Iron-catalyzed dehydrocoupling/dehydrogenation of amine-boranes

Article abstract of DOI:10.1021/ja410129j

The readily available iron carbonyl complexes, [CpFe(CO)₂] ₂ (1) and CpFe(CO)₂I (2) (Cp = η-C₅H ₅), were found to be efficient precatalysts for the dehydrocoupling/dehydrogenation of the amine-borