Metal oxide-triazole hybrids as heterogeneous or reaction-induced self-separating catalysts

Article abstract of DOI:10.1016/j.jcat.2016.06.005

The hybrid metal oxide-triazole materials [MoO₃(trz)_0.5] (1) and [W₂O₆(trz)] (2) (trz?=?1,2,4-triazole) have been hydrothermally synthesized and characterized by different techniques (TGA, SEM, ¹H and ¹³C MAS NMR, FT-IR spectroscopy, and structure determination by Rietveld analysis of high resolution synchrotron powder XRD data). Materials 1 and 2 display distinct behaviors when applied as catalysts for oxidation reactions with alcohol, aldehyde, olefin and sulfide substrates, and are more effective with hydrogen peroxide as the oxidant than with tert-butylhydroperoxide. The Mo^VI hybrid 1 transforms into soluble active species during cis-cyclooctene epoxidation with H₂O₂. When consumption of H₂O₂ reaches completion, spontaneous reassembly of the 2-dimensional molybdenum oxide network of 1 takes place and the hybrid precipitates as a microcrystalline solid that can be easily separated and recycled. Reaction-induced self-separation behavior occurs with 1, H₂O₂ and other substrates such as methyl oleate and methylphenylsulfide. The W^VI hybrid 2 behaves differently, preserving its structural features throughout the heterogeneous catalytic process.

Full text of DOI:10.1016/j.jcat.2016.06.005

Journal of Catalysis 340 (2016) 354–367
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Metal oxide-triazole hybrids as heterogeneous or reaction-induced
self-separating catalysts
1
1
⇑
⇑
Tatiana R. Amarante , Patrícia Neves , Anabela A. Valente , Filipe A. Almeida Paz , Martyn Pillinger,
Isabel S. Gonçalves
⇑
Department of Chemistry, CICECO – Aveiro Institute of Materials, University of Aveiro, Campus Universitário de Santiago, 3810-193 Aveiro, Portugal
a r t i c l e i n f o
a b s t r a c t
Article history:
The hybrid metal oxide-triazole materials [MoO3(trz)0.5] (1) and [W2O6(trz)] (2) (trz = 1,2,4-triazole) have
been hydrothermally synthesized and characterized by different techniques (TGA, SEM, 1H and 13C MAS
NMR, FT-IR spectroscopy, and structure determination by Rietveld analysis of high resolution syn-
chrotron powder XRD data). Materials 1 and 2 display distinct behaviors when applied as catalysts for
oxidation reactions with alcohol, aldehyde, olefin and sulfide substrates, and are more effective with
hydrogen peroxide as the oxidant than with tert-butylhydroperoxide. The MoVI hybrid 1 transforms into
soluble active species during cis-cyclooctene epoxidation with H2O2. When consumption of H2O2 reaches
completion, spontaneous reassembly of the 2-dimensional molybdenum oxide network of 1 takes place
and the hybrid precipitates as a microcrystalline solid that can be easily separated and recycled. Reaction-
induced self-separation behavior occurs with 1, H2O2 and other substrates such as methyl oleate and
methylphenylsulfide. The WVI hybrid 2 behaves differently, preserving its structural features throughout
the heterogeneous catalytic process.
Received 26 April 2016
Revised 3 June 2016
Accepted 4 June 2016
Keywords:
Organic–inorganic hybrid materials
Molybdenum oxide
Catalytic oxidations
Hydrogen peroxide
Catalyst self-separation
Ó 2016 Elsevier Inc. All rights reserved.
1. Introduction
of the oxidant led to spontaneous precipitation of the catalyst
which could be recovered and used again. Later, Dioumaev and
An important goal in transition metal catalysis science is the
development of systems that combine the advantages of homoge-
neous catalysts (high activities and selectivities, absence of diffu-
sion limitations, ability to fine-tune performance through ligand
modification) with those of heterogeneous catalysts (ease of
catalyst-product separation and catalyst recycling) [1,2]. The prin-
cipal approach to solving this problem has been the immobiliza-
tion of molecular catalysts onto organic, inorganic or hybrid
organic–inorganic supports [3–6]. However, the resultant materi-
als suffer from a number of drawbacks, which include catalyst
‘‘leaching”, leading to deactivation, and poor accessibility of the
active sites for the substrate. In the period 2001–2003 two papers
reported on a new approach involving catalyst self-separation or
self-precipitation. The first report, by Xi et al., described reaction-
controlled phase-transfer catalysis for propylene oxidation to
propylene oxide [7]. The insoluble W-based catalyst precursor
formed a soluble active species by the action of H2O2. Exhaustion
Bullock developed an alternative solid–liquid–solid phase separa-
tion procedure (for the solvent-free hydrosilylation of carbonyl
compounds) that relied on differences in solubility of the catalyst
in the liquid substrate and product [8]. The tungsten carbonyl cat-
alyst was initially soluble in the polar substrate but then precipi-
tated upon complete conversion of the substrate and formation
of the non-polar liquid product. Since the publication of these
two landmark papers, progress on self-precipitating transition
metal catalysts has been slow. With some exceptions, such as a
redox-switchable phase-tagged ruthenium-based catalyst [9], the
solid–liquid–solid phase separation procedures are either
reaction-controlled [10–20] or thermoregulated [21–24], and
involve polyoxometalate salts of large organic cations [25].
In the present work, we describe a new type of self-separating
catalyst based on a molybdenum oxide hybrid material. The mate-
rial [MoO3(trz)0.5] (trz = 1,2,4-triazole), which was first reported by
Zubieta and co-workers [26], was chosen for study as part of our
ongoing investigations into the catalytic properties of molybde-
num(VI) and tungsten(VI) oxide-organonitrogen hybrid materials
[27–32]. Depending on the structure and composition of these
hybrids, as well as the catalytic reaction conditions, the materials
typically act either as sources of soluble active species or (more
⇑
Corresponding authors.
E-mail addresses: atav@ua.pt (A.A. Valente), filipe.paz@ua.pt (F.A.Al. Paz),
igoncalves@ua.pt (I.S. Gonçalves).
1
These authors contributed equally to this work.
http://dx.doi.org/10.1016/j.jcat.2016.06.005
0021-9517/Ó 2016 Elsevier Inc. All rights reserved.

