Improved performance of TiO2 in the selective photo-catalytic oxidation of cyclohexane by increasing the rate of desorption through surface silylation

Article abstract of DOI:10.1016/j.jcat.2010.05.006

The effect of silylation on the performance of an anatase TiO₂ catalyst in the selective photo-oxidation of cyclohexane was investigated using attenuated total reflection Fourier transform infrared (ATR-FTIR) spectroscopy and an illuminated slu

Full text of DOI:10.1016/j.jcat.2010.05.006

Journal of Catalysis 273 (2010) 116–124
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Improved performance of TiO in the selective photo-catalytic oxidation of
2
cyclohexane by increasing the rate of desorption through surface silylation
a
a
a
b,*
Ana Rita Almeida , Joana T. Carneiro , Jacob A. Moulijn , Guido Mul
a
Catalysis Engineering, Delft University of Technology, Julianalaan 136, 2628 BL Delft, The Netherlands
PhotoCatalytic Synthesis Group, IMPACT Institute, Faculty of Science and Technology, University of Twente, 7500 AE Enschede, The Netherlands
b
a r t i c l e i n f o
a b s t r a c t
Article history:
The effect of silylation on the performance of an anatase TiO2 catalyst in the selective photo-oxidation of
cyclohexane was investigated using attenuated total reflection Fourier transform infrared (ATR-FTIR)
spectroscopy and an illuminated slurry reactor. The rate of cyclohexanone formation showed a depen-
dency on the availability of surface active OH sites and on the desorption rate of cyclohexanone. Two clas-
ses of catalysts could be identified: (1) containing less than 1.0 wt.% Si where the cyclohexanone
formation rate is decreased by silylation due to a decrease in OH availability and (2) containing more than
Received 24 February 2010
Revised 12 May 2010
Accepted 13 May 2010
Available online 19 June 2010
Keywords:
1.0 wt.% Si where the improved desorption rate becomes dominant over the decreasing OH availability,
TiO
2
and the cyclohexanone formation rate observed in an illuminated slurry reactor is increased. ATR-FTIR
results confirmed the linear increase in the rate of cyclohexanone desorption as a function of increasing
Deactivation
Silylation
ATR-FTIR
2
Silane content of the TiO surface. Because of this enhanced desorption, silylation also resulted in a
Cyclohexane
Cyclohexanone
Photo-catalysis
decrease in rate of formation of surface deactivating carbonate and carboxylate species on TiO .
2
Ó 2010 Published by Elsevier Inc.
1
. Introduction
commonly applied for Ti-substituted mesoporous silica catalysts
to reduce desorption limitations of polar products in oxidation
reactions [14–16]. In these cases, silanol groups from the mesopor-
ous support are exchanged with methylated silane groups, like
Photo-catalysis is widely applied in the decomposition of con-
taminants from water and air [1]. Among various semi-conductor
materials, TiO is the most investigated catalyst and used in com-
mercial water purification [2]. The use of TiO to selectively syn-
thesize organic compounds at mild conditions is also of interest
3]. However, low conversions and strong deactivation of TiO cat-
alysts prevent these photo-catalytic processes from being commer-
cially applied [4–8]. The reason for TiO deactivation is the
2
2
trimethylsilanes. But also silylation of commercial TiO has been
2
applied, and in general, a decrease in surface hydrophilicity and
affinity to polar molecules has been observed [17,18]. It will be
demonstrated in the present paper by data obtained from in situ
[
2
2
ATR-FTIR studies that the hydrophilicity of a commercial TiO cat-
2
alyst is indeed decreased by coupling trimethylsilane groups to its
surface, stimulating desorption of cyclohexanone formed by
photo-catalytic cyclohexane oxidation and thus reducing the rate
of catalyst deactivation.
formation of surface-bound carbonates and carboxylates, as ob-
served in ATR-FTIR studies. The formation of these species in the
selective cyclohexane oxidation was proposed to be the result of
slow desorption of cyclohexanone, which leads to over-oxidation
[
8]. Transient studies confirmed that cyclohexanone desorption is
slow, and requires in the dark almost 30 min to desorb from a
pre-illuminated TiO catalyst into cyclohexane [9]. In order to de-
2
. Experimental
2
2.1. Catalyst synthesis
crease the rate of deactivation in cyclohexane oxidation, desorp-
tion of cyclohexanone should be stimulated. Several strategies
can be followed to achieve this, such as the use of solvents [10–
The photo-catalyst used in this study was Hombikat UV100 TiO
2
(
Sachtleben) of 100% anatase crystallinity (determined by XRD), a
1
2], the use of elevated temperatures [13] or catalyst modification.
In this research, the last option has been evaluated, in particular
the modification of the TiO surface by silylation. This method is
2
SBET of 337 m /g and a primary particle size of approximately
5
nm [7]. Hexamethyldisilazane (HMDS) from Fluka was chosen
2
as the silylating agent, because of its high reactivity [19,20]. The
silylation method applied is similar to the one described by Fadeev
et al. [21]. The catalyst was dried overnight at 120 °C, and 0.5 g of
*
Corresponding author. Fax: +31 53 4892882.
E-mail addresses: G.Mul@utwente.nl, g.mul@tudelft.nl (G. Mul).
2
TiO was mixed with different volumes of HMDS in 10 mL of
0
021-9517/$ - see front matter Ó 2010 Published by Elsevier Inc.
doi:10.1016/j.jcat.2010.05.006

A.R. Almeida et al. / Journal of Catalysis 273 (2010) 116–124
117
toluene (99.5% from Sigma Aldrich), in a magnetically stirred round
bottom flask with a reflux cooler. The synthesis was carried out at
room temperature for 24 h, after which the silylated TiO materials
2
were filtered, washed with toluene and dried overnight at 120 °C.
