Ti(IV)/trialkanolamine catalytic polymeric membranes: Preparation, characterization, and use in oxygen transfer reactions

Article abstract of DOI:10.1016/j.jcat.2005.11.044

New heterogeneous oxidation catalysts have been obtained by entrapping Ti(IV)/trialkanolamine complexes within polymeric membranes. The catalytic membranes were prepared by a nonsolvent-induced phase-inversion technique. Three polymers, polyvinylidene fluoride (PVDF), a modified polyetherketone (PEEK-WC), and polyacrylonitrile (PAN), with different functional groups and chemical-physical properties, were used to tune the reactivity of the catalytic polymeric membranes in the stereoselective oxidation to sulfoxide and chemoselective oxidation of secondary amines to nitrones by alkyl hidroperoxides. The chemical-physical analysis of the new catalytic membranes was carried out by SEM, EDX, IR, CAM, and XPS techniques. In particular, the XPS spectra showed a very interesting orientation effect of PVDF membranes on the entrapped Ti(IV)/trialkanolamine complex. PVDF-based catalytic membranes gave the best results, affording products in shorter reaction times, higher yields, and better selectivity compared with the corresponding homogeneous systems. The membranes can be recycled up to five runs with no loss of activity.

Full text of DOI:10.1016/j.jcat.2005.11.044

Journal of Catalysis 238 (2006) 221–231
www.elsevier.com/locate/jcat
Ti(IV)/trialkanolamine catalytic polymeric membranes:
Preparation, characterization, and use in oxygen transfer reactions
a,∗
a
b
b
Maria Giovanna Buonomenna , Enrico Drioli , Renzo Bertoncello , Laura Milanese ,
b
a,b
b
Leonard J. Prins , Paolo Scrimin , Giulia Licini
a
ITM-CNR, via Bucci 17/C, I-87030 Arcavacata Di Rende, CS, Italy
b
Dipartimento Scienze Chimiche, Università di Padova, ITM-CNR, sezione di Padova, via Marzolo 1, I-35131, Padova, Italy
Received 10 October 2005; revised 29 November 2005; accepted 29 November 2005
Abstract
New heterogeneous oxidation catalysts have been obtained by entrapping Ti(IV)/trialkanolamine complexes within polymeric membranes.
The catalytic membranes were prepared by a nonsolvent-induced phase-inversion technique. Three polymers, polyvinylidene fluoride (PVDF),
a modified polyetherketone (PEEK-WC), and polyacrylonitrile (PAN), with different functional groups and chemical–physical properties, were
used to tune the reactivity of the catalytic polymeric membranes in the stereoselective oxidation to sulfoxide and chemoselective oxidation of
secondary amines to nitrones by alkyl hidroperoxides. The chemical–physical analysis of the new catalytic membranes was carried out by SEM,
EDX, IR, CAM, and XPS techniques. In particular, the XPS spectra showed a very interesting orientation effect of PVDF membranes on the
entrapped Ti(IV)/trialkanolamine complex. PVDF-based catalytic membranes gave the best results, affording products in shorter reaction times,
higher yields, and better selectivity compared with the corresponding homogeneous systems. The membranes can be recycled up to five runs with
no loss of activity.
©
2005 Elsevier Inc. All rights reserved.
Keywords: Catalytic polymeric membranes; Ti(IV)/trialkanolamines complex; Polyvinylidene fluoride
1
. Introduction
in polymeric membranes represent a situation in which single
site heterogeneous catalysts (SSHCs) may be designed.
At low temperature (from room temperature up to about
Most industrial catalysts are high-surface area solids onto
◦
which an active component is dispersed in the form of very
small particles. To produce selectivities approaching 100%, het-
erogeneous catalysts will need to share a key characteristic with
the homogeneous and biological catalyst: they must present
uniform sites and uniform contact patterns as an additional cat-
alyst engineering constraint [1,2].
The capability to produce a well-defined, porous matrix that
can serve as catalyst support from a wide variety of inorganic
materials is contributing to the production of single-site cata-
lysts in which all of the active sites closely resemble each other
150 C), polymeric membranes have some advantages over the
most expensive inorganic membranes made from ceramic or
metals. Most of the polymers can be easily manufactured in
different shapes (e.g., hollow, spiral wound, flat sheet); they are
elastic, resist fatigue, have satisfactory diffusion and sorption
coefficients, and can be produced with incorporated catalysts
as nanosized dispersed metallic clusters [8,9], zeolites and acti-
vated carbons [10], or metallic complexes [11–13].
Polydimethylsiloxane (PDMS) is the most widely used poly-
mer system for catalytic polymeric membranes, combining high
thermal and mechanical stability with chemical resistance [14].
This dense hydrophobic elastomer has been applied success-
fully in the immobilization of oxygenation catalysts such as
metallo porphyrins [15], epoxidation catalysts such as Mn-salen
[
3–7]. From this perspective, homogeneous catalysts entrapped
*
Corresponding author.
E-mail address: mg.buonomenna@itm.cnr.it (M.G. Buonomenna).
0021-9517/$ – see front matter © 2005 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2005.11.044

222
M.G. Buonomenna et al. / Journal of Catalysis 238 (2006) 221–231
Scheme 1.
Scheme 2.
[
16,17], and Mn-bipyridines in zeolite [18]. The hydropho-
2
. Experimental
bicity of the polymer favors the sorption of apolar compo-
nents, resulting in the suppression of the oxidant decomposi-
tion.
2.1. Chemicals
This paper describes and discusses the preparation of new
catalytic membranes incorporating Ti(IV) complex and their
successful use in two model reactions, the sulfoxidation of sul-
fides to sulfoxides and the secondary amines oxidation to ni-
trones.