T.R. Amarante et al. / Journal of Catalysis 340 (2016) 354–367
355
rarely) as heterogeneous catalysts. For example, when the materi-
als [Mo2O6(2-(1-pentyl-3-pyrazolyl)pyridine)] [30], [Mo3O9(2-
[3(5)-pyrazolyl]pyridine)] [31] and [MO3(2,20-bipy)] (M = Mo,
W; 2,20-bipy = 2,20-bipyridine) [32] are used in catalytic olefin
epoxidation with either tert-butylhydroperoxide (TBHP) or hydro-
gen peroxide as oxidant, they are irreversibly converted into
soluble oxodiperoxo complexes of the type [MO(O2)2(L)], which
are responsible for the homogeneous catalytic reaction of the
olefin. The efficient recovery and reuse of these complexes is not
trivial, typically requiring precipitation with organic solvents.
Herein, we describe an unprecedented behavior for the hybrid
material [MoO3(trz)0.5] when used in catalytic olefin epoxidation
with H2O2 as oxidant. In a manner similar to that reported
with polyoxometalate salts [7,10–20] a solid–liquid–solid phase
transfer takes place, with spontaneous reassembly and self-
precipitation of the original molybdenum oxide-triazole solid upon
completion of the reaction. Results with the molybdenum(VI)
hybrid are compared with those for the corresponding tungsten
(VI) compound, and the catalytic performances of both materials
have been further examined for the oxidation of benzyl alcohol
and benzaldehyde. The crystal structures of both hybrids have
been determined through Rietveld analysis of high-resolution syn-
chrotron X-ray diffraction data.
recorded using a Bruker Avance 400 spectrometer (9.4 T) at
100.62 MHz with 3.25
1H 90° pulses, 2 ms contact time, spin-
ning rates of 10–12 kHz, and 3–4 s recycle delays. Chemical shifts
ls
are quoted in parts per million (ppm) from tetramethylsilane.
2.2. [MoO3(trz)0.5] (1)
A mixture of MoO3 (0.14 g, 0.97 mmol), 1,2,4-triazole (0.05 g,
0.72 mmol) and water (15 mL) was heated in a rotating (15 rpm)
Teflon-lined stainless steel digestion bomb at 180 °C for 1 h. After
cooling down to ambient temperature, the resultant bluish white
microcrystalline solid was separated from the aqueous liquor by
filtration, washed with an excess of water and diethyl ether
(4 ꢁ 10 mL), and finally dried at ambient temperature. Yield:
0.14 g, 83% (based on Mo). Anal. Calcd for CH1.5N1.5O3Mo: C,
6.73; H, 0.85; N, 11.77. Found: C, 6.65; H, 1.13; N, 11.73. FT-IR
(KBr, cmꢀ1):
m = 329 (m), 374 (s), 437 (m), 595 (vs, br), 626 (sh),
842 (vs, br), 937 (s), 964 (m), 1000 (m), 1058 (m), 1141 (m),
1176 (w), 1222 (w), 1253 (w), 1307 (m), 1425 (m), 1519 (m),
1779 (m), 2300–3000 (several weak bands due to trz), 3116 (s).
13C{1H} CP MAS NMR: d = 145.8 ppm.
2.3. [W2O6(trz)] (2)
A mixture of H2WO4 (0.49 g, 1.96 mmol), 1,2,4-triazole (0.20 g,
2.90 mmol) and water (25 mL) was heated in a rotating (15 rpm)
Teflon-lined stainless steel digestion bomb at 180 °C for 93 h. After
cooling down to ambient temperature, the resultant yellowish
white microcrystalline solid was separated from the aqueous
liquor by filtration, washed with an excess of water and diethyl
ether (4 ꢁ 10 mL), and finally dried at ambient temperature. Yield:
0.46 g, 89% (based on W). Anal. Calcd for C2H3N3O6W2: C, 4.51; H,
0.57; N, 7.89. Found: C, 4.45; H, 0.55; N, 7.97. FT-IR (KBr, cmꢀ1):
2. Experimental
2.1. Materials and methods
For synthesis, MoO3 (Analar, BDH Chemicals, 99.5%), H2WO4
(puriss p.a., Fluka), 1,2,4-triazole (98%, Sigma–Aldrich), and diethyl
ether (puriss p.a., Sigma–Aldrich) were acquired from commercial
sources and used as received. For the catalytic experiments, the
substrates cis-cyclooctene (95%), anhydrous benzyl alcohol
(99.8%), benzaldehyde (P99%), cyclohexene (99%), trans-2-octene
(97%), DL-limonene (P95%), methyl oleate (99%), methylphenylsul-
fide (99%) and benzothiophene (95%) were obtained from Sigma–
Aldrich and used as received. Acetonitrile (99.9%, Panreac), ethanol
(99.9%, Carlo Erba), ethyl acetate (99.9%, Sigma–Aldrich), 5.5 M
tert-butylhydroperoxide in decane (Sigma–Aldrich), 70 wt.% aq.
tert-butylhydroperoxide (Sigma–Aldrich), and 30% aq. H2O2
(Sigma–Aldrich) were acquired from commercial sources and used
as received.
m
= 306 (vs), 372 (m), 406 (w), 426 (w), 447 (w), 622 (s), 684 (sh,
br), 732 (vs, br), 906 (s, br), 950 (m), 975 (m), 1002 (w), 1062
(m), 1145 (w), 1226 (w), 1247 (w), 1263 (w), 1311 (m), 1427
(m), 1513 (m), 1521 (m), 1533 (w), 1789 (w), 2500–3000 (several
weak bands due to trz), 3106 (s), 3436 (m, br). 13C{1H} CP MAS
NMR: d = 145.1 ppm.
2.4. Synchrotron powder X-ray diffraction studies
High-resolution synchrotron PXRD data for 1 and 2 were col-
lected at low temperature (100 K; cooling device from Oxford
Instruments) on the powder diffractometer assembled at ID22
[33] at the European Synchrotron Radiation Facility (ESRF), Greno-
ble, France. The beam line receives X-rays from the synchrotron
source, operating with an average energy of 6 GeV and a beam cur-
rent of typically 200 mA, from an undulator device. The high
signal-to-noise ratio of the data is due to the high brilliance of
the synchrotron beam in combination with a Si(111) crystal
multi-analyzer.
The monochromatic wavelength was fixed at 0.495958(7) Å and
calibrated against the Si standard NIST 640c [certified cell param-
eter a = 5.4311946(92) Å]. Hard X-rays were selected for data col-
lection in order to significantly reduce radiation damage, an
occurrence observed in previous investigations using related mate-
rials but under different experimental conditions [27,29,31]. Even
at low temperature the high brilliance of the synchrotron source
led to visible damage of the samples. To minimize such effects con-
secutive data collections were performed on fresh portions of the
samples by translating the capillaries by ca. 1.3 mm.
Elemental analysis for C, H, and N was performed at the Univer-
sity of Aveiro with a Leco TruSpec 630–200–200 analyzer. Routine
powder X-ray diffraction (PXRD) data were collected at ambient
temperature on a Philips Analytical Empyrean (h/2h) diffractome-
ter equipped with a PIXcel1D detector, with automatic data acqui-
sition
(X’Pert
Data
Collector
software
v4.2)
using
monochromatized Cu K
a
radiation (k = 1.5406 Å). Intensity data
were collected by the step-counting method (step 0.01°), in contin-
uous mode, in the ca. 5 6 2h 6 50° range. Scanning electron micro-
scopy (SEM) images were collected using
a
Hitachi S4100
microscope operating at 25 kV. Samples were prepared by deposi-
tion on aluminum sample holders followed by carbon coating
using an Emitech K 950 carbon evaporator. Thermogravimetric
analysis (TGA) was carried out using a Shimadzu TGA-50 instru-
ment, from ambient temperature to ca. 800 °C, under a continuous
stream of air at a flow rate of 20 mL minꢀ1, and a heating rate of
5 °C minꢀ1
.
FT-IR spectra were collected using KBr (Sigma–Aldrich, 99%, FT-
IR grade) pellets and a Mattson-7000 infrared spectrophotometer.
Attenuated total reflectance (ATR) FT-IR spectra were measured
using a Specac Golden Gate Mk II ATR accessory having a diamond
top plate and KRS-5 focusing lenses. Solid state 13C cross-
polarization (CP) magic-angle-spinning (MAS) NMR spectra were
Finely powdered samples of compounds 1 and 2 were placed
inside a Hilgenberg borosilicate glass capillary (ca. 0.9 mm in
diameter) which was spun during data collection to improve pow-
der averaging over the individual crystallites, ultimately removing

356
T.R. Amarante et al. / Journal of Catalysis 340 (2016) 354–367
textural effects such as preferred orientation. Data were collected
in continuous mode with accumulation times increasing with the
scattering angle. The counts of the six detectors (covering roughly
5.5° 2h) were rebinned and normalized to give the equivalent step
scans (0.002°) suitable for further structural analyses.
carrier gas. The identification of the reaction products was carried
out by GC–MS (GC-qMS Agilent Technologies, 6890 N Network GC
system, 5973 Network mass selective detector, equipped with a
DB-1 capillary column (30 m ꢁ 0.25 mm ꢁ 0.10
lm)), using He as
the carrier gas.
The collected high-resolution PXRD patterns were indexed
using the LSI-Index algorithm implemented in TOPAS-Academic
V5 [34,35] and a whole-powder-pattern Pawley fit allowed the
orthorhombic space groups Pbcm and P212121 to be unequivocally
confirmed as the most suitable for compounds 1 and 2, respec-
tively. Pattern indexation and symmetry search permitted, on the
one hand, to identify 1 as isotypical with that reported by Hagrman
et al. [26], despite the presence of a residual amount of an uniden-
tified crystalline impurity (ten excluded zones in Fig. 1). On the
other hand, this initial procedure revealed that the replacement
of MoVI by WVI (to isolate 2) induced some modifications on the
local symmetry elements of the compound. In bulk 2, a residual
amount of an unidentified impurity could also be observed as indi-
cated by the single excluded region in Fig. 2.
After a 24 h batch run of cis-cyclooctene (Cy) epoxidation, the
solid was separated from the reaction mixture by centrifugation
(3500 rpm), thoroughly washed with pentane or ethanol, and
vacuum-dried at 60 °C for 1 h. The recovered solid was reused
twice for Cy epoxidation under typical conditions (the initial mass
ratios of olefin:oxidant:metal compound were the same for all
runs).
Contact tests (ct) were carried out for 1 and 2 as follows. The
compound was treated with oxidant (H2O2) and solvent (CH3CN)
under similar conditions to those used for a typical batch run,
but without substrate, for 24 h at 70 °C. Subsequently, the mixture
was cooled to ambient temperature and centrifuged (3500 rpm).
The liquid phase (L) was passed through a filter equipped with a
0.2 lm nylon membrane (giving 1-L-ct and 2-L-ct for 1 and 2,
The two crystal structures were determined in TOPAS-Academic
V5 [34] by using a simulated annealing approach. The adopted
strategy combined the use of individual chemical moieties for
the crystallographically independent metal centers (Mo and W)
and individual (terminal and bridging) oxygen atoms, and also
the use of Fenske–Hall Z-matrices for the N,N-bridging organic
linkers. This complex strategy greatly facilitates the mobility of
all chemical entities inside the unit cell boundaries during the glo-
bal optimization processes, while also allowing the inclusion of
various antibump and distance restraints for the optimization
procedures.
Rietveld structural refinements were performed with TOPAS-
Academic V5 [34] using a Chebyshev polynomial throughout the
entire angular range to model the background contribution. The
peak shapes for the powder pattern were described using the fun-
damental parameters approach [36], with preferred orientation
effects being modeled using either a 4th or 8th order spherical har-
monics approach, for 1 and 2 respectively. The final Rietveld refine-
ment was performed considering the presence of crystalline
impurity phases contributing to both powder patterns: a total of
ten (for 1) and one (for 2) excluded zones were included in the final
models (see Figs. 1 and 2 for additional details).
respectively), and used for the homogeneous catalytic reaction of
Cy with H2O2; Cy was added to the liquid to give an initial sub-
strate concentration of 1.2 M, and this solution was stirred for
24 h at 70 °C.
The contact test with 1 led to a yellow solution (1-L-ct); partial
evaporation of this solution at 70 °C resulted in the precipitation of
a yellow solid denoted PS-(1-L-ct). The contact test with 2 led to a
biphasic solid–liquid mixture; the solid was recovered in a similar
fashion to that described above for catalyst reuse, giving 2-S-ct.
The solids PS-(1-L-ct) and 2-S-ct were tested as catalysts for the
reaction of Cy with H2O2 at 70 °C (the initial olefin:oxidant:metal
compound mass ratios were the same as those used under the typ-
ical reaction conditions). After 24 h catalytic batch runs with these
solids, biphasic solid–liquid mixtures were obtained; the solids
were recovered in a similar fashion to that described above for cat-
alyst reuse, giving (PS-(1-L-ct))-run1 and (2-S-ct)-run1,
respectively.
All of the recovered solids obtained in the catalytic and contact
tests were characterized by ATR FT-IR spectroscopy and/or PXRD.
Selected solids were also studied with SEM.
Table S1 in the Supplementary Material gathers all the details
pertaining to the ID22 high-resolution synchrotron PXRD data col-
lection, crystal data and structure refinement details for com-
pounds 1 and 2. CCDC 1436239 (1) and 1436238 (2) contain the
supplementary crystallographic data for this paper. These data
can be obtained free of charge from The Cambridge Crystallo-
graphic Data Centre via https://summary.ccdc.cam.ac.uk/struc-
ture-summary-form.
3. Results and discussion
3.1. Catalyst synthesis and characterization
The molybdenum oxide hybrid [MoO3(trz)0.5] (1) was obtained
in very good overall yield by the reaction of MoO3, 1,2,4-triazole
(trz) and H2O in the mole ratio of ca. 1:0.75:830 at 180 °C for 1 h
in a Teflon-lined digestion bomb. Previously, Zubieta and co-
workers reported the isolation of 1 as light yellow blocks in 74%
yield by the reaction of Na2MoO4, trz, NiCl2, and H2O in the mole
ratio 1:0.95:1.67:1465 at 200 °C for 136 h [26]. In the present
work, only a dynamic hydrothermal synthesis (i.e., with continu-
ous rotation of the bomb) led to complete reaction of MoO3 to give
1. The corresponding tungsten oxide hybrid, [W2O6(trz)] (2), which
has not been previously reported, was prepared in a similar fashion
with 89% yield by heating H2WO4, trz, and H2O in the mole ratio
1:1.5:1400 for 93 h at 180 °C.
TGA for compounds 1 and 2 showed no weight loss up to about
305 °C for 1 and 400 °C for 2 (Fig. S1 in the Supplementary Mate-
rial). Thermal decomposition of the organic ligand above these
temperatures leads to residual masses of 79.9% at 420 °C for 1
and 86.8% at 480 °C for 2, which are in excellent agreement with
the calculated values of 80.6% and 87.0% for the stoichiometric for-
mation of MO3 (M = Mo, W). Whereas the residual WO3 undergoes
2.5. Catalytic tests
Catalytic reactions were carried out at 70 °C in 5 mL borosilicate
reactors equipped with valves for sampling and a PTFE magnetic
stirring bar. Typically, an amount of hybrid compound equivalent
to 18 lmol of Mo or W, 1.8 mmol of substrate (1.2 M), and 1 mL
of cosolvent (CH3CN) was added to the reactor, which was then
immersed in a temperature-controlled oil bath. After stirring
(1000 rpm) for 10 min, oxidant (2.75 mmol) was added, and the
reaction time was counted from this instant.
The evolution of the catalytic reactions was monitored by gas
chromatography (GC). Individual experiments were performed
for a given reaction time; prior to sampling, the reactors were
cooled to ambient temperature. The samples were analyzed using
a
Varian 3900 GC equipped with
a DB-5 capillary column
(30 m ꢁ 0.25 mm ꢁ 0.25 m) and a FID detector, with H2 as the
l

