'Click' silica immobilisation of metallo-porphyrin complexes and their application in epoxidation catalysis

Article abstract of DOI:10.1016/j.jorganchem.2009.02.020

We present the synthesis, via Adler condensation reactions, of mono- and tetrakis-4-(ethynyl-phenyl)porphyrin ligands and the zinc and manganese complexes thereof. The formed complexes were immobilised on silica by reacting the ethynyl groups with azide-f

Full text of DOI:10.1016/j.jorganchem.2009.02.020

Journal of Organometallic Chemistry 694 (2009) 2153–2162
Contents lists available at ScienceDirect
Journal of Organometallic Chemistry
journal homepage: www.elsevier.com/locate/jorganchem
‘Click’ silica immobilisation of metallo-porphyrin complexes and their application
in epoxidation catalysis
*
Aidan R. McDonald, Nicole Franssen, Gerard P.M. van Klink, Gerard van Koten
Chemical Biology and Organic Chemistry, Faculty of Science, Utrecht University, Padualaan 8, 3584 CH Utrecht, The Netherlands
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 3 June 2008
Received in revised form 12 February 2009
Accepted 24 February 2009
Available online 1 March 2009
We present the synthesis, via Adler condensation reactions, of mono- and tetrakis-4-(ethynyl-
phenyl)porphyrin ligands and the zinc and manganese complexes thereof. The formed complexes were
immobilised on silica by reacting the ethynyl groups with azide-functionalised silica in a copper(I) cat-
alysed Huisgens 1,3-dipolar cycloaddition reaction. The synthesised metallo-porphyrin containing mate-
rials were thoroughly characterised using various solid-state techniques (NMR, IR, UV–Vis, elemental
content analysis). The manganese containing materials were applied as catalysts in the epoxidation of
various alkenes (cyclooctene, cyclohexene, styrene) with various oxidants (iodosylbenzene, tert-butyl-
peroxide). The heterogenised homogeneous catalysts show diminished activity and yields compared to
the analogous homogeneous catalysts (71% yield cf. 92% for cyclooctene epoxidation, TOF 82 hÀ1 cf.
230 hÀ1). Upon recycling, the heterogenised catalysts become gradually less active over five cycles until
they are catalytically inactive. The deactivation process is discussed, with spectroscopy suggesting that
the catalysts themselves are intact and thus stable to the reaction conditions and recycling, however
there is likely some decomplexation, and also both chemical and mechanical decomposition of the silica
support resulting in inaccessibility to the catalytic site.
Keywords:
Homogeneous catalysis
Epoxidation
Immobilisation
Click chemistry
Catalyst recycling
Ó 2009 Elsevier B.V. All rights reserved.
1. Introduction
According to a recent report asymmetric epoxidation is the sec-
ond most widely used asymmetric process in industry [3]. Chiral
The immobilisation of homogeneous catalysts, and the stability
of the catalyst itself, are important issues in the development of
homogeneous catalytic systems for industrial application [1]. The
immobilisation of homogeneous catalysts allows for ease of sepa-
ration of catalyst from reactants and products, while catalyst sta-
bility is essential to recycling procedures and for the recovery of
precious metal complexes. The synthesis of many of the fine chem-
icals essential for life and leisure involve at least one or more steps
that involve the use of homogeneous catalysis [2]. The reason for
this is relatively simple, selectivity. Homogeneous catalysts pro-
vide the required selectivity for certain end-products, that hetero-
geneous catalysts cannot offer. Thus the benefits of using
heterogeneous catalysts, high stability and recyclability, are out-
weighed by the necessity to have selective syntheses using homo-
geneous catalysts. However, the cost of synthesis and recovery of
homogeneous catalysts or their derivatives mean they are not
widely applied. One manner to overcome this would be to immo-
bilise homogeneous catalysts such that they can be treated in the
same manner as heterogeneous catalysts, i.e. they can be recycled.
epoxides and their derivatives are extremely high value products
and are the basis of many enantiopure compounds. For this reason
the immobilisation of chiral epoxidation catalysts has been inves-
tigated thoroughly in recent times [4]. The field of supported
metallo-porphyrins for epoxidation catalysis [5] and in particular
the immobilisation of porphyrin complexes on inorganic supports
and their application in alkane/alkene oxidation have been investi-
gated [6]. What comes from these previous reports and is a general
trend in the field of homogeneous catalyst immobilisation, is that
covalent immobilisation of catalysts on inorganic supports is essen-
tial for good catalyst recycling results, and minimal influences on
the catalytic site by the support. It is also apparent that a simple
manner for the immobilisation of such homogeneous catalysts on
inorganic supports is necessary. A key point to add to this is that
catalyst stability should be taken into account before any attempts
are made to covalently immobilise a homogeneous catalyst. Previ-
ous work has shown that highly stable coordination complexes
should be used, because labile coordination complexes do not
withstand the relatively harsh conditions used for catalyst immo-
bilisation, and subsequently for the recycling of the heterogenised
catalyst. The conditions for immobilisation in some systems can be
detrimental to the most stable of complexes, and ideally a modular
system, comprising support, linker and catalyst would be a goal, in
* Corresponding author. Fax: +31 30 252 3615.
E-mail addresses: mcdon701@umn.edu (A.R. McDonald), g.vankoten@uu.nl
(G. van Koten).
0022-328X/$ - see front matter Ó 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.jorganchem.2009.02.020

2154
A.R. McDonald et al. / Journal of Organometallic Chemistry 694 (2009) 2153–2162
Column chromatography was used to isolate the mono-ethynyl
porphyrin manganese complex from the other products. However,
it was not possible to separate and [MnCl(TPP)]
6
(TPP = 5,10,15,20-tetrakisphenylporphyrin). Nevertheless, for the
immobilisation of 6 on silica, the 6/[MnCl(TPP)] mixture could be
used, because the [MnCl(TPP)] could be washed off the silica sup-
port after 6 had been tethered to the support. In all three complex-
ation reactions almost all TMS-protecting groups were cleaved
during the reaction. Tetrabutylammonium fluoride (TBAF) was
added to the reaction mixtures to ensure cleavage of all TMS
groups, but in most instances was not necessary.
Scheme 1. Copper(I) thiolate catalysed ‘Click’ coupling of ethynyl organic moieties
to silica support.