Four silylation batches were made with 0.02, 0.06, 0.08 and
ment. UV-induced oxidation of cyclohexane was continued for
90 min, taking a spectrum every minute. The photo-catalytic reac-
tion was initiated by an assembly of 7 UV LEDs with 375-nm wave-
ꢀ
9
length emission with an incident photon flux of 9 ꢁ 10 Einstein
ꢀ2 ꢀ1
cm
s
at the surface of the catalyst coating. The first minute of
0
.15 g of HMDS.
collection was performed in the dark. The background and the
sample spectra were averaged from 64 and 32 spectra, respec-
ꢀ1
tively. A mirror velocity of 0.6329 cm s
and a resolution of
2.2. Catalyst characterization
ꢀ
1
4
cm were used for all measurements. All the spectra were cor-
rected by subtracting the features of the rotational modes of water
vapor.
X-ray fluorescence (XRF) measurements were done on a Pana-
lytical PW 1480 spectrometer, and the silicon content (wt.%) in
the silylated TiO catalysts was calculated with the so-called
Superq routine. The silylated TiO catalysts were characterized
2
2.4. Cyclohexane photo-catalytic oxidation – slurry reactor
2
by diffuse reflectance infrared spectroscopy (DRIFTS) on a Bruker
IFS 66 spectrometer. The catalyst powders were introduced into
a DRIFTS cell (Spectratech) and dried for 30 min at 120 °C. All
absorbance and single-beam spectra were collected with 128 aver-
Photo-catalytic tests were performed in a slurry reactor illumi-
nated from the top, which will be referred to as top illumination
reactor and is described elsewhere [22]. The pyrex reactor con-
tained 100 mL of cyclohexane and 1 g/L of catalyst in all measure-
ments. Air, pre-saturated with cyclohexane was bubbled through
the liquid at a flow rate of 30 mL/min, assuring constant saturation
ꢀ
1
aged scans and 4 cm
spectral resolution against a KBr back-
ground. Thermogravimetric analysis (TGA) was performed on a
Mettler Toledo TGA/SDTA851e. The samples were heated from 25
to 600 °C at a heating rate of 10 K/min under 10 mL/min He flow.
2
of cyclohexane with O (0.012 mol/L). Illumination was provided
2
9
Solid-state Si magic angle spinning nuclear magnetic resonance
by a 50-W high pressure mercury lamp (HBO50W from ZEISS),
2
9
ꢀ7
ꢀ2 ꢀ1
(
Si MAS NMR) was performed on a Bruker Avance 400 spectrom-
with incident photon flux of 2.10 ꢁ 10 Einstein cm
s
in all
eter, using a 5-mm zirconium rotor, with spinning speed of 11 kHz.
The spectral window was 25,063 Hz, with 50 ms acquisition time
and 15 s acquisition delay.
measurements. The reaction was carried out for 100 min, and li-
quid GC samples were analyzed by a flame ionization detector
(Chromopack CPwax 52CB).
2.3. Cyclohexane photo-catalytic oxidation – ATR-FTIR
3. Results
The catalysts were dried at 120 °C for 1 h in static air and sus-
3.1. Characterization
pended in toluene at a concentration of 3.67 g/L. The suspensions
were treated for 30 min in a 35-kHz Elmasonic ultrasonic bath;
The applied variation of the silylation method yielded a series of
2
mL of this suspension was spread on a ZnSe crystal and dried
2
materials with different silane loading on TiO : TiSi0.1, TiSi1.0,
in vacuum overnight.
TiSi1.6 and TiSi2.1, which contain 0.1, 1.0, 1.6 and 2.1 wt.% Si, as
determined by XRF analysis. Bare Hombikat, of 100% anatase crys-
tallinity, is referred to as TiSi0. The TGA analysis of TiSi0, TiSi1.0,
TiSi1.6 and TiSi2.1 recorded in He is shown in Fig. 1. The negative
contributions in this figure correspond to the weight loss divided
by initial mass of catalyst and are displayed as a function of
increasing temperature. At temperatures below 120 °C, weakly ad-
sorbed water is removed from the catalyst surface [15]. More
strongly adsorbed water is removed at temperatures between
120 and 300 °C [23], where particle sintering also starts to occur.
In the range of 150–400 °C, an extra contribution related to decom-
position of HMDS residue may contribute to the weight loss [19].
Above 400 °C, a negative feature is observed for the silylated
The in situ ATR-FTIR setup used for the photo-catalytic oxida-
tion of cyclohexane is described in detail elsewhere [8]. Briefly, a
volume of 50 mL of cyclohexane (99.0% from Sigma Aldrich) was
used for the photo-oxidation reactions. Cyclohexane was dried
over Molsieve (type 4A) overnight before use, to remove traces of
2
water, and saturated with O by bubbling dry air at 7.65 mL/min.
The oxygen-saturated cyclohexane was flown at 4 mL/min through
the ATR cell by means of a series II high performance liquid chro-
matography pump. Prior to the photo-catalytic oxidation experi-
ments, adsorption of cyclohexane on the TiO
monitored for 60 min. A spectrum of adsorbed cyclohexane on
TiO was collected as background for the photo-oxidation experi-
2
coating was
2
0.0
0.0
-
-
-
-
4
4
4
4
-4
-1.0x10
-2.0x10
-3.0x10
-4.0x10
-1.0x10
-4
-2.0x10
-4
-3.0x10
(
a)
(b)
-4
-4.0x10
0
100
200
300
400
500
0
100
200
300
400
500
T (ºC)
T (ºC)
Fig. 1. Thermogravimetric analysis (TGA) under He flow of (a) TiSi0 (black line), TiSi1.0 (gray line) and (b) TiSi1.6 (black line) and TiSi2.1 (gray line).

1
18
A.R. Almeida et al. / Journal of Catalysis 273 (2010) 116–124
11.69
hydroxylation nor polymerization of surface silane groups has oc-
curred to a large extent in the synthesis.