T.R. Amarante et al. / Journal of Catalysis 340 (2016) 354–367
357
Fig. 1. Final Rietveld plot of [MoO3(trz)0.5] (1). Observed data points are indicated as a blue line, and the best fit profile (upper trace) and difference pattern (lower trace) are
drawn as solid red and gray lines, respectively. Blue vertical bars indicate the angular positions of the allowed Bragg reflections for 1. Refinement details are given in Table S1
in the Supplementary Material. The inset depicts a perspective view along the [100] direction of the unit cell.
Fig. 2. Final Rietveld plot of [W2O6(trz)] (2). Observed data points are indicated as a blue line, and the best fit profile (upper trace) and difference pattern (lower trace) are
drawn as solid red and gray lines. Blue vertical bars indicate the angular positions of the allowed Bragg reflections for 2. Refinement details are given in Table S1 in the
Supplementary Material. The inset depicts a perspective view along the [001] direction of the unit cell.
no further weight loss up to 800 °C, the residual MoO3 starts to
sublime above 700 °C.
the latter band is m(NH). Such relatively low frequencies for m
(NH) are possible when the group is involved in a significant
hydrogen bonding interaction [38]. As will be described below, this
is the case for 1 (involving an interaction between the NAH group
of a triazole of one layer and a bridging oxo group of an adjacent
layer), but not for 2. Therefore the assignment of the sharp band
SEM of 1 showed that the morphology consists of rectangular
blocks with typical widths in the range of 1.5–2.5
lm and lengths
in the range of 5–9 m (Fig. 3). By contrast, compound 2 exhibits a
l
needle-like morphology, quite similar to that found recently for
[WO3(2,20-bipyridine)] [32].
to
1 and 2 display some notable differences. The hybrid 1 exhibits a
strong band at 937 cmꢀ1, assigned to
(Mo@O), and two very
strong, broad bands at 595 and 842 cmꢀ1
assigned to
(MoAOAMo). The corresponding bands for 2 can be tentatively
identified as a medium intensity band at 950 [ (W@O)] and a very
strong, broad band at 732 cmꢀ1, which has a broad shoulder
m
(CH) seems more plausible. Below 1000 cmꢀ1, the spectra of
The FT-IR spectra of 1 and 2 are similar in the range of 1000–
3200 cmꢀ1, displaying bands that are attributed mainly to the tri-
azole ligand (Fig. S2 in the Supplementary Material). In particular,
we may single out medium-to-strong bands in the 1500–
m
,
m
1540 cmꢀ1 range assigned to
3110 cmꢀ1 assigned to
m
(C@N) and a sharp band at about
m
m
(CH) [37]. An alternative assignment for

358
T.R. Amarante et al. / Journal of Catalysis 340 (2016) 354–367
Fig. 3. SEM images of (a) compound 1, (c) 1/H2O2-run1, (d) PS-(1-L-ct), (e) compound 2, (g) 2/H2O2-run1, and (h) 2-S-ct. The EDS elemental distribution maps corresponding
to the images (a) and (e) are shown in (b) and (f), respectively.
centered at ca. 680 cmꢀ1
[m(WAOAW)]. Both of the spectra for 1 and
data collected at the ESRF (Grenoble, France). As clearly observed
in both collected patterns, the materials contain trace amounts of
crystalline impurities (Figs. 1 and 2), which remain unidentified
to date. None of the impurities correspond to the starting materials
used in the synthesis or possible secondary products with crys-
talline structures deposited in the Cambridge Structural Database.
Topologically, the replacement of MoVI centers by WVI does not
affect the main structural features of the materials, with the most
striking motif present in both crystal structures being neutral two-
2 display a sharp band at about 625 cmꢀ1 assigned to a triazole ring
torsion vibration [39]. In a study of the infrared spectra of several
transition metal triazole complexes, Haasnoot et al. concluded that
the presence of one ring torsion band in the 600–700 cmꢀ1 range,
typically around 630 cmꢀ1, is indicative of 4-H,1,2-coordinating tria-
zole rather than 1-H,2,4-coordinating triazole [39], and therefore the
spectra for 1 and 2 are consistent with the presence of metal centers
bridged by 1,2-bicoordinating triazole ligands.
The solid-state 1H and 13C{1H} MAS NMR spectra of 1 and 2
were very similar, which is consistent with both compounds hav-
ing the same type of topological motif containing bridging trz
units. Hence, the 13C{1H} CP MAS NMR spectra display a single
sharp resonance in the 145–146 ppm range attributed to the car-
bon atoms C3 and C5, which are chemically equivalent due to
the N1,N2 bridging mode adopted by the ligand (Fig. S3 in the Sup-
plementary Material). The 1H MAS NMR spectra display one reso-
nance in the 9.6–9.9 ppm range and another in the 13.7–
14.6 ppm range assigned to the CH (H3/H5) and NH protons,
respectively. An additional signal at 5.1 ppm can be attributed to
residual amounts of physisorbed water molecules.
2
dimensional ½MoO3ðtrzÞ0:5ꢂ and 21½W2O6ðtrzÞꢂ layers placed in the
1
ac and bc planes of the respective unit cells (see insets in Figs. 1 and
2). The major differences lie in the small crystallographic details.
While 1 is based on a single crystallographically independent Mo
center with the asymmetric unit containing only half of one
1,2,4-triazole molecule (the mirror plane of the Pbcm space group
bisects the organic bridge as depicted in Fig. 4a), the modification
of the space group for 2 (while the Laue symmetry remains identi-
cal) leads to a different asymmetric unit containing instead two
metal centers and an entire organic linker (Fig. 4b). Attempts to
model 2 with the same symmetry elements as those found in the
Mo-based material failed, clearly indicating that the material has
a completely distinct space group (this was true for XRD data col-
lected at both 100 K and at ambient temperature using standard
laboratory equipment). We attribute this modification in the space
group to the small, but significantly larger, effective ionic radius of
hexacoordinated WVI which, ultimately, leads to a distinct crystal
3.2. High-resolution synchrotron PXRD studies
The structures of compounds 1 and 2, which were isolated as
microcrystalline powders, were investigated on the basis of PXRD

