Mechanistic study of propane selective oxidation with H2 and O2 on Au/TS-1

Article abstract of DOI:10.1016/j.jcat.2008.04.004

The selective oxidation of propane to acetone and 2-propanol with H₂ and O₂ was studied on Au/TS-1 by kinetic and spectroscopic analysis. A kinetic study using a factorial design at conditions where the catalyst was stable and gave propane conversions of <6% and combined oxygenates selectivity of >90%, resulted in a power-rate law expression of the form r_oxyg. = k_oxyg.(H₂)^0.74(O₂)^0.36(C₃H₈)^0.29. In situ Au L₃-edge X-ray absorption near-edge spectroscopy (XANES) measurements showed activation of O₂ on Au, whereas in situ ultraviolet-visible (UV-vis) spectroscopy evidenced the presence of Ti-hydroperoxo species. The role of the Ti-hydroperoxo species was probed by a transient technique in which changes in Ti K-edge XANES spectra were used to determine the evolution of coverage with time (d θ / d t). It was shown that the rate of reaction by XANES (6.7 × 10^-4 s^-1) was close to the turnover rate measured in a catalytic flow reactor (5.6 × 10^-4 s^-1), indicating that the hydroperoxo species were true intermediates in the reaction. A proposed reaction sequence in which H₂O₂ forms on Au sites and propane is partially oxidized on Ti centers accounts for the spectroscopic results and the reaction orders obtained experimentally for the power-rate law expression.

Full text of DOI:10.1016/j.jcat.2008.04.004

Journal of Catalysis 257 (2008) 32–42
Contents lists available at ScienceDirect
Journal of Catalysis
www.elsevier.com/locate/jcat
Mechanistic study of propane selective oxidation with H2 and O2 on Au/TS-1
Juan J. Bravo-Suárez a,1, Kyoko K. Bando a, Tadahiro Fujitani a, S. Ted Oyama a,b,∗
a
Research Institute for Innovation in Sustainable Chemistry, National Institute of Advanced Industrial Science and Technology, AIST Tsukuba West, 16-1 Onogawa, Tsukuba,
Ibaraki 305-8569, Japan
b
Environmental Catalysis and Nanomaterials Laboratory, Department of Chemical Engineering (0211), Virginia Polytechnic Institute and State University, Blacksburg, VA 24061, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
The selective oxidation of propane to acetone and 2-propanol with H2 and O2 was studied on Au/TS-1 by
kinetic and spectroscopic analysis. A kinetic study using a factorial design at conditions where the catalyst
was stable and gave propane conversions of <6% and combined oxygenates selectivity of >90%, resulted
in a power-rate law expression of the form roxyg. = koxyg.(H2)0.74(O2)0.36(C3H8)0.29. In situ Au L3-edge
X-ray absorption near-edge spectroscopy (XANES) measurements showed activation of O2 on Au, whereas
in situ ultraviolet–visible (UV–vis) spectroscopy evidenced the presence of Ti-hydroperoxo species. The
role of the Ti-hydroperoxo species was probed by a transient technique in which changes in Ti K-edge
Received 7 February 2008
Revised 6 April 2008
Accepted 8 April 2008
Available online 20 May 2008
Keywords:
Gold
XANES spectra were used to determine the evolution of coverage with time (dθ/dt). It was shown that
TS-1
Propane
Acetone
2-Propanol
−4
the rate of reaction by XANES (6.7×10
−4
s
−1), indicating that the hydroperoxo species were true intermediates in the reaction.
s
−1) was close to the turnover rate measured in a catalytic flow
reactor (5.6 × 10
A proposed reaction sequence in which H2O2 forms on Au sites and propane is partially oxidized on Ti
centers accounts for the spectroscopic results and the reaction orders obtained experimentally for the
power-rate law expression.
Kinetics
Ultraviolet–visible spectroscopy
X-ray absorption near edge spectroscopy
True intermediate
© 2008 Elsevier Inc. All rights reserved.
Oxidation
1. Introduction
hydroxylations, alkane oxidations, and alkene epoxidations, us-
ing H2O2 as the oxidant [13–18]. Compared with alkene epox-
idation, the selective oxidation of low-cost abundant alkanes,
such as propane, with H2O2 or H2–O2 mixtures has not been
as widely studied. The catalytic oxidation of propane in an au-
toclave with H2O2 on TS-1 resulted in oxygenates (acetone and
2-propanol) with high selectivities at a space–time yield (STY) of
The discovery that nanometer-sized, highly dispersed supported
gold materials can serve as efficient catalysts for several oxidation
reactions has stimulated much work in the area [1–3]. In the liquid
phase, Au/C catalyzes the epoxidation of cis-cyclooctene [4], and
Au–Pd/TiO2 catalyzes the oxidation of alcohols to aldehydes [5].
In the gas phase, Au supported on transition metal oxides (e.g.,
TiO2, Fe2O3, Co2O3) has high activity for CO oxidation even at sub-
ambient temperatures [1,2,6,7], and Au supported on Ti-containing
oxides has high selectivity for propylene epoxidation with H2 and
O2 [8,9]. Recently, it has been demonstrated that Au/TiO2 [10]
and Au/TS-1 [11] are efficient catalysts for the partial oxidation of
propane with H2 and O2. On Au/TiO2, propane is efficiently con-
verted into propylene [10], whereas on Au/TS-1 catalysts, propane
is transformed with high selectivity (∼95%) to acetone and 2-
propanol [11].
−1
−1
about 80 g kgcat h
[14]. The selective oxidation of propane in a
flow system with H2–O2 mixtures on Au/TS-1 catalysts has been
shown to also produce acetone and 2-propanol at high selectivi-
ties (∼95%) with STYs (∼40 g kg−ca1t h−1) [11] comparable to those
obtained in a batch reactor with H2O2 [14]. Currently, acetone is
formed as a coproduct in the production of phenol by the cumene
process [19,20], with a worldwide production of about 5.2 million
metric tons in 1999 [21]. There is increasing interest in the direct
synthesis of acetone from propane because of the possibility of
commercialization of a process for the direct synthesis of phenol
from benzene [22–24].
In the liquid phase, propane oxidation with H2O2 results in
the formation of acetone and 2-propanol by attack on the sec-
ondary C–H bond of propane. When TS-1 is used as a catalyst,
propane activation has been proposed to occur via a 2-propoxy
intermediate [13]. In the gas phase, the activation of propane on
Au-supported Ti-containing catalysts with H2 and O2 also likely
occurs through the formation of a 2-propoxy intermediate species
from the reaction of adsorbed propane and H2O2 produced from
Titanium-substituted silicalite (TS-1) [12] has been shown to
catalyze a wide range of selective oxidations, including aromatic
Corresponding author at: Department of Chemical Engineering (0211), Virginia
Polytechnic Institute and State University, Blacksburg, VA 24061, USA. Fax: +1 540
*
231 5022.
E-mail address: oyama@vt.edu (S.T. Oyama).
Present address: Department of Chemical Engineering, University of California,
1
Berkeley, CA 94720, USA.
0021-9517/$ – see front matter © 2008 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2008.04.004

J.J. Bravo-Suárez et al. / Journal of Catalysis 257 (2008) 32–42
33
H2 and O2 on Au [11]. Based on our spectroscopic and kinetic data,
here we propose that the likely steps leading to the synthesis of
acetone from propane, H2, and O2 involve the following reactions
that are known to occur on gold or Ti sites:
vacuum-dried overnight at 298 K, and calcined in air at 673 K
(∼3 K min−1) for 3 h.
2.2. Catalytic testing
• Synthesis of H2O2 from H2 and O2 on Au nanoparticles [25–
27].
Propane partial oxidation experiments were carried out in a
quartz tubular microreactor (6 mm i.d., 180 mm long) using a
powder catalyst sample diluted with SiC (Strem) (particle size 63–
212 μm; 150 mg of Au/TS-1 and 450 mg of SiC). The empty space
before and after the catalyst sample was filled with glass wool
to prevent gas-phase reactions. Flow rates of H2 (purity ꢀ99.99%,
from a hydrogen generator, OPGU-2100S, Shimadzu), O2 (purity
ꢀ99.5%, Hitachi Sanso), C3H8 (purity ꢀ99.5%, Takachiho Chemical),
and Ar (purity ꢀ99.9997%, Suzuki Shokan) were regulated by mass
flow controllers. The reactor was equipped with an axial quartz
thermocouple well (2 mm o.d.), which allowed monitoring of the
catalyst bed temperature. Before reaction, the catalyst was pre-
treated under Ar and held at 443 K for 0.5 h. A typical standard re-
action condition used for spectroscopic studies was H2/O2/C3H8/Ar
= 1/1/1/7 at a total flow rate of 30 cm3 min−1, 443 K, and 0.1 MPa
(Table 1, run S). The space velocity at this standard condition var-
• Formation of Ti-hydroperoxo (Ti(OOH)) species from H2O2 on
tetrahedral Ti(OH) sites [28–34].
