r
n
Applied Catalysis B: Environmental 91 (2009) 470–480
cr
ru
n
in
nd
xi
d
io
Contents lists available at ScienceDirect
Applied Catalysis B
l
1. Introduction
Oxide supported Au catalysts have been demonstrated to be
highly active already at low temperatures for various oxidation and
reduction reactions,mostprominently theCOoxidation reaction [1–
3], which in turn is of interest in environmental catalysis
applications such as the preferential oxidation of CO in H2 feed
gases for fuel cells. It turned out, that in addition to other properties,
the nature of the supporting oxide has a significant effect on the
activity of the catalyst [4–7]. Recently, mesoporous oxides were
introduced as alternative support materials, which may be
attractive, e.g., because of their high surface area or the stabilization
of the Au nanoparticles in the mesopores of the support [7–15]. For
example, Au catalysts supported on mesoporous aluminosilicates
showed a higher COoxidation activity [9–11] thanAu/SiO2 catalysts
[6,16]. Similar observations of a higher activity were reported also
for the much more active mesoporous Au/TiO2 catalysts [8,12–
15,17], when comparing with highly active Au/TiO2 catalysts based
on non-porous titania [2,18–24]. Comparing different mesoporous
Au catalysts, Overbury et al. found a much higher CO oxidation
activity on Au/TiO2 than on Au/SiO2 in temperature screening
measurements [7]. Furthermore, it was reported that also the
crystalline phase of the (non-porous or mesoporous) TiO2 support
influences the COoxidation activity [13,25–27]. Schwarz et al. found
that the activity of non-porous Au/TiO2 catalysts decreased from
brookite via anatase to rutile for catalysts activated by calcination at
300 8C, while after reduction at 150 8C the activities were compar-
able [25], and similar observations were reported by Yan et al. for
mesoporous TiO2 supported catalysts [26]. These authors explained
this by an increasing tendency for Au particle sintering during
calcinationat300 8C intheorderbrookite < anatase< P25(anatase/
rutile)� rutile Au/P25 (non-porous rutile/
anatase) > Au/rutile (mesoporous) [13]. In a comparative study on
the activity of porous and non-porous Au/TiO2 catalysts prepared
from different crystalline phases of TiO2 (anatase, rutile, P25),
Comotti et al. concluded that the crystalline phase has no significant
influence on the catalytic properties of the unconditioned catalysts
[27]. This was different when the catalysts were calcined before
reaction,where the P25 based catalyst showeda significantlyhigher
thermostability than catalysts supported on anatase or rutile.While
after calcination at 250 8C, the activities are comparablewith that of
the unconditioned catalysts and similar for all catalysts, calcination
at 350 8C leads to lower activities for the (mesoporous) rutile or
(mesoporous) anatase supported catalysts, while the P25 supported
catalyst retains its activity. The decay in activity, however, is not
caused by Au nanoparticle sintering, since the mean particle size of
�3.0 nm was identical for 250 and 350 8C calcination and all
catalysts.On theotherhand,Wangetal. reporteda significanthigher
Deactivation
Au/TiO2
CO oxidation
� 2009 Elsevier B.V. All rights reserved.
* Corresponding author. Tel.: +49 731 50 254 50; fax: +49 731 50 254 52.
E-mail address: juergen.behm@uni-ulm.de (R.J. Behm).
0926-3373/$ – see front matter � 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.apcatb.2009.06.016
Influence of the crystalline phase and su
the CO oxidation activity of mesoporous
Y. Denkwitz a, M. Makosch a, J. Geserick b, U. Ho¨rma
a Institute of Surface Chemistry and Catalysis, Ulm University, D-89069 Ulm, Germany
b Institute of Inorganic Chemistry I, Ulm University, D-89069 Ulm, Germany
cCentral Facility of Electron Microscopy, Ulm University, D-89069 Ulm, Germany
A R T I C L E I N F O
Article history:
Received 13 March 2009
Received in revised form 15 June 2009
Accepted 18 June 2009
Available online 24 June 2009
Keywords:
Mesoporous catalysts
Activity
A B S T R A C T
The influence of the TiO2
deactivation behavior of st
loading in the CO oxidatio
reaction conditions and by
were controlled by the pH a
loading of themesoporous o
catalysts. The resulting tren
stability/deactivation behav
journa l homepage: www.e
face area of the TiO2 support on
Au/TiO2 catalysts
n c, S. Selve c, U. Kaiser c, N. Hu¨sing b, R.J. Behma,*
ystalline phase and of the surface area on the activity and stability/
cturally well-defined mesoporous Au/TiO2 catalysts with comparable Au
reaction was investigated by kinetic measurements under differential
situ DRIFTS. The crystalline phase and surface area of the TiO2 substrate
the type of the structure-directing surfactants applied in the synthesis. Au
des was performed by the same deposition–precipitation procedure for all
s in the CO oxidation behavior, including the TOF based activities and the
r are discussed.