Chiral sulfoxides are an important class of compounds that
have increasing applications as chiral auxiliaries in asymmet-
ric synthesis [19] and recently as biological active molecules
Commercially available reagents and solvents were used
as received without further purification for the preparation of
polymeric catalytic membranes. PEEKWC was supplied by the
Chanchung Institute of Applied Chemistry, Academia Sinica.
PVDF 20 10 Solef was supplied by Solvay, Inc. PAN was sup-
plied from SNIA. Ti-2 was prepared as described previously
[
23]. Cumene hydroperoxide (Fluka) was stored under molecu-
◦
[
20]. The oxidation of amines to nitrones is a very appeal-
lar sieves at 0 C. 1,2-dichloroethane was washed three times
with 10% H2SO4 and with several time until pH reached 7,
dried overnight over CaCl distilled over P O , and stored over
ing reaction because of the relevance of the metabolic fate of
these compounds in vivo and the synthetic interest of the reac-
tion products. Important enzymes are involved in the metabolic
oxidation of amines. For example, flavin mono-oxygenase and
related compounds, such as 5-ethyl-4a-hydro-peroxyflavin, ox-
idize secondary amines to nitrones [21]. The simulation of the
activity of these enzymes using metal complexes is of great
interest, potentially providing mimetic methods for catalytic
oxidation of the amines [22].
We have recently reported that tetradentate alkoxide lig-
ands, namely C3-symmetric trialkanolamines 1 (Scheme 1),
provide very stable Ti(IV) homogeneous complexes [23,24].
In the presence of alkyl hydroperoxides, such species are able
to catalyze the asymmetric sulfoxidation of alkyl aryl sulfides
with ee’s up to 84% and with unprecedented catalytic effi-
ciency, reaching turnover numbers (TONs) of 1000 [25] as
well as the oxidation of secondary amines to nitrones with
high chemoselectivities, quantitative yields, and TONs up to
2
2
5
molecular sieves.
2
.2. Preparation of catalytic polymeric membranes
Flat-sheet PVDF-, PAN-, and PEEKWC-based membranes
were prepared by phase-inversion process induced by nonsol-
vent. The membranes were prepared using dimethylacetam-
mide (DMA) as a solvent and distilled water as a nonsolvent.
For each polymer, a solution (10.0 wt%) was prepared by
dissolving the polymer in DMA (74 wt%) by magnetic stir-
ring at room temperature; then Ti(IV)/trialkanolamine complex
(
16 wt%) was added, and the resulting solution was stirred for
an additional 24 h. The solution was cast on a glass plate by set-
ting the knife gap at 250 µm, and the cast film was immediately
immersed for 10 min in a coagulation bath containing distilled
◦
water at 15 C. The membranes were removed from the coagu-
1
400 [26]. Besides the fact that in both cases the oxygen trans-
lation bath and stored in distilled water for 24 h to completely
fer process affords corresponding products with high chemical
yields and good selectivities, the Ti(IV) catalyst was shown to
be robust under the reaction conditions, which require the pres-
ence of large quantities of alkylperoxide [23]. In addition, it
affords a stoichiometric amount of water in the case of sec-
ondary amine oxidation [24]. Therefore, titanatrane complex 2
seems to be a good candidate for preparing Ti(IV) polymeric
catalytic membranes to be used in selective oxygen trans-
fer reactions. They should survive in the conditions required
for membrane synthesis because of their intrinsic stability.
The membranes are prepared using polymers with differ-
ent hydrophobicity/hydrophilicity properties: a fluoropoly-
mer, polyvinylidene fluoride (PVDF), modified polyether-
etherketone (PEEK-WC) and hydrophilic polyacrylonitrile
remove DMA. Then the membranes were dried in an oven at
◦
6
0 C under vacuum for 24 h.
An additional preparation method was used to prepare
PVDF-based catalytic membranes using a mixture of two sol-
vents, DMA and acetone; and water as a nonsolvent, follow-
ing the procedure reported by Pintauro et al. [27]. A solution
(
(
10.0 wt%) was prepared by dissolving the polymer in DMA
20 wt%) and Acetone (54 wt%) by magnetic stirring at room
temperature; then Ti(IV)/trialkanolamine complex (16 wt%)
was added and the resulting solution was left under stirring for
◦
additional 24 h at 60 C. The solution was cast on a glass plate
by setting the knife gap at 250 µm, and the cast film was kept
at air for 3 h. The membrane was immersed in a coagulation
◦
bath containing distilled water at 15 C. After the membranes
(
PAN) (Scheme 2). All of these polymers are chemically sta-
were removed from the coagulation bath, the procedure above
reported was followed.
ble in the reaction conditions used.

M.G. Buonomenna et al. / Journal of Catalysis 238 (2006) 221–231
223
2
.3. Phase diagrams of the Ti-2/water/DMA/PVDF system
Table 1
Surface tension parameters for different test liquids according to van Oss,
◦
LW
−
◦
Chaudhury, and Good a 20 C: γ
= Lifshitz–van der Waals component, γ
,
Phase diagrams, at room temperature (ca. 20 C), were de-
+
γ
= electron donor or proton acceptor (basicity or acidity parameters) com-
termined by visual observation of the cloud point during slow
addition of the nonsolvent to the stirred polymer solution con-
taining Ti-2.
ponents of the liquid surface tension γ for different test liquids [28,29]
S
LW
AB
−
+
γ
Test liquid
γ
γ
γ
γ
2
2
2
2
2
(
mJ/m ) (mJ/m ) (mJ/m ) (mJ/m ) (mJ/m )
Water
Glycerol
Diiodomethane
72.8
64
50.8
21.8
34
50.8
51.0
30
0
25.5
3.92
−
25.5
57.4
−
2
.4. Membrane characterisation
The formed membranes were characterized by the following
methods:
Once the γ LW of membrane surface has been measured, it
ab
+
is possible to calculate the other components (γ , γ , and
(
(
1) The morphology of the membranes was evaluated by
means of SEM at 20 kV (Cambridge Instruments Stere-
oscan 360).
