
used solid acids and acid supports in chemical industry.5,17 The
surface acidity is generally tuned via changing the Si/Al ratio
during the synthesis. Then, the supported metal catalysts will
be produced after postloading metal precursors on these acidic
Received: February 6, 2016
© XXXX American Chemical Society
2372
DOI: 10.1021/acscatal.6b00396
ACS Catal. 2016, 6, 2372−2381

ACS Catalysis
Research Article
supports. It is a time-consuming process and takes days to
obtain supported catalysts. In addition, the formation of the
metal particles could also be influenced by the supports’ acidity
due to interactions between metal precursors and surface acid
sites. Thus, it is hard to produce uniform metal particles on the
supports. Recently, an ultrafast synthesis method based on the
single-flame spray pyrolysis (FSP) has been used to produce
supported metal catalysts in microseconds directly from their
mixed metals and support precursors.1,9,10,18,19 However, some
metal particles may be inserted inside the bulk of the support
and may not be available for the surface reactants. In addition,
the metal particle size was strongly influenced by the support
acidity, making it hard to control their catalytic performance
due to the size-induced catalytic effects.
Most recently, we developed supported Pd metal catalysts via
decoupling of the support and metal particle formation using
the double-flame spray pyrolysis (double FSP, one spray for
support’ precursors and another spray for the metal precursor),
in which the uniform metal particle size and catalytic properties
have been obtained with various support acidities. Therefore,
these catalysts kept high chemoselectivity during the
acetophenone hydrogenation, where the conversion increased
5 times due to the enhanced acidity of the supports.13 Similar
to most of the current methods, the acidity of the supports was
tuned via the Si/Al ratio in the support precursors. It has to be
noticed that the FSP is a continuous process to produce
catalysts combining ultrafast nucleation, calcination, and
aggregation under flame conditions. The operation parameters,
such as gas or precursor flow rate, could affect the formation of
catalyst particles and their structures to further tune the surface
acidity due to the correlation between acidity and structure. It
may act as an alternative method to tune support acidity and
catalyst properties without changing the precursors.
into two independent nozzles (placed 11 cm apart) directed at
an angle of 20° to the base, where the aerosol streams were
homogeneously mixed at a distance of 16 cm perpendicular to
the base as described in our earlier works.13,20 Two syringe
pumps supplied the liquid precursors separately through two
independent nozzles for combustion, and each flame had a
separate set of fuel and O2 (1.5 L/min/3.2 L/min) flow
combinations. For monitoring the Pd particles on the support
surface, different O2 flow rate (7, 5, 3 mL/min) and precursor
flow rate (3, 5, 7 mL/min) were injected into the flame. The
nomenclature of the 5% Pd/silica−alumina powders is defined
as Pd/X-(m/n), where X is the support composition (Al2O3,
SiO2, silica−alumina (SA-30 with Si/Al = 70/30)) and (m/n) is
the indication of the oxygen flow rate for the Pd flame (m)
divided by the precursor flow rate (n) (7/3, 5/5, 3/7). For
example, to obtain Pd/SiO2 (7/3), 50 mL of 2.4 mM Pd2+
solution was injected into one nozzle with an oxygen flow rate
of 7 mL/min, and a flow rate of 3 mL/min was applied. At the
same time, 50 mL of 0.5 M Si4+ precursor solution was injected
into the other one at the same oxygen (7 mL/min) and
precursor flow rate (3 mL/min).
Physical Characterization. BET Measurements. Nitrogen
adsorption−desorption measurements (BET) were performed
at 77 K using a 4000e NOVA system to determine the specific
surface areas. Before the measurements, the samples were
degassed at 200 °C under vacuum. The primary particle size
and specific surface areas were determined by BET measure-
ment and summarized in columns 1 and 2 of Table 1. The BET
surface area measurement is related to the average equivalent
primary particle size by the equation21 dBET = 6000/(ρp·SA) (in
nm), where dBET is the average diameter of a spherical particle,
SA represents the measured surface area of the powder in m2/g,
and ρp is the theoretical density in g/cm3.
In the present investigation, we synthesized a series of SiO2-,
Al2O3-, and silica−alumina supported Pd catalysts using the
double FSP technique with different oxygen and precursor flow
rates. It was observed that various flow rates used for the
synthesis of catalysts with the same composition influenced the
support acidity but did not affect the electronic properties of
the Pd particles. The catalysts were analyzed using BET, X-ray
diffraction (XRD), HRTEM, XPS, and solid-state NMR
spectroscopy for investigating and demonstrating the structure,
Pd electron properties, support acidity, and the local framework
atom coordination. The hydrogenation of acetophenone was
utilized as a reference test reaction for the competitive
hydrogenation between carbonyl and aromatic groups. In
addition, the selective hydrogenation of acetophenone has
practical relevance for the production of phenylethanol, which
is used in perfumery products and ethylbenzene for polymer
and fuel industry.5
X-ray Characterization. Siemens X-ray Diffractometer
(XRD-D5000) with Cu Kα radiation operated at 40 kV with
a current of 30 mA, at a scan range from 1 to 70°, and with
continuous scanning mode at a rate of 2° min−1.
TEM Characterization. For preparing the TEM samples, a
small amount of the sample material was dispersed in 5 mL of
ethanol in an ultrasonic bath and sonicated for 60 min. A drop
of the colloidal suspension was placed on a copper grid coated
with carbon film. The samples were dried in ambient air. Large
regions of the sample were scanned before the investigation of
the particle morphology. The low magnification TEM, the
corresponding selected area electron diffractions (SAED), and
high-resolution transmission electron microscopy (HRTEM)
studies were carried out using an FEI Titan 80/300 ST unit.
The instrument was equipped with a Cs corrector for spherical
aberration of the objective lens. For imaging, a high-angle
annular dark field (HAADF) detector and a GATAN
postcolumn imaging filter was used. In all investigations, the
microscope was operated at an acceleration voltage of 300 kV.
