The control of selectivity in benzene hydroxylation catalyzed by TS-1: The solvent effect and the role of crystallite size

Article abstract of DOI:10.1016/j.jcat.2010.07.030

This paper deals with a study on benzene hydroxylation with hydrogen peroxide, catalyzed by TS-1. The reaction scheme consists of two kinetically parallel primary reactions, leading either to phenol or to benzoquinone. The two diphenol isomers, hydroquinone and catechol, were secondary products, formed on the external sites of TS-1 crystallites (intercrystalline reactivity), and in part in the bulk liquid phase also, because of thermally activated radical reactions. The role of the solvent was examined; because of the presence of methanol/water co-solvents mixture, in three-phase conditions (with two liquid phases) the formation of tar and by-products predominated over the formation of phenol. On the other hand, in two-phase conditions (with a single liquid phase), the selectivity to phenol was higher than 90% at low benzene conversion, but rapidly declined in conditions leading to benzene conversion higher than 3-4%. The kinetically parallel - although chemically consecutive - formation of benzoquinone was minimized by controlling the TS-1 crystallite size; in fact, small-sized TS-1 crystallites produced the highest primary selectivity to phenol. The possible influence of both different titanium environments and defective hydroxyls population was evaluated by means of diffuse reflectance UV-Vis and IR spectroscopy, respectively.

Full text of DOI:10.1016/j.jcat.2010.07.030

Journal of Catalysis 275 (2010) 158–169
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
The control of selectivity in benzene hydroxylation catalyzed by TS-1:
The solvent effect and the role of crystallite size
a
a
a
b
b
D. Barbera a, F. Cavani a, , T. D’Alessandro , G. Fornasari , S. Guidetti , A. Aloise , G. Giordano ,
*
M. Piumetti c, B. Bonelli c, C. Zanzottera c
a Dipartimento di Chimica Industriale e dei Materiali, ALMA MATER STUDIORUM Università di Bologna, Viale Risorgimento 4, 40136 Bologna, Italy
b Dipartimento di Ingegneria Chimica e dei Materiali, Università della Calabria, 87036 Arcavacata di Rende (CS), Italy
c Dipartimento di Scienza dei Materiali ed Ingegneria Chimica, Politecnico di Torino, Corso Duca degli Abruzzi 24, 10129 Turin, and INSTM Unit Torino Politecnico, Italy
a r t i c l e i n f o
a b s t r a c t
Article history:
This paper deals with a study on benzene hydroxylation with hydrogen peroxide, catalyzed by TS-1. The
reaction scheme consists of two kinetically parallel primary reactions, leading either to phenol or to
benzoquinone. The two diphenol isomers, hydroquinone and catechol, were secondary products, formed
on the external sites of TS-1 crystallites (intercrystalline reactivity), and in part in the bulk liquid phase
also, because of thermally activated radical reactions. The role of the solvent was examined; because of
the presence of methanol/water co-solvents mixture, in three-phase conditions (with two liquid phases)
the formation of tar and by-products predominated over the formation of phenol. On the other hand, in
two-phase conditions (with a single liquid phase), the selectivity to phenol was higher than 90% at low
benzene conversion, but rapidly declined in conditions leading to benzene conversion higher than 3–4%.
The kinetically parallel – although chemically consecutive – formation of benzoquinone was minimized
by controlling the TS-1 crystallite size; in fact, small-sized TS-1 crystallites produced the highest primary
selectivity to phenol. The possible influence of both different titanium environments and defective
hydroxyls population was evaluated by means of diffuse reflectance UV–Vis and IR spectroscopy,
respectively.
Received 9 June 2010
Revised 30 July 2010
Accepted 30 July 2010
Keywords:
Benzene hydroxylation
Phenol
Titanium–silicalite
TS-1
Hydrogen peroxide
Diphenols
Ó 2010 Elsevier Inc. All rights reserved.
1. Introduction
oxidized products – i.e. hydroquinone, catechol and p-benzoqui-
none – are formed too, and therefore a strict control over both
One reaction of great interest for the chemical industry is the di-
rect hydroxylation of benzene to phenol with hydrogen peroxide
(HP) [1–4]. This process, if carried out with good yield and selectiv-
ity, could become competitive with the current production of phe-
nol, the main drawback of which is the co-formation of acetone.
A problem with the direct benzene oxidation is the cost of HP;
however, this drawback might be overcome by integrating the HP
production with the downstream use in benzene oxidation, in the
same way as is now done industrially in propene epoxidation. An
alternative approach would be the in situ generation of HP (directly
in the reaction environment where the benzene hydroxylation also
takes place), by reaction between hydrogen and oxygen, catalyzed
by noble metals, but the selectivity to HP achieved in this reaction
is still too low [1,3,5].
reaction conditions and benzene conversion is necessary to achieve
an acceptable selectivity to phenol.
Several catalysts have been reported for this reaction [6–19],
but one of the best-performing systems is TS-1 [20–29]. However,
looking at the literature published on benzene hydroxylation with
TS-1, several discrepancies are found, especially with regard to the
reaction network, the effect of solvents, and the effect of having
either bi-phase or tri-phase reaction systems. Therefore, we
decided to carry out a study on the reactivity of TS-1, with the
aim of providing a contribution to the understanding of these as-
pects, the comprehension of which is crucial for developing a more
selective and sustainable direct benzene hydroxylation process.
2. Experimental
Another problem in benzene hydroxylation with HP is the low
selectivity to phenol, mainly due to the fact that phenol is more
reactive toward oxidation than benzene [1]. This means that over-
Various TS-1 samples were prepared (Table 1). Sample TS-1A
was prepared following the procedure described in the ENI patent
[30]; the crystallite size of this sample was very low (less than
0.2 micron). The slurry containing TS-1 crystallites was spray-
dried, after the addition of silica Ludox (10 wt.% with respect to
*
Corresponding author. Fax: +39 051 209 3680.
E-mail address: cavani@ms.fci.unibo.it (F. Cavani).
0021-9517/$ - see front matter Ó 2010 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2010.07.030