2) The success of Ti-2 entrapment and its uniform dispersion
in the different polymeric membranes was evaluated by
EDX and BSE techniques. EDX microanalyses were per-
formed using a Philips EDAX analysis system.
−
γ ) using two polar liquids, as glycerol and water,
γl(1 + cosϑ)
ꢀꢀ
ꢁ
ꢀ
ꢁ
ꢁ
LW LW 0.5
+
− 0.5
−
+ 0.5
=
2 γ
γ
+ (γ γ ) + γ γ
s
l
s
l
s
l
and
ab
s
+
− 0.5
.
γ
= 2(γ γ )
(
3) X-Ray diffractograms were collected by positioning the
PVDF films on a sample holder mounted on the Philips PW
s
s
The values of surface tension components of the liquids
used in the experiments were taken from the literature
1
820 powder diffractometer, using Cu-Kα radiation, with
the generator working at 40 kV and 40 mA. Intensities were
measured in the range 15 < 2θ < 55 , typically with step
scans of 0.01 (10 s/step). Only the region 15 < 2θ < 35
[
28,29] and are summarized in Table 1.
◦
◦
◦
◦
◦
was analyzed (the outermost region being quite noisy and
featureless for all the samples).
2
.5. Reaction procedures
(
4) FTIR spectra of the prepared films were recorded with
a Perkin–Elmer Spectrum One in the range of 4000–
The performance of the catalytic membranes was investi-
gated under batch conditions using an equal active surface and
comparable total catalyst loading. A typical procedure for oxi-
dation of sulfides is as follows. Benzylphenylsulfide (0.1 mmol)
is dissolved in dry 1,2-dichloroethane, in a 1.0-mL volumet-
ric flask with an amount of polymer membrane containing
−1
4
00 cm
.
(
5) Information about the chemical state, quantitative surface
analyses, and depth profiles were obtained by an XPS
spectrometer (ꢀ 5600 ci; Perkin–Elmer) equipped with
Al/Mg anodes working with Al-Kα (1486.6 eV) and Mg-
Kα (1253.6 eV) sources at 20 mA and 14 kV. The analyzed
areas were 0.8-mm-diameter circles, and the working pres-
◦
0.01 mmol of Ti-2. After cooling at 0 C, cumylhydroperox-
ide (CHP) (0.1 mmol) is added under magnetic stirring. The
course of the reaction was monitored by gas chromatography
(Varian GC equipped with a SE-30 column) and HPLC (R,R-
Dach DNB; n-hexene:isopropanol 80:20).
−
7
sure was about 10 Pa. The take-off angle was varied
◦
◦
between 45 and 15 to obtain different analysis depths
angle-resolved XPS). The quantitative analysis data are re-
(
For oxidation of secondary amines, the substrate (0.1 mmol)
was dissolved in CDCl (or CD CN), in a 1.0 ml volumetric
ported as atomic percentage of elements.
3
3
(
6) The acid and base components of the membrane surface
tension were obtained by contact angle measurements.
Contact angles of various test liquid droplets on the mem-
brane surfaces were measured by the sessile drop method
using a CAM 200 contact angle meter (KSV Instruments,
Helsinki, Finland). Three reference liquids (distilled water,
diiodomethane, and glycerol) were used to determine the
flask containing activated molecular sieves (25 mg/mmol) and
an amount of polymer membrane containing 0.01 mmol of Ti-2.
After 30 min, CHP (0.4 mmol) was added. Then the solution
was transferred in a screw cap NMR tube and warmed to the
reaction temperature, and the course of the reaction was moni-
1
tored by H NMR.
LW
ab
+
apolar γ , acid–base γ , acid (electron acceptor) γ ,
3. Results and discussion
−
and base (electron donor) γ components of surface free
energy by the Good–van Oss–Chaudhury method [28]. The
Lifshitz–van der Waals component, γ , of the membrane
3.1. Phase diagram of Ti-2/water/DMA/PVDF system
LW
surface tension, reflecting the dipole interactions, was cal-
culated from the measured diiodomethane contact angles
under the assumption that diiodomethane is an apolar test
liquid,
The experimental phase diagram of the Ti-2/water/DMA/
PVDF system is shown in Fig. 1. Line 0% represents the
gelation-phase boundary with no catalyst being added. The
measured crystallization-induced gelation agrees with data re-
ported by Bottino et al. [30]. On addition of catalyst, the gela-
tion line shifts toward nonsolvent axis and the homogeneous
LW
= γlLW(1 + cosϑ) /4.
2
γ
s

224
M.G. Buonomenna et al. / Journal of Catalysis 238 (2006) 221–231
solution (one phase) region becomes larger. Such a shift of the
binodal demixing curve to higher nonsolvent concentrations is
often related to a reduced solvent–nonsolvent interaction and
typically leads to delayed demixing [31,32]. In our case, this
suggests that in the presence of Ti-2, PVDF becomes more
soluble in DMA solvent. This is probably due to specific inter-
actions between the catalyst and fluoride groups of the polymer.
Some polymer/solvent systems have been reported to exhibit
this behavior using salts as additives in a casting solution [33].