CO Chemisorption. The Pd dispersion was investigated by
CO-pulse chemisorption on a Micromeritics Autochem II 2920
unit. The off-gas was analyzed via a mass spectrometer to derive
the amount of chemisorbed CO. The number of exposed Pd
metal atoms was calculated on the basis of a chemisorption
stoichiometry of Pd/CO = 2. Prior to the measurements, all
samples were freshly reduced for 30 min at 400 °C under
flowing hydrogen (5 mol % in Ar, 20 mL min−1) and then
cooled down to 40 °C during flushing with He.
EXPERIMENTAL SECTION
■
Catalyst Preparation by the Double FSP Synthesis
Method. Metallorganic-based precursors, such as palladium-
(II) acetate trimer (Pd 47.36%, Johnson Matthey Catalysts),
aluminum secondary butoxide, tetraethyl orthosilicate (TEOS)
(99.9% purity, Sigma-Adrich), were used for the synthesis of
the catalysts. For the synthesis of 5%Pd/Al2O3 and 5%Pd/SiO2
nanoparticles, aluminum secondary butoxide and TEOS were
dissolved in xylene (0.5 M concentration by metal). In the case
of 5%Pd/silica−alumina (Si/Al = 70/30), aluminum and silicon
precursor were mixed in the ratio of 3:7 by volume. Then, the
support precursor and Pd(OAc)2 dissolved in xylene were fed
XPS Characterization. The Pd surface analysis was
performed by XPS (ESCALAB250Xi spectrometer, Thermo
2373
DOI: 10.1021/acscatal.6b00396
ACS Catal. 2016, 6, 2372−2381

ACS Catalysis
Research Article
Table 1. Properties of the Supported Pd Catalysts Synthesized by the Double FSP Method Using Different Oxygen and
a
Precursor Flow Rates
b
ABET
primary particle
size (nm)
amount of SiOH groups amount of Brønsted acid sites content of Brønsted acid
AlIV
(at.%)
AlV (at.
%)
AlVI
(at.%)
c
samples
m2/g
(mmol/g)
(mmol/g)
sites (%)
Pd/SA-30
(7/3)
303
195
145
118
174
207
227
159
126
5.7
0.86
7.6 × 10−2
1.4
38.2
38.1
30.5
-
29.7
32.1
27.2
29.4
-
Pd/SA-30
(5/5)
8.8
0.44
0.10
1.2
1.5
1.1
-
4.7 × 10−2
10.74
34.7
Pd/SA-30
(3/7)
11.8
16.3
11.1
9.3
0.3 × 10−2
8.73
40.1
Pd/SiO2
(7/3)
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
Pd/SiO2
(5/5)
-
-
Pd/SiO2
(3/7)
-
-
Pd/Al2O3
(7/3)
7.1
-
-
Pd/Al2O3
(5/5)
10.1
12.7
-
34.5
32.2
66.5
67.8
Pd/Al2O3
(3/7)
-
a
b
BET surface areas, primary particle size, concentration of OH group and Brønsted acid sites. ABET is specific surface area determined by nitrogen
c
adsorption/desorption isotherms. Brønsted acidic OH (mol %) = (number of Brønsted acidic OH groups/total number of OH groups on supports)
× 100%.
Scientific, U.K.) employing a Al Kα (1486.68 eV) X-ray source
operated at 15.2 kV and 164 W (10.8 mA). The photoelectron
takeoff angle was 90° measured with respect to the surface of
the sample. The typical operating pressure in the vacuum
chamber was 2 × 10−9 mbar.
delay of 20 s were applied. To separate the different signals and
for the quantitative evaluation of spectra, the data were
processed using the Bruker software WINNMR and WINFIT.
Catalytic Hydrogenation. Prior to the hydrogenation
reactions, the catalyst materials were reduced in a quartz reactor
under a hydrogen flow rate of 50 mL/min for 1 h at 573 K.
After the materials were cooled to room temperature under a
nitrogen flow rate of 50 mL/min, the reduced catalysts were
immediately transferred to a 50 mL stainless steel autoclave,
which was purged with nitrogen followed by hydrogen for three
times. Acetophenone (60 mg) was dissolved in hexane (6 mL)
and added to the autoclave. The molar ratio of Pd on the
catalyst to reactant was 1:101. The reaction was carried out at a
hydrogen pressure of 3 bar at 15 °C under continuous stirring
for 3 h. The product mixtures were collected after reaction
times of 10, 20, 30, 60, 90, 120, and 180 min. The conversion
and composition of the product mixtures were analyzed using a
Shimadzu GCMS-QP2010 Ultra with a RTX-5MS capillary
column (30 m × 0.25 mm × 0.25 μm) and quantified by a
Shimadzu GC-FID equipped with a RTX-5 capillary column
(30 m × 0.32 mm × 3 μm). Bicyclohexyl (0.1 mmol) was used
as an internal standard. The selectivity to a specific product i
(Si) was calculated as Si (%) = (Ci)/[(CAPh)0-(CAPh)], where
(Ci) is the molar concentration of the products, such as 1-
phenylehtanol (PhE), and (CAPh)0 and (CAPh) correspond to
the molar concentration of acetophenone before and after the
reaction, respectively.
Solid-State NMR Studies. For the 27Al and 29Si MAS
NMR investigations, all samples were exposed to the saturated
vapor of a Ca(NO3)2 solution at ambient temperature
1
overnight in a desiccator for full hydration. Before the H
MAS NMR experiments, the samples were dehydrated at 723 K
in vacuum with a pressure less than 10−2 bar for 12 h in glass
tubes. The dehydrated samples were sealed or utilized for in
situ loading with ammonia on a vacuum line, followed by
evacuation at 393 K for 1 h (for ammonia) to remove
physisorbed molecules. Subsequently, the samples were trans-
ferred into the MAS NMR rotors under dry nitrogen gas inside
a glovebox.