D. Barbera et al. / Journal of Catalysis 275 (2010) 158–169
159
Table 1
Catalysts prepared and main characteristics.
Sample
Ti/Si (wt ratio %)
TPAOH/Si (molar ratio)
Preparation procedure
Crystallite size (
lm)
Surface area (m2/g)
UV–Vis bands (nm)
TS-1A
TS-1B
TS-1C
TS-1D
TS-1E
2.0
2.4
3.2
3.8
4.4
0.25
1.63
2.50
3.00
1.63
Conventional
Modified
Modified
Modified
Modified
<0.2c
0.45
0.40
0.31
0.45
500
615
510
510
534
210a
210a
210a
210a, 270b (weak)
210a, 270b (strong), 300 (sh)
a
b
c
Assigned to TiIV in tetrahedral coordination.
Assigned to TiIV in coordination higher-than-four [22,23].
This sample was used in the form of aggregated particles, with an average particle size of 25 micron.
TiO2), by using a lab-scale spray-drier LabPlant SD-05 (T of nebuli-
zation chamber 200 °C, nozzle diameter 2 mm). The final average
Catalytic tests were performed using the following procedure:
first, the reactor was loaded with benzene, solvent, and TS-1 cata-
lyst. Then the mixture was heated up to the desired reaction tem-
perature; at this point, the HP aqueous solution was fed to the
reactor under stirring, for 30 min, by means of a peristaltic pump;
after that, the reaction was left to proceed for 1.5 h more, for an
overall reaction time that in standard conditions was 2 h. During
this time, HP was also almost completely consumed, as confirmed
by iodometric titration; this is also important in order to avoid
reactions due to residual unconverted HP when injecting the solu-
tion containing the reaction products for GC analysis [31,32], caus-
ing misleading determination of the products selectivity. As long as
there is no residual HP in the sample, the GC analysis gives highly
accurate results as by means of HPLC [28]. On the other hand, we
found, by means of preliminary calibration tests, that problems
occurring during injection in the GC due to the presence of uncon-
verted HP are avoided when less than 800 ppm HP is present in the
injected sample. Therefore, our analytical procedure involved the
dilution of the final solution with methanol; this way, even if some
residual HP was left after reaction (as inferred by iodometric titra-
tion), its concentration was decreased by dilution down to less
than 800 ppm. An alternative approach uses MnO2 for the catalytic
decomposition of residual HP into oxygen and water [28]; this
method, however, must be used carefully, since the oxidation of
some phenolic compound may also occur.
size of agglomerated particles was ꢀ25
lm. The agglomerated
samples were then calcined in air at 550 °C.
Conversely, samples TS-1B, C, D and E were prepared using a
modified procedure, aimed at obtaining materials characterized
by crystallite size in the range of 0.3–0.6 micron. The main param-
eter changed, when compared to the standard preparation of TS-1,
was the TPAOH-to-TEOS ratio in the synthesis sol. In a typical pro-
cedure, tetraethylorthotitanate (TEOTi) was dissolved in tetraeth-
ylorthosilicate (TEOS); the clear yellow solution obtained was
stirred for 15–20 min, then TPAOH (20% aqueous solution) was
added. After further stirring, the resulting synthesis gel had the fol-
lowing composition: 1SiO2ꢁaTiO2ꢁbTPAOHꢁ23H2O (a = 0.0149, or
0.026 or 0.046; b = 1.63, or 2.5 or 3.0). The batch was kept in a
water bath at 80 °C for 4 h (for TEOS hydrolysis), then the mixture
was topped up to its original volume with deionized water, trans-
ferred into a Teflon-lined, stainless-steel autoclave, and heated at
180 °C for 24 h under autogeneous pressure. The autoclave was
then cooled and the white gelatinous solution formed was centri-
fuged at 10,000 rpm for 10 min. The supernatant liquid was dis-
carded, deionized water was added, and the process was
repeated 4–5 times until the pH of the supernatant solution was
close to neutral. Samples were activated by thermal treatment at
550 °C in air.
The final crystal morphology and dimensions of the TS-1 sam-
ples were studied by scanning electron microscopy (SEM) JEOL
JSM-T330-A. The silicalite structure was checked by means of pow-
der X-ray diffraction (XRD); diffractograms were recorded using a
The following amounts of reactants, solvent and catalysts were
used, unless otherwise specified: solvent volume (either methanol,
or methanol + water, or sulfolane + methanol) 52 mL; HP aqueous
solution (35 wt.%) 8.7 mL; benzene 0.1 mol; catalyst 1.17 g. The li-
quid phase reaction was conducted in a 150-mL batch autoclave
made of glass, equipped with a stirrer; heating was provided by
means of electric heating elements. The reactor operated at autog-
enous pressure (4–5 bar for the reaction carried out at 80 °C). We
decided to use a sealed reactor (after preliminary tests carried
out in an open system, equipped with a refrigerated condenser),
because we soon realized that it was not possible to avoid benzene
evaporation during reaction. This caused errors in the determina-
tion of benzene conversion and therefore made the determination
of C balances impossible. With the sealed reactor, C balances were
found to always fall between 95% and 105%. In the case of tests car-
ried out in three-phase conditions – which also led to a non-negli-
gible formation of tar – we evaluated the amount of tar produced
using a thin film evaporator, which allowed both the removal of
monoaromatic compounds and the calculation of the wt yield of
tar formed. This method was also systematically applied for the
determination of tar in phenol hydroxylation to diphenols. When
the reaction mixture was triphasic, we added methanol to the cold
slurry containing the reaction components before analysis, and in
order to obtain a single liquid phase, the catalyst was separated
by means of centrifugation.
Phillips PW 1710 diffractometer with Cu Ka radiation.
Fourier transform infrared (FT-IR) spectra were collected at
2 cmꢁ1 resolution on a Bruker FT-IR Equinox 55 spectrophotometer
equipped with an MCT detector. Samples were studied as either
dispersed in KBr pellets or self-supporting wafers. To prepare the
latter, powders were pressed into thin pellets (density of about
5 mg cmꢁ2) and pretreated under high vacuum (residual pressure
<10ꢁ3 mbar) using a standard vacuum frame, in an IR cell equipped
with KBr windows. To remove moisture and other atmospheric
contaminants and to study the hydroxylation degree of the surface,
wafers were outgassed for 1 h at 150 °C and 450 °C. To study both
Lewis and Brønsted acidic sites, ammonia was used as a probe mol-
ecule: in a typical experiment, increasing doses of ammonia (in the
0.01–15 mbar equilibrium pressure range) were introduced into
contact with the sample, and then the reversibility of adsorption
was checked by 300 evacuation at room temperature (r.t.).
Diffuse reflectance (DR) UV–Vis spectra were recorded using a
Cary 500 Scan UV–Vis NIR spectrophotometer (VARIAN), equipped
with an integrating sphere: about 400 mg powder samples were
outgassed under high vacuum (residual pressure <10ꢁ3 mbar) for
1 h at 150 °C and dosed with ammonia at room temperature
(1.0–15 mbar equilibrium pressure range). Afterward, reversibility
of adsorption was checked by 300 evacuation at room temperature
(r.t.).
For phenol reactivity tests, the following conditions were used:
phenol 4.3 mmol; solvent volume (methanol) 50 mL; HP aqueous
solution (35 wt.%) 5.6 mL; catalyst 0.45 g. The reactivity test was

160
D. Barbera et al. / Journal of Catalysis 275 (2010) 158–169
then conducted with the same procedure used for benzene
hydroxylation.
biphasic system
triphasic system
8
6
4
2
0
8
6
4
2
0
Before injection into the GC instrument, samples were diluted
with methanol (typically, 0.5 g of sample was diluted with 25 mL
of methanol), in order to decrease the concentration of unconverted
HP– thelatterhavingbeendeterminedbymeansofiodometrictitra-
tion) down to less than 800 ppm. The reaction products were ana-
lyzed by means of GC, using a Thermo instrument equipped with
FID and a capillary column HP5; oven temperature was programmed
from 50° to 250 °C (internal standard: undecane). The error made
during catalytic experiments was evaluated after repeating an
experiment three times, at the following given conditions: HP
(35 wt.%) 8.7 mL; benzene 0.1 mol (8.9 mL, 7.8 g); solvent [water/
(methanol + water) = 0.1, vol. fraction] 52 mL; catalyst 1.17 g
(TS-1A); temperature 80 °C; reaction time 2 h; benzene/HP 1/1
(molar ratio). Results achieved were: benzene conversion 6.4 0.8
and phenol selectivity 71.3 2.5.
Competitive adsorption experiments were carried out by equil-
ibrating a solution of benzene–phenol (15–1 wt.%) in different sol-
vents (methanol only, methanol/water 5.5/1 wt ratio, methanol/
sulfolane 3.8/1 wt ratio) in the presence of TS-1A (6 wt.%), but
without HP, for 2 h at 80 °C. Then, the analysis of the liquid phase
as determined by GC (internal standard: undecane) was compared
with that of the original liquid phase.
0
0.2
0.4
0.6
0.8
1
Volumetric fraction of water
20
16
12
8
100
80
60
40
20
0
4
0
3. Results
0
0.2
0.4
0.6
0.8
1
Volumetric fraction of water
3.1. The role of solvent: biphasic or triphasic?
Fig. 1. Effect of the volumetric fraction of water in water + methanol solvent
mixture on catalytic performance in benzene hydroxylation. Symbols: yield to
phenol (j), CT (ꢀ), HQ (d) BQ (N), benzene conversion (ꢂ), and selectivity to phenol
(h). Dotted line (right-hand scale in top figurer): volumetric water/benzene ratio
(included water from HP solution). Reaction conditions: HP (35 wt.%) 8.7 mL;
benzene 0.1 mol (8.9 mL, 7.8 g); solvent (water + methanol) 52 mL; catalyst 1.17 g
(TS-1A). Temperature 80 °C; reaction time 2 h; benzene/HP 1/1 (molar ratio).
One controversial point of TS-1 reactivity during benzene oxi-
dation concerns how the presence of either a triphasic or a biphasic
system affects the performance. In order to elucidate this aspect,
we planned specifically designed experiments. Fig. 1 shows the re-
sults of tests carried out keeping the benzene-to-HP molar ratio
fixed, and the overall solvent volume also, while varying the rela-
tive amount of methanol and water in the solvent mixture (all tests
shown in Figs. 1–7 were carried out using the TS-1A sample). The
system was biphasic in the left-hand part of the figure (where
methanol was the predominant solvent), but was triphasic in the
right-hand part (with water as the predominant solvent). Reaction
products were phenol, hydroquinone (HQ), catechol (CT) and p-
benzoquinone (BQ), as is well known from the literature [1–3];
worth of note, there was no formation of o-benzoquinone, or of
any other oxygenated by-product, that also agrees with literature.
In triphasic conditions, tar also formed.
An increasing amount of water led first to an increase in phenol
yield, whereas the overall yield to by-products remained substan-
tially constant (this clearly implies an increase in both benzene
conversion and phenol selectivity from the initial value of 71% up
to 82%, Fig. 1 [bottom]); the maximum yield to phenol achieved
was 7.2%. After that, i.e. when the system became triphasic, there
was a dramatic increase in the yield of by-products, especially to
HQ and CT, and a fall in the yield (and selectivity as well) of phenol;
the reaction was also accompanied by the significant formation of
tar. BQ formed in amounts greater than diphenols in biphasic med-
ium, but the opposite was true when the reaction was carried out
in triphasic conditions. As for the reaction in the biphasic zone,
there was an optimum relative amount of methanol and water;
with low water content, the yield of phenol was low, and it in-
creased when the molar fraction of methanol was decreased.
In Fig. 2, we report the results of experiments aimed at examin-
ing again the effect of an increased water-to-benzene ratio on cat-
alytic behavior; however, tests were carried out by keeping the
volume of methanol solvent constant (unlike what we had done
in the case of the tests shown in Fig. 1), that is the methanol-to-cat-
20
16
12
8
100
80
60
40
20
0
4
0
0
2
4
6
8
10
12
14
Water/benzene (vol ratio)
Fig. 2. Effect of the water/benzene volumetric ratio on catalytic performance in
benzene hydroxylation. Symbols: benzene conversion (ꢂ), selectivity to phenol (h),
CT (e), HQ (s) and BQ (4). Reaction conditions: HP (35 wt.%) 8.7 mL; benzene
0.1 mol; methanol 52 mL; catalyst 1.17 g (TS-1A); the amount of water was
increased (therefore, the total volume changed). Temperature 80 °C; reaction time
2 h; benzene/HP 1/1 (molar ratio); biphasic system.
alyst ratio, and the benzene-to-HP molar ratio as well, while the
amount of water added (and the overall volume solvent) was in-
creased. With these experiments, it was possible to highlight the
role of water, while maintaining a biphasic system (because of
the presence of an amount of methanol sufficient to dissolve all
components); indeed, in the experiments of Fig. 1, the formation

