Research Collection Doctoral Thesis Hydrogenation over supported noble metal catalysts From characterization to design Author(s): Makosch, Martin Publication Date: 2012 Permanent Link: https://doi.org/10.3929/ethz-a-007603659 Rights / License: In Copyright - Non-Commercial Use Permitted This page was generated automatically upon download from the ETH Zurich Research Collection . For more information please consult the Terms of use . ETH Library
138
Embed
"Hydrogenation over supported noble metal catalysts: from ...
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Research Collection
Doctoral Thesis
Hydrogenation over supported noble metal catalystsFrom characterization to design
whereas n(Sub)t0 resembles the molar quantity of the substrate before the reaction. The rate of
reaction at a certain time point t based on the conversion of the substrate in mmol gcat-1 s-1 was
established according to:
������� � ��� ���! ��� �����$!%�&��� 2.2-5
2.0 2.5 3.0 3.5 4.0 4.5
3.0
3.5
4.0
4.5
5.0
5.5
6.0
6.5
7.0
7.5
n(S
)/n
(IS
)
A(S)/A(IS)
y = 1.6x - 0.19
R2 = 0.9973
Chapter 2
18
For the rate of formation of intermediate or product a respectively the same equation was
used except that (nat0-nat) was employed. For the chemoselective hydrogenations in Chapter 6
and 7 the selectivity Sela towards a certain product a in relation to all the other intermediates
and products b, c, … x was calculated according to:
���� � � ����������…��" � �!! 2.2-6
All these factors (C-balance, rate, selectivity) were used to compare the performance of the
catalysts used in this work and to establish structure-performance relationships.
2.3. Transmission electron microscopy
Transmission electron microscopy (TEM) evolved to a powerful tool in material science and
improvement of TEM equipment is an active field of research. The breakthrough of the
aberration correction leads to a detection limit in the sub Å regime. Improvement of electron
microscopy also lead to different measurement modes. State of the art electron microscopy is
scanning transmission electron microscopy (STEM) where a focused coherent beam is
scanned over a defined area of the sample. In catalysis the electron microscopy technique is
used to investigate e.g. the morphology of zeolite crystals and most important the elucidation
of the particle size, morphology of the particles and stability of supported noble metal
catalysts. Figure 2.4 shows a schematic of a conventional microscope and the difference
between TEM and STEM.
Figure 2.4 a) Schematic of a conventional transmission electron microscope (CTEM), adapted from 59; b) illustration of the principle of a transmission electron microscope (TEM), adapted from 60; c)
illustration of the principle of a scanning transmission electron microscope (STEM), adapted from 61.
a) b) c)
Methods and experimental
19
The electron beam is generated via “electron guns” made from W- or LaB6 cathodes
generating electrons either by heating (thermoionic guns) or by applying an extraction
voltage (field emission guns). The electron beam gets accelerated towards the anode and
channeled via a condenser system towards the sample. A projector system allows either
imaging or diffraction whereas the image or diffraction pattern is projected onto a viewing
screen. The contrast visible in TEM pictures does not originate from a loss of intensity of the
electron beam as the samples in most cases are not thicker than 100 Å, but due to a phase
difference (to the primary beam) caused by the interaction of the electron beam with the
atoms of the sample.59 Here the nature and the thickness of the sample play an important role:
the scattering is higher when many atoms or heavy atoms interact with the primary electron
beam. An objective aperture tunes the contrast via interference of the diffracted beam and the
primary beam. Is only the primary beam taken into account the image results from a bright
field image; if only one or more diffracted beams are taken into account the mode is called
dark field image. In a transmission electron microscope a large area of the sample is
illuminated and the magnification of the image is achieved via a projector system after the
sample, thus the whole image is recorded at once (Figure 2.4 b). A disadvantage of TEM is
that image contrast is a function of sample thickness and focus and thus no unique image
represents a “simple” representation of the sample. In contrast to TEM the focus of the STEM
electron beam is done before the sample to form a 1 Å probe (Figure 2.4 c). This beam is then
scanned via scanning coils over a defined small area of the sample. The scattered electrons
are then collected via the simultaneous combination of various detectors to form the image as
a function of the position in a straight forward way. A high-angle annular dark field
(HAADF) detector collects electrons scattered at high angles. According to Rutherford high
Z elements scatter to high angles more strongly than light ones and thus the representing
image is achieved by Z contrast. The combination of STEM and energy dispersive X-ray
spectroscopy (EDXS), electron energy-loss spectroscopy (EELS) respectively is a powerful
tool to investigate structure morphology and composition as function of position of e.g.
supported noble metal catalysts at the nano scale.
In this work electron microscopy was used to determine the particle sizes of all supported
metal catalysts used in this work. Electron microscopy measurements were performed on a
HD2700CS (Hitachi, aberrationcorrected dedicated scanning transmission electron
microscope (STEM), cold FEG, 200 kV) or a Tecnai F30 ST (FEI, FEG, 300 kV). The high-
resolution capability of HD2700CS 27 (shown to be better than 0.1 nm) is due to a probe
corrector (CEOS) that is incorporated in the microscope column between the condenser lens
Chapter 2
20
and the probe-forming objective lens so that a beam diameter of ca. 0.1 nm can be achieved.62
A special bright field setting allows one to record highly-resolved phase-contrast STEM (PC-
STEM) images (similar to HRTEM) without delocalization artifacts. The catalyst was
suspended in ethanol and a drop of the suspension was supported on a perforated copper grid
by evaporation of the solvent. To determine the particle size distribution, TEM pictures were
analyzed with the Image J software and 200 particles distributed over various micrographs
were taken into account for each measurement.
2.4. Infrared spectroscopy
Infrared (IR) spectroscopy is a powerful tool which is extensively used in all fields of
chemistry. Especially in catalysis infrared spectroscopy, as a non invasive technique, is
applied for catalysts characterization as well as for following the reaction under working
conditions as characteristic IR vibrations can be used to distinguish substrates, intermediates
and products. The infrared region (14300 cm-1 to 100 cm-1) which is located between the
ultraviolet-visible and the microwave region of the electromagnetic spectrum is commonly
divided in three sub-categories: near-, mid- and far-IR are found in the regions of 14300-4000
cm-1, 4000-400 cm-1 and 400-100 cm-1 respectively. For general application the mid-IR
region is used as most organic chemicals have characteristic vibration features in this range
and a variety of commercial available spectrometers exist. Several experimentally different
applications of IR spectroscopy exist which are tailored to the corresponding application
among them transmission Fourier transform infrared spectroscopy (FT-IR), diffuse
reflectance infrared Fourier transform spectroscopy (DRIFTS) and attenuated total
reflectance infrared (ATR-IR) spectroscopy. The intensity of the IR beam before and after the
sample are given by Lambert-Beer´s law: the intensity I0 of the incident beam diminishes
exponentially over the path length l depending on the sample concentration [J] and ε the
molar absorption coefficient to the intensity I according to:
��& ��! � �' ()*� 2.4-1
In spectroscopy the term “absorption” (A) is more general, relates to A = ε[J]l and is
dimension less as the unit of ε is L mol-1 cm-1. The ratio ++, is referred to as “transmission” (T)
and relates to A as log 0 � �1. As the concentration [J] is part of Lambert Beer´s law, IR can
be used for qualitative as well as for quantitative analysis but requires calibration.
Methods and experimental
21
In this work transmission FT-IR and ATR IR spectroscopy were used in Chapters 3, 4, 6 and
7 and thus will be explained in more detail in the following.
State of the art commercially available transmission Fourier transform spectrometers apply
Fourier transformation for detection and analysis of spectra, which enhances the detection
limit compared to spectrometers with monochromators, as the entire emitted radiation of the
sample is used for detection.63 For transmission FT-IR the sample is placed between the
incident IR beam, which is modulated via a Michelson-interferometer, and the detector. A
Michelson interferometer divides the incident IR beam into two beams, introduces a path
difference p so that destructive and constructive interference can occur and recombines the
beams.64 Doing so, an interferometer transforms a certain wavenumber ν of the incident beam
into a variation of the intensity of the exiting beam and thus the different wavenumbers of a
coherent source become distinguishable according to:
��2� � 3 ��4�∞
! �� 5 ��%�642��4 2.4-2
Fourier transformation of the measured I(p) yields the spectrum I(ν) according to:
��4� � 73 8��2� � �� ��!�9∞
! ��%�642�2 2.4-3
Transmission IR is always applied when the sample itself or the supporting material is
transparent for the IR beam so that good spectrum resolution can be achieved within a
reasonable time. For recording transmission IR spectra the sample is generally pressed to
self-supporting or KBr-supported pellets and then mounted via various experimental cells
into the IR beam.
In this work transmission FT-IR spectroscopy was used to characterize the organic thiol
modified supported Pt catalysts in chapter 6 and 7. Fourier transform infrared measurements
were recorded on a Bruker Equinox 55 FT-IR spectrometer. For that, pellets of 3 mg of the
corresponding catalyst mixed with 97 mg KBr were pressed at 5 tons for 1 min. Prior to the
measurement, a background spectrum in air was recorded with 50 scans per spectrum.
Samples were recorded between ν = 4000 and 1000 cm-1 for the supported Pt catalysts with a
resolution of 1 cm-1. A total of 1000 scans per spectrum were averaged. The spectrum of the
modified supported Pt catalyst was background corrected with the spectrum of the
Chapter 2
22
unmodified catalyst and a constant air background. The raw data was smoothed by 25 points
using the OPUS software.
2.4.2. Attenuated total reflectance infrared spectroscopy
Attenuated total reflectance infrared spectroscopy (ATR IR) is one of the most applied
techniques where the sample (solid or liquid) is placed in close contact to a sensing device (in
most cases a crystal). ATR IR can be applied without sample preparation or dilution and even
in case the sample itself allows no transmission of IR beams. By coating an ATR crystal it is
even possible to measure adsorbents on the catalyst surface in the liquid phase as well as in
the gas phase.29,65,66
The ATR IR concept can be explained by total reflection. Figure 2.5 illustrates the path of
radiation from a dense optical medium 1 to the rare optical medium 2.
Figure 2.5 Representation of Snell´s law for radiation passing from an optical dense medium n1 to a
optical rare medium n2. Incident beams at or above the critical angel θc do not obey Snell´s law but
reflect internally in the rare optical medium. Adapted from 67.
