Controlling the hydrogenolysis of silica-supported tungsten pentamethyl leads to a class of highly electron deficient partially alkylated metal hydrides

Article abstract of DOI:10.1039/c5sc03490f

The well-defined single-site silica-supported tungsten complex [(Si-O-)W(Me)₅], 1, is an excellent precatalyst for alkane metathesis. The unique structure of 1 allows the synthesis of unprecedented tungsten hydrido methyl surface complexes via a controlled hydrogenolysis. Specifically, in the presence of molecular hydrogen, 1 is quickly transformed at -78 °C into a partially alkylated tungsten hydride, 4, as characterized by ¹H solid-state NMR and IR spectroscopies. Species 4, upon warming to 150 °C, displays the highest catalytic activity for propane metathesis yet reported. DFT calculations using model systems support the formation of [(Si-O-)WH₃(Me)₂], as the predominant species at -78 °C following several elementary steps of hydrogen addition (by σ-bond metathesis or α-hydrogen transfer). Rearrangement of 4 occuring between -78 °C and room temperature leads to the formation of an unique methylidene tungsten hydride [(Si-O-)WH₃(CH₂)], as determined by solid-state ¹H and ¹³C NMR spectroscopies and supported by DFT. Thus for the first time, a coordination sphere that incorporates both carbene and hydride functionalities has been observed.

Full text of DOI:10.1039/c5sc03490f

Chemical
Science
View Article Online
View Journal
EDGE ARTICLE
Controlling the hydrogenolysis of silica-supported
tungsten pentamethyl leads to a class of highly
electron deficient partially alkylated metal
hydrides†
Cite this: DOI: 10.1039/c5sc03490f
a
a
a
Niladri Maity,‡ Samir Barman,‡ Emmanuel Callens, Manoja K. Samantaray,a
*
Edy Abou-Hamad,a Yury Minenkov,a Valerio D'Elia,ab Adam S. Hoffman,c
Cory M. Widdifield, Luigi Cavallo, Bruce C. Gates and Jean-Marie Basset
d
a
c
a
*
*
The well-defined single-site silica-supported tungsten complex [(^Si–O–)W(Me)5], 1, is an excellent
precatalyst for alkane metathesis. The unique structure of 1 allows the synthesis of unprecedented
tungsten hydrido methyl surface complexes via a controlled hydrogenolysis. Specifically, in the presence
of molecular hydrogen, 1 is quickly transformed at ꢀ78 ꢁC into a partially alkylated tungsten hydride, 4,
as characterized by 1H solid-state NMR and IR spectroscopies. Species 4, upon warming to 150 ꢁC,
displays the highest catalytic activity for propane metathesis yet reported. DFT calculations using model
systems support the formation of [(^Si–O–)WH3(Me)2], as the predominant species at ꢀ78 ꢁC following
several elementary steps of hydrogen addition (by s-bond metathesis or a-hydrogen transfer).
Rearrangement of 4 occuring between ꢀ78 ꢁC and room temperature leads to the formation of an
unique methylidene tungsten hydride [(^Si–O–)WH3(]CH2)], as determined by solid-state 1H and 13C
NMR spectroscopies and supported by DFT. Thus for the first time, a coordination sphere that
incorporates both carbene and hydride functionalities has been observed.
Received 16th September 2015
Accepted 25th November 2015
DOI: 10.1039/c5sc03490f
www.rsc.org/chemicalscience
tungsten hydrides includes complexes graed to oxide supports
and synthesized by a surface organometallic chemistry (SOMC)
strategy.15,16
Introduction
Since the discovery of [Cp2ReH] by Green and Wilkinson,1
transition metal hydrides have played an important role in
organometallic chemistry. Their reactivity in many chemical
transformations has been elucidated, including their involve-
ment in the elementary steps of numerous industrial catalytic
processes.2 Notably, transition metal hydrides have been
extensively investigated for the activation of nearly inert C–H
and C–C bonds in alkanes.3–5 Highly coordinated tungsten
hydrides bearing phosphine ligands have been widely studied,
and a few tungsten hydrides have been found to be catalytically
active for the dehydrogenation of alkanes.6–14 The family of
This approach has led to the synthesis of numerous well-
dened supported species that have allowed the elucidation
of elementary reaction steps in heterogeneous catalytic
processes.17 Compared to the phosphine-chelated tungsten
hydrides used in solution, the surface organometallic species
are generally highly reactive and electron-decient. They were
typically prepared and stabilized on supports including SiO2,
Al2O3, and SiO2/Al2O3 by the hydrogenolysis of the graed
Schrock-type tungsten neopentyl/neopentylidyne complex
[W(^C–t-Bu)(CH2–t-Bu)3] (Bu ¼ butyl) at 150 ꢁC.18–20 These
supported metal hydrides catalyse challenging and potentially
valuable transformations including the metathesis of
alkanes,21,22 the hydrogenolysis of saturated hydrocarbons,23,24
and the non-oxidative coupling of methane.25,26
Herein, we address alkane metathesis, dened as a reaction
that transforms an alkane into its lower and higher homo-
logues. In this regard, the immobilized tungsten hydrides serve
as multifunctional precatalysts that engage in separate,
successive elementary steps. These involve propagative species
that are proposed to display both hydridic (for the C–H bond
activation and olen hydrogenation) and carbene functional-
ities (for the olen metathesis steps).27 The SiO2/Al2O3 or Al2O3
aKing Abdullah University of Science & Technology, KAUST Catalysis Center (KCC),
23955-6900 Thuwal, Saudi Arabia. E-mail: jeanmarie.basset@kaust.edu.sa;
emmanuel.callens@kaust.edu.sa
bDepartment of Materials Science and Engineering, Vidyasirimedhi Institute of Science
and Technology, 21210, Rayong, Thailand
cDepartment of Chemical Engineering and Materials Science, University of California
at Davis, Davis, California 95616, USA. E-mail: bcgates@ucdavis.edu
dDepartment of Chemistry, Durham University, Stockton Road, Durham DH1 3LE, UK
† Electronic supplementary information (ESI) available: Complete experimental
and quantum chemical procedures and characterization data for the prepared
compounds. Additional information pertaining the calculation of NMR
parameters. See DOI: 10.1039/c5sc03490f
‡ These authors contributed equally.
This journal is © The Royal Society of Chemistry 2015
Chem. Sci.