D. Barbera et al. / Journal of Catalysis 275 (2010) 158–169
161
10
8
100
80
60
40
20
0
20
16
12
8
100
80
60
40
20
0
6
4
2
4
0
0
0
0.5
1
1.5
2
0
20
40
60
80
100
HP/benzene (molar ratio)
Sulfolane (wt % in solvent)
20
16
12
8
100
Fig. 3. Effect of the HP/benzene molar ratio on catalytic performance in benzene
hydroxylation. Symbols: benzene conversion (ꢂ), selectivity to phenol (h) and BQ
(4). Reaction conditions: HP 8.75 mL (at varying concentrations); benzene 0.1 mol;
methanol 52 mL; catalyst 1.17 g (TS-1A). Temperature 60 °C; reaction time 2 h.
Note that due to the large excess of methanol, the system is always biphasic (single
liquid phase).
80
60
40
20
0
20
16
12
8
100
80
60
40
20
0
4
0
0
20
40
60
80
100
Sulfolane (wt % in solvent)
Fig. 6. Effect of the sulfolane content in sulfolane/methanol solvent mixture on
catalytic performance in benzene hydroxylation with 7.5% HP (top figure) and with
3% HP (bottom figure). Symbols: benzene conversion (ꢂ), selectivity to phenol (h),
CT + HQ (s), and BQ (4). Reaction conditions: benzene 0.1 mol; solvent (metha-
nol + sulfolane) 52 mL; catalyst 1.17 g (TS-1A). Temperature 80 °C; reaction time:
2 h; benzene/HP 1/1 (molar ratio).
4
0
0
1
2
3
Reaction time (h)
20
16
12
8
100
80
60
40
20
0
Fig. 4. Effect of reaction time on catalytic performance in benzene hydroxylation
with diluted HP solution (3 wt.%). Symbols: benzene conversion (ꢂ), selectivity to
phenol (h), CT (}), HQ (s) and BQ (4). Reaction conditions: HP (3 wt.%) 50 mL;
benzene 0.1 mol; methanol 52 mL; catalyst 1.17 g (TS-1A). Temperature 80 °C;
benzene/HP 1/1 (molar ratio).
20
16
12
8
100
80
60
40
20
0
4
0
0
3
6
9
12
15
Water/benzene (vol. ratio)
Fig. 7. Effect of the water/benzene volumetric ratio on catalytic performance in
benzene hydroxylation. Symbols: benzene conversion (ꢂ), selectivity to phenol (h),
CT (}), HQ (s) and BQ (4). Reaction conditions: benzene 0.1 mol; sulfolane 32 mL;
catalyst 1.17 g (TS-1A); the amount of water was increased (therefore, the total
volume changed). Temperature 80 °C. Benzene/HP 1/1 (molar ratio).
4
0
0
20
40
60
80
100
Sulfolane (wt % in solvent)
when the water-to-benzene ratio was increased. The effect on
selectivity was less pronounced than in the tests of Fig. 1.
Fig. 5. Effect of the sulfolane content in sulfolane/methanol solvent mixture on
catalytic performance in benzene hydroxylation. Symbols: benzene conversion (ꢂ),
selectivity to phenol (h), CT + HQ (s), and BQ (4). Reaction conditions: benzene
0.1 mol; solvent (methanol + sulfolane) 52 mL; catalyst 1.17 g (TS-1A). Temperature
80 °C; reaction time: 2 h; benzene/HP 1/1 (molar ratio).
3.2. Analysis of the reaction network
Since the formation of BQ should normally occur by means of a
two-step phenol oxidation (first to HQ and then to BQ), it is reason-
able to expect that the concentration of HP may affect the distribu-
tion of the products. Therefore, tests were carried out using various
of the triphasic system, for intermediate values of water molar
fraction, may interfere with the effect of solvent variation. Once
again, as for tests shown in Fig. 1, benzene conversion increased

162
D. Barbera et al. / Journal of Catalysis 275 (2010) 158–169
concentrations of HP; the molar ratio between benzene and HP
was changed, but the overall volume of the aqueous solution of
HP was kept constant. Reaction time and temperature (2 h and
60 °C) were chosen so as to obtain low benzene conversion and
to best highlight the effect of HP concentration on the selectivity
to the products of primary oxidation. Moreover, the overall amount
of water was kept low, in order to develop biphasic systems
throughout all the experiments.
only sulfolane solvent, however, the selectivity to BQ became nil,
whereas that to diphenols increased remarkably.
Finally, Fig. 7 sums up the results achieved with 90–100% sulfo-
lane solvent (that is, with nil or a very small amount of methanol),
while increasing the HP concentration (that is, decreasing the
water/benzene ratio, while keeping the HP/benzene molar ratio
constant). It is interesting to compare this figure with the results
plotted in Fig. 2, where the effect of the same parameter, water/
benzene ratio, on the reactivity behavior is reported, for tests car-
ried out in methanol solvent. It is shown that in both cases an in-
crease in the water/benzene ratio led to an increase in benzene
conversion.
Fig. 3 reports the results of these experiments. In these condi-
tions, there was no formation of CT and HQ, but only of phenol
and BQ; the selectivity to this latter compound increased when
the molar ratio between benzene and HP was decreased, and cor-
respondingly the selectivity to phenol declined. The increase in
HP concentration also accelerated benzene transformation.
These experiments suggest that the phenol formed inside the
pores is oxidized to HQ – the formation of CT being unfavored
for steric reasons – and HQ is then very quickly transformed into
BQ, before it can counterdiffuse toward the bulk phase. In order
to confirm this hypothesis, we carried out tests at increasing reac-
tion times by using an HP concentration that permitted only the
formation of phenol (the first experimental point in Fig. 3), in order
to distinguish between truly primary and secondary products.
Fig. 4 shows the results of this experiment, that is, the effect of
the reaction time on the selectivity to products. These data confirm
the hypothesis stated: CT and HQ both showed the classical behav-
ior of a kinetically secondary product, obtained by the consecutive
oxidation of phenol; conversely, BQ was clearly a primary product
of oxidation, as it was phenol.
3.4. Competitive adsorption of benzene and phenol
Competitive adsorption tests were carried out by equilibrating
liquid mixtures of different compositions in the presence of TS-
1A, without HP. Mixtures were mixed for 2 h at 80 °C, in a closed
vessel; then the final bulk liquid phase was analyzed and its com-
position was compared with that of the starting mixture. The aim
was to find out how the nature of solvents used affected the distri-
bution of selected compounds between the liquid phase and TS-1.
Results of these experiments are compared in Table 2, which
shows the relative variation in concentration of the aromatic com-
ponents in the bulk liquid phase after equilibration. The behavior
of the mixtures containing either sulfolane or water as co-solvents
for methanol appears to be very similar, whereas the use of meth-
anol, the less polar solvent, alone led to the preferred adsorption of
phenol, which showed the greatest relative change of concentra-
tion in the liquid phase. This indicates that the competitive adsorp-
tion in hydrophobic TS-1 was controlled by the solvent
characteristics.
These data suggest that the formation of HQ and CT indeed
might occur either at the intercrystalline (external) Ti sites or even
in the bulk liquid phase by means of radical reactions between
ꢁ
phenol and OH radicals, because phenol is much more reactive
than benzene. Blank experiments made by reacting phenol with
hydrogen peroxide, in the same experimental conditions as used
for benzene hydroxylation, but without catalyst, led to the follow-
ing result: phenol conversion 0.8%, with comparable selectivity to
CT and HQ. This confirms that a small contribution to phenol con-
secutive hydroxylation to diphenols may also derive from ther-
mally activated homogeneous reactions.
3.5. The role of crystallite size
The preparation of Ti–silicalite by the conventional synthesis
method (sample TS-1A) leads to crystallites of an extremely low
size, which must be agglomerated into bigger particles in order
to allow the filtration of the used catalyst from the reaction batch.
In order to study the effect of the crystallite size on the primary
selectivity in benzene oxidation, we set up a modified synthetic
protocol for TS-1. This protocol allowed the formation of larger
crystallites than those obtained with the conventional procedure
and, lastly, permitted the use of TS-1 for reactivity experiments,
without the need for agglomeration into larger particles. Further-
more, we found that by using a TPAOH-to-TEOS ratio higher than
1, it was possible to control the crystallite size of TS-1. This is
shown in Table 1, which summarizes the main features of the
samples synthesized with the modified procedure; XRD, FT-IR
and Raman spectroscopy all confirmed the formation of original,
well-crystallized TS-1 samples, with no formation of segregated
anatase (with the exception of sample TS-1E).
3.3. The use of sulfolane co-solvent
Fig. 5 compiles the results aimed at understanding the role of
sulfolane as the co-solvent; tests were carried out by changing
the sulfolane content in the sulfolane/methanol solvent mixture
and using 35% HP solution. It is shown that the effect of added sul-
folane on BQ selectivity was minimal, whereas the selectivity to
diphenols went down to almost zero. The selectivity to phenol
was the higher with 25–50 wt.% sulfolane in the solvent mixture.
There was also an effect on conversion, which decreased; however,
this change was not great enough to be the only one responsible
for the drastic fall in selectivity to diphenols.
Results of experiments carried out with diluted HP solutions are
illustrated in Fig. 6a (HP solution 7.5%) and 6b (HP solution 3%); in
these experiments, the benzene-to-HP ratio was always 1, whereas
the water-to-benzene ratio increased, due to the use of the diluted
HP solution. The effect of sulfolane with the 7.5% HP solution was
the opposite of that obtained with the concentrated HP solution
(shown in Fig. 5). There was a remarkable effect on BQ selectivity
(which decreased when the 50% sulfolane was used), whereas
there was no effect on the selectivity to CT + HQ. Less clear is the
effect of sulfolane with extremely diluted HP solutions, that is, in
conditions leading to the triphasic system (Fig. 6b). In this case,
the presence of 50% sulfolane led to a drastic fall in benzene con-
version and to an increase in the selectivity to phenol, with a cor-
responding decrease in selectivity to both BQ and CT + HQ; with
Table 2
Results of competitive adsorption experiments with TS-1A.
Experiment
number
Composition of the
initial mixture
Change of aromatics
concentration in the
bulk liquid phase
after equilibration
1
2
3
Methanol, benzene,
phenol, TS-1
Benzene – 1%
Phenol – 21%
Methanol, water, benzene,
phenol, TS-1
Benzene – 12%
Phenol – 14%
Methanol, sulfolane,
benzene, phenol, TS-1
Benzene – 14%
Phenol – 13%