The Snell´s law describes the angle at which radiation is refracted by passing from one
transparent medium to another with different refractive indices:
�� %��:� � �� %��:� 2.4-4
with n1 and n2 as the refractive index of medium 1 and 2 and θ1 as the angle of incidence and
θ2 as the angle of refraction with respect to the normal to the interface. If the refractive index
Methods and experimental
23
of medium 1 is bigger than the index of medium 2, n1 > n2, then the former is the dense
optical medium and the latter the rare optical medium. If radiation passes from an optical
dense medium 1 to an optical rare medium 2, the incident angle θ1 will always be smaller than
θ2. If the angle of incidence will be further increased the critical angle θc will be reached
where according to Snell´s law the refracted angle θ2 will eventually reach 90 °, i.e. refracted
parallel to the media interface. In practice only incident angles below θc obey Snell´s law,
whereas incident radiation ≥ θc will internally reflect in the optical dense medium with the
same angle as the incident angle. The critical angle θc can be easily calculated by knowing
the refractive indices of medium 1 and 2 according to:
:� � %�� � ���� 2.4-5
Materials that show internal reflection are known as internal reflection elements (IRE). At the
point of the reflection the electrical field of the IR beam´s photons extends perpendicular to
the beam direction into the rare optical medium whereas the IR beam is confined to the IRE.
The field in the optical rare medium is also referred to as evanescence wave and
exponentially decays over distance z according to:
; � ;!� <= 2.4-6
where E0 is the strength of the electrical field at the surface and γ a constant. The strength of
E (around 1/e of E0) which effectively interacts with the optical rare medium containing the
sample is known as the penetration depth dp and directly depends on the wavelength of the
incident IR beam. The penetration depth dp can be calculated via the incident wavelength λ of
the IR beam in a vacuum, the refractive indices n1 and n2 and the angle of incidence θ1
according to:
�2 � >�6��?%���:� �������
2.4-7
As the penetration depth is responsible for the interaction of the IR beam and the sample and
thus the performance of the ATR IR device the material of the IRE which defines the
refractive index has to be chosen accordingly. Most IREs are made of zinc sulfide (ZnS),
germanium (Ge) or diamond because all these materials have a high refractive index n1. ATR
IR devices are available in various experimental setups such as vertical variable angle
attenuated total reflection (VATR), horizontal attenuated total reflection (HATR), in situ
cylindrical internal reflection cell for liquid evaluation (CIRCLE) and DiComp sensors. In
Chapter 2
24
this work a DiComp probe connected to the spectrometer via a fiber optic was used as this
setup can be operated remotely (from spectrometer) under in situ conditions at high pressure,
temperature and realistic reaction conditions in the liquid phase as the IRE is made of
diamond. Figure 2.6 shows a vertical cut of a DiComp probe.
Figure 2.6 Vertical cut through a DiComp probe: The IR beam is channeled from the spectrometer
via fiber optics and a supporting ZnS crystal towards the diamond IRE. The beam is multiple reflected
in the diamond crystal and then guided back to the detector. The ZnS and diamond crystals are
contained in a stainless steel case. Adapted from 67.
The beam is channeled through a fiber optic from the detector through a ZnSe crystal to the
diamond IRE. As both materials have similar refractive indices (ZnS :2.2, diamond: 2.4) the
transmission from ZnS to the diamond IRE works with minimal refractive losses. The IR
beam is multiple reflected in the diamond IRE which is in direct contact with the sample. At
each point of reflection an evanescent wave interacts with the sample (red waves). The
advantage of multiple reflections is that the absorbance of the sample is much higher
compared to a single reflection and is employed for most commercial ATR IR devices.
Multiple reflections are achieved via the geometry of the IRE (e.g. parallelpiped or vertical
truncated triangular crystals). After passing the diamond IRE the IR beam is guided through
the ZnS crystal and via the fiber optic back to the detector. Directly below the diamond
crystal is a cavity filled with air, achieved via a cut in the ZnS crystal, so that the IR beam is
reflected only in the IRE.
In this work ATR IR was used in combination with a custom made in situ cell for pressurized
liquid phase reaction (see Chapter 3)68 in Chapters 3 and 4 to monitor the conversion of
nitrobenzene in situ by using a commercial ATR IR system from Mettler Toledo with a
Methods and experimental
25
DiComp optical fiber immersion probe attached to a ReactIR 45 spectrometer. One spectrum
per minute (average of 16 scans) was recorded, with a resolution of 2 cm-1 from ν = 2000 to
750 cm-1. Background spectra of the catalyst suspension were collected in toluene at
respective temperatures and subtracted via the ICIR software.
2.5. X-ray absorption spectroscopy
X-ray absorption spectroscopy (XAS) is a well established technique that allows to collect
electronic and structural properties in situ of catalytic reactions under working conditions.69
The combination of using hard X-rays (> 2500 eV) with suitable in situ cells is a powerful
tool to establish structure-performance relationships and help to identify active sites of a
heterogeneous catalyst. In X-ray absorption, a photon is absorbed by an atom initiating an
electron transition from the core state to empty states above the Fermi level. The core state
energy is unique for each element making XAS an element specific-method. By passing
through a material the incident energy I of the X-ray photons will decrease according to the
absorption properties of the material. The decrease dI of the photon intensity for a certain
path length dl through the material is given by
�� � �@�;���� 2.5-1
Where µ(E) resembles the linear absorption coefficient as a function of photon energy. By
integration of that one obtains the Lambert´s law (compare 2.4-1):
� � �!� @�;�� 2.5-2
A sharp rise in the absorption intensity will occur upon the absorption of the photon and the
excitation of the core electron. This sharp rise in intensity is referred to as the absorption edge
(Figure 2.7). The edges are named after the core state from where the electrons are excited.
Thus the K-edge reflects the excitation of a 1s core electron, the LIII, LII and LI edge the
excitation from 2p3/2, 2p1/2 and 2s core levels respectively. All excitations and transitions
obey the quantum mechanical selection rules, whereas transitions for which the orbital
quantum number of the final state differs from the initial state by 1 (∆L= ±1) and 2 (∆L= ±2)
are called dipole respectively quadrupole transitions. In general the dipole transitions are
more intense than quadrupole transitions. Thus p density of states (DOS) are probed at the K
and LI edges and d density of states can be probed at the LII and LIII edges. Figure 2.7 shows a
XAS spectrum of the Pt LIII edge.
Chapter 2
26
Figure 2.7 Characteristic regions of a XAS spectrum. Adapted from 70.
The XAS spectrum can be divided into three regions: the pre-edge region, the X-ray
absorption near-edge structure (XANES) region and the extended X-ray absorption fine
structure region (EXAFS). The first sharp feature after the absorption edge is called the white
line. The energy position of the absorption edge, the whiteline intensity and the shape of the
spectrum in the XANES region can be used to investigate the electronic and local geometric
structure of the absorbing atom. In heterogeneous catalysis the number of d-electrons is
related to the catalytic activity.71 As the shape of the XANES spectrum reflects the density of
empty states, d-states can be probed by using 2p or 3p core-sates according to the dipole
selection rule as explained above. The shape of the spectrum also reveals the presence or
absence of adsorbates even without long range order making is especially attractive for
supported noble metal catalysts.
The kinetic energy of the electron, Ek, is defined to be equal to 0 at the absorption edge;
therefore the kinetic electron energy, Ek, above the edge is given by
;A � B4 � ;������& 2.5-3
The outgoing photoelectron can be described as a spherical wave with the wavelength λ as
follows:
> � �6A 2.5-4
where k is the wave-vector:
Methods and experimental
27
A � ?�C6�B� " DB4 5 ;! � ;��&�E 2.5-5
with m the electron mass and h Planck´s constant. The transition probability of the
photoelectric effect is proportional to the linear absorption coefficient µ(E) and according to
Fermi´s Golden Rule a function of the initial- and finale-state wave function72
@�;� � �FG�H|�J�|KG�F�L�;� � ;� � B4� 2.5-6
where ê is the electric field polarization vector of the photon and r the coordinate vector of
the electron. This dipole approximation is only valid when the wavelength of the photons is
larger than the size of the absorbing atom.69 Two main parts the outgoing electron wave,
ψoutgoing, and the backscattered electron wave ψbackscattered contribute to the final wave function
ψf
G� � G� �&���& 5G���A%�������� 2.5-7
The fine structure in the XANES and EXAFS region are due to variation in the transition
probability as function of energy that arise from the interference between these final-state
wave functions.
2.5.1. High-energy resolution fluorescence detected X-ray absorption near edge
spectroscopy (HERFD XANES)
XAS is still an emerging field and new approaches lead to new applications in the field of
synchrotron based measurements. In conventional XAS, the transmitted photons or the
radiative and/or non-radiative decay of the sample is monitored as function of energy of the
incident photons. A disadvantage of conventional XAS is that the energy resolution is limited
to the life-time broadening of the core hole in the exited state resulting in broad features in
the spectra. This limitation can be circumvented by selectively detecting a fluorescence decay
channel. The fluorescence decay results in a final state that has a longer life-time and thus
less broadening, which leads to spectra with a higher energy resolution and sharper
features.73,74 Figure 2.8 shows a schematic representation of the life time broadening and the
HERFD XAS principle.
Chapter 2
28
Figure 2.8 Schematic representation of the HERFD XAS principle. In conventional XAS (left) the
incident X-ray (hν1) excites an electron from a core level. The life-time of this core level leads to a
broadening in the spectra. For HERFD XAS (right) the fluorescence line (hν2) of the decay is
monitored. The life time of this hole is much longer and thus leads to decreased broadening. Adapted
from 4.
Due to the delocalized character of the 5d electrons of the 5th row elements the final state
effects are almost negligible making HERFD XAS especially efficient for those elements.
Besides higher energy resolution another important advantage of HERFD XAS over
conventional XAS is that the selective detection of a fluorescence decay channel is element
specific such that EXAFS spectra can be recorded with extended in k range even for edges
which would normally interfere with edges of other elements. For example full EXAFS
analysis of the Mn K edge could be recorded for a multiprotein PS II complex via HERFD
XAS without the interference of the Fe K edge, which would normally arise in the region of
the Mn K edge in conventional XAS. Practically HERFD XAS is achieved via a X-ray
emission spectrometer (Figure 2.9).