The DRIFTS analysis of the catalysts is shown in Fig. 3a and b.
Though different OH active sites have been identified on commer-
cial TiO
as acidic or basic OH. The surface OH density was determined by
the Fe(acac) adsorption method [24], which accounts for all types
of OH sites, of which an acidic contribution was measured by NH
2
surfaces, for simplicity, we refer to the surface active sites
3
3
–
TPD [7]. The remaining contribution will be referred to as basic OH
groups. An acidic/basic OH ratio of 3.3 has been reported for this
commercial TiO
2
[7], corresponding to ꢂ75% surface coverage by
acidic OH sites. Stretching vibrations of the acidic OH appear in
3
0
20
10
0
-10
all spectra, as a number of overlapping bands in the 3690–
ppm
ꢀ1
3
630 cm region [25] (Fig. 3a). A broad contribution of H
2
O is also
ꢀ1
Fig. 2. 29Si MAS NMR of the TiSi1.6 catalyst.
apparent in the 3600–3000 cm region. The basic OH group is H-
bridged with adsorbed water molecules, and its stretching vibra-
tion is expected to absorb around 3415 cm but is overlapped
by strong water contributions [25]. The peak area of the acidic
OH is represented in Fig. 3c as a function of silane content, and a
decreasing trend can be observed. The TiSi0 sample shows small
methyl vibrations below 3000 cm , characteristic of this commer-
cial TiO . A reference spectrum of liquid HMDS is shown in Fig. 3a
with specific CH
ꢀ1
catalysts, which corresponds to the decomposition of methyl
groups from the surface silanes [19].
Si MAS NMR was performed on the TiSi1.6 catalysts to under-
stand which type of Si bonding was induced by the synthesis meth-
od. In Fig. 2, a single peak is observed at 11.69 ppm that
2
9
ꢀ
1
2
ꢀ1
corresponds to (TiO)Si(CH
were absent, like those of Si–OH and Si–O–Si groups, so neither
3
)
3
[14,16,19]. Additional NMR features
3
stretching vibrations at 2954 and 2897 cm . The
silylated materials show similar IR bands but shifted to higher
2
972
1266
(
a)
(b)
2915
Acidic
TiO-H str.
TiSi2.1
TiSi1.6
TiSi2.1
TiSi1.0
TiSi0.1
2
958
TiSi1.6
TiSi1.0
2905
1
158
TiSi0
1253
1249
CH
3
str.
TiSi0.1
TiSi0
2954
CH
3
NH
bend
def.
2897
HMDS
1178
HMDS
3600
3400
3200
3000
1400
1300
1200
-
1
Wavenumbers (cm-1)
Wavenumbers (cm )
4
3
2
1
0
0
0
0
(
c)
0
0.0
0.5
1.0
1.5
2.0
Si (wt%)
Fig. 3. DRIFTS measurements recorded in He flow of TiSi0, TiSi0.1, TiSi1.0, TiSi1.6 and TiSi2.1 at RT after heating to 120 °C, including a reference spectrum of liquid HMDS,
ꢀ
1
ꢀ1
measured by ATR-FTIR in the (a) 3800–2800 cm wavenumbers region and (b) the 1450–1100 cm wavenumbers region. In (c), the peak area of the acidic TiO–H stretching
vibrations is represented as a function of silane content.

A.R. Almeida et al. / Journal of Catalysis 273 (2010) 116–124
119
wavenumbers, 2958 and 2905 cm 1. The shift is more intense for
TiSi2.1, in which the CH stretching vibrations of silane occur at
972 and 2915 cm . The typical Si–OH stretching vibration ex-
ꢀ
The relative absorption of bulk/adsorbed carbonyl bands depends
on the degree of silylation. In TiSi0, the bulk cyclohexanone band
intensity is much lower than the bands corresponding to adsorbed
cyclohexanone. Silylated samples show a similar bulk cyclohexa-
none band as observed for TiSi0, but lower contributions of ad-
sorbed cyclohexanone. The formation of other species is also
3
ꢀ
1
2
ꢀ1
pected at 3675 cm [26] is not visible in any of the silylated cat-
alysts, in agreement with the 29Si MAS NMR result.
The spectral features of the samples in the range of 1450–
100 cm are shown in Fig. 3b. TiSi0 does not show any particular
ꢀ1
ꢀ1
1
apparent in the region below 1600 cm , where a broad absorption
band in this region, besides the broad feature of strong Ti–O–Ti
stretching modes below 1000 cm . The silylated TiO catalysts
2
show additional IR bands, which are indicative of the presence of
surface silane. The reference spectrum of HMDS has specific IR
of at least three main bands is apparent. Different species can con-
tribute to these absorptions, including symmetric vibrations of car-
ꢀ1
ꢀ1
boxylates [31,32], carbonates (around 1589 cm
)
[33] and
[34]. These species adsorb
surface and are therefore referred to as deac-
tivating species. The CH bending vibration of the silane groups
ꢀ1
bicarbonates (around 1555 cm
strongly on the TiO
)
ꢀ
1
bands at 1249 and 1178 cm , which correspond to a CH
deformation mode [27] and to NH bending modes [28], respec-
tively. The silylated TiO materials show the CH group deforma-
tion mode [27] at 1253 cm and a number of overlapping bands
3
group
2
3
2
3
shows a negative absorption after the photo-catalytic reaction.
ꢀ
1
ꢀ1
The position of these negative bands is at 1247 cm for samples
ꢀ1
ꢀ1
at lower wavenumbers, of which the one occurring at 1158 cm
is the most pronounced. These bands occur in the NH bending re-
gion and are probably related to small amounts of adsorbed NH
of low silane content and at 1256 cm for TiSi2.1. These bands
are slightly red shifted in comparison with the ones observed in
Fig. 3.