D. Barbera et al. / Journal of Catalysis 275 (2010) 158–169
163
The values reported in Table 1 show that
Ti content and the presence of pentacoordinated Ti species, which
are considered to be especially active species [22,23]. One distin-
guishing feature of TS-1D was the lower crystallite size, which
means a shorter mean intracrystalline path for reactants and prod-
ucts and thus a lower effective reaction time constant. In the case
of TS-1E, not only the high Ti content and the presence of a fairly
intense band at 270 nm, but also a large crystallite size, are all fac-
tors that may contribute to obtaining the highest conversion of
benzene.
However, for our purposes, the conversion achieved itself is not
so important; the aim of these experiments, in fact, was that of
comparing the selectivity to the products of primary benzene oxi-
dation: Fig. 8 shows that there was a clear inverse relationship be-
tween the crystallite size and the selectivity to phenol.
1. The crystal size was affected exclusively by the TPAOH/TEOS
molar ratio (that is, the TPAOH/Si ratio), regardless of the Ti/Si
ratio used (compare samples TS-1B and TS-1E). An increase in
the TPAOH/Si ratio led to a non-negligible, systematic effect
on the crystal size; the latter was the smallest in samples pre-
pared using the highest TPAOH/Si ratio. Therefore, it is evident
that large amounts of TPAOH facilitate the nucleation process
when compared to the crystallite growth process.
2. All samples show the band at 210 nm, which is attributed to
tetrahedral TiIV incorporated within the silicalite framework.
However, in samples TS-1D and TS-1E (especially the latter),
an additional band at 270 nm was present: in the literature, this
event is attributed to the formation of a Ti center with an
enlarged coordination sphere, but still isolated in the silicalite
matrix [28,33–35].
3.6. The reactivity in phenol hydroxylation
We conducted experiments aiming to confirm the role of crys-
tallite size on catalytic behavior in benzene hydroxylation. Specif-
ically, reactivity tests were carried out using phenol as reactant,
under the same reaction conditions as used for benzene hydroxyl-
ation experiments. It should be noted that these conditions are not
the optimal ones typically used for the synthesis of diphenols by
phenol hydroxylation with TS-1; in the latter case, in fact, a mix-
ture of acetone and water is the reaction solvent, and a 60% HP
solution is used.
Results of experiments are summarized in Table 3. The main
differences when compared to the synthesis of diphenols, when
carried out at the optimal reaction conditions, concern the forma-
tion of BQ [2,36,37]. The latter forms in negligible amounts in the
process for diphenol production, but it is the prevailing product
in the experiments described in the present work. Also, the consid-
erable formation of tar is attributable to the fact that water and
methanol, unlike acetone, are not good solvents for tar [37]. As
for differences among the various TS-1 samples, the following is
worth mentioning: (a) the yield to BQ was by far the lowest with
TS-1D, which also produced no diphenols, but a significant amount
of tar; (b) TS-1B, TS-1C and TS-1E showed a similar behavior with
regard to the formation of BQ, but the yield to diphenols and tar
shown by these samples was very different.
Catalytic experiments carried out on samples synthesized with
the modified procedure are summarized in Fig. 8, plotting the
crystallite size, the conversion of benzene, and the selectivity to
phenol on the basis of the Ti/Si wt ratio. There was no clear rela-
tionship between the overall benzene conversion and sample
features, probably because conversion was a complex function of
various structural and chemical features, i.e. the Ti/Si ratio (which
also affected the hydrophilicity of samples, see characterization of
Ti species) and the crystallite size. In fact, TS-1C was more active
than TS-1B: this may be attributed to the higher amount of frame-
work Ti (in both samples, the only Ti species was the tetrahedral
one), but TS-1D was as active as TS-1B, despite both the higher
0.5
TS-1B
TS-1E
0.45
0.4
TS-1C
0.35
0.3
3.7. UV–Vis and IR spectroscopic characterization of Ti species
TS-1D
Fig. 9a shows the DR UV–Vis spectra of TS-1B, TS-1D and TS-1E
powders both (i) outgassed at 150 °C and (ii) under approximately
15 mbar ammonia equilibrium pressure (bold curves).
2
3
4
5
Ti/Si (wt ratio, %)
The spectra of samples outgassed at 150 °C show a main
absorption band at 205–210 nm, assigned in the literature to
Ti4+O2ꢁ ? Ti3+Oꢁ charge-transfer of highly dispersed tetrahedral
TiO4 species, incorporated within the framework [38,39]; as ex-
pected, the overall intensity of absorption increases with the tita-
nium content (Table 1). The two samples with the highest
titanium content, i.e. TS-1D and TS-1E, show an additional band
at 270 nm that, according to the literature, are assigned to very
5
4
3
2
1
0
100
96
92
88
84
80
reactive titanium species in
a coordination higher-than-four
formed in the TS-1 obtained by the modified synthetic procedure
[22,23]; other possible assignments for the 270 nm band are either
Table 3
2
3
4
5
Reactivity of TS-1 catalysts in phenol hydroxylation.
Ti/Si (wt ratio, %)
Sample
BQ yield (%)
CT + HQ yield (%)
Tar yield (%)
Fig. 8. Plot of crystal size (top figure) and of benzene conversion (j) and phenol
selectivity (h) (bottom figure) in function of the Ti/Si ratio used for samples
preparation (see Table 1 for main samples characteristics). Reaction conditions: HP
(35 wt.%) 8.7 mL; benzene 0.1 mol (8.9 mL, 7.8 g); methanol 52 mL; catalyst 1.17 g
(TS-1A). Temperature 80 °C; reaction time 2 h; benzene/HP 1/1 (molar ratio).
TS-1B
TS-1C
TS-1D
TS-1E
16
24
4
2
6
0
0
41
10
26
17
16