T.R. Amarante et al. / Journal of Catalysis 340 (2016) 354–367
359
2
2
Fig. 4. Schematic representation of portions of the 2D neutral ½MoO3ðtrzÞ0:5ꢂ (a) and ½W2O6ðtrzÞꢂ (b) layers present in the crystal structures of 1 and 2, respectively,
1
1
emphasizing the crystallographically independent Mo and W coordination environments and the coordination mode of the trz ligand. For 1, the mirror plane along the c-axis
bisects the organic ligand (indicated by the dashed gray line); for 2, the local symmetry is changed with displacement of the organic ligand and increased distortion of the
coordination polyhedra. Selected bond lengths and angles are given in Tables 1 and 2. Symmetry transformations used to generate equivalent atoms: For 1, (i) ꢀ1 + x, y, z; for
2, (i) x, y, ꢀ1 + z, (ii) 1 ꢀ x, ꢀ½ + y, 1.5 ꢀ z.
Table 1
packing. This additional distortion is also clearly noticeable when
Selected bond lengths (Å) and angles (degrees) for the crystallographically indepen-
attention is focussed on the bridging organic linker itself
(Fig. 4b): when looking at analogous positions within the neutral
dent Mo metal center present in [MoO3(trz)0.5] (1).a
Mo1AO1
Mo1AO2
Mo1AO3
Mo1AO4
Mo1AO1i
Mo1AN1
1.739(2)
1.677(2)
1.9250(13)
1.8855(6)
2.218(2)
2.331(3)
O1AMo1AO2
O1AMo1AO3
O1AMo1AO4
O1AMo1AN1
O1AMo1AO1i
O1iAMo1AO2
O1iAMo1AO3
O1iAMo1AO4
O1iAMo1AN1
O2AMo1AO3
O2AMo1AO4
O2AMo1AN1
O3AMo1AO4
O3AMo1AN1
O4AMo1AN1
104.51(11)
96.57(13)
96.91(13)
90.03(11)
167.21(10)
88.23(10)
80.65(12)
81.54(12)
77.18(9)
96.43(15)
100.51(9)
164.40(11)
154.94(14)
75.97(14)
82.98(7)
layers, in 2 the bridging 1,2,4-triazole is tilted and closer to one
metal center.
The coordination environments of all metal centers in 1 and 2
closely resemble distorted octahedra, in which each metal center
is bound to four l2-bridging oxo groups, one terminal oxo moiety
and one nitrogen atom of the N,N-bridging trz ligand. The terminal
oxo group appears trans coordinated to the organic linker, clearly
exerting the well-known trans effect and, thus, contributing to
the overall distortion of the {MoNO5} and {WNO5} octahedra. For
1, the MoA(N,O) bond distances range from 1.677(2) to 2.331
(3) Å, and the cis and trans (N,O)AMoA(N,O) octahedral internal
angles are found in the 77.18(9)–104.51(11)° and 154.94(14)–16
7.21(10)° ranges, respectively (Table 1 and Fig. 4a). These refined
values are in very good agreement with the geometrical data
reported by Zubieta and co-workers for a single-crystal XRD study
of the structure of 1 [26], with the analogous intervals being
1.700(3)–2.354(3) Å, 75.95(15)–104.80(15)° and 153.15(17)–167.
26(19)°, respectively. The two crystallographically independent
{WNO5} octahedra present in 2 (W1/W2 – Fig. 4b) have relatively
similar coordination environments: the WA(N,O) bond distances
were found in the 1.63(2)–2.44(2)/1.692(17)–2.34(3) Å ranges,
and the cis and trans (N,O)AWA(N,O) octahedral angles refined
within the 72.7(7)–104.2(8)/76.7(19)–105.2(7)° and 156.0(8)–
168.7(9)/158.8(9)–164.9(7)° intervals, respectively (Table 2). As
mentioned above, the distinct W1AN and W2AN bond distances
are a clear consequence of the local distortion associated with
the way the organic ligand bridges neighboring metals.
a
Symmetry transformation used to generate equivalent atoms: (i) ꢀ1 + x, y, z.
Table 2
Selected bond lengths (Å) and angles (degrees) for the two crystallographically
independent W metal centers present in [W2O6(trz)] (2).a
W1AO1
W1AO2
2.090(18)
1.80(3)
1.844(15)
2.046(15)
1.63(2)
2.44(2)
158.3(9)
90.7(6)
98.2(6)
104.2(8)
74.4(9)
79.3(8)
84.4(8)
96.9(11)
85.9(12)
156.0(8)
72.7(7)
88.5(7)
102.8(10)
96.6(9)
W2AO1
1.692(17)
1.99(3)
1.80(2)
1.964(15)
1.874(15)
2.34(3)
158.8(9)
96.3(8)
W2AO2ii
W1AO3
W2AO5
W1AO3i
W2AO6
W1AO4
W1AN1
W2AO6i
W2AN3
O1AW1AO2
O1AW1AO3
O1AW1AO3i
O1AW1AO4
O1AW1AN1
O2AW1AO3
O2AW1AO3i
O2AW1AO4
O2AW1AN1
O3AW1AO3i
O3iAW1AN1
O3AW1AN1
O4AW1AO3
O4AW1AO3i
O4AW1AN1
O1AW2AO2ii
O1AW2AO5
O1AW2AO6
O1AW2AO6i
O1AW2AN3
O2iiAW2AO5
O2iiAW2AO6
O2iiAW2AO6i
O2iiAW2AN3
O5AW2AO6
O5AW2AO6i
O5AW2AN3
O6AW2AO6i
O6AW2AN3
O6iAW2AN3
75.8(6)
92.0(6)
To help compare these two isotypical materials and to further
evaluate a potential variation in the degree of distortion of the
metal octahedra between 1 and 2, we decided to employ an adap-
tation of the mathematical method proposed by Baur [40] to calcu-
late a distortion index (DI) for bonds and angles:
76.7(10)
104.0(10)
105.2(7)
82.9(7)
82.1(12)
102.1(9)
88.0(9)
163.6(12)
164.9(7)
90.7(11)
77.6(11)
P
n
i¼1jdðMetal ꢀ ðN; OÞÞi ꢀ dðMetal ꢀ ðN; OÞÞa eragej
v
DIBond
¼
P
ð1Þ
n
i¼1jdðMetal ꢀ ðN; OÞÞij
P
n
i¼1j\ððN; OÞ ꢀ Metal ꢀ ðN; OÞÞi ꢀ \ððN; OÞ ꢀ Metal ꢀ ðN; OÞÞideal
j
DIAngle
¼
P
n
168.7(9)
i¼1j\ððN; OÞ ꢀ Metal ꢀ ðN; OÞÞij
a
ð2Þ
Symmetry transformations used to generate equivalent atoms: (i) x, y, ꢀ1 + z;
(ii) 1 ꢀ x, ꢀ½ + y, 1.5 ꢀ z.

360
T.R. Amarante et al. / Journal of Catalysis 340 (2016) 354–367
where the subscripts i and average/ideal correspond to individual
and average/ideal bond lengths and angles (see Tables 1 and 2 for
individual data). As in our previous studies using these distortion
degrees, the value of DIAngle should be, for an octahedron, calculated
independently for the cis and trans angles. Although the model by
Baur [40] suggested that DIAngle should be calculated with respect
to the average values observed for the polyhedra, we shall compare
instead with the ideal values of an octahedron (i.e., 90° and 180° for
the cis and trans angles, respectively). Having this in mind, the
DIBond values for the MoVI and WVI (W1/W2) centers are 0.11 and
0.11/0.08, respectively. The corresponding DIAngle values for the
cis/trans angles are 0.09/0.11, 0.10/0.12 and 0.10/0.11 for Mo1,
W1 and W2, respectively. These results clearly demonstrate that,
as a whole, the coordination environments of all metal centers
belonging to both structures are roughly equally distorted.
as catalyst instead of 1, under similar reaction conditions. The
epoxidation reaction with MoO3 was slower (67% at 24 h) than
with 1, and furthermore the reaction mixture was always homoge-
neous and did not exhibit reaction-induced self-separating
behavior.
The catalytic results for 2/H2O2 were compared with literature
data for the hybrid material [WO3(2,20-bipy)]ꢃnH2O (n = 1–2) and
the oxodiperoxo complex [WO3(O2)2(2,20-bipy)] tested as catalysts
under identical Cy reaction conditions [32]. Thꢀe1latter two tungsten
compounds led to TON(24 h) = 98 molCy molMo, which is slightly
ꢀ1
higher than that for 2/H2O2 [TON(24 h) = 85 molCy mol ].
Mo
The superior catalytic results for (1 or 2)/H2O2 over (1 or 2)/
TBHP suggest that the two hybrid compounds are more efficient
in activating H2O2 than TBHP. This is interesting since the conver-
sion of H2O2 gives water as the coproduct, which is more eco-
friendly than tert-butanol, the coproduct of TBHP conversion.
The higher catalytic activity of 1/H2O2 when compared with 2/
H2O2 is somewhat unusual since molybdenum-based catalysts fre-
quently display a lower activity than equivalent tungsten-based
catalysts for olefin epoxidation in the presence of H2O2. The situa-
tion is often reversed when TBHP is the oxidant, with the
molybdenum-based catalysts being superior. Examples of com-
pounds that exhibit this behavior (i.e., better catalytic results with
Mo/TBHP and W/H2O2) include the homogeneous catalysts [MO2-
Cl2(OPMePh2)2] [44], [MO2Cl2(dppmO2)] (dppmO2 = bis(diphenyl
phosphineoxide)methane) [44], [Cp⁄2M2O5] [45], and [MO2L]
(L = diamine bis(phenolate) ligand) [46], oxodiperoxo complexes of
the type [MO(O2)2(L)] supported on mesoporous SBA-15 materials
[47], and the hybrid materials [MO3(2,20-bipy)]ꢃnH2O (n = 0–2)
[32]. As mentioned in the introduction, when aqueous H2O2 is used
as the oxidant, the latter materials are converted to the partially sol-
uble oxodiperoxo complexes [MO(O2)2(2,20-bipy)], which are the
active species responsible for the epoxidation reaction in homoge-
neous phase.
The major consequence of the different space groups for 1 and 2
concerns the way the neutral two-dimensional 21½MoO3ðtrzÞ0:5ꢂ and
2
½W2O6ðtrzÞꢂ layers close pack in the solid state. While the former
1
pack on top of each other such that a strong and directional
NAHꢃ ꢃ ꢃO interplanar hydrogen bond is established between adja-
cent layers (dashed green lines in Fig. 5a), the molecular geometry
of the latter does not permit the existence of such an interaction
(Fig. 5b). Indeed, while in 1 the NAH group of the bridging trz
points directly to a l2-bridging oxo group of the neighboring layer,
in 2 this moiety is instead pointing to the void space in between
metal centers, thus not allowing a hydrogen-bonding interaction
to be established.
3.3. Catalytic investigations
3.3.1. Epoxidation of cis-cyclooctene
Compounds 1 and 2 were tested as (pre)catalysts for the epox-
idation of cis-cyclooctene (Cy), chosen as a model substrate, at
70 °C using tert-butylhydroperoxide or hydrogen peroxide as oxi-
dant, and acetonitrile as cosolvent [41]. The catalytic performances
of 1 and 2 strongly depended on the type of oxidant used (Table 3).
With TBHP [either aqueous (TBHPaq) or decane (TBHPdec) solu-
tions] as oxidant, the reaction of Cy was sluggish, giving 20–35%
and 3–9% conversion for 1 and 2, respectively, at 24 h reaction,
with a selectivity of 100% for cyclooctene oxide (CyO).
The catalytic performances improved considerably when aq.
H2O2 was used as oxidant instead of TBHP, leading to 98% and
85% conversion for 1 and 2, respectively, at 24 h reaction, with
CyO again being the only product (Table 3, Fig. 6). The catalytic
performance of 1/H2O2 is superior to that reported for different
molybdenum-based compounds tested as catalysts under similar
Cy reaction conditions. Epoxide yields in the range of 54–81%
Computational studies have suggested that the reaction mech-
anism for Mo or W-catalyzed olefin epoxidation with H2O2 as oxi-
dant involves the primary activation of the oxidant, where an oxo
ligand (M@O) is protonated with concomitant formation of a
hydroxide-hydroperoxo intermediate possessing the moiety {M
a
(OH)(O ObH)} [48–50]. Higher activity of W species is sometimes
claimed and associated with a lower activation barrier for the
a
rate-limiting step involving the transfer of the O atom to the ole-
fin [48,50]. Specifically, natural bond orbital and natural popula-
tion analysis studies suggested that the tungsten center may
withdraw electron density from the
r bonding [OAO] orbital more
strongly than a molybdenum center, lowering the r⁄ [OAO] orbital
energy and thus facilitating nucleophilic attack of the olefin. There
is, however, added complexity since an alternative pathway exists
involving external Ob transfer accompanied by a proton transfer
(100% selectivity) and turnover numbers calculated at 24 h [TON
(24 h)] in the range of 54–81 molCy mol were obtained for the
from Ob to O , and the interaction strength between the W center
ꢀ1
a
Mo
a
complexes [MoO(O2)2(L)] and [Mo2O4(
4,40-di-tert-butyl-2,20-bipyridine), the octanuclear complex [Mo8-
24(4,40-di-tert-butyl-2,20-bipyridine], and the hybrid material
l
-O)Cl2(L)2] (L = 2,20-bipy,
and O may influence the Ob transfer process [49]. This pathway is
not possible for an intermediate possessing the moiety {M(OH)
a
O
(O ObtBu)} formed in the case of TBHP. The operative oxygen
{[MoO3(2,20-bipy)][MoO3(H2O)]}n [42]. High catalytic activity was
recently reported for the polyoxometalate salt [1-hexyl-3-
methylimidazolium]4Mo8O26 (98% Cy conversion at 1 h, 60 °C); a
relatively high molybdenum loading was used (12 molMo%, com-
pared to 1 molMo% in the present work), which led to ꢀa1turnover
frequency calculated at 1ꢀh1 [TOF(1 h)] of 8 molCy molMo hꢀ1 (for
1, TOF(1 h) = 20 molCy molMo hꢀ1) [20]. The hybrid material [Mo2-
O6(trpzH)(H2O)2] [trpzH = 4-(3,5-dimethyl-1H-pyrazol-4-yl)-1,2,4
-triazole] was recently investigated as a catalyst under identical
transfer mechanism will clearly play a major role in determining
the activities of Mo/W catalysts. On the other hand, an important
factor which is sometimes discarded is the catalyst stability which
may be influenced by the oxidizing conditions used.
In order to check whether compounds 1 and 2 possess catalase-
type activity leading to non-productive decomposition of H2O2
(into O2 and H2O), iodometric titrations were carried out for the
mixtures of (1 or 2)/H2O2/CH3CN (without olefin) after stirring
for 24 h at 70 °C. The H2O2 decomposition was ca. 90% for 1 and
60% for 2 (ca. 10% experimental error); the thermal decomposition
of H2O2 (without catalyst and olefin) was negligible. These results
indicate that compounds 1 and 2 possess catalase-type activity,
which may explain the pressure relief observed upon opening
Cy reactꢀio1n conditions, and led to a higher TOF (1 h) of 66 molCy
-
molMꢀ1o h than 1 (20 molCy mol hꢀ1), but the former did not
ꢀ1
Mo
exhibit reaction-induced self-separating behavior [43]. For com-
parison, a commercial sample of MoO3 (AnalaR, 99.5%) was tested