• Formation of 2-propoxy species from Ti(OOH) and adsorbed
C3H8 [10,11,13].
• Synthesis of acetone from oxidation of 2-propoxy species with
H2O2 [10,11].
• Decomposition of H2O2 to H2O [31,32,35].
In this work, the kinetics of propane selective oxidation with
H2 and O2 on Au/TS-1 (Ti/Si = 3/100) were also determined. Under
certain reaction conditions, acetone and 2-propanol were produced
−1
at STYs of about 170 and 15 g kgcat h−1, respectively. These results
are much higher than those obtained in a batch reactor with ex-
pensive H2O2 [14]. The partial pressure dependence of the rate of
formation of acetone on the reactants, H2, O2, and C3H8 were ob-
tained using a factorial design method, and the resulting kinetics
are in agreement with the proposed sequence of steps.
−1
−1
ied from 9000 to 90,000 cm3 h
g
as the weight of the catalyst
cat
used for UV–vis and XAFS measurements differed. Reaction prod-
ucts were analyzed online using two gas chromatographs (Shimazu
GC-14), one equipped with a flame ionization detector (FID) and a
thermal conductivity detector (TCD) using a FFAP capillary column
(0.32 mm × 60 m) and a Porapak Q packed column (3 mm × 2 m),
respectively, and the other equipped with an FID attached to a MS-
5A 60/80 packed column (3 mm × 2 m) and TCD attached to a
Gaskuropak 54 84/100 packed column (3 mm×2 m). The FFAP and
Porapak Q columns were used to detect oxygenates (e.g., acetalde-
hyde, propylene oxide, acetone, propionaldehyde, acrolein, acetic
acid, and 2-propanol) and CO2 and H2O, respectively, whereas the
Gaskuropak 54 84/100 and MS-5A 60/80 columns were used to de-
tect hydrocarbons (e.g., propane, propylene, ethylene, and ethane)
and H2, O2, CO, and methane, respectively.
The reaction also was studied in detail with spectroscopic tech-
niques at reaction conditions using a novel transient kinetic anal-
ysis. The adsorption of O2 on Au was demonstrated by in situ
UV–vis and Au L3-edge XANES spectroscopies, the formation of
Ti(OOH) species was investigated by in situ UV–vis spectroscopy,
and the Ti(OOH) were identified as true intermediates by a tran-
sient Ti K -edge XANES spectroscopic technique [34]. The spec-
troscopic results support the aforementioned sequence of steps.
The reaction sequence involves the migration of hydrogen peroxide
formed on gold sites to form a reactive hydroperoxide intermedi-
ate on Ti sites, and the subsequent reaction of the hydroperoxide
species with propane to form a 2-propoxide intermediate that pro-
duces oxygenated products [11].
Because the major products observed were acetone (CH3CO-
CH3), 2-propanol (CH3CH(OH)CH3), CO2, and H2O, the net reac-
tions occurring on the catalysts were taken as [10]:
2. Experimental
C3H8 + 2H2 + 2O2 → CH3COCH3 + 3H2O,
C3H8 + H2 + O2 → CH3CH(OH)CH3 + H2O,
C3H8 + 5O2 → 3CO2 + 4H2O,
2.1. Catalyst preparation
The TS-1 (Ti/Si = 3/100) zeolite support was prepared by the
method of Khomane et al. [36] In a typical preparation, 2.0 g of
Tween 20 (Aldrich) was dissolved in a solution of 24 g of water
(Ultrapure, Wako) and 27.3 g of tetrapropylammonium hydrox-
ide (20–25 wt% aqueous solution; TCI). Then 36.6 g of tetraethy-
lorthosilicate (+95%; Wako) was added dropwise under vigorous
stirring. After continuous stirring for 1 h, a clear solution was ob-
tained. A solution of about 1.8 g of tetrabutylorthotitanate (TCI)
dissolved in 18 g of 2-propanol (Wako, +99.5%) was added drop-
wise under vigorous stirring. After another 1 h of stirring, the
mixture was transferred to a Teflon-lined autoclave and placed in
an oven for hydrothermal crystallization at 433 K for 18 h. The fi-
nal product was recovered by centrifugation, washed thoroughly
with water (Ultrapure, Wako), dried overnight at 383 K, and cal-
cined in air at 813 K (∼2 K min−1) for 4 h.
and
H2 + 1/2O2 → H2O.
Based on these reactions, propane conversion, selectivity to CH3-
COCH3 and CH3CH(OH)CH3, and H2 efficiency (selectivity) were
defined as follows:
mol of (C3 products + CO2/3)
Propane conversion =
× 100,
× 100,
mol of propane in feed
mol of C3 product
C3 product selectivity =
mol of (C3 products + CO2/3)
hydrogen consumed by desired reactions
total hydrogen consumption
H2 efficiency =
× 100
The Au supported on TS-1 catalyst was prepared by a deposi-
tion–precipitation method. In a typical preparation, 100 ml of a
2 × mol of CH3COCH3 + mol of CH3CH(OH)CH3
=
× 100.
total hydrogen consumption
−3
gold solution (3 g of HAuCl4·4H2O in 1 L of water, 7.2 × 10
M)
was heated to 343 K under vigorous stirring. The pH of the solu-
tion was adjusted to 7.0 within 10 min by addition of 1 M NaOH
(Wako, 97.0+%), followed by addition of 1.0 g of TS-1. Stirring was
continued for 1 h, after which the suspension was cooled to room
temperature. Solids were separated by filtration and washed with
0.3 L of water (Ultrapure, Wako). The solid thus obtained was
The kinetic study was carried out at a temperature of 443 K and
a total pressure of 0.1 MPa by varying the flow rate of reactants
while keeping the total flow rate at 45 cm3 min
by adjusting the
−1
flow rate of Ar. The catalyst was stable, and measurements were
started after about 7 h of reaction at the initial reaction conditions.
A 33–1 fractional factorial design methodology was used to obtain

34
J.J. Bravo-Suárez et al. / Journal of Catalysis 257 (2008) 32–42
Table 1
Summary of catalytic results at different temperatures and with various combinations of reactants following a 33–1 fractional factorial design
Code
Partial pressure
(kPa)
C3H8
conv. (%)
Selectivity
(%)
H2 eff.
(%)
Formation rate
−1
−1
)
(mmol kgcat
h
H2
O2
C3H8
Acet.
85
2-Pr.
11
Other
0
CO2
4
Acet.
842
2-Pr.
111
CO2
36
H2O
S
10.1
10.1
10.1
2.0
18
10,800
T 4
T 1
T 3
R0
R1
12.7
12.7
12.7
12.7
12.7
12.7
12.7
12.7
12.7
12.7
12.7
12.7
12.7
12.7
12.7
1.8
1.0
1.9
1.6
1.7
85
77
85
83
85
7
17
10
13
12
1
3
1
0
0
7
3
4
4
3
10
13
13
14
14
1410
681
1450
1120
1320
116
149
177
192
194
106
29
71
50
46
35,800
10,300
26,200
17,300
21,300
L7
L5
L3
L2
L9
R2
L8
L1
L6
L4
21.8
12.7
3.5
3.5
12.7
21.8
12.7
21.8
12.7
12.7
3.5
21.8
21.8
21.8
12.7
12.7
12.7
3.5
3.5
3.5
12.7
0.9
1.2
0.5
0.6
3.5
1.9
5.2
1.0
4.6
1.1
77
84
75
80
90
87
92
93
92
83
19
15
15
13
7
11
4
6
0
0
5
1
0
0
0
0
0
0
4
1
5
6
3
2
4
1
4
4
9
16
16
13
15
15
7
8
9
10
1090
1590
577
273
290
114
73
238
187
49
18
44
135
56
24
41
33
87
34
59
2
26,500
22,600
8070
3.5
440
8000
21.8
12.7
21.8
3.5
12.7
12.7
2900
1530
1260
254
1120
826
46,500
23,400
52,500
7460
32,100
17,700
21.8
3.5
4
13
48
39
R3
12.7
12.7
12.7
1.7
83
12
1
4
13
1280
179
67
21,800
Notes. Conv.: conversion, Eff.: efficiency, Acet.: acetone, 2-Pr.: 2-propanol. The symbol
S indicates standard conditions; L represents Taguchi orthogonal array elements
with numeric labels;
R
denotes repeats; T indicates measurements at different temperatures, T 1 = 428 K, T 3 = 458 K, T 4 = 473 K, and the remaining runs at 443 K.
−1
Catalyst = 0.52 wt% Au/TS-1 (Ti/Si = 3/100), catalyst weight = 150 mg, total flow rate = 45 cm3 min
(for standard conditions is 30 cm3 min−1), and total pressure =
0.1 MPa.
Table 2
Dependencies of formation rates on partial pressure of reactants
Formation
rate
ka
Dependency
R2
95% confidence
H2 (l)
O2 (m)
C3H8 (n)
Acetone (p)
H2 (l)
O2 (m)
C3H8 (n)
Acetone (p)
Oxygenates
Acetone
42.6
41.2
0.74
0.76
0.36
0.38
0.29
0.21
0.974
0.963
0.14
0.17
0.14
0.17
0.14
0.17
CO2
0.358
0.82
3.06
0.66
1.71
0.33
0.48
0.81
0.37
0.738
0.853
0.647
0.69
2.98
0.69
1.50
0.89
0.69
0.74
0.82
−8
5.21 × 10
−2.76
0.87
3.61
0.98
66.1
H2O
1540
2330
2150
0.87
0.88
0.89
0.22
0.22
0.23
0.951
0.989
0.989
0.18
0.095
0.54
0.18
0.095
0.27
−0.18
−0.18
0.095
0.14
−0.014
0.66
Units will vary so that the formation rate is given in mmol kg−ca1t h−1. Catalyst = 0.52 wt% Au/TS-1 (Ti/Si = 3/100), catalyst weight = 150 mg, total flow rate =
a
45 cm3 min−1, H2, O2, and C3H8 partial pressure range = 3.5–21.8 kPa, temperature = 443 K, and total pressure = 0.1 MPa.
rate parameters over a fairly large range of reactant partial pres-
sures [37]. Three levels of partial pressure—3.5, 12.7, and 21.8 kPa—
were used for each reactant, H2, O2, and C3H8. The partial pressure
combinations were set such that they followed a Taguchi orthogo-
nal array L9 (L1 to L9 in Table 1). Measurements at the central
level (H2/O2/C3H8 = 12.7/12.7/12.7 kPa) were also carried out
at other temperatures (T 1 = 428 K, R = 443 K, T 3 = 458 K, and
T 4 = 473 K). Each measurement was taken after stable rates were
achieved, which took about 1 h. The order of the experiments
was varied randomly, and the experiments were carried out se-
quentially as shown in the first column of Table 1. The highest
temperature point was measured first, to minimize possible cata-
lyst changes. The first entry (run S) of Table 1 is typical of in situ
spectroscopic experiments, entries 2–5 are used for the estimation
of activation energies, and entries 7–16 are used to fit the studied
models. Despite the limited number of experimental data points
(10), sufficient points are available to obtain relevant results as the
number of degrees of freedom, data points (N) minus model pa-
rameters (p), is positive and >5 [32]. To verify the lack of catalyst
deactivation, runs at the central level at 443 K (R0) were repeated
during and after the experiments (R1, R2, and R3). Power-rate law
expressions (Table 2) were obtained by simultaneously fitting the
logarithms of the partial pressures (kPa) of each reactant and the
formation rate (mmol kg−ca1t h−1) data by linear least squares regres-
sion analysis using the POLYMATH 5.1 program [38]. Internal mass
transfer limitations were neglected, because the Weisz–Prater pa-
rameter (CWP) was estimated to be 5 × 10−4 < 1 [39].
2.3. Characterization
Transmission electron microscopy (TEM) images were obtained
in a microscope (UT-Philips CM200) operated at 200 kV. In situ
UV–vis spectra were collected under reaction conditions using
a large-compartment spectrometer (Varian Cary 5000) equipped
with a Harrick Scientific reaction chamber (model HVC-DRP) and
a praying mantis diffuse reflectance attachment (DRP-XXX). Pow-
der catalyst (∼50 mg) was loaded into the reaction chamber, the
temperature of which was monitored by a thermocouple added to
the Harrick cell just under the center of the sample. Reaction con-
ditions (e.g., temperature, pressure, gas flow rates) were the same
as for the reactivity measurements (Table 1, run S). Spectra were
referenced to the same material under Ar just before the start of
the reaction (difference spectra) or to BaSO4. Diffuse reflectance
spectra were analyzed using the Kubelka–Munk function, F(R ),
α
calculated from absorbance data [40]. In situ Au L3-edge X-ray ab-
sorption fine structure (XAFS) measurements under propane partial
oxidation conditions (H2/O2/C3H8/He = 1/1/1/7, 443 K, 0.1 MPa)
were carried out at beamline BL9A of the Photon Factory in the
Institute of Materials Structure Science, High-Energy Accelerator

J.J. Bravo-Suárez et al. / Journal of Catalysis 257 (2008) 32–42
35
Fig. 1. TEM micrograph and particle size distribution of Au/TS-1 catalyst.
Research Organization (PF-IMSS-KEK) in Tsukuba, Japan. In situ Ti
K -edge XAFS measurements were also carried out at two different
reaction conditions, one in the presence of H2 and O2 (H2/O2/He
= 1/1/8, 443 K, 0.1 MPa) and the other in the presence of H2, O2,
and C3H8, (H2/O2/C3H8/He = 1/1/1/7, 443 K, 0.1 MPa). Ti K -edge
XAFS spectra were obtained every 150 s in a step-scanning mode
using the same pretreatment and reaction conditions as used in
the catalytic testing reactor (Table 1, run S), but with He as the
inert gas. All spectra were obtained in transmission mode using
an in situ XAFS cell provided with a flow delivery system. Sam-
ple wafers (diameter = 1 cm) were prepared from 200 mg of
catalyst for Au L3-edge measurements and 20 mg of catalyst for
Ti-edge XAFS measurements. Pretreatment and reaction conditions
were monitored and controlled from outside the radiation shield
hutch. Gold content in the catalyst was estimated from the Au L3-
edge XAFS edge-jump absorption intensity by comparison with a
gold-supported TS-1 sample of known Au concentration. XAFS data
were analyzed with commercially available software (REX, Rigaku
Co.).
Fig. 2. Propane conversion and C3-oxygenates selectivity over Au/TS-1 cat-
alyst with reaction time on stream. Reaction conditions: H2/O2/C3H8/Ar
−1
= 5.6/5.6/5.6/26.8 cm3 min . Space velocity = 18,000 cm3
h
g
−1, T = 443 K,
−1
cat
P = 0.1 MPa.
3. Results
total pressure of 0.1 MPa. The data were obtained at propane con-
versions <6% and C3-oxygenates selectivity >90%. Propane conver-
sion increased with H2 and O2 partial pressures but appeared to
decrease with propane partial pressure. The C3-oxygenates com-
bined selectivity remained high (>90%) during the experimental
study. Acetone was the main product, with selectivities >75%,
whereas 2-propanol selectivities reached values as high as ca. 20%.
Here the rates are expressed in mmol kg−ca1t h−1, but they can be
The catalyst used in this study comprised nanometer-sized gold
particles highly dispersed on a microporous titanosilicate, TS-1
(Ti/Si = 3/100), prepared by deposition-precipitation at a pH of 7
using NaOH as the neutralizing agent. Au L3-edge XAFS measure-
ments of the Au/TS-1 catalyst indicated that it had a gold loading
of 0.52 wt%, and a Au coordination number (CN) of 10.4, corre-
sponding to a particle size (Dp) of 3.4 nm assuming Au fcc packing
and spherical particles. The error introduced by estimating the
Au content of the catalyst (0.52 wt%) from the Au L3-edge XAFS
edge-jump absorption intensity would be expected to be small
in comparison with a reference sample, a gold supported TS-1 of
similar composition and gold content (0.11 wt%). The nominal Ti
content for this catalyst with a Ti/Si ratio of 3/100 is estimated as
converted to turnover frequencies (TOF; s−1) by the conversion
−1
factor (1/S) (1/3600), where S is the surface gold (10.6 μmol gcat
)
or titania (477 μmol g−ca1t ) content. The formation rate of oxygenates
(acetone and 2-propanol) did not track with propane conversion,
because different inlet propane partial pressures were used. In gen-
eral, oxygenates formation rate increased with H2, O2, and C3H8
partial pressures (see Supporting Information). Hydrogen efficiency
(selectivity to oxygenates) varied depending on the reaction con-
ditions, but appeared to increase at low H2, high O2, and high
C3H8 partial pressures (runs L1, L2, and L3). Under reaction condi-
tions typical of in situ spectroscopy experiments (Table 1, run S),
at steady state, propane conversion was 2% with a selectivity to
oxygenates (acetone and 2-propanol) of 96%, resulting in an STY of
56 g kg−ca1t h−1. The steady-state oxygenates turnover rate based on
−1
477 μmol gcat . Fig. 1 presents a high-resolution TEM image of the
Au/TS-1 catalyst and its particle size distribution, which was cen-
tered at 3.5 nm. Gold particles with a diameter of 3.5 nm exhibited
a dispersion of about 40%, resulting in an exposed gold content of
−1
10.6 μmol gcat
.