164
D. Barbera et al. / Journal of Catalysis 275 (2010) 158–169
indicating that this moiety interacts strongly with tetrahedral Ti
species, increasing their coordination state. The 270 nm band
seems to be not too affected by the presence of ammonia, thus con-
firming that the corresponding titanium species are in a higher-
than-four coordination state and therefore less prone to interact
with another ligand.
The presence of framework tetrahedral titanium species is con-
firmed in Fig. 9b by the IR spectra of the samples dispersed in KBr
pellets: in addition to the bands of asymmetric and symmetric
stretch vibrations of Si–O–Si groups in the silica matrix [42], the
spectra reported in Fig. 9b show a band at 960 cmꢁ1, readily as-
signed to the asymmetric stretch mode of Si–O–Ti bridges. Accord-
ing to the literature, the latter band may be considered
a
fingerprint of the presence of tetrahedral Ti species in the frame-
work [42]. On the other hand, the shift of the asymmetric Si–O–
Si stretch band toward lower wavenumbers with increasing tita-
nium loading suggests the formation of extra-framework TiOx
species.
Fig. 10a reports IR spectra in the OH stretch range (3800–
3000 cmꢁ1) of self-supporting wafers outgassed at 150 °C and
450 °C: all spectra were normalized to unit weight to allow com-
parison and were stacked for the sake of clarity. The spectra of
samples outgassed at 150 °C (upper curves) show components at
3740, 3727, 3700 cmꢁ1 and a broad absorption extending below
3600 cmꢁ1. These features are assigned respectively to the pres-
ence of (i) free isolated silanols on the outer surface of crystals
(3740 cmꢁ1); (ii) free isolated silanols inside Ti–silicalite cavities
(3727 cmꢁ1); (iii) terminal H-bonded silanols of hydroxyl nests lo-
cated inside micro-channels (3700 cmꢁ1); and (iv) H-bonded sila-
nols inside hydroxyl nests [43]. Outgassing at 450 °C (lower
curves) brings about an overall decrease in band intensity due to
dehydration, in particular as far as OH species belonging to hydro-
xyl nests are concerned, in keeping with their high reactivity.
The comparison among samples shows that, as a whole, TS-1E
has a higher amount of defective hydroxyls, as indicated by the
much higher intensity of the broad absorption extending below
1.5
1.0
0.5
0.0
b
TS-1B
TS-1D
TS-1E
3600 cmꢁ1
.
960
The acidity of both Lewis and Brønsted surface species was
studied by means of ammonia adsorption at room temperature:
Fig. 10b shows IR difference spectra recorded after dosing NH3
(equilibrium pressures in the 0.1–15 mbar range) on the TS-1E
sample outgassed at 150 °C. Difference spectra were obtained by
subtracting the spectrum of the bare sample reported in Fig. 10a;
therefore, the negative bands in Fig. 10b correspond to species dis-
appearing upon interaction with NH3 molecules and positive bands
are the corresponding features of ammonia adducts.
1400
1300
1200
1100
1000
900
800
Wavenumbers (cm-1)
Dosage of NH3 brings about the formation of two main bands at
1457 cmꢁ1 (with a component forming at higher pressure at ap-
prox. 1490 cmꢁ1) and at 1607 cmꢁ1. The former is due to ammo-
nium species formed by the interaction of ammonia with
Brønsted acidic sites, as confirmed by the corresponding negative
bands observed in the OH stretch region. It may be inferred that
at least two kinds of Brønsted hydroxyls with different acidic
strengths occur on the surface of the TS-1E sample: this is shown
by both the presence of two components (at 1457 and
1490 cmꢁ1) and the fact that after 300 evacuation at room temper-
ature, the 1457 cmꢁ1 band is still present (dotted spectrum). The
latter band is therefore assigned to ammonium species, which
are formed after reaction with more acidic hydroxyls, most proba-
bly TiOH species, and the component at 1490 cmꢁ1 to ammonia
molecules reacting with less acidic species, i.e. nest hydroxyls
[44]. The presence of Lewis acidic sites along with Brønsted
hydroxyls is confirmed by the band at 1607 cmꢁ1, assigned to
ammonia molecules interacting with TiIV sites [45].
Fig. 9. (a) DR UV–Vis spectra of TS-1B, TS-1D and TS-1E powders outgassed at
150 °C and under approximately 15 mbar NH3 equilibrium pressure (curves labeled
as TS-1B NH3, TS-1D NH3 and TS-1E NH3). (b) FT-IR spectra, in the 1400–750 cmꢁ1
range of TS-1B (continuous line), TS-1D (bold line) and TS-1E (dotted line) powders
dispersed in KBr pellets.
partly aggregated hexacoordinated Ti–O–Ti species or isolated sin-
gle-bonded TiOx species attached to the zeolite lattice [40]. With
TS-1E, a further absorption is observed above 300 nm, indicating
that the formation of Ti–O–Ti groups (e.g. extra-framework TiO2)
[41] occurs at the highest titanium loading.
Dosage of ammonia brings about an overall absorption intensi-
fication and a red-shift of the 210 nm band, with the formation of a
shoulder at about 245 nm, due to an increase in titanium coordina-
tion that can lead to the formation of octahedral Ti complexes in
which two ammonia molecules are coordinated by each TiIV site
[38]. Ammonia adsorption was not reversible by outgassing at
room temperature (spectra not reported for the sake of clarity),
At higher pressures another component shows up at 1625 cmꢁ1
due to ammonia molecules H-bonded to less acidic isolated silanols
,

D. Barbera et al. / Journal of Catalysis 275 (2010) 158–169
165
1.0
b
TS-1E
1607
1457
1490
0.5
1625
3740
3727
a
0.0
TS-1B
1.0
0.5
0.0
TS-1D
TS-1E
3700
-0.5
3600 3300 3000 2700 1800
1600
1400
Wavenumbers (cm-1)
0.10
0.05
0.00
150 °C
450 °C
TS-1E
TS-1D
TS-1B
c
1607
3800
3600
3400
Wavenumbers (cm
3200
3000
-1
)
1700
1650
1600
1550
1500
1450
1400
Wavenumbers (cm-1)
Fig. 10. (a) FT-IR spectra in the 3800–3000 cmꢁ1 range of TS-1B (continuous line), TS-1D (bold line) and TS-1E (dotted line) self-supporting wafers outgassed at 150 °C (upper
curves) and 450 °C (lower curves). (b) Difference spectra recorded after dosing increasing pressures of ammonia (0.1–15 mbar range) on TS-1E sample outgassed at 450 °C;
the dotted line corresponds to the spectrum recorded after outgassing for about 30’ at room temperature. (c) Comparison of normalized difference spectra in the NH bending
region recorded on samples TS-1B (continuous line), TS-1D (bold line) and TS-1E (dotted line) outgassed for about 300 at room temperature after dosing approximately
15 mbar ammonia.
originally absorbing at 3740 cmꢁ1 (negative band in the figure) [45].
After prolonged evacuation at room temperature (dotted curve), the
adsorption of ammonia proved to be not completely reversible, thus
indicating the presence of rather strong Brønsted and Lewis sites,
due to TiOH and TiIV species, respectively.
sites, showing that a fraction of titanium is not accessible to this
probe and therefore is less reactive.
4. Discussion
Dosage of ammonia on the TS-1E sample outgassed at 450 °C
(spectra not reported for the sake of brevity) shows the presence
both of Lewis acidic sites (band at 1607 cmꢁ1) and of a band at
1480 cmꢁ1 due to fewer ammonium species: this latter phenome-
non is due to surface dehydroxylation, with the removal of most
acidic (i.e. reactive) hydroxyls, since the component at
1457 cmꢁ1 due to TiOH species is no longer seen and only the fea-
tures of ammonia reacting with silanols are observed. The latter
interaction is indeed reversible by evacuation at room tempera-
ture. Similar IR features were observed with both the TS-1B and
TS-1D samples: Fig. 10c compares difference spectra in the
1700–1375 cmꢁ1 range recorded on samples evacuated at room
temperature after ammonia dosage; difference spectra were ob-
tained by subtracting the spectra of the bare samples outgassed
at 150 °C which are shown in Fig. 10b. The same features ascribable
to the presence of titanium are seen: two main bands are observed
at 1607 cmꢁ1 and 1457 cmꢁ1, assigned respectively to ammonia
molecules interacting with TiIV Lewis sites and to ammonium spe-
cies formed upon interaction with TiOH Brønsted hydroxyls.
Although sample TS-1E has the highest Ti content, it does not show
the most intense bands due to ammonia adsorbed on titanium
4.1. The reaction scheme in benzene hydroxylation: intraparticle and
interparticle products
From the literature, it is not possible to infer a clear picture of
the reaction network for benzene hydroxylation [1–3]; in general,
it is reported that by-products of the reaction are HQ, CT, BQ and
tar, and that the less sterically hindered HQ and BQ may preferen-
tially form inside TS-1 pores, whereas the formation of bulkier CT
occurs on external sites. However, no scheme showing kinetic rela-
tionships between the various products has been reported.
Experiments reported in Fig. 4 clearly indicate that the only pri-
mary products of benzene oxidation are phenol and BQ (at least,
when diluted HP solutions are used). Therefore, a fraction of phe-
nol, once formed at intracrystalline Ti sites, is oxidized into HQ
(the formation of CT being disadvantaged because of steric rea-
sons) and then quickly to BQ before it can leave the porous TS-1
structure. The fast oxidation of HQ into BQ entails the homolythic
scission of the HO–u–O–H bond to generate HO–u
–Oꢁ, catalyzed
inside TS-1 pores by Ti–Oꢁ sites. With regard to the primary forma-
tion of phenol, this product may be formed either on external,