3
.2. Membrane morphology and catalyst distribution
The Ti-2 complex has been entrapped in different poly-
meric membranes using the phase-inversion technique induced
by nonsolvent. Some examples of asymmetric membranes pre-
pared by PVDF, PEEKWC and PAN using nonsolvent induced
phase-inversion technique have been reported previously [30–
33,36–47].
◦
Fig. 1. Phase diagrams of Ti-2/DMA/PVDF/water systems at 20 C. (Ti-2 con-
tent: 0 and 1.6 wt%) determined by visual cloud point observation at room
temperature. Closed symbols represent the clear–turbid transition upon nonsol-
vent addition, open symbols represent the turbid–clear transition upon solvent
addition.
(a)
(b)
Fig. 2. SEM photomicrographs of the membranes prepared using PVDF, PAN and PEEKWC (10 wt%) with (a) or without (b) Ti-2 (1.6 wt%).

M.G. Buonomenna et al. / Journal of Catalysis 238 (2006) 221–231
225
Fig. 3. Cross section of PVDF-Ti membrane in back scattered electron (BSE) mode (a) and corresponding EDX analyses at different depths of membrane cross
section (b).
Cross-sections of the polymeric membranes with and with-
out catalyst are shown in Figs. 2a and 2b. According to the
synthetic procedure, the catalyst embedding results in the for-
mation of asymmetric hybrid membranes. PEEKWC-Ti mem-
brane has nearly the same morphology of PEEKWC membrane;
in both cases, observed asymmetric membranes were character-
ized by elongated macrovoids. PVDF-Ti membrane has a cav-
ern structure with macrovoids without particulate morphology
composed of crystalline particles [32] observed in the analogue
PVDF membrane. For membranes synthesized from semicrys-
talline polymers such as PVDF, crystallization can be enhanced
to dominate the precipitation process, thereby yielding partic-
ulate morphologies composed of crystalline particles. Of all
of the parameters that affect the morphology of precipitated
membranes, the composition of the casting solution is the most
crucial. In our case, adding the catalyst in the casting solution
produced a very unusual morphology without macrovoids, as
shown from the morphology in Fig. 2. In addition, for the PAN
membranes, different morphologies were observed in the pres-
ence of the catalyst in the casting solution. PAN-Ti membrane
has a nearly symmetric structure, in contrast to the completely
asymmetric PAN membrane. In all likelihood the catalyst de-
lays the liquid–liquid demixing process favoring macrovoid
suppression [33].
A uniform catalyst distribution was observed in all of the
catalytic polymeric membranes prepared using solvent DMA
only in casting solution.
Fig. 3a shows a cross-section of PVDF-Ti membrane in
back-scattering mode (BSE). The uniform catalyst distributions
were confirmed by EDX analyses of membrane cross-sections.
In fact, the intensities of Ti with respect to fluoride peak (cho-
sen as reference) were constant in EDX spectra at different
depths of membrane cross-sections (Fig. 3b). Fig. 4 shows
the surface of the PVDF-Ti membrane in BSE and the corre-
sponding X-ray maps. The surface was dense and clean. Ti-2
was observed, suggesting that Ti-2 was not recrystallized into

226
M.G. Buonomenna et al. / Journal of Catalysis 238 (2006) 221–231
(a)
(a)
(b)
(b)
(c)
(c)
Fig. 4. X-Ray maps of PVDF-Ti surface (Ti, a; F, b) and corresponding BSE
analysis (c).
Fig. 5. Surface characterization of PVDF catalytic membranes prepared by
using two solvent in casting solution (DMA and acetone). X-Ray maps of
PVDF-Ti surface (Ti, a; F, b) and corresponding BSE analysis (c).
large particles, but rather was highly dispersed as fine particles
throughout the polymeric matrix. In contrast, no uniform cat-
alyst distribution was observed in PVDF catalytic membranes
prepared using a mixture of two solvents (i.e., DMA and ace-
tone) in the casting solution. However, there are visible large
spots (≈150 µm) due to recrystallization of the catalyst on the
membrane surface (Fig. 5) These results are explained by the
volatility difference between DMA and acetone: Acetone goes
out from the nascent membrane for evaporation (dry step), cre-

M.G. Buonomenna et al. / Journal of Catalysis 238 (2006) 221–231
227
Scheme 3.
Scheme 4.
Table 2
Oxidation of benzyl phenyl sulfide with CHP catalyzed by Ti(IV) complexes
a
Table 3
Effect of the catalyst on reaction times and product distribution in the Ti-
catalyzed oxidation of dibenzylamine with CHP
#
Catalyst
Solvent
Time
h)
Conversion
(%)
SO:SO
e.e. (S)
(%)
2
a
(
1
2
3
4
Ti(IV)-2
PVDF-Ti
Ti(IV)-2
DCE
DCE
4
5
1
94
93
15
8
56:44
59:41
90:10
100:0
69
67
44
40
CH CN
3
PEEKWC-Ti
CH CN
100
3
Time
(h)
Catalyst (product ratio: 4a:5a:6a:7a:8)
a
Reaction conditions: [substrate] : = 0.1 M; [CHP] = 0.1 M, [catalyst] =
0
0
Ti-2
PVDF-Ti
PAN-Ti
PEEKWC-Ti
◦
0
.01 M, T = 0 C.
3
7
25
4:5:74:5:3
1:4:78:4:13
38:12:44:6:0
0:7:82:5:6
54:6:25:11:4
49:6:27:12:7
9:0:40:18:33
92:1:6:1:0
82:1:14:3:0
58:1:34:7: 0
43:2:34:14:6
ating catalyst microdomains on the membrane surface, whereas
the membrane final structure is due to DMA/water diffusion in
the coagulation bath (wet step).
