Combined experimental and theoretical study on the reductive cleavage of inert C-O bonds with silanes: Ruling out a classical Ni(0)/Ni(II) catalytic couple and evidence for Ni(I) intermediates

Article abstract of DOI:10.1021/ja311940s

A mechanistic and computational study on the reductive cleavage of C-OMe bonds catalyzed by Ni(COD)₂/PCy₃ with silanes as reducing agents is reported herein. Specifically, we demonstrate that the mechanism for this transformation does not proceed via oxidative addition of the Ni(0) precatalyst into the C-OMe bond. In the absence of an external reducing agent, the in-situ-generated oxidative addition complexes rapidly undergo β-hydride elimination at room temperature, ultimately leading to either Ni(0)-carbonyl- or Ni(0)-aldehyde-bound complexes. Characterization of these complexes by X-ray crystallography unambiguously suggested a different mechanistic scenario when silanes are present in the reaction media. Isotopic-labeling experiments, kinetic isotope effects, and computational studies clearly reinforced this perception. Additionally, we also found that water has a deleterious effect by deactivating the Ni catalyst via formation of a new Ni-bridged hydroxo species that was characterized by X-ray crystallography. The order in each component was determined by plotting the initial rates of the C-OMe bond cleavage at varying concentrations. These data together with the in-situ-monitoring experiments by ¹H NMR, EPR, IR spectroscopy, and theoretical calculations provided a mechanistic picture that involves Ni(I) as the key reaction intermediates, which are generated via comproportionation of initially formed Ni(II) species. This study strongly supports that a classical Ni(0)/Ni(II) for C-OMe bond cleavage is not operating, thus opening up new perspectives to be implemented in other related C-O bond-cleavage reactions.

Full text of DOI:10.1021/ja311940s

Article
pubs.acs.org/JACS
Combined Experimental and Theoretical Study on the Reductive
Cleavage of Inert C−O Bonds with Silanes: Ruling out a Classical
Ni(0)/Ni(II) Catalytic Couple and Evidence for Ni(I) Intermediates
Josep Cornella,† Enrique Gomez-Bengoa,*,‡ and Ruben Martin*,†
́
†Institute of Chemical Research of Catalonia (ICIQ), Av. Països Catalans 16, 43007, Tarragona, Spain
‡Department of Organic Chemistry I, Universidad País Vasco, UPV/EHU, Apdo. 1072, 20080, San Sebastian, Spain
S
* Supporting Information
ABSTRACT: A mechanistic and computational study on the
reductive cleavage of C−OMe bonds catalyzed by Ni(COD)2/
PCy3 with silanes as reducing agents is reported herein.
Specifically, we demonstrate that the mechanism for this
transformation does not proceed via oxidative addition of the
Ni(0) precatalyst into the C−OMe bond. In the absence of an
external reducing agent, the in-situ-generated oxidative
addition complexes rapidly undergo β-hydride elimination at
room temperature, ultimately leading to either Ni(0)−
carbonyl- or Ni(0)−aldehyde-bound complexes. Character-
ization of these complexes by X-ray crystallography unambiguously suggested a different mechanistic scenario when silanes are
present in the reaction media. Isotopic-labeling experiments, kinetic isotope effects, and computational studies clearly reinforced
this perception. Additionally, we also found that water has a deleterious effect by deactivating the Ni catalyst via formation of a
new Ni-bridged hydroxo species that was characterized by X-ray crystallography. The order in each component was determined
by plotting the initial rates of the C−OMe bond cleavage at varying concentrations. These data together with the in-situ-
1
monitoring experiments by H NMR, EPR, IR spectroscopy, and theoretical calculations provided a mechanistic picture that
involves Ni(I) as the key reaction intermediates, which are generated via comproportionation of initially formed Ni(II) species.
This study strongly supports that a classical Ni(0)/Ni(II) for C−OMe bond cleavage is not operating, thus opening up new
perspectives to be implemented in other related C−O bond-cleavage reactions.
all these methodologies are not yet entirely attractive from an
atom-economical point of view due to generation of large
amounts of byproduct waste and the need for derivatization of
the starting phenol. Therefore, the use of more attractive C−O
electrophiles that would meet these challenges will be highly
desirable. In this regard, activation of C(sp2)−OMe bonds is
particularly attractive (Scheme 1, right) due to the fact that (a)
aryl methyl ethers are commercially available and inexpensive
building blocks and (b) aryl methyl ethers are the simplest
derivatives from phenols, thus becoming the most atom- and
step-economical counterparts in the phenol series.3
INTRODUCTION
■
In recent years, the use of phenol derivates as aryl C(sp2)−O
electrophiles in cross-coupling reactions have emerged as a high
cost-effective and environmentally friendly alternative to aryl
halide counterparts.1,2 Among their advantages is that halide
waste is avoided and that phenols are more readily available than
the corresponding aryl halides.2 Despite the advances realized,
the use of C(sp2)−O electrophiles is rather limited to its
sulfonate, carbamates, carbonates, esters, or phosphate deriva-
tives (Scheme 1, left).1 This is probably due to their remarkable
ability to act as leaving groups, thus facilitating the oxidative
addition step within the catalytic cycle. Unfortunately, however,
Unlike the use of other C(sp2)−O electrophiles, catalytic
cross-coupling reactions based upon cleavage of C(sp2)−OMe
bonds are still scarce.4,5 Indeed, chemists have the general
perception that C(sp2)−OMe bonds are rather inert, as
illustrated by the fact that the vast majority of metal-catalyzed
reactions perfectly tolerate the presence of aryl methyl ethers.2
Not surprisingly, the few existing methodologies for C(sp2)−
OMe bond cleavage mostly utilize highly reactive organometallic
species in stoichiometric amounts to overcome the remarkable
Scheme 1. C−O Electrophiles in Cross-Coupling Reactions
Received: December 14, 2012
Published: January 14, 2013
© 2013 American Chemical Society
1997
dx.doi.org/10.1021/ja311940s | J. Am. Chem. Soc. 2013, 135, 1997−2009

Journal of the American Chemical Society
Article
activation energy required for C(sp2)−OMe cleavage.4,5
Although no mechanistic studies have been conducted for C−
OMe bond-cleavage reactions,6 it is believed that all these
methodologies follow a classical catalytic cycle consisting of an
initial oxidative addition, transmetalation, and final reductive
elimination (Scheme 2).4,5
more complex organic frameworks such as quinine or estradiol-
type backbones (Scheme 3, bottom).7
Intriguingly, our Ni-catalyzed reaction of aryl methyl ethers
with silanes exhibited unexpected reactivity depending on the
catalyst and/or substrates of choice:7
(a) Use of Ni precatalysts other than Ni(COD)2 provided
trace amounts of the final products.
(b) Control experiments indicated that the reaction did not
proceed in the absence of catalyst or silane, thus ruling out
simple nucleophilic aromatic substitution or β-hydride
elimination pathways.
Scheme 2. Accepted Mechanism for C−OMe Cleavage
(c) Extended π systems gave consistently better reactivity than
simple anisole derivatives.
(d) Bulky and electron-rich monodentate PCy3 was critical for
success; under our reaction conditions, no other ligand,
neither monodentate nor bidentate, afforded the desired
coupling products, thus showing the subtleties of our
catalytic protocol.
Despite the recent advances in the field,4,5 a mechanistic
understanding of the Ni-catalyzed C−OMe bond cleavage still
remains elusive and speculative at this time.6 This is likely due to
in-situ formation of short-lived nickel species, which are
exceptionally sensitive as compared to their Pd analogues.
Recently, Agapie and co-workers reported an elegant stoichio-
metric, intramolecular mechanistic study on the Ni-mediated
reductive cleavage of C−O bonds with pincer-type ligands.15,16
However, all bidentate or pincer-type ligands analyzed were found to
be inert for the intermolecular Ni-catalyzed cleavage of C(sp2)−
OMe bonds promoted by silanes;7,11 as indicated above,
monodentate PCy3 was the only ligand that showed activity in a
catalytic fashion. Unlike the use of bidentate or pincer-type
ligands,17 however, the available literature data reveals limited
examples in which reaction mechanisms have been studied with
monodentate ligands in an intermolecular fashion.18 Most likely,
ligand dissociation, the presence of low-coordinate species, the
lack of a driving force for preparing metal complexes, and the
lower redox ability of the metal chelates as compared with the
bidentate or the pincer-type backbones constitutes serious
barriers for studying reaction mechanisms using monodentate
ligands.
Recently, we reported a Ni-catalyzed reductive cleavage of C−
OMe bonds7 using TMDSO8 as hydride donor (Scheme 3).9,10
Scheme 3. Ni-Catalyzed Reductive C−OMe Cleavage
Although we anticipated that a complete mechanistic
understanding for cleavage of C−OMe bonds in an intermolecular
fashion would be far from trivial, we decided to shed light into this
fast-growing and challenging area of expertise by studying the
reaction mechanism with the real catalytic system based on
monodentate PCy3 as the supporting ligand.7 Herein, we describe
our mechanistic investigations on the Ni-catalyzed reductive
cleavage of C−OMe bonds from an experimental and computa-
tional point of view. We demonstrate that, in the presence of
silanes, the widely accepted mechanism for C−OMe cleavage4,5
consisting of a Ni(0)/Ni(II) couple is not operating and that a
new pathway involving Ni(I) species is responsible for the
observed reactivity. We believe our results might have a
significant impact in other related C−O bond-cleavage reactions,
suggesting that mechanistic scenarios other than the classical
Ni(0)/Ni(II) couple could be conceivable as well.
As for other C−O bond-cleavage reactions, the use of PCy3 was
found to be the most critical factor for success.5 Shortly after,
Chatani and Tobisu11 independently described a related Ni-
catalyzed procedure of aryl methyl ethers and aryl pivaloates
using (MeO)2MeSiH as the reducing agent. Subsequently,
Hartwig reported a Ni-catalyzed protocol that reductively
cleaved C(sp2)−OAr and some C−OAlkyl bonds in the presence
of H2.12 Overall, this recent literature data clearly illustrates the
prospective impact of this methodology in the field of
homogeneous catalysis.13,14 Our catalytic protocol based on
the use of TMDSO was characterized by its wide substrate scope
and excellent chemoselectivity profile, allowing for coupling of a
wide variety of aryl methyl ethers with a different set of
substitution patterns (Scheme 3).7 Our methodology was found
to preferentially activate C(sp2)−OMe bonds in the presence of
C(sp3)−OMe bonds, allowing for site-selectivity approaches in
RESULTS AND DISCUSSION
■
Study of the Reactivity of Putative Reaction Inter-
mediates and Catalyst Deactivation. Prompted by our initial
results7 and available literature data,4−6 we formulated a working
hypothesis for our Ni-catalyzed reductive cleavage of C(sp2)−
1998
dx.doi.org/10.1021/ja311940s | J. Am. Chem. Soc. 2013, 135, 1997−2009