3
The peak heights of the 1691, 1579 and 3630 cm 1 bands after
90 min of photo-catalytic reaction were determined for all cata-
lysts and are shown in Fig. 5, as a function of silane content. The
ꢀ
[
28,29]. TiSi2.1 shows only the CH
3
group deformation mode [27]
ꢀ1
at 1266 cm
and no clear NH contribution. Other important
3
ꢀ1
vibrations like Si–C rocking around 860 cm [27], Ti–O–Si stretch-
ing vibration at 910–960 cm or Si–OH at 980 cm [30] are not
visible due to the strong absorption modes of TiO
ꢀ1
ꢀ1
2
.
20
3
.2. Photo-catalytic cyclohexane oxidation
1
1
5
0
5
0
Fig. 4 shows the ATR-FTIR absorption spectra of the catalysts in
cyclohexane flow after 90 min of UV illumination. The acidic OH
sites on TiO are depleted during the photo-catalytic reaction [8],
as witnessed for TiSi0 by the negative vibration centered at
2
ꢀ
1
3
630 cm . The same feature is present in the spectra of the silylat-
ed samples, with the higher the silane content, the lower the ex-
tent of depletion. The center of the (negative) acidic OH peak
shifts from 3630 cm on TiSi0 to 3650 cm on silylated catalysts,
and the relative contribution of this shifted band increases with
silane content. In the infrared fingerprint region, below
ꢀ1
ꢀ1
0
0.5
1
1.5
2
ꢀ1
1
800 cm , the main products formed during reaction can be ob-
ꢀ
1
Si (wt% )
served. The band observed at 1710 cm corresponds to the C@O
stretching vibration of cyclohexanone, when this molecule is des-
orbed and dissolved in the reaction mixture. The bands observed
at 1691 and 1682 cm are also corresponding to the C@O stretch-
ing vibration of cyclohexanone but indicate that the molecule is
adsorbed on the catalyst surface through hydrogen bonding [8].
Fig. 5. ATR-FTIR bands after 90 min of cyclohexane photo-oxidation for TiSi0, 0.1,
.0, 1.6 and 2.1 divided by the mass of catalyst. The black symbols correspond to
deactivating species (1579 cm ), the empty symbols to adsorbed cyclohexanone
(1691 cm ) and the gray symbols to the positive values of acidic OH depletion
1
ꢀ
1
ꢀ1
ꢀ
1
ꢀ
1
(
3630 cm ).
C=O str.
OH
C=O str.
3
CH def.
TiO-H str.
1
691
682
1
1
710
3630
TiSi0
TiSi0.1
TiSi1.0
TiSi1.6
TiSi0
1
247
TiSi0.1
TiSi1.0
TiSi1.6
TiSi0
TiSi0.1
TiSi1.0
TiSi1.6
TiSi2.1
1256
1250
TiSi2.1
TiSi2.1
3
650
3
800 3700 3600 1800
1700
1600
1500
1200
-1
Wavenumbers (cm-1)
Wavenumbers (cm-1)
Wavenumbers (cm )
Fig. 4. ATR-FTIR spectrum after 90 min of cyclohexane photo-oxidation on TiSi0, TiSi0.1, TiSi1.0, TiSi1.6 and TiSi2.1.

1
20
A.R. Almeida et al. / Journal of Catalysis 273 (2010) 116–124
band height for adsorbed cyclohexanone decreases with silane
content. The peak height of the acidic OH depletion is represented
by positive values, showing a decrease with surface silane concen-
tration. Both adsorbed cyclohexanone and acidic OH depletion
show a similar trend. From the intensity of the band of deactivating
species represented by the absorption at 1579 cm , it can be con-
cluded that the amount of deactivating species formed decreases
only above 1.0% silane content.
stant of cyclohexanone, measured in 1/min, while H is the
ꢀ
1
1691 cm peak height in time, H
and Hend the peak height at t = 60 min, all determined in absor-
bance/gcatalyst. TiSi0 starts from a larger value of H but for all cases
0
the peak height at t = 0 min
0
a similar Hend is reached after 60 min of dark desorption. The kdes
values deduced from the exponential decay in all catalysts are rep-
resented in Fig. 6b as a function of silane content, and a linear rela-
tion was obtained.
ꢀ1
A similar result was obtained for water desorption after reac-
tion, shown in Fig. 7a. The amount of water desorbing from TiSi0
is much larger than for silylated materials, indicating a higher de-
gree of surface hydration. The exponential curves fitted through
the data provide an estimate for the desorption constant of water
from the surface, which is represented in Fig. 7b as a function of
silane content. As for cyclohexanone, it can be observed that kdes
for water correlates linearly with surface silylation.
3
.3. Cyclohexanone and water desorption under dark conditions
Cyclohexanone and water desorption were analyzed after reac-
tion in the ATR-FTIR setup, by continuing the cyclohexane flow un-
der dark conditions. Cyclohexanone desorption is characterized by
a decay in the intensity of the 1691 cm band in time, which is
shown in Fig. 6a, for the TiSi0, TiSi1.0 and TiSi2.1 catalysts.
Although carboxylates and water might have a small contribution
ꢀ1
ꢀ
1
to the peak height of the 1691 cm
band, exponential curves
3.4. Top illumination reactor
could be well fitted through the data, allowing estimation of the
desorption constants of cyclohexanone through the following
To compare the performance of the different catalysts in a
quantitative way, cyclohexane photo-oxidation tests were
equation: H = H
0
exp(kdest) + Hend, where kdes is the desorption con-
8
0.2
(
a)
(b)
6
4
2
0
0.1
0.0
0
10
20
30
40
50
60
0
0.5
1
1.5
2
tdesorption (min)
wt% Si
Fig. 6. (a) Cyclohexanone desorption (dark conditions) on TiSi0 (open symbols), TiSi1.0 (gray symbols) and TiSi2.1 (black symbols) measured by the 1691 cmꢀ1 band height
divided by catalyst mass (the lines correspond to exponential fits) and (b) cyclohexanone desorption constant (determined from the exponential curves) as a function of
silane content (wt.%).
tdesorption (min)
2
0
2
4
6
8
0.1
.05
0
(
a)
(b)
0
10
20
30
40
50
60
-
-
-
-
0
-
-
10
12
0
0.5
1
1.5
2
wt% Si
Fig. 7. (a) Water desorption (dark conditions) from TiSi0 (open symbols), TiSi1.0 (gray symbols) and TiSi2.1 (black symbols) measured by the 3350 cmꢀ1 band height divided
by catalyst mass (the lines correspond to exponential fits) and (b) water desorption constant (determined from the exponential curves) as a function of silane content (wt.%).