Table 1 gives the factorial design used for the kinetic study with
the Au/TS-1 catalyst, comprising 10 runs with various combina-
tions of reactant partial pressures at a temperature of 443 K and a

36
J.J. Bravo-Suárez et al. / Journal of Catalysis 257 (2008) 32–42
surface gold atoms was 2.5 × 10−2 s−1, and that based on total Ti
by the absence of a near-edge resonance around 11,920 eV similar
to that observed in a HAuCl4 reference sample. Fig. 8 presents the
in situ Au L3-edge XANES difference spectra μ(Reaction) – μ(He, at
443 K before reaction), clearly showing that a transient resonance
centered at about 11,921 eV developed in the initial stages of the
reaction. It initially grew during the first 0.1 h of reaction, but then
decreased, reaching a relative stable value after 1.0–1.5 h of reac-
tion. The feature can be related to the formation of Au–O species
by interaction of gold with adsorbed oxygen [44,45], whereas the
decrease indicates that the Au–O complex was reduced under re-
action conditions.
−1
was 5.6 × 10−4 s
.
Fig. 2 shows the reactivity over the Au/TS-1 catalyst as a func-
tion of reaction time. Propane conversion and oxygenates (acetone
and 2-propanol) selectivity remained relatively stable for up to 7 h.
The catalyst also remained stable during the course of the kinetic
study, as verified by repetition of the central level run at the be-
ginning, during, and at the end of the experiments (Table 1, runs
R0, R1, R2, and R3).
Table 2 presents the dependence of the formation rates of C3-
oxygenates, acetone, CO2, and H2O on the nominal inlet partial
pressures of H2, O2, and C3H8 as a power rate law expression
r = k(H2)l(O2)m(C3H8)n. Water was found to have little effect on
the formation rates when introduced in the feed as a vapor (2.2%),
and thus it was not considered in the model fits. The best fits for
the C3-oxygenates, acetone, and H2O were
Fig. 9 shows the in situ Ti K -edge XANES spectra for the
Au/TS-1 catalyst under reaction conditions (H2/O2/C3H8/He =
1/1/1/7) as a function of time. The spectra present a Ti pre-
edge peak centered at 4969.0 eV that decreased with reaction
time. This pre-edge feature is characteristic of fourfold-coordinated
Ti sites [46,47]. The fractional decrease in the Ti pre-edge peak
area from its value before the introduction of the reactants is de-
fined as the coverage, θ, on the Ti centers. The decrease in the
pre-edge area is a direct result of an increase in the coordina-
tion of Ti. Fig. 10 shows the increased coverage on Ti during the
first 450 s of reaction with H2/O2/He (1/1/8) and H2/O2/C3H8/He
(1/1/1/7). The initial rate of increase of coverage was faster with
H2/O2/He than with H2/O2/C3H8/He. The slopes at time zero for
roxyg. = koxyg.(H2)0.74(O2)0.36(C3H8)0.29
,
racet. = kacet.(H2)0.76(O2)0.38(C3H8)0.21
,
and
−0.18
rwater = kwater(H2)0.89(O2)0.23(C3H8)
.
The overall models fits were relatively good, with R2 values
>0.960. The oxygenates formation rate depended mainly on H2
partial pressure, and also on O2 and C3H8 partial pressures.
Fig. 3 presents power-rate law model parity plots and residuals
of the formation rates of H2O, CO2, and oxygenates. The parity plot
lines pass through the origin and show no systematic deviations.
The residual plots, expressed as a percentage of the experimen-
tal rate, show that the errors in the data are randomly distributed.
Fig. 4 shows Arrhenius plots for the formation rates of H2O, ace-
−4
the H2/O2/He and H2/O2/C3H8/He transient curves were 7.4×10
−5
and 7.3 × 10
s
−1, respectively.
4. Discussion
The catalyst used in this work, Au/TS-1, consisted of gold nano-
particles deposited on a microporous titanosilicate (TS-1; Ti/Si =
3/100) support [12,17,41,48]. The gold content in the catalyst of
0.52 wt% corresponded to gold nanoparticles with an average par-
ticle size of 3.5 nm as estimated from TEM measurements (Fig. 1).
This result is in excellent agreement with the particle size of
3.4 nm estimated from Au L3-edge XAFS measurements. The ca.
tone, and CO2, which yielded corresponding activation energies of
−1
42, 38, and 40 kJ mol
.
The in situ UV–vis spectra of Au/TS-1 under reaction condi-
tions as a function of time exhibited bands characteristic of the
gold nanoparticles and the TS-1 support (see Supporting Informa-
tion). A broad band at around 520 nm (2.38 eV) was due to a
plasmon resonance of the gold nanoparticles, whereas the bands
at 240 nm (5.17 eV) and 350 nm (3.54 eV) can be assigned to
tripodal Ti-hydroperoxo species (Ti(OOH)) or tripodal hydroxo Ti
(Ti(OH)(OSi)3), and hydrated Ti-hydroperoxo species (Ti(OOH)), re-
spectively [11,41]. Fig. 5 shows the gold plasmon resonance (PR)
position changes during in situ UV–vis measurements under re-
action conditions on the Au/TS-1 catalyst as a function of time.
The PR position showed an initial increase (red shift) during the
first 0.5 h, followed by a decrease (blue shift), reaching a constant
value of 520.3 nm after 2–3 h of reaction. The initial red shift
is assigned to adsorption of O2 on the gold nanoparticles (Au–O
species) [42,43], whereas the blue shift is explained by the partial
reduction of these Au–O species, likely with H2 present in the reac-
tant gas. The PR position under reductive H2 flow (H2/Ar, 10 vol%)
measured after reaction was 516.0 nm.
40% dispersion for this size Au particle corresponds to an exposed
−1
Au content of 10.6 μmol gcat
.
Propane selective oxidation with H2 and O2 on the Au/TS-1
catalyst was stable (Fig. 2), enabling the kinetic measurements
described in this study (Table 1). Under the standard reaction
conditions typical of the in situ spectroscopic measurements (Ta-
ble 1, run S), propane conversion was 2% and oxygenates (acetone
and 2-propanol) selectivity was 96%. From these values, the oxy-
−4 −1
based on total Ti
genates TOF was calculated as 5.6 × 10
s
−2 −1
based on surface gold atoms. The TOF based
and 2.5 × 10
s
on Ti was lower than on Au, because the catalyst contained about
50 times more Ti atoms (477 μmol g−ca1t ) than surface gold atoms
−1
(10.6 μmol gcat ). It has recently been shown that TOF also can be
measured independently by appropriate in situ spectroscopic tech-
niques for reacting species by transient measurements of coverage
versus time (dθ/dt) [34]. If the TOF of the reacting species mea-
sured spectroscopically is of similar magnitude to the TOF mea-
sured in the conventional reactor, then the reacting species can be
considered a true intermediate [34,49–51]. In the present study,
we demonstrated this property for the hydroperoxide species on
the Ti sites through in situ XAFS spectroscopy measurements, as
discussed later in more detail.
Fig. 6 shows the in situ difference spectra for the Au/TS-1
catalyst under reaction conditions as
a function of time. As
can be seen, three main bands evolved with reaction time:
a band at 237 nm (5.23 eV) due to tripodal Ti hydroperoxo sites
(Ti(OOH)(OSi)3); a shoulder at around 272 nm (4.56 eV), assigned
to hydrated tripodal Ti tetrahedral sites (Ti(OH)(H2O)2(OSi)3); and
a broad band centered at around 355 nm (3.49 eV) assigned to
hydrated Ti-hydroperoxo species (Ti(OOH)(H2O)2) [11,33,41].
Fig. 7 presents the in situ Au L3-edge XANES spectra, normal-
ized to the same edge jump, for Au/TS-1 under reaction condi-
tions. The catalyst spectra show three near-edge features at 11,935,
11,949, and 11,970 eV, which are characteristic of metallic gold. No
cationic gold was observed during reaction conditions, as indicated
The maximum rate of oxygenates (acetone and 2-propanol)
formation on the Au/TS-1 catalyst was 3140 mmol kg−ca1t h−1, cor-
−1
−1
(Table 1, run L9). This value is
responding to 183 g kgcat
h
much higher than the values previously reported in the liter-
ature for similar catalysts for propane oxidation with reduced
activated O2 or H2O2; for instance, a Au/TiO2/SiO2 catalyst us-
ing H2 and O2 oxidized propane to produce acetone (∼50%) at
−1
−1
[8]; a H2–O2 cell system provided with
a STY of < 1 g kgcat
h

J.J. Bravo-Suárez et al. / Journal of Catalysis 257 (2008) 32–42
37
Fig. 3. Parity plots and residuals of the calculated and experimental oxygenates, CO2, and H2O formation rates.
Fig. 5. In situ Au plasmon resonance position for Au/TS-1 catalyst under reaction
conditions as a function of time.