166
D. Barbera et al. / Journal of Catalysis 275 (2010) 158–169
intercrystalline Ti sites or inside the porous structure of TS-1,
thereafter counterdiffusing toward the liquid phase before the
occurrence of any further oxidation (e.g. primary oxidation might
occur in pore mouths).
fusional constraints for the conversion of phenol and geometric
constraints for the formation of CT were highlighted; on the other
hand, with small crystals, the external surface sites were reported
to contribute significantly to the total rate of reaction and its selec-
tivity. The inertization or selective poisoning of the external sur-
face of TS-1 led to a decrease in the formation of tar and a slight
increase in the selectivity to HQ; however, the role of external sites
was dependent on the type of solvent used. While HQ was the pre-
dominant product in the pores of TS-1, the main product formed on
the external surface was CT in acetone solvent and HQ in protic sol-
vents [37].
In our case, the ratio between the two isomers, formed during
benzene hydroxylation, changed a great deal depending on the
type of solvents used (see, for instance, Figs. 1, 2 and 7): an obser-
vation in line with literature findings. However, the plot in Fig. 4,
reporting the effect of reaction time on the selectivity to products
with only methanol solvent (apart from the water contained in the
diluted HP solution), shows that the selectivity to CT was twice
that of HQ over the entire reaction time range examined. This sup-
ports the absence of any shape-selectivity effect, as well as the
hypothesis that the two diphenol isomers are formed mainly as
kinetically secondary products on the external Ti sites.
With regard to the behavior of HQ and CT, results shown in
Figs. 1 and 2 evidence similar behaviors for the two isomers in re-
sponse to changes in solvent composition, thus suggesting that the
formation of both compounds occurs over the same types of sites,
either intracrystalline or intercrystalline. After phenol has formed,
it may either diffuse into the porous structure and therein react – a
phenomenon especially favored in water solvent [37] – or more
likely react over the external sites, or even in the bulk liquid phase
by thermal activation, finally yielding the two dihydroxylated iso-
mers. In fact, if the consecutive oxidation of phenol to HQ occurred
inside pores, the latter should be oxidized quickly up to BQ; as a
matter of fact, our results clearly indicate that there is no kinetic
relationship between HQ and BQ. It is worth noting that a similar
behavior (i.e. primary formation of BQ, secondary formation of
both HQ and CT) was recently reported for the same reaction when
catalyzed by TS-1 PQ [28], with an initial selectivity to phenol and
BQ higher than zero; however, this result was not discussed in that
paper.
An important implication for this result is that the only intra-
crystalline product of the reaction is BQ (and maybe also phenol),
whereas CT and HQ are likely to be intercrystalline products.
Therefore, it is expected that the crystal size of TS-1 may affect
the ratio between phenol and BQ, because smaller crystals will lead
to a higher initial selectivity to phenol, and a lower selectivity to
BQ. This may be due to the fact that either a smaller intracrystal-
line residence time would limit any further oxidation of phenol
to HQ and then to BQ (in fact, a shorter mean path for the counter-
diffusion of phenol toward the bulk liquid phase would make any
consecutive phenol oxidation less likely), or the contribution of
external Ti sites would be increased with smaller crystallites: an
event that might foster the formation of phenol with respect to
BQ if the former compound is formed over external Ti sites.
In the case of the TS-1A sample used for these experiments –
which is made of particles of ꢀ25-micron average size and pre-
pared by spray-drying and agglomeration of very small crystallites
(of a size smaller than 0.2 micron) – the mean path of products is
evidently very long: something that – in principle – should en-
hance the formation of BQ, with a lower primary selectivity to
the desired phenol. These results anticipate that one key event
for increasing the selectivity to phenol is shortening the mean path
within pores. On the other hand, the formation of the kinetically
secondary products, HQ and CT, is expected to depend mainly on
both benzene conversion and the characteristics of external surface
sites.
Our data demonstrate that the solvent affects both conversion
and selectivity to products in benzene hydroxylation. Literature
is in agreement on the fact that the nature of the solvent is essen-
tial and is also the major parameter affecting catalytic performance
[1,2]. Solvents capable of homogenizing the hydrophobic substrate
and the aqueous HP, such as acetone, acetonitrile or t-butanol,
make it possible to achieve optimum performances. In general,
more polar solvents have little interaction with the hydrophobic
surface of TS-1, especially in samples having fewer defects and sur-
face hydroxyl groups. Therefore, with polar solvents the pores are
preferentially filled with apolar hydrocarbons (e.g. benzene). The
opposite is true with less polar solvents (e.g. t-butanol); the parti-
tion coefficient is modified, and this clearly has implications on the
reaction rate. Another important point is the interaction between
the solvent and the active sites. For example, it is well known that
methanol itself, although showing little affinity for the TS-1 due to
its predominantly polar characteristics, may interact with TiIV and
contribute to the formation of a pentacoordinated site in which the
positive charge on the O atom of the hydroperoxo species forms by
interaction with the methanol proton [2].
With regard to the effect of having either a biphasic or a tri-
phasic system, Kumar et al. [24–27] reported that the presence
of a triphasic system (two liquid phases and the solid catalyst)
yields an excellent performance. Other authors, however, did
not confirm this result [22,28]. Our results indeed show that in
the conditions examined, the presence of a triphasic system has
negative effects on the catalytic behavior (Fig. 1). As for the reac-
tion in the biphasic zone, with low water content, the yield of
phenol was low; benzene conversion and yield to phenol in-
creased when the molar fraction of methanol was decreased
within the biphasic zone. The same effect is observed in the case
of experiments reported in Fig. 2, carried out under conditions
leading to a biphasic system; benzene conversion increased when
the water-to-benzene ratio was increased; however, for a similar
interval of variation in this feed ratio (i.e. from 1 to 3), there was
only a lower increase in conversion compared to that observed
for the tests in Fig. 1.
4.2. The role of solvent: (a) water vs methanol and (b) sulfolane as
methanol co-solvent, in relation to the reaction network
In the case of phenol hydroxylation into diphenols, phenol
transformation was found to be governed by intracrystalline diffu-
sion in large crystals of TS-1 [36]; a strong diffusion-limited trans-
formation of phenol supports our hypothesis that with larger
crystallite size the consecutive transformation of HQ into BQ has
the time to occur, finally yielding BQ as a primary product of ben-
zene oxidation. The same authors excluded a major influence of
external surface activity. Tuel et al. [46] concluded that in phenol
hydroxylation the product ratio of HQ to CT is mainly controlled
by the external surface activity; HQ and CT were thought to be
the preferred reaction products in the micropores of TS-1 and on
the external surface Ti sites, respectively. Wilkenhöner et al. [37]
investigated the effect of solvent and of TS-1 crystallite size on
the catalytic behavior in phenol hydroxylation. The presence of dif-
The increase in benzene conversion, observed in the biphasic
zone for increasing amounts of water, was due to the fact that a
more polar solution (because of an higher molar fraction of water)
may increase the concentration of benzene within pores, an effect
attributable to the zeolite property of performing a selection of mol-
ecules from the liquid phase by acting as an additional solvent for
the reaction [47]. This leads to a higher rate of benzene conversion