A.R. Almeida et al. / Journal of Catalysis 273 (2010) 116–124
121
0
0
0
0
.04
.03
.02
.01
0
0
0
0
.04
.03
.02
0.003
(a)
(b)
0
0
.002
.001
0.01
0
0.000
0
0.1
1
1.6
2.1
0
20
40
60
80
100
t(min)
Si ( wt%)
Fig. 8. GC analysis of the products of cyclohexane photo-oxidation performed in a top illumination reactor. (a) Formation of cyclohexanone (in mmol) in time for TiSi0 (open
symbols), TiSi1.0 (gray symbols) and TiSi2.1 (black symbols) and (b) cyclohexanone (black) and cyclohexanol (gray) in mmol, formed after 100 min of reaction, for all
catalysts.
performed in a top illumination reactor. Fig. 8a shows the photo-
catalytic formation of cyclohexanone in mmol, for TiSi0, TiSi1.0
and TiSi2.1. The initial rate of cyclohexanone formation increases
in the following order: TiSi0 > TiSi2.1 > TiSi1.0. At illumination
times longer than 40 min, the slope of the TiSi0 and TiSi1.0 curves
decreases due to deactivation, and the rate of cyclohexanone for-
mation for TiSi2.1 becomes the highest. Fig. 8b shows the GC re-
sults after 100 min of reaction for all the catalysts, for both
cyclohexanone and cyclohexanol (in mmol). The latter was not
clearly observed in the ATR-FTIR results, but in the slurry reactor
cyclohexanol is observed, though in very low amounts for all sam-
ples. Cyclohexanone formation decreases by addition of small
amounts of silane to the TiO surface, but over 1.0 wt.% Si the
2
amount of cyclohexanone formed increases again. For TiSi2.1, the
ketone formed after 100 min of photo-oxidation is similar to that
2
obtained for bare TiO . The selectivity toward cyclohexanone for-
mation is similar, ranging between 95% and 97%, suggesting that
the reaction mechanism is not affected by silylation.
by the 29Si MAS NMR (Fig. 2) spectra, in which isolated Ti–
O–Si(CH groups were identified for TiSi1.6. Since not all
3 3
)
2
9
catalysts were analyzed by Si MAS NMR, other types of silane
coupling for samples containing higher silane content cannot be
excluded. Silylation is confirmed by DRIFTS analysis (Fig. 3), where
the presence of methyl groups with wavenumbers corresponding
to those of HMDS on the surface is evident. Also secondary changes
in the surface properties are in agreement with surface silylation.
TGA analysis in Fig. 1 shows a lower amount of weakly and
2
strongly adsorbed water on the surface, indicating that TiO cata-
lysts become less hydrophilic after the silylation procedure. This
is confirmed by the decreasing intensity of absorptions assigned
to water vibrations in the DRIFT spectra (Fig. 3). Furthermore, from
the DRIFT spectra, it can be deduced that acidic OH groups are con-
verted, the amount changing linearly with silane content (Fig. 3c).
The analysis of basic OH sites is not possible under the applied con-
ditions, so the possibility of silylation of basic OH sites cannot be
fully ruled out. But knowing that (i) only ꢂ25% of the OH sites
existing on the surface are basic [7], (ii) the maximum coverage
with silylated species is 60% (for TiSi2.1) and (iii) a linear relation
with the amount of acidic sites exists, we conclude that mainly
acidic sites are silylated. It is therefore proposed that silanes bind
4
. Discussion
HMDS is expected to react with a TiO
equation described in Fig. 9 [18].
We provide significant evidence that indeed this reaction has
occurred on the Hombikat surface. The XRF analysis proves the
2
surface through the
2
to the TiO surface preferentially through the acidic OH sites, as
demonstrated in Fig. 9. It can be concluded that for TiSi2.1 the
acidic OH is not fully depleted upon silylation. This is important
because both acidic and basic OH sites are involved in cyclohexane
photo-oxidation, as proposed in a previously reported reaction
mechanism [8]. Another factor preventing full silylation is the cov-
presence of silicon on the TiO
.1 wt.% Si loading, corresponding to a maximum surface coverage
of 60%. Silane coupling on the surface is furthermore demonstrated
2
surface, with a range of 0 up to
2
erage of the TiO
2
surface with up to three hydrogen-bonded water
CH3
CH3
Si
H
N
H C
Si
CH3 (HMDS)
3
CH3
CH3
CH3
Si
CH3
+
H C
CH3
3
H C
CH3
3
H
H
H
H
Si
OH
Ti
OH
Ti
OH
OH
OH
OH
Ti
OH
Ti
OH
O+
O+
O+
O+
O+
O+
-
NH3
Ti
Ti
Ti
Ti
surface and ammonia.
Fig. 9. Expected reaction of Hexamethyldisilazane (HMDS) with TiO
2
, producing a partially silylated TiO
2

1
22
A.R. Almeida et al. / Journal of Catalysis 273 (2010) 116–124
layers at ambient conditions, of which some is strongly adsorbed
23,35]. A pre-treatment at temperatures higher than 400 °C,
needed for complete dehydration (Fig. 1), was not applied in this
study, in order to prevent morphological changes and to maintain
a comparable activity of the samples. Summarizing, the synthesis
characteristic of the (silane-free) TiSi0 sample, upon UV illumina-
tion of silylated catalysts. Perhaps the best strategy is to apply
more stable silane groups, e.g., fluorine-containing silylation
agents, that have been confirmed to be very stable under photo-
catalytic conditions [36].