Ethynyl-functionalised porphyrin ligand 2 and metallo-porphy-
rins 3, 4, and 6 were reacted with azide-functionalised silica in the
presence of a copper catalyst (2-[dimethylamino)methyl]-1-thio-
phenolato-copper(I)) at room temperature in MeCN for four days.
With free-base porphyrin 2 (TMS groups cleaved using TBAF) sev-
eral attempts to immobilise on silica using ‘click’ techniques were
made. However, we were unsuccessful with both the aforemen-
tioned copper(I) thiolate complex and also [Cu(MeCN)4]PF6 as cat-
alyst. We found that transmetallation occurs between the copper(I)
catalysts and the porphyrin ligand, thus deactivating the copper
catalyst which resulted in very low yields of immobilised porphy-
rin ligand 2. This led us to synthesise the ‘protected’ porphyrin
material 4, because, contrary to the copper porphyrin complexes,
zinc containing porphyrins are known to be relatively labile and
can be applied in demetallation and transmetallation reactions.
Material 4 was reacted with azide-functionalised silica yielding
material 8. We attempted to demetallate the zinc containing mate-
rial 8 using trifluoro-acetic acid (TFA). Spectroscopic studies (UV–
Vis) of the resulting material confirmed that the demetallation
process to yield immobilised free-base porphyrin material 11
was successful. However, we found it very difficult to remove all
the formed non-porphyrin zinc(II) salts from the silica material.
The zinc elemental content only dropped from 0.24% to 0.14% even
after carrying out several thorough washing techniques, and IR
spectroscopy gives indications of acetates on the silica surface.
Materials 7, 8, and 9 were successfully synthesised (Scheme 3).
Ethynylbenzene was subsequently added to the reaction mixture
to ensure ‘capping’ of any remnant azide groups on the silica sur-
face. In the case of materials 7 and 8 it was difficult to gather
experimental proof that all ethynyl groups had reacted with sur-
face azide groups. Therefore, extra free hexylazide was added to
the reaction mixture to ensure any remnant ethynyl groups were
‘capped’ to ensure they did not influence any future applications
of the materials. The obtained materials were either dark green
in colour (manganese containing materials 7 and 9) or light pink
(zinc containing material 8). Model system 10 was also synthesised
for spectroscopic purposes, to monitor both the electronic effects
of having triazole-substituents bound to the porphyrin backbone
and to study the influence of the support material on the porphyrin
metal complex itself.
which mixing, under mild conditions, of the various components
together resulted in the immobilised catalyst.
We have recently demonstrated such a system in which the
immobilisation of homogeneous catalysts on an inorganic support
is mediated using ‘click’ immobilisation techniques involving an
azide linked to a high density silica support (Fig. 1) [7]. ‘Click’
immobilisation is defined by high yielding, highly atom efficient
reactions for the tethering of an organic or organometallic moiety
to an inorganic support. Azide-functionalised silica has been used
to couple ethynyl-functionalised compounds to high density silica
using a copper(I) catalysed 1,3-Huisgens dipolar cycloaddition
reaction. The Huisgens dipolar cycloaddition is very prominent in
organic chemistry presently, because of its high yielding, high
atom efficiency and selectivity. The reaction has been classified
as a ‘Click’ reaction, because of its high yielding, efficient nature,
and its low by-product production [8] (see Scheme 1).
In this report immobilisation of manganese and zinc porphyrin
complexes was mediated through the copper(I) thiolate catalysed
reaction of an acetylene-functionalised porphyrin complex with
the aforementioned azide-functionalised silica support. The novel
materials were fully characterised using solid-state spectroscopic
techniques. The synthesised manganese materials were applied
as catalysts in the epoxidation of various alkenes showing results
comparable to the homogeneous catalyst system. However, upon
recycling the catalytic material showed considerably diminished
activity. The porphyrin ligand system is acknowledged to be highly
stable and rigid, however, the epoxidation reactions that metal
porphyrin complexes catalyse use oxidants that can have great ef-
fects on complex stability, but also that give side reactions and we
believe this is one of the sources of the deactivation of the catalytic
manganese materials.
2. Results and discussion
2.1. Synthesis of silica-bound porphyrin complexes
For ethynyl-functionalised porphyrin synthesis (Scheme 2),
mono-trimethylsilyl (mono-TMS) protected ethyne was coupled
with 4-bromo-benzaldehyde yielding 4-[(trimethylsilyl)ethy-
nyl]benzaldehyde (1) in a palladium catalysed Sonigashira cou-
pling reaction (79% yield). Aldehyde 1 was applied in an Adler
porphyrin synthesis in two manners [9]. Similar compounds have
been previously synthesised [10]. Firstly, 1 was mixed in a 1:1 ratio
with pyrrole in refluxing propionic acid yielding 5,10,15,20-tetra-
kis-(4-[(trimethylsilyl)ethynyl]phenyl)-porphyrin (2). Compound
2 was reacted with both manganese(II) and zinc(II) acetates to
yield complexes 3 and 4, respectively, in relatively high yields. Sec-
ondly compound 1 was reacted with pyrrole and benzaldehyde
(1:6:7 molar ratio, respectively) to synthesise 5-(4-[(trimethyl-
silyl)ethynyl]phenyl)-10,15,20-trisphenylporphyrin (5).
It is important to note that test reactions were carried out to
monitor if transmetallation between the copper(I) thiolate com-
plex (‘click’ catalyst) and the porphyrin–manganese/zinc com-
plexes had occurred during the immobilisation procedure. This
was found not to be the case, demonstrating the effectiveness of
using the copper catalysed coupling reaction for high yielding,
clean, side-product-less immobilisation. No signs of transmetalla-
tion were observed (MALDI-ToF mass analysis) when a stoichiom-
etric 1:1 molar mixture of copper(I) thiolate and metallo-
porphyrins 3, 4, or 6, were mixed together under standard ‘click’
reaction conditions. No signals pointing to copper porphyrin for-
mation were observed, and likewise no manganese or zinc thiolate
signals were observed. Furthermore, it was found that neither
complexes 3, 4, or 6 catalyse the 1,3-Huisgens dipolar cycloaddi-
tion reaction.
Compound 5 could not be completely purified (TPP, with small
amounts of bis-, tris- and tetrakis- substituted porphyrins were
also formed in the synthetic process) and complexation with
Mn(OAc)2 was carried out on the mixture of functionalised ligands.

A.R. McDonald et al. / Journal of Organometallic Chemistry 694 (2009) 2153–2162
2155
Scheme 2. Synthesis of mono- and tetra- functionalised porphyrin complexes for tethering to silica support material.