View Article Online
Chemical Science
Edge Article
supported tungsten hydrides were found to be more efficient change of the material from pale yellow to dark brown. Methane
precatalysts (with higher numbers of catalytic turnovers)19,20 was evolved (detected by gas chromatography (GC)), corre-
than their silica-supported analogues due to the higher stability sponding to 4.9 ꢂ 0.3 molecules per tungsten atom (matching
of the surface species on the former supports. The lower within the error of theoretically expected C/W ratio of 5). The IR
stability of the silica-supported tungsten hydrides was attrib- spectrum of 2 (Fig. 1(Ib)) clearly revealed the almost complete
uted either to the sintering of tungsten during hydrogenolysis of disappearance of the characteristic n(C–H) (3014–2878 cmꢀ1) and
the precursor complex (at 150 ꢁC) or to the transfer of hydrides d(C–H) (1410 cmꢀ1) vibrational modes assigned to the CH3
from tungsten to silicon atoms of the support. Both effects lead groups of 1 (Fig. 1(Ia)), as an effect of hydrogenolysis and the
to a decrease of the number of active sites available for catalysis. appearance of more than one (weak) new bands in the 1993–
We recently demonstrated that the silica-supported tungsten 1905 cmꢀ1 range. The multiple broad bands at 1993, 1978, 1953,
pentamethyl [(^Si–O–)W(Me)5] (1) (Me ¼ CH3) is a precatalyst and 1905 cmꢀ1 (Fig. 2(II(b-a))), fall in the frequency region ex-
for the metathesis of propane.28 As the initial C–H bond acti- pected for W(VI) hydride(s) (W–H) (i.e., with tungsten in its
vation occurs by sigma bond metathesis,3 we postulated that highest oxidation state).30,31 Moreover, these values are close to
this step would be more favourable by employing tungsten those determined experimentally for distorted trigonal pris-
hydride precatalysts rather than tungsten methyl species. We matic tungsten hexahydrides by Wang and Andrews.30,31 The
also anticipated that hydrogenolysis of the W–Me moieties in 1 stronger absorptions are assigned to the symmetric and anti-
should take place more readily than for the analogous Schrock symmetric (W–H) bond stretching vibrations. Contacting the
species,29 thus reducing the tendency of the W atoms towards sample of 2 with D2 at 80 ꢁC led to the disappearance of all W–H
sintering and H-transfer to the SiO2 surface.
vibrational absorptions and the emergence of a new band at
Herein, we demonstrate that the controlled hydrogenolysis 1403 cmꢀ1 (Fig. 1(II)). This band is attributed to the W–D bond
of 1 at various temperatures (Scheme 1) leads to a variety of stretching and conrms the presence of a tungsten hydride
different surface complexes in terms of structure (as supported species (on the basis of the harmonic approximation, the ex-
by DFT computations, infrared (IR) and solid-state NMR spec- pected value of nW–D would be 1402 cmꢀ1 for the nW–H band
troscopies, and elemental microanalysis) and reactivity. observed at 1978 cmꢀ1). This deuterium/hydrogen exchange
Amongst them, unprecedented partially alkylated tungsten was found to be reversible, further corroborating the identi-
hydride species 4, obtained from the hydrogenolysis of 1 at low cations of the silica-supported tungsten hydrides.
Moreover, the band at 2276 cmꢀ1 in the IR spectrum of 2
ꢁ
temperature (<ꢀ70 C), is the most active supported tungsten
hydride species yet reported for single metal propane metath- supports the formation of ^Si–H species possibly arising via
esis under batch conditions. The specic structural features of hydride transfer from tungsten to the silica surface. This is
surface complex 4 are investigated here with the aid of solid- a well-documented rearrangement whereby a hydride transfers
state NMR spectroscopy, DFT calculations and EXAFS spec- to a neighbouring siloxane bridge, leading to the formation of
troscopy (using 1 as a model substrate) leading to a deeper supported bipodal metal hydride species at high tempera-
insight into the catalytically relevant intermediates generated tures.32 Taken together, the data strongly support the formation
upon its thermal rearrangement.
ꢁ
of W–H and Si–H bonds upon hydrogenolysis of 1 at 150 C.
Hydrogenolysis at 25 ꢁC. In agreement with our initial
hypothesis, the hydrogenolysis of 1 could be carried out also at
lower temperature. When 1 is allowed to react with H2 at 25 ꢁC
for 4 min, the IR spectrum of the resulting material 3 (Fig. 2(Ic)
Results and discussion
IR spectroscopic characterization of the tungsten hydrides
prepared from [(^Si–O–)W(Me)5] (1) at three different
hydrogenolysis temperatures
and (II(c-a))) clearly shows n(W–H) bands at 1978 and 1959 cmꢀ1
.
Moreover, the absence of a band at 2276 cmꢀ1 for n(Si–H), which
is indirectly associated with a bipodal tungsten species,32 rules
out the transfer of hydrides to the silica surface at room
temperature and below.
Hydrogenolysis at 150 ꢁC. 1 was initially treated with
hydrogen gas (at 750 mbar and 150 C for 15 h) to produce 2.
ꢁ
Hydrogenolysis at ꢀ78 ꢁC. This facile hydrogenolysis at 25 ꢁC
prompted us to investigate the same process at even lower
temperatures (0, ꢀ20, and ꢀ78 ꢁC).29 The tungsten hydrides
obtained at 0 and at ꢀ20 ꢁC showed similar multiple IR bands
(with slight alterations in intensity when compared to 2 and 3)
corresponding to tungsten hydrides (Fig. S1†). In contrast, the
material obtained at ꢀ78 ꢁC (4), under otherwise identical
hydrogenolysis conditions, showed mainly one intense band, at
1959 cmꢀ1, which is assigned to a n(W–H) mode, along with
residual n(C–H) bands in the 3014–2853 cmꢀ1 region (Fig. 2(Id)
and (II(d-a))). Therefore, the IR spectrum of 4 shows the coex-
istence of hydride functionality and residual or unreacted alkyl
groups. 4 was observed to be stable at temperatures below ꢀ78
ꢁC for more than 2 h (Fig. S2†).
This hydrogenolysis treatment led to an immediate color
Scheme 1 Synthesis of silica-supported tungsten hydrides [WHx/SiO2]
(2–4) by treatment of 1 at various hydrogenolysis temperatures.
Chem. Sci.
This journal is © The Royal Society of Chemistry 2015

View Article Online
Edge Article
Chemical Science
Fig. 1 (I) IR spectra in the 3900–1400 cmꢀ1 region recorded under vacuum (pressure <10ꢀ5 mbar) at 25 ꢁC: (a) species 1 and (b) after hydro-
genolysis at 150 ꢁC (2). (II) IR spectra in the 3900–1400 cmꢀ1 region of species 2 before and after the addition of D2 at 80 ꢁC, followed by
evacuation and then further addition of H2. This cycle of operations was repeated twice.
Overall, the IR investigation shows that the controlled
The presence of residual alkyl groups in 4 hints at a partial
hydrogenolysis of 1 at different temperatures leads to unam- hydrogenolysis of 1 at low temperature resulting in the presence
biguously different surface species. Hydrogenolysis at 25 ꢁC of unreacted methyl groups at the tungsten centres along with
prevents the formation of the ^Si–H species observed at 150 ꢁC, the newly introduced hydride functionalities. Partial hydro-
which implies the lack of hydride transfer from tungsten to the genolysis was also conrmed by elemental microanalysis, which
support. On the other hand, residual alkyl species are still gave a C/W atomic ratio of 0.6 and 1.3 measured for 2 and 3,
ꢁ
present on the surface aer hydrogen treatment at ꢀ78 C (4) respectively (Table S1†). This ratio further increases to 2.1 for 4
along with a single strong n(W–H) band observed at 1959 cmꢀ1. In (measured at room temperature). Therefore, it is evident that
subsequent sections, we provide discussion and support for the decreasing the hydrogenolysis temperature leads to an increase
various possible surface species at each reaction temperature.
in the amount of residual carbon. The microanalysis data are
Fig. 2 (I) IR spectra in the 3900–1400 cmꢀ1 region recorded under vacuum (10ꢀ5 mbar) and 25 ꢁC: (a) species 1 and (b) after hydrogenolysis of 1
at 150 ꢁC (2); (c) after hydrogenolysis at 25 ꢁC (3) and (d) after hydrogenolysis at ꢀ78 ꢁC (4) (this spectrum was uniquely recorded at a temperature
below ꢀ78 ꢁC). (II) Difference IR spectra in the 2260–1850 cmꢀ1 region obtained by subtraction of the spectra of hydrogenolysis products (2–4)
and 1, namely, (b-a) (representing the difference between spectrum b and spectrum a), (c-a), and (d-a).
This journal is © The Royal Society of Chemistry 2015
Chem. Sci.

View Article Online
Chemical Science
Edge Article
complemented by the CH4 gas quantication analysis aer (vide infra). However, in this case, it was not possible to conrm
hydrogenolysis of 1. At 150 ꢁC, the loss of nearly 5 molecules of this attribution on the basis of the 13C spectrum of 2 (Fig. 4(b)).
CH4 per tungsten atom was measured for 2. This number
Because tungsten hydrides are expected to be highly ux-
decreased to 4.0 for 3 and to 1.7 for 4. In agreement with our ional on the NMR time scale, and are known to be found within
initial hypothesis, the silica-supported Schrock analogue18 of 1 a wide 1H chemical shi range (i.e., ranging from approximately
did not show any trace of hydrogenolysis when reacted with H2 +24 to ꢀ10 ppm),7,35,36 their detection and spectroscopic quan-
at ꢀ78 ꢁC under identical conditions as per 1. At 25 ꢁC, partial tication can oen be difficult.19,37 Indeed, an expanded view of
hydrogenolysis of the silica-supported Schrock analogue was the solid-state 1H NMR spectrum (inset of Fig. 3(b)) reveals
observed, but to a lesser extent than for 1 (see IR spectra in several low-intensity resonances with positive chemical shis.
Fig. S3 in the ESI†).
Consequently, we used density functional theory (DFT) calcu-
lations which included relativistic effects (both scalar and spin–
orbit) using the zeroth-order regular approximation (ZORA) to
predict the magnetic shielding and chemical shi values (vide
infra) of a variety of tungsten hydrides on silica surface models.
A complete specication of the protocol used here builds upon
that established in an earlier study of 1.28 In brief, we carefully
benchmarked the ZORA DFT method using several well-char-
acterized systems that can be found at the various points in the
W–H shi range denoted above (see ESI, additional experi-
mental discussion, Fig. S12–S15 and Tables S4 and S5†). The 1H
chemical shis for models of the W(VI) pentahydride species on
silica have been calculated to occur in the range of 12–16 ppm
(see the ESI†), depending upon the surface chosen to model the
silica and also by considering the slightly different shis for
each of the 5 1H sites in [(^Si–O–)WH5].
Solid-state NMR characterization of silica-supported tungsten
hydride species 2, 3 and 4
Further spectroscopic analyses were also conducted with solid-
state NMR. The 1H magic-angle spinning (MAS) solid-state NMR
spectrum of 1 (Fig. 3(a)) displays one signal at 2.0 ppm, attrib-
uted to the ve dynamically equivalent methyl groups,28 which
overlaps with the Si–OH signal.33 However, the 1H MAS NMR
spectrum of 2 (Fig. 3(b), scaled vertically by a factor of 8) indi-
cates a large reduction in the intensity at 2.0 ppm and shows
a very weak signal in the 1H–1H double-quantum/single-
quantum (DQ/SQ) spectrum (Fig. S5†). These results indicate
that most of the methyl groups were consumed during the
reaction with H2 and that the signal at 2.0 ppm is mostly
1
attributed to unreacted silanol (Si–OH).33 The H signal at 1.4
Intriguingly, the variation in these chemical shis appears to
be nearly fully attributable to spin–orbit (i.e., relativistic) effects,
with variation in the other magnetic shielding mechanisms (i.e.,
non-relativistic) being on the order of a few tenths of one ppm.
This aspect is discussed more fully in the ESI, Section 6.†
Hence, the weak high-frequency 1H signals at 11.6 and 18.8
ppm are assigned to surface WHx, consistent with the results of
the DFT calculations. This is also in agreement with the pres-
ence of multiple IR bands of 2 assigned for n(W–H) in the range
1993–1905 cmꢀ1 (Fig. 2(II(b-a))). The presence of multiple
hydride bands in the FT-IR spectrum of 2, and the likely
ppm, that shows a strong autocorrelation in the DQ/SQ NMR
spectrum, is attributed to the W–Me moieties present in the
newly formed surface species. The 1H resonance observed at 0.0
ppm is assigned to methyl groups attached to silicon atoms of
the support surface, these can originate from methyl transfer
from tungsten to the silica surface with a change of podality at
the W-center (vide infra).28,34 The resonance at 4.2 ppm is
assigned to the development of ^Si–H through a bipodal
structural transformation, which is consistent with the IR band
observed at 2276 cmꢀ1 (Fig. 1(Ib)).32 The signal at 7.2–7.7 ppm,
showing autocorrelation in the DQ/SQ NMR spectrum
(Fig. S5†), falls in the region of the W methylidene species
Fig. 3 One-dimensional (1D) 1H solid-state NMR spectra recorded at Fig. 4 13C CP/MAS NMR spectra of 13C-enriched 1 recorded at room
room temperature for species 1 (a) and 2 (b) acquired at 500 MHz (i.e., temperature (a) and its hydride analogue 2 (b) acquired at 125 MHz (B0
within a magnetic field, B0, of 11.7 T) with a 10 kHz MAS frequency for ¼ 11.7 T) with a 10 kHz MAS frequency, 4 s repetition delay, and 2 ms
(a) and 12.5 kHz MAS frequency for (b), a repetition delay of 5 s, 32 contact time. 40 000 scans were recorded to produce (a) and 26 500
scans for (a), and 128 scans for (b).
were recorded to yield (b).
Chem. Sci.
This journal is © The Royal Society of Chemistry 2015

View Article Online
Edge Article
Chemical Science
formation of ^Si–H and ^Si–CH3 moieties observed respec-
tively by FT-IR and 1H solid-state NMR demonstrate that
ꢁ
hydrogenolysis at 150 C leads to the formation of a variety of
species with little control on structure and identity of the
ensuing surface complexes.
The 13C cross-polarization (CP)/MAS NMR spectrum of 2
(Fig. 4(b)), prepared from 95% 13C enriched 1, displayed no
detectable signals in the +250 to ꢀ35 ppm range, even aer
>26 000 scans, consistent with the inference that most of the
tungsten methyl groups had been hydrogenolysed. Solid-state
1H NMR spectra of 3 and 4 were compared with the spectrum
obtained for 2 formed in the typical hydrogenolysis treatment at
high temperatures. A set of proton resonances similar to that of
2 was detected in the spectrum of 3, with the exception of the
signal at 4.2 ppm corresponding to the ^Si–H moieties of 2.
Furthermore, the solid-state 13C CP/MAS NMR spectrum of 3
shows signals at 42 and 47 ppm (Fig. S6†) attributable to
unreacted tungsten methyl groups (W–CH3) which are le as
a consequence of the partial hydrogenolysis of 1.
Because species 4 was prepared at low temperature, we
recorded the solid-state 1H NMR spectrum using a low-
temperature probe, at 100 K. The spectrum displays multiple
signals in the low-frequency range of 0–2 ppm, and at high
frequency, weak signals, at 8.7 and 16.3 ppm. Assignments of
these signals are presented aer discussion of the various
hydrogenolysis pathways starting with 1.
Evaluation of hydrogenolysis pathways by DFT calculations
To provide a better understanding on how the hydrogenolysis of
the silica-supported tungsten complex [(^Si–O–)W(Me)5], 1,
takes place, we proceeded with a DFT approach using a cluster
model of the silica support. To ensure that our relatively small
cluster model could be used to simulate the reactions on the
silica surface, we validated it against periodic DFT and using
slab silica models (see ESI† for details).38 For clarity in what
follows, we use Roman numerals to denote all the various
minimum-energy species. Transition states are described with
the notation exemplied by [I–II]‡, referring to the transition
state characterizing the transformation of I to II.
Starting from the silica-supported [(^Si–O–)W(Me)5]
precursor 1 depicted as I in Scheme 2, the calculations show that
the reaction with molecular hydrogen leads to the formation of
a
dihydrogen bonded (h2–H2) heptacoordinated tungsten
˚
species, with predicted W–H distances of 1.84 A, and the H–H
39
˚
˚
bond elongated to 0.86 A from 0.741 A in gaseous H2. The H2
coordination is endergonic by 14.8 kcal molꢀ1. The hydro-
genolysis of I occurs via transition state (TS) [I–IV]‡ and requires
an activation energy of 15.9 kcal molꢀ1, making this process
possible at room temperature. This step corresponds to a s-bond
metathesis event with the release of a CH4 molecule and forma-
tion of the tungsten monohydride [(^Si–O–)WH(Me)4], IV, which
is 18.2 kcal molꢀ1 lower in free energy than I.
Scheme 2 Proposed mechanism for hydrogenolysis of 1 (i.e., I) into
various tungsten hydrides; the values shown are free energies, in kcal
molꢀ1. The energetically most favourable hydrogenolysis pathway is
highlighted by the structures inside the boxes.
process is also exergonic, with a Gibbs free energy release of
In addition to this step, two other intramolecular trans- 18.1 kcal molꢀ1, but the associated transition state [I–II]‡ is 30.7
formations of I were explored. First, I can form tungsten kcal molꢀ1 higher in energy than I, and it is also 14.8 kcal molꢀ1
methylidene II via a-H elimination from a methyl group with higher in energy than transition state [I–IV]‡, making a-H
the simultaneous release of one methane molecule. This elimination highly disfavoured. Second, I can also undergo an
This journal is © The Royal Society of Chemistry 2015
Chem. Sci.

View Article Online
Chemical Science
Edge Article
intramolecular a-H transfer to the metal, with migration of
The above characterization of transformations from I to
another methyl group to form a tungsten ethyl hydride, III, via XVIII suggests that hydrogenolysis of I is an exergonic process
transition state [I–III]‡. This process is endergonic by 5.7 kcal (DG z ꢀ80 kcal molꢀ1), leading to the formation of silica-
molꢀ1, with an associated free energy barrier of 42.7 kcal molꢀ1
which excludes it from the reactivity scenario.
,
supported tungsten pentahydride as the thermodynamic
product. The proposed mechanism is a cascade or sequence of
The subsequent transformation of IV to VIII was found to be hydrogen addition reactions forming the series of supported
even easier than the addition of the rst hydrogen molecule to I, metallo hydrides ([(^Si–O–)WHx(Me)y], x ¼ 1–5, y ¼ 4–0) along
with a free energy barrier of only 12.0 kcal molꢀ1. This step is with the elimination of free methane molecules. The lowest
again exergonic, and the product [(^Si–O–)WH2(Me)3], VIII, is barrier (+12 kcal molꢀ1) corresponds to the reaction with the
more stable than IV by 18.0 kcal molꢀ1. Similar to the previous second H2 molecule. The highest barrier was calculated for the
ndings for I, the formation of a methylidene species (VI) or an addition of the fourth hydrogen molecule (during trans-
ethyl hydride (VII) requires too high of an activation energy and formation of XII to XVI) and amounts to more than 20 kcal
these processes can thus be excluded from further consider- molꢀ1, suggesting that this step might be rate determining. On
ation. However, in contrast to I, IV can undergo reductive the basis of these calculations it is possible to suggest that the
elimination to form a triplet W(IV) species [(^Si–O–)W(Me)3], V, silica-supported [(^Si–O–)WH3(Me)2] (XII) might represent the
with the release of one methane molecule. This process is fav- predominant surface species if 1 is hydrogenolysed to afford 4
oured thermodynamically, with a free energy change of at ꢀ78 ꢁC. Other mechanistic pathways involving the reduction
ꢀ23.1 kcal molꢀ1. The associated energy barrier estimated at of W(VI) species have been found to be characterized by higher
the minimum energy crossing point (MECP) along the CH3– activation barriers.
H(W) bond stretching mode turned out to be only 14.8 kcal
molꢀ1. However, because direct hydrogen addition occurs via
a free energy barrier lower by ꢃ3.0 kcal molꢀ1, the conversion of
Evaluation of propane metathesis activities of catalysts
formed from precatalysts 2, 3, and 4
IV to VIII is the most likely route for further hydrogenolysis. The
addition of the third, fourth, and h hydrogen atoms occurs
via direct successive hydrogen additions. All other competing
reactions again were found to require higher activation energies
than the hydrogenation steps. Even reductive elimination, with
formation of a W(IV) complex (which was found to have an
activation energy of 14.8 kcal for IV), requires 33.3, 38.4, and
26.3 kcal molꢀ1 for complexes VIII, XII, and XVI, respectively.
Eventually, the nal step is the formation of a silica-supported
tungsten pentahydride [(^Si–O–)WH5], XVIII, which was found
to be the most stable species of all those considered, with
a Gibbs free energy of ꢀ77.9 kcal molꢀ1 relative to I.
The catalytic performances of the various tungsten hydride
species (2–4, Table 1) were examined for propane metathesis in
a batch reactor. The results were compared with previously re-
ported data for tungsten hydrides (entry 5). A relatively high
turnover number (TON) (i.e., 104) was obtained with 2 (prepared
at 150 ꢁC) (entry 1) compared with the tungsten hydrides
prepared from a graed neopentyl organometallic precursor.19
ꢁ
For species 3 (prepared at 25 C), a further increase in activity
for propane metathesis was observed (154, entry 2). This result
could be attributed to a higher number of active tungsten
hydride sites at this treatment temperature (25 ꢁC), inasmuch as
Table 1 Comparative propane metathesis activities expressed as TON and product selectivities of tungsten species on silica and their corre-
sponding hydrides in a batch reactor at 150 ꢁC for 5 daysa
Product selectivityc (%)
Entry
Precatalyst
TONb (% conversion)
Methane
Ethane
Butanesd
Pentanesd
Ref.
1
2
3
4
5
6
[WH@150 ꢁC] (2)
[WH@25 ꢁC] (3)
[WH@ꢀ78 ꢁC] (4)e
[WMe5/SiO2] (1)
[WH@150 ꢁC]f
104 (11.5)
154 (17)
261 (29)
127 (12)
8 (1.2)
4
3
3
2
5.7
3
50
56
58
54
56
55
28/15
30/5
27/5
33/4
29/2.8
29/5.4
2/1
4/2
4.5/2
6/1
5.1/1.4
5/2.4
This work
This work
This work
28
19
[WH@ꢀ78 ꢁC] (4)g
166 (18.4)
This work
a
In a typical procedure, propane gas (20.4 mmol) was introduced into the reactor at 25 ꢁC. b TON is expressed in (mol of propane transformed)/(mol
c
of W) assuming that all the W atoms on the support are active. The selectivities are dened as the number of mols of product per mol of total
d
e
f
product. Ratio of linear to branched alkanes. Propane (20.4 mmol) was introduced into the reactor at 77 K. Data taken from the following
communication (Angew. Chem., Int. Ed., 2005, 44, 6755) whereby the tungsten hydride species were synthesized by hydrogenolysis of [(^Si–O–)
W(^C–t-Bu)(CH2–t-Bu)2]. g Propane was introduced aer allowing 4 to warm to room temperature.
Chem. Sci.
This journal is © The Royal Society of Chemistry 2015

View Article Online
Edge Article
Chemical Science
no apparent hydride transfer to the silica surface was observed
by solid-state 1H NMR and IR spectroscopy (Fig. 2(Ic)). Notably,
ꢁ
precatalyst 4, prepared at an even lower temperature (ꢀ78 C),
gave the best activity for this transformation (entry 3, TON: 261).
Signicantly, the catalytic activity of species 4, when it was
allowed to warm to room temperature (Table 1, entry 6) before
addition of propane, is comparable to that of 3. This observa-
tion is in agreement with the structural rearrangement of 4 as
discussed in the following paragraphs and demonstrates that
the structure obtained at low temperature plays a key role in the
catalytic activity. In general, the catalytic activity increases from
2 to 4, and, therefore, with the number of unreacted methyl
functionalities on the W-centre. Therefore, it appears that
tuning the degree of hydrogenolysis by reducing the hydro-
genolysis temperature is a crucial point in allowing the forma-
tion of highly active, partially alkylated hydrides. As shown
below with the aid of DFT calculations, partially alkylated
hydrides are required to generate key methylidene hydride
intermediates. Moreover, it has to be taken into account that the
amount of H2 generated in the reactor upon contact of the
sample with propane is different for completely and partially
alkylated species. The presence of H2 is related to the amount of
Fig. 5 1H MAS solid-state NMR spectra of fresh 4 (a) and after warming
to room temperature and holding for ca. 10 minutes (b). The spectra
were acquired at 400 MHz (B0 ¼ 9.4 T) with a 10 kHz MAS frequency,
a repetition delay of 5 s, 8 scans, and at ꢀ173 ꢁC.
alkenes, key intermediates in this alkane transformation.40 It a bipodal species under the experimental conditions by the
has also been demonstrated that, under dynamic conditions, transfer of a methyl group to the silica support.28,43 This trans-
a competitive reaction, the hydrogenolysis of alkanes, occurs.41 formation is not surprising as this behaviour has also been
Thus, the presence of partially hydrogenolysed species could observed by EXAFS spectroscopy for the parent material [(^Si–
have also an effect on the amount of H2 released in the reactor O–)W(Me)5] (vide infra). Allowing the rotor sample to warm to
and favor alkane metathesis.
room temperature for 10 min (outside the NMR probe) and then
reintroducing it into the cold NMR probe led to the complete
disappearance of the H chemical shi at 0.8 ppm and to the
1
Structural investigation on species 4
increase of the signal at 0.4 ppm (Fig. 5b). Additionally, we
To have a better understanding of the activity enhancement in observed the appearance of a signal at 8.7 ppm and a shi in the
propane metathesis for pre-catalyst 4 and on how structural W hydride signal (from 16.3 ppm to 14.7 ppm, Fig. 5b). All the
features are intertwined with catalytic activity we examined its observations support the progressive rearrangement of 4 at
thermal stability without substrate. First, an in situ IR variable temperatures higher than ꢀ78 ꢁC. In a different experiment, the
1
temperature investigation of 4 was carried out. When 4, solid-state H and 13C NMR spectra of 4 were measured aer
prepared at ꢀ78 ꢁC, was allowed to warm to room temperature, allowing the sample to warm to room temperature (Fig. 6a) and
the initial W–H stretching band at 1959 cmꢀ1 decreased with with a longer acquisition time. We clearly observed an intense
time, along with a decrease in intensity of the nC–H bands at signal in the W hydride region at 15.2 ppm that allowed us to
3014–2853 cmꢀ1. As a result, a new, weaker band centred at measure the average of spin–lattice relaxation T1 value for these
1978 cmꢀ1 was observed (Fig. S4†). These data strongly support protons of greater than 1500 ms (Fig. S7†) as we tted the data
the evolution of a new tungsten hydride species at temperatures with a stretched exponential decay.44 This value is characteristic
ꢁ
ranging from ꢀ78 C to room temperature.
of tungsten hydride species without the formation of tungsten
dihydrogen complexes (i.e., without W(h2–H2)).13
Second, this process was monitored by solid-state 1H and 13
CP/MAS NMR spectroscopies. Knowing the relatively poor
C
The 1H NMR spectrum in Fig. 6a along with the corre-
thermal stability of 4, we initially recorded the 1H NMR spectra sponding 1H–13C HETCOR (Fig. S8†) spectrum shows a clear
using a pre-cooled system with the probe temperature set at 100 indication of static disorder that we tentatively attribute to the
K (Fig. 5). In the solid-state proton NMR spectrum, 1H reso- formation of bipodal species upon warming.46 Under these
nances in the 0.4 to 16.3 ppm region are evident (Fig. 5a). The conditions we observed as well a signal at 7.3 ppm. The 13C CP/
1H NMR signals at 0.4 and 0.8 ppm should correspond MAS NMR spectrum (Fig. 6b) displayed signals at 40–48 ppm,
respectively to ^Si–CH3 (ref. 42) and to the unreacted W–CH3 attributed, as before, to tungsten methyl (W–CH3) moieties, and
moieties in 4. The signal at 1.9 ppm probably corresponds to a distinct signal at 231 ppm. The 1H peak at 7.3 ppm correlates
1
unreacted silanols,42 whereas the weak signal at 16.3 ppm can only with the carbon resonance at 231 ppm in the 2D H–13C
be attributed to tungsten hydrides according to the calculations HETCOR NMR spectrum (Fig. S8†). These correlated signals are
discussed above. The presence of ^Si–CH3 or of a different typical of a tungsten methylidene (W]CH2) moiety.26,47 This
kind of methyl group functionalities implies that the initial signal was not observed in the 13C solid-state NMR spectrum of
monopodal species 4 might have partially rearranged into species 3 generated directly at room temperature (Fig. S6†). No
This journal is © The Royal Society of Chemistry 2015
Chem. Sci.

View Article Online
Chemical Science
Edge Article
Fig. 7 Plausible surface structures of tungsten hydride species in 4 (a)
before and (b) after warming from ꢀ78 ꢁC to RT.
proposal is also in agreement with the DFT calculations that
show a relatively high energy barrier of 20.3 kcal molꢀ1 for the
reaction of [(^Si–O–)WH3(Me)2] with the fourth molecule of
hydrogen (transition state XII–XVI of Scheme 2). Other surface
species, that are consistent with the solid state NMR investi-
gation reported above (and with the X-ray absorption spectros-
copy results reported in the next paragraph), and are proposed
to form on the surface from the evolution of 4 (when allowed to
warm to room temperature), are depicted with the selected
characteristic NMR signals assigned (Fig. 7b).
Fig. 6 (a) One-dimensional (1D) 1H solid-state NMR spectrum of 4
recorded at room temperature; data acquired at 400 MHz with a 10
kHz MAS frequency, a repetition delay of 5 s, and 8 scans. (b) 1D 13C
CP/MAS NMR spectrum of 4 recorded at room temperature; data
acquired at 100 MHz with a 10 kHz MAS frequency, 4 s repetition delay,
2 ms contact time, and 20 000 scans. *The peak at ꢀ24 ppm is
attributed to adsorbed ZnMe2, which was used in the preparation of
the precursor complex WMe6.45
X-ray absorption spectroscopy
The transformation of the well-dened monopodal species into
a bipodal surface complex, accompanied by the transfer of
methyl groups to the silicon atoms of the support, is invoked in
the text as a major decomposition pathway for 4 when it is
allowed to warm to room temperature. To provide further
other correlation was observed between the 13C resonance evidence for this chemical transformation, the evolution of 1 as
signal at 231 ppm and the signals in the 30–50 ppm region, a model compound was monitored by X-ray absorption spec-
excluding the co-existence of a W-methyl within this W-meth- troscopy (XAS) at room temperature with the sample in a stream
ylidene species. Therefore, the structure proposed for this of dry helium. As reported previously, microelemental analyses
species is that of a methylidene hydride. Notwithstanding the of 1 indicate that the tungsten pentamethyl moiety is anchored
apparent absence of spectroscopic correlation also between the via only one isolated silanol group of the surface, maintaining
W]CH2 and W–H signals, the formation of a methylidene its hexacoordination around the W center.
hydride from partially alkylated W-hydrides is supported by
The structural parameters determined by the best t of the
DFT calculations (vide infra). The W-alkylidene species formed data of the rst W LIII-edge EXAFS scan of 1 in owing helium
was found by 1H NMR spectroscopy to be stable at temperatures strongly support the inference of two absorber–backscatterer
up to 150 ꢁC, but it decomposed readily upon introduction of D2 pairs, W–C and W–O, with internuclear distances consistent
gas at 80 ꢁC. Moreover, at room temperature we observed with typical sigma bonds. The W–C and W–O coordination
a proton chemical shi at 0.0 ppm corresponding to ^Si–CH3 numbers were found to be 4.9 and 1.0, respectively. Within the
moieties. This result is supported by the appearance of a peak at expected error (ꢂ20%), these values are consistent with the
ꢀ12 ppm (attributed to surface ^Si–CH3) in the 29Si CP/MAS initial supported species 1 being monopodal on the surface.
NMR spectrum of 4 recorded at room temperature (Fig. S9†).28 Furthermore, there was no detectable W–W contribution in the
Thus, on the basis of DFT calculations and a systematic solid- t, indicating that, within error, the tungsten species remained
state NMR investigation we infer that warming of 4 to room mononuclear.
temperature would lead to various monopodal and bipodal
structures on the amorphous silica surface (Fig. 7).
When 1 was present in the ow-through cell in the presence
of helium (ow rate: 1.0 mL minꢀ1), with continuous exposure
To summarize our spectroscopic and DFT investigation, we to the X-ray beam, changes in the X-ray absorption near edge
represent species 4 as [(^Si–O–)WH3(Me)2] (Fig. 7a); whereas structure (XANES) region of the X-ray absorption spectrum show
the coexistence of alkyl and hydride moieties has been observed that the supported tungsten species was transformed.
by IR (Fig. 2(Id)) and by solid state NMR spectroscopies (Fig. 5
The XANES region (Fig. 8) shows the edge energy increasing
and 6a), the average C/W ratio obtained by microanalysis (ca. slightly, from 10 208.9 to 10 209.9 eV. This change suggests
2 : 1, measured at room temperature) is in agreement with the a structural rearrangement and/or an increase in the tungsten
inferred structure as a main surface component. Considering oxidation state during the treatment. Isosbestic points in the
the low hydrogenolysis temperature of ꢀ78 ꢁC, the structural XANES spectra (Fig. 8) indicate a stoichiometric change from
Chem. Sci.
This journal is © The Royal Society of Chemistry 2015

View Article Online
Edge Article
Chemical Science
Table 2 EXAFS fit parametersa characterizing the tungsten species
supported on SiO2 at the W LIII edgeb
103 ꢄ Ds2
2
˚
˚
Species
Shell
N
R [A] [A ]
DE0 [eV]
1
W–Osupport 1.03 1.81
W–C 4.91 2.02
1 aer 45 min in W–Osupport 1.56 1.82
3.70
8.56
9.78
ꢀ3.73
4.68
ꢀ4.23
owing helium
W–C
3.77 2.00
12.0
14.6
(1 bar) at 298 K
a
Notation: N, coordination number; R, distance between absorber and
backscatterer atoms; Ds2, disorder term; DE0, inner potential
correction. Error bounds characterizing the structural parameters are
2
˚
estimated to be as follows: N, ꢂ20%; R, ꢂ0.02 A; Ds , ꢂ20%; and DE0,
b
ꢂ20%. Details pertaining to the best ts can be found in the ESI
(Tables S2, S3 and Fig. S10 and S11).
Fig. 8 XANES region of the X-ray absorption spectra at the W LIII edge
characterizing the sample initially present as the silica-supported
tungsten complex [(^Si–O–)W(Me)5] as it underwent changes in the
presence of flowing helium at 298 K and 1 bar. Isosbestic points
(circled) indicate a nearly stoichiometric conversion of the initial
species. The blue, red, green and purple spectra represent the first to
fourth scans, respectively.
DFT calculation to support the formation of tungsten
methylidene (W]CH2) moiety by warming 4 from ꢀ78 C to
ꢁ
room temperature
Motivated by the structural rearrangements in the structure of
4, inferred from the solid-state 1H and 13C NMR data, we
investigated the formation of the supported tungsten methyl-
idene species (i.e., XIV, Scheme 3) by DFT, envisaging two
possible routes for its formation. The rst corresponds to a-H
elimination from XII with release of a CH4 molecule to produce
a carbene, XIV, whereas the second corresponds to an a-H
transfer from XVI. Focusing on the former, we found that the
conversion of XII to XIV is an exergonic process (ꢀ10.1 kcal
molꢀ1), but the associated activation barrier for the direct
conversion via transition state [XII–XIV] is quite high (43.1 kcal
molꢀ1) to be accessible under the reaction conditions used. In
contrast, the conversion of XVI to XIV is exergonic by only
the initial species. Because the changes occurred slowly, the
rst EXAFS spectrum of the sample in owing helium (which
was recorded during the rst 15 min of exposure to the X-ray
beam) is considered to be essentially indicative of the initial
supported tungsten complex. The two EXAFS spectra recorded
aer the rst one show that the tungsten coordination sphere
was changing during the experiment. The second and third
scans were only slightly different from the rst, but the fourth
was sufficiently different to justify a detailed analysis.
This fourth EXAFS spectrum was recorded starting 45 min
aer the initial exposure of the sample to the X-ray beam. The
data were of lower quality than those of the rst scan, so that
the t parameters were determined with less condence.
Several plausible models were tested in the tting of the data
characterizing the rst and fourth spectra. The ts according
to the recommended models (see Tables 2, S2, S3 and Fig. S10,
S11 in ESI†) are characterized by physically realistic values of
the tting parameters. Attempts were made to obtain satis-
factory ts with alternative structural models. However, all the
alternative models converged to the best t models previously
stated. Details of the data and analyses, including ts
according to structural models that were rejected, are shown
in the ESI.† The relatively large DE0 value determined in the
best t of the W–C shell of species 1 aer 45 min in contact
with owing helium (Table 2) is attributed to the limited
number of shells permitted in a statistically justied t
(indicated by the overall data quality and the Nyquist
theorem;48 details are given below).