1H, 27Al, and 13C MAS NMR investigations were carried out
on a Bruker Avance III 400 WB spectrometer at resonance
frequencies of 400.1, 104.3, and 100.6 MHz with the sample
spinning rate of 8 kHz using 4 mm MAS rotors. Spectra were
recorded after single-pulse π/2 and π/6 excitation with
1
repetition times of 20 and 0.5 s for studying H and 27Al
nuclei, respectively. The 2D 27Al MQMAS NMR spectra were
recorded at a Bruker AVANCE III 700 MHz spectrometer at
the resonance frequrency of 182.5 MHz using a three-pulse z-
filter pulse sequence with pulse lengths of 3.0, 0.8, and 8.0 μs,
the repetition time of 200 ms, 64 to 128 t1 increments, and the
1
RESULTS AND DISCUSSION
sample spinning rate of 20 kHz. Quantitative H MAS NMR
■
measurements were performed using zeolite H,Na−Y (35%
ion-exchanged) as an external intensity standard, which
contains 58.5 mg zeolite H,Na−Y with 1.776 mmol OH/g.
13C Cross-polarization (CP) MAS NMR spectra were recorded
with the contact time of 4 ms and the repetition time of 4 s. 29Si
MAS NMR experiments were performed with the same
spectrometer at the resonance frequency of 79.5 MHz and
the sample spinning rate of 4 kHz using 7 mm MAS rotors. For
recording the 29Si MAS NMR spectra, single-pulse π/2
excitation and high-power proton decoupling with a recycle
Catalyst Morphology. The BET surface areas of the
prepared catalysts are summarized in Table 1. The BET surface
area of silica−alumina or alumina supported catalysts increased
from 145 to 303 m2/g or 126 and 227 m2/g with a decreasing
precursor flow rate, consistent with flame-made vanadia/titania
in a previous study.22 Due to the shorter residence time of
precursor in the flame and the rapid atomic nucleation at high-
temperatures, the particle coagulation and coalescence are
caused at high precursor flow rate, reducing the overall specific
surface area and increasing primary particle size in the flame.23
2374
DOI: 10.1021/acscatal.6b00396
ACS Catal. 2016, 6, 2372−2381

ACS Catalysis
Research Article
In the case of silica-supported catalyst, however, the surface
area was increased from 118 to 207 m2/g with an increasing
precursor flow rate. Accordingly, the average particle size was
decreased from 16.3 to 9.3 nm. The possible reason is that the
high precursor velocity during the injection into the flame could
promote the higher dispersion of Si atoms and hinder the
coagulation. Unlike the longer residence time required for the
formation of Al2O3 clusters (pure Al precursor) or the Si−O−
Al network (mixed Si and Al precursors), the short residence
time or the high density of Si atoms in the system may be
adequate for the formation of small amorphous silica particles.
It was partially similar to the previous observation for FSP-
made Pd/silica−alumina that the surface areas of Pd/SA
materials increased with the silica content.10,13
was in the range of 1.8−3.5 nm. HRTEM images revealed the
spherical Pd particle morphology with a similar particle size
dispersed on diverse supports. The mean particle size for Pd/
SA-30, Pd/Al2O3, and Pd/SiO2 synthesized at the highest
oxygen flow (7 mL/min) showed mean particles sizes of 2.2,
1.9, and 1.8 nm, respectively. When decreasing the oxygen flow
rate and increasing precursor flow rate, the average Pd particle
size was slightly increased up to 2.4−2.6 nm and 3.0−3.5 nm at
flow rates of 5/5 and 3/7, respectively. No strongly aggregated
Pd particles were observed. The lattice structure of the Pd
particles was not affected by the oxygen flow and precursor flow
rate.
Local Structure and Surface Properties of Supports.
The above-mentioned investigations have shown that the
oxygen and precursor flow rates during the double FSP
synthesis method obviously influence the particle size of the
supports. As the surface properties of the support materials
mainly depend on their local structure, the solid-state NMR
spectroscopy has been used to clarify the Al and Si coordination
and to characterize the properties of the surface active sites.
According to the 27Al MAS NMR spectrum in Figure 3d, the Al
coordination in alumina has an AlIV to AlVI ratio of about 32 to
68% (similar as the γ-Al2O313,25).
Similarly, the XRD patterns of all prepared catalysts in Figure
1 show that the change of the oxygen and precursor flow ratios
For pure silica utilized as support material, 29Si MAS NMR
spectroscopy was employed to investigate the silicon
coordination. The 29Si MAS NMR signals at −110 and −100
ppm in Figure 4 were assigned to dominating Q4 species
(Si(OSi)4) and Q3 species (Si(OSi)3OH), respectively, along
with the Q2 species (Si(OSi)2(OH)2), occurring at −90 ppm.26
Among them, Q3 and Q2 species are the origin of the surface
SiOH groups on silica materials.10,27 For a lower precursor flow
rate and a higher oxygen flow rate, the signal of Q3 species was
increased in comparison with the signal of Q4 species, which
indicates that changing the precursor/oxygen flow rates
enhances the formation of Q3 species and could change the
surface acidity, although the amount of silicon precursor is the
same for all the Pd/SiO2 catalysts under study.
Figure 1. XRD patterns of different types of supported Pd-
nanoparticle catalysts synthesized with different flow rates using the
double-flame spray pyrolysis method.
After dehydration of the Pd/SiO2 catalysts at 400 °C, the 1H
MAS NMR spectra in Figure 5 consisted of a strong signal at
ca. 1.8 ppm, which was assigned to silanol groups, and a
shoulder at ca. 2.7 ppm resulting from hydrogen-bonded silanol
groups, which were also observed in previous studies.10,26 The
hydrogen bonds between SiOH groups have been observed for
all Pd/SiO2 materials and were found to increase with an
increasing fraction of Q3 species (Figure 4 and Figure 5).