Fig. 4. Arrhenius plots of formation rates of CO2, acetone, and H2O.
conversion was <6%, the selectivity to C3-oxygenates was >90%,
and the H2 efficiency varied between about 5 and 20%. It should
be noted that these results are similar to those for propylene epox-
idation to propylene oxide (PO) on gold-supported catalysts at
low temperature (<473 K) which presented propylene conversions
as high as 8%, PO selectivities >90%, and H2 efficiencies <40%
[28–32].
a Pd + VO(acac)2/VGCF cathode oxidized propane to form ace-
−1
−1
[52];
tone (39%) and acetaldehyde (13%) at a STY of 6 g kgcat
h
a three-phase reactor using a catalytic hollow fiber membrane
(HF15/Nafion/PEEK-WC) and the Fenton system, Fe2+–H2O2, car-
ried out the propane oxidation to oxygenates (96%; acetone,
2-propanol, 1-propanol, and propionic aldehyde) at
a
STY of
−1
−1
5 g kgmembrane
h
[53]; and a batch reactor using TS-1 and expen-
From Table 1, it can be seen that the rate of C3-oxygenates
formation was increased by simultaneously increasing the partial
pressures of H2, O2, and C3H8 (run L1, H2/O2/C3H8 = 3.5/3.5/3.5,
and run R, H2/O2/C3H8 = 12.7/12.7/12.7). The partial pressures of
sive H2O2 produced oxygenates (∼100%; acetone and 2-propanol)
−1
−1
[14]. Table 1 also summarizes the reac-
at a STY of 79 g kgcat
h
tion rate results for different partial pressures of reactants used to
determine the kinetics of propane oxidation on Au/TS-1. Propane

38
J.J. Bravo-Suárez et al. / Journal of Catalysis 257 (2008) 32–42
Fig. 8. In situ Au L3-edge XANES difference spectra, μ(Reaction) − μ(He) at 443 K
before reaction, for Au/TS-1 under propane selective oxidation conditions as a func-
tion of reaction time.
Fig. 6. In situ UV–vis difference spectra for Au/TS-1 catalyst under reaction con-
ditions as a function of time. Spectra referenced to catalyst before reaction in Ar
(difference spectra).
Fig. 9. In situ Ti K -edge XANES spectroscopy results for Au/TS-1 under propane
Fig. 7. In situ Au L3-edge XANES for Au/TS-1 under propane selective oxidation con-
selective oxidation conditions as a function of reaction time.
ditions as a function of reaction time.
rates exhibited R2 values >0.96, with the fitting for CO2 forma-
tion rates not as good. This was because of the relatively large
variance of the experimental error (Fig. 3), probably due to the
low sensitivity in the analytical system at the rather low CO2 se-
lectivities (<6%) obtained in this study. Consequently, we do not
discuss CO2 formation models any further in this paper. Fig. 3
presents parity plots and residuals between measured and calcu-
lated values. An apparent increasing trend for the residuals at the
highest level of oxygenates formation can be explained by the lim-
ited number of experimental points under such conditions. For
the rate of oxygenates formation, the dependencies on H2, O2,
and C3H8 were 0.74, 0.36, and 0.29, respectively (Table 2). Simi-
lar dependencies were obtained for acetone formation (0.76, 0.38,
and 0.21, respectively), the main oxygenate product. These posi-
tive fractional dependencies orders indicate that H2, O2, and C3H8
were adsorbed on the surface of the catalyst, but not necessar-
ily on the same sites [32,57,58]. Additionally, they indicate that
the adsorbed species should be involved in steps that determine
the formation rate [57,58]. We provide a more detailed discussion
H2 and O2 had particularly strong effects on C3-oxygenates forma-
−1
−1
tion; for example, the lowest oxygenates STY (272 mmol kgcat h
)
occurred at low H2 and O2 partial pressures (run L1, H2/O2/C3H8
= 3.5/3.5/3.5), whereas the largest STY (3140 mmol kg−ca1t h−1) was
obtained at high H2 and O2 partial pressures (run L9, H2/O2/C3H8
= 12.7/21.8/21.8); see Supporting Information.
These results suggest that both H2 and O2 may be involved in
the formation of a reaction intermediate, such as a hydroperoxide
species [33,54]. This species has been detected on a Au/TiO2 cat-
alyst by inelastic neutron scattering (INS) experiments (at 20 K)
after reaction with flowing H2/O2 at 423 K [55]; on a Pd/TS-1 cat-
alyst by means of EPR and UV–vis spectroscopy after contact with
H2, followed by O2 at 323–343 K [56]; on a Au/(mesoporous)Ti–
SiO2 by EPR after gas-phase reaction with H2 and O2 at 423 K; by
in situ UV–vis [33]; and on a Au–Ba/Ti-TUD by in situ UV–vis and
XAFS spectroscopy under reaction conditions with H2 and O2 [34].
Table 2 gives the dependencies of formation rates on the par-
tial pressure of reactants as calculated from regression analysis.
Power-rate law fitting for oxygenates, acetone, and H2O formation

J.J. Bravo-Suárez et al. / Journal of Catalysis 257 (2008) 32–42
39
The reaction rate for propylene oxide formation appears to be
determined by two irreversible steps: the production of H2O2 on
a Au site and the oxidation of the propylene by a hydroperox-
ide species on a Ti site [31,32,34,35,60]. In the case of propane
oxidation, a similar set of steps is likely, with propane proba-
bly reacting with a hydroperoxide species to form an 2-propoxy
intermediate [11,13]. This species can readily desorb to form 2-
propanol or undergo further oxidation with H2O2 to produce ace-
tone [11,13,62].
The likely sequence of steps for propane oxidation can be de-
duced from a combination of in situ spectroscopic techniques and
kinetic measurements. In the case of H2 oxidation and propylene
epoxidation with H2 and O2 on gold catalysts, significant advances
have been realized through in situ UV–vis and XAFS spectro-
scopic techniques [30,33,34] augmented by density functional the-
ory (DFT) calculations [35]; for example, for propylene epoxidation,
Ti-hydroperoxo species recently was found to be a true intermedi-
ate by means of transient in situ XANES spectroscopy [34]. In the
case of propane oxidation, recent evidence for adsorption and ac-
tivation of O2 on Au has been provided in an Au/TiO2 catalyst by
in situ UV–vis and XANES spectroscopy [10] as well as in an Au–
Ba/TS-1 catalyst by in situ UV–vis spectroscopy [11]. In situ UV–vis
spectroscopy also has provided evidence for the formation of Ti-
hydroperoxo and tripodal Ti(OH) sites in a Au–Ba/TS-1 catalyst [11].
The present study provides additional evidence in support of the
presence of adsorbed O2 on Au and the formation of Ti(OOH) and
Ti(OH) sites, and proof that Ti(OOH) species is a true reaction in-
termediate.
In situ UV–vis spectra of Au/TS-1 as a function of reaction
time during propane oxidation conditions demonstrated notice-
able changes in three main regions: (1) a band at about 520 nm
(2.38 eV) due to a plasmon resonance (PR) of Au nanoparticles;
(2) a band increase at 350 nm (3.54 eV), assigned to hydrated
Ti(OOH) species; and (3) a band increase at 240 nm (5.17 eV), as-
signed to tripodal Ti(OH) sites (see Supporting Information). The
Au PR feature was symmetric, indicating that the gold particles had
a regular shape and a relatively narrow size distribution, consistent
with the TEM results (Fig. 1) [63,64]. Tracking of the Au PR posi-
tion as a function of reaction time provided additional information
regarding the adsorptive properties of the Au nanoparticles (Fig. 5).
The Au PR position shifted toward higher wavelengths (red shift)
initially (in the first 0.5 h), but then moved toward lower wave-
lengths (blue shift) thereafter, reaching an almost constant value
after about 2 h of reaction. The initial red shift of the Au PR was
due to O2 adsorption on Au and can be explained by the lower
density of free electrons in the gold as a result of charge transfer
from the surface metal atoms to adsorbed O2 [42,43]. The subse-
quent blue shift possibly occurred because the adsorbed O2 on Au
reacted with H2 as steady conditions were reached. Some adsorbed
O2 remained, as suggested by the higher Au PR position (520.3 nm)
compared with that observed for the catalyst under reductive H2
flow (516.0 nm), which served as the source of the steady produc-
tion of the hydroperoxide species. The in situ UV–vis difference
spectra in Fig. 6 show the simultaneous formation of hydrated
Ti(OOH)(H2O)2 (355 nm) and tripodal Ti(OOH)(OSi)3 (237 nm) sites
more clearly. A shoulder at around 272 nm is likely related to hy-
drated tripodal Ti(OH)(OSi)3 [41]. The TS-1 (Ti/Si = 3/100) support
has isolated tetrahedral Ti(OSi)4 sites incorporated in the frame-
work of a crystalline silica zeolite (structure type MFI) [12,17,41,
48]. These tetrahedral Ti(OSi)4 sites can be readily transformed by
hydrolysis into tripodal Ti(OH)(OSi)3 sites, as supported by theo-
retical calculations [65]. The participation of this Ti(OH)(OSi)3 site
in hydrocarbon oxidation with H2O2 also has been suggested re-
cently based on theoretical calculations [66,67]. The small increase
in the Ti(OOH) signal over 5 h corresponds to the small increase
in conversion over the same time period (Fig. 2). This is likely due
Fig. 10. Coverage on 4-fold coordinated Ti (θ, fractional decrease of pre-edge
peak area in the Ti K -edge XANES spectra) for Au/TS-1 under (a) H2/O2/C3H8/He
= 1/1/1/7 (bottom) and (b) H2/O2/He = 1/1/8 (top) at 443 K and 0.1 MPa during
the first 450 s of reaction.
of a possible sequence of steps during propane oxidation, includ-
ing H2, O2, and C3H8 species later in this section. For the rate of
H2O formation, the respective dependencies on H2, O2, and C3H8
were 0.88, 0.22, and −0.18 (Table 2). The positive fractional ex-
ponents indicate that H2O was formed mainly by reaction of H2
and O2, whereas the negative fractional exponent suggests that
propane was detrimental for this reaction, probably because it cap-
tures some of the H2O2 that otherwise would decompose to H2O.