Methods and experimental
29
Figure 2.9 Schematic representation of a vertical-plane Rowland circle X-ray emission spectrometer.
The incident X-ray beam causes fluorescence decay in the sample. The fluorescence line is selectively
collected via the analyzer crystals and channeled onto the detector. Adapted from 75.
The incident X-ray beam causes the fluorescence decay in the sample. The analyzer crystals
focus the corresponding fluorescence line with very high energy resolution to the detector.
Sample, analyzer crystals and X-ray detector are located on the so called “Rowland circle”.
Thus only photons resulting from the fluorescence decay are detected.
In situ high energy resolution fluorescence detected X-ray absorption near edge spectroscopy
(HERFD XANES) measurements were performed in Chapter 3, 4 and 7. For that a cell which
will be described in detail in Chapter 3 was employed for all measurements.68 A suspension
of 300 mg catalyst in 25 g of toluene (puriss > 99 %, Fluka Analytical) was put into the cell
and purged three times with 10 bars H2 at 80 °C and finally pressurized to 10 bars. HERFD
XANES spectra were recorded under stirring for 30 min with a time interval of 1 min per
spectrum. After that the cell was opened and 300 mg substrate (4-nitrostyrene puriss > 95 %
in Chapter 7, TCI; nitrobenzene, > 99.5%, Sigma–Aldrich in Chapter 3 and 4) were added to
the reaction mixture. The cell was closed and purged 3 times with 10 bars H2 under stirring
maintaining the temperature at 80 °C. The reaction started and HERFD XANES spectra were
recorded during the whole reaction with a time interval of 1 min per spectrum. All
experiments were recorded at beamline ID26 of the European Synchrotron Radiation Facility
(ESRF) in Grenoble, France. The ring operated in uniform mode at a ring current of 200 mA.
Three coupled undulators using the third harmonic were employed for the HERFD XANES
measurements. The incident energy was monochromatized by a pair of Si(111) crystals.
Three Pd/Cr mirrors positioned at 2.5 mrad relative to the incident beam were used to
Chapter 2
30
suppress higher harmonics and focus the beam on the sample with a size of 600 µm
horizontal by 200 µm vertical. The estimated flux was 8x1013 photons s-1. HERFD XANES
spectra were measured by using a vertical-plane Rowland circle X-ray emission spectrometer
in combination with an avalanche photodiode (APD, Perkin Elmer).76 The scattering angle in
the horizontal plane was about 130°. The spectrometer was tuned to the Au Lα1 fluorescence
line (9713 eV) respectively the Pt Lα1 fluorescence line (9442 eV) using the [660] reflection
of four spherically bent Ge crystals, that is, working at a Bragg angle of about 80 °. A total
resolution of 2.1 for gold respectively 1.93 eV for Pt (FWHM) was obtained. The raw
HERFD XANES spectra were treated with the Athena software.77 After background
subtraction the raw data were normalized to the last point of each spectrum and 30 spectra
were averaged. Exposure of the slurry to X-rays did not cause any changes to the spectra and
thus beam damage did not occur. Special pretreatments or variations of the parameters
described above will be described in detail in the experimental section of the corresponding
chapter.
Chapter 3
Design and application of HERFD XAS/ATR FT-IR
batch reactor cell
(Makosch, M.; Kartusch, C.; Sa, J.; Duarte, R. B.; van Bokhoven, J. A.; Kvashnina, K.; Glatzel, P.; Fernandes, D. L. A.; Nachtegaal, M.; Kleymenov, E.; Szlachetko, J.; Neuhold, B.; Hungerbühler, K. Phys. Chem. Chem.
Phys. 2012, 14, 2164 - Reproduced by permission of the PCCP Owner Societies) [Martin Makosch performed the experiments, did the data analysis and wrote the manuscript]
Chapter 3
32
3.
3.1. Introduction
Pressurized reactors for liquid phase reactions (autoclaves) are commonly used for the
preparation of fine chemicals. Industrial reactions generally use a catalyst, a liquid medium,
and substrates. These reactions are often performed under pressure. Because the inside of an
autoclave is often inaccessible to spectroscopy, the elucidation of the reaction mechanism and
the catalyst structure is challenging under these conditions. This gave autoclaves the
denomination of black boxes. Understanding the role of catalysts under relevant conditions is
a relatively recent but strongly growing area of catalysis.1 Discovery and improvement of
catalysts rely on accurate determination of the reaction mechanism and its relation to the
structure of the catalyst and active site. Therefore, characterization instrumentation must
enable monitoring catalytic performance of a particular active site in real-time and in a
spatially resolved way under realistic catalytic conditions of pressure and temperature.2 The
major advantage of a combined approach or single mode operation is that all measurements
are carried out with the same setup, which ensures that measurements are performed on the
same catalytic system. Any added characterization technique should yield complementary
information and/or be a monitor of the influence of the main technique on the catalytic
performance or the catalyst structure (e.g. X-ray radiation damage). The choice of the setup
should be based on the catalytic application, and not on the characterization method. An
operando setup consists of a combination of spectroscopic methods which can follow the
reaction kinetics and the catalyst structure to identify reaction intermediates and active sites
and ultimately link the reaction mechanism to the active site. For gas phase reactions there
are numerous applications using in situ cells, which successfully showed their feasibility in in
situ and operando spectroscopy.3–6 In contrast, relatively little is published about reactions in
the liquid phase. Nevertheless there are some interesting setups for such measurements. An
example for measuring X-ray absorption spectroscopy (XAS) of a liquid/solid reaction is a
setup where the reaction mixture is pumped from a slurry reactor into a small stainless steel
compartment. This compartment contains a Millipore filter at the bottom where the catalyst
slurry remains whereas the liquid flows back into the slurry reactor. The compartment is
sealed by mylar windows transparent for X-rays which are in line with the Millipore filter.7
For XAS in the liquid phase at high pressure, there is a cell, which enables to study metals in
the liquid phase and the solid/liquid phase, respectively.8 This cell enables measuring in
supercritical CO2 and can sustain very high pressure and therefore has a more oval than round
Design and application of HERFD XAS/ATR FT-IR batch reactor cell
33
shape characteristic of autoclave reactors. Another cell has been constructed for
measurements in supercritical media.9 The fixed bed reactor of the continuous flow cell lies
in a sapphire capillary which can sustain high pressures (400 bars) and temperatures (400 °C)
and is transparent for X-rays. Another example is an in situ infrared (IR) cell for
measurements in the liquid phase, where the reaction solution is forced through an IR
transparent capillary, which subsequently flows back into the mixing chamber.10 Recently, a
similar example of a recirculation reactor setup for operando fluorescence XAS was
presented.11 The reaction mixture is pumped from a stirred reactor vessel via a peristaltic
pump and a capillary through a small PTFE cell with Kapton windows and back into the
stirring vessel. All these approaches are however very different to working conditions of a
stirred slurry in an autoclave. Despite the importance of time-resolved studies in catalysis
(seconds or minutes timescale), XAS measurements in the liquid phase are often performed at
the steady-state or under differential kinetic conditions and relatively large amounts of
catalysts are used to increase the signal to noise ratio. To the best of our knowledge there is
no cell available that allows studying reactions in autoclaves with fluorescence XAS and
attenuated total reflectance Fourier transform infrared (ATR FT-IR) spectroscopy. We
describe a cell that combines XAS and ATR FT-IR for simultaneously measuring the catalyst
structure and the reaction mechanism in real time. The cell enables the determination of the
catalysts oxidation state, structure and catalytic performance under exact catalytic conditions
of pressure, temperature, and medium without interfering with the reaction itself and without
altering the basic reactor design. Via ATR FT-IR, which is well established to monitor
reactions in the liquid phase,12 we are able to follow the conversion of the educts to the
products and the occurrence of possible reaction intermediates in the liquid phase. Also,
modification of the setup will enable determining surface adsorbed intermediates at the
liquid/solid interphase, which provides information about the reaction kinetics and
mechanism.13 The cell is designed such that high energy resolution fluorescence detection
(HERFD XAS) is possible. Fluorescence detected absorption spectroscopy is used for dilute
samples.14 By detecting a fluorescence line with an instrumental energy bandwidth on the
order of the core hole lifetime broadening HERFD XAS spectra are obtained with a good
signal to background ratio with line sharpened absorption features.15–18 The combination of
HERFD XAS and ATR FT-IR is a powerful combination to establish structure–performance
relationships in liquid phase reactions. Both techniques allow performing time resolved
studies at a sub-minute timescale, which is crucial to understand dynamic changes in a
catalyst structure during pretreatment and reaction. The hydrogenation of nitrobenzene over
Chapter 3
34
gold catalysts, which will be used to show the feasibility of our cell, has drawn increased
attention since its discovery in 2006.19 Gold catalysts are an environmental friendly
alternative to common catalysts used for this reaction such as Pd and Pt supported on active
carbon or CaCO3, 20 because they do not require environmental harmful additives to achieve
high selectivity or low phenylhydroxylamine concentrations.21 The mechanism of this
reaction and the active site of the gold catalysts are still under debate. In this study we show
that during the hydrogenation of nitrobenzene over Au/CeO2 catalysts the main oxidation
state of the active metal is Au0.
3.2. Experimental section
For all experiments a 1 wt% Au/CeO2 catalyst was used, which was prepared via deposition
precipitation with urea.22 In a Teflon container, 3 g of the support (CeO2 MicroCoating
Technology) was dissolved in 300 ml of deionised water containing 0.9 g of urea. To this
solution 50.97 mg of HAuCl4.3H2O (Au: 49 %, 99.9 % metal basis, ABCR-Chemicals) was
added. Subsequently, the mixture was stirred for 16 h at 80 °C under the exclusion of light.
The precipitate was filtered, washed 3 times with water to prevent chloride contamination and
dried under vacuum overnight. The resulting catalysts are referred to as ‘‘as-prepared’’.