Considering the effect of the oxygen and precursor flow rates
on the formation of Q3 species (vide supra), changing these
synthesis parameters can also tune the properties of the surface
SiOH groups. Although Pd/SiO2 (7/3) has the highest fraction
of Q3 species contributing to SiOH groups, their total number
was limited by its lowest surface area (Table 1). Pd/SiO2 (5/5)
showed the highest amount of SiOH groups.
Interestingly, both the Al and Si coordinations of the silica−
alumina supports have been obviously changed with modifying
the synthesis parameters, even though the chemical composi-
tions were the same. As shown in the 27Al MAS NMR spectra
of Pd/SA-30 in Figure 3a−c, the penta-coordinated (AlV)
aluminum at ca. 28 ppm increased while the tetrahedrally
coordinated (AlIV) aluminum and octahedrally coordinated
(AlVI) aluminum at ca. 54−55 and 0.6−1.6 ppm were slightly
decreased with decreasing the oxygen flow rates. As evidenced
by the two-dimensional 27Al MQMAS NMR spectrum of the
hydrated samples in Figures 6a−c, up to three peaks occur,
had nearly no influence on the bulk structure of Pd/SiO2, a
slight influence on the Pd/SA-30, but strongly changes the
structure of Pd/Al2O3. Obviously, the residence time is
significant for the formation of alumina and silica−alumina
supports rather than for the formation of silica. Silica-
containing catalysts (Pd/SA-30 and Pd/SiO2) were highly
amorphous in nature. A broad peak at 2θ = 25° revealed the
amorphous nature of silica.24 Only a single reflection peak of 2θ
= 35° was detected for Pd/SA-30 (3/7) due to the formation of
small alumina clusters. The lower oxygen flow rate and the
longer particle residence time influenced the segregation of the
alumina clusters during the synthesis. For Pd/Al2O3, the
diffraction patterns were similar to the catalysts prepared under
various flow rates and similar to the previously observed for
flame-derived aluminum supported Pd catalysts.19 However,
the reflection intensity decreased with increasing oxygen flow
rates. Of course, the smaller alumina particles, which were
synthesized under this flow rate made them undetectable by
XRD. For all catalysts, the Pd particles were too small to be
observed by XRD.
As shown in the TEM and HRTEM images in Figure 2, very
fine and uniform Pd particles were homogeneously dispersed
on the all supports. The size distribution of Pd particles was
similar (2.2−3.0 nm) on all supports, as shown in Figure S2 of
Supporting Information. The average size of the Pd catalysts
2375
DOI: 10.1021/acscatal.6b00396
ACS Catal. 2016, 6, 2372−2381

ACS Catalysis
Research Article
Figure 2. TEM and HRTEM of Pd/SA-30 (7/3) (a,d,g), Pd/SiO2 (7/3) (b,e,h), and Pd/Al2O3 (7/3) (c,f,i).
Figure 3. 27Al MAS NMR spectra of Al containing supported Pd
catalysts.
which are caused by three well-separated signals with isotropic
chemical shifts of ca. 0, 30, and 55 ppm. The above-mentioned
chemical shift values are typical for octahedrally coordinated
(AlVI), pentacoordinated (AlV), and tetrahedrally coordinated
aluminum atoms (AlIV), respectively. As previously discussed,
AlV species may coordinate to Si−O species and form a
Figure 4. 29Si MAS NMR spectra of dehydrated Pd/SiO2 with m/n =
3/7 (a), 5/5 (b), and 7/3 (c).
2376
DOI: 10.1021/acscatal.6b00396
ACS Catal. 2016, 6, 2372−2381

ACS Catalysis
Research Article
Figure 7. 29Si MAS NMR spectra of Pd/SA-30 with m/n = 3/7 (a)
and 7/3 (b).
1H MAS NMR spectroscopy was carried out to investigate
the hydroxyl coverage and the acidity of the Pd/SA-30 catalysts.
The H MAS NMR spectra in Figure 8 are dominated by a
1
Figure 5. H MAS NMR spectra of dehydrated Pd/SiO2 with m/n =
3/7 (a), 5/5 (b) and 7/3 (c).
1
pseudobridging bond with an involved silanol group.13,28 It was
also confirmed by the 29Si MAS NMR spectra of Pd/SA-30 in
Figure 7 that the center of gravity of the 29Si MAS NMR signals
was slightly shifted to lower field with increasing oxygen flow
rate and decreasing the precursor flow rate, which hints to a
slightly enhanced incorporation of aluminum into the silica
network.17 If an increasing number of AlIV and AlV species are
located near surface SiOH groups (Pd/SA-30 catalysts), then
probability of pseudobridges between AlV species and these
silanols should also increase. In this case, the electron-
withdrawing character of AlV species would lead to an
enhancement of the acid strength of neighboring surface
SiOH groups having the potential for acting as Brønsted acid
sites.10,13,26
strong signal at δ1H = 1.8 ppm, which is due to surface silanol
groups. A signal of surface AlOH groups at δ1H = 0.1 ppm was
only detected in Figure 8a. The formation of AlOH groups at
low oxygen and high precursor flow rate could be caused by the
formation of a large amount of surface aluminum species. On
the other hand, the 27Al MAS NMR spectra showed only the
slight change of the fraction of three Al species (Figure 3).
Obviously, the change of the flow rates during the double FSP
synthesis did not cause a significant change of Al coordination
but changed their distribution in the support particles (i.e.,
caused an enrichment of these species at the outer surface in
the vicinity of SiOH groups). By adsorption of ammonia as
probe molecules, it was found that only small amount of
Brønsted acidic SiOH groups (BAS) exist on Pd/SA-30 (3/7),
Figure 6. 2D 27Al MQMAS NMR spectra of Al containing supported Pd catalysts.