2.2. Elemental content analysis of Mn–porphyrin materials
2.3. Surface area measurements
The manganese loading of materials 7 and 9 was found to be
approximately 0.1% per weight of the total silica mass
(1.8 Â 10À5 mol/g Mn on SiO2). The nitrogen loading of materials
7, 8, and 9 is less than the azide-functionalised silica material
(0.5–1.0% vs. 1.39% for starting azide material). However, the de-
crease in nitrogen loading is to be expected because the ethy-
BET surface area measurements were used to analyse the start-
ing azide material [11]. The starting azide material was chosen be-
cause it gives a reliable and general indication of the surface
landscape relative to all materials presented herein. The surface
area of the unfunctionalised silica starting material was approxi-
mately 218 m2/g. The azide-functionalised material showed a sur-
face area of 208 m2/g. This is an expected decrease, which comes
with the coupling of organic moieties to the silica surface, thus
decreasing the surface area. The pore diameter and volume mea-
surements also gave valuable information about catalyst site
accessibility. We set out to immobilise catalysts without having
influences from the support material. This is reflected in the ob-
served pore volumes for the azide material vs. the starting silica
material. On loading the silica material with the tethering func-
tionality, the pore volume drops minimally. The observed large
pore volumes (ꢀ1 cm3/g) and the observed pore size (16–17 nm)
nyl–phenyl functionalised manganese porphyrin is
a large
macromolecule with a large carbon content, which would be ex-
pected to result in diminished relative nitrogen content mea-
surement (even though it contains four nitrogen’s itself). This
is backed up by the relatively large carbon content (18% for
material 7). The relative nitrogen to manganese loadings would
suggest that there are 20 nitrogen atoms for every manganese
atom in material 7 – suggesting complete tethering of all four
ethynyl groups to the silica surface (expected 16 N for every 1
Mn).
Scheme 3. Immobilisation of functionalised porphyrin complexes on silica support using copper(I) thiolate catalysed click chemistry.

2156
A.R. McDonald et al. / Journal of Organometallic Chemistry 694 (2009) 2153–2162
Fig. 1. Soluble model system for immobilised manganese complexes, 10, and insoluble immobilised free-base porphyrin material 11.
means it is highly unlikely that the support has much influence on
the kinetics (mass transport) of the catalytic reactions (see
Table 1).
both reference complexes 10 and [MnCl(TPP)]. Similarly, the Q-
band region of the spectrum shows results comparable to those
of the untethered complexes. Material 9 shows contrasting results
to material 7 with a considerable blue shift in both the Soret band
and the Q-bands compared to the reference compounds. The strong
blue shifts observed upon immobilisation indicate aggregation of
By combining the nitrogen content analysis results of the azide
material with the BET surface area results we can get an indication
of the ‘area’ per a single catalytic site, and the proximity of catalytic
sites to each other. This point is more relevant to material 9
(mono-dentate to silica surface) than 7 (multi-dentate to silica sur-
face). This relies on the assumption that all azide sites are available
for reaction with ethynyl-functionalised organometallic species,
and the entire surface of the material is accessible to the organic
moieties used to functionalise the material. The nitrogen content
is 1.39% of the total loading, which gives 4.76 Â 10À6 mol/m2 nitro-
gen loading. This can be converted to functionality loading by
dividing by three (three nitrogen’s for every azide) which gives
1.59 Â 10À6 mol/m2 (1.59 Â 10À26 mol/Å2) azide loading. Using
Avogadro’s constant this gives an approximation that every azide
group sits on 104.4 Å2. Simplifying the calculations even further,
we can estimate the ‘ideal situation’ distance between azide sites.
Assuming that all azide sites are equally deposited on the silica
surface, we treat every functionality site as a cylinder lying perpen-
dicular to the silica surface. The estimated radius (r) of the area
porphyrins by
p-stacking on the silica surface [12]. This observa-
tion can tell us a lot about both materials 7 and 9. It appears that
with only one site for tethering of the porphyrin ring to the inor-
ganic support (material 9), high levels of inter-porphyrin interac-
tion on the surface are possible (see surface area analysis
results). Because no blue shift is observed in the spectrum of 7
we assume the multiple anchoring of the porphyrin prevents this
stacking. This would suggest that inter-porphyrin interactions are
highly likely when there is a single appendage between the por-
phyrin ring and the silica surface, because the porphyrin lies per-
pendicular to the silica surface, and is relatively mobile. In
material 7, because there are four possible appendages between
the porphyrin ring and the silica surface, the porphyrin is probably
forced to be parallel to the silica surface, and thus inter-porphyrin
interactions are unlikely.
Material 8 was also characterised using solid-state UV–Vis spec-
troscopy and the subsequent demetallation of material 8 was mon-
itored. Fig. 3 depicts the solid-state UV–Vis reflectance spectrum of
material 8 and the solid-state spectrum of material 11, which is
formed after demetallation of material 8. The latter material shows
characteristic absorptions of free-base porphyrin species. The fact
that positioning of the Soret band and the Q-bands have shifted
considerably with respect to 8 is characteristic of an uncoordinated
porphyrin ligand.
around a catalytic site could be acquired from p
r2 = area. Therefore,
the average distance between catalytic sites is equal to 2r, and thus
is 11.3 Å. This distance is relatively small and has consequences for
both materials 7 and 9. The relatively short distance suggests that
material 7 can readily bind to the silica surface with more than one
of its ethynyl functionalities. Thus confirming the elemental anal-
ysis results which suggest the same from the Mn:N content. In
material 9 the relatively short distance would allow for easy in-
ter-catalytic site interactions which is discussed further later in
this report.