Reduction experiments were performed in toluene, isopropanol, cyclohexane, and
tetrahydrofuran. For each experiment, 300 mg of the ‘‘as-prepared’’ catalyst was put into the
reactor in 25 g of solvent. After that, the reactor was flushed 3 times with 5 bars helium and
finally pressurized to 5 bars helium. Subsequently the mixture was heated to 60 °C under
mechanical stirring at 1500 rpm. After reaching that temperature, the cell was flushed 3 times
with 10 bars H2 under mechanical stirring at 1500 rpm and finally pressurized to
10 bars H2. After introducing H2, HERFD XAS scans were recorded for 1 h with a time
interval of 1 min while heating. Hydrogenation of nitrobenzene was performed in water,
resulting in an emulsion with the substrate. For this experiment, the catalyst was pretreated
only in helium. A total of 300 mg of the ‘‘as-prepared’’ catalyst was heated to 60 °C in 5 bars
of helium and then kept at 60 °C in 5 bars He for 1 h. Subsequently, 300 mg of nitrobenzene
were added. The cell was flushed 3 times with 10 bars H2. After that the reaction mixture was
heated to 120 °C and maintained at that temperature in 10 bars H2 under constant mechanical
stirring at 1500 rpm. An ATR FT-IR DiComp optical fiber immersion probe (Mettler Toledo)
combined with a ReactIR 45 spectrometer was employed to monitor the evolution of the
nitrobenzene conversion. A time resolution of 1 min was achieved by averaging 16 spectra in
the range of 750–2000 cm-1. For background subtraction, spectra at the corresponding
Design and application of HERFD XAS/ATR FT-IR batch reactor cell
35
temperature were recorded of the catalyst in solvent. All experiments were recorded at
beamline ID26 of the European Synchrotron Radiation Facility (ESRF) in Grenoble, France.
The ring operated at an energy of 6.0 GeV and at a ring current of 200 mA. Two u35
undulators using the third harmonic were employed for the HERFD XAS measurements. The
incident energy was monochromatized by a pair of Si(111) crystals. Three Pd/Cr mirrors
positioned at 2.5 mrad relative to the incident beam were used to suppress higher harmonics.
The size of the X-ray beam measured 0.3 mm horizontal and 1 mm vertical, with a total flux
of 5x1012 photons s-1. HERFD XAS spectra were measured by using a vertical-plane
Rowland circle X-ray emission spectrometer in combination with an avalanche photodiode
(APD, Perkin Elmer). The scattering angle in the horizontal plane was ~130°. The
spectrometer was tuned to the Au Lα1 fluorescence line (9713 eV) using the [660] reflection
of four spherically bent Ge crystals. A total resolution of 2.1 eV was obtained. The raw
HERFD XAS spectra were treated with the ID26 matlab code. This code splits the raw data
into equal energy steps and normalizes absorption to the incoming photon flux. After
background subtraction the raw data were normalized in the range between 11.98 and 12
keV. Due to the good spectra quality, it was possible to obtain the fraction of metallic to
oxidic gold at different stages of the pretreatment using linear combinations of standard
spectra. The reference spectra were gold foil and bulk Au2O3 for Au0 and Au3+, respectively.
All samples used were checked for beam damage. Exposure of the slurry to X-rays in the
absence of hydrogen did not cause reduction of the gold precursor.
3.3. Cell description
A 50 ml autoclave used for liquid/solid reactions was modified without altering the shape of
the reaction vessel. This was achieved by cutting a 2.1 cm broad and 5 mm high oval opening
in the stainless steel container which surrounds the reaction container (we refer to that
opening as ‘‘window’’). Figure 3.1 shows the outer view, a vertical cut and the single parts of
the cell.
Chapter 3
36
Figure 3.1 Schematic drawing of the HERFD XAS/ATR FT-IR cell. Outer view of the cell (a),
vertical cut view of the cell (b), detail drawing of the cell parts (c).
The reaction container consists of polyetheretherketon (PEEK), which has a density of 1.3 g
cm-3, a wall thickness of 1.5 mm, and is pressure proof up to at least 20 bars, which was
determined via pressurizing to 20 bars for 24 hours. No pressure loss was detected during this
period. PEEK is attractive, because it has a high chemical resistance against acids, organic
solvents, and alkaline media; 1.5 mm PEEK transmits about 45 % of incident X-ray flux at
energies above 9 keV. The same is true for emitted fluorescence X-rays. The window allows
X-rays to enter and exit the reaction mixture through the PEEK. Via the closure head on top
of the cell, two ATR FT-IR probes (Mettler-Toledo DiComp immersion probe connected via
a flexible AgX fibre conduit to a ReactIR 45m spectrometer) can be inserted and sealed via
Swagelok connections. The stainless steel container can be heated by two heating rods up to
250 °C. The heat is transferred from the stainless steel container via the PEEK insert to the
liquid. The temperature of the steel container is controlled according to the real temperature
inside the
reaction mixture, which is measured with NiCr/Ni thermocouples. Due to absorption of X-
rays by solvents used in chemical reactions, a relatively large pathway through the reaction
mixture and the low concentration of the catalyst in solution, the cell can only be used in
fluorescence mode when measuring XAS at the Au LIII edge. The cell is sealed via a closure
head which contains Kalrez O-rings above and below the PEEK sealing plate. This sealing
plate acts both as a sealer as well as a thermal isolator. The closure head contains two tubes,
one to pressurize the cell and the other to measure the pressure via a Keller Mano 2000 LEO
3 manometer with a range of 0 to 30 bars. In the center, a mechanical stirrer is attached which
reaches up to 3000 rpm. One ATR FT-IR probe can be exchanged with a sample tube, which
Design and application of HERFD XAS/ATR FT-IR batch reactor cell
37
allows taking aliquots of the liquid phase during the reaction for GC analysis. The whole cell
can be fixed onto a carrier connected to a xyz sample stage for alignment at the beamline. All
parameters such as pressure, temperature of the steel container, temperature of the reaction
mixture, and stirring speed are remotely controlled.
3.4. Results and discussion
3.4.1. Reduction of as-prepared Au/CeO2 catalysts in different solvents
Heterogeneous metal catalysts used in organic synthesis are often reduced in situ before
reaction. The reduction temperature is generally based on gas-phase temperature-
programmed reduction, which does not take into account any influence of solvent. Recently,
we showed that the extend of reduction of a Pt–Re catalyst is dependent on the media used.23
At a moderate temperature, complete reduction of the metals was only achieved when the
reduction was performed in the gas phase. In the liquid phase the reduction was incomplete.
To illustrate the relevance of monitoring catalysts during pretreatment, the as-prepared
Au/CeO2 catalyst was reduced for 1 h in 10 bars of hydrogen at 60 °C in various solvents.
Figure 3.2 shows the evolution of the Au LIII edge HERFD XAS signal collected in situ
during the reduction.
Chapter 3
38
Figure 3.2 Evolution of the Au LIII edge HERFD XAS signal during reduction at 60 °C in 10 bars H2
as a function of time for the different solvents used: (a) cyclohexane, (b) tetrahydrofuran, (c)
isopropanol, and (d) toluene. One spectrum per minute was recorded. The spectra at t = 0 corresponds
to the catalyst in 5 bars of helium at 60 °C (compare the experimental part).
The initial spectrum (t = 0), recorded before adding the hydrogen, showed an intense first
feature at ~11.92 keV, which is called the whiteline. Its intensity reflects the number of holes
in the d-band and high intensity is typical of a high fraction of cationic gold (Au3+).17 This is
not surprising since the catalyst was prepared by deposition precipitation, which yields Au3+
hydroxide. Noticeable peaks at ~11.93, ~11.95 and ~11.97 keV that are typical of metallic
gold were also present. This suggests that part of the initial gold was already reduced to the
metal state. The result is not surprising since cationic gold is very unstable and can be
reduced even under oxidizing environments.24 After pressurizing to 10 bars hydrogen at 60
°C, the intense whiteline decreased and the peaks at ~11.93, ~11.95 and ~11.97 keV became
more intense. This occurred in all solvents, however the speed at which this happened was
different. When the experiment was conducted in cyclohexane (Figure 3.2a) the whiteline at
~11.92 keV slowly decreased within 30 min on stream. A similar evolution was observed for
the reduction in tetrahydrofuran (Figure 3.2b) though the loss in intensity of the whiteline at
~11.92 keV was slower and was complete between 40 and 50 min. Also the increase of
Design and application of HERFD XAS/ATR FT-IR batch reactor cell
39
intensity at ~11.95 keV was not as fast as in cyclohexane. A different behaviour was
observed when the reduction was performed in isopropanol and toluene (Figure 3.2 c and d).
The signal for the whiteline at ~11.92 keV diminished during the reduction in isopropanol as
well as in toluene within 10 min on stream. Compared to the reduction in cyclohexane and
tetrahydrofuran the signals at ~11.95 and ~11.97 keV in isopropanol and toluene remained
more or less constant from the beginning of the reduction. As expected, the percentage of
Au3+ diminished during reduction yielding metallic gold. Quantification of gold species was
carried out by linear combination fitting using Au foil (Au0) and Au2O3 (Au3+) as reference
spectra. Figure 3.3 shows an example of a linear fit for one spectrum recorded during the
reduction in tetrahydrofuran.
Figure 3.3 Linear fitting of a Au LIII edge HERFD XAS spectrum of Au/CeO2 after reduction in
tetrahydrofuran in 10 bars of pure H2 at 60 °C after 37 min on stream. (●) original data; (—) overall
fitting; (…) Au foil reference and (---) Au2O3 reference spectra.
The two components reproduced the measured spectra well (R2 = 0.993). This was valid for
the large majority of the data, with some exceptions where small residues could not be
accurately fitted. The small discrepancies were detected during less abrupt transitions of Au3+
to Au0, which might suggest that some Au+ might be present as an intermediate, however, its
amount was too small to be quantified. Figure 3.4 shows the evolution of the percentages of
cationic and metallic gold during 1 h reduction in different solvents at 60 °C.