A comparison of the data from the rst and fourth scans
shows that, within error, the supported tungsten complexes lost
methyl groups during the treatment in helium but remained
six-coordinate, as a new W–support-oxygen bond formed. Thus,
the EXAFS data are in agreement with the rearrangement of 1
with a proximate Si–O–Si bridge from monopodal to bipodal at
ambient temperature (Scheme S1†).
Scheme
3 Proposed mechanism for the formation of tungsten
methylidene hydrides from XII or XVI based on DFT calculations.
This journal is © The Royal Society of Chemistry 2015
Chem. Sci.

View Article Online
Chemical Science
Edge Article
2.3 kcal molꢀ1, and the associated free energy barrier for the Computational resources for selected DFT calculations were
ˆ
´
´
direct conversion via transition state [XVI–XIV] and the release provided by the Pole Scientique de Modelisation Numerique
of a H2 molecule amounts to just 22.3 kcal molꢀ1. This barrier (PSMN).
can be overcome at room temperature, suggesting that the
dominant route for the formation of XIV starts from XVI.
Notwithstanding the result that XII is the predominant
species at low temperatures according to our spectroscopic
Notes and references
investigation and DFT calculations, species XVI can be formed
by a less than stoichiometric conversion in the hydrogenolysis
of XII to XVI, which we calculated (Scheme 2) to be a facile
process.
1 M. L. H. Green, L. Pratt and G. Wilkinson, J. Chem. Soc., 1958,
3916.
2 G. S. McGrady and G. Guilera, Chem. Soc. Rev., 2003, 32, 383.
3 J. A. Labinger and J. E. Bercaw, Nature, 2002, 417, 507.
4 P. L. Watson and G. W. Parshall, Acc. Chem. Res., 1985, 18, 51.
5 G. W. Parshall, Acc. Chem. Res., 1975, 8, 113.
6 J. R. Moss and B. L. Shaw, Chem. Commun., 1968, 632.
7 D. Gregson, J. A. K. Howard, J. N. Nicholls, J. L. Spencer and
D. G. Turner, J. Chem. Soc., Chem. Commun., 1980, 572.
8 M. B. Hursthouse, D. Lyons, M. Thorntonpett and
G. Wilkinson, J. Chem. Soc., Chem. Commun., 1983, 476.
9 R. H. Crabtree and G. G. Hlatky, J. Organomet. Chem., 1982,
238, C21.
Thus, the DFT calculations are consistent with the formation
of an unprecedented silica-supported tungsten methylidene
hydride species (Scheme 3) following hydrogenolysis of 1 at
ꢁ
ꢀ78 C and warming to room temperature.
Conclusion
The controlled hydrogenolysis of silica-supported tungsten
pentamethyl [(^Si–O–)W(Me)5] (1) has been investigated at
various temperatures. The hydride products have been charac- 10 G. G. Hlatky and R. H. Crabtree, Coord. Chem. Rev., 1985, 65,
terized by means of elemental microanalysis, IR and solid-state 1.
NMR spectroscopies revealing that hydrogenolysis is tempera- 11 M. L. H. Green and G. Parkin, J. Chem. Soc., Chem. Commun.,
ture dependent and leads to different structural features. The 1985, 1383.
results demonstrate that the hydrogenolysis of [(^Si–O–) 12 M. L. H. Green and G. Parkin, J. Chem. Soc., Dalton Trans.,
ꢁ
W(Me)5] at ꢀ78 C leads to an unprecedented silica-supported
1987, 1611.
partially methylated tungsten hydride species. On the basis of 13 A. Sattler, A. A. Zuzek and G. Parkin, Inorg. Chim. Acta, 2014,
DFT calculations, we propose that [(^Si–O–)WH3(Me)2] is the
422, 102.
predominant surface species at this temperature. The thus- 14 R. H. Crabtree, H. Felkin, G. E. Morris, T. J. King and
synthesized tungsten hydride represents the precursor of the J. A. Richards, J. Organomet. Chem., 1976, 113, C7.
most active catalyst for propane metathesis yet reported. The 15 C. Coperet, M. Chabanas, R. P. Saint-Arroman and
thermal rearrangement of this partially methylated species, J. M. Basset, Angew. Chem., Int. Ed., 2003, 42, 156.
when allowed to warm to room temperature, has been observed 16 Modern Surface Organometallic Chemistry, ed. J. M. Basset, R.
´
to proceed via divergent pathways. This change involves methyl
transfer to the silica surface, evidenced by the presence of ^Si–
Psaro, D. Roberto and R. Ugo, Wiley-VCH Verlag GmbH & Co.
KGaA, Weinheim, 2009.
CH3 species indicated by solid-state NMR spectroscopy, trans- 17 P. Serna and B. C. Gates, Acc. Chem. Res., 2014, 47, 2612.
forming the initial monopodal surface complex 4 into a bipodal 18 E. Le Roux, M. Taouk, M. Chabanas, D. Alcor, A. Baudouin,
´
species. Such a change in podality was indicated by EXAFS
C. Coperet, J. Thivolle-Cazat, J. M. Basset, A. Lesage,
spectra determined with 1 as a model substrate. Furthermore,
S. Hediger and L. Emsley, Organometallics, 2005, 24, 4274.
´
the formation of a tungsten alkylidene (methylidene) hydride 19 E. Le Roux, M. Taouk, C. Coperet, A. de Mallmann,
intermediate, which has always been considered in alkane
metathesis but never been observed, has been proposed on the
basis of solid-state NMR spectroscopy and DFT calculations.
J. Thivolle-Cazat, J. M. Basset, B. M. Maunders and
G. J. Sunley, Angew. Chem., Int. Ed., 2005, 44, 6755.
´
20 E. Le Roux, M. Taouk, A. Baudouin, C. Coperet, J. Thivolle-
Cazat, J. M. Basset, B. M. Maunders and G. J. Sunley, Adv.
Synth. Catal., 2007, 349, 231.
21 V. Vidal, A. Theolier, J. Thivolle-Cazat and J. M. Basset,
Science, 1997, 276, 99.
Acknowledgements
The research was supported by the King Abdullah University of
´
Science and Technology. The authors acknowledge Prof. Lyn- 22 J. M. Basset, C. Coperet, D. Soulivong, M. Taouk and
don Emsley and Dr Aaron J. Rossini (EPFL) for their assistance. J. Thivolle-Cazat, Acc. Chem. Res., 2010, 43, 323.
The work at the University of California was supported by the 23 S. Norsic, C. Larabi, M. Delgado, A. Garron, A. de Mallmann,
U.S. Department of Energy (DOE), Office of Science, Basic
Energy Sciences, Grant DE-FG02-04ER15513. We acknowledge
C. Santini, K. C. Szeto, J. M. Basset and M. Taouk, Catal. Sci.
Technol., 2012, 2, 215.
beam time at beamline 4-1 at the Stanford Synchrotron Radia- 24 V. R. Dufaud and J. M. Basset, Angew. Chem., Int. Ed., 1998,
tion Lightsource supported by the DOE Division of Materials 37, 806.
Sciences. We thank the beam line staff for valuable support. C. 25 D. Soulivong, S. Norsic, M. Taouk, C. Coperet, J. Thivolle-
´
M. W. acknowledges the Natural Sciences and Engineering
Council of Canada (NSERC) for a postdoctoral fellowship.
Cazat, S. Chakka and J. M. Basset, J. Am. Chem. Soc., 2008,
130, 5044.