117
a
Reaction conditions: [substrate] = 0.1 M; [CHP] = 0.4 M, [catalyst] =
◦
0
.01 M; CD CN; T = 60 C.
3
3
.3. Crystal structure of PVDF-based membranes
phenyl sulfoxide was produced with stereoselectivities compa-
rable to those of the homogeneous systems, even if reactivities
were rather diverse. Despite the fact that reactions in DCE are
intrinsically faster than those in CH3CN, PVDF-Ti catalytic
membrane (Table 2, entry 2) undoubtedly gave the best results,
affording high conversion into products with ee’s comparable
to those of the homogeneous system (Table 2, entry 1). In con-
trast, PEEKWC-Ti afforded only the sulfoxide in much lower
yields (8%) and at much longer reaction times (100 h) (Table 2,
entry 3). These data indicate that both membranes maintain
catalyst stability and that in the case of more hydrophobic mem-
brane PVDF-Ti, the reactivity of the system is not significantly
depleted.
Both PVDF-based membranes shown in Fig. 2 are inherently
crystalline, although their bulk morphologies exhibit different
features, as before observed. Identification of the crystalline
phases was carried out with FTIR spectroscopy and X-ray dif-
fraction. Fig. 6 shows the FTIR spectra of PVDF-Ti and PVDF
membranes. Sharp β-phase characteristic peaks at 470, 511,
−1
and 840 cm were observed in both membranes. Comparing
the IR absorption bands characteristic of the α, β, and γ phases
for PVDF reported in the literature [34] shows that PVDF-TI
catalytic film and PVDF membranes exhibit the same β-type
crystalline phase regardless of the Ti-2 addition. This was also
confirmed by means of XRD spectra (data not shown). The peak
◦
The screening of activity of the different catalytic mem-
branes (PVDF-Ti, PEEKWC-Ti, and PAN-Ti) was also as-
sessed in the oxidation of dibenzylamine, which was chosen
as the model substrate (Scheme 4). To better compare the re-
sults, the reactions were carried out under the best conditions
selected for the homogeneous reactions. Due to the solubility
of PEEK-WC in chlorinated solvents (the most suitable sol-
vents for homogeneous reactions), the reactivity was initially
explored in acetonitrile, and the course of the reactions was
at 20.66 is attributed to the reflection of the (110) plane and
thus to β-type crystallites [35].
3
.4. Membrane activity
Initially, the reactivity of the catalytic membranes was tested
in the sulfoxidation reaction to provide indirect information of
the nature of the catalyst and active species embedded in the
different membranes (Scheme 3). Preliminary tests indicated
that catalyst leaching was not a serious problem for the new
catalytic membranes. In fact, after the membrane was washed
twice with the solvent of the reaction, no significant amount of
catalyst or free trialkanolamine ligand could be detected in so-
1
monitored via H-NMR. Table 3 reports the product distribu-
tions at increasing reaction times for the oxidations catalyzed
by PVDF-Ti, PEEKWC-Ti, PAN-Ti, and Ti-2 homogeneous
system. Significantly, all three Ti-based catalytic membranes
could activate CHP for the oxygen transfer process, although
the polymeric matrix plays an important role in determining
performance and product distribution.
Reactions performed with PAN-Ti and PEEK-WC-Ti gave
slow conversions of the reagent, giving nitrone in comparable
amounts with N-benzylidene benzylamine and benzaldehyde.
Furthermore, at long reaction times, high amounts of benzalde-
hyde were recovered due to further oxidation/hydrolysis of the
products. In contrast, PVDF-Ti efficiently catalyzed oxidation,
giving nitrone in good yields (82%) in reasonable reaction
times, almost comparable to those of the homogeneous system.
1
lution via H-NMR.
In the sulfoxidation reaction, due to the occurrence of two
cooperative stereoselective processes, (i.e., asymmetric oxida-
tion of the sulfide and subsequent kinetic resolution for the
overoxidation to sulfone), the enantiopurity of the sulfoxide in-
creased during the course of the reaction.
The results obtained with PVDF-Ti and PEEKWC-Ti, to-
gether with results obtained under homogeneous conditions, are
reported in Table 2. It is noteworthy that in both cases the cat-
alytic membranes were able to activate CHP and that (S)-benzyl

228
M.G. Buonomenna et al. / Journal of Catalysis 238 (2006) 221–231
Fig. 6. FTIR spectra of PVDF-Ti and PVDF membranes.
Fig. 7. XPS survey spectra of PEEKWC-Ti membrane and homogeneous catalyst Ti-2.
The oxidation with catalytic membrane PVDF-Ti was also
Table 4
Oxidation of dibenzylamine with CHP catalyzed by PVDF-Ti, catalyst recy-
tested in chloroform. In chloroform, the reaction occurs much
faster (t1/2(CDCl3) = 65 min vs t1/2(CH3CN) = 140 min), and
nitrone was recovered at a high chemical yield (92%) with al-
most complete chemoselectivity (with only traces of imine and
hydroxylamine detected).
The performance of the PVDF-Ti catalytic membrane was
also evaluated on the basis of system recycling. A series of ex-
periments were performed to examine this membrane’s activity
along five oxidation runs. After each experiment, the membrane
was removed from the reaction vessel, washed in chloroform to
remove adsorbed reagent or products, and recycled. The results
are reported in Table 4.
cling: substrate: [substrate] = 0.1 M; [CHP]0 = 0.4 M, [catalyst] = 0.01 M;
0
◦
a
T = 60 C, CDCl , m.s. = 250 mg/mmol in run 1, 500 mg/mmol runs 2–5
3
Run
T
(min)
Nitrone (%)
1
/2
1
2
3
4
5
39
10
14
14
23
92
90
90
90
90
a
In the recycling, the presence of twice amount of ms proved to be beneficial
in decreasing the formation of benzaldehyde.