2377
DOI: 10.1021/acscatal.6b00396
ACS Catal. 2016, 6, 2372−2381

ACS Catalysis
Research Article
Figure 9. Binding energies of the Pd 3d5/2 and Pd 3d3/2 lines of
supported Pd catalysts synthesized by the double FSP method.
induced the ionic effects for the metal particles. The Pd binding
energies of Pd/SA-30 catalysts were shifted to 334.3 and 339.7
eV, respectively, from 334.8 and 340.2 eV for Pd/SiO2 and Pd/
Al2O3. Pd/SA-30 (3/7) had relatively lower amount of BAS
compared with other two Pd/SA-30 catalysts, but showed the
similar ionic effects. It was similar as the previous report that
the density of BAS on the supports could not change the ionic
effects for the supported metal particles.34 The Pd particle size
was slightly changed when changing the oxygen and precursor
flow ratio from 7/3 to 3/7 (means size from 1.8 to 3.0 nm);
however, no change for the electronic properties has been
observed on those Pd catalysts.
Figure 8. 1H MAS NMR spectra of Pd/SA-30 with m/n = 3/7 (a), 5/5
(b) and 7/3 (c), recorded before (bottom) and after (top) loading
with NH3 and subsequent evacuation of physisorbed NH3 at 373 K for
1 h.
which are able to protonate ammonia to ammonium ions
1
causing a H MAS NMR signal at ca. 7 ppm (Figure 8a). For
Pd/SA-30 (5/5) and Pd/SA-30 (7/3), however, a significant
amount of BAS was observed for both catalysts. Although all
Pd/SA-30 catalysts have the same support composition (silica
70: aluminum 30), these materials exhibit different surface
properties depending on the conditions of their double FSP
synthesis, especially the oxygen and precursor flow rates. By
quantitative evaluation of the 1H MAS NMR signals of the OH
groups before (Figure 8, bottom) and after NH3 adsorption
(Figure 8, top), the total number of surface OH groups and the
amount of Brønsted acid sites were determined. Both these
values increased from the Pd/SA-30 (3/7) to Pd/SA-30 (7/3)
with increasing oxygen flow rate and decreasing precursor flow
rate (see Table 1, columns 4 and 5, respectively). Similar as in
previous studies,10,13 the Pd/SA-30 (7/3) catalyst had the
highest amount of BAS due to the highest amount of total
surface SiOH groups (Figure 1 and Table 1).
Surface Electronic Property of Pd Particles. The
catalytic property of surface metal particles is based on their
surface electronic properties, which are generally determined by
XPS.29,30 As shown in Figure 9, the binding energies of Pd
nanoparticle on silica and alumina supports were similar at
334.8 (Pd 3d5/2) and 340.2 eV (Pd 3d3/2). It was widely
reported that the surface acid sites could withdraw the electrons
from the metal surface and change the surface electronic
properties of metal particles via ionic effects.10,31−34 Pd/SiO2
has only silanol groups including very weakly acidic SiOH
groups from Q3 species. The acidity of SiOH groups is not
strong enough to withdraw the electron from the metal surface
and its binding energies were identical to those of palladium
metal. Interestingly, Pd/Al2O3 catalysts (Lewis acid sites on the
supports) showed similar binding energies like Pd/SiO2
catalysts. Therefore, the Lewis acid sites on the supports
might not influence the electronic properties of metal
nanoparticles. However, the BAS sites on supports obviously
Chemoselective Hydrogenation of Acetophenone. As
discussed above, Pd supported nanocatalysts via the decoupling
of the support and metal particle formation using the double
FSP synthesis method exhibited the similar particle size of Pd
particles on different supports (silica, alumina, or silica−
alumina) at the same flow ratio of the precursors (7/3, 5/5, and
3/7, respectively). After testing the chemoselective hydro-
genation of acetophenone (Aph), all catalysts offered the
similar chemoselectivity for the hydrogenation of the CO
group to form phenylethanol (PE) and ethylbenzene (EB) as
shown in Table 2. It was also observed previously4,13 that the
chemoselectivity mainly relies on the Pd surface properties, as
the hydrogenation reaction occurs on the metal surface. As
shown in Figure 10, the support properties have significant
effects for the catalytic reactions. Pd particles on silica−alumina
(containing dominant BAS with mediate strength) showed the
much higher conversion of Aph compared to Pd particles with
the same size distribution on silica (neutral SiOH) and alumina
(strong LAS). As summarized in Table 2, the TOFs were 1.1 ×
10−2 − 5.1 × 10−2, 0.9 × 10−2 − 1.6 × 10−2, and 0.6 × 10−2
−
0.7 × 10−2 s−1 for Pd/SA-30, Pd/SiO2, and Pd/Al2O3,
respectively. It was reported that the support acidity could
withdraw the electrons from the metal surface and enhance the
catalytic property of catalysts.10,29,32,34 The present study
demonstrates that not all acid sites can contribute the ionic
effects for the metal surface. Combined with the above XPS
investigation, it is concluded that only BAS on support can
induce the ionic effect for the metal surface and enhance the
catalytic performance of metal in the hydrogenation. The
strong LAS of the Al2O3 support are not able to withdraw the
electrons from the metal surface and result in the ionic effects.
It was also confirmed by the above-mentioned XPS study (in
the section Surface Electronic Property of Pd Particles) that the
2378
DOI: 10.1021/acscatal.6b00396
ACS Catal. 2016, 6, 2372−2381

ACS Catalysis
Research Article
Figure 10. Conversion of acetophenone (Aph) over Pd catalysts as a
function of the reaction time at a hydrogen pressure of 3 bar and at 15
°C.
electronic property of Pd on Al2O3 was the same as it on
neutral SiO2 with SiOH groups only. Interestingly, Pd/SiO2
obviously showed the higher activity during the hydrogenation
and a higher Aph conversion compared to Pd/Al2O3 (shown in
Figure 10). It indicates that neutral surface hydroxyl groups,
such as SiOH groups, performed much better than strong LAS,
such as surface Al cations, associated with Pd particles for the
Aph hydrogenation. The possible reason is that the flexible
surface SiOH groups easily interact with the metal particles on
the surface, which are hardly available for LAS.