D. Barbera et al. / Journal of Catalysis 275 (2010) 158–169
167
and phenol formation (provided HP does not become the limiting
reactant). This hypothesis is confirmed by the results of competitive
adsorption tests reported in Table 2. The concentration of benzene
and phenol inside TS-1 is affected by the nature of solvent used;
with more polar solvent mixtures (either methanol/water or meth-
anol/sulfolane), benzene and phenol are equally retained inside
TS-1 pores, whereas when only methanol is used, the interaction
of TS-1 with phenol predominates over that of benzene. In other
words, with more polar solvent mixtures, benzene has more affinity
for TS-1 than for the liquid bulk phase, whereas with a less polar sol-
vent, benzene does not compete so much with phenol for diffusion
inside the TS-1 pores. On the other hand, phenol is soluble both in
more polar methanol/water(sulfolane) mixtures and in methanol
only.
Sulfolane co-solvent also shows a significant effect on catalytic
behavior. In literature, it is reported that the coordination of
phenol by sulfolane inside the TS-1 pores limits its consecutive
hydroxylation to CT and HQ [1,3,22]. Authors from the ENI re-
search team hypothesize that the increased selectivity may be
due to the formation of a sterically hindered species that cannot
enter the TS-1 pores, thus allowing phenol to remain relatively
protected against further oxidation [22,23]. The hypothesis of a
bulky complex with sulfolane would fit perfectly with our results
and with the hypothesis of BQ as the only intracrystalline by-
product, in the case sulfolane had an effect only on CT and HQ
(formed on Ti intercrystalline sites), and not on BQ (formed inside
the TS-1 microporosity). In other words, we hypothesize that the
coordination effect of sulfolane plays out of TS-1 pores, and not
inside them, thus keeping phenol from consecutive oxidation to
either CT or HQ. This is not so much due to steric reasons, but
rather probably to the fact that sulfolane is a polar molecule (not-
withstanding the apolar C4 part) which is soluble in water, and its
access in TS-1 hydrophobic pores is probably not allowed, at least
as long as less polar molecules (e.g. methanol) are present in the
reaction medium. Less polar molecules have preferential access to
the TS-1 pores, and then sulfolane would remain in the bulk li-
quid. Results shown in Fig. 5 – obtained under conditions inhib-
iting the access of sulfolane to the TS-1 porous system, because
concentrated HP was used, and the amount of methanol solvent
was largely predominant over the amount of water - support
our hypothesis. The main effect of sulfolane co-solvent in metha-
nol/sulfolane mixtures was that of decreasing the selectivity to
diphenols that went down to almost zero, whereas the effect on
BQ selectivity was minimal.
On the other hand, when sulfolane was used as co-solvent in
diluted HP solutions (Fig. 6), with the amount of water solvent
predominating over the amount of methanol, e.g. with diluted
HP solutions (while the HP-to-benzene ratio was constant), under
conditions more favorable for the access of sulfolane to pores, the
effect of sulfolane was quite different from that obtained with
negligible amount of water solvent. In fact, there was a remark-
able decrease in BQ selectivity, whereas there was no effect on
selectivity to diphenols. This provides a clear indication that in
reaction conditions which are more conducive to the diffusion
of sulfolane into the TS-1 pores, sulfolane yields an enhancement
effect on the selectivity to the by-products formed within pores
(BQ), but a less marked effect on those formed outside the TS-1
crystallites.
As summarized in Fig. 7, the relative amounts of sulfolane,
methanol and water have a profound influence on catalytic behav-
ior; higher water-to-benzene ratio (at low HP concentration) leads
to higher values of benzene conversion. The amount of water in the
reaction medium influences the amount of both methanol and sul-
folane inside TS-1 pores; therefore, depending on the amount of
water used, sulfolane plays its protecting role on phenol either in
the bulk liquid phase or inside TS-1 pores.
4.3. How prepare a more selective TS-1 catalyst: the role of crystallite
size and of nature of Ti sites
Data reported in Fig. 8 show that the crystallite size is an impor-
tant parameter affecting catalytic behavior. It is worth noting that
all these samples showed a lower activity than that of sample TS-
1A. In conditions that, with the latter catalyst, led to a benzene
conversion of 6.4%, with 75% selectivity to phenol (selectivity to
BQ 12%, to CT + HQ 13%) (see Fig. 1 bottom), all samples prepared
with the modified procedure resulted in less than 4.5% benzene
conversion. However, the low conversion achieved makes it possi-
ble to compare samples in conditions that led to the exclusive for-
mation of BQ and phenol, with nil formation of CT + HQ. In fact, in
the conversion range of all our samples, the only reactions occur-
ring are the parallel ones leading to phenol and BQ, with nil contri-
bution of consecutive reactions leading to CT and HQ (the
selectivity to these latter compounds was nil). In other words, in
the conversion range examined, the selectivity to phenol (and to
BQ as well) was not a function of benzene conversion, but only
of the ratio between the rates of the two kinetically parallel
reactions:
1. Benzene ? phenol
2. Benzene ? (phenol ? HQ ?) BQ
in which reaction 2 is made up of chemically consecutive steps, but
appears as a direct, single-step reaction from a kinetic standpoint.
This allows comparing the selectivity to the primary products in
function of the crystallite size. The inverse relationship found be-
tween the latter and the selectivity to phenol (Fig. 8) is in agree-
ment with our hypothesis, that the counterdiffusion path length
of phenol toward the bulk phase greatly affects the probability that
phenol may undergo the chemically consecutive transformation
into HQ and then BQ; the longer the path, the higher the selectivity
to BQ and the lower to phenol.
Tests made by reacting phenol confirm this hypothesis (Table 3).
TS-1D, the sample showing the higher initial selectivity to phenol
and nil formation of BQ in benzene hydroxylation, produced the
smallest amount of BQ when the reaction was carried out starting
from phenol. Notably, the same catalyst also produced a negligible
amount of diphenols but rather significant amounts of tar. There-
fore, because of its lower crystallite size, and despite the relatively
high amount of Ti, the rate of phenol transformation to HQ and BQ
inside the TS-1D pores is low, because the effective intracrystalline
contact time is low if compared to TS-1 samples having greater
crystallite size. On the other hand, reactions occurring on external
sites lead to tar formation.
In samples with the largest crystallite size, the formation of BQ
by the two-step intracrystalline oxidation of phenol is much more
significant than with TS-1D. On the other hand, samples TS-1B, TS-
1C, and TS-1E show more important differences with regard to the
formation of diphenols and tar. This supports the hypothesis that
the formation of BQ (not very different in the three samples) occurs
on different sites than the formation of the other by-products.
The reactivity of TS-1E appears anomalous. On the one hand,
the yield to BQ is the same as that of TS-1B, which is perfectly in
line with the similar crystallite size of the two samples (which also
show similar selectivity to phenol and BQ in benzene hydroxyl-
ation, Fig. 8). On the other hand, the selectivity to CT and HQ is
nil (as it is also for TS-1D), and the yield to tar is rather significant.
IR spectroscopy has shown that TS-1E, the sample with the
highest titanium content, has also the highest amount of hydroxyls
nests: according to the literature [48], the intensity of the silanol
nests band in TS-1 should decrease with an increasing amount of
framework titanium, since Ti insertion into the TS-1 network is
accompanied by the progressive reduction of Si vacancies, i.e. hy-

168
D. Barbera et al. / Journal of Catalysis 275 (2010) 158–169
droxyl nests, with formation of both [TiO4] and [(SiO)3TiOH] sites.
Previous EXAFS studies also indicated that in TS-1 samples pre-
pared by traditional synthesis the fraction of defective [(SiO)3TiOH]
sites progressively decreases upon increasing Ti content [48]. In
the present case, however, the modified synthesis procedure
adopted could lead to slightly different titanium environments
along with different hydroxyl populations, as well. In fact, although
TS-1E has the highest titanium content, it shows the most intense
3600 cmꢁ1 band; that is, it contains a higher defective hydroxyl
population. On the other hand, DR UV–Vis spectra showed that
the latter sample contains a higher amount of titanium species in
coordination higher-than-four (270 nm band). In conventional
TS-1, such species develop by treatment of TS-1 at 80 °C with an
aqueous solution of NH4HF2 and H2O2; this treatment leads to
the extraction of some tetrahedral Ti sites in the form of
(NH4)3Ti(O2)F5 [22,23]. In the present case, the use of both a
TPAOH/Si ratio higher than 1 and a large amount of Ti loading
(conditions met in samples TS-1D and TS-1E) may facilitate the
development of this peculiar Ti species even during the synthesis
of TS-1. It is worth noting that Ti loadings higher than 3, but low
TPAOH/Si preparation ratios, usually lead to the additional forma-
tion of anatase and not of this peculiar Ti species.
Therefore, unlike TS-1 samples prepared by traditional methods
[22,23], in the present case species absorbing at 270 nm seem not
to be very reactive and should likely be assigned to partly extra-
framework TiOx species. Such a result is in agreement with the
abundance of hydroxyl nests in sample TS-1E, due to the fact that
not all titanium is entering the framework: such hydroxyl nests be-
have indeed as hydrophilic species inside channels and, therefore,
they should affect the diffusion of polar molecules such as, for
example, phenol formed during the reaction under study, thus
increasing their residence time inside channels and therefore con-
tributing to the decreased selectivity of the TS-1E catalyst (Fig. 8b).
Adsorption of ammonia, as followed by IR spectroscopy, showed
that although TS-1E sample has the highest nominal titanium con-
tent, it does not show the most intense bands related to titanium,
thus confirming that in the latter sample, in addition to framework
species, titanium occurs also as less reactive extra-framework tita-
nium oxide (also detected by DR UV–Vis spectroscopy). This not
only explains the relatively low activity of TS-1E (Fig. 8a), despite
the presence of both a large amount of titanium and a relatively
low amount of octahedral TiIV, but also provides an interpretation
for the results obtained during phenol hydroxylation. It is possible
that the predominant highly coordinated Ti sites and the lower
concentration of tetrahedral TiIV sites on the external of TS-1E crys-
tallites may hinder the formation of diphenols and facilitate the
formation of tar as a consecutive reaction on the BQ formed.
bulk liquid phase by a simple thermally activated reaction.
Although it is a kinetically primary product, benzoquinone is
chemically consecutive to phenol; therefore, the likelihood of
achieving this by-product depends on the counterdiffusion path
length of phenol inside TS-1 pores. In fact, the primary selectivity
to phenol, and to benzoquinone as well, was found to be a clear
function of the average crystallite size of TS-1. The reactivity
behavior was also affected by the solvent used, as shown in tests
carried out by using different methanol-to-water ratios, in biphasic
conditions. The development of a triphasic, water-rich solvent sys-
tem led to a remarkable lowering of the selectivity to phenol, with
the formation of large amounts of both diphenols and tar. Lastly,
we produced experimental evidences that the phenol-protecting
role of sulfolane co-solvent – the use of which is known to cause
a higher selectivity to phenol – was different depending on the
conditions used. When the solvent used inhibits the diffusion of
sulfolane inside TS-1 pores, the protective effect on phenol limits
the occurrence of the consecutive hydroxylation to CT and HQ.
Conversely, when the conditions used facilitated the diffusion of
sulfolane inside pores, the protecting effect inhibited the intracrys-
talline oxidation of phenol to benzoquinone.
Lastly, a modified synthesis procedure using a large excess of
TPAOH allows the control of both the crystallite size of TS-1 and
the nature of the Ti species. In fact, the use of high Ti/Si atomic ra-
tios and high TPAOH/Si molar ratios led to the formation of Ti sites
with TiIV sitting in coordination higher-than-four, most probably as
extra-framework species. Correspondingly, a higher population of
defective hydroxyls was observed: the latter species may indeed
be responsible for a higher affinity of the catalysts for polar mole-
cules, thus affecting the product distribution, with a lower initial
selectivity to phenol, but did not contribute to benzene conversion.
Acknowledgments
The Ministero dell’Università e della Ricerca (MIUR) is acknowl-
edged for sponsoring this research activity (PRIN 2006), as well as
for the PhD Grant to SG (Progetto Giovani).
References
[1] M. Ricci, D. Bianchi, R. Bartolo, in: F. Cavani, G. Centi, S. Perathoner, F. Trifirò
(Eds.), Sustainable Industrial Chemistry. Principles, Tools and Industrial
Examples, Wiley-VCH, Weinheim, 2009, pp. 507–528.
[2] M.G. Clerici, Top. Catal. 15 (2001) 257–263.
[3] G. Centi, S. Perathoner, Catal. Today 143 (2009) 145–150.
[4] C. Perego, A. Carati, P. Ingallina, M.A. Mantegazza, G. Bellussi, Appl. Catal. A 221
(2001) 63–72.
[5] M.G. Clerici, Catal. Today 41 (1998) 351–364.
[6] D. Dumitriu, R. Bârjega, L. Frunza, D. Macovei, T. Hu, Y. Xie, V.I. Pârvulescu, S.
Kaliaguine, J. Catal. 219 (2003) 337–351.
[7] X. Qi, J. Li, T. Ji, Y. Wang, L. Feng, Y. Zhu, X. Fan, C. Zhang, Micropor. Mesopor.
Mater. 122 (2009) 36–41.
[8] D. Bianchi, M. Bertoli, R. Tassinari, M. Ricci, R. Vignola, J. Mol. Catal. A 200
(2003) 111–116.
5. Conclusions
This work reports the identification of the reaction network in
benzene hydroxylation to phenol with hydrogen peroxide, cata-
lyzed by Titanium–silicalite (TS-1), as well as the effect of the crys-
tallite size on the primary selectivity to phenol. Benzene is
hydroxylated to phenol inside TS-1 pores, but phenol undergoes
consecutive oxidation to hydroquinone; hydroquinone, however,
is rapidly transformed into benzoquinone inside the restricted
environment of the silicalite, before it may conterdiffuse toward
the bulk liquid phase. Phenol formed on the external surface of
Ti sites may undergo consecutive hydroxylation to diphenols
(which are kinetically secondary products); hydroquinone, how-
ever, is not so efficiently transformed into benzoquinone as it is
when formed in the TS-1 porosity. Therefore, phenol and benzoqui-
none are the two only kinetically primary products of benzene oxi-
dation. Phenol may then undergo consecutive oxidation also in the
[9] J.-S. Choi, T.-H. Kim, K.-Y. Choo, J.-S. Sung, M.B. Saidutta, S.-O. Ryu, S.-D. Song, B.
Ramachandra, Y.-W. Rhee, Appl. Catal. A 290 (2005) 1–8.
[10] N.K. Renuka, J. Mol. Catal. A 316 (2010) 126–130.
[11] F. Gao, R. Hua, Appl. Catal. A 270 (2004) 223–226.
[12] X. Gao, J. Xu, Appl. Clay Sci. 33 (2006) 1–6.
[13] T.K. Si, K. Chowdhury, M. Mukherjee, D.C. Bera, R. Bhattacharyya, J. Mol. Catal.
A 219 (2004) 241–247.
[14] S. Feng, S. Pei, B. Yue, L. Ye, L. Qian, H. He, Catal. Lett. 131 (2009) 458–462.
[15] N.A. Alekar, V. Indira, S.B. Halligudi, D. Srinivas, S. Gopinathan, C. Gopinathan, J.
Mol. Catal. A 164 (2000) 181–189.
[16] Y. Leng, H. Ge, C. Zhou, J. Wang, Chem. Eng. J. 145 (2008) 335–339.
[17] Y. Zhu, Y. Dong, L. Zhao, F. Yuan, J. Mol. Catal. A 315 (2010) 205–212.
[18] J.K. Joseph, S. Singhal, S.L. Jain, R. Sivakumaran, B. Kumar, B. Sain, Catal. Today
141 (2009) 211–214.
[19] J. He, W.-P. Xu, D.G. Evans, X. Duan, C.-y. Li, Micropor. Mesopor. Mater. 44-45
(2001) 581–586.
[20] A. Keshavaraja, V. Ramaswamy, H.S. Soni, A.V. Ramaswamy, P. Ratnasamy, J.
Catal. 197 (1995) 501–511.
[21] D. Bianchi, R. D’Aloisio, R. Bortolo, M. Ricci, Appl. Catal. A 327 (2007) 295–299.