It is noteworthy that the catalytic activity of the PVDF-Ti
membrane was maintained for all five runs, affording compa-

M.G. Buonomenna et al. / Journal of Catalysis 238 (2006) 221–231
229
Table 5
Effect of substrate/catalyst ratio on oxidation of dibenzylamine catalyzed by
Table 6
a
Catalytic oxidation of secondary amines by CHP
◦
a
PVDF-Ti membrane or the homogeneous catalyst Ti(IV)-2 in CDCl at 60 C
3
b
Substrate Product
T
1
Yield
/2
[
Sub] , M
Sub/Cat
Catalyst
Time
h)
Nitrone
(%)
(min)
(%)
0
(
3
9
92
b
0.1
1.0
1.0
10:1
Ti-2
3
2.5
1
3
100
71
100
92
100
90
68
65
PVDF-Ti
b
50
80
93
100:1
1000:1
Ti-2
PVDF-Ti
b
c
Ti-2
20
d
PVDF-Ti
a
Conditions: [catalyst] = 0.01 M, dibenzylamine, CHP (4 equiv) and 4A
a
Reaction conditions: [substrate] = 0.10 M, [CHP] = 0.40 M; [catalyst] =
0
0
◦
molecular sieves in CDCl at 60 C.
3
b
◦
0.01 M, 4A molecular sieve, CHCl at 60 C.
3
See Ref. [26].
b
Based on isolated product after chromatography. Reactions were carried
out on mmol scale. In all cases nitrones were readily isolated in synthetically
significant chemical yields (80–93%) via removal of the solvent under vacuum
followed by chromatography over silica gel.
c
Complete reagent conversion; other products: N-benzylidene benzylamine
(
15%), PhCHO (15%).
d
Complete reagent conversion; other products: N-benzylidene benzylamine
(
10%), PhCHO (22%). DCE as solvent.
of conversion and selectivity; in contrast, reactions performed
with PAN-Ti and PEEKWC-Ti gave low and slow conversion
of the reagent. This difference was ascribed to the different
chemical–physical properties of the polymeric membranes (sur-
face tension differences). The catalytic membranes investigated
rable yields in nitrone (90%). Interestingly, reaction rates were
higher in the recycling experiments than in the initial experi-
ment. This increased reactivity could originate from modifica-
tions of the polymeric membrane that make available a larger
quantity of the catalyst or enhance the Lewis acidity of the
metal complex. This important aspect will be addressed in fu-
ture studies.
The effect of catalyst loading for the PVDF-Ti membrane
was also tested at substrate/catalyst ratios of 10:1–1000:1 (Ta-
ble 5). The results demonstrate that complete conversion of
dibenzylamine and >90% selectivity for nitrone could also be
achieved in the presence of 1% of catalyst. However, further in-
creasing the ratio to s/c = 1000 (which necessarily decreased
the catalyst concentration to 0.001 M) resulted in a severe de-
terioration of chemoselectivity, as was also observed in the ho-
mogeneous system [26].
+
exhibited different values of acidic component γ due to their
different chemical structures (Table 7). In fact, as Della Volpe
and Siboni suggested, one “cannot compare the acid and base
components of the same solvent (or substrate), but eventually
the acid (or base) components of different substrates can be
compared” [48].
+
PVDF-Ti exhibited the highest value of γ , due to the fluo-
ride atoms of the PVDF. With its acidic character, PVDF prob-
ably favors the nucleophilic attack of dibenzylamine more than
+
−
PEEKWC and PAN do, with lower γ and higher γ values,
respectively.
The conditions for the highest catalyst loading (10%) from
Table 5 were adopted to explore the scope of the reaction
with other secondary amines (Table 6). Benzylic or branched
nitrones were highly stable in the oxidative environment; in
all cases they were readily isolated in synthetically significant
chemical yields (80–93%).
3
.6. X-Ray photoelectron spectroscopy
The physicochemical interactions of Ti-2 with the polymeric
membrane were explored by XPS analyses. XPS analyses of
homogeneous catalyst, PVDF-Ti, and PEEKWC-Ti membranes
revealed a unique titanium species present in all of the sys-
tems (Figs. 7 and 8). This experimental evidence agrees with
the comparable stereoselectivity observed in asymmetric sul-
foxidation using these catalytic membranes.
3
.5. Contact angle measurements
Our study showed that all of the catalytic membranes pre-
Furthermore, a very peculiar and interesting orientation of
the catalyst has been found in the PVDF-Ti membrane. It was
possible to obtain quantitative surface analyses by XPS. As
shown in Table 8, in the solid catalyst, a Ti/N ratio of about 1.0
serve the structure of Ti-2 catalyst, as indirectly observed on
the basis of reactivity tests. However, the membrane reactivi-
ties are rather diverse. PVDF-Ti gave the best results in terms
Table 7
LW
Contact angles in water (W), glycerol (Gl), diiodomethane (DIM) and calculated surface tension parameters of different catalytic polymeric membranes: γ
=
+
−
Lifshitz–van der Waals component, γ = electron-acceptor (acid parameter) component and γ = electron (base parameter) of the liquid surface tension γ . The
S
surface tension and its parameter were calculated according to Good–Van Oss equation by the average values of measured contact angles
LW
−
+
AB
γ
Membrane
ϑDIM
ϑW
( )
ϑGl
( )
γ
γ
γ
γ
S
(mJ/m2)
S
S
S
S
◦
◦
◦
(
)
2
2
2
2
(
mJ/m )
(mJ/m )
(mJ/m )
(mJ/m )
PVDF-Ti
PEEKWC-Ti
PAN-Ti
58
62
65
123
84.2
51
130
73
54
29.5
27.4
25.7
2.9
7.9
32.8
9.7
0.48
1.75
10.6
3.8
15.1
40.05
31.3
40.8

230
M.G. Buonomenna et al. / Journal of Catalysis 238 (2006) 221–231
Fig. 8. XPS survey spectra of PVDF-Ti membrane and homogeneous catalyst Ti-2.