Journal of the American Chemical Society
Article
OMe bonds promoted by silanes. Thus, we believed that the
reaction was initiated by oxidative addition into the C(sp2)−
OMe bond (II), followed by σ-bond metathesis and a final
reductive elimination from a Ni(II)−hydride, delivering the final
product while regenerating the active Ni(PCy3)2 (I) species
(Scheme 4).
Scheme 5. Stoichiometric Studies
Scheme 4. Initial Proposed Mechanistic Hypothesis
geometry (Figure 1). With substantial amounts of 3 in hand, we
hypothesized that the oxidative addition species 4 could be easily
To gain insights into this mechanism, we decided to focus our
attention on the oxidative addition complex II (Scheme 4).
Beyond any reasonable doubt, a deeper understanding of the
reactivity of II will have significant ramifications in other related
C−O bond-cleavage processes described in the literature as
oxidative addition has invariably been proposed to be the first
step within the catalytic cycle.4,5 Given that extended π systems
have systematically been found to be much more reactive than
regular anisole derivatives in C(sp2)−OMe cleavage reac-
tions,4,5,7,11 we started our study with 2-methoxynaphthalene
(1). Unfortunately, all our attempts of isolating II by reacting 1
with Ni(COD)2/PCy3 were not successful. A simple spectro-
scopical analysis of the crude reaction mixture revealed a
significant amount of Ni(COD)2, suggesting incomplete
conversion to the active Ni(PCy3)2 species (I). These results
questioned the simple and commonly accepted assumption that
PCy3 easily replaces COD under the reaction conditions,5
arguably indicating that COD competes with substrate binding.
Indeed, we performed titration experiments that showed an
equilibrium between Ni(COD)2 and Ni(PCy3)2 (I) by NMR
spectroscopic analysis, favoring the former over the desired 14-
electron species I.19 While this might seem undesirable at first
sight, the presence of COD in the reaction medium actually turns
out to be highly advantageous as other Ni precatalysts such as
NiCl2 or NiCl2·glyme in combination with Zn as reducing agent
were not as reactive as Ni(COD)2.19 We tentatively propose that
COD might indeed serve as ancillary ligand to protect the
propagating and highly reactive Ni(PCy3)2 species I in the
resting state of the catalyst.
Figure 1. ORTEP diagram of 3. Selected bond lengths (Angstroms) and
angles (degrees): Ni1−Cl1, 2.2252(8); P2−Ni1, 2.2583(8); P1−Ni1,
2.2430(8); C11−Ni1, 1.927(10); C1−Ni1−P1, 174.71(9); C1−Ni1−
P2, 90.10(10).
obtained by anion metathesis with NaOMe (Scheme 5, top).22
According to our expectations, 3 was rapidly converted at room
temperature in less than 2 h to a new species by treatment with
NaOMe in THF.23 Surprisingly, however, naphthalene (5) was
formed almost quantitatively and 31P NMR spectroscopic data
did not match with the expected oxidative addition species 4. The
identity of this new compound was finally revealed by X-ray
structure analysis. As shown in Figure 2, this complex turned out
to be the symmetrical trigonal Ni(PCy3)2CO complex 6.
Similarly, exposure of 1 to NaOEt cleanly produced a new
complex at room temperature after 2 h reaction time (Scheme 5,
bottom) that did not correspond to the oxidative addition
complex 7. X-ray crystallography unambiguously identified this
complex as 8 (Figure 3). In contrast to 6, however, the molecular
structure of 8 shows that an ethanal molecule is coordinated to
the Ni(0) center in a η2-fashion, resulting in a distorted
tetrahedral geometry with different Ni−O and Ni−C distances
(Figure 3).24 As expected from a nonsymmetrical geometry, 8
In light of these observations, an indirect route for the
oxidative addition species II was envisaged (Scheme 5). Reaction
of (2-naphthyl)bis(triphenylphosphine)nickel(II) chloride 2,
easily obtained from 2-chloronaphthalene and Ni(PPh3)4,20
with 2 equiv of PCy3 in acetone cleanly afforded 91% of 3.21 Such
complex was unambiguously characterized by X-ray crystallog-
raphy, showing that the Ni atom is in a square-planar geometry
surrounded by two phosphorus atoms in trans coordination
1999
dx.doi.org/10.1021/ja311940s | J. Am. Chem. Soc. 2013, 135, 1997−2009

Journal of the American Chemical Society
Article
event initiated by oxidative addition into the aldehydic C−H
bond ultimately leads to 6.25 Unlike the dehydrogenation event
in route to 6, however, extrusion of methane via decarbonylation
of ethanal (R = Me) was not observed. Thus, complex 8 was
isolated in analytically pure form without traces of 6 detected by
NMR spectroscopy of the crude reaction mixture, even after
longer reaction times. While β-hydride elimination from
alkoxide-bound metal complexes has been reported in the
literature with pincer-type ligands, these reactions usually
required high temperatures;15−26 it is noteworthy that in the
presence of PCy3, however, β-hydride elimination occurred at an
exceptional rate at room temperature. These results evidence the
known ability of bidentate ligands to significantly retard β-
hydride elimination.27
The data summarized above arguably illustrate that the
oxidative addition species 4 and 7 underwent β-hydride
elimination with exceptional ease in the absence of silanes,
even at room temperature (Scheme 5). We speculated that
preparation of structurally well-characterized Ni(PCy3)2 (I)
species, which are more amenable to mechanistic studies, would
shed light into these intriguing observations. To such end, the
known complex [Ni(PCy3)2]2N2 (9)28 seemed ideally suited for
our purpose as 9 generates, in situ, Ni(PCy3)2 in argon
atmospheres. Importantly, treatment of 1 with 9 (0.50 equiv)
in toluene at 100 °C afforded 5 and 6 in 54% yield in the absence
of silane (Scheme 7).
Figure 2. ORTEP diagram of 6. Selected bond lengths (Angstroms) and
angles (degrees): Ni1−C37, 1.719(5); P2−Ni1, 2.2047(7); P1−Ni1,
2.2223(7); C37−Ni1−P2, 117.09(8); Ni1−C37−O, 170(3).
Scheme 7. Reactivity of 1 with 9 in the Absence of Silane
Figure 3. ORTEP diagram of 8. Selected bond lengths (Angstroms) and
angles (degrees): Ni1−P1, 2.2326(17); Ni1−P2, 2.1410(16); Ni−O1,
1.872(4); Ni1−C37, 1.933(6); C37−Ni1−P1, 133.56(18); O1−Ni1−
P1, 94.30(13); O1−Ni1−C37, 40.2(2).
Although the data in Schemes 5 and 7 showed that β-hydride
elimination pathways are indeed conceivable in a stoichiometric
fashion, we believe that this particular manifold represents a
minor contribution, if any, under our catalytic conditions. This
assumption is supported by the fact that not even traces of
naphthalene (5) were observed upon exposure of 1 to our optimized
catalytic protocol based upon Ni(COD)2 and PCy3 in the absence of
silane (Table 1, entry 1). As shown in entries 2 and 3, control
experiments in the absence of Ni(COD)2, PCy3, or silane did not
afford the desired product. Similarly, no reaction was observed in
the presence of H2 (entry 4).12 In sharp contrast, addition of 1
equiv of TMDSO or 2 equiv of Et3SiH afforded naphthalene in
quantitative yields (entry 5). In line with the stoichiometric
results in Scheme 8, use of catalytic amounts of 9 in the absence
of silane resulted in non-negligible amounts of 5 (entry 6).
Likewise, the behavior of Ni(PCy3)2(C2H4) (10)28b followed a
similar pattern (entry 7); in this particular case, however, 5 was
formed at much less extent, reinforcing the notion that ancillary
ligands such as COD or ethylene might indeed play a critical role
for preventing, or at least retarding, β-hydride elimination. The
fact that not full conversion is achieved for both 9 and 10
precatalysts in the absence of silane (entries 6 and 7) contributes
to the perception that 6, the Ni complex that is generated upon β-
hydride elimination and dehydrogenation events (Schemes 6 and
7), was not catalytically active. As expected, 6 was not a suitable
Ni precatalyst under our reaction conditions. In sharp contrast,
both 9 and 10 were catalytically competent in the presence of
TMDSO, resulting in a 54% and 90% yield of 5, respectively
(entries 8 and 9).
showed two pair of doublets at 48.5 and 35.9 ppm by 31P NMR
spectroscopy.
Formation of 6 and 8 is best explained by an initial halogen
metathesis leading to III. Upon β-hydride elimination, III
generates a naphthyl−Ni(II) hydride IV that rapidly undergoes
reductive elimination, giving rise to naphthalene (5) and
Ni(PCy3)2 that likely remains bound to the aldehyde template
in a η2-fashion (Scheme 6). Subsequently, a dehydrogenation
Scheme 6. Mechanistic Proposal for Complexes 5 and 6
2000
dx.doi.org/10.1021/ja311940s | J. Am. Chem. Soc. 2013, 135, 1997−2009

Journal of the American Chemical Society
Article
a
Table 1. Reaction of 1 with Different Ni(0) Precatalysts
b
entry
[Ni]
TMDSO (equiv)
GC yield (%)
1
2
3
Ni(COD)2/PCy3
Ni(COD)2
none
0
0
0
0
1
none
1
c
4
Ni(COD)2/PCy3
Ni(COD)2/PCy3
[(Cy3P)2Ni]2N2 (9)
(Cy3P)2Ni(C2H4) (10)
[(Cy3P)2Ni]2N2 (9)
(Cy3P)2Ni(C2H4) (10)
Ni(COD)2/PCy3
none
d
5
6
7
8
9
1
0
0
1
1
1
99
6
2
54
90
5
e
10
a
Reactions were carried out with 1 (0.25 mmol), TMDSO (0.25
b
mmol), and Ni(0) precatalyst (10 mol %) at 110 °C for 14 h. Yields
determined by GC analysis using decane as internal standard.
c
d
Figure 4. ORTEP diagram of 11. Selected bond lengths (Angstroms)
and angles (degrees): P1A−Ni1A, 2.1660(15); Ni1A−O1A, 1.897(17);
Ni1A−C1, 1.888(4); P1A−Ni1A−O1A, 97,0(5); P1A−Ni1A−O1,
170.0(4); P1A−Ni1A−C1, 95,70(13).
Reaction was carried out under 1 atm of H2. Quantitative yield
e
was also observed when Et3SiH (2 equiv) was utilized. A 1 equiv
amount of degassed H2O was added.
Scheme 8. Isotope-Labeling Experiments
Taken together, the above data show not only the striking
different reactivity of Ni(0) precatalysts (Table 1) but also the
deleterious effect that water might have in the reaction outcome
(Figure 4).19 The difference in reactivity is particularly
pronounced with Ni(COD)2 as the catalyst. The superior
catalytic activity of the latter illustrates the crucial role of COD,
acting as a noninnocent ancillary ligand that stabilizes the resting
state of the catalyst and retards the subsequent oxidative addition
step.31 However, the ability of the oxidative addition species to
rapidly undergo β-hydride elimination in the absence of silane,
even at room temperature, is rather striking and certainly does
not clarify the role of silane in our catalytic protocol.
Isotope-Labeling Studies and Kinetic Isotope Effects.
In order to fully understand the role of the silane on the Ni-
catalyzed reductive cleavage of C(sp2)−OMe bonds we carried
out isotope-labeling studies with Et3SiD and 1-D (Scheme 8).32
If the reaction commences with oxidative addition, then one
might anticipate some degree of deuterium scrambling in the
product as we demonstrated that β-hydride elimination rapidly
takes place at room temperature (Scheme 5). As shown in
Scheme 8, exclusive formation of 2-deuteronaphthalene (5-D)
was observed by reacting 1 with Et3SiD, employing a catalytic
system based upon either Ni(COD)2/PCy3 or 9 (99% and 55%,
respectively; Scheme 8, top right). A similar argument applies for
the observed opposite labeling pattern by reacting 2-deuter-
omethoxynaphthalene (1-D) with Et3SiH (Scheme 8, top left).
Within the limits of detection, no sign of deuterium scrambling
was observed in either case. These experimental results provide
strong evidence that silanes are indeed the hydride sources
responsible for yielding naphthalene and that a β-hydride
elimination pathway is unlikely in our catalytic protocol.
Notably, we found that that addition of 1 equiv of water to the
reaction of 1 under our optimized conditions resulted in almost
no conversion to 5 (Table 1, entry 10). Likewise, traces of water
had also a detrimental effect in this reaction, resulting in much
slower rates and lower yields of 5.19 Interestingly, NMR
spectroscopy of the crude reaction mixtures showed significant
amounts of Et3SiOSiEt3, suggesting catalyst decomposition in
the presence of trace amounts of water. Beyond any doubt, a
better understanding of this decomposition pathway would be
critical for designing more efficient and improved catalysts.
Gratifyingly, careful crystallization at low temperatures of a
stoichiometric reaction in the presence of water furnished
crystals suitable for X-ray crystal structure analysis, in which the
Ni(II) complex 11 was obtained (Figure 4).29 Such complex
contains two hydroxo bridging groups in addition to a naphthyl
unit and PCy3. While identification of 11 shows that catalyst
decomposition might occur if not rigorous anhydrous conditions
are utilized, its molecular structure suggests that the elusive
oxidative addition species 4 might contain bridging alkoxy groups
as well.30
In line with our previous observations in which an ancillary
ligand is not present (Scheme 7), reaction of 1-D with
stoichiometric amounts of 9 in the absence of silane delivered
exclusively 5-D (Scheme 8, bottom right); note, however, that in
the presence of Et3SiH, 1-D gave rise to 5 with no traces of 5-D
being detected in the crude reaction mixture (Scheme 8, bottom
left). We believe these results clearly show the intriguing
2001
dx.doi.org/10.1021/ja311940s | J. Am. Chem. Soc. 2013, 135, 1997−2009

Journal of the American Chemical Society
Article
dichotomy on the reaction outcome depending on whether
silanes are present or not in the reaction media.
More interestingly, no kinetic isotope effect (kH/kD = 1) was
observed when comparing the initial rates of 1 with Et3SiH and
Et3SiD, suggesting that Si−H bond cleavage is not involved in the
rate-determining step;33,34 if our initially postulated mechanism
is correct (Scheme 4), the observed kinetic isotope effect
together with the results shown in Scheme 8 suggested that σ-
bond metathesis must proceed at a faster rate as compared to β-
hydride elimination. This assumption is, however, somewhat
controversial given the exceptional rate for β-hydride elimination
in our stoichiometric studies, thus leaving some doubt with
regard to the proposed catalytic cycle based upon an oxidative
addition followed by σ-bond metathesis (Scheme 4).
Kinetic Studies: Insights into the Turnover-Limiting
Step. The results in Schemes 5 and 7 together with the kinetic
isotope effects suggest that our initial mechanistic hypothesis
(Scheme 4) might have been premature and that further
experiments would deserve more careful consideration. Con-
vinced of the relevance of this study, we continued performing
more systematic investigations in order to elucidate how silanes
are involved and what their exact role is in our reductive catalytic
protocol. Thus, we turned our attention to determine the order
of all reaction components as it might give us valuable
information about the species that are involved in the rate-
determining step.
Figure 6. Plot of initial rates vs concentration of catalyst.
Initial rates were monitored by taking aliquots from the
reaction mixture and analyzed by gas chromatography, changing
the concentration of each reactant using 1 and Ni(COD)2/PCy3
as substrate and catalyst. Given the similar reactivity of TMDSO
and Et3SiH under our reaction conditions, we chose Et3SiH as
the silane source. As shown in Figures 5 and 6, the reaction
Figure 7. Plot of initial rates vs concentration of Et3SiH.
it was highly informative to observe that our model reaction was
partially inhibited upon increasing the naphthalene (5)
concentration;19 such inverse dependence implies that product
5 competes with 1 for substrate binding.
Theoretical Calculations for a Mechanism Consisting
of a β-Hydride Elimination or σ-Bond Metathesis. We
performed DFT calculations with the Gaussian 09 suite of
programs.35 Optimizations were carried out using the standard 6-
31G(d,p) basis set for C, H, O, P, and Si and the LANL2DZ (Hay
and Wadt)36 basis including an effective core potential for Ni.
Single-point calculations were performed on the optimized
structures with B3LYP37 and Truhlar’s M0638 functionals in
combination with different large basis sets, like 6-311++G(d,p)
for C, H, O, P, and Si atoms and SDD (Stuttgart−Dresden
effective core potential)39 for Ni, or def2-TZVPP basis set. The
values reported in this work, which are given in kcal/mol,
correspond to Gibbs free energies (G) and include thermal and
zero-point vibrational corrections (ZPVE). When necessary, a
solvent model system (IEFPCM, toluene)40 was also included.
We used HSiMe3 as a computational model of HSiEt3 to avoid
the high conformational complexity of the latter.41 We initially
Figure 5. Plot of initial rates vs concentration of 1.
exhibits a first-order dependence on both catalyst and substrate
1. While these results might indeed account for oxidative
addition being rate determining, the surprising first-order
dependence on Et3SiH (Figure 7) reinforces the perception
that a different mechanism is operative in our reaction
conditions. These striking results are in sharp contrast with the
originally postulated mechanism in which oxidative addition was
believed to be rate determining (Scheme 5). On the other hand,
2002
dx.doi.org/10.1021/ja311940s | J. Am. Chem. Soc. 2013, 135, 1997−2009

Journal of the American Chemical Society
Article
Scheme 9. Gibbs Free Energy Surfaces of the Favored Pathways for C−OMe Bond Cleavage via Oxidative Addition Followed by σ-
a
Bond Metathesis or β-Hydride Elimination
a
Energies are in kcal/mol and calculated using M06/6-311++G**(SDD). Values in parentheses correspond to B3LYP/def2-TZVPP.
used PMe3 as a simplified model of the more demanding ligand
PCy3; after observing divergent dissociation energy values
between both phosphines, however, we decided to include
PCy3 in the most critical steps of the mechanism.
Ni(COD)2 as compared to Ni(PCy3)2 (I).42 As shown in
Scheme 9, displacement of the two COD ligands by 1 and PCy3 is
highly disfavored (INT-1, +9.8 kcal/mol, M06). Furthermore,
oxidative addition into the C(sp2)−OMe bond from INT-1 had a
high activation Gibbs free energy of 30.6 kcal/mol, affording a
total activation barrier of 40.4 kcal/mol from Ni(COD)2 to TS1.
This high computed energy contributes to the notion that a
mechanism based upon a classical oxidative addition into the C−
OMe bond using Ni(COD)2 as precatalyst is highly unlikely
(Scheme 4). In sharp contrast, the M06 functional predicted an
activation barrier of 35.1 kcal/mol for oxidative addition from
Ni(PCy3)2 (TS1), making such process feasible at high
temperatures. This is in agreement with the experimental
observation that [Ni(PCy3)2]2N2 (9) facilitates oxidative
addition at 110 °C (Scheme 7). Once again, B3LYP failed to
predict this experimental evidence43 as oxidative addition from
Ni(COD)2 has a remarkable low activation Gibbs free energy of
26.4 kcal/mol from Ni(COD)2 (TS1). It is worth noting that the
dissociation energy values obtained for PMe3 or PCy3 toward
formation of Ni(PR3)2 are different enough to discourage the use
of PMe3 as a valid model of PCy3, at least regarding comparison
of Ni species of different coordination number.
We next studied the ability of the key oxidative addition
intermediate INT-2 to form naphthalene (5), in both the
absence or the presence of silane (Scheme 9). Intermediate INT-
2 bears a coordination vacancy that can serve both to
accommodate the forming hydride in the β-elimination event
(absence of silane) or to coordinate the silane and promote the σ-
bond metathesis process. In the absence of silane (Scheme 9, red
lines) we confirmed computationally that the transformation of
INT-2 into 5 follows an exothermic pathway through a series of
low-energetic transition states, namely, β-hydride elimination
(TS2), C−H oxidative addition (TS3), and α-elimination (TS4)
Our available data unambiguously suggested that a different
mechanistic scenario came into play. We anticipated that
computational studies would allow gathering indirect but rather
important evidence for such hypothesis. Specifically, we aimed to
shed light into the following striking observations: (a) no
reaction is observed when utilizing a catalytic system based upon
Ni(COD)2/PCy3 until silane is added to the reaction mixture;
(b) isotope-labeling studies illustrated that the proposed σ-bond
metathesis should occur at a faster rate than a β-hydride
elimination pathway; (c) the lack of a kinetic isotope effect
indicated that the Si−H bond was not irreversibly cleaved during
the rate-determining step, and (d) the reaction showed a first-
order dependence in substrate, catalyst, and silane concentration.
The controversial findings that 1 afforded cleanly naphthalene
(5) when operating with 9 and not with our optimized protocol
based upon Ni(COD)2/PCy3 (Scheme 7 and Table 1) prompted
us to study computationally the more accessible pathway toward
the putative oxidative addition species 4 depending on the
catalyst of choice.
We experimentally demonstrated by titration studies that
PCy3 does not easily replace COD from Ni(COD)2 complex to
form the 14-electron Ni(PCy3)2 species (I).19 We first computed
the equilibrium energies between both Ni species and also the
coordination of 1 to form INT-1 (Scheme 9). While B3LYP
showed that the equilibrium is shifted to Ni(PCy3)2 (I), M06
functional theory correctly predicted that Ni(PCy3)2 is 5.3 kcal/
mol higher in energy than Ni(COD)2 (Keq = 1.14 × 10−4). These
results corroborated that the B3LYP functional was not
appropriate to predict the experimental greater stability of
2003
dx.doi.org/10.1021/ja311940s | J. Am. Chem. Soc. 2013, 135, 1997−2009

Journal of the American Chemical Society
Article
to afford naphthalene 5 and the experimentally observed
(PCy3)2Ni−CO complex 6. In the presence of silane (Scheme
9, blue lines), the σ-bond metathesis is fast, leading to a highly
reactive intermediate INT-5 (TS5), which suffers an instante-
nous C−H bond-reductive elimination to INT-6. Subsequently,
INT-6 would form either Ni(PCy3)2 (I) or INT-1 by
coordination with either PCy3 or 1, respectively. In the event
of a hypothetical competition between both pathways, the σ-
bond metathesis process (TS5) has a barrier of 3.4 kcal/mol
lower than the β-hydride elimination (TS2). These results are in
line with the observed isotope-labeling studies (Scheme 8). More
interestingly, both pathways should be relatively facile at room
temperature as soon as INT-2 is formed in the reaction medium.
These results are in perfect agreement with the experimental
observation that 3 rapidly reacts with NaOMe at room
temperature, giving rise to 6 (Scheme 5).
While the M06 functional predicted that σ-bond metathesis
was faster than β-hydride elimination (TS5 vs TS2), there were
still several inconsistencies that left some doubt about the
proposed mechanism in Scheme 4: (a) the activation barrier in
route to INT-2 from Ni(COD)2 was too high to be achievable
under reaction conditions based upon Ni(COD)2 and PCy3
(TS1, 40.4 kcal/mol); note, however, that the reaction can occur
at some extent from Ni(PCy3)2 (TS1, 35.1 kcal/mol). This
argument should be valid independently on whether a silane is
present or not as INT-2 is the common intermediate of both
scenarios; (b) according to the theoretical calculations in Scheme
9, the silane does not participate during or before the rate-
determining step (TS1).44 These results are clearly inconsistent
with the observed first-order dependence on silane concen-
tration.
Figure 8. In-situ monitoring by FTIR spectroscopy.
On the basis of the above experimental mechanistic data, we
hypothesized that at initial stages Et3SiH might bind reversibly to
Ni(PCy3)2 (I) in a η2-fashion (VI) (Scheme 10). Related Ni
Scheme 10. Formation of Ni(I)−H and Ni(I)−SiR3 Species
a
from Ni(COD)2
In-Situ Monitoring: Evidence for Ni(I) Intermediates. In
view of the controversial results gathered by our kinetic and
computational studies, we set out to explore and clarify in more
detail the critical role of the silane in our catalytic protocol.
Accordingly, we turned our attention to in situ monitor the
course of a catalytic reaction of 1 with Ni(COD)2/PCy3 by gas
chromatography. Interestingly, we observed that reaction with
Et3SiD evidenced a slightly longer induction period as compared
with its Et3SiH analogue, thus evidencing the critical role of
silanes for generating the active propagating species.19 In order to
study in more depth the striking difference in these observed
induction periods, we further monitored the model reaction by
1H NMR spectroscopy at 100 °C in a stoichiometric fashion.
Notably, we observed that rapid consumption of Et3SiH
occurred when mixing Et3SiH (2 equiv), 1 (1 equiv), Ni(COD)2
(1 equiv), and PCy3 (2 equiv), even when naphthalene (5) was
not yet formed. Additionally, we also monitored the reaction by
29Si NMR spectroscopy during the induction period and showed
no other species than unreacted Et3SiH. Taken together, we
believe these results suggest the intermediacy of paramagnetic
species present in the reaction mixture. It is also worth noting
that during the course of the reaction where naphthalene (5) was
being formed the only species observed by 29Si NMR were
Et3SiOMe.
a
Gibbs free energies computed at M06/6-311++(d,p) (SDD) (blue).
complexes have previously been reported, although low
temperatures are generally required due to their high
instability.46,47 We hypothesized that VI presumably undergoes
oxidative addition into the Si−H bond, affording a Ni(II)
intermediate VII.48 Subsequently, a comproportionation of these
species and Ni(PCy3)2 (I)49 would lead to a Ni(I)−hydride
(VIII) and Ni(I)−SiEt3 (IX).50 Consistent with the available
literature data on related Ni(I)−hydrides, we believe that VIII
are diamagnetic species that contain a NiI−NiI bonding
interaction.51 This hypothesis is supported by the fact that H
1
NMR spectroscopy of a crude reaction mixture showed a
characteristic signal at −15.8 ppm during the course of the
reaction.19 This value is in analogy with the chemical shift of an
otherwise similar Ni(I)−hydride complex 12 containing a
bidentate linkage with tricyclohexylphosphine units that was
prepared following a slightly modified literature procedure.52,53
Careful crystallization from pentanes at −40 °C furnished
crystals suitable for X-ray structure analysis (Figure 9).54 Still,
however, our experiments did not allow us to rule out an
equilibrium of VIII with the corresponding monomeric and,
therefore, paramagnetic Ni(I)−H species. As for Ni(I)−SiEt3
species, we speculate that the steric bulk imposed by the bulky
PCy3 results in only monomeric, tricoordinated paramagnetic
species IX.50
To shed more light into the initial consumption of Et3SiH at
short reaction times, we also monitored the reaction by in-situ IR
spectroscopy (Figure 8).45 As expected from the observed H
1
NMR spectroscopical data, depletion of the signal corresponding
to the Si−H at 812 cm−1 was observed (Figure 8, blue line). It is
important to highlight that during the induction period in which
consumption of Et3SiH is observed, no naphthalene (5) was detected
by either gas chromatography or 1H NMR spectroscopy.
2004
dx.doi.org/10.1021/ja311940s | J. Am. Chem. Soc. 2013, 135, 1997−2009

Journal of the American Chemical Society
Article
Since our available experimental data did not allow us to
rigorously distinguish whether Ni(I)−H or Ni(I)−SiR3 was the
actual propagating species, we turned our attention to theoretical
calculations to confirm the role of Ni(I) intermediates in our
catalytic protocol.
Theoretical Calculations for the Intermediacy of
Catalytic Ni(I) Species. As shown in Scheme 10, we confirmed
computationally our in-situ-monitoring studies that suggested
the intermediacy of Ni(I) species (blue values). While the
transformation of Ni(COD)2 into the INT-7 was initially
disfavored, this effect was compensated by the large energy gain
measured in the comproportionation event between INT-7 and
Ni(PMe3)2, thus rendering the Ni(I)−H dimer INT-8 and
Ni(I)−SiMe3 species INT-9.56
In principle, two different mechanisms could be envisioned
depending on the Ni(I) source. The high stability of the Ni(I)−
H dimer (INT-8, Scheme 11) suggested that the reaction should
be initiated via monomeric Ni(I)−H species INT-10. Sub-
sequently, displacement of PCy3 by 2-methoxynaphthalene (1)
would render a new Ni(I) species (INT-11). At this stage, several
potential pathways could be conceivable for reaction of INT-11
with the aromatic backbone, including σ-bond metathesis or
oxidative addition across the C−OMe bond, among others.19
Among these, the lowest activation energies corresponded to
Ni−H insertion into the naphthyl ring57 (TS7, 18.1 kcal/mol
from INT-11) to form a reactive intermediate, wherein the Ni(I)
center and the methoxy group are positioned in contiguous
carbons in an anti fashion.
Figure 9. ORTEP diagram of 12. Selected bond lengths (Angstroms)
and angles (degrees): Ni1−P1, 2.1350(7); Ni1−P2, 2.1356(7); Ni1−
Ni2, 2.4078(5); P1−Ni1−P2, 91,52(3); P1−Ni1−Ni2, 133.63(2); P2−
Ni1−Ni2, 134.73(2).
To gain more insight on whether paramagnetic species are
present in our reaction mixture, electron paramagnetic resonance
(EPR) was performed at low temperatures by taking aliquots of a
stoichiometric and a catalytic reaction of 1 with Ni(COD)2/
PCy3 and Et3SiH during both the induction period and at low
conversions to naphthalene (5). Interestingly, a characteristic
EPR spectrum for Ni(I) species was obtained in all cases,19,55 even
after the induction period in which naphthalene (5) was being
formed. We believe these experiments advocate the notion that
Ni(I) intermediates are indeed responsible for the catalytic
activity in our protocol (Scheme 10).
We located a transition state (TS8) in which the Ni(I) atom
migrates from the α- to the β-position of the naphthyl ring, hence
triggering elimination of methanol from the β-carbon (Scheme
a
Scheme 11. Formation of Naphthalene (5) via Catalytic Ni(I)−H Species
a
Gibbs free energies calculated at M06/6-311++G(d,p) (SDD). Bold values correspond to PCy3 and values in italic to PMe3.
2005
dx.doi.org/10.1021/ja311940s | J. Am. Chem. Soc. 2013, 135, 1997−2009

Journal of the American Chemical Society
Article
a
Scheme 12. Formation of Naphthalene Through Ni(I)−SiR3 Species
a
Gibbs free energies calculated at M06/6-311++G(d,p) (SDD). Bold values correspond to PCy3 and values in italic to PMe3.
11). It is worth noting that a substantial amount of negative
charge is developed in TS8; while this is, not surprisingly,
penalized in the gas phase, a solvent model was applied to acquire
a more accurate energy, rendering a value of 31.1 kcal/mol, which
is of similar magnitude to TS7. The reaction further evolves by
forming (PCy3)2Ni−OMe (INT-14), which reacts with silane
through a low in energy σ-bond metathesis-type transition state
TS10, thus recovering back the catalyst INT-10. It is worth
noting that although we computed the full scheme with PMe3 as
the ligand, we checked the critical points with PCy3 and
confirmed that similar conclusions can be drawn for both
phosphines. Nevertheless, we believe the accuracy of these
results does not have any influence in the mechanism due to the
following: (a) (R3P)2Ni(I)−H dimers (INT-8) are remarkably
more stable than the corresponding monomeric (R3P)2Ni(I)−H
species (INT-10). As judged by the calculations using PMe3, this
seemingly simple transformation is heavily penalized by the 15.5
kcal/mol difference (Scheme 11, bottom left). (b) Ni−H and
Si−H bonds are broken and formed during the rate-determining
step TS7 (or TS8), thus suggesting that a positive kinetic isotope
effect should have been observed; as already demonstrated (see
above), this was not the case. (c) The silane does not participate
during or before the rate-detemining step; and therefore, it is not
consistent with the first-order dependence on silane concen-
tration (Figure 7).
INT-16 into the naphthyl ring with an activation energy of 21.7
kcal/mol (Scheme 12, TS12). The intermediate INT-17 then
evolves by MeOSiMe3 elimination and Ni(I) migration to form
INT-18 that easily incorporates an additional ligand on its
structure to afford INT-15. Subsequently, INT-15 undergoes a
reversible σ-bond metathesis (TS11), thus giving rise to
naphthalene (5) and recovering back the Ni(I)−SiR3 species
INT-9.
As shown in Scheme 12, it becomes apparent that the
migratory insertion into the naphthalene backbone represents
the rate-detemining step of the reaction (TS12, 32.9 kcal/mol
operating with PCy3). Significantly, the lowest energy
intermediate of the proposed mechanism does not correspond
to the initial Ni(I)−SiR3 species (INT-9) but rather the 2-
naphthyl−Ni(I)(PCy3)2 complex (INT-15). As a result, the
latter can formally be considered the resting state of the catalytic
cycle. Taking a closer look into the mechanism shown in Scheme
12, our results also suggest that both 2-methoxynaphthalene (1)
and the silane actively participate from the lowest reaction
intermediate (INT-15) to the highest data point of the catalytic
cycle (TS12). Such assumption cannot be underestimated as it
virtually explains, for the first time, the observed first-order
dependence on both 1 and Et3SiH. Moreover, no Ni−H or Si−H
is broken or formed during the rate-determining step (TS12), an
observation that is in perfect in agreement with the lack of kinetic
isotope effect when comparing the initial rates of R3SiH vs R3SiD.
Additionally, the means to dearomatize an arene ring at the rate-
determining step (TS12) nicely explains the greater reactivity of
extended π systems as compared to simpler anisole deriva-
tives.4,5,7,11
According to the results shown in Scheme 11, we concluded
that our protocol for cleavage of aryl methyl ethers in the
presence of silanes could not operate under a catalytic regime
based upon Ni(I)−H species. Therefore, we next turned our
attention to a mechanistic alternative in which Ni(I)−SiR3
species participate in the catalytic cycle (Scheme 12). As for
the Ni(I)−H catalytic proposal (Scheme 11), the lowest energy
pathway corresponded to the Ni(I)−SiR3 bond insertion from
As a result and taking into consideration all the experimental
data and theoretical calculations, we believe that the mechanism
from which our reaction operates is initiated by Ni(I)−SiEt3
2006
dx.doi.org/10.1021/ja311940s | J. Am. Chem. Soc. 2013, 135, 1997−2009