Figure 11. Conversion of acetophenone (Aph) over Pd catalysts as a
function of the reaction time at a hydrogen pressure of 3 bar and at 15
°C.
The Pd/SA catalysts containing BAS showed a much higher
activity in the hydrogenation reactions (apart from Pd/SA-30
(3/7), containing the lower BAS) compared with Pd/SiO2 and
Pd/Al2O3 (Figure 11). It is similar to a previous study,34 which
showed that increasing the density of uncovered BAS on the
support could significantly enhance the conversion of Aph
hydrogenation on supported Pt catalysts but keep the same and
high chemoselectivity. When most of the Aph converted to PE,
some of PE products would be further converted to EB as
shown in Figure S3. The selectivity of EB was increased
proportionally with the consumption of PE. Because nearly no
hydrogenation of the aromatic ring has been observed during
the reaction process, it is certificated that the catalysts prepared
via the double FSP synthesis method has desired catalytic
properties. The selectivity was determined at 30% conversion of
Aph to obtain comparable results. As shown in Table 2 and
Figure S3a,b,c in Supporting Information, all Pd/SA-30
catalysts revealed a similar chemoselectivity in the Aph
hydrogenation, generating only carbonyl hydrogenated prod-
ucts, such as PE (selectivity of 88.2−98%) and EB (selectivity
of 2−11.8%). Additionally, there was no selectivity to
hydrogenation of the aromatic ring (cyclohexylmethylketone
(CMK), ethylcyclohexane (EC), and cyclohexylethonol (CE)).
The BAS on the silica−alumina support of the Pd/SA-30
catalysts could have more surface SiOH groups with enhanced
acid strength allowing them to react with the C−OH groups of
PE after completed reaction of Aph due to the strong electron
donor property, which should enhance their interaction with
dissociated hydrogen.10 All Pd/SA-30 catalysts showed similar
chemoselectivity in the hydrogenation of Aph and exhibited a
preferred hydrogenation activity for carbonyl groups rather
than for aromatic groups. This could be explained by the similar
surface electronic properties of Pd particles as observed in XPS
investigation, which was similar to our earlier work.13
In addition, supported Pd catalysts with the same
composition prepared by the double FSP technique with
different oxygen and precursor flow rates showed different
support properties (such as support size, variable density, and
type of surface OH groups) but the same electronic properties
of Pd particles (as shown by the XPS investigation in the
section Surface Electronic Property of Pd Particles). The same
surface properties of Pd catalysts contributed the similar
product chemoselectivity during Aph hydrogenation as shown
in Figure S3. As described above, LAS on Al2O3 cannot
associate Pd for the hydrogenation and promote the reaction
performance. Even more LAS are available on the surface when
Al2O3 support size decreased from 12.7 to 7.1 nm (surface area
increased from 126 to 227 m2/g), both Aph conversion and
product selectivity nearly had not changed during the
hydrogenation (shown in Figure 11 and Figure S3). For the
hydrogenation on Pd/SiO2, the catalyst containing the highest
amount of SiOH groups (Pd/SiO2 (5/5)) showed the highest
activity during the hydrogenation reaction as shown in Figure
11. Because the same surface activity was observed for Pd on
SiO2, it could be concluded that the flexible surface OH groups
can associate the Pd for the catalytic hydrogenation reactions. It
was reported that the SiOH groups of Q3 silicon species may
contribute to the weak acidity of SiO2 surfaces.10,27 Pd/SiO2
(7/3) has a similar total number of SiOH groups but more Q3
silicon species than Pd/SiO2 (3/7). The existence of more
acidic SiOH groups on Pd/SiO2 (7/3) could enhance its
catalytic performance as shown in Figure 11, and the TOF was,
therefore, increased from 0.7 × 10−2 to 0.9 × 10−2 s−1.
2379
DOI: 10.1021/acscatal.6b00396
ACS Catal. 2016, 6, 2372−2381

ACS Catalysis
Research Article
Table 2. Conversion, Product Selectivities, and Turnover Frequencies (TOF) for the Chemoselective Hydrogenation of
Acetophenone (Aph) over Pd Catalysts Synthesized by the Double FSP Method with Different Oxygen and Precursor Flow
Rates
a
b
b
b
c
d
sample
conversion
%
SPhE
%
SEB
%
SCMK
%
dPd
%
TOFs (s−1
)
Pd/SA-30(7/3)
Pd/SA-30(5/5)
Pd/SA-30(3/7)
Pd/SiO2 (7/3)
Pd/SiO2 (5/5)
Pd/SiO2 (3/7)
Pd/Al2O3(7/3)
Pd/Al2O3 (5/5)
Pd/Al2O3 (3/7)
100
100
61
88.2
91
11.8
9
-
-
-
-
-
-
50
47
42
52
42
39
50
44
34
5.1 × 10−2
3.9 × 10−2
1.1 × 10−2
0.9 × 10−2
1.6 × 10−2
0.7 × 10−2
0.7 × 10−2
0.6 × 10−2
0.6 × 10−2
97.7
100
98.8
100
97.6
98.3
98.2
2.3
0
53
83
1.2
0
30
39
0
2.4
1.7
1.8
31
0
26
0
a
b
Conversion is at the reaction time of 3 h. SphE, SEB, and SCMK are the selectivity to PhE, EB and CMK, respectively, obtained at the Aph conversion
c
d
of 30% Pd dispersion is determined by CO chemisorption. TOFs for Pd/supported catalysts were calculated based on Pd dispersion obtained by
CO chemisorption at the Aph conversion of 30%
Combining the results of spectroscopic and catalytic character-
ization as well as the similar surface electronic properties could
indicate a similar hydrogen dissociation energy and adsorp-
tion−desorption balance for the hydrogenation of Aph.