Chapter 3
40
Figure 3.4 Evolution of a Au3+ signal during the reduction at 60 °C in 10 bars H2 as a function of
Each profile contains ca. 60 spectra, corresponding to 1 spectra per minute. The initial ratio
of Au3+/Au0 at the beginning of the reduction at 60 °C in 10 bars of hydrogen was about 1,
meaning that half of the gold in the catalyst was metallic after heating the reaction mixture to
60 °C in 5 bars of helium. Gold reduced extensively and rather easily in all solvents even at
60 °C.25 During the reduction in toluene and isopropanol, the percentage of Au3+ decreased
below 15 % after 15 min. After that the fraction of Au3+ was more or less stable until the end
of the measurement. Gold reduced at a much slower rate when the reduction was carried out
in tetrahydrofuran and cyclohexane. Only after 45 min for both solvents the percentage of
Au3+ was below 15 %. It is not clear at the moment which solvent parameters affect the
reduction of gold, since the differences cannot be assigned to a single contribution, such as
H2 solubility and polarity. A possible reason for the different reduction rates could be that the
degree of suspension of the Au/CeO2 catalyst differed in various solvents. The results
highlight the importance of examining the state of the catalyst under reaction conditions. At
the end of the pretreatment gold was found to be more than 90 % in the metallic phase in all
solvents. When reduction was carried out in the gas phase, 100 % metallic gold is achieved
within 5–10 min as the hydrogen is not diluted in the solvent.26
3.4.2. Hydrogenation of nitrobenzene over Au/CeO2
We operated the setup in operando mode by coupling the HERFD XAS with the ATR FT-IR
probe, which enabled us to monitor simultaneously the gold oxidation state and catalytic
Design and application of HERFD XAS/ATR FT-IR batch reactor cell
41
reactivity. We monitored the system from pretreatment to reaction. Figure 3.5 shows the
evolution of the Au LIII edge HERFD XAS signal during the pretreatment of the as-prepared
Au/CeO2 catalyst in water at 60 °C in 5 bars of He.
Figure 3.5 Evolution of the Au LIII edge HERFD XAS signal during reduction in water at 60 °C in 5
bars of He.
Once again the initial spectrum showed an intense whiteline and some peak characteristics of
Au3+ and Au0, respectively. Time on stream revealed a fast disappearance of the whiteline
and increase of the peaks, which reached a plateau after 10 min in 5 bars of helium at 60 °C.
Linear fitting (not shown) revealed that all gold was present in the metallic state after 10 min
of pretreatment. Hence in water the as-prepared Au/CeO2 fully reduced to Au0 (via the
production of O2 and H2O) without the presence of H2. As expected, no noticeable changes
were observed in the ATR FT-IR spectra. After the helium pretreatment, nitrobenzene and 10
bars of pure hydrogen were added to the mixture to evaluate the catalytic performance of the
catalyst in the reduction of nitrobenzene. In this reaction, nitrobenzene reacts to form short
lived intermediates, which in the case of Au/CeO2 condense to form azoxybenzene.27 The
azoxybenzene is reduced in consecutive steps to azobenzene and hydrazobenzene and finally
to aniline. The condensation route is one of the routes proposed by Haber in 1898.28,29 The
unusual reactivity of gold catalysts is commonly assigned to the size of its particles and to its
oxidation state. In terms of particle size, the literature is consensual that to be active, gold
must be present in small clusters30 or as nanoparticles31 and the interface with the support is
often suggested to be the active site. The active oxidation state of gold during the reaction
however has been a topic of extensive discussion and to date no consensus has been achieved.
Cationic gold has been often assigned to be the active site for several reactions32–34 based on
the observation of its presence at the onset of reaction and pretreatment. Herein, the power of
Chapter 3
42
performing characterization under relevant working conditions is illustrated. The oxidation
state of gold and the concentration of the chemical species in solution were monitored
simultaneously using HERFD XAS and ATR FT-IR, respectively. Figure 3.6 shows the
evolution of the gold signal during the hydrogenation of the emulsion of nitrobenzene in
water over Au/CeO2 at 120 °C in 10 bars of hydrogen.
Figure 3.6 Evolution of the Au LIII edge HERFD XAS signal during the hydrogenation of
nitrobenzene over Au/CeO2 in water at 120 °C in 10 bars of H2.
No visible changes in the whiteline region at ~11.92 keV or in the characteristic nanoparticles
region at ~11.93, ~11.95 and ~11.97 keV occurred throughout the reaction. The spectra
recorded were nearly constant and not affected by the reduction of nitrobenzene. Figure 3.7
shows a characteristic IR spectrum at ~5 min of reaction and the evolution of the chemical
species during the hydrogenation of nitrobenzene recorded by ATR FT-IR simultaneously by
measuring the spectra of Figure 3.6.
Figure 3.7 Characteristic IR spectrum after ~5 min of reaction (left) and evolution of chemical
species (right) during nitrobenzene hydrogenation monitored by ATR FT-IR. Reaction was carried out
in water in 10 bars of pure H2 at 120 °C. (■) Nitrobenzene (NB) and (●) azoxybenzene (AOB).
Design and application of HERFD XAS/ATR FT-IR batch reactor cell
43
The evolution of the band at 1531 cm-1 (blue), which is assigned to the asymmetric stretch
vibration of the nitro-group35 of nitrobenzene, linearly decreased with time and only slightly
deviated in the first 15 min of the reaction. After 5 min, a signal at 1477 cm-1 (red) appeared
which is assigned to azoxybenzene.36 The intensity of this band increased strongly to a
maximum at ~20 min after which it stayed constant. After about 40 min the intensity started
to decrease. Thus nitrobenzene was converted to azoxybenzene, which after 40 min on stream
started to react to form azobenzene for which traces were observed by GC analysis. The
amount of catalyst to substrate was too low to observe further consecutive reaction products.
The gold oxidation state (Au0) remained constant during the reaction. Linear fitting revealed
that gold was present only in the metallic form throughout the entire reaction period.
3.5. Conclusions
We showed for the first time that in situ and operand HERFD XAS coupled with ATR FT-IR
can be performed in pressurized liquid batch reactors without changing the reactor geometry.
The combined techniques yield complementary information about the chemical state of the
active center and catalytic performance of the catalyst, which enables the deduction of
structure–reactivity relations without altering the reactor design. These methods are
minimally invasive and have good time resolution (<1 min/full spectrum) to monitor
dynamics of the system as the reaction takes place. Thus conditions outside of the steady-
state can be monitored. We observed that the rate of reduction of gold in Au/CeO2 in
different solvents depends on the solvent and is different from the gas phase. The final
catalyst structure was found to be practically the same. The combination of HERFD XAS/
ATR FT-IR enabled monitoring the evolution of chemical species and the oxidation state of
gold during the hydrogenation of nitrobenzene. ATR FT-IR confirmed that the reaction
follows a stepwise mechanism, indicated by the formation of an azoxybenzene intermediate.
HERFD XAS indicated that gold remained completely metallic throughout the catalytic run.
Our results indicate that within about 1 % accuracy, only metallic gold was present in the
hydrogenation of nitrobenzene over Au/CeO2. No other oxidation state of gold was detected
throughout the reaction.25 The amount of any, if any, active cationic species is below the
instrumentation detection limit (~1%) and/or too short lived to be detected.
Chapter 3
44
Chapter 4
The dynamic structure of gold supported on ceria in
the liquid phase hydrogenation of nitrobenzene
(Reprinted with permission from Kartusch, C., Makosch, M., Sá, J., Hungerbühler, K., van Bokhoven, J. A. ChemCatChem 4 (2012) 236. Copyright 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim)
[Martin Makosch and Christiane Kartusch performed the experiments and did the data analysis to equal parts.
Christiane Kartusch wrote the manuscript]
Chapter 4
46
4.
4.1. Introduction
Bulk gold is the most inert metal. However, when finely dispersed on a support, it is a very
active catalyst in many reactions,51 and molecules such as hydrogen, oxygen, and carbon
monoxide chemisorb on the gold surface. Examples of such reactions are CO oxidation,78-81
water gas shift reaction,82-84 selective alcohol oxidation,85-87 ethyne hydrochlorination,88,89
propene epoxidation,90,91 and hydrogenation reactions, such as the hydrogenation of
alkenes,92,93 alkadiens, 94-97 alkynes, 98-101 α,β-unsaturated carbonyl compounds, 102-105 and
nitro- compounds. 26,43,48 A unique property of gold is its high chemoselectivity due to the
preferential adsorption of oxygen-containing groups,104 in contrast to conventional
hydrogenation catalysts, such as palladium, platinum, and ruthenium, which must be
modified. Gold is, therefore, a promising catalyst in the development of new, clean, and
sustainable industrial processes with a minimal formation of byproducts. Although many
studies aim to describe the catalytically active sites, there is disagreement on the nature of
catalytically active sites in heterogeneous gold catalysis. Various suggestions have been made
to explain the catalytic activity of supported gold catalysts, for example, the size, structure,
and morphology of the supported gold species; the interface between gold particles and the
support; and the oxidation state(s) of gold in the catalysts. The oxidation state of active gold
species is still unclear. The catalytically active gold species in the well-studied CO oxidation
have been proposed to be cationic,106-110 fully reduced,81,111,112 and negatively polarized
gold.113-116 Aberration-corrected scanning transmission electron microscopy revealed that
gold bilayer clusters, approximately 0.5 nm in size and containing approximately 10 gold
atoms on FeOx supports, are highly active in CO oxidation.117 Cationic gold has been
proposed to be essential to reach high catalytic activity in the water gas shift reaction over
gold supported on nanocrystalline La-doped CeO2.82 Although hydrogenation has not been
studied to the same extent as oxidation, there is still no consensus with regard to the oxidation
state of active gold species. Cationic92-96,118 and reduced97 gold have been associated with
high catalytic activity. In conclusion, the nature of active species in heterogeneous catalysis
by gold is still unclear. In hydrogenation reactions, in situ studies92,93,119 determining the
oxidation state of gold during the process are rare. The aim of our work was to study in situ
the electronic properties of a supported gold catalyst (Au/CeO2) in a liquid phase
hydrogenation reaction, namely, the hydrogenation of nitrobenzene to form aniline. This
reaction has been studied extensively, both mechanistically and kinetically, and is used in
The dynamic structure of gold supported on ceria in the liquid phase hydrogenation of nitrobenzene
47
large-scale industrial processes to produce 90–95 % of the world’s aniline.120 Corma et al.26
discovered that supported gold catalysts catalyze the liquid phase hydrogenation of aromatic
nitrocompounds under mild conditions (100–120 °C, 10 bar H2). By using a newly
constructed in situ cell, we recorded simultaneously and in situ high-resolution X-ray
absorption near-edge structure (XANES) spectra at the Au LIII edge in the fluorescence mode
of the catalyst in the reaction mixture and monitored the reaction by means of attenuated total
reflectance Fourier transform infrared (ATR FT-IR) spectroscopy with a probe dipped into
the slurry.29,121 The cell is an autoclave reactor modified to achieve access of the X-rays to
the inner part of the reactor.68 X-ray absorption spectroscopy (XAS) is a very powerful
technique to determine the structural and electronic properties of catalysts under reaction
conditions.69 The X-ray absorption near-edge structure part of an XAS spectrum gives
information about the oxidation state and the local geometry of the absorbing atom. It reflects
the empty density of states of the electronic transition.3 XANES spectra were measured in the
high-energy-fluorescence detection (HERFD) mode, which gives much better resolution of
the spectra with sharper features than is possible with standard detection methods.4,74,76,119,122
ATR FT-IR spectroscopy is a reliable technique for monitoring online reactions.65,66,123
Molecular vibrations that lead to changes in the molecule’s dipole momentum, induced by
mid-IR radiation (ν = 4000-400 cm-1), enable us to identify and quantify most of the organic
compounds. Thus, we determined the electronic properties of gold supported on ceria and the
changes it undergoes under different pretreatment and reaction conditions and simultaneously
monitored the course of the reaction.