[
procedure leads to a successful partial silylation of the TiO
2
surface.
4.2. Effect of silylation on the reaction rate
4.1. Cyclohexanone and water desorption
In silylating the catalyst surface, the number of available active
The main objective of this study was to decrease the surface
hydrophilicity of TiO , by coupling hydrophobic groups on the sur-
face and stimulate desorption of photo-catalytically formed cyclo-
hexanone. The polar character of cyclohexanone and water
OH sites is reduced, so an effect on the rate of cyclohexanone for-
mation is expected [22]. In Fig. 10a, the initial rate of cyclohexa-
none formation is compared to the rate over partially deactivated
catalysts, as determined from the tangents of the exponential fits
at t = 0 min and t = 100 min in Fig. 8a. The initial reaction rate is
greatly decreased for catalysts with low silane content, but at
silane loadings above 1.0% there is an increase in the initial rate
of cyclohexanone formation. The large difference in initial rate ob-
served for TiSi0 relative to the silylated samples suggests the pres-
ence of a fraction of highly reactive sites on the surface. These
highly reactive sites are apparently preferentially converted to
2
2
molecules leads to desorption times from TiO into an organic
medium of over 30 min [9]. As desired, the cyclohexanone and
water desorption constants were significantly increased by silyla-
tion (Figs. 6 and 7). Since the desorption constants of both cyclo-
hexanone and water increase linearly with surface silane content,
we propose that the adsorption enthalpy of these molecules on
surface OH sites decreases by the presence of neighboring silane
groups. A progressive shift of the negative band from 3630 to
3 3 3
Ti–O–Si(CH ) groups upon silylation or blocked by NH adsorption
ꢀ1
3
650 cm with increasing silane content (Fig. 4), suggesting a de-
during silylation (Fig. 3). We conclude that these highly active sites
are very susceptible to deactivation, explaining the large difference
between initial cyclohexanone formation rate and the rate over
partially deactivated TiSi0 catalyst, compared to the silylated
materials. In the following, we provide a simplified explanation
to the trend in cyclohexanone formation rate over partially deacti-
vated surfaces, on the basis of the trend in desorption rate con-
stant, and availability of acidic OH sites, taken as a measure of
the photo-catalytic sites (Fig. 10b).
crease in acid strength of the OH groups, is in agreement with
weaker product adsorption involving H bridges.
The increased rate of desorption indicated earlier leads to a low-
er amount of deactivating species on the surface of catalysts with
more than 1.0 wt.% Si (Figs. 4 and 5). This can be explained as fol-
lows. It has been previously proposed that the main route for car-
boxylate formation is through the oxidation of surface-bound
Å
cyclohexanone by OH radicals [8]. The increased hydrophobicity
of the surface decreases the amount of adsorbed water, thus
The rate of cyclohexanone formation, r in mmol/min, can be de-
scribed by Eq. (1):
Å
decreasing the concentration of OH radicals formed during illumi-
Å
nation. The smaller availability of OH radicals contributes to the
K
Cane Cane
C
observed larger surface selectivity.
r ¼ k
f
N
T
ð1Þ
1
þ KCane
C
Cane þ KCnone
C
Cnone
Though the cyclohexanone desorption rates are increased and,
as a consequence, catalyst deactivation is reduced, surface silanes
proved to be unstable under photo-catalytic conditions (Fig. 4).
where k
f
(min/mmol) is the cyclohexanone formation constant, N
T
the number of available active sites (OH groups, lowered in amount
by catalyst deactivation and precursors for the active site formed by
The negative trend in the CH
the silane groups are oxidized, as observed in previous publications
17]. Analyzing the catalyst coating before and after reaction (not
3
deformation vibration indicates that
3
light activation), KCane, CCane (mmol/cm ) and KCnone, CCnone (mmol/
3
[
cm ) the Langmuir constants and concentrations of cyclohexane
shown), we observed that the decomposition of the silanes from
the surface is not complete, explaining the observed remaining po-
sitive effects. No evidence was found for the recovery of OH sites
and cyclohexanone, respectively. Because of the low temperatures,
it is assumed that the occupancy is high, implying that the fraction
of empty sites is negligible. Two limiting cases are considered.
5
4
0
0
(a)
(b)
6
4
2
.E-04
.E-04
.E-04
0
0
.2
.1
30
2
1
0
0
0.E+00
0
0.0
0
0.5
1
1.5
2
0
0.5
1
1.5
2
Si (wt%)
Si (wt%)
Fig. 10. (a) Initial cyclohexanone formation rate (black symbols) and rate over partially deactivated surfaces (gray symbols) during photo-catalytic cyclohexane oxidation, as
determined in the top illumination reactor. The arrow emphasizes the large difference in initial cyclohexanone formation rate between TiSi0 and TiSi0.1. (b) Peak area of
acidic OH (black) and cyclohexanone desorption constant (gray) as a function of silane content, compared with the cyclohexanone formation rate over partially deactivated
surfaces (dashed gray line).

A.R. Almeida et al. / Journal of Catalysis 273 (2010) 116–124
123
When KCane
C
Cane ꢃ [1 + KCnone
C
Cnone], Eq. (1) can be simplified
(0.00108 in mol Einsteinꢀ1) than for TiSi0 (0.00062 in mol Ein-
stein ), the consequence of the much slower rate of deactivation
of the silylated catalyst.