Chem. Sci.
This journal is © The Royal Society of Chemistry 2015

View Article Online
Edge Article
Chemical Science
26 K. C. Szeto, S. Norsic, L. Hardou, E. Le Roux, S. Chakka,
J. Thivolle-Cazat, A. Baudouin, C. Papaioannou,
J. M. Basset and M. Taouk, Chem. Commun., 2010, 46, 3985.
also: V. D'Elia, H. Dong, A. J. Rossini, C. M. Widdield,
S. V. Vummaleti, Y. Minenkov, A. Poater, E. Abou-Hamad,
J. D. Pelletier, L. Cavallo, L. Emsley and J. M. Basset, J. Am.
Chem. Soc., 2015, 137, 7728.
´
27 J. M. Basset, C. Coperet, L. Lefort, B. M. Maunders, O. Maury,
E. Le Roux, G. Saggio, S. Soignier, D. Soulivong, G. J. Sunley, 39 J. S. Craw, G. B. Bacskay and N. S. Hush, Inorg. Chem., 1993,
M. Taouk and J. Thivolle-Cazat, J. Am. Chem. Soc., 2005,
127, 8604.
28 M. K. Samantaray, E. Callens, E. Abou-Hamad, A. J. Rossini,
32, 2230.
40 N. Riache, E. Callens, M. K. Samantaray, N. M. Kharbatia,
M. Atiqullah and J. M. Basset, Chem.–Eur. J., 2014, 20, 15089.
C. M. Widdield, R. Dey, L. Emsley and J. M. Basset, J. Am. 41 S. Soignier, G. Saggio, M. Taouk, J. M. Basset and
Chem. Soc., 2014, 136, 1054. J. Thivolle-Cazat, Catal. Sci. Technol., 2014, 4, 233.
29 A. J. Shortland and G. Wilkinson, J. Chem. Soc., Dalton Trans., 42 Y. Chen, B. Zheng, E. Abou-Hamad, A. Hamieh,
1973, 872.
B. Hamzaoui, K. W. Huang and J. M. Basset, Chem.
Commun., 2014, 50, 11721.
30 X. F. Wang and L. Andrews, J. Am. Chem. Soc., 2002, 124,
5636.
43 H. Ahn and T. J. Marks, J. Am. Chem. Soc., 2002, 124, 7103.
31 X. F. Wang and L. Andrews, J. Phys. Chem. A, 2002, 106, 6720. 44 (a) G. Lapadula, D. Trummer, M. P. Conley, M. Steinmann,
´
32 F. Rataboul, A. Baudouin, C. Thieuleux, L. Veyre, C. Coperet,
Y.-F. Ran, S. Brasselet, Y. Guyot, O. Maury, S. Decurtins,
´
J. Thivolle-Cazat, J. M. Basset, A. Lesage and L. Emsley, J. Am.
Chem. Soc., 2004, 126, 12541.
S.-X. Liu and C. Coperet, Chem. Mater., 2015, 27, 2033; (b)
G. Lapadula, A. Bourdolle, F. Allouche, M. P. Conley, I. del
Rosal, L. Maron, W. W. Lukens, Y. Guyot, C. Andraud,
33 A. P. Legrand, H. Hommel, A. Tuel, A. Vidal, H. Balard,
E. Papirer, P. Levitz, M. Czernichowski, R. Erre,
H. Vandamme, J. P. Gallas, J. F. Hemidy, J. C. Lavalley,
O. Barres, A. Burneau and Y. Grillet, Adv. Colloid Interface
Sci., 1990, 33, 91.
´
S. Brasselet, C. Coperet, O. Maury and R. A. Andersen,
´
Chem. Mater., 2014, 26, 1062; (c) R. Metivier, I. Leray,
J.-P. Lefevre, M. Roy-Auberger, N. Zanier-Szydlowski and
B. Valeur, Phys. Chem. Chem. Phys., 2003, 5, 758; (d)
´
¨
¨
34 Y. Chen, E. Callens, E. Abou-Hamad, N. Merle, A. J. P. White,
D. Gajan, M. Schwarzwalder, M. P. Conley, W. R. Gruning,
´
M. Taouk, C. Coperet, E. Le Roux and J. M. Basset, Angew.
A. J. Rossini, A. Zagdoun, M. Lelli, M. Yulikov, G. Jeschke,
´
C. Sauvee, O. Ouari, P. Tordo, L. Veyre, A. Lesage,
Chem., Int. Ed., 2012, 51, 11886.
´
35 R. R. Schrock, K. Y. Shih, D. A. Dobbs and W. M. Davis, J. Am.
Chem. Soc., 1995, 117, 6609.
C. Thieuleux, L. Emsley and C. Coperet, J. Am. Chem. Soc.,
2013, 135, 15459.
36 D. A. Dobbs, R. R. Schrock and W. M. Davis, Inorg. Chim. 45 (a) M. V. Luzgin, V. A. Rogov, S. S. Arzumanov, A. V. Toktarev,
Acta, 1997, 263, 171.
37 (a) J. Joubert, F. Delbecq, P. Sautet, E. Le Roux, M. Taouk,
A. G. Stepanov and V. N. Parmon, Angew. Chem., Int. Ed.,
2008, 47, 4559; (b) J. F. Wu, W. D. Wang, J. Xu, F. Deng
and W. Wang, Chem.–Eur. J., 2010, 16, 14016; (c) U. Patil,
Y. Saih, E. Abou-Hamad, A. Hamieh, J. D. A. Pelletier and
J. M. Basset, Chem. Commun., 2014, 50, 12348; (d)
A. Hamieh, Y. Chen, S. Abdel-Azeim, E. Abou-hamad,
S. Goh, M. Samantaray, R. Dey, L. Cavallo and J. M. Basset,
ACS Catal., 2015, 5, 2164.
´
C. Thieuleux, F. Blanc, C. Coperet, J. Thivolle-Cazat and
J. M. Basset, J. Am. Chem. Soc., 2006, 128, 9157; (b) E. Le
Roux, M. Chabanas, A. Baudouin, A. de Mallmann,
´
C. Coperet, E. A. Quadrelli, J. Thivolle-Cazat, J. M. Basset,
W. Lukens, A. Lesage, L. Emsley and G. J. Sunley, J. Am.
Chem. Soc., 2004, 126, 13391.
38 The relative electronic energies obtained for other 46 A. Badia, L. Demers, L. Dickinson, F. G. Morin, R. B. Lennox
intermediates are also similar between those two models; and L. Reven, J. Am. Chem. Soc., 1997, 119, 11104.
the difference is within 5.0 kcal/mol. Because such 47 (a) F. Blanc, R. Berthoud, C. Coperet, A. Lesage, L. Emsley,
´
a difference is acceptable and equal to what one would
expect from simply changing of the DFT functional and
basis set, and also due to much larger CPU resources/time
required for periodic simulations, it was decided to use
R. Singh, T. Kreickmann and R. R. Schrock, Proc. Natl.
Acad. Sci. U. S. A., 2008, 105, 12123; (b) A. J. Jiang,
¨
J. H. Simpson, P. Muller and R. R. Schrock, J. Am. Chem.
Soc., 2009, 131, 7770.
the cluster model in the present study. For recent 48 F. W. Lytle, D. E. Sayers and E. A. Stern, Phys. B, 1989, 158,
computational applications of the same cluster model see
701.
This journal is © The Royal Society of Chemistry 2015
Chem. Sci.