T.R. Amarante et al. / Journal of Catalysis 340 (2016) 354–367
361
2
2
Fig. 5. Side view schematic representation of the mutual distribution of the 2D neutral ½MoO3ðtrzÞ0:5ꢂ (a) and ½W2O6ðtrzÞꢂ (b) layers present in compounds 1 and 2,
1
1
respectively. The dislocation of the layers on top of their growing planes leads to the loss of the strong and directional NAHꢃ ꢃ ꢃO interplanar hydrogen bonds (dashed green
line indicated with an arrow) on going from 1 to 2.
the reactors (after cooling to room temperature), i.e. the excess
pressure was likely due to O2 formed from the decomposition of
H2O2. The final epoxide yield will depend at least partly on the rel-
ative rates of the Cy reaction (epoxidation activity) and H2O2
decomposition (catalase-type activity). The catalase-type activity
may contribute to the slowdown of the reaction observed after
6 h, especially in the case of 1 (Fig. 6). Nevertheless, fairly high
Cy conversions were reached after 24 h reaction (>85%).
Table 3
Epoxidation of cis-cyclooctene with different oxidants in the presence of 1 or 2.a
Sample
Oxidant
Cy conversion (%)
or TON (molCy mol
ꢀ1
b
)
Mo
1
1
TBHPdec
TBHPaq
H2O2
H2O2
H2O2
TBHPdec
TBHPaq
H2O2
H2O2
H2O2
35
20
98
62
59
9
1 (run1)
1/H2O2-run2
1/H2O2-run3
2
2
3
3.3.2. Catalyst stability and formation of active species
2 (run1)
2/H2O2-run2
2/H2O2-run3
85
81
80
For a typical batch run of Cy epoxidation at 70 °C with 1/H2O2,
the reaction mixture initially consisted of a white solid suspended
in a colorless liquid phase, which after several minutes turned into
a homogeneous transparent yellow solution (Fig. 7). After 24 h, the
phase characteristics of the mixture had reverted back to that
observed during the initial instants of the reaction, i.e. a white solid
suspended in a colorless liquid phase. For 2/H2O2 and the reactions
performed with 1 or 2 and TBHPaq/TBHPdec, these phase transfor-
mations were never observed, i.e. the reaction mixtures always
consisted of a white solid suspended in a colorless liquid phase.
For all reactions with 1 or 2 and the different oxidants, the reac-
tion mixtures were centrifuged after 24 h in order to recover the
solid phases. All of the recovered solids, which are referred to as
1/oxidant-run1 and 2/oxidant-run1 (oxidant = TBHPdec, TBHPaq or
H2O2), exhibited FT-IR spectra and PXRD patterns that were similar
to those of the corresponding original compounds 1 (Figs. 8 and 9)
and 2 (Figs. S5 and S6 in the Supplementary Material). On the other
hand, while SEM revealed an identical morphology for 2 and 2/
H2O2-run1, the morphologies for 1 and 1/H2O2-run1 were strik-
ingly different, changing from regular micro-sized parallelepipeds
for 1 to aggregates of smaller irregular particles for 1/H2O2-run1
(Fig. 3).
a
Initial mole ratios of metal:olefin:oxidant = 1:100:152, initial concentration of
Cy = 1.2 M, 1 mL CH3CN, 24 h reaction, 70 °C.
b
Cy conversion at 24 h reaction; the values of conversion coincide with those of
turnover number (TON), at 24 h; CyO selectivity was always 100%.
The solids 1/H2O2-run1 and 2/H2O2-run1 were reused for two
consecutive 24 h batch runs (Table 3). For 1, Cy conversion and
epoxide yield decreased between runs 1 and 2, but then remained
roughly constant at around 60%. For 2, the catalytic performance
was steady for the three runs (80–85% Cy conversion), which is
consistent with the similar FT-IR spectra (Fig. S5), PXRD patterns
(Fig. S6) and particle morphologies (Fig. 3) observed for as-
synthesized 2 and the recovered solids.
Fig. 6. cis-Cyclooctene epoxidation with H2O2 in the presence of 1 (d) or 2 (N) at
70 °C (CH3CN as cosolvent). The dashed lines are visual guides.

362
T.R. Amarante et al. / Journal of Catalysis 340 (2016) 354–367
Fig. 7. Photographs and corresponding GC traces of the epoxidation of cis-
cyclooctene with aq. H2O2 at 70 °C in the presence of 1 (CH3CN as cosolvent):
1 min reaction, suspension of 1; 2 h reaction, yellow homogeneous solution; 24 h
reaction, compound 1 has precipitated.
The observation of a solid–liquid–solid phase transfer for the
catalytic system 1/H2O2/Cy/CH3CN/70 °C, together with the fact
that as-synthesized 1 and 1/H2O2-runi (i = 1, 2, 3) exhibited similar
FT-IR spectra and PXRD patterns, suggest that during the catalytic
reaction compound 1 is transformed into a different species (sol-
uble and yellow in color) which, as the reaction approaches com-
pletion, is reversibly converted to a molybdenum oxide-triazole
hybrid of the type 1. This system therefore appears to exhibit the
features required for a reaction-controlled solid–liquid–solid phase
separation procedure. Several triazolylmolybdenum(VI) oxide
hybrids recently investigated did not exhibit this behavior [43].
For example, the solid [Mo2O6(trpzH)(H2O)2] (mentioned above)
dissolved completely in the reaction mixture but did not self-
precipitate [43].
Clearly, the type of oxidant influences the stability of 1, and, on
the other hand, H2O2 plays a role in the reversible formation of 1.
When 1 was mixed with Cy/CH3CN/H2O (without H2O2), and stir-
red for 24 h at 70 °C, no yellow coloring of the mixture was
observed, and the FT-IR spectrum of the recovered solid (not
shown here) matched that for 1. Hence, while 1 is hydrolytically
stable, in the presence of H2O2 it is transformed into a different
compound. On the other hand, no reaction of Cy occurred for the
Cy/CH3CN/H2O mixture, indicating that H2O2 acts as oxygen
atom-donor in the catalytic cycle. The reaction of Cy with H2O2
as limiting reagent (initial Cy:H2O2 mole ratio of 2.5) led to incom-
plete Cy reaction (Cy conversion at 24 h was 44% when based on
the initial number of moles of Cy, and 100% when based on the ini-
tial number of moles of H2O2). For this experiment, a yellow col-
ored liquid phase was initially obtained, but at 24 h the mixture
consisted of a white solid suspended in a colorless liquid. These
results suggest that compound 1 is reversibly formed upon the
complete consumption of the oxidant. Furthermore, under the nor-
mal catalytic reaction conditions with a starting catalyst:olefin:ox-
Fig. 8. ATR FT-IR spectra for the ligand 1,2,4-triazole (a), 1 (b) and the correspond-
ing recovered solids 1/TBHPdec-run1 (c), 1/TBHPaq-run1 (d), 1/H2O2-run1 (e), 1/
H2O2-run2 (f), 1/H2O2-run3 (g), PS-(1-L-ct) (h), PS-(1-L-ct)-run1 (i), 1-(methyl
oleate)/H2O2 (j), 1-PhCH2OH/H2O2 (k), and 1-PhMeS/H2O2 (l).
idant mole ratio of 1:100:152, the catalase-type activity discussed
above may contribute to the complete consumption of the oxidant.
To further understand the catalytic properties of 1 and 2, a ser-
ies of contact tests (without olefin) were performed. Compounds 1
or 2 were treated with oxidant (H2O2) and solvent (CH3CN) for 24 h
at 70 °C. Subsequently, the resultant liquid (1-L-ct, 2-L-ct) and
solid (2-S-ct; no solid phase was present with 1) phases were sep-
arated for further study. Catalytic tests with 1-L-ct and 2-L-ct (per-
formed at 70 °C after addition of the substrate Cy to the liquid
phases) indicated negligible reaction of Cy. The solution 1-L-ct
was yellow in color, suggesting that it contained dissolved molyb-
denum species. The poor results for 1-L-ct may therefore be due to
the very low initial concentration of oxidant (according to the
iodometric titrations). In the case of 2-L-ct, the poor results of
the catalytic test may be due to the absence of soluble active spe-
cies (the solution was colorless), suggesting that the catalytic reac-
tion with 2 is heterogeneous in nature. The catalytic test with 2-S-
ct gave similar results to those obtained with 2 (96% CyO selectiv-
ity at 85% conversion). These results are consistent with the similar
FT-IR spectra (Fig. S5), PXRD patterns (Fig. S6), and particle mor-
phologies (Fig. 3) of 2-S-ct and 2. Moreover, the 24 h batch run
with 2-S-ct led to a solid [(2-S-ct)-run1] which was recovered
and found to exhibit similar FT-IR (Fig. S5) and PXRD (Fig. S6) data
to those displayed by 2. Hence, taking into account the similar