In summary, the data show that a relatively high level of silyla-
tion is required to create improved catalysts. Two regimes can be
defined:
ꢀ1
to:
r ¼ k
f
N
T
ð2Þ
When KCnoneCCnone ꢃ [1 + KCane
CCane], Eq. (1) can be simplified to:
K
Cane
C
C
Cane
f
r ¼ k N
T
ð3Þ
K
Cnone
Cnone
(1) At Si content <1.0 wt.%, the rate of cyclohexanone formation
follows the decrease in available OH sites. So the improve-
ment in cyclohexanone desorption rate constant is not
enough to compensate for the decrease in surface OH
groups.
Cyclohexanone is expected to have a much stronger adsorption
enthalpy than cyclohexane, so the second limiting case is most
realistic and considered in the following. For simplicity, we assume
that on bare TiO
rate in steady-state conditions (NT,TiS0 = N
three types of sites are distinguished: isolated OH sites (N
sites with a neighboring silane group (N ) and silylated inactive
sites (NSi), = N + N So for
+ N
This leads to the following rate equation:
2
only one set of OH sites determines the reaction
). On a silylated surface,
), OH
(
2) At Si content >1.0 wt.%, the rate of cyclohexanone formation
increases as a result of a further increased cyclohexanone
desorption rate constant. Under these conditions, the
desorption rate becomes dominant and over-compensates
the negative impact of the decrease in surface OH groups.
At higher silane content, the still available OH sites are char-
acterized by a higher turnover frequency than the original
free OH sites on TiSi0.
1
1
2
N
T
1
2
+ NSi
.
a
silylated surface
[
N
1
2
] < NT,TiSi0.
K
Cane
C
C
Cane
Cnone
K
Cane Cane
C
C
r ¼ k
f
N
1
þ k
f
N
2
ð4Þ
K
Cnone;1
K
Cnone;2 Cnone
There should be an optimum silane loading that combines the
increased surface hydrophobicity, with a sufficient number of
available surface active sites. Based on the data provided here,
the silylation loading should be at least 2.1 wt.%.
For TiSi0 this reduces to:
K
Cane
C
Cane
r
TiSi0 ¼ k
f
N
T;TiSi0
ð5Þ
K
Cnone;1
C
Cnone
The trends in surface OH sites (N
acidic OH groups represent the active sites, and desorption con-
1
+ N
2
), assuming that the
5
. Conclusions
stants (kdes) as a function of silylation are shown in Fig. 10b.
2 3 3
A commercial TiO catalyst was modified by coupling ASi(CH )
groups to its surface. The silane groups, which seems to react pref-
erentially with acidic OH sites, increase the hydrophobicity of the
1 2
The data shows for TiSi2.1: (N + N ) = 0.6NT,TiSi0. It should be
noted that for TiSi2.1, a surface coverage of 60% was obtained
and, as a consequence, the number of OH active sites which are
2
silylated-TiO catalysts. Desorption rates of cyclohexanone were
neighboring a silane group is predominant, so N
ꢂ 0.6NT,TSi0. For the rate of desorption of TiSi2.1, an analogous
relation holds: which corresponds to
Cnone,TiSi2.1 = (1/2.3)KCnone,TiSi0 (when kads is constant). So Eq. (4)
for TiSi2.1 can be simplified to:
2
1
ꢃ N and
increased by surface modification, and a linear relation between
silane loading and cyclohexanone desorption rate constant was
observed. Accordingly, during cyclohexane photo-oxidation, the
silylated catalysts showed lower tendency for deactivation. At
low silane contents, the rate of cyclohexanone formation over par-
tially deactivated surfaces decreased due to the dependency on the
number of OH groups. However, for higher silane content
N
2
kdes,TiSi2.1 = 2.3 kdes,TiSi0,
K
K
Cane Cane
C
r
TiSi2:1 ¼ k
f
0:6NT;TiSi0
¼ 1:4rTiSi0
C
Cnone
ð6Þ
ð1=2:3ÞKCnone;1
(
Si > 1.0 wt.%), the increased desorption rate of cyclohexanone
assuming that the cyclohexanone formation rate constant is the
same for TiSi0 and TiSi2.1. So the rate of cyclohexanone formation
is higher for TiSi2.1 than for TiSi0 due to the improved desorption
constant, despite the decrease in available active sites. Accordingly,
the available OH sites of TiSi2.1, taking into consideration that
approximately 60% of the OH groups is converted by silylation,
overcomes the lower number of surface OH sites with relatively
high stability.
The silylated surface groups were not stable under photo-cata-
lytic conditions. Improvement of the surface modification method
should include the use of fluorine-containing silylation agents,
which have been proven to resist photo-catalytic conditions.
show
a
higher turnover frequency than of TiSi0, i.e.,
ꢀ
3
ꢀ1
ꢀ3
ꢀ1
9
.3 ꢁ 10 min compared to 5.6 ꢁ 10 min , respectively. The
Acknowledgment
turnover numbers are 0.8 and 0.3 for TiSi2.1 and TiSi0 for the short
illumination time applied here. Previous work has shown that TiSi0
at longer illumination times shows a TON above unity and that
We thank STW, the Netherlands (Project DPC.7065) for financial
2
9
support. Thanks to Kristina Djanashvili for the Si MAS NMR mea-
surements and to Louise Vrielink (Utwente) for XRF measure-
ments. Thanks to Mariette de Groen for her contributions to this
work.
(
2
silylated) TiO is indeed a catalyst for the target reaction, albeit
deactivating.
As stated previously, Eq. (6) applies only to the rate of cyclohex-
anone formation at prolonged illumination times, after which a
certain amount of active sites is deactivated. To analyze the initial
rate of cyclohexanone formation on TiSi0, two sites should be con-
References
sidered: very reactive, rapidly deactivating OH sites (N
stable OH sites contributing to the conversion over partially deac-
tivated catalysts (N ). The cyclohexanone formation constant for N
) is apparently considerably larger than for N (k ). This has an
3
) and the
[
1] A. Fujishima, X.T. Zhang, C.R. Chim 9 (2006) 750.