Journal of the American Chemical Society
Article
(IX)50 that are generated from Ni(COD)2, PCy3, and Et3SiH via
a comproportionation event (Scheme 13). Such assumption
Ni(COD)2 in solution, thus leaving some doubt about the
accuracy of such test in our protocol.60,61 Additionally, we found
that upon filtration of the mixture after 50 min reaction time,
further conversion to naphthalene (5) was observed in a linear
relationship over time (Figure 10b);19 interestingly, no induction
period was observed, suggesting the presence of homogeneous
catalysts in the reaction mixture. Nevertheless, at present we
cannot completely rule out a leaching from heterogeneous
species, thus acting as a reservoir of the corresponding Ni
homogeneous active species. In addition, TEM and ESEM
analysis of different aliquots during the course of the reaction
revealed formation of Ni−P particles with an average size of
200−2000 nm.19 Arguably, such particle size rules out the
intermediacy of Ni nanoparticles and suggests that if Ni−
aggregates are present, these likely coexist in equilibrium with the
homogeneous catalysts.62
Scheme 13. Mechanistic Proposal
CONCLUSIONS
■
This study provides compelling experimental and computational
evidence that the reported Ni-catalyzed C−OMe bond-cleavage
reactions occur along more than one pathway depending on the
catalyst of choice. We demonstrate that the presence of ancillary
ligands such as COD or ethylene play an important, if not critical,
role in the reaction outcome, hence stabilizing the resting state of
the catalyst and therefore retarding the subsequent oxidative
addition event. On the contrary, the absence of ancillary ligands
makes oxidative addition feasible. This later premise is
corroborated by the unusual molecular structures 6 and 8,
unambiguously indicating that in-situ-formed alkoxy-bound Ni
complexes are able to rapidly undergo at room temperature β-
hydride elimination in the absence of ancillary ligand. The fact
that 6 and 8 are not catalytically competent contributed to the
perception that β-hydride elimination was not responsible for the
catalytic activity. Additionally, isolation of 11 showed that
decomposition pathways might compete with the desired
catalytic transformation under nonrigorous anhydrous con-
ditions.
Isotope-labeling studies, kinetic experiments, as well as
computational studies univocally demonstrated that our Ni-
catalyzed protocol for cleavage of C−O bonds with silanes as
reducing agents does not operate via putative Ni(II)−oxidative
addition complexes. The first-order dependence on silane
concentration as well as the observed silane depletion in the
induction period suggested that a distinctive mechanism came
into play. In-situ monitoring by NMR and EPR spectroscopy
identified the presence of Ni(I) species that are likely generated
by a comproportionation event. Computational and experimen-
tal studies suggested Ni(I)−SiR3 as the key propagating species
with migratory insertion into the naphthalene backbone being
the rate-determining step. Such assumption is in perfect analogy
with all the recorded kinetic data, isotope-labeling studies, and
the significantly greater reactivity of extended π systems as
compared with simpler anisole derivatives.
exploits a previously unrecognized opportunity in which Ni(I)−
SiEt3 (IX) are the key catalytic species, therefore constituting an
opportunity for research in other fields of expertise. We believe
that initially generated IX coordinates to 1 in a η2-fashion and
after a migratory insertion event, a benzyl nickel species X is
generated (Scheme 13). Elimination of MeOSiR3 with
concomitant migration of the Ni atom results in XI that undergo
a reversible σ-bond metathesis, delivering the final product 5
while generating back the Ni(I)−SiEt3 species IX.
Homogeneous vs Heterogeneous Catalysis. We next
focused our attention to study whether heterogeneous catalysts
could be responsible for the observed catalytic activity.58 To such
end, we turned our attention to mercury poisoning experi-
ments59 since suppression of catalytic activity in the presence of
Hg(0) is generally indicative of heterogeneous catalysis via
forming an amalgam with the metal(0) heterogeneous catalysts.
Interestingly, we observed that reaction of 1 with Et3SiH was shut
down in the presence of mercury under our optimized reaction
conditions (Figure 10). Care must be taken in interpreting these
results, as a control experiment revealed that mercury reacts with
Although additional investigations are warranted to expand the
scope and improve even further the catalytic performance, we
believe this study represents a significant step toward under-
standing the cleavage of inert C−O bonds in a catalytic fashion.
Additionally, our investigations study, for the first time, the Ni-
catalyzed C−O bond cleavage employing the catalytically active
monodentate PCy3 in an intermolecular fashion. Overall, our
data suggest that related intermolecular catalytic C−O bond-
cleavage reactions might indeed operate by mechanisms other
than the classical Ni(0)/Ni(II) couple. We anticipate that our
Figure 10. Plot of naphthalene (5) GC yield vs time from reaction of 1
with Et3SiH (a) under the exact optimized reaction conditions (red), (b)
upon filtration after 50 min reaction time, and (c) in the presence of Hg
from the beginning.
2007
dx.doi.org/10.1021/ja311940s | J. Am. Chem. Soc. 2013, 135, 1997−2009