4.2. Experimental section
Au/CeO2 was prepared through deposition-precipitation with urea.55 CeO2 (99.9 %, ABCR
Chemicals) was calcined at 500 °C for 5 h to remove any residual nitrates. To prepare 0.8
The dynamic structure of gold supported on ceria in the liquid phase hydrogenation of nitrobenzene
53
At a reaction temperature of 60 °C, the intensity of the whiteline also decreased continuously
as the reaction proceeded, though at a slower rate than that obtained during hydrogenation at
100 °C. Starting from 24 % of Au3+ after pretreatment, the Au3+ fraction was 20 % after 10
min and 10 % after 30 min of the reaction. After 60 min, reduction was complete and the
spectra resembled the spectrum of the gold foil. Shown in Figure 4.6 is the conversion of
nitrobenzene at 100 °C after pretreatment at 60 and 100 °C as determined from the intensity
of the IR band of the asymmetric stretching of the nitro-group at ν = 1530 cm-1.
Figure 4.6 Hydrogenation of nitrobenzene after pretreatment of Au/CeO2 at 60 °C (red triangles) and
100 °C (black squares) at 100 °C and 10 bar H2 as determined by the decreasing amount of
nitrobenzene in the reaction. The corresponding fractions of Au3+ (yellow triangles for Au/CeO2
pretreated at 60 °C and black crossed squares for Au/CeO2 pretreated at 100 °C) are plotted against
the reaction time. For both pretreatments, two straight lines between 0 and 10 min and between 10 and
30 min, respectively, represent the conversion of nitrobenzene over time.
The corresponding fractions of Au3+ are plotted against the reaction time. The conversion
curves after each pretreatment can be divided into two more or less linear segments between
0 and 10 min and between 10 and 30 min reaction time, respectively. The initial activity (0-
10 min) of the catalyst pretreated at 100 °C, at which cationic gold was not detected, was
higher (0.19 mmol gAu-1 s-1) than that of the catalyst reduced at 60 °C, which contained
significant amounts of cationic gold (0.14 mmol gAu -1 s-1). Thus, it is clear that the cationic
gold present after treatment at 60 °C does not lead to high catalytic activity. After
approximately 10 min, the activity of Au/CeO2, pretreated at 60 °C, decreased to 40 % of its
initial activity, which results in 0.06 mmol gAu-1 s-1. Concurrently, the reduction of Au3+ was
complete. However, the catalytic activity of the fully reduced Au/CeO2 after pretreatment at
100 °C showed exactly the same behavior and also decreased to 40 % after approximately 10
Chapter 4
54
min, which results in 0.08 mmol gAu-1 s-1. Thus, no correlation has been found between the
catalytic activity and the amount of cationic gold; the decrease in activity is the same
irrespective of whether the sample contains measurable amounts of cationic gold.
4.4. Discussion
As shown by the intensive whiteline in the Au LIII HERFD spectra, as-prepared Au/CeO2
mainly contained gold in the 3+ oxidation state. During pretreatment at 100 °C, the reduction
of Au3+ was fast and complete. Pretreatment at 60 ° C led to slower reduction, and after
pretreatment, approximately 25–30 % cationic gold was left. The liquid phase hydrogenation
of nitrobenzene was performed at 100 °C and 10 bar H2, which are mild conditions, as
commonly reported in the literature for such catalytic systems.26 The catalyst without
measurable amounts of cationic gold was more active than the catalyst with cationic gold.
Thus, the observed cationic gold does not show high catalytic activity. Moreover, the fraction
of cationic gold present in Au/CeO2 pretreated at 60 °C was not maintained under reaction
conditions; it gets converted to Au0, and after 10 min, reduction was complete. Concurrently,
the activity of this catalyst decreased by 60 %. However, Au/CeO2, pretreated at 100 °C, was
fully reduced from the beginning and showed exactly the same deactivation behavior. After
approximately 10 min, the catalytic activity decreased by 60 %. Thus, the decrease in
catalytic activity is not related to the reduction of cationic gold, which is probably due to
poisoning of the catalyst surface by reaction intermediates and/or deposition of carbonaceous
species.29 After both pretreatments, the reaction eventually proceeded without detectable
amounts of cationic gold. The above results were confirmed by performing an additional
experiment. The hydrogenation of nitrobenzene over Au/CeO2, pretreated at 60 °C, was
performed at 60 °C instead of 100 °C. Reduction of Au3+ also occurred under these
conditions but was slower than that under reaction conditions at 100 °C. Significant amounts
of Au3+ were present for at least 30 min. The conversion of nitrobenzene was low but
constant during this period, although the fraction of Au3+ decreased from 25 to 10 %. Thus,
varying amounts of Au3+ did not influence the catalytic activity. If undetected cationic gold
were responsible for the catalytic reaction, then it would have been located at an undetectable
site and would have been very stable to maintain its oxidation state. We estimate that about 1
% of cationic gold can be found by using HERFD. For hydrogenation reactions over
supported gold catalysts, cationic and reduced gold can be highly active. Our results are in
good agreement with those of Hensen et al.,97 who studied the structure and oxidation state of
gold supported on ceria in the selective gas phase hydrogenation of 1,3-butadiene by means
The dynamic structure of gold supported on ceria in the liquid phase hydrogenation of nitrobenzene
55
of XAS. They found that the catalytic activity of 0.08 wt.% Au/CeO2 obtained after leaching
with use of NaCN, which comprised solely isolated Au3+, increased strongly with the
reduction temperature, that is, the amount of reduced gold. They further observed an
increasing catalytic activity of the as-prepared non-pre-reduced catalyst with time on stream,
which was ascribed to a slow reduction of Au3+ under reaction conditions. Thus, they also
observed changes in the catalyst structure under reaction conditions. Cationic gold prepared
through deposition-precipitation is inactive in the hydrogenation of nitrobenzene and 1,3-
butadiene. Our results contrast with those of Guzman and Gates,92,93 who concluded that
mononuclear Au3+ species supported on MgO were the catalytically active species in the
hydrogenation of ethylene. A cationic gold complex [Au(CH3)2(C5H7O2)] was deposited on
MgO. Extended X-ray absorption fine structure spectroscopy under working conditions did
not show Au–Au contributions; however, the XANES spectra exhibit a decrease in the white-
line intensity during the first 30 min of the reaction. The authors attribute this decrease to
changes in site symmetry of mononuclear gold species rather than to changes in the oxidation
state of gold, such as partial reduction. Homogeneous cationic complexes of gold are active
catalysts.126 This suggests that cationic gold in a heterogeneous catalyst might be active when
prepared from an appropriate cationic gold complex. However, many homogeneous reactions
proceed with both Au0 and Au3+ precatalysts, and the oxidation state of active species has not
yet been identified.126 Zhang et al.94 investigated the nature of active gold species in the
hydrogenation of 1,3-butadiene over Au/ZrO2 catalysts by preparing catalysts with different
fractions of Au3+. They concluded that site-isolated Au3+ ions, which they observed in
Au/ZrO2 catalysts with loadings lower than 0.1 wt.%, were the active sites for the
hydrogenation of 1,3-butadiene. In contrast, the catalysts were characterized only prior to the
reaction (temperature-programmed reduction and X-ray photoelectron spectroscopy), not
during or after the reaction. They further varied the calcination temperature of a 0.8 wt.%
Au/ZrO2 catalyst between 473 and 773 K to obtain catalysts with different Au3+/Au0 ratios,
whereas the fraction of cationic gold decreased with increasing calcination temperature, and
so did the catalytic activity. The decrease in catalytic activity was attributed to the decrease in
the fraction of cationic gold. However, with increasing calcination temperature, the mean
particle size increased significantly from 4 nm at a calcination temperature of 473 K to 7 nm
at 573 K and 12 nm at 773 K. Thus, the decrease in activity might also be a result of the
increasing particle sizes. With regard to the activation of hydrogen on supported gold
catalysts, several studies suggested that the presence of metallic gold particles is essential.