: Environmental
sev ier .com/ locate /apcatb
Table 1
TiO2 crystallite size, BET surface area, pore diameter, pore volume, Au loading and weight loss (TGA) of Au/TiO2 catalysts based on different (mesoporous) support materials.
Catalyst TiO2 support
synthesized
with
pH Crystalline
phase
TiO2 crystallite
size (nm)
BET surface
area (m2 g�1)
Pore
diameter
(nm)
Pore volume
(cm3 g�1)
Au loading
(wt.%)
Weight loss
<500 8C (wt.%)
Weight loss
<990 8C (wt.%)
Au/TiO2(P25) – – Anatase/rutile 21/30 58 – – 3.1 2.7 2.8
Au/TiO2(1) Brij56, EGMT 2 Anatase 9.9 106 7.5–8.2 0.18 3.0 4.6 4.8
Au/TiO2(2) Brij56, EGMT 2 Anatase 9.9 106 7.5–8.2 0.18 2.6 3.9 4.1
Au/TiO2(3) SDS, EGMT 2 Anatase 6.6 175 4.4–6.3 0.28 2.6 8.5 9.2
Au/TiO2(4) SDS, EGMT �1.2 Anatase/rutile 5.6 240 3.5 0.32 4.2 11.0 11.8
Au/TiO2(5) SDS, EGMT 0 Rutile 6.7 160
a 0.22 4.8 6.8 9.3
a Not well defined.
Y. Denkwitz et al. / Applied Catalysis B: Environmental 91 (2009) 470–480 471
activity of Au catalysts based on mesopous rutile (5% anatase) after
calcination at 500 8C compared to Au/P25 [14]. Moreau et al.
reported a higher activity of a non-conditioned Au/rutile catalyst
(100 m2 g�1) compared to a similarly treated Au/anatase catalyst
(90 m2 g�1) [24]. These authors also studied the influence of
the surface area of the anatase support on the CO oxidation activity,
and found that the activity decreases with increasing surface area
in the order 37 m2 g�1 > 45 m2 g�1 > 90 m2 g�1 > 240 m2 g�1 >
305 m2 g�1. Comparable observations of a decreasing activity with
increasing BET surface area were reported also by Zhu et al. for
mesoporousAu/TiO2 catalysts, comparing a55m
2 g�1 catalyst anda
141 m2 g�1 catalyst [15]. Summarizing these results, theactivityand
stability of mesoporous Au/TiO2 catalysts seem to be affected by
various properties, including (i) the size of the oxide crystallites and
Au nanoparticles and the respective surface areas, (ii) the structure
and size of the oxide pores, and (iii) the crystalline phase of the TiO2
support (anatase/rutile). Clear trends, however, can hardly be
extracted from these data because of the wide variations in
parameters.
For a more detailed understanding, we started a systematic
study on the CO oxidation behavior of Au catalysts based on
mesoporous TiO2 whose structural properties were systematically
varied. Recently we reported that mesoporous Au/TiO2 catalysts
prepared by sol–gel processing and subsequent Au deposition on
the mesoporous TiO2 via a deposition–precipitation procedure
show a higher CO oxidation activity than catalysts prepared in a
similar way from a commercially available (P25, Degussa) non-
porous support, with a comparable Au particle size and Au metal
loading [12]. In the present paper, we report on the influence of the
TiO2 crystalline phase of mesoporous Au/TiO2 catalysts on their CO
oxidation activity and deactivation behavior. Mesoporous TiO2
materials with high surface areas and of different crystalline
phases were synthesized via sol–gel processing as described in Ref.