The exponent values of H2 and O2 (0.88 and 0.22, respectively) are
consistent with those reported previously for H2 oxidation on a
0.15 wt% Au on TS-1 (1 wt% Ti) catalyst (0.76 and 0.17, respec-
tively) [35]. The dependency of H2O production on the acetone
partial pressure was found to be −0.014. This small value indicates
a quite small contribution of propane partial oxidation to water
formation in comparison with the oxidation of H2.
The Arrhenius plots for the formation of H2O, acetone, and CO2
−1
produce apparent activation energies of 42, 38, and 40 kJ mol
,
respectively (Fig. 4). The apparent activation energy for H2O for-
mation (42 kJ mol−1) is in good agreement with the value of
−1
37 kJ mol
reported for H2 oxidation on a 0.15 wt% Au on TS-1
−1
(1 wt% Ti) [35] and with the values of 38–42 kJ mol
for H2 ox-
idation on Au/TiO2 [59]. The apparent activation energy for C3H8
−1
partial oxidation to acetone of 38 kJ mol
values of 43–48 kJ mol
agrees well with the
−1
for C3H8 total combustion on Au/TiO2
catalysts [59]. Similar values have been obtained for propylene
epoxidation with H2 and O2 on Au–Ba/Ti-TUD (43 kJ mol−1) [31],
Au/TS-1 (35 kJ mol−1) [60], and Au/TiO2 (47 kJ mol−1) [61]. The
apparent activation energy for CO2 (40 kJ mol−1), although smaller
than that determined during propylene epoxidation on Au–Ba/Ti-
TUD (80 kJ mol−1) [31], falls within the energy range for CO oxida-
−1
tion on Au/TiO2 catalysts of 25–72 kJ mol
[59].
Further discussion on the kinetics of propane selective oxida-
tion with H2 and O2 requires knowledge of the reaction mech-
anism. Although little is known about this reaction on Au cat-
alysts, the similarities in the reaction orders of propane partial
oxidation using H2 and O2 (roxyg. = koxyg.(H2)0.74(O2)0.36(C3H8)0.29
)
with those for H2 oxidation (rH = kH O(H2)0.76(O2)0.17
) [35]
2O
2
and propylene epoxidation with H2 and O2 on Au/TS-1 (rPO
=
kPO(H2)0.60(O2)0.31(C3H8)0.18) [60] suggest a possibly similar se-
quence of steps. Recently, reaction pathways for H2 oxidation and
propylene epoxidation to propylene oxide using H2 and O2 with
Au supported on Ti-containing catalysts have been proposed [31,
34,35,60].

40
J.J. Bravo-Suárez et al. / Journal of Catalysis 257 (2008) 32–42
to a secondary effect involving the Ti centers, which are somewhat
distant from the Au particles. The Au particles first supply H2O2
to neighboring Ti centers, which are responsible for the first im-
mediate reaction products, but as supply and demand for H2O2
becomes balanced for those sites, the more distant centers come
into play.
In the absence of separate experiments to determine absolute
coverages, proving that these observed species are true interme-
diates and not spectators is not possible with UV–vis spectroscopy
alone. Later in the paper, we use in situ Ti K -edge XANES measure-
ments to demonstrate that Ti(OOH) is a true intermediate species
during propane oxidation.
The net rate of oxidation was calculated as the difference between
the two slopes,
roxid = (dθ/dt)synthesis − (dθ/dt)oxid
.
The method described here assumes that measurements of the
changes in the pre-edge peak intensity of the Ti K -edge region
are indicative of adsorption. This assumption is reasonable, because
the pre-edge peak is characteristic of the tetrahedral geometry on
the Ti sites before adsorption, and changes in coordination result in
extinction of the peak intensity [47]. It also assumes that the step
involving the reaction of the Ti hydroperoxide species is a slow
step in the reaction, which is consistent with the kinetic results, as
we discuss later.
Gold particles were found in the vicinity of Ti, although not
necessarily bonded directly to the tetrahedral Ti sites [29]. The in
situ Au L3-edge XANES results for Au/TS-1 showed that the Au
in the catalyst was mostly metallic (Fig. 7), and that the pres-
ence of oxidized Au, characterized by a near-edge resonance of
ca. 11,920 eV (e.g., HAuCl4), was not observed. In situ Au L3-edge
XANES difference spectra, μ(Reaction) −μ(He, at 443 K before
reaction) (Fig. 8), showed the evolution of a near-edge transient
resonance (white line) at about 11,921 eV during the first 0.1 h
of reaction, which later decreased to reach a near-stable value af-
ter 1.0–1.5 h of reaction. The intensity of this white line in the
Au L3-edge arises from unoccupied d-states; therefore, interactions
between adsorbed molecules and gold clusters should have an ef-
fect on this resonance [44,45]. The initial increase in intensity of
the white line can be related to the chemisorption of O2 on Au
particles (Au–O complex formation) as the d-orbital electron count
of gold is decreased by charge transfer to the 2π∗ orbital of O2
[44,45]. The subsequent decrease in intensity of the white line as
the reaction reaches a near-steady value corresponds to a reduc-
tion of the Au–O complex (in the air-calcined sample) by reaction
with H2. This behavior is similar to that observed for propane
oxidative dehydrogenation with H2 and O2 on Au/TiO2 catalysts,
which showed an initial increase of the Au L3-edge white line, fol-
lowed by a decrease due to reduction of the Au–O complex with
H2 [10]. Similar trends were also found for a 4 wt% Au/Al2O3 cata-
lyst subsequently exposed to O2 and CO [45]. The Au L3-edge white
line exhibited an increase in intensity during O2 exposure, fol-
lowed by a decrease due to reaction of the formed Au–O complex
with CO to produce CO2. Overall, the in situ Au L3-edge XANES dif-
ference spectra results are in line with Au PR measurements under
reaction conditions, which also indicates initial adsorption of O2
on Au and its further reaction with H2 to likely form hydroperox-
ide species.
The pre-edge peak in the Ti K -edge XANES spectra for Au/TS-1
centered at 4969.0 eV is a salient feature; see Fig. 9. This pre-edge
peak arises from the normally symmetry forbidden 1s→3d tran-
sition because of mixing of sp-states in tetrahedrally coordinated
Ti (Ti(Td)), and it provides a wealth of information regarding the
Ti oxidation state and local geometry [46,47,68,69]. Under propane
oxidation conditions with the standard H2/O2/C3H8/He = 1/1/1/7
mixture, a simultaneous decrease in the Ti pre-edge normalized
absorption and increase in the intensity of the feature at 4985.5 eV
indicates the formation of higher-coordinated Ti species as a re-
sult of adsorption. Similar results were obtained for the in situ Ti
K -edge XANES measurements with the H2/O2/He = 1/1/8 mixture
and are not shown for the sake of brevity. The fractional decrease
in the Ti pre-edge area from its value before the introduction of re-
actants is defined as the coverage on the Ti(Td) centers, θ. Fig. 10
shows the coverage on fourfold-coordinated Ti for Au/TS-1 under
H2/O2/C3H8/He = 1/1/1/7 and H2/O2/He = 1/1/8 mixtures. Fol-
lowing the method described earlier, the initial net rate of propane
−4 −1
oxidation based on Ti coverage, θ, is estimated as 6.7 × 10
s
(Fig. 10). This value is similar to the TOF for propane oxidation
−4 −1
based on total Ti
to oxygenates at steady-state of 5.6 × 10
s
(Table 1, run S), indicating that the Ti-hydroperoxide species is
a true intermediate in the gas-phase oxidation of propane with
H2 and O2. Comparing the initial transient rate by XANES to the
steady-state rate is justified because the Ti sites are isolated, and
thus interactions are not expected.