T.R. Amarante et al. / Journal of Catalysis 340 (2016) 354–367
363
talline structural features (discussed above). Specifically, the
reversible transformation of 1 during the catalytic reaction seems
to be accompanied by a reduction in the average particle size
(Fig. 3).
As noted above, the FT-IR spectrum of PS-(1-L-ct) is clearly dis-
tinct from that for 1, showing differences in both the ligand modes
and the MoAO bands (Fig. 8). New, weak bands are present at 3505
and 3600 cmꢀ1, possibly due to NAH and/or OAH stretching vibra-
tions (the latter may arise from coordinating or crystallization
molecules of water). A comparison of the spectra for 1 and PS-(1-
L-ct) in the 1000–1550 cmꢀ1 range confirms that the organic ligand
is retained in the yellow species isolated from the contact test,
although some of the bands are shifted by up to 17 cmꢀ1 when
compared with the corresponding bands for 1. A new medium
intensity band at 1599 cmꢀ1 for PS-(1-L-ct) may be due to
m
(C@N) of triazole or, possibly, an NAH or water bending vibration
(if weakly hydrogen-bonded). Below 1000 cmꢀ1, the IR spectrum of
PS-(1-L-ct) displays two strong bands at 910 and 947 cmꢀ1, which
may both be due to
the 2D network structure of 1 to give PS-(1-L-ct) is evident from
the disappearance of the very strong, broad (MoAOAMo) bands
at 595 and 842 cmꢀ1. Instead, PS-(1-L-ct) exhibits a medium-
strong band at 860 cmꢀ1 and a very strong band at 528 cmꢀ1
which are attributed to (MoAO)peroxo and (OAO) vibrations,
m
(Mo@O) (cf. 937 cmꢀ1 for 1). The disruption of
m
,
m
m
respectively. Hence, PS-(1-L-ct) is most likely a mononuclear or
polynuclear oxo-peroxo molybdenum species. Although PS-(1-L-
ct) may be indicative of the type of active species present in solu-
tion under the normal catalytic reaction conditions, the actual spe-
cies present may fluctuate according to changes in the composition
of the reaction mixture caused by consumption of the oxidant and
conversion of Cy to CyO. This was evident when performing the
contact test with shorter reaction times (less than 24 h), which
typically led to the isolation of species with FT-IR spectra different
from that for PS-(1-L-ct).
The behavior of 1 in liquid-phase catalytic olefin epoxidation
parallels that recently reported for octamolybdate-based self-
separating catalysts of general formula [cation]4Mo8O26
Fig. 9. PXRD patterns for the ligand 1,2,4-triazole (a), 1 (b) and the corresponding
recovered solids 1/TBHPdec-run1 (c), 1/TBHPaq-run1 (d), 1/H2O2-run1 (e), 1/H2O2-
run2 (f), 1/H2O2-run3 (g), PS-(1-L-ct) (h), PS-(1-L-ct)-run1 (i), 1-(methyl oleate)/
H2O2 (j), 1-PhCH2OH/H2O2 (k), and 1-PhMeS/H2O2 (l).
(cation = 1-hexyl-3-methylimidazolium,
1,2-dimethyl-3-
hexylimidazolium, or 1-hexylpyridinium) [20]. Thus, with Cy as
substrate, the octamolybdate salts dissolved in the presence of
H2O2 and CH3CN to give yellow homogeneous solutions, possibly
containing oxodiperoxomolybdenum active species. As found for
1, self-precipitation of the starting octamolybdate salts took place
after the completion of reaction took place. Several mechanistic
hypotheses have been put forward in the literature (mainly based
on gas-phase density functional theory) for olefin epoxidation with
hydroperoxides in the presence of oxoperoxomolybdenum species;
characterization data and steady catalytic performance of the
solids reused in consecutive catalytic runs, compound 2 seems to
behave as a fairly stable heterogeneous catalyst for Cy epoxidation
with aq. H2O2.
A yellow solid [denoted PS-(1-L-ct)] was precipitated from the
solution 1-L-ct after concentrating the mixture by partial evapora-
tion of the solvent at 70 °C. In comparison with 1, the characteriza-
tion data for PS-(1-L-ct) were different (Figs. 3, 8 and 9). In a
catalytic test with PS-(1-L-ct), the solid completely dissolved and
led to a Cy conversion of 69% at 24 h, with 100% CyO selectivity.
Hence, the yellow compound possesses intrinsic activity as a
homogeneous catalyst. The lower conversion reached for PS-(1-L-
ct) in comparison with 1 (98% at 24 h, Table 3) may be partly
due to differences in the initial molar concentration of catalyst
since the catalytic tests were compared on the basis of the same
initial mass of solid. As found for 1, the catalytic test with PS-(1-
L-ct) gave a yellow solution during the early stages of the reaction,
but at 24 h the mixture had turned into a white solid suspended in
a colorless liquid phase. FT-IR and PXRD data for the recovered
solid [PS-(1-L-ct)-run1] were similar to those for 1 (Figs. 8 and
9). These results support the hypothesis that 1 is reversibly con-
verted into a different (yellow colored, soluble) compound during
the catalytic reaction. The reversibility of this process may partly
explain why 1 undergoes morphological changes during the cat-
alytic reaction, despite the retention of the chemical and crys-
reaction mechanisms may involve O-atom transfer from a
g
2-O2
peroxo ligand [51–53] or a hydroperoxo (or alkylperoxo) ligand
[52,54,55] to the olefin (which may or may not be coordinated to
the metal center).
3.3.3. Catalytic performance with different substrates
The liquid-phase oxidation of benzyl alcohol (PhCH2OH) to ben-
zaldehyde (PhCHO) is an important reaction since PhCHO is com-
monly used as a food flavorant, perfume ingredient, industrial
solvent, and as an intermediate in the synthesis of other organic
compounds, such as pharmaceuticals, plastic additives, and aniline
dyes [56]. Further oxidation of PhCHO gives benzoic acid (PhCO2H),
which is used as a preservative in food, drugs, and personal care
products, and as an intermediate in the synthesis of various chem-
icals such as e-caprolactam and terephthalic acid [57]. The value of
PhCHO and PhCO2H has motivated the development of catalytic
systems for the oxidation of PhCH2OH with aq. H2O2 [58,59].
The catalytic performances of 1 and 2 were explored for the oxi-
dation of PhCH2OH and PhCHO with H2O2 at 70 °C. The reaction of