Mohr et al.127,128 identified the edges of single crystalline cubocatahedral gold nanoparticles
Chapter 4
56
with a mean diameter of 9 nm supported on ZnO as active sites for the selective
hydrogenation of the CO group of acrolein to allyl alcohol. Through selective decoration of
the gold faces by indium and leaving the edges uncovered, an increased selectivity to the
desired allyl alcohol was observed. Bus et al.129 investigated the interaction of hydrogen with
Au/Al2O3 and Au/SiO2 catalysts combining XAS, hydrogen chemisorption, and hydrogen–
deuterium (H/D) exchange experiments. They found that with decreasing particle size,
increasing amounts of hydrogen were chemisorbed and an increasing fraction adsorbed
strongly. The Au/Al2O3 catalyst with the smallest particle size of about 1 nm exhibited the
highest hydrogen uptake per surface atom. At this size, most of the surface consists of atoms
at corner and edge positions. Thus, hydrogen atoms may adsorb only at the edges and corners
of the gold particles. Additional H/D exchange experiments showed that hydrogen adsorbed
dissociatively on gold and that the adsorption was activated. By combining isotopic H/D
exchange experiments with IR and DFT results, Boronat et al.130,131 demonstrated that among
the different gold sites identified, only low coordinated, neutral gold atoms located at corner
or edge positions of Au/TiO2 catalysts were able to dissociate H2. Fujitani et al.132 studied the
H2/D2 exchange reaction over Au/TiO2 (110) surfaces with different gold particle sizes. With
decreasing gold particle size, the rate of H/D formation increased. For particle sizes below 2
nm, a marked increase in activity was observed. However, the apparent activation energies
for the H2/D2 exchange reaction were almost identical for all Au/TiO2 (110) model catalysts,
irrespective of the differences in gold particle sizes. Thus, the authors concluded that the
nature of the active sites for the dissociation of H2 over Au/TiO2 (110) was the same
irrespective of gold particle sizes and proposed that the gold atoms at the metal/support
interface were the catalytically active sites. Shimizu and co-workers133 investigated the
influence of the particle size and the nature of the support on the chemoselective
hydrogenation of nitroaromatics over supported gold catalysts. With regard to the gold
particle size, they observed an increasing activity for the OH/D2 exchange reaction with
decreasing particle size. Gold nanoparticles of similar mean particle sizes were found to be
most active when supported on an acid-base bifunctional support (Al2O3) rather than on a
basic (MgO) or acidic (SiO2) support. Thus, the authors concluded that these surface acid-
base pair sites were required for the dissociation of hydrogen and proposed that the gold
atoms at the metal-support interface were the catalytically active sites. We determined in situ
the oxidation state of gold in our highly responsive catalysts and simultaneously monitored
the conversion of nitrobenzene. In our system, when cationic gold remained after
pretreatment, it was reduced under reaction conditions. We found no evidence that the
The dynamic structure of gold supported on ceria in the liquid phase hydrogenation of nitrobenzene
57
detected cationic gold contributed to catalytic activity. However, the fully reduced catalyst
was more active than the partially oxidized catalyst.
4.5. Conclusions
In situ high-energy-fluorescence detection X-ray absorption near-edge structure spectroscopy
at the Au LIII edge during catalyst pretreatment and the liquid phase hydrogenation of
nitrobenzene revealed large changes in the oxidation state of gold in Au/CeO2 catalysts. Ex
situ characterization is not quantitative for the structure under catalytic conditions. The liquid
phase pretreatment of as-prepared Au/CeO2 at 100 °C led to complete reduction of gold
species, whereas pretreatment at 60 °C resulted in the incomplete reduction of gold species.
Reduced Au/CeO2 was more active in the liquid phase hydrogenation of nitrobenzene at 100
°C than was cationic gold containing Au/CeO2. When cationic gold was present at the
beginning of the reaction, it reduced under reaction conditions, which was accompanied by a
loss of 60 % activity. However, deactivation is not related to the amount of cationic gold,
because Au/CeO2 that was fully reduced at the beginning of the reaction became deactivated
in the same manner, probably as a result of blocking of the active surface. We did not find
that cationic gold prepared through deposition-precipitation on ceria is related to catalytic
activity.
Chapter 4
58
Chapter 5
Hydrogenation of nitrobenzene over Au/MeOx
catalysts - a matter of the support
(Reprinted with permission from Makosch, M., Sa, J., Kartusch, C., Richner, G., van Bokhoven, J. A. and Hungerbühler, K. ChemCatChem 4 (2012) 59. Copyright 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim)
[Martin Makosch performed the experiments, did the data analysis and wrote the manuscript]
Chapter 5
60
5.
5.1. Introduction
The heterogeneous hydrogenation of substituted nitrobenzenes is a reaction of great interest,
because aniline and its derivates are valuable substances in the chemical industry for the
production of polymers, pharmaceuticals, herbicides, and dyes.27 The state-of-the-art
catalysts are mostly active metals, such as Pt, Pd, Ni, Cu, and Ir, which are supported on
various materials, such as activated C, CaCO3, and SiO2, depending on their application.20 To
achieve high selectivity to substituted anilines in the presence of other reducible groups and
to prevent arylhydroxylamine accumulation in the reaction mixture, state-of-the-art catalysts
are often modified with environmentally harmful additives, such as Pb and V promoters and
Fe salts.34 Since the discovery that Au, when present as nanoparticles in the range of 1–3 nm,
catalyzes CO oxidation, more and more reactions have been shown to be catalyzed by Au,104
among them the hydrogenation of nitrobenzene.26 Hydrogenation of nitroaromatics
containing additional unsaturated groups over unmodified Au/TiO2 and Au/Fe2O3 shows a
high selectivity to the nitro-group. Thus, Au/MeOx (Me corresponds to a metal) catalysts
have been presented as a “green” alternative in reactions where a high selectivity under
moderate reaction conditions is required. Haber proposed a reaction scheme (Scheme 5.1) for
the electrochemical hydrogenation of nitrobenzene and its derivates in 1898;28 however, there
is an ongoing debate about the reaction mechanism over heterogeneous catalysts.
Hydrogenation of nitrobenzene over Au/MeOx catalysts – a matter of the support
61
Scheme 5.1 Possible reaction pathways for the hydrogenation of aromatic nitrocompounds to the
The evolution of the species in solution for a reaction of azoxybenzene over Au/CeO2, is
shown in Figure 5.2.
Figure 5.2 Evolution of substrate, intermediates, product, and C-balance in the liquid-phase
hydrogenation of azoxybenzene (0.4 mmol substrate, 10 bar H2, 100 °C) over Au/CeO2 (y-axis
relative to the stoichiometric coefficient of aniline); (�) azoxybenzene, (�) azobenzene, (■) aniline,
and (�) C-balance.
The amount of azoxybenzene decreased exponentially until azoxybenzene was fully
converted after 250 min. Azobenzene was the intermediate product in the formation of
aniline. Both azobenzene and aniline were detected in the liquid phase already after 5 min of
reaction. The azobenzene concentration reached a maximum at approximately 90 min, after
which it steadily dropped until the end of the reaction. At 250 min, the azobenzene
concentration was approximately 25 % of its maximum concentration at 90 min. The C-
balance slightly fluctuated in the first 15 min of the reaction and remained above 95 %
throughout the whole reaction. The evolution of the different species detected in the liquid
0 50 100 150 200 2500.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
C-b
ala
nce
/ %
n /
mm
ol
t / min
0
20
40
60
80
100
Chapter 5
66
phase using nitrosobenzene as the starting material is shown in Figure 5.3a for Au/TiO2 and
in Figure 5.3b for Au/CeO2.
Figure 5.3 Evolution of substrate, intermediates, product, and C-balance in the liquid-phase
hydrogenation of nitrosobenzene (0.4 mmol substrate, 10 bar H2, 100 °C) over a) Au/TiO2 and b)
Au/CeO2 (y-axis relative to the stoichiometric coefficient of aniline); (�) nitrosobenzene, (�)
azoxybenzene, (�) azobenzene, (■) aniline, and (�) C-balance.
Because of earlier results, which showed that a high nitrosobenzene concentration poisoned
the reaction29, we decreased the concentration of nitrosobenzene in this experiment from 0.8
to 0.4 mmol, which did not result in catalyst poisoning. All other reaction conditions were the
same as in Figure 5.1. Over Au/TiO2 (Figure 5.3a), the nitrosobenzene concentration dropped
to zero within the first 10 min of the reaction. After 5 min, azoxybenzene, azobenzene, and
aniline were already detected in the liquid phase. Both azoxybenzene and azobenzene reached
maxima after 5 min of reaction. Azoxybenzene was already completely converted after 15
min, whereas the amount of azobenzene constantly dropped throughout the reaction and
reached 0 mmol after 120 min. The aniline concentration increased steadily during the
reaction. The C-balance strongly decreased during the first 20 min of the reaction to 70 %,
but then increased with a similar slope as the aniline concentration, and finally reached a
value exceeding 90 %. The conversion rate of nitrosobenzene was lower over Au/CeO2
(Figure 5.3b). Again, the maximum azoxybenzene concentration was reached after 5 min. In
contrast to Au/TiO2, the azobenzene concentration reached its maximum after 15 min, and
azobenzene was slowly converted into aniline. The C-balance had an unconventional
evolution, as it started at 60 % and rose throughout the reaction to 100 % at the end of the
reaction. Intermediates probably formed at the beginning of the reaction, which were not
detected because adsorption on the catalyst led to an imperfect C-balance in the liquid phase.
0 20 40 60 80 1000.0
0.1
0.2
0.3
0.4
0.5
0 40 80
b)
C-b
ala
nce / %
n / m
mol
a)
t / min
0
20
40
60
80
100
Hydrogenation of nitrobenzene over Au/MeOx catalysts – a matter of the support
67
Nitrosobenzene is a candidate for such an adsorbed intermediate. The concentration profile of
the nitrobenzene hydrogenation over Au/TiO2 with addition of CeO2 is shown in Figure 5.4a.
Figure 5.4 a) Evolution of substrate, intermediates, product, and C-balance in the liquid-phase
hydrogenation of nitrobenzene (0.8 mmol substrate, 10 bar H2, 100 °C) on a Au/TiO2 catalyst with the
addition of an equal mass amount of pure CeO2 (y-axis relative to the stoichiometric coefficient of
aniline); (�) nitrobenzene, (�) azoxybenzene, (�) azobenzene, (■) aniline, and (�) C-balance. b)
Evolution of the nitrobenzene concentration in the liquidphase hydrogenation of nitrobenzene (0.8
mmol substrate, 10 bar H2, 100 °C) on a Au/TiO2 catalyst with (����) and without the addition of pure
CeO2 (����).
In contrast to the CeO2 free reaction, azoxybenzene and azobenzene were detected in a
similar time dependence as the reaction of nitrobenzene over Au/CeO2 (Figure 5.1b). The C-
balance was > 95 % throughout the reaction, except after 5 min, when it dropped to
approximately 80 % at the onset of azoxy- and azobenzene production. In contrast to the
reaction over pure Au/TiO2 (Figure 5.1a), the nitrobenzene and aniline concentrations did not
mirror each other, which would be indicative for the formation of reaction intermediates.