Journal of the American Chemical Society
Article
(6) For a computational study dealing with C−OAc bond cleavage,
see: Li, Z.; Zhang, S.-L.; Fu, Y.; Guo, Q.-X.; Liu, L. J. Am. Chem. Soc.
2009, 131, 8815.
(7) Alvarez-Bercedo, P.; Martin, R. J. Am. Chem. Soc. 2010, 132, 17352.
(8) TMDSO = 1,1,3,3-tetramethyldisiloxane.
(9) For selected reviews: (a) Ng, S.-S.; Ho, C.-Y.; Schleicher, K. D.;
Jamison, T. F. Pure. Appl. Chem. 2008, 80, 929. (b) Montgomery, J. Acc.
Chem. Res. 2000, 33, 467.
(10) For a related C−SMe bond-cleavage procedure, see: Barbero, N.;
Martin, R. Org. Lett. 2012, 14, 796.
(11) Tobisu, M.; Yamakawa, K.; Shimasaki, T.; Chatani, N. Chem.
Commun. 2011, 47, 2946.
study will lead to new knowledge in catalyst design, thus opening
up new perspectives and stimulating new concepts within the
field of C−O bond-cleavage reactions.
ASSOCIATED CONTENT
* Supporting Information
■
S
Experimental procedures, spectral data, and crystallographic data
(CIF); complete ref 34; computational details, Cartesian
coordinates, and energies of all located structures. This material
is available free of charge via the Internet at http://pubs.acs.org.
(12) Sergeev, A. G.; Hartwig, J. F. Science 2011, 332, 439.
(13) Tobisu, M.; Chatani, N. ChemCatChem. 2011, 3, 1410.
(b) McGlaken, G. P.; Clarke, S. L. ChemCatChem. 2011, 3, 1260.
(14) For examples using stoichiometric metal species (a) Azzena, U.;
Dettori, G.; Idini, M. V.; Pisano, L.; Secchi, G. Tetrahedron 2003, 59,
7961. (b) Casado, F.; Pisano, L.; Farriol, M.; Gallardo, I.; Marquet, J.;
Melloni, G. J. Org. Chem. 2000, 65, 322. (c) Dabo, P.; Cyr, A.; Lessard, J.;
AUTHOR INFORMATION
Corresponding Author
rmartinromo@iciq.es; enrique.gomez@ehu.es
■
Funding
The authors declare no competing financial interests
Notes
́
Brossard, L.; Menard, H. Can. J. Chem. 1999, 77, 1225. (d) Maercker, A.
Angew. Chem., Int. Ed. 1987, 26, 972.
The authors declare no competing financial interest.
(15) Kelly, P.; Lin, S.; Edouard, G.; Day, M. W.; Agapie, T. J. Am. Chem.
ACKNOWLEDGMENTS
Soc. 2012, 134, 5480.
■
(16) For other stoichiometric hydrogenolysis of alkoxy metal
complexes with pincer-type ligands, see: (a) Fulmer, G. R.; Herndon,
A. N.; Kaminsky, W.; Kemp, R. A.; Goldberg, K. I. J. Am. Chem. Soc.
2011, 133, 17713. (b) Fulmer, G. R.; Muller, R. P.; Kemp, R. A.;
Goldberg, K. I. J. Am. Chem. Soc. 2009, 131, 1346.
We thank the ICIQ Foundation, the European Research Council
(ERC-277883) and MICINN (CTQ2012-34054) for financial
support. Johnson Matthey, Umicore, and Nippon Chemical
Industrial are acknowledged for a gift of metal and ligand sources.
R.M. thanks MICINN for a RyC fellowship. We sincerely thank
(17) For selected reviews: (a) van Leeuwen, P. W. N. M.; Kamer, P. C.
́
Nuria Clos & Guillem Aromi for EPR measurements at the
́
J.; Claver, C.; Pam
(b) Selander, N.; Szabo,
̀
ies, O.; Dieg
́
uez, M. Chem. Rev. 2011, 111, 2077.
K. J. Chem. Rev. 2011, 111, 2048. (c) Choi, J.;
Unitat de Mesures Magnet
̀
iques (UB) as well as Rita Marimon
Moncusi for TEM and ESEM analysis at Unitat de
ia I Tecniques Nanometriques (URV). We thank
́
́
MacArthur, A. H. R.; Brookhart, M.; Goldman, A. S. Chem. Rev. 2011,
111, 1761. (d) van der Boom, M. E.; Milstein, D. Chem. Rev. 2011, 111,
1759.
(18) For a selection of mechanistic studies using monodentate
phosphines, see: (a) Tan, Y.; Barrios-Landeros, F.; Hartwig, J. F. J. Am.
Chem. Soc. 2012, 134, 3683. (b) García-Melchor, M.; Fuentes, B.;
and Merce
Microscop
̀
̀
̀
̀
Eddy Martin and Eduardo Escudero for all X-ray crystallographic
data, Fernando Bozoglian for kinetic studies, the NMR unit at
ICIQ for our in situ monitoring studies, and N. Barbero for
preliminary experiments. We are indebted to Atsushi Urakawa
and Vladimir Grushin (ICIQ) for insightful discussions. We also
thank SGI/IZO-SGIker UPV/EHU for allocation of computa-
tional resources.
́
Lledos, A.; Casares, J. A.; Ujaque, G.; Espinet, P. J. Am. Chem. Soc. 2011,
133, 13519. (c) Liu, Z.; Yamamichi, H.; Madrahimov, S. T.; Hartwig, J.
F. J. Am. Chem. Soc. 2011, 133, 2772. (d) Alexanian, E. J.; Hartwig, J. F. J.
Am. Chem. Soc. 2008, 130, 15627. (e) Biscoe, M. R.; Barder, T. E.;
Buchwald, S. L. Angew. Chem., Int. Ed. 2007, 46, 7232. (f) Macgregor, S.
A.; Roe, D. C.; Marshall, W. J.; Bloch, K. M.; Bakhmutov, V. I.; Grushin,
V. V. J. Am. Chem. Soc. 2005, 127, 15304.
REFERENCES
■
(1) For selected reviews, see: (a) Rosen, B. M.; Quasdorf, K. W.;
Wilson, D. A.; Zhang, N.; Resmerita, A.-M.; Garg, N. K.; Percec, V.
Chem. Rev. 2011, 111, 1346. (b) Li, B.-J.; Yu, D.-G.; Sun, C.-L.; Shi, Z.-J.
Chem.Eur. J. 2011, 17, 1728. (c) Yu, D.-G.; Li, B.-J.; Shi, Z.-J. Acc.
Chem. Res. 2010, 43, 1486.
(19) See Supporting Information for details.
(20) Zeller, A.; Herdtweck, E.; Strassner, T. Eur. J. Inorg. Chem. 2003, 9,
1802.
(21) Retbøll, M.; Edwards, A. J.; Rae, A. D.; Willis, A. C.; Bennett, M.
A.; Wenger, E. J. Am. Chem. Soc. 2002, 124, 8348.
(2) In Metal-catalyzed Cross-Coupling Reactions; Diederich, F.; Stang, P.
J., Eds.; Wiley-VCH: New York, 1998.
(22) For some selected references dealing with salt metathesis using
bidentate ligands, see: (a) Breitenfeld, J.; Scopelliti, R.; Hu, X.
Organometallics 2012, 31, 2128. (b) Mrutu, A.; Dickie, D. A.; Goldberg,
K. I.; Kemp, R. A. Inorg. Chem. 2011, 50, 2729. (c) Vechorkin, O.; Csok,
Z.; Scopelliti, R.; Hu, X. Chem.Eur. J. 2009, 15, 3889.
(23) The reaction was found to be slower when stoichiometric
amounts of NaOMe were used. In any case, the product of oxidative
addition 4 was observed.
(3) For Ni-catalyzed cross-coupling of phenoxy derivates, see: (a) Yu,
D.-G.; Li, B.-J.; Zheng, S.-F.; Guan, B.-T.; Wang, B.-Q.; Shi, Z.-J. Angew.
Chem., Int. Ed. 2010, 49, 4566. (b) Yu, D.-G.; Shi, Z.-J. Angew. Chem., Int.
Ed. 2011, 50, 7097.
(4) For Ru-catalyzed C(sp2)−OMe arylation, see: (a) Ueno, S.;
Mizushima, E.; Chatani, N.; Kakiuchi, F. J. Am. Chem. Soc. 2006, 128,
16516. (b) Kakiuchi, F.; Usui, M.; Ueno, S.; Chatani, N.; Murai, S. J. Am.
Chem. Soc. 2004, 126, 2706.
(24) For related η2-bound Ni(0) carbonyl compounds prepared by
other means, see: (a) Ohashi, M.; Saijo, H.; Arai, T.; Ogoshi, S.
Organometallics 2010, 29, 6534. (b) Tamaki, T.; Nagata, M.; Ohashi, M.;
Ogoshi, S. Chem.Eur. J. 2009, 15, 10083. (c) Ogoshi, S.; Arai, T.;
Ohashi, M.; Kurosawa, H. Chem. Commun. 2008, 1347.
(5) For Ni-catalyzed C(sp2)−OMe arylation methodologies, see:
(a) Wang, C.; Ozaki, T.; Takita, R.; Uchiyama, M. Chem.Eur. J. 2012,
18, 3482. (b) Wang, Z.-X.; Xie, L.-G. Chem.Eur. J. 2011, 17, 4972.
(c) Shimasaki, T.; Konno, Y.; Tobisu, M.; Chatani, N. Org. Lett. 2009,
11, 4890. (d) Tobisu, M.; Shimasaki, T.; Chatani, N. Angew. Chem., Int.
Ed. 2008, 47, 4866. (e) Guan, B.-T.; Xiang, S.-K.; Wu, T.; Sun, Z.-P.;
Wang, B.-Q.; Zhao, K.-Q.; Shi, Z.-J. Chem. Commun. 2008, 1437.
(f) Tobisu, M.; Shimasakai, T.; Chatani, N. Angew. Chem., Int. Ed. 2008,
47, 4866. (g) Dankwart, J. W. Angew. Chem., Int. Ed. 2004, 43, 2428.
(h) Wenkert, E.; Michelotti, E. L.; Swindell, C. S. J. Am. Chem. Soc. 1982,
104, 1670.
(25) A similar pathway has been observed for pincer-type ligands, see
ref 15.
(26) For selected references, see: (a) Matas, I.; Campora, J.; Palma, P.;
Alvarez, E. Organometallics 2009, 28, 6515. (b) van der Boom, M. E.;
Liou, S.-Y.; Ben-David, Y.; Shimon, L. J. W.; Milstein, D. J. Am. Chem.
Soc. 1998, 120, 6531. (c) Gambarotta, S.; Floriani, C.; Chiesi-Villa, A.;
Guastini, C. Organometallics 1986, 5, 2425. (d) Bianchini, C.; Meli, A.
2008
dx.doi.org/10.1021/ja311940s | J. Am. Chem. Soc. 2013, 135, 1997−2009