Table 8
Atomic compositions (%) derived from XPS spectra
Sample
C
O
Si
Cl
F
Ti
N
Ti/N
Ti-2
PVDF-Ti
PEEKWC-Ti
73.7
57.6
83.1
17.0
12.1
14.6
5.3
0.8
0.3
0.2
0.0
0.0
0.0
24.7
0.0
1.7
2.8
0.7
2.1
2.0
1.3
0.8
1.5
0.5
Table 9
Atomic compositions (%) of untested and tested PVDF-Ti membranes from angle-dependent XPS spectra
◦
◦
Composition at 15 take off angle
Sample
Composition at 45 take off angle
C
O
Si
Cl
F
Ti
N
Ti/N
C
O
Si
Cl
F
Ti
N
Ti/N
PVDF-Ti
Untested
57.6
12.1
0.8
0.0
24.7
2.8
2.0
1.5
56.0
12.4
0.0
0.0
26.9
3.1
1.6
1.9
PVDF-Ti
After II run
50.9
15.8
1.4
1.5
24.5
3.5
2.4
1.4
51
17.7
1.7
1.4
22.4
4.1
1.8
2.3
(
0.8) was found, due to the 1-to-1 titanium–nitrogen stoichiom-
increased at lower take-off angles, confirming that the Ti atoms
are nearer to the surface than the N atoms. It is noteworthy
that after run II, this orientation effect of the Ti-2 in PVDF-
based membrane is enhanced. This result provides important
experimental evidence for the observed increased reactivity of
PVDF-Ti after run II.
etry. In the PVDF-Ti membrane, a 1.5 Ti/N ratio was found.
These results clearly indicate that in this membrane, Ti is more
exposed than N on the surface, or, in other words, Ti atoms
with the labile monodentate alkoxy ligand lie on the PVDF
membrane surface. For the PEEKWC-Ti membrane, a 0.5 ratio
was found, indicating that in this case, N is more exposed to the
membrane surface than Ti.
4. Conclusion
We performed a qualitative angle-resolved study to confirm
that Ti lies on the surface of PVDF-Ti. As the take-off angle
In this work we have achieved the first heterogenization of
Ti(IV) catalysts. Polymeric catalytic membranes consisting of
Ti(IV)/trialkanolamine complex entrapped in PVDF, PAN, and
PEEKWC membranes are active for the oxidation of dibenzyl-
amine to nitrone. For all of the catalytic membranes, SEM,
EDX, and BSE analyses exhibited a uniform catalyst distri-
(
the angle between the surface and the photoelectrons accepted
by the analyzer) decreases, the surface sensitivity of XPS in-
creases [49]. Table 9 gives the atomic compositions derived
from survey spectra at two take-off angles of untested and tested
(
after run II) PVDF-Ti membranes. As expected, the Ti/N ratio

M.G. Buonomenna et al. / Journal of Catalysis 238 (2006) 221–231
231
bution. XRD and FTIR analyses showed that the crystalline
structure of PVDF-Ti was unchanged by addition of the cata-
lyst.
[14] I.F.J. Vankelecom, P.A. Jacobs, Catal. Today 56 (2000) 147.
[15] P.E.F. Neys, I.F.J. Vankelecom, R.F. Parton, W. Dehaen, G. L’abbé,
P.A. Jacobs, J. Mol. Catal. 126 (1997) L9.
[16] R.F. Parton, I.F.J. Vankelecom, D. Tas, K.B. Janssen, P.P. Knops-Gerrits,
P.A. Jacobs, J. Mol. Catal. 113 (1996) 283.
The reactivity screening allowed us to: (i) select the PVDF-
Ti catalytic membrane as the best system in terms of conversion
and selectivity, (ii) exclude the occurrence of competing homo-
geneous pathways, and (iii) recycle the catalytic membrane for
five runs without loss of activity. XPS analyses done to explore
the nature of the catalyst embedded in the polymeric matri-
ces confirmed the preservation of the catalyst structure based
on reactivity tests. A very interesting orientation of the Ti(IV)
complex was observed on the PVDF membrane surface, which
is the origin of the higher activity of PVDF-Ti compared with
that of the analogue homogeneous system.
[17] I.F.J. Vankelecom, D. Tas, R.F. Parton, V. Van de Vyver, P.A. Jacobs,
Angew. Chem. Int. Engl. 35 (1996) 1346.
[18] P.P. Knops-Gerrits, I.F.J. Vankelecom, E. Beatse, P.A. Jacobs, Catal. To-
day 32 (1996) 63.
[19] M.C. Carreno, Chem. Rev. 95 (1995) 1717.
[
20] P. Pichen, Asymmetric Synthesis of Sulfoxides: Two Case Studies in Chi-
rality in Industry II, Wiley, New York, 1997, p. 381.
21] F.P. Ballistreri, U. Chiacchio, A. Rescina, G. Tomaselli, R.M. Toscano,
Tetrahedron 48 (1992) 8677.
[
[22] F.P. Ballistreri, R. Bianchini, C. Pinzino, G. Tomaselli, R.M. Toscano,
J. Phys. Chem. A 104 (2000) 2710.
[23] W.A. Nugent, R.L. Harlow, J. Am. Chem. Soc. 116 (1994) 6142.