364
T.R. Amarante et al. / Journal of Catalysis 340 (2016) 354–367
PhCH2OH gave PhCHO and PhCO2H, which were formed with a
total selectivity of 100% (Fig. 10). The dependence of product selec-
tivity on PhCH2OH conversion was roughly comparable for 1 and 2,
until ca. 75% conversion. Initially, PhCHO was the only product
(100% selectivity at 10–11% conversion), and subsequently PhCHO
selectivity decreased with the concomitant increase of PhCO2H
selectivity, leading to 45% PhCHO and 30% PhCO2H yield for 1,
and 40% PhCHO and 56% PhCO2H yield for 2, at 48 h. These results
are consistent with the conversion of PhCH2OH to PhCO2H via the
intermediate formation of PhCHO. The oxidation of PhCHO (as sub-
strate) with H2O2, in the presence of 1 or 2 at 70 °C, gave PhCO2H as
the only product (80% and 72% yield, respectively, at 24 h reaction,
Fig. 11). While for the reaction of PhCH2OH, 2 led to higher conver-
sions than 1 (96% and 75% conversion, respectively, at 48 h reac-
tion), the opposite was observed for the reaction of PhCHO (80%
and 72% conversion for 1 and 2, respectively, at 24 h). These results
may be due to different reaction mechanisms involved and/or dif-
ferent rates of substrate oxidation relative to H2O2 decomposition.
The 1/H2O2 system was further investigated for the conversion
of cyclohexene and trans-2-octene, and the biomass-derived ole-
fins DL-limonene and methyl oleate, at 70 °C (Table 4). In general,
the corresponding epoxide was formed, as well as the diol product
via hydrolysis of the oxirane ring, e.g. 59% cyclohexane-1,2-diol
selectivity at 90% cyclohexene oxide conversion, and 56%
limonene-1,2-diol selectivity at 55% limonene conversion. Hence,
the 1/H2O2 system promotes one-pot epoxidation and hydrolysis
reactions for olefins which give relatively reactive epoxides. Side-
reactions included allylic oxidation of CAH bonds to give alcohol
and carbonyl products, e.g. cyclohex-2-en-1-ol and cyclohex-2-
en-1-one were formed from cyclohexene, and 2-methyl-5-(prop-
1-en-2-yl)cyclohex-2-en-1-ol (carveol) and 2-methyl-5-(prop-1-
en-2-yl)cyclohex-2-en-1-one (carvone) were formed from DL
-
limonene. The ability of the 1/H2O2 system to convert alcohols to
the corresponding carbonyls is consistent with its ability to oxidize
PhCH2OH to PhCHO.
The catalytic versatility of the 1/H2O2 system was explored for
the oxidation of sulfides into sulfoxides or sulfones (Table 4). This
transformation has received much interest in the context of pro-
cesses for the oxidative desulfurization of liquid fuels [60]. More-
over, sulfoxides and sulfones are versatile intermediates in
organic synthesis [61]. The reaction of methylphenylsulfide
(PhMeS) with H2O2 in the presence of 1 led to 99% conversion at
6 h, and gave the corresponding sulfoxide and sulfone with 39%
and 61% selectivity, respectively. With benzothiophene (BzS), the
sulfone (1-benzothiophene 1,1-dioxide) was formed with 100%
selectivity at 72% conversion (24 h). The conversion of the sub-
strates to the corresponding sulfones involves the intermediate
formation of the sulfoxide [61]. For the two substrates, no oxida-
tion of the phenyl CAH bonds was observed.
Fig. 10. (A) Conversion of benzyl alcohol (d, s) and (B) selectivity to benzaldehyde
(N, 4) and benzoic acid (j, h) for the reaction of benzyl alcohol with H2O2 in the
presence of 1 (filled symbols) or 2 (open symbols) at 70 °C. The dashed lines are
visual guides.
There are few reports that describe the use of molybdenum
oxide-organic hybrids or oxoperoxo compounds as (pre)catalysts
for the oxidation of PhMeS and BzS with aqueous H2O2 [62–66].
Organotin molybdates of the type [(R3Sn)2MoO4]ꢃnH2O
(R = methyl, n-butyl, cyclohexyl, phenyl, benzyl) were tested as
catalysts for the BzS reaction with H2O2 at 35 °C (Mo:substrate:
H2O2 = 1:100:200) [62]. With CH3CN as solvent and the benzyl
derivative as catalyst, the corresponding sulfone was formed with
ca. 100% selectivity at 65% conversion (7 h). The oxodiperoxo com-
plex [MoO(O2)2(3-methylpyrazole)] was tested as catalyst for the
reactions of PhMeS and BzS in an ionic liquid medium at 20 °C
(Mo:substrate:H2O2 = 1:40:40) [63]. With PhMeS as substrate, the
corresponding sulfoxide was formed with 96% selectivity at 80%
conversion (1 h), while with BzS as substrate the sulfoxide was
formed with 1% selectivity at 65% conversion, with no formation
of sulfone being reported. A molybdenum based metallomicellar
catalyst, (C19H42N)2[MoO(O2)2(C2O4)]ꢃH2O, was tested with PhMeS
Fig. 11. Reaction of benzaldehyde with H2O2 in the presence of 1 (d) or 2 (s) at
70 °C. Selectivity to benzoic acid was always 100%. The dashed lines are visual
guides.
as substrate at room temperature, leading to 95% sulfoxide yield at
30 min (Mo:substrate:H2O2 = 1:40:40); increasing the amount of
oxidant (Mo:substrate:H2O2 = 1:40:120) promoted the formation

T.R. Amarante et al. / Journal of Catalysis 340 (2016) 354–367
365
Table 4
Conversion of different substrates using 1/H2O2.a
Substrate
Conv. (%)b
98
Product selectivities (%)c
cis-cyclooctene
(100)
90
55
cyclohexene
(10%)
(24%)
(59%)
(4%)
(12%)
DL-Limonene
(8%)
(56%)
42
39
(9%)
trans-2-octene
(31%)
(50%)
methyl oleate
(10%)
(61%)
99d
methylphenylsulfide
(39%)
72
benzothiophene
(100%)
a
Initial Mo:substrate:H2O2 molar ratio = 1:100:152, initial concentration of substrate = 1.2 M, 1 mL CH3CN, 70 °C.
Substrate conversion at 24 h reaction (unless otherwise specified).
Product selectivity indicated in parentheses; for the substrates cyclohexene and limonene, epoxide was formed in less than 1% yield; other products were formed which
b
c
were not clearly identified.
d
Reaction time of 6 h.
of the sulfone (97% yield at 10 min) [64]. In a different approach,
and somewhat in parallel with that observed for 1/H2O2, peroxo-
molybdenum species may be formed in situ from pre-catalysts,
promoting sulfoxidation reactions with H2O2 [65,66]. For example,
cis-[Mo(CO)4(eppa)] {eppa = ethyl[3-(2-pyridyl)-1-pyrazolyl]acet
ate} was converted to [MoO(O2)2(eppa)] in situ during PhMeS reac-
tion at 35 °C, leading to 100% sulfoxide selectivity at 88% conver-
sion (Mo:substrate:H2O2 = 1:40:40; 6 h) [65]. For 1/H2O2, using a
Mo:substrate:H2O2 molar ratio of 1:100:152 and higher reaction
temperature of 70 °C shifted the sulfide-to-sulfoxide-to-sulfone
cascade reaction system toward the formation of the end product
(61% sulfone yield at 99% conversion, 6 h). When the reaction tem-
perature was decreased from 70 to 35 °C and the amount of oxi-
dant was decreased (Mo:substrate:H2O2 = 1:40:40), the catalytic
system 1/H2O2 led to 99% PhMeS conversion at 6 h, and the
corresponding sulfoxide and sulfone products were formed with
89% and 11% selectivity. Hence, under similar conditions to those
used for [MoO(O2)2(eppa)] [65], the PhMeS reaction was faster
for 1, promoting the formation of the sulfoxide.
3.3.4. Catalyst behavior and recovery with different substrates
With PhCH2OH as substrate and 1 or 2 as (pre)catalyst, the cat-
alyst behavior (solid–liquid–solid phase transfer vs. heteroge-
neous) paralleled that observed for 1 and 2 in the reaction of Cy.
Accordingly, the FT-IR spectra and PXRD patterns of the original
and recovered solids were similar (Figs. 8 and 9 for 1, and
Figs. S5 and S6 in the Supplementary Material for 2). The
reaction-induced self-separating behavior was also observed for
the 1/H2O2 system and other olefin/sulfide substrates; Figs. 8 and
9 show representative FT-IR and PXRD data for the solids recovered