A possible simplified sequence of steps for propane oxidation
with H2 and O2 on Au/TS-1 consistent with the kinetic results, the
in situ spectroscopic study, and previously reported observations is
as follows:
∗
∗−→
O + ◦
O
(1)
←−
2
2
−→
H + 2 • ◦ 2H •
(2)
We used a previously reported transient technique to identify
reaction intermediates [34]. The technique involves measuring the
initial transient coverage changes (dθ/dt) on fourfold-coordinated
Ti as determined from intensity changes in the Ti pre-edge peak
centered at about 4968.9 eV in the in situ Ti K -edge XANES spec-
tra using two different reacting environments: a H2/O2/He = 1/1/8
gas mixture, which corresponds to hydroperoxide species synthe-
sis conditions, and a H2/O2/C3H8/He = 1/1/1/7 gas mixture, which
corresponds to propane oxidation conditions. The net reaction of
the hydroperoxide species includes its formation (rfor) and decom-
position rates (rdecomp):
←−
2
∗
∗
−→
O + H • ◦ HOO + •
(3)
←−
2
∗
HOO + H2 + • −∧→ H2O2 + ∗ + H •
(4)
−→
H O + Ti(OH) ◦ Ti(OH)(HOOH)
(5)
(6)
←−
2
2
−→
Ti(OH)(HOOH) ◦ Ti(OH )(OOH)
←−
2
−→
Ti(OH2)(OOH) + CH3CH2CH3
◦
Ti(OH )(OOH)(HCH(CH ) ) (7)
←−
2
3 2
Ti(OH2)(OOH)(HCH(CH3)2) −∧→ Ti(OH2)(OCH(CH3)2) + H2O
Ti(OH2)(OCH(CH3)2)
(8)
(9)
−→
←−
Ti(OH) + CH CH(OH)CH
3
3
(dθ/dt)synthesis = rfor − rdecomp
,
Ti(OH2)(OCH(CH3)2) + H2O2 −→ Ti(OH) + 2H2O + CH3COCH3 (10)
whereas the net reaction of propane oxidation includes not only
hydroperoxide species formation (rfor) and decomposition rates
(rdecomp), but also the reaction rate to form oxygenates from
propane (roxid),
The first part of this reaction sequence (steps (1)–(4)) involves
the formation of H2O2 on two types of Au sites (∗ and •), as
suggested by previous studies of H2 oxidation and propylene epox-
idation with H2 and O2 [31,32,35,60]. The second part of the
reaction sequence (steps (5)–(8)) involves the formation of Ti-
hydroperoxide species from tetrahedral Ti sites and H2O2 [66,67],
(dθ/dt)oxid = rfor − rdecomp − roxid
.

J.J. Bravo-Suárez et al. / Journal of Catalysis 257 (2008) 32–42
41
which then react with propane to produce a 2-propoxy interme-
5. Conclusion
diate species [11,13]. This 2-propoxy species can then form 2-
propanol (step (9)) or acetone (step (10)) [13,62]. Assuming that
steps (4) (formation of H2O2) and (8) (formation of 2-propoxy
species) are slow steps, that H2O2 can decompose to H2O via a
parallel first-order pathway, and that the rate of formation of the
2-propoxy species (step (8)) mostly determines the rate of forma-
tion of oxygenates, we can derive a power-rate law expression of
the form roxyg. = koxyg.(H2)k(O2)l(C3H8)m based on the foregoing
sequence of steps. As shown elsewhere, the derivation of a rate
expression for a similar sequence of steps in propylene epoxida-
tion with H2 and O2 on a Au–Ba/Ti-TUD catalyst [31,32] allows us
to solve the following set of equations simultaneously to provide a
final expression:
This work demonstrated the efficient production of oxygenates
(acetone and 2-propanol) from propane, H2, and O2 on a 0.52 wt%
Au/TS-1 (Ti/Si = 3/100) catalyst at 443 K and 0.1 MPa. The com-
bined selectivity for oxygenates was always >90%, with ace-
tone (75–93%) as the main product, followed by 2-propanol (4–
19%) and then CO2 (1–6%). We studied the reaction using ki-
netic and spectroscopic techniques. The rate of formation of oxy-
genates is well described by a power-rate model of the form
roxyg. = koxyg.(H2)0.74 (O2)0.36(C3H8)0.29. Apparent activation ener-
gies for the formation of H2O, acetone, and CO2 were 42, 38, and
40 kJ mol−1, respectively. In situ Au L3-edge XANES and UV–vis
spectroscopy under reaction conditions demonstrated the forma-
tion of O2 adsorbed on Au, which may react with H2 to form
hydroperoxide species. Ti-hydroperoxo species were detected by
in situ UV–vis spectroscopy and were demonstrated to be true
intermediates by a transient Ti K -edge XANES technique. A pos-
sible sequence of steps has been proposed: (1) H2O2 is formed
from H2 and O2 on gold; (2) H2O2 is transferred to Ti centers
to form Ti-hydroperoxo species; (3) Ti-hydroperoxo species re-
act with propane to form 2-propoxy intermediate species; and
(4) 2-propoxy species form acetone and 2-propanol. The proposed
reaction pathway produced a rate of formation of oxygenates con-
sistent with the experimental power-rate law kinetic results.
• From steps (1) to (4),
α(O2)(H2)3/2
rH
2O2
=
,
(11)
[1 + β(O2) + γ (O2)(H2)1/2][1 + δ(H2)1/2
]
where α, β, γ , and δ are constant quantities resulting from
combinations of the total number of Au adsorption sites for
H2 (•) and O2 (∗) and reaction rate coefficients of steps (1) to
(4).
• From steps (5) to (8),
κ(H2O2)(C3H8)
roxyg.
=
,
(12)
Acknowledgments
[1 + λ(H2O2) + μ(H2O2)(C3H8)]
where κ, λ, and μ are also constant quantities resulting from
combinations of the total number of Ti(OH) adsorption sites
and reaction rate coefficients of steps (5) to (8).
• Finally, accounting for a first-order decomposition of H2O2,
The authors acknowledge financial support from the Ministry of
Economy, Trade and Industry (Minimum Energy Chemistry project)
and the National Science Foundation (CBET 0651238). Financial
support also was provided by the Japan Society for the Pro-
motion of Science through the postdoctoral fellowship for for-
eign researcher program (P05627, to J.J.B.-S.) and the invited fel-
low program (to S.T.O.). The authors thank Ms. E. Kobayashi and
Dr. K. Kawasaki for the TEM measurements. The XAFS experi-
ments were conducted under the approval of PF-PAC (proposals
2004G304 and 2006G362).
rH
2O2
= roxyg. + rdecomp = roxyg. + kD (H2O2).
(13)
The foregoing set of equations can be further simplified by
assuming that in the denominator of (12), the term λ(H2O2) +
μ(H2O2)(C3H8) may be <1 if H2O2 concentrations are relatively
low, resulting in
Supplementary material
roxyg. ≈ κ(H2O2)(C3H8).
(14)
Solving (11), (13), and (14) simultaneously yields
The online version of this article contains additional supple-
mentary material.
α(O2)(H2)(H2)1/2
Please visit DOI: 10.1016/j.jcat.2008.04.004.
roxyg.
=
[1 + β(O2) + γ (O2)(H2)1/2][1 + δ(H2)1/2
]
(C3H8)
References
×
.
(15)
[kD /κ + (C3H8)]
[1] M. Haruta, Catal. Today 36 (1997) 153.
[2] M. Haruta, Cattech 6 (2002) 102.
The left-side term of (15) is similar to the rate of formation of
H2O from H2 and O2 on Au catalysts reported earlier [35]. As-
suming that δ(H2)1/2 > 1, it will cancel the (H2)1/2 in the nu-
[3] A.S.K. Hashmi, G.J. Hutchings, Angew. Chem. Int. Ed. 45 (2006) 7896.
[4] M.D. Hughes, Y.-J. Xu, P. Jenkins, P. McMorn, P. Landon, D.I. Enache, A.F. Carley,
G.A. Attard, G.J. Hutchings, F. King, E.H Stitt, P. Johnston, K. Griffin, C.J. Kiely,
Nature 437 (2005) 1132.
[5] D.I. Enache, J.K. Edwards, P. Landon, B. Solsona-Espriu, A.F. Carley, A.A. Herzing,
M. Watanabe, C.J. Kiely, D.W. Knight, G.J. Hutchings, Science 311 (2006) 362.
[6] M. Haruta, N. Yamada, T. Kobayashi, S. Iijima, J. Catal. 115 (1989) 301.
[7] G.R. Bamwenda, S. Tsubota, T. Nakamura, M. Haruta, Catal. Lett. 44 (1997) 83.
[8] T. Hayashi, K. Tanaka, M. Haruta, Prep.-Am. Chem. Soc., Div. Pet. Chem. 41
(1996) 71.