2.5. Infra-red analysis of porphyrin materials
2.4. UV–Vis analysis of Mn– and Zn–porphyrin materials
Porphyrin containing materials were also studied with ATR-IR
[13] and DRIFT [14] techniques. Fig. 4 depicts the DRIFT difference
spectra pertaining to material 7. The results observed are indica-
tive for all porphyrin bound materials synthesised (7–11). The
DRIFT spectrum of the starting silica is subtracted from the spec-
trum of 7 giving an indication of what has been ‘added’ to the silica
(Fig. 4a). The spectra show that a trough in the difference spectrum
is observed at 3738 cmÀ1 which corresponds to a clear decrease in
the intensity of silanol stretches demonstrating that all linker
groups are covalently bound to the silica support. The decrease
in silanol groups is a result of the synthesis of azide-functionalised
material (Fig. 1). It demonstrates a covalent linkage between the
silica support and the propyl-triazole linker is present. Fig. 4b con-
firms the reaction of ethynyl groups of the porphyrin complex with
the azide group on the silica surface. The trough at 2037 cmÀ1 is
the azide stretch of the starting material which has disappeared
in material 7. In both spectra a less intense resonance peak is ob-
Solid-state diffuse-reflectance UV–Vis spectroscopy was used to
analyse materials 7, 8, and 9. The acquired spectra for 7 and 9 were
compared to solution-state transmission spectra of compound 10
and [MnCl(TPP)] respectively (Fig. 2). The solid-state spectra con-
firm that metallo-porphyrin species have been immobilised on
the silica surface, with minimal changes in the photophysical prop-
erties of the coordination complexes bound to the support. Mate-
rial 7 shows a Soret band at 479 nm, which is comparable with
Table 1
BET surface area, pore volume and size measurements.
Material
Surface area (m2/g)
Pore volume (cm3/g)
Pore size (nm)
Starting SiO2
Azido-SiO2
218.40
208.20
1.13
0.92
20.61
17.62

A.R. McDonald et al. / Journal of Organometallic Chemistry 694 (2009) 2153–2162
2157
Fig. 2. Solid- and solution-state UV–Vis spectra of materials 7, 9, 10 and [MnCl(TPP)], respectively.
Fig. 3. UV–Vis solid-state reflectance spectra of materials 8 and 11.
served at 1450 cmÀ1 which corresponds to the triazole groups of
the ‘clicked’ final product (confirmed by ATR spectrum of 10).
These resonances are not observed in the starting material spectra.
Due to the low amount of characteristic, intense IR resonances of
metallo-porphyrin complexes, it was difficult to assign the remain-
der of the observed peaks. However, the broad resonances ob-
served in the 3000–3500 cmÀ1 and 900–1300 cmÀ1 range are
characteristic of the silica support material.
port. The 29Si spectrum of organometallic containing material 7
showed several peaks that indicated the presence of different
forms of silicon species at the surface and in the bulk of the mate-
rial (Fig. 5). Peaks at À101 and À111 ppm are characteristic of Q
type (Si-(O-)4) silicates, which are in the bulk material. Peaks at
À61 and À67 ppm are typical of T-type (R1-Si-(O-)3) silicates, that
is, the silicon bound to the propyl group at the surface of the sup-
port. This is a clear indication that the catalyst is covalently bound
to the support. The presence of double peaks in the À60 ppm (T2 at
61 ppm, T3 at 67 ppm) region indicates that not all ethoxy groups
of functionalised propyl-1-triethoxysilane have reacted in the
grafting process, and in some cases only two (thus nomenclature
T2) of the three have reacted. The single peak at 13 ppm is typical
2.6. NMR analysis
29Si CP-MAS NMR spectroscopy was applied to analyse the ef-
fects of carrying out organic synthetic techniques on the silica sup-

2158
A.R. McDonald et al. / Journal of Organometallic Chemistry 694 (2009) 2153–2162
Fig. 4. IR DRIFT difference spectra of material 7: (a) 7 minus SiO2, (b) 7 minus azide-functionalised material.
of M-type (R3-Si-(O-)) silicates (where R is the methyl groups of the
trimethylsilyl capped silanol groups).
82 turnovers hÀ1). The observed decrease in catalyst activity (on
a per Mn site basis) is to be expected with the heterogenisation
of homogeneous catalysts. However, the yield observed was 71%
compared to 92% yield for the homogeneous system, which was
unexpected (based on conversion of oxidant, alkene in excess).
Material 9 showed low conversion levels and activity in the epox-
idation of COE. This is surprising when compared to material 7,
however we believe it to be as a result of inter-porphyrin site inter-
2.7. ‘Heterogeneous’ oxidation catalysis with Mn–porphyrin materials
7 and 9
Materials 7 and 9 were applied as heterogenised catalysts in the
epoxidation of various substrates. Various oxidants were used to
monitor the effects of the oxidant properties on the stability and
reusability of the heterogenised catalysts. Materials 7 and 9 were
initially applied in the epoxidation of cyclooctene (COE) using iod-
osylbenzene (PhIO) as oxidant, and with imidazole introduced as a
catalytic co-ligand (Table 2). Material 7 showed diminished activ-
ity compared to its homogeneous analogue, [MnCl(TPP)] (230 cf.
actions, resulting in intermolecular
l-oxo-bridged dimer forma-
tion (see following section). In fact the colour of the material 9
was quite brown compared to the dark green hue it had prior to
catalysis. This would suggest MnO2 formation. In all cases the ki-
netic profile of the catalysed epoxidation reaction is the same as
the kinetic profile of the homogeneous catalyst, [MnCl(TPP)] (see
Fig. 6).
Material 7 was applied as an epoxidation catalyst using other
substrates to demonstrate the general applicability of these cata-
lytic materials. Cyclohexene (CHE) and styrene (STE), which are
less reactive substrates to manganese porphyrin catalysed epoxi-
dation, were both tested using iodosylbenzene as oxidant. CHE
was converted to cyclohexene epoxide (CHO) in very ordinary
yields (ꢀ10%) with a TOF of 7.5 hÀ1. Likewise STE was converted
to styrene-epoxide (STO) in average yield (47%), however with a
much higher TOF (58 hÀ1) (see Table 3).
After the epoxidation of COE material 7 was recycled by simple
filtration of the catalyst and subsequent washing of the solid rem-
nants with CH2Cl2 and MeOH. A fresh solution of substrate, oxi-
dant, and imidazole was then introduced to the catalytic
material, and the catalysed reaction was restarted. In the epoxida-
tion of COE catalyst recycling was carried out over four cycles. After
every cycle the TOF decreased slightly and the yield decreased con-
siderably, such that by the fourth cycle the TOF was under 20 hÀ1
and the yield was below 10% (Table 2). The post-catalysis filtrate
was charged with extra alkene and oxidant, however showed no
extra turnovers. This demonstrated that no Mn–porphyrin was
leaching into the reaction solution. The observed gradual decrease
in TOF and yields was a cause for concern, and thus we investi-
gated the effects of other oxidants on the stability and durability
Table 2
Application of materials 7 and 9 in the epoxidation of COE.