[28]. For comparison, we also include a standard Au/TiO2 catalyst
based on a P25 support. The structure and surface composition of
the catalysts were characterized by X-ray diffraction (XRD),
transmission electron microscopy (TEM), X-ray photoelectron
spectroscopy (XPS) and temperature programmed desorption
(TPD) (Section 3.1). Stimulated by our recent finding of a
pronounced effect of a catalyst drying step prior to catalyst
Table 2
Mean Au particle size, Au(4f):Ti(2p) intensity ratio, absolute and relative activity for C
calcination in 10% O2/N2 at 400 8C (30 min) (N15-O400) and after re-calcination in 10%
Catalyst Au loading
(wt.%)
Au-particle
size (nm)
Au(4f):Ti(2p)
ratio
Initial/fina
(�10�4 mo
Au/TiO2(P25) 3.1 2.9 � 0.8 0.13 34.3/9.4
Au/TiO2(1) 3.0 4.1 � 0.8 0.05 84.1/27.9
Au/TiO2(2) 2.6 3.7 � 0.9 0.04 87.1/25.7
Au/TiO2(3) 2.6 3.0 � 0.7 0.04 37.2/12.4
Au/TiO2(4) 4.2 6.1 � 1.3 0.04 33.9/6.0
Au/TiO2(5) 4.8 6.1 � 1.5 0.03 34.8/9.2
conditioning on the CO oxidation behavior of P25 supported
Au/TiO2 catalysts [29], we also evaluated the influence of a similar
catalyst pre-treatment on the reaction behavior (Section 3.2). The
activities as well as the deactivation and re-activation behavior
during CO oxidation were evaluated by kinetic measurements
under differential reaction conditions (Section 3.3). Mechanistic
informationwas obtained from in situ IRmeasurements performed
under similar reaction conditions, employing diffuse reflectance IR
spectroscopy (DRIFTS).
2. Experimental set-up and procedures
2.1. Preparation of the mesoporous TiO2 support and the Au catalysts
The mesoporous TiO2 support material was prepared via an
ethylene glycol-modified titanium precursor (bis(2-hydro-
xyethyl)titanate, EGMT) [28]. EGMTwas synthesized bymodifying
a procedure described by Xia et al. [30]. For the catalysts (1) and
(2), the mesoporous TiO2 support was synthesized by dissolving
polyethylene oxide hexadecylether (Brij56) in dilute hydrochloric
acid (pH 2). Subsequently, EGMT was added, and the solution was
ultrasonicated for 5 h at 60 8C. The resulting suspension was aged
at 60 8C for 24 h, and then centrifuged. The precipitate was washed
withwater three times and subsequently calcined at 400 8C for 4 h.
The mesoporous support for the catalysts (3)–(5) was synthesized
with sodium dodecyl sulfate (SDS) and EGMT to obtain higher
surface areas. The different polymorphs of titania (anatase and
rutile) were accessible by variation of the pH-value (see Table 1).
The Au/TiO2 catalysts were prepared via a deposition–
precipitation procedure described in [23,31]. In short, TiO2 was
suspended in water at 60 8C at a pH of 5–5.5, followed by addition
of tetrachloroauric acid (HAuCl4�4H2O). A constant pH value was
maintained by addition of Na2CO3 solution. After additional 30 min
of stirring, the precipitate was cooled to room temperature,
filtered, washed and dried overnight at room temperature under
vacuum. The catalysts were stored in the dark at �10 8C. The Au
metal loading (Table 2) was determined by inductively coupled
plasma atom emission spectroscopy (ICP-AES). If not mentioned
otherwise, the raw catalysts were dried in 20 N ml min�1 N2 at
100 8C for 15 h, then calcined (10% O2 in N2) at 400 8C for 30 min,
O oxidation of various Au/TiO2 catalysts after drying in N2 at 100 8C (15 h) plus
O2/N2 at 400 8C (30 min).
l activity
l s�1 gAu
�1)
Initial/final
TOF (s�1)
Activity
after 1000 min
reaction (%)
Initial activity
after re-calcination (%)
2.0/0.5 28 n.m.