D. Barbera et al. / Journal of Catalysis 275 (2010) 158–169
169
[22] D. Bianchi, L. Balducci, R. Bortolo, R. D’Aloisio, M. Ricci, G. Spanò, R. Tassinari, C.
Tonini, R. Ungarelli, Adv. Synth. Catal. 349 (2007) 979–986.
[23] L. Balducci, D. Bianchi, R. Bortolo, R. D’Aloisio, M. Ricci, R. Tassinari, R.
Ungarelli, Angew. Chem. Int. Ed. 42 (2003) 4937–4940.
[24] P. Mukherjee, A. Bhaumik, R. Kumar, Ind. Eng. Chem. Res. 46 (2007) 8657–
8664.
[25] R. Kumar, P. Mukherjee, A. Bhaumik, Catal. Today 49 (1999) 185–191.
[26] A. Bhaumik, P. Mukherjee, R. Kumar, J. Catal. 178 (1998) 101–107.
[27] R. Kumar, A. Bhaumik, Micropor. Mesopor. Mater. 21 (1998) 497–504.
[28] P. Chammingkwan, W.F. Hoelderich, T. Mongkhonsi, P. Kanchanawanichakul,
Appl. Catal. A 352 (2009) 1–9.
[37] U. Wilkenhöner, G. Langhendries, F. van Laar, G.V. Baron, D.W. Gammon, P.A.
Jacobs, E. van Steen, J. Catal. 203 (2001) 201–212.
[38] S. Bordiga, A. Damin, F. Bonino, G. Ricchiardi, A. Zecchina, R. Tagliapietra, C.
Lamberti, Phys. Chem. Chem. Phys. 5 (2003) 4390–4393.
[39] M. Anpo, M. Takeuchi, J. Catal. 216 (2003) 505–516.
[40] F.-Z. Zhang, X.-W. Guo, X.-S. Wang, G. Li, J.-C. Zhou, J.-Q. Yu, C. Li, Catal. Lett. 2
(2001) 235.
[41] E. Duprey, P. Beaunier, M.-A. Springuel-Huet, F. Bozon-Verduraz, J. Fraissard,
J.-M. Manoli, J.-M. Brégeault, J. Catal. 165 (1997) 22–32.
[42] G. Ricchiardi, A. Damin, S. Bordiga, C. Lamberti, G. Spanò, F. Rivetti, A. Zecchina,
J. Am. Chem. Soc. 123 (2001) 11409–11419.
[29] J.F. Bengoa, N.G. Gallegos, S.G. Marchetti, A.M. Alvarez, M.V. Cagnoli, A.A.
Yeramián, Micropor. Mesopor. Mater. 24 (1998) 163–172.
[30] M. Taramasso, G. Perego, B. Notari, US Patent 4410,501, 1983 (assigned to
EniChem).
[31] T. Yokoi, P. Wu, T. Tatsumi, Catal. Commun. 4 (2003) 11–15.
[32] N. Ma, Z. Ma, Y. Yue, Z. Gao, J. Mol. Catal. A 184 (2002) 361–370.
[33] G. Leofanti, F. Genoni, M. Padovan, G. Petrini, G. Trezza, A. Zecchina, Stud. Surf.
Sci. Catal. 62 (1991) 553–564.
[43] S. Bordiga, I. Roggero, P. Ugliengo, A. Zecchina, V. Bolis, G. Artioli, R. Buzzoni, G.
Marra, F. Rivetti, G. Spanò, C. Lamberti, J. Chem Soc. Dalton Trans. (2000)
3921–3929.
[44] B. Bonelli, L. Forni, A. Aloise, J.B. Nagy, G. Fornasari, E. Garrone, A. Gedeon, G.
Giordano, F. Trifirò, Micropor. Mesopor. Mater. 101 (2007) 153.
[45] B. Bonelli, M. Cozzolino, R. Tesser, M. Di Serio, M. Piumetti, E. Garrone, E.
Santacesaria, J. Catal. 246 (2007) 293–300.
[46] A. Tuel, S. Moussa-Khouzami, Y. Ben Taarit, C. Naccache, J. Mol. Catal. 68 (1991)
45.
[47] E.G. Derouane, J. Mol. Catal. A 134 (1998) 29–45.
[48] S. Bordiga, F. Bonino, A. Damin, C. Lamberti, Phys. Chem. Chem. Phys. 9 (2007)
4854–4878.
[34] I. Halasz, M. Agarwal, E. Senderov, B. Marcus, App. Catal. A 241 (2003).
[35] C. Lamberti, S. Bordiga, A. Zecchina, G. Artioli, G. Marra, G. Spanò, J. Am. Chem.
Soc. 123 (2001) 2204–2212.
[36] A.J.H.P. van der Pol, A.J. Verduyn, J.H.C. van Hooff, Appl. Catal. A 92 (1992) 113-.