Catalyst
Yield epoxide (%)a
TOFb
[MnCl(TPP)]
7
9
92
71
13
230
82
8.5
a
Based on iodosylbenzene consumed.
b
Based on rate of epoxide formation (hÀ1).
Fig. 5. Solid-state CP-MAS 29Si NMR spectrum of material 7.

A.R. McDonald et al. / Journal of Organometallic Chemistry 694 (2009) 2153–2162
2159
2.8. Catalysts deactivation discussion
Supported Mn-complex deactivation is a very common occur-
rence in epoxidation catalysis. Several reports have commented
on Mn–salen systems bound to supports which show gradual deac-
tivation upon recycling of the catalytic materials [15]. In these sys-
Fig. 6. Epoxidation of COE using catalytic materials 7 or 9.
tems, the deactivation was believed to be as
a result of
decomplexation of the Mn–salen complexes bound to the silica
support. Likewise Mn–porphyrins supported on polymeric sup-
ports have shown deactivation upon recycling in epoxidation catal-
ysis.[16] In some cases the deactivation is put down to bleaching of
the manganese–porphyrin complexes by intermolecular oxidation,
resulting in decomplexation. The precursor to the decomplexation
Table 3
Recycling of materials 7 in the epoxidation of cyclooctene
Cycle #a
PhIO TOFb
BPO TOFc
1
2
3
4
5
76
66
40
20
–
3.4
2.9
1.7
1.0
0.4
is a
l-oxo-bridged dimer which is also inactive in epoxidation
catalysis. In most cases a convincing reason for deactivation is
not given.
In the case of material 7, we do not believe that any intermolec-
ular interactions between porphyrin complexes on the silica sur-
face are occurring. Firstly no UV–Vis evidence for this was
observed, and secondly, the multi-dentate manner in which the
complex is bound to the surface probably means the porphyrin
ring lies in such a position that inter-porphyrin site interactions
are unlikely. Obviously this does not tell us about all porphyrin
sites on the silica surface, but gives an indication that most have
retained their structural integrity. However, the solid-state UV–
Vis technique employed is quite crude. In previous reports accu-
mulation of by-products was believed to deactivate the catalyst.
PhIO reacts with itself giving iodobenzene (PhI) and iodoxyben-
zene (PhIO2). It is believed that PhIO2, having low solubility in
the reaction solvent, accumulates on the support surface and
blocks access to the active site [17]. However, here, post-catalytic
elemental analysis shows no iodine content, thus proving conclu-
sively that iodoxybenzene plays no role in the deactivation of the
catalytic material. Activation of the active site by imidazole coordi-
nating to the manganese centre is possibly inhibited by the manner
by which the complex is bound to the silica support. The incoming
imidazole binds either on the face of the porphyrin which points
away or towards the silica surface and thus activates the opposite
axial site for oxidation. It is possible that if the imidazole binds the
face pointing away from the silica surface, that the active site may
be prone to interactions with the silica surface or functional groups
on the silica surface. Oxidation of these functional groups would
likely create a hydroxyl moiety which could bind the Mn site, or
if not bind, block access to the active site. However, we presently
have no concrete evidence for this hypothesis. Finally, magnetic
stir bars are used to stir the catalytic solution. This often leads to
pulverisation of the catalytic material, and possibly mechanical
decomposition of the silica material. As all spectroscopic
a
Catalyst recycled by simple filtration and washing.
Based on rate of iodosylbenzene consumption (hÀ1).
Based on rate of t-butylperoxide consumption (hÀ1).
b
c
of the catalytic material 7. The milder oxidising agent t-butylper-
oxide (BPO) was applied in the epoxidation of COE using material
7 as a catalyst. The initial reaction using BPO showed considerably
lower conversion levels and TOF’s compared to PhIO (Table 2). This
is to be expected and is due to the better oxidising powers of PhIO
compared to BPO. Material 7 was recycled in the same manner and
fresh substrate and BPO were added to the separated catalytic
material. However, a similar deactivation trend was observed.
Material 9 was not recycled because of the poor catalytic ability
of the material, and furthermore the visible decomposition of the
manganese–porphyrin complex on the silica surface.
The properties of material 7 after application as a catalyst in the
epoxidation of COE were investigated to gain insights into the deac-
tivation pathways. The elemental content of the post-catalysis
material showed that the Mn loading had not changed at all, even
after five catalyst cycles (0.1% Mn before catalysis, 0.1% after cataly-
sis). Solid-state UV–Vis and DRIFT-IRspectroscopy were usedto ana-
lysethematerialas well(Fig. 7). Inparticular, theUV–Vis spectrumis
comparable to that of material 7 and [MnCl(TPP)], however this
result must be tempered by the fact that the spectrum is very broad,
which thus limits the accuracy. Both spectra suggest that the
integrity of the manganese–porphyrin complex has been retained
and the integrity of the silica support has been maintained – no extra
silanol resonance were observed in the IR spectrum. The DRIFT-IR
difference spectrum show that nothing new has been added to the
silica surface, and that the triazole species remains intact.
Fig. 7. Post-catalysis solid-state UV–Vis and IR spectra of material 7.

2160
A.R. McDonald et al. / Journal of Organometallic Chemistry 694 (2009) 2153–2162
techniques indicate that both the manganese complex, and silica
support remain sound during catalysis, we propose that a combi-
nation of the above-mentioned decomposition pathways is the rea-
son for gradual catalyst deactivation over five cycles.
In the case of material 9 we did not attempt recycling experi-
ments because of the poor results observed. Material 9 gives max-
imum epoxide yields of 13%, and shows no catalytic turnovers
upon the addition of fresh substrate. We believe the deactivation
to be as a result of inter-porphyrin interactions that are resulting
in active site blockage and eventual decomplexation.
was colourless. The remaining purple solid was recrystallised from
CH2Cl2 and MeOH (yield: 4.7–8.2%). 1H NMR (CDCl3, ppm): d À2.84
(NH, s, 2H), 0.372 (Si-CH3, s, 36H), 7.86 (m-ArH, d, 3J = 7.80 Hz, 8H);
8.14 (o-ArH, d, 3J = 7.80 Hz, 8H); 8.81 (b-H, s, 8H). 13C NMR (CDCl3,
ppm): 0.253 (Si-CH3); 40 ((CH3)3Si-C„C); 51 ((CH3)3Si-C„C); 120
(Cmeso); 127 (ArC3,5); 128 (ArC4); 131 (Cb); 135 (ArC2,6); 139 (ArC1);
142 (C ); UV–Vis (CH2Cl2): kmax/nm (log e): 422 (5.78); 517 (4.38);
a
553 (4.17); 592 (3.87); 648 (3.87). MALDI-TOF (m/z): 1000.02 (calc.