+
In addition, analyses of the surface acid parameters (γ ) of
[24] F. Di Furia, G. Licini, G. Modena, R. Motterle, W.A. Nugent, J. Org.
Chem. 61 (1996) 5175.
the PVDF-, PEEKWC-, and PAN-based catalytic membranes
provides a supporting argument for the observed differences in
reactivity. PVDF-Ti membranes exhibited the highest value of
[25] G. Licini, M. Bonchio, G. Modena, W.A. Nugent, Pure Appl. Chem. 71
(1999) 463.
+
γ , due to the fluoride atoms of the PVDF. This may favor nu-
[26] M. Forcato, W.A. Nugent, G. Licini, Tetrahedron Lett. 44 (2003) 49.
27] K. Jian, P.N. Pintauro, J. Membr. Sci. 85 (1993) 301.
[
cleophilic attack of the substrate compared with the PEEKWC-
+
[28] R.J. Good, C.J. van Oss, in: M.E. Scrader, G.L. Loeb (Eds.), Modern Ap-
proaches to Wettability, Plenum, New York, 1992.
Ti and PAN-Ti membranes, with lower γ values and higher
−
γ
values, respectively.
[
29] C.J. van Oss, Interfacial Forces in Aqueous Media, Dekker, New York,
994.
30] A. Bottino, G. Camera Roda, G. Capannelli, S. Munari, J. Membr. Sci. 57
1991) 1.
[31] H.K. Lonsdale, J. Membr. Sci. 10 (1982) 81.
1
Acknowledgments
[
(
The authors are grateful to the Italian Ministry of University
and Research for supporting this research through the MIUR-
FIRB RBNE03JCR5 project.
[32] J.H. Kim, B.R. Min, J. Won, H.C. Park, Y. Kang, J. Membr. Sci. 187
(2001) 47.
[
[
[
33] F.M. Gray, Solid Polymer Electrolytes, VCH, New York, 1991.
34] W.M. Prest, D.J. Luca, J. Appl. Phys. 46 (1975) 4136.
35] R. Hasegawa, Y. Takahashi, Y. Chatani, H. Tadokoro, Polymer 3 (1972)
References
6
00.
36] R.D. Lundberg, F.E. Bailey, R.W. Callard, J. Polym. Sci. A 1 (4) (1966)
563.
[
[
[
[
[
[
1] A.T. Bell, Science 299 (2003) 1688.
2] M.A. Barteau, J.E. Lyons, I.K. Song, J. Catal. 216 (2003) 236.
3] J. Corker, Science 271 (1996) 966.
4] V. Vidal, A. Theolier, J. Thivolle-Cazat, J.M. Basset, Science 276 (1997)
9
[
1
37] H.C. Shih, Y.S. Yeh, H. Yasuda, J. Membr. Sci. 50 (1990) 299.
38] A. Bottino, G. Capannelli, O. Monticelli, P. Piaggio, J. Membr. Sci. 166
9.
5] C. Nozkaki, C.G. Lugmair, A. Bell, T.D. Tilley, J. Am. Chem. Soc. 124
2002) 13194.
(2000) 231.
[
[
[
[
[
[
[
[
[
39] D. Lin, C. Chang, F.M. Huang, L.P. Cheng, Polymer 44 (2003) 413.
40] B. Jung, J.K. Yoon, B. Kim, H.W. Rhee, J. Membr. Sci. 243 (2004) 45.
41] Y. Xiuli, C. Hangbin, W. Xiu, Y. Yongxin, J. Membr. Sci. 146 (1998) 179.
42] N. Scharnagal, H. Buschtz, Desalination 139 (2001) 191.
43] S. Yang, Z. Liu, J. Membr. Sci. 222 (2003) 87.
44] E. Drioli, H.C. Zhang, Chimicaoggi 11 (1989) 59.
45] F. Tasselli, J.C. Jansen, E. Drioli, J. Appl. Polym. Sci. 91 (2004) 841.
46] M.G. Buonomenna, A. Figoli, J.C. Jansen, E. Drioli, J. Appl. Polym.
Sci. 92 (2004) 576.
(
[
[
[
[
6] D. Kolb, Surf. Sci. 500 (2002) 722.
7] K.P. De Jong, J.W. Geus, Catal. Rev. Sci. Eng. 42 (2000) 481.
8] D. Fritch, K.-V. Peinemann, Catal. Today 25 (1995) 277.
9] J.F. Ciebien, R.E. Cohen, A. Duran, Supramol. Sci. 5 (1998) 31.
[
10] J. Vital, A.M. Ramos, I.F. Silva, H. Valente, J. E Castanheiro, Catal. To-
day 56 (2000) 167.
11] I.F.J. Vankelecom, N.M.F. Moens, K.A.L. Vercruysse, R.F. Parton, P.A. Ja-
cobs, in: H.U. Blaser, A. Baiker, R. Prins (Eds.), Heterogeneous Catalysis
and Fine Chemicals, IV, Elsevier, Amsterdam, 1997, p. 437.
12] I.F.J. Vankelecom, Chem. Rev. 102 (2002) 3779.
[
[
47] M. Mulder, Basic Principles of Membrane Technology, Kluwer, Dor-
drecht, 1991.
[
[
13] P.E.F. Neys, A. Severeyns, I.F.J. Vankelecom, E. Ceulemans, W. Dehaen,
P.A. Jacobs, J. Mol. Catal. 144 (1999) 373.
[48] C. Della Volpe, S. Siboni, J. Colloid Interface Sci. 195 (1997) 121.
[49] D. Briggs, M.P. Seah, Practical Surface Analysis, Wiley, Chichester, 1983.