Therefore, the reaction and chemoselectivities of Aph on all
of Pd/SA-30 catalysts under study should be similar.
leading to a decreased particle size of the alumina support and
to an enhancement of the amount of surface LAS could not
increase the catalystic performance for hydrogenation of Aph
on Pd/Al2O3 catalysts. All Pd/Al2O3 catalysts in this research
showed similar activity.
The present work has demonstrated that after selecting
suitable supports, it is possible to produce tunable nanocatalysts
with the same chemical composition but very different catalytic
performance via the ultrafast double FSP synthesis method,
which is promising for the development of custom-made
catalysts for high-performance reaction processes.
CONCLUSIONS
■
The double-flame spray pyrolysis was used to prepare Pd
nanocatalysts on three most popular supports (silica−alumina,
Al2O3, and SiO2) with the same precursor composition but
different ratios of the oxygen and precursor flow rates.
Changing the flow rates leads to changes of both the Al and
Si coordination in spite of the same chemical composition,
resulting in different structural and surface properties, such as
support size, as well as various densities and types of surface
OH group. In the case of silica−alumina-supported Pd
nanocatalysts (Pd/SA), the density of BAS was strongly
increased from 0.3 × 10−2 mmol/g up to 7.6 × 10−2 mmol/g
under the same chemical composition with the change of the
specific surface area and the coordination of aluminum species
in silica network, when the ratio of the oxygen and precursor
flow rates was changed from 3/7 to 7/3. This increased BAS
density strongly enhanced the catalytic performance in
chemoselective hydrogenation of acetophenone from a
conversion of 32% at 120 min on Pd/SA-30 (3/7) up to
100% at same time on Pd/SA-30 (7/3). All samples showed
similar selectivity due to the same surface electronic property as
confirmed by XPS. A similarly enhanced catalytic performance
has also been observed for the Pd/SiO2 catalysts after tuning
the ratio of the oxygen and precursor flow rate for the same
quality and quantity of the silicon precursor. Pd/SiO2 (5/5) has
the highest amount of SiOH groups on the support surface and
showed the highest activity of all Pd/SiO2 catalysts under study.
Pd/Al2O3 catalysts containing surface Lewis acid sites had a
similar Pd surface electronic property as the Pd/SiO2 catalysts
containing exclusively nonacidic SiOH groups. However, the
Aph conversion was much lower on Pd/Al2O3 than that on Pd/
SiO2 catalysts, and the Pd/SA-30 catalysts containing BAS
showed the highest activity. This observation indicates that BAS
and SiOH groups could associate and activate the Pd
nanoparticles for the hydrogenation reaction, whereas LAS
(even strong LAS as γ-Al2O3 in this research) could not
enhance the conversion of Aph. Therefore, changing the
different flow conditions during the double FSP synthesis
ASSOCIATED CONTENT
■
S
* Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acscatal.6b00396.
TEM images, Pd particle size distribution, and chemo-
selectivity (PDF)
AUTHOR INFORMATION
■
Corresponding Authors
*E-mail: michael.hunger@itc.uni-stuttgart.de (M.H.).
*E-mail: lmaedler@iwt.uni-bremen.de (L.M.).
*E-mail: jun.huang@sydney.edu.au (J.H.).
Author Contributions
¶These authors contributed equally to this work (K.D.K. and
S.P.).
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
J.H. acknowledges the financial supports from Australian
Research Council Discovery Projects (DP150103842), the
International Project Development Funding, and Faculty’s
MCR scheme at the University of Sydney, and M.H. thanks
Deutsche Forschungsgemeinschaft for financial support (INST
41/879-1 FUGG). S.P. and L.M. gratefully acknowledge the
financial support of the Deutsch Forschungsgemeinschaft
(DFG) within the SPP 1613 (MA 3333/6-1) and the Priority
Program "Partikel im Kontakt -Mikromechanik, Mikroprozess-
dynamik und Partikelkollektiv" (SPP 1486) under grants MA
3333/3.
2380
DOI: 10.1021/acscatal.6b00396
ACS Catal. 2016, 6, 2372−2381

ACS Catalysis
REFERENCES
Research Article
(31) Koningsberger, D. C.; de Graaf, J.; Mojet, B. L.; Ramaker, D. E.;
Miller, J. T. Appl. Catal., A 2000, 191, 205−220.
(32) Lepage, M.; Visser, T.; Soulimani, F.; Beale, A. M.; Iglesias-Juez,
A.; van der Eerden, A. M. J.; Weckhuysen, B. M. J. Phys. Chem. C 2008,
112, 9394−9404.
(33) Lepage, M.; Visser, T.; van der Eerden, A. M. J.; Soulimani, F.;
Weckhuysen, B. M. Vib. Spectrosc. 2008, 48, 92−100.
(34) Wang, Z. C.; Kim, K. D.; Zhou, C. F.; Chen, M. M.; Maeda, N.;
Liu, Z. W.; Shi, J.; Baiker, A.; Hunger, M.; Huang, J. Catal. Sci. Technol.
2015, 5, 2788−2797.
■
(1) Schimmoeller, B.; Hoxha, F.; Mallat, T.; Krumeich, F.; Pratsinis,
S. E.; Baiker, A. Appl. Catal., A 2010, 374, 48−57.
(2) White, R. J.; Luque, R.; Budarin, V. L.; Clark, J. H.; Macquarrie,
D. J. Chem. Soc. Rev. 2009, 38, 481−494.
(3) Maki-Arvela, P.; Hajek, J.; Salmi, T.; Murzin, D. Y. Appl. Catal., A
2005, 292, 1−49.