999.55).
4.3. 5,10,15,20-Tetrakis-(4-(ethynyl-phenyl)porphyrinmanganese(III)
chloride (3)
3. Conclusions
To
a purple solution of 5,10,15,20-tetrakis-(4-[(trimethyl-
Metallo-porphyrins have been immobilised on silica using
‘click’ techniques. The ‘click’ immobilisation procedure is highly
efficient, and demonstrates a simple lab shelf to reaction vessel
procedure for the application of heterogenised homogeneous cata-
lysts. No side products are formed in the immobilisation proce-
dure, no activating reagents are required, and there are no traces
of the copper catalyst used in the coupling reaction. This click
immobilisation alleviates the necessity for extreme reaction condi-
tions, or even the use of base or acid to allow coupling between
complex and support. The developed materials contain complexes
bound to silica in both a monodentate and polydentate manner,
and a notable difference is observed between the two. Monoden-
tate appendage to the support allows freedom of movement of
the porphyrin complexes, resulting in inter-porphyrin interactions,
and deactivation. Polydentate immobilisation of porphyrin com-
plexes yields active catalytic materials. The catalytic materials be-
come deactivated over a number of catalyst recycles, and this is as
a result of side-product accumulation on the silica surface. The
polydentate immobilisation may enhance the deactivation of the
immobilised catalysts because the active site may be trapped be-
tween the silica surface and the porphyrin ring system.
silyl)ethynyl]phenyl)porphyrin (50 mg 0.05 mmol) in DMF (ca.
30 mL), Mn(OAc)2 Á 4H2O (250 mg, 1 mmol) was added and the
mixture was stirred at reflux temperature overnight, during which
it turned dark green. The solution was allowed to cool to room
temperature. It was added to an ice cooled NaCl-solution (10 g/
40 mL) and a precipitate a dark green precipitate formed that
was recovered by filtration. The green solid was dissolved in
dichloromethane and saturated NaCl-solution was added. After
stirring at room temperature for 3 h the organic layer was sepa-
rated dried over MgSO4. Cleavage of the TMS groups was checked
using MALDI-TOF. If the cleavage was not complete, an extra reac-
tion step was performed in which the dichloromethane was re-
moved in vacuo and the remaining green solid was dissolved in
THF (ca. 10 mL). A solution of tetra-butylammonium fluoride
(TBAF) in THF (1 M, 0.5 mL, 0.5 mmol) was added. The reaction
was quenched by the addition of water and a precipitate formed.
The liquid was removed by filtration after which the green solid
was dissolved in dichloromethane and dried over MgSO4. The
green solution was concentrated in vacuo before it was loaded
on a column containing dry, neutral alumina. After elution with
5% MeOH in CHCl3 a small purple fraction was collected, followed
by a green product fraction. The solvent was removed in vacuo,
4. Experimental
yielding 44% of a dark green solid. UV–Vis (CH2Cl2): kmax/nm (log e):
376 (4.44); 403 (439); 421 (4.39); 478 (4.72); 526 (3.52); 585
(3.68); 619 (3.75). MALDI-TOF (m/z): 763.75 (calc. 799.20; without
Cl 763.75).
4.1. General considerations
Standard Schlenk procedures under N2 were carried out
throughout. Reagents were used as supplied from Acros BV or Sig-
ma–Aldrich, unless otherwise stated. 1H, 13C and 31P solution NMR
was carried out on a Varian Inova 300 spectrometer or a Varian Ox-
ford AS400. CP-MAS NMR was carried out using a Bruker AV 750.
Elemental analyses were performed by Dornis und Kölbe, Mikro-
analytisches Laboratorium, Mülheim a. d. Ruhr, Germany. MS mea-
surements were carried out on an Applied Biosystems Voyager
DE-STR MALDI-TOF MS. IR spectra were recorded on a Perkin–El-
mer SpectrumOne FT-IR Spectrometer. GC analyses were per-
formed on a Perkin-Elmer AutosystemXL Gas Chromatograph.
Solution UV–Vis spectra were recorded on a Varian Cay 50 Scan
UV–Vis spectrophotometer and diffuse-reflectance UV–Vis spectra
were recorded on a Cary 5 UV–Vis – NIR spectrophotometer.
4.4. 5,10,15,20-Tetrakis-(1-hexyl-4-phenyl-1,2,3-triazolo)porphyrin-
manganese(III) chloride (10)
To a solution of 5,10,15,20-tetrakis-(4-(ethynyl-phenyl)porphy-
rinmanganese(III) chloride (50 mg, 0.075 mmol) in dry and degassed
CH3CN was added hexylazide (65
lL, 0.45 mmol), followed by
addition of 2,6-lutidine (8.5 L, 0.075 mmol) and a spatula tip of
l
[Cu(MeCN)4]PF6. The mixture was stirred at room temperature
for 3 days, after which the CH3CN was removed in vacuo. The
remaining green oil was dissolved in dichloromethane and washed
with water (2Â). After drying over MgSO4 the dichloromethane
was removed in vacuo, yielding a green oily solid 30%. UV–Vis
(CH2Cl2): kmax/nm (log e): 381 (5.32); 418 (5.40); 479 (5.56); 528
(3.59); 583 (3.65); 621 (3.72). MALDI-TOF (m/z): 1272.35 (calc.
4.2. 5,10,15,20-Tetrakis-(4-[(trimethylsilyl)ethynyl]phenyl)porphyrin
1307.95, –Cl 1272.50).