366
T.R. Amarante et al. / Journal of Catalysis 340 (2016) 354–367
from the 1-(methyl oleate)/H2O2 and 1-PhMeS/H2O2 systems. For
the latter, the intermediate formation of metal species of the type
PS-(1-L-ct) (as isolated in the 1-Cy/H2O2 system) was confirmed by
precipitating a yellow solid from the liquid phase; the FT-IR spec-
trum of this solid was similar to that for PS-(1-L-ct).
References
[1] W.A. Herrmann, B. Cornils, Angew. Chem., Int. Ed. Engl. 36 (1997) 1048–1067.
[2] D.J. Cole-Hamilton, Science 299 (2003) 1702–1706.
[3] P. McMorn, G.J. Hutchings, Chem. Soc. Rev. 33 (2004) 108–122.
[4] A. Corma, H. Garcia, Adv. Synth. Catal. 348 (2006) 1391–1412.
[5] K.R. Jain, F.E. Kühn, Dalton Trans. (2008) 2221–2227.
[6] S. Shylesh, M. Jia, W.R. Thiel, Eur. J. Inorg. Chem. (2010) 4395–4410.
[7] Z. Xi, N. Zhou, Y. Sun, K. Li, Science 292 (2001) 1139–1141.
[8] V.K. Dioumaev, R.M. Bullock, Nature 424 (2003) 530–532.
[9] M. Süßner, H. Plenio, Angew. Chem., Int. Ed. 44 (2005) 6885–6888.
[10] X. Yang, S. Gao, Z. Xi, Org. Proc. Res. Dev. 9 (2005) 294–296.
[11] H. Chen, W.-L. Dai, X.-L. Yang, R. Gao, Y. Cao, H. Li, K. Fan, Appl. Catal. A: Gen.
309 (2006) 62–69.
[12] Z. Weng, G. Liao, J. Wang, X. Jian, Catal. Commun. 8 (2007) 1493–1496.
[13] Z. Weng, J. Wang, X. Jian, Catal. Commun. 9 (2008) 1688–1691.
[14] Y. Ding, W. Zhao, H. Hua, B. Ma, Green Chem. 10 (2008) 910–913.
[15] B. Ma, W. Zhao, F. Zhang, Y. Zhang, S. Wu, Y. Ding, RSC Adv. 4 (2014) 32054–
32062.
[16] Y. Qiao, Z. Hou, H. Li, Y. Hu, B. Feng, X. Wang, L. Hua, Q. Huang, Green Chem. 11
(2009) 1955–1960.
[17] Y. Leng, J. Wang, D. Zhu, X. Ren, H. Ge, L. Shen, Angew. Chem., Int. Ed. 48 (2009)
168–171.
[18] P. Zhao, J. Wang, G. Chen, Y. Zhou, J. Huang, Catal. Sci. Technol. 3 (2013) 1394–
1404.
[19] Y. Chen, R. Tan, W. Zheng, Y. Zhang, G. Zhao, D. Yin, Catal. Sci. Technol. 4 (2014)
4084–4092.
4. Conclusions
In this work the tungsten analogue of the known molybdenum
oxide-triazole hybrid material [MoO3(trz)0.5] has been successfully
prepared by hydrothermal synthesis. Under the reaction condi-
tions used, the W and Mo phases were obtained as microcrystalline
powders. Therefore, the solid-state structures were determined by
Rietveld analysis of high-resolution powder X-ray diffraction data,
supported by spectroscopic data (1H and 13C CP MAS NMR, and FT-
IR). The structure for [MoO3(trz)0.5] (1) is identical to that previ-
ously found by single-crystal X-ray diffraction [26]. The W phase
(2) has a different space group, but displays the same type of undu-
lating two-dimensional (2D) metal oxide network structure. Com-
pounds 1 and 2 have been shown to display distinct catalytic
behaviors with aqueous H2O2 as oxidant. Whereas the hybrid 2
behaves as a fairly stable heterogeneous catalyst, the hybrid 1 dis-
plays reaction-controlled phase-transfer (solid–liquid–solid)
behavior, and thus it can be used as a self-separating catalyst.
The self-separating behavior of 1/H2O2 was observed with different
substrates and reaction systems. To the best of our knowledge, the
hybrid 1 is the first example of a self-precipitating catalyst that has
an extended 1D, 2D or 3D crystalline structure. The reasons why 1
displays this unprecedented behavior are not yet fully understood,
but possible factors include the structure-directing influence of the
ligand trz, which leads to unique structures for 1 and (presumably)
the active species present in solution, and the dual epoxidation and
catalase-like activities. We plan to investigate other types of metal
oxide-triazole hybrid materials in anticipation of finding new
examples of phase transfer catalysts that combine the advantages
of homogeneous and heterogeneous catalysts.
[20] M.-D. Zhou, M.-J. Liu, L.-L. Huang, J. Zhang, J.-Y. Wang, X.-B. Li, F.E. Kühn, S.-L.
Zang, Green Chem. 17 (2015) 1186–1193.
[21] Y. Ding, W. Zhao, J. Mol. Catal. A: Chem. 337 (2011) 45–51.
[22] W. Zhao, C. Yang, Y. Ding, B. Ma, New J. Chem. 37 (2013) 2614–2618.
[23] K. Li, L. Chen, H. Wang, W. Lin, Z. Yan, Appl. Catal. A: Gen. 392 (2011) 233–237.
[24] D. Liu, J. Gui, F. Iu, Z. Sun, Y.-K. Park, Catal. Lett. 142 (2012) 1330–1335.
[25] Y. Zhou, Z. Guo, W. Hou, Q. Wang, J. Wang, Catal. Sci. Technol. 5 (2015) 4324–
4335.
[26] P.J. Hagrman, R.L. LaDuca Jr., H.-J. Koo, R. Rarig Jr., R.C. Haushalter, M.-H.
Whangbo, J. Zubieta, Inorg. Chem. 39 (2000) 4311–4317.
[27] M. Abrantes, T.R. Amarante, M.M. Antunes, S. Gago, F.A.A. Paz, I. Margiolaki, A.
E. Rodrigues, M. Pillinger, A.A. Valente, I.S. Gonçalves, Inorg. Chem. 49 (2010)
6865–6873.
[28] M. Abrantes, I.S. Gonçalves, M. Pillinger, C. Vurchio, F.M. Cordero, A. Brandi,
Tetrahedron Lett. 52 (2011) 7079–7082.
[29] T.R. Amarante, P. Neves, A.A. Valente, F.A.A. Paz, A.N. Fitch, M. Pillinger, I.S.
Gonçalves, Inorg. Chem. 52 (2013) 4618–4628.
[30] T.R. Amarante, P. Neves, F.A.A. Paz, A.A. Valente, M. Pillinger, I.S. Gonçalves,
Dalton Trans. 43 (2014) 6059–6069.
[31] T.R. Amarante, P. Neves, A.C. Gomes, M.M. Nolasco, P. Ribeiro-Claro, A.C.
Coelho, A.A. Valente, F.A.A. Paz, S. Smeets, L.B. McCusker, M. Pillinger, I.S.
Gonçalves, Inorg. Chem. 53 (2014) 2652–2665.
[32] T.R. Amarante, M.M. Antunes, A.A. Valente, F.A.A. Paz, M. Pillinger, I.S.
Gonçalves, Inorg. Chem. 54 (2015) 9690–9703.
Acknowledgments
[33] A.N. Fitch, J. Res. Natl. Inst. Stand. Technol. 109 (2004) 133–142.
[34] A. Coelho, Topas Academic, Version 5.0, Coelho Software, Brisbane, 2013.
[35] A. Coelho, J. Appl. Crystallogr. 36 (2003) 86–95.
[36] R.W. Cheary, A.A. Coelho, J. Appl. Crystallogr. 25 (1992) 109–121.
[37] F. Billes, H. Endrédi, G. Keresztury, J. Mol. Struct. (Theochem) 530 (2000) 183–
200.
We acknowledge funding by FEDER (Fundo Europeu de Desen-
volvimento Regional) through COMPETE (Programa Operacional
Factores de Competitividade). National funding through the FCT
(Fundação para a Ciência e a Tecnologia) within the projects
[38] D. Bougeard, N.L. Calvé, B.S. Roch, A. Novak, J. Chem. Phys. 64 (1976) 5152–
5164.
FCOMP-01-0124-FEDER-029779
(FCT
ref.
PTDC/QEQ-
[39] J.G. Haasnoot, G. Vos, W.L. Groeneveld, Z. Naturforsch. B 32 (1977) 1421–1430.
[40] W.H. Baur, Acta Crystallogr., Sect. B: Struct. Sci. 30 (1974) 1195–1215.
[41] As detailed in Sections 2.4 and 3.2, the high-resolution synchrotron PXRD
studies indicated the presence of trace amounts of unidentified crystalline
impurities in the as-synthesised samples for 1 and 2 (not detectable through
standard in-house PXRD measurements). Although we cannot exclude the
possibility that these impurities could be catalytically active in the reaction
systems studied, we consider it reasonable to discount any possible catalytic
contribution due to the very low amounts present (<1%).
[42] C.A. Gamelas, A.C. Gomes, S.M. Bruno, F.A.A. Paz, A.A. Valente, M. Pillinger, C.C.
Romão, I.S. Gonçalves, Dalton Trans. 41 (2012) 3474–3484.
[43] A.B. Lysenko, G.A. Senchyk, K.V. Domasevitch, J. Hauser, D. Fuhrmann, M.
Kobalz, H. Krautscheid, P. Neves, A.A. Valente, I.S. Gonçalves, Inorg. Chem. 54
(2015) 8327–8338.
SUP/1906/2012) and FCOMP-01-0124-FEDER-041282 (FCT ref.
EXPL/CTM-NAN/0013/2013) is thanked. This work was developed
in the scope of the project CICECO – Aveiro Institute of Materials,
POCI-01-0145-FEDER-007679 (FCT ref. UID/CTM/50011/2013),
financed by national funds through the FCT/MEC and co-financed
by FEDER under the PT2020 Partnership Agreement. The FCT and
the European Union are acknowledged for postdoctoral grants to
P.N. (SFRH/BPD/73540/2010) and T.R.A. (SFRH/BPD/97660/2013)
cofunded by MCTES and the European Social Fund through the pro-
gram POPH of QREN. The authors are grateful to the ESRF (Greno-
ble, France) for approving the experiment CH-4254 (ID22).
[44] A. Jimtaisong, R.L. Luck, Inorg. Chem. 45 (2006) 10391–10402.
[45] P. Sözen-Aktasß, E. Manoury, F. Demirhan, R. Poli, Eur. J. Inorg. Chem. (2013)
2728–2735.
[46] F. Madeira, S. Barroso, S. Namorado, P.M. Reis, B. Royo, A.M. Martins, Inorg.
Chim. Acta 383 (2012) 152–156.
Appendix A. Supplementary material
[47] J. Tang, L. Wang, G. Liu, Y. Liu, Y. Hou, W. Zhang, M. Jia, W.R. Thiel, J. Mol. Catal.
A: Chem. 313 (2009) 31–37.
[48] C. Dinoi, M. Ciclosi, E. Manoury, L. Maron, L. Perrin, R. Poli, Chem. Eur. J. 16
(2010) 9572–9584.
[49] C. Dinoi, R. Poli, L. Perrin, L. Maron, Dalton Trans. 41 (2012) 1131–1133.
[50] A. Comas-Vives, A. Lledós, R. Poli, Chem. Eur. J. 16 (2010) 2147–2158.
[51] M. Herbert, E. Álvarez, D.J. Cole-Hamilton, F. Montilla, A. Galindo, Chem.
Commun. 46 (2010) 5933–5935.
CIF files for 1 and 2, additional characterization data for 1 and 2
(TGA curves, FT-IR spectra, solid-state 13C and 1H MAS NMR spec-
tra), and for the solids recovered after the contact tests and cat-
alytic runs using 2 (ATR FT-IR spectra and PXRD patterns).
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.jcat.2016.06.005.

T.R. Amarante et al. / Journal of Catalysis 340 (2016) 354–367
367
[52] V. Yudanov, J. Struct. Chem. Suppl. 48 (2007) S111–S124.
[53] M. Drees, S.A. Hauser, M. Cokoja, F.E. Kühn, J. Organomet. Chem. 748 (2013)
36–45.
[54] M. Herbert, F. Montilla, E. Álvarez, A. Galindo, Dalton Trans. 41 (2012) 6942–
6956.
[55] M.J. Calhorda, P.J. Costa, Curr. Org. Chem. 16 (2012) 65–72.
[56] R.V. Sharma, K.K. Soni, A.K. Dalai, Catal. Commun. 29 (2012) 87–91.
[57] E. Arceo, J.A. Ellman, R.G.A. Bergman, ChemSusChem 3 (2010) 811–813.
[58] G. Ming-Lin, L. Hui-Zhen, Green Chem. 9 (2007) 421–423.
[59] S. Xiao, C. Zhang, R. Chen, F. Chen, New J. Chem. 39 (2015) 4924–4932.
[60] V.C. Srivastava, RSC Adv. 2 (2012) 759–783.
[62] M. Abrantes, A.A. Valente, M. Pillinger, I.S. Gonçalves, J. Rocha, C.C. Romão,
Chem. Eur. J. 9 (2003) 2685–2695.
[63] C.J. Carrasco, F. Montilla, E. Álvarez, C. Mealli, G. Manca, A. Galindo, Dalton
Trans. 43 (2014) 13711–13730.
[64] R.D. Chakravarthy, V. Ramkumar, D.K. Chand, Green Chem. 16 (2014) 2190–
2196.
[65] P. Neves, T.R. Amarante, A.C. Gomes, A.C. Coelho, S. Gago, M. Pillinger, I.S.
Gonçalves, C.M. Silva, A.A. Valente, Appl. Catal. A: Gen. 395 (2011)
71–77.
[66] L.F. Veiros, C.A. Gamelas, M.J. Calhorda, C.C. Romão, Organometallics 30 (2011)
1454–1465.
[61] S. Caron, R.W. Dugger, S.G. Ruggeri, J.A. Ragan, D.H.B. Ripin, Chem. Rev. 106
(2006) 2943–2989.