Adding TiO2 to Au/CeO2 did not lead to any changes in the concentration profile of pure
Au/CeO2 (not shown). The nitrobenzene concentration profiles of the experiments with
(hexagons) and without (pentagons) the addition of CeO2 are shown in Figure 5.4b. The
nitrobenzene concentration dropped faster in the first 20 min when CeO2 was added to the
mixture. The enhanced conversion of nitrobenzene after addition of CeO2 correlated
quantitatively with the amounts of azo- and azoxybenzene, suggesting that CeO2 initiated an
additional parallel reaction channel in the reaction. Hydrogenation of all the known
intermediates in the direct and condensation route over bare TiO2 and CeO2 were tested to
identify the differences between the two supports. The only difference was found in the
0 20 40 60 800.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0 20 40 60 80 100
C-b
ala
nce / %
b)n
/ m
mol
a)
t / min
0
20
40
60
80
100
Chapter 5
68
reaction of phenylhydroxylamine, which is a rather unstable species and which decomposed
faster into nitrosobenzene and aniline in the presence of CeO2 than in the presence of TiO2.
5.4. Disucssion
For the hydrogenation of nitrobenzene, both direct and condensation routes were observed.
Azoxybenzene could be formed through the condensation of nitrosobenzene and
phenylhydroxylamine. Thus, for the direct route to occur, nitrosobenzene and
phenylhydroxylamine should be present in low concentrations to prevent azoxybenzene
formation. However, the intermediates phenylhydroxylamine and nitrosobenzene can also
form azoxybenzene by themselves. Phenylhydroxylamine easily disproportionates into
nitrosobenzene and aniline.136 The so-formed nitrosobenzene can then condensate with
another molecule of phenylhydroxylamine to form azoxybenzene. In the case of
nitrosobenzene, two molecules can form a dimer and then react with another molecule of
nitrosobenzene to form azoxybenzene and nitrobenzene.137 Thus, for the condensation route
to occur, accumulation of nitrosobenzene and/or phenylhydroxylamine is necessary to form
azoxybenzene according to one of the ways described above, which is subsequently
converted into aniline via azobenzene. The direct route dominates over Au/TiO2,48,136 and
accumulation of phenylhydroxylamine on the surface of the catalyst has been observed.29 We
found that hydrogenation was sufficiently fast over Au/TiO2 to prevent condensation of
phenylhydroxylamine with nitrosobenzene, which was formed either from nitrobenzene or by
phenylhydroxylamine decomposition, but slower than over Au/CeO2. Phenylhydroxylamine
has been found as an intermediate on the surface of the catalyst during the liquid-phase
hydrogenation of nitrobenzene by using liquid-phase attenuated-total-reflectance (ATR FT-
IR) measurements.29 FT-IR measurements have revealed that nitrosobenzene reacts to aniline
via phenylhydroxylamine in the gas phase.48 In contrast, we identified azoxybenzene as the
primary product by using liquid-phase experiments, which was transformed into aniline via
azobenzene over both Au/TiO2 and Au/CeO2. Most likely, the condensation route was
suppressed because the molecules that form azoxybenzene could not condensate in the gas
phase. Thus, the reaction over Au/TiO2 can also proceed through the condensation route
when the concentration of nitrosobenzene is high enough. Au/TiO2 rapidly converts
nitrosobenzene, which is also converted rapidly over Au/CeO2; however, the large loss of C-
balance suggests that initially a large amount adsorbs on CeO2, which leads to a high surface
concentration. Hydrogenation of nitrobenzene over Au/CeO2 (Figure 5.1b) proceeds through
the condensation route and at a considerably slower rate than over Au/TiO2. This lower
Hydrogenation of nitrobenzene over Au/MeOx catalysts – a matter of the support
69
hydrogenation rate and the fast decomposition of phenylhydroxylamine could lead to a
buildup of nitrosobenzene molecules on the surface of CeO2, which allows for the
condensation route to occur. Adsorption of large amounts of nitrosobenzene can be assumed
because the C-balance is reduced directly after exposing CeO2 to nitrosobenzene (Figure
5.3b). Adding CeO2 to Au/TiO2 yields the condensation products azoxybenzene and
azobenzene, which are probably formed through nitrosobenzene accumulation on the CeO2
surface. The nitrobenzene concentration drops faster when CeO2 is added to the reaction
mixture when Au/TiO2 is used as the catalyst. In addition to the direct route that occurs over
Au/TiO2, CeO2 catalyzes the condensation route through accumulation of
phenylhydroxylamine and its decomposition into nitrosobenzene or through accumulation of
nitrosobenzene directly. The result is a reaction that occurs through the condensation route.
The direct route is not strongly affected, probably because the surface concentration of
phenylhydroxylamine on Au/TiO2 remains sufficiently high to poison the reaction.29
5.5. Conclusion
The hydrogenation of nitrobenzene over Au/TiO2 proceeds through the direct route, whereas
the hydrogenation reaction over Au/CeO2 proceeds through the condensation route. For the
condensation route to occur, a high (surface) nitrosobenzene concentration is necessary. In
the case of Au/TiO2, nitrosobenzene is rapidly converted into phenylhydroxylamine, which
accumulates on the surface and is then transformed to aniline. The concentration of
nitrosobenzene is never high enough to form azoxybenzene. For Au/CeO2, the rate of
hydrogenation is considerably lower, and the conversion of nitrobenzene and nitrosobenzene
are slower; as a result, nitrosobenzene can accumulate and form condensation intermediates.
An additional path to nitrosobenzene is the decomposition of phenylhydroxylamine, which is
especially fast over the CeO2 support. Furthermore, the CeO2 support catalyzes the
condensation. The support has a direct impact on the reaction mechanism and actively
changes the reaction route.
Chapter 5
70
Chapter 6
Organic thiol modified Pt/TiO2 catalysts to control
chemoselective hydrogenation of substituted
nitroarenes
(Reprinted with permission from Makosch, M.; Lin, W.-I.; Bumbálek, V.; Sá, J.; Medlin, J. W.; Hungerbühler, K.; van Bokhoven, J. A. ACS Catal. 2012, 2079. Copyright 2012 American Chemical Society.) [Martin Makosch performed the experiments, did the data analysis and wrote the manuscript]
Chapter 6
72
6.
6.1. Introduction
Hydrogenation reactions over heterogeneous catalysts are of industrial as well as of scientific
interest.21 An optimal hydrogenation catalyst combines high activity, selectivity, and stability.
For generally employed supported metals in hydrogenation catalysts such as Pt and Pd,
selectivity control is an issue when more reducible groups are present in the same molecule.
An example for such a reduction is the hydrogenation of substituted nitroarenes to the
corresponding anilines which are important substrates for pharmaceuticals, dyes and
pigments.27 The conventional Pt and Pd catalysts simultaneously reduce the nitro- and all
other reducible groups in the molecule. A way to circumvent this problem is to use less active
metals, such as gold26,37,136 or to modify the more active catalysts.20,34,35 There are various
surface modifications reported in the literature for a variety of catalytic challenges.23,138-141
Modification via ligands is a well established method to tune the performance of catalysts.
Recently PVP stabilized Rh nanoparticles were modified via phosphine ligands to tune the
performance during the hydrogenation of substituted aromatics.141 Upon modification with
several different bulky phosphine ligands, the selectivity towards ring hydrogenation could be
increased to values greater than 90 % during the liquid phase hydrogenation of
phenylacetone. In this study, we report a new simple surface modification procedure for
supported Pt particles on TiO2 employing organic thiols to selectively hydrogenate 4-nitro- to
4-aminostyrene in the liquid phase. Scheme 6.1 summarizes the effect of our modification.
Scheme 6.1 Products observed during the liquid phase hydrogenation of 4-nitrostyrene at 80 °C under
10 bars H2 in toluene over an unmodified Pt/TiO2 (left) and an organic thiol modified Pt/TiO2 catalyst
(right).
Performing the hydrogenation over an unmodified Pt/TiO2 catalyst yields 4-
ethylnitrobenzene and 4-ethylaniline simultaneously. Upon modification with organic thiols
the selectivity of the catalyst can be switched so that the primary product of the
hydrogenation reaction is exclusively 4-aminostyrene.
Organic thiol modified Pt/TiO2 catalysts to control chemoselective hydrogenation of substituted nitroarenes
73
6.2. Experimental section
6.2.1. Catalyst preparation and modification
All Pt/TiO2 catalysts used in this work were synthesized via incipient wetness impregnation.
To obtain a nominal 1 wt% Pt-loading, 80.2 mg tetra-amine-platinum (II) nitrate
(Pt(NH3)4(NO3)2, Aldrich Chemicals) were dissolved in 3.7 ml deionized water. This solution
was added dropwise to 4 g of support (TiO2 P25 Acros) under vigorous mixing. The resulting
powder was heated to 200 °C at a rate of 5 °C min-1 and kept at this temperature for 4 hours.
Subsequently, the powder was further heated to 400 °C at a rate of 5 °C min-1 and kept at this
temperature for 4 hours and then cooled down to room temperature. We refer to this catalyst
in the following as “as prepared”. Prior to reaction and modification, the as prepared catalyst
was pretreated in a flow of 100 ml min-1 5 % H2/He (v/v) at 250 °C (heating rate 2 °C min-1)
for 2 h. We refer to this catalyst as Pt/TiO2H250. A total of 600 mg of the freshly reduced
Pt/TiO2H250 catalyst was added to 100 ml of a 30 mM solution of the corresponding thiol in
ethanol (ethanol absolute, analytical grade, Scharlau) to obtain a nominal Pt/thiol ratio of
1:100. Prior to the addition of the catalyst the thiol/ethanol solution was always purged with
nitrogen under magnetic stirring (750 rpm) for 30 min. After the addition of the Pt/TiO2H250
catalyst the suspension was stirred (750 rpm) for 16 h under argon purging at room
temperature, filtered and washed three times with 125 ml ethanol and dried in vacuum
overnight.
6.2.2. Kinetic measurements
All hydrogenation reactions were performed in 50 ml Premex stainless steel autoclaves with
polyetheretherketone (PEEK) inlets. A typical reaction composition consisted of solvent (20