(2)
4.5. 5,10,15,20-Tetrakis-(4-[(trimethylsilyl)ethynyl]phenyl)porphyrin-
zinc(II)
A solution of 4-[(trimethylsilyl)ethynyl]-benzaldehyde (8.0 g,
39.5 mmol) in propionic acid (ca. 180 mL) was brought to reflux
temperature. Pyrrole (3.2 mL, 49 mmol) was added and the mix-
ture turned immediately black. The mixture was stirred at reflux
for 1 h after which it was allowed to cool to room temperature. A
black solid precipitated and the suspension was allowed to settle
overnight. The black liquid was removed by filtration and the
remaining black solid was washed with methanol until the filtrate
5,10,15,20-tetrakis-(4-[(trimethylsilyl)ethynyl]phenyl)porphy-
rin (63 mg, 0.08 mmol) was dissolved in CH2Cl2 (ca. 50 mL) and a
solution of Zn(OAc)2 Á 2H2O (365 mg, 1.66 mmol, 20 eq,) in MeOH
(ca. 5 mL) was added. After 2 h stirring at room temperature, the
solvent was removed in vacuo and the remaining purple-pink solid
was dissolved in CH2Cl2 (5 mL) and eluted through a silica column

A.R. McDonald et al. / Journal of Organometallic Chemistry 694 (2009) 2153–2162
2161
with CH2Cl2 as eluent. A pink fraction was collected. The pink solu-
tion was concentrated in vacuo and recrystallised from CH2Cl2/
hexanes (yield: 34.4%). 1H NMR (CDCl3, ppm): d 0.378 (Si-CH3, s,
36H), 7.88 (m-ArH, d, 3J = 7.50 Hz, 8H); 8.16 (o-ArH, d,
2-[dimethylamino)methyl]-1-thiophenolato-copper(I)
catalyst.
After 5 days stirring under N2 at room temperature, the solvent
was removed by filtration and the green solid was washed with
CH2Cl2 (2Â), water and hot MeOH (2Â) and dried in vacuo. The
immobilised porphyrin was then again suspended in dry toluene
and degassed for 15 min ethynylbenzene (0.5 mL) was added as
well as a spatula tip of copper(I) aminoarenethiolate catalyst and
the mixture was stirred for 3 days under N2 at room temperature.
The liquid was removed by filtration, followed by washing with
CH2Cl2 (2Â), hot MeOH (2Â) and water. The remaining solid was
dried in vacuo. The solid was once again suspended in toluene
and degassed for 15 min. Hexylazide (1 mL) was added together
with a small amount of copper(I) aminoarenethiolate catalyst
and the mixture was stirred for 3 days under N2 at room tempera-
ture. The liquid was removed by filtration and the solid was
washed with CH2Cl2 (2Â), hot MeOH (2Â) and hexanes. Drying in
vacuo yielded the final product. The loadings of the different cata-
lysts were quantified by elemental analysis.
3J = 7.20 Hz, 8H); 8.93 (b-H, s, 8H). UV–Vis (CH2Cl2): kmax/nm
:
423; 550; 589.
4.6. 5,10,15,20-Tetrakis-(4-(ethynyl)phenyl)porphyrin-zinc(II) (4)
TBAF (1 M in THF, 120 lL, 0.12 mmol) was added to a solution of
(5,10,15,20-tetrakis-(4-ethynylphenyl)porphyrin-Zn(II) (29.3 mg,
0.03 mmol) in dry THF (ca. 20 mL). After stirring at room tempera-
ture for 30 min, the reaction was quenched with water and a pink
solid precipitated. The solvent was removed by filtration, yielding a
pink solid (yield: 95%). 1H NMR (DMSO, ppm): d 4.46 (acetyleneH.
s,
4
H); 7.92 (m-ArH, d, 3J = 7.80 Hz, 8H); 8.20 (o-ArH, d,
3J = 8.10 Hz, 8H); 8.81 (b-H, s, 8H). UV–Vis (CH2Cl2): kmax/nm (log
e
):
422 (5.74); 548 (4.35); 587 (3.70). MALDI-TOF (m/z): 772.41 (calc.
772.16).
4.9.1. Material 7
4.7. 5-(4-[(Trimethylsilyl)ethynyl]phenyl)-10,15,20-
trisphenylporphyrin (5)
29Si MAS CP (ppm): d À111 (Q4 silica); À101 (Q3 silica); À61 (T2,
T3 RSi(O-)3); 16 (-CH2SiO-). UV-VIS: kmax/nm: 479; 526 (sh); 581;
619. IR (DRIFT, difference, cmÀ1): mCH 2967; mtriazole 1460–1440.
Elemental analysis found: C, 18.03; H, 0.95; N, 0.50; Mn, 0.10%.
A
solution containing 4-[(trimethylsilyl)ethynyl]-benzalde-
hyde (296 mg, 1.34 mmol, 1 equiv.) and benzaldehyde (0.8 mL,
7.8 mmol, 5.8 equiv.) in propionic acid (ca. 40 mL) was brought
to reflux temperature. Pyrrole (0.7 mL, 9.1 mmol, 6.8 equiv.) was
added and the mixture turned immediately black. The mixture
was stirred for 1 h after which it was allowed to cool to room tem-
perature. A black solid precipitated and the suspension was al-
lowed to settle overnight. The black liquid was removed by
filtration and the remaining black solid was washed with methanol
until the filtrate was colourless. The crude product was used in fur-
ther reaction steps.
4.9.2. Material 8
UV–Vis: kmax/nm: 428; 561; 605. IR (DRIFT, difference, cmÀ1):
mCH 2961; mtriazole 1460–1440. Elemental analysis found: C, 9.49;
H, 2.06; N, 0.96; Zn, 0.24%.
4.9.3. Demetallation of 8, material 11
TFA (30 lL, 0.42 mmol) was added to a suspension of X in
CH2Cl2 (ca. 5 mL) upon which the mixture turned green immedi-
ately. After stirring for 1 h at room temperature the mixture was
poured in saturated NaHCO3 (50 mL) solution. The mixture turned
pink again upon stirring and a pink solid precipitated. The liquid
was removed by filtration and the solid was dried in vacuo. UV–
Vis: kmax/nm: 424; 545; 584 (sh); 653. IR (DRIFT, difference,
cmÀ1): mNH 3449; mCH 2951. Elemental analysis found: C, 10.50;
H, 2.62; N, 0.61; Zn, 0.14%.
4.8. (5-(4-[(Trimethylsilyl)ethynyl]phenyl)-10,15,20-
trisphenylporphyrin)manganese(III)chloride (6)
The remaining solid (from porphyrin synthesis reaction) was
dissolved in DMF (ca. 40 mL) and Mn(OAc)2 Á 4 H2O (250 mg,
1 mmol) was added. The mixture was refluxed overnight. The
formed green solution was allowed to cool to room temperature.