6.4/2.1 33 n.m
6.3/1.9 30 n.m.
2.2/0.7 33 94
4.0/0.7 18 82
4.2/1.1 26 84
Administrator
打字机
Administrator
打字机
Administrator
打字机
乙二醇钛
Administrator
打字机
Administrator
打字机
Administrator
打字机
Administrator
打字机
Administrator
打字机
Administrator
打字机
Administrator
打字机
Administrator
打字机
Administrator
打字机
adsorbate concentration [36]. For possible deviations from this
linear behavior see [37]. Background subtraction and normal-
ization of the spectra was performed by subtracting spectra
recorded in a flow of N2 at the reaction temperature directly before
starting CO oxidation. To remove changes in the reflectivity of the
respective catalysts, the spectra were scaled to similar background
intensity at 2430 cm�1, which did not interfere with any other
signals and where the shape of the raw spectrum changed little
during the measurements. The gas phase CO signal was removed
by subtracting the spectral region of gas phase CO (2040–
2240 cm�1) from a spectrum recorded on pure a-Al2O3 in CO
containing atmosphere, which was verified in test experiments to
not adsorb CO under present reaction conditions.
3. Results and discussion
3.1. Catalyst surface properties
In the following, we present and discuss results on the structure
and surface composition of the different Au/TiO2 catalysts, which
were obtained by different techniques including X-ray diffraction,
Y. Denkwitz et al. / Applied Catalysis B: Environmental 91 (2009) 470–480472
and finally cooled down to the reaction temperature in N2 before
the reaction measurement. The concentration of Cl was below the
detection limit of the ICP-AES and XPS measurements.
2.2. Catalyst characterization
The surface area and the pore diameter of the mesoporous
support material and of the different catalysts were determined by
N2 sorption measurements (Autosorb MP1, Quantachrome). The
resulting surface areas were calculated using the Brunauer–
Emmett–Teller (BET) relation in the p/p0 range of 0.05–0.30. The
pore size distributionwas evaluated from the desorption branch of
the isotherms, using the procedure developed by Barrett, Joyner
and Halenda (BJH) [32]. XRD measurements were performed on a
PANalytical MPD PRO instrument, using Cu-Ka radiation
(l = 0.154 nm) to evaluate the titania structure and the titania
particle sizes. The size and particle distribution of the Au
nanoparticles were derived from TEM images obtained in a Philips
CM 20 instrument (200 kV). Typically, several hundred Au
nanoparticles (>400) were evaluated per sample for determining
the size distribution of the Au nanoparticles. The relative surface
concentrations and the oxidation states of Au, Ti, and O were
determined by X-ray photoelectron spectroscopy, using mono-
chromatized Al-Ka radiation for excitation (PHI 5800 ESCA system,
Physical Electronic). The binding energies (BEs) were calibrated
with respect to the bulk Ti(2p3/2) signal at 459.2 eV for Ti
4+ [33].
Subtraction of a Shirley background and peak fitting were
performed using a public domain XPS peak fit program
(XPSPEAK4.1 by R. Kwok). Thermogravimetry (TGA) measure-
ments were performed on a STA 449C Jupiter instrument (Netsch
Gera¨tebau GmbH). Finally, temperature programmed desorption
measurements were performed in a quarz tube micro reactor,
which was connected to a mass spectrometer (Atomica IMR-MS
1500). TPD spectra were recorded in a N2 stream with a
temperature ramp of 5 8C min�1, starting at 30 8C.
2.3. Activity measurements
The catalytic activity of the catalysts was determined in
reaction measurements performed at atmospheric pressure in a
quartz tube micro-reactor under differential reaction conditions,
using typically 65–70 mg diluted catalyst powder. In order to limit
the conversion to values between 5 and 20%, the catalyst was
diluted with a-Al2O3, which is not active for CO oxidation under
present reaction conditions. The measurements were carried out
with a gas flow of 60 N ml min�1 in 1 kPa CO, 1 kPa O2 and balance
N2 at 80 8C. Incoming and effluent gases were analyzed by on-line
gas chromatography (Dani GC 86.10HT), using H2 as carrier gas.