[9] T. Hayashi, K. Tanaka, M. Haruta, J. Catal. 178 (1998) 566.
[10] J.J. Bravo-Suárez, K.K. Bando, J.Q. Lu, T. Fujitani, S.T. Oyama, J. Catal. 255 (2008)
114.
[11] J.J. Bravo-Suárez, K.K. Bando, T. Fujitani, S.T. Oyama, Chem. Commun. (2008), in
press.
merator. The final result is an expression of the form roxyg.
=
koxyg.(H2)k(O2)l(C3H8)m, where the order for H2 is between 0.5
and 1 and the orders for O2 and C3H8 are between 0 and 1, which
correspond well with the respective power-rate fitting results of
0.74, 0.36, and 0.29. Overall, the suggested reaction pathway is in
close agreement with these kinetic results and the spectroscopic
observations of activation of oxygen on gold, as well as the forma-
tion of reactive hydroperoxide species on Ti, and provides a good
description of the reaction of propane oxidation on Au/TS-1. Addi-
tional support for this model comes from the fact that it provides
a better fit to the reaction data than the generic power-rate law
equation (see Supporting Information). Our kinetic and spectro-
scopic results can be used to provide support for a mechanism
involving the reaction of hydroperoxide species with propane to
form oxygenates.
[12] M. Taramasso, G. Perego, B. Notari (Enichem), US Patent 4,410,501 (1983).
[13] C.B. Khouw, C.B. Dartt, J.A. Labinger, M.E. Davis, J. Catal. 149 (1994) 195.
[14] M.G. Clerici, Appl. Catal. 68 (1991) 249.
[15] M.G. Clerici, P. Ingallina, Catal. Today 41 (1998) 351.
[16] D.R.C. Huybrechts, L. De Bruycker, P.A. Jacobs, Nature 345 (1990) 240.
[17] B. Notari, Adv. Catal. 41 (1996) 253.

42
J.J. Bravo-Suárez et al. / Journal of Catalysis 257 (2008) 32–42
[18] R. Murugavel, H.W. Roesky, Angew. Chem. Int. Ed. 36 (1997) 477.
[19] W.L. Howard, Acetone, Kirk Othmer Encyclopedia of Chemical Technology, Wi-
ley & Sons, 2002.
[20] R.J. Schmidt, Appl. Catal., A 280 (2005) 89.
[21] Z.-C. He, Xiandai Huagong/Mod. Chem. Ind. 25 (11) (2005) 61.
[22] G.I. Panov, Cattech 4 (2000) 18.
[43] W. Cai, H. Hofmeister, T. Rainer, W. Chen, J. Nanopart. Res. 3 (2001) 443.
[44] N. Weiher, A.M. Beesley, N. Tsapatsaris, L. Delannoy, C. Louis, J.A. Van
Bokhoven, S.L.M. Schroeder, J. Am. Chem. Soc. 129 (2007) 2240.
[45] J.A. Van Bokhoven, C. Louis, J.T. Miller, M. Tromp, O.V. Safonova, P. Glatzel,
Angew. Chem. Int. Ed. 45 (2005) 4651.
[46] M. Fernández-García, Catal. Rev. 44 (2002) 59.
[47] F. Farges, G.E. Brown Jr., J.J. Rehr, Phys. Rev. B 56 (1997) 1809.
[48] G.N. Vayssilov, Catal. Rev.—Sci. Eng. 39 (1997) 209.
[49] K. Tamaru, Dynamic Heterogeneous Catalysis, Academic Press, New York, 1978.
[50] C. Reed, Y. Xi, S.T. Oyama, J. Catal. 235 (2005) 378.
[23] P.P. Notté, Top. Catal. 13 (2000) 387.
[24] New Technology, Benzene-to-phenol process ready for commercialization, Fo-
cus Catal. (2004) 6.
[25] P. Paredes Olivera, E.M. Patrito, H. Sellers, Surf. Sci. 313 (1994) 25.
[26] M. Okumura, Y. Kitagawa, K. Yamaguchi, T. Akita, S. Tsubota, M. Haruta, Chem.
Lett. 32 (2003) 822.
[27] A. Corma, M. Boronat, S. González, F. Illas, Chem. Commun. (2007) 3371.
[28] N. Yap, R.P. Andres, W.N. Delgass, J. Catal. 226 (2004) 156.
[29] B. Taylor, J. Lauterbach, W.N. Delgass, Appl. Catal., A 291 (2005) 188.
[30] B. Chowdhury, J.J. Bravo-Suárez, M. Daté, S. Tsubota, M. Haruta, Angew. Chem.
Int. Ed. 45 (2006) 412.
[51] S.T. Oyama, W. Li, Top. Catal. 8 (1999) 75.
[52] I. Yamanaka, S. Hasegawa, K. Otsuka, Appl. Catal., A 226 (2002) 305.
[53] C. Espro, F. Arena, F. Tasselli, A. Regina, E. Drioli, A. Parmeliana, Catal. Today 118
(2006) 253.
[54] E.E. Stangland, K.B. Stavens, R.P. Andres, W.N. Delgass, J. Catal. 191 (2000) 332.
[55] C. Sivadinarayana, T.V. Choudhary, L.L. Daemen, J. Eckert, D.W. Goodman, J. Am.
Chem. Soc. 126 (2004) 38.
[56] V.N. Shetti, P. Manikandan, D. Srinivas, P. Ratnasamy, J. Catal. 216 (2003) 461.
[57] M.A. Vannice, Kinetics of Catalytic Reactions, Springer, New York, 2005.
[58] G. Djéga-Mariadassou, M. Boudart, J. Catal. 216 (2003) 89.
[59] K. Ruth, M. Hayes, R. Burch, S. Tsubota, M. Haruta, Appl. Catal., B 24 (2000)
L133.
[60] B. Taylor, J. Lauterbach, G.E. Blau, W.N. Delgass, J. Catal. 242 (2006) 142.
[61] T.A. Nijhuis, B.M. Weckhuysen, Catal. Today 117 (2006) 84.
[62] J.E. Rekoske, M.A. Barteau, J. Catal. 165 (1997) 57.
[31] J.Q. Lu, X. Zhang, J.J. Bravo-Suárez, S. Tsubota, J. Gaudet, S.T. Oyama, Catal. To-
day 123 (2007) 189.
[32] J.J. Bravo-Suárez, J.Q. Lu, C.G. Dallos, T. Fujitani, S.T. Oyama, J. Phys. Chem. C 111
(2007) 17427.
[33] B. Chowdhury, J.J. Bravo-Suárez, N. Mimura, J.Q. Lu, K.K. Bando, S. Tsubota, M.
Haruta, J. Phys. Chem. B 110 (2006) 22995.
[34] J.J. Bravo-Suárez, K.K. Bando, J.Q. Lu, M. Haruta, T. Fujitani, S.T. Oyama, J. Phys.
Chem. C 112 (2008) 1115.
[63] U. Kreibig, M. Vollmer, Optical Properties of Metal Clusters, Springer, Berlin,
1995.
[64] L.M. Liz-Marzán, Langmuir 22 (2006) 32.
[35] D.G. Barton, S.G. Podkolzin, J. Phys. Chem. B 109 (2005) 2262.
[36] R.B. Khomane, B.D. Kulkarni, A. Paraskar, S.R. Sainkar, Mater. Chem. Phys. 76
(2002) 99.
[65] P.E. Sinclair, G. Sankar, C.R.A. Catlow, J.M. Thomas, T. Maschmeyer, J. Phys.
Chem. B 101 (1997) 4232.
[37] NIST/SEMATECH e-Handbook of Statistical Methods, http://www.itl.nist.gov/
div898/handbook/ (accessed Feb. 06, 2007), Section 5.3.3.10.
[38] M. Shacham, M.B. Cutlip, M. Elly, Polymath, Copyright 2006.
[39] H.S. Fogler, Elements of Chemical Reaction Engineering, third ed., Prentice–Hall
PTR, Upper Saddle River, NJ, 2006.
[40] G. Kortüm, Reflectance Spectroscopy, Springer, Berlin, 1969.
[41] P. Ratnasamy, D. Srinivas, H. Knözinger, Adv. Catal. 48 (2004) 1.
[42] W. Cai, H. Hofmeister, T. Rainer, Physica E 11 (2001) 339.
[66] R.R. Sever, T.W. Root, J. Phys. Chem. B 107 (2003) 4080.
[67] D.H. Wells Jr., W.N. Delgass, K.T. Thomson, J. Am. Chem. Soc. 126 (2004) 2956.
[68] F.M.F. de Groot, A. Knop-Gericke, T. Ressler, J.A. van Bokhoven, in: B.M. Weck-
huysen (Ed.), In-situ Spectroscopy of Catalysts, American Scientific Publishers,
Stevenson Ranch, CA, 2004, p. 107.
[69] J.M. Thomas, G. Sankar, Acc. Chem. Res. 34 (2001) 571.