It was added to an ice cooled NaCl-solution (10 g/40 mL) and a
precipitate was formed that was removed by filtration. The green
solid was dissolved in CH2Cl2 and saturated NaCl-solution was
added. After stirring at room temperature for 3 h the organic layer
was separated dried with MgSO4. The green solution was concen-
trated in vacuo before it was loaded on a silica column prepared
with a mixture of hexanes and CH2Cl2 (3:1 v/v). First, a purple
fraction was eluted followed by a small green fraction. Methanol
(5%) was added to the eluent and a green product fraction was
eluted, followed by a dark green band. The presence, in one of
the fractions, (5-(p-acetylenephenyl)-10,15,20-triphenylporphy-
rin)Mn(III)Cl was confirmed by MALDI-TOF. However, the same
fraction also contained [Mn(III)(TPP)Cl]. This could not be sepa-
rated from the other compound. The liquid was removed and the
green solid was dried in vacuo. The mixture of functionalised and
unfunctionalised porphyrin was used in the immobilisation pro-
cess. MALDI-TOF (m/z): 667.70 ([MnCl(TPP)]), 691.36 (calc.
727.13; without Cl 691.68).
4.9.4. Material 9
UV–Vis (CH2Cl2): kmax/nm: 461; 529; 572; 608. IR (DRIFT, differ-
ence, cmÀ1): marCH 3057; mCH 2964; mtriazole 1440–1420. Elemental
analysis found: C, 17.71; H, 1.59; N, 0.69; Mn, 0.05%.
4.10. Standard alkene epoxidation catalysis conditions
Cyclooctene, cyclohexene and styrene were distilled prior to
use. Cyclooctene and cyclohexene were filtered over neutral alu-
mina prior to every catalytic reaction to remove trace amounts of
epoxide. Iodosylbenzene was synthesised according to the litera-
ture procedure and t-butylperoxide was used as solution in decane.
Iodosylbenzene: all experiments were performed at room temper-
ature in small GC-vessels with magnetic stirring. All reagents were
added as solutions in CH2Cl2, except iodosylbenzene and the sup-
ported catalysts. Catalyst (0.25
with the desired substrate (0.5 mmol), imidazole (5
l
mol active species) was stirred
mol) and
l
iodosylbenzene (0.03 mmol) in CH2Cl2. Product formation was
monitored by taking small aliquots for GC analysis with 1,2-dibro-
mobenzene as internal standard. t-Butylperoxide: all experiments
were performed at room temperature in small GC-vessels with
magnetic stirring. All reagents were added as solutions in MeCN
except the supported catalysts. Substrate (0.3 mmol), catalyst
4.9. Procedure for porphyrin immobilisation
A porphyrin bearing free acetylene groups (50 mg) was dis-
solved in dry, degassed CH3CN (10 mL). TMS-capped-3-azidopro-
pyl-1-silica (500 mg) was added, together with a spatula tip of
(1
lmol), imidazole (0.02 mmol) and t-butylperoxide (0.02 mmol)
were stirred in dichloromethane (2 mL). Product formation was

2162
A.R. McDonald et al. / Journal of Organometallic Chemistry 694 (2009) 2153–2162
(b) J.R. Lindsay Smith, Y. Iamamoto, F.S. Vinhado, J. Mol. Catal. A: Chem. 252
(2006) 23–30.
[7] A.R. McDonald, H.P. Dijkstra, B.M.J.M. Suijkerbuijk, M. Lutz, A.L. Spek, G.P.M.
van Klink, G. van Koten, submitted for publication.
checked by taking small aliquots for GC analysis with phenyl bro-
mide as internal standard.
[8] H.C. Kolb, M.G. Finn, K.B. Sharpless, Angew. Chem., Int. Ed. 40 (2001) 2004–
2021.
4.11. Catalyst recycling
[9] A.D. Adler, F.R. Longo, J.D. Finarelli, J. Goldmacher, J. Assour, L. Korsakoff, J. Org.
Chem. 32 (1967) 476.
[10] J. Rochford, D. Chu, A. Hagfeldt, E. Galoppini, J. Am. Chem. Soc. 129 (2007)
4655–4665.
[11] S. Brunauer, P.H. Emmett, E.J. Teller, J. Am. Chem. Soc. 60 (1938) 309–319.
[12] F. Odobel, E. Blart, M. Lagrée, M. Villieras, H. Boujtita, N. El Mur, S. Caramori,
C.A. Bignozzi, J. Mater. Chem. 13 (2003) 502.
After each oxidation step, the reaction solution was removed by
simple filtration and the catalytic material (solid) was washed with
methanol. Fresh reagents were added and a subsequent catalytic
experiment was performed.
[13] ATR-IR (attenuated total reflection) is a technique whereby transmission of an
IR beam through a sample is measured.
Acknowledgement
[14] DRIFT-IR (diffuse reflectance Fourier transform) is a surface technique which
analyses reflected IR beam from the surface/near surface of the material.
[15] (a) M.D. Angelino, P.E. Laibinis, Macromolecules 31 (1998) 7581–7587;
(b) T.S. Reger, K.D. Janda, J. Am. Chem. Soc. 122 (2000) 6929–6934.
[16] (a) P.R. Cooke, J.R. Lindsay Smith, J. Chem. Soc., Perkin Trans. 1 (1994) 1913–
1923;
The Dutch Technology Foundation (STW, DPC-5772) are kindly
acknowledged for financial support.
References
(b) F.G. Doro, J.R. Lindsay Smith, A.G. Ferreira, M.D. Assis, J. Mol. Catal. A:
Chem. 164 (2000) 97–108.
[17] (a) S. Camperstrini, B. Meunier, Inorg. Chem. 31 (1992) 1999;
(b) H.C. Sacco, Y. Iamamoto, J.R. Lindsay Smith, J. Chem. Soc., Perkin Trans. 2
(2001) 181–190;
[1] B. Cornils, W.A. Herrmann, J. Catal. 216 (2003) 23–31.
[2] D.E. De Vos, I.F.J. Vankelecom, P.A. Jacobs (Eds.), Chiral Catalyst Immobilization
and Recycling, Wiley-VCH, Weinheim, 2000.
[3] H.-U. Blaser, Chem. Commun. (2003) 293–296.
(c) M. Bonnet, L. Schmid, A. Baiker, F. Diederich, Adv. Funct. Mater. 12 (2002)
39–42.
[4] C. Baleizao, H. Garcia, Chem. Rev. 106 (2006) 3987–4043.
[5] E. Brule, Y.R. de Miguel, Org. Biomol. Chem. 4 (2006) 599–609.
[6] (a) P. Battioni, J.-P. Lallier, L. Barloy, D. Mansuy, Chem. Commun. (1989) 1149–
1151;