Journal of the American Chemical Society
Article
Organometallics 1985, 4, 1537. For a review dealing with alcoxy metal
species, see: Bryndza, H. E.; Tam, W. Chem. Rev. 1988, 88, 1163.
(27) Hartwig, J. F. Inorg. Chem. 2007, 46, 1936.
(50) For isolation of three-coordinate Ni(I)−silyl complexes, see: Iluc,
V. M.; Hillhouse, G. L. J. Am. Chem. Soc. 2010, 132, 11890.
(51) (a) Minidola, D. J.; Hillhouse, G. L. J. Am. Chem. Soc. 2001, 123,
4623. (b) Vicic, D. A.; Jones, W. D. J. Am. Chem. Soc. 1997, 119, 10855.
(52) Jonas, K.; Wilke, G. Angew. Chem., Int. Ed. 1970, 9, 312. We found
that 12 was not catalytically competent in our reaction protocol.
(53) For selected applications of Ni(I)−H dimers utilizing bidentate
backbones, see: (a) Smith, E. E.; Du, G.; Fanwick, P. E.; Abu-Omar, M.
M. Organometallics 2010, 29, 6527. (b) Torres-Nieto, J.; Brennessel, W.
W.; Jones, W. D.; Garcia, J. J. J. Am. Chem. Soc. 2009, 131, 4120.
(c) García, J. J.; Brunkan, N. M.; Jones, W. D. J. Am. Chem. Soc. 2002,
124, 9547. (d) Edelbach, B. L.; Lachicotte, R. J.; Jones, W. D.
Organometallics 1999, 18, 4040.
(28) (a) Jolly, P. W.; Jonas, K.; Kruger, C.; Tsay, Y.-H. J. Organomet.
̈
Chem. 1971, 33, 109. (b) Aresta, M.; Nobile, C. F.; Sacco, A. Inorg. Chim.
Acta 1974, 12, 167.
(29) Analytically pure 11 was only obtained in trace amounts (<5%
yield) after recrystallization.
(30) Carmona, E.; Marin, J. M.; Palma, P.; Paneque, M.; Poveda, M. L.
Inorg. Chem. 1989, 28, 1895.
(31) For an example, see: Furstner, A.; Majima, K.; Martin, R.; Krause,
̈
H.; Kattnig, E.; Goddard, R.; Lehmann, W. J. Am. Chem. Soc. 2008, 130,
1992.
(54) The structure of 12 was discussed in ref 52, but no X-ray
crystallography was shown.
(32) Taking into consideration that TMDSO and Et3SiH were equally
effective for promoting the Ni-catalyzed reductive cleavage of C−OMe
bonds, we decided to employ for our isotope-labeling studies
commercially available Et3SiH and Et3SiD
(33) (a) Simmons, E. M.; Hartwig, J. F. Angew. Chem., Int. Ed. 2012, 51,
́
3066. (b) Gomez-Gallego, M.; Sierra, M. A. Chem. Rev. 2011, 111, 4857.
(34) Reaction of 1 with Ni(COD)2 (5 mol %), PCy3 (10 mol %), and
Et3SiH:Et3SiD (1:1, 2 equiv) gave rise to 5 and 5-D (1:1 ratio)
(35) For a full reference of Gaussian-09, see Supporting Information.
(36) (a) Hay, P. J.; Wadt, W. R. J. Chem. Phys. 1985, 82, 270. (b) Wadt,
W. R.; Hay, P. J. J. Chem. Phys. 1985, 82, 284. (c) Hay, P. J.; Wadt, W. R.
J. Chem. Phys. 1985, 82, 299.
(37) (a) Lee, C.; Yang, W.; Parr, R. G. Phys. Rev. B 1988, 37, 785.
(b) Becke, A. D. J. Chem. Phys. 1993, 98, 5648. (c) Kohn, W.; Becke, A.
D.; Parr, R. G. J. Phys. Chem. 1996, 100, 12974.
(38) Zhao, Y.; Truhlar, D. G. Theor. Chem. Acc. 2008, 120, 215.
(39) (a) Fuentealba, P.; Preuss, H.; Stoll, H.; Szentpaly, L. V. Chem.
Phys. Lett. 1982, 89, 418. (b) Szentpaly, L. V.; Fuentealba, P.; Preuss, H.;
Stoll, H. Chem. Phys. Lett. 1982, 93, 555.
(55) Our results are in analogy with other Ni(I) complexes, see, for
example: (a) Uyeda, C.; Peters, J. C. Chem. Sci. 2013, 4, 157. (b) Klein,
A.; Vicic, D. A.; Biewer, C.; Kieltsch, I.; Stirnat, K.; Hamacher, C.
Organometallics 2012, 31, 5334. (c) Kraikivskii, P. B.; Saraev, V. V.;
Meusinger, R.; Bocharova, V. V.; Ushakov, I. A.; Petrovskii, S. K. J.
Organomet. Chem. 2012, 715, 43. (d) Bai, G.; Wei, P.; Stephan, S. W.
Organometallics 2005, 24, 5901. (e) Holland, P. L.; Cundari, T. R.; Perez,
L. L.; Eckert, N. A.; Lachicotte, R. J. J. Am. Chem. Soc. 2002, 124, 14416.
(56) Unfortunately, the Ni(I)−H dimer with PCy3 is not computa-
tionally tractable, as it contains four PCy3 and more than 200 atoms
(57) For a Ni(I)−H insertion into arene motifs, see: Lin, S.; Day, M.
W.; Agapie, T. J. Am. Chem. Soc. 2011, 133, 3828.
(58) For very recent heterogeneous Ni-catalyzed reductive cleavage of
C−OAr bonds, see: (a) Sergeev, A. G.; Hartwig, J. F. J. Am. Chem. Soc.
2012, 134, 20226. (b) Parsell, T. H.; Owen, B. C.; Klein, I.; Jarrell, T. M.;
Marcuum, C. L.; Hupert, L. J.; Amundson, L. M.; Kenttamaa, H. I.;
̈
Ribeiro, F.; Miller, J. T.; Abu-Omar, M. M. Chem. Sci. 2013, 4, 806.
(c) He, J.; Zhao, C.; Lercher, J. A. J. Am. Chem. Soc. 2012, 134, 20768.
(59) (a) Crabtree, R. H. Chem. Rev. 2012, 112, 1536. (b) Widegren, J.
A.; Finke, R. G. J. Mol. Catal. A 2003, 198, 317. (c) Yu, K. Q.; Sommer,
W.; Weck, M.; Jones, C. W. J. Catal. 2004, 226, 101. (d) Foley, P.;
Dicosimo, R.; Whitesides, G. M. J. Am. Chem. Soc. 1980, 102, 6713.
(60) For similar discrepancies regarding the Hg drop test reacting with
homogeneous catalysts, see: Stein, J.; Lewis, L. N.; Gao, Y.; Scott, R. A. J.
Am. Chem. Soc. 1999, 121, 3693.
(40) (a) Cances
3032. (b) Cossi, M.; Barone, V.; Mennuci, B.; Tomasi, J. Chem. Phys.
Lett. 1998, 286, 253. (c) Tomasi, J.; Mennucci, B.; Cances, E. J. Mol.
Struct. (THEOCHEM) 1999, 464, 21.
̀
, E.; Mennucci, B.; Tomasi, J. J. Chem. Phys. 1997, 107,
̀
(41) No significant differences in free activation energies were
observed when computing Et3SiH vs Me3SiH.
(42) B3LYP has been shown to underestimate the dissociation
energies of PCy3 in other organometallic complexes, see: Zhao, Y.;
Truhlar, D. G. J. Chem. Theory Comput. 2009, 5, 324.
(43) For this reason, only M06 values will be used for the mechanistic
discusion. B3LYP values are collected in the Supporting Information
(44) These results are in sharp contrast with the proposed mechanism
for cleavage of C−OAc bonds (ref 6) in which transmetalation and not
oxidative addition is rate determining.
(61) For a selected number of references that leave some doubt about
the Hg drop test for identifying whether heterogeneous systems are
present or not in other homogeneous reactions, see: (a) Esfandiari, N.
M.; Blum, S. A. J. Am. Chem. Soc. 2011, 133, 18145. (b) Widegren, J. A.;
Finke, R. G. J. Mol. Catal. A. 2003, 198, 317. (c) van Asselt, R.; Elsevier,
C. J. J. Mol. Catal. 1991, 65, L13. (d) Jones, R. A.; Real, F. M.; Wilkinson,
G.; Galas, A. M. R.; Hurthouse, M. B. J. Chem. Soc., Dalton Trans. 1981,
126.
(62) At present we cannot rule out the presence of Ni(I) clusters that
are in equilibrium with their corresponding Ni(I) homogeneous
monomers. Recently, related Ni(I) clusters have been reported in the
literature, see: Beck, R.; Shoshani, M.; Johnson, S. A. Angew. Chem., Int.
Ed. 2012, 51, 11753.
(45) Similar experiments were performed for other reductive cleavage
protocols, see: Baxter, R. D.; Montgomery, J. J. Am. Chem. Soc. 2011,
133, 5728.
(46) For selected references, see: (a) Tanabe, M.; Yumoto, R.;
Osakada, K.; Sanji, T.; Tanaka, M. Organometallics 2012, 31, 6787.
(b) Beck, R.; Johnson, S. A. Organometallics 2012, 31, 3599. (c) Tanabe,
M.; Yumoto, R.; Osakada, K. Chem. Commun. 2012, 48, 2125.
(d) Takaya, J.; Iwasawa, N. Dalton Trans. 2011, 40, 8814. (e) Iluc, V.
M.; Hillhouse, G. L. J. Am. Chem. Soc. 2010, 132, 11890. (f) Chen, W.;
Shimada, S.; Tanaka, M.; Kobayashi, Y.; Saigo, K. J. Am. Chem. Soc. 2004,
126, 8072.
(47) For a related σ-bound Ni(0) borane complex, see: Crestani, M.
́
G.; Munoz-Hernandez, M.; Arevalo, A.; Acosta-Ramirez, A.; Garcia, J. J.
̃
J .Am. Chem. Soc. 2005, 127, 18066.
(48) Iluc, V. M.; Hillhouse, G. L. Tetrahedron 2006, 63, 7577.
(49) For selected comproportionation events in route to Ni(I) species,
see: (a) Velian, A.; Lin, S.; Miller, A. J. M.; Day, M. W.; Agapie, T. J. Am.
Chem. Soc. 2010, 132, 6296. (b) Phapale, V. B.; Bunuel, E.; García-
̃
́
Iglesias, M.; Cardenas, D. J. Angew. Chem., Int. Ed. 2007, 46, 8790.
(c) Jones, G. D.; Martin, J. L.; McFarland, C.; Allen, O. R.; Hall, R. E.;
Haley, A. D.; Brandon, R. J.; Konovalova, T.; Desrochers, P. J.; Pulay, P.;
Vicic, D. A. J. Am. Chem. Soc. 2006, 128, 13175.
2009
dx.doi.org/10.1021/ja311940s | J. Am. Chem. Soc. 2013, 135, 1997−2009