(4) Chen, M. M.; Maeda, N.; Baiker, A.; Huang, J. ACS Catal. 2012,
2, 2007−2013.
(5) Huber, G. W.; Iborra, S.; Corma, A. Chem. Rev. 2006, 106, 4044−
4098.
(6) Minnermann, M.; Grossmann, H. K.; Pokhrel, S.; Thiel, K.;
Hagelin-Weaver, H.; Baumer, M.; Madler, L. Catal. Today 2013, 214,
̈
90−99.
(7) Minnermann, M.; Pokhrel, S.; Thiel, K.; Henkel, R.; Birkenstock,
J.; Laurus, T.; Zargham, A.; Flege, J. I.; Zielasek, V.; Piskorska-
Hommel, E.; Falta, J.; Madler, L.; Baumer, M. J. Phys. Chem. C 2011,
̈
115, 1302−1310.
(8) Handjani, S.; Marceau, E.; Blanchard, J.; Krafft, J. M.; Che, M.;
Maki-Arvela, P.; Kumar, N.; Warna, J.; Murzin, D. Y. J. Catal. 2011,
282, 228−236.
(9) Hoxha, F.; Schimmoeller, B.; Cakl, Z.; Urakawa, A.; Mallat, T.;
Pratsinis, S. E.; Baiker, A. J. Catal. 2010, 271, 115−124.
(10) Huang, J.; Jiang, Y. J.; van Vegten, N.; Hunger, M.; Baiker, A. J.
Catal. 2011, 281, 352−360.
(11) Li, J. J.; Xu, X. Y.; Jiang, Z.; Hao, Z. P.; Hu, C. Environ. Sci.
Technol. 2005, 39, 1319−1323.
(12) Ramaker, D. E.; Oudenhuijzen, M. K.; Koningsberger, D. C. J.
Phys. Chem. B 2005, 109, 5608−5617.
(13) Wang, Z. C.; Pokhrel, S.; Chen, M. M.; Hunger, M.; Madler, L.;
̈
Huang, J. J. Catal. 2013, 302, 10−19.
(14) Ji, Y. Y.; van der Eerden, A. M. J.; Koot, V.; Kooyman, P. J.;
Meeldijk, J. D.; Weckhuysen, B. M.; Koningsberger, D. C. J. Catal.
2005, 234, 376−384.
(15) Oudenhuijzen, M. K.; van Bokhoven, J. A.; Miller, J. T.;
Ramaker, D. E.; Koningsberger, D. C. J. Am. Chem. Soc. 2005, 127,
1530−1540.
(16) Stakheev, A. Y.; Zhang, Y.; Ivanov, A. V.; Baeva, G. N.; Ramaker,
D. E.; Koningsberger, D. C. J. Phys. Chem. C 2007, 111, 3938−3948.
(17) Sun, B.; Pokhrel, S.; Dunphy, D. R.; Zhang, H.; Ji, Z.; Wang, X.;
Wang, M.; Liao, Y. P.; Chang, C. H.; Dong, J.; Li, R.; Madler, L.;
̈
Brinker, C. J.; Nel, A. E.; Xia, T. ACS Nano 2015, 9, 9357−9372.
(18) Pratsinis, S. E. Prog. Energy Combust. Sci. 1998, 24, 197−219.
(19) Strobel, R.; Stark, W. J.; Madler, L.; Pratsinis, S. E.; Baiker, A. J.
̈
Catal. 2003, 213, 296−304.
(20) Pokhrel, S.; Nel, A. E.; Madler, L. Acc. Chem. Res. 2013, 46,
̈
632−641.
(21) Pokhrel, S.; Birkenstock, J.; Schowalter, M.; Rosenauer, A.;
Madler, L. Cryst. Growth Des. 2010, 10, 632−639.
̈
(22) Schimmoeller, B.; Schulz, H.; Ritter, A.; Reitzmann, A.;
Kraushaar-Czametzki, B.; Baiker, A.; Pratsinis, S. E. J. Catal. 2008,
256, 74−83.
(23) Strobel, R.; Pratsinis, S. E. J. Mater. Chem. 2007, 17, 4743−4756.
(24) Gerardin, C.; Sundaresan, S.; Benziger, J.; Navrotsky, A. Chem.
Mater. 1994, 6, 160−170.
(25) Feng, R.; Liu, S. T.; Bai, P.; Qiao, K.; Wang, Y. H.; Al-Megren,
H. A.; Rood, M. J.; Yan, Z. F. J. Phys. Chem. C 2014, 118, 6226−6234.
(26) Huang, J.; van Vegten, N.; Jiang, Y. J.; Hunger, M.; Baiker, A.
Angew. Chem., Int. Ed. 2010, 49, 7776−7781.
(27) Rosenholm, J. M.; Czuryszkiewicz, T.; Kleitz, F.; Rosenholm, J.
B.; Linden, M. Langmuir 2007, 23, 4315−4323.
(28) Chizallet, C.; Raybaud, P. Angew. Chem., Int. Ed. 2009, 48,
2891−2893.
(29) Mojet, B. L.; Miller, J. T.; Ramaker, D. E.; Koningsberger, D. C.
J. Catal. 1999, 186, 373−386.
(30) Zhang, H.; Pokhrel, S.; Ji, Z.; Meng, H.; Wang, X.; Lin, S.;
Chang, C. H.; Li, L.; Li, R.; Sun, B.; Wang, M.; Liao, Y. P.; Liu, R.; Xia,
T.; Madler, L.; Nel, A. E. J. Am. Chem. Soc. 2014, 136, 6406−6420.
̈
2381
DOI: 10.1021/acscatal.6b00396
ACS Catal. 2016, 6, 2372−2381