High-purity reaction gases (CO: 4.7, O2: 5.0 and N2: 6.0;
Westphalen) were passed through a moisture filter (Varian GC-
MS filter CP 17973) to ensure water concentrations of below
0.1 ppm. The reaction rates were determined from the CO2 partial
pressure; for further information on experimental details see ref.
[34].With rates below 10�5 mol s�1 cm�3, mass and heat transport
problems were negligible [35].
2.4. Infrared measurements
In situ IR measurements were performed by diffuse reflectance
infrared fourier transform spectroscopy using a commercial in situ
reaction cell (Harricks, HV-DR2). The spectra were recorded in a
Magna 560 spectrometer (Nicolet) equipped with a MCT narrow
band detector. About 20 mg of a-Al2O3 diluted catalyst (dilution
1:2) were used as catalyst bed. Typically, 400 scans (acquisition
time 3 min) were co-added per spectrum. The intensities were
evaluated in Kubelka-Munk units, which are linearly related to the
N2 sorption, thermogravimetric analysis, infrared spectroscopy,
transmission electron microscopy as well as X-ray photoelectron
spectroscopy.
Based on the characteristic reflections in the diffractograms (a)
and (b) of the as-synthesized catalysts in Fig. 1, the synthesis of
mesoporous TiO2 with EGMT and Brij56 or SDS at pH 2 results in
anatase. In contrast, for TiO2 synthesized at pH < 0, only the
characteristic rutile reflections are observed (Fig. 1, diffractogram
(d)), while for synthesis at pH � 1.2 reflections of anatase and rutile
coexist (Fig. 1, diffractogram (c)). Obviously, the resulting crystal-
line phase of TiO2 depends sensitively on the pH during the
synthesis (see also Table 1) [28], in addition to the choice of the
titania precursor and the temperature [38]. Additionally, a catalyst
supported on commercial, non-porous TiO2 (P25, Degussa) was
characterized (Fig. 1e), which shows reflections of anatase and
rutile [26]. Based on the width of the reflections, the crystallite
sizes of the mesoporous supports are rather small, about 10 nm for
the TiO2 obtained when using Brij56 as surfactant and about 6–
7 nmwhen using SDS, independent of the crystalline phase of TiO2.
Since no Au reflections were observed in the raw catalysts by XRD,
the Au particles are too small for detection by XRD. For the Au/
TiO2(1) and Au/TiO2(2) catalysts, which are based on the same
support material, we obtained BET surface areas of 106 m2 g�1 and
Fig. 1. XRD pattern of the different as-synthesized Au/TiO2 catalysts: (a) Au/TiO2(2),
which is supported on the same material as catalyst (1), (b) Au/TiO2(3), (c) Au/
TiO2(4), (d) Au/TiO2(5), (e) Au/TiO2(P25).
pore diameters of 8 nm. The interaction of SDS as the structure-
directing agent in the synthesis protocol with the growing titania
species is due to its ionic character completely different from that
of the previously applied Brij56 as a non-ionic surfactant (samples
(1) and (2)). In our case, this interaction results in larger specific
surface areas of 175, 240 and 160 m2 g�1 for samples (3)–(5),
respectively.
The thermal stability of the different unconditioned catalysts
was characterized by TGA. With increasing BET surface area, the
weight loss of the mesoporous catalysts increased (see Table 1).
This weight loss is mainly attributed to the desorption of
molecularly adsorbed water and recombination of OH groups
[39], whose amount increases with increasing BET surface area. At
higher temperatures (>500 8C), the catalyst weight loss (0.7–2.6%,
see Table 1) is caused by reactive desorption of carbon containing
species, which are residuals from the TiO2 synthesis (see also
Section 3.2). These species are also visible in IR spectra recorded on
the different catalysts (Fig. 2). Prior to the IR measurements, the
catalysts were dried at 100 8C for 15 h and calcined at 400 8C.
Signals at 2352 and 1358 cm�1 belong to the carbon containing
species, which are not observed on the P25 based catalyst (Fig. 2,
spectrum (e)). The first signal is usually related to gas phase CO2
resulting from CO oxidation (CO2 was