[2] A. Mills, R.H. Davies, D. Worsley, Chem. Soc. Rev. 22 (1993) 417.
[3] M.A. Fox, M.T. Dulay, Chem. Rev. 93 (1993) 341.
[
[
[
[7] P. Du, J.A. Moulijn, G. Mul, J. Catal. 238 (2006) 342.
[8] A.R. Almeida, J.A. Moulijn, G. Mul, J. Phys. Chem. C 112 (2008) 1552.
[
1
3
4] J. Peral, X. Domenech, D.F. Ollis, J. Chem. Technol. Biotechnol. 70 (1997) 117.
5] H. Einaga, S. Futamura, T. Ibusuki, Appl. Catal. B – Environ. 38 (2002) 215.
6] S.A. Larson, J.L. Falconer, Catal. Lett. 44 (1997) 57.
(k
3
1
1
effect on the photonic efficiency, determined by the ratio of the ini-
tial rate of product formation and the incident photon flux, which
ꢀ1
is higher for TiSi0 (0.0025 in mol Einstein ) than for TiSi2.1
9] T.J.A. Renckens, A.R. Almeida, M.R. Damen, M.T. Kreutzer, G. Mul, Catal. Today,
in press, doi:10.1016/j.cattod.2009.12.002.
ꢀ1
(
0.0016 in mol Einstein ). It should be mentioned that in the final
stages of the reaction, the photonic efficiency is higher for TiSi2.1
[10] C.B. Almquist, P. Biswas, Appl. Catal. A – Gen. 214 (2001) 259.

1
24
A.R. Almeida et al. / Journal of Catalysis 273 (2010) 116–124
[
11] M.A. Brusa, Y. Di Iorio, M.S. Churio, M.A. Grela, J. Mol. Catal. A – Chem. 268
2007) 29.
[24] J.A. Rob van Veen, Fred T.G. Veltmaat, Gert Jonkers, J. Chem. Soc. Chem.
Commun. (1985) 1656.
(
[
[
[
12] P. Boarini, V. Carassiti, A. Maldotti, R. Amadelli, Langmuir 14 (1998) 2080.
13] W. Mu, J.M. Herrmann, P. Pichat, Catal. Lett. 3 (1989) 73.
14] N. Igarashi, K. Hashimoto, T. Tatsumi, Micropor. Mesopor. Mater. 104 (2007)
[25] G. Martra, Appl. Catal. A – Gen. 200 (2000) 275.
[26] T.A. Jurgens, J.W. Rogers, J. Phys. Chem. 99 (1995) 731.
[27] L.J. Bellamy (Ed.), The Infrared Spectra of Complex Molecules Advances in
Infrared Group Frequencies, Chapman and Hall Ltd., London, 1975.
[28] G. Busca, H. Saussey, O. Saur, J.C. Lavalley, V. Lorenzelli, Appl. Catal. 14 (1985)
245.
[29] K. Hadjiivanov, O. Saur, J. Lamotte, J.C. Lavalley, Z. Phys. Chem. – Int. J. Res.
Phys. Chem. Chem. Phys. 187 (1994) 281.
[30] C. Beck, T. Mallat, T. Burgi, A. Baiker, J. Catal. 204 (2001) 428.
[31] P.Z. Araujo, C.B. Mendive, L.A.G. Rodenas, P.J. Morando, A.E. Regazzoni, M.A.
Blesa, D. Bahnemann, Colloid Surf. A – Physicochem. Eng. Aspects 265 (2005)
73.
2
69.
[
[
[
15] M.V. Cagnoli, S.G. Casuscelli, A.M. Alvarez, J.F. Bengoa, N.G. Gallegos, M.E.
Crivello, E.R. Herrero, S.G. Marchetti, Catal. Today 107–08 (2005) 397.
16] M.L. Pena, V. Dellarocca, F. Rey, A. Corma, S. Coluccia, L. Marchese, Micropor.
Mesopor. Mater. 44 (2001) 345.
17] S. Ikeda, Y. Kowata, K. Ikeue, M. Matsumura, B. Ohtani, Appl. Catal. A – Gen.
2
65 (2004) 69.
[
[
18] M.I. Baraton, L. Merhari, J. Eur. Ceram. Soc. 24 (2004) 1399.
19] X.F. Li, H.X. Gao, G.J. Jin, L. Chen, L. Ding, H.Y. Yang, Q.L. Chen, J. Mol. Struct. 872
(
2008) 10.
20] A. Quintanilla, J.J.W. Bakker, M.T. Kreutzer, J.A. Moulijn, F. Kapteijn, J. Catal. 257
2008) 55.
[32] D.J. Yates, J. Phys. Chem. 65 (1961) 746.
[33] W.G. Su, J. Zhang, Z.C. Feng, T. Chen, P.L. Ying, C. Li, J. Phys. Chem. C 112 (2008)
7710.
[
(
[
[
21] A.Y. Fadeev, R. Helmy, S. Marcinko, Langmuir 18 (2002) 7521.
22] J.T. Carneiro, C.C. Yang, J.A. Moma, J.A. Moulijn, G. Mul, Catal. Lett. 129 (2009)
[34] M. Primet, P. Pichat, M.V. Mathieu, J. Phys. Chem. 75 (1971) 1221.
[35] A.Y. Nosaka, Y. Nosaka, Bull. Chem. Soc. Jpn. 78 (2005) 1595.
[36] Y. Kuwahara, K. Maki, Y. Matsumura, T. Kamegawa, K. Mori, H. Yamashita, J.
Phys. Chem. C 113 (2009) 1552.
1
2.
[
23] J. Soria, J. Sanz, I. Sobrados, J.M. Coronado, A.J. Maira, M.D. Hernandez-Alonso,
F. Fresno, J. Phys. Chem. C 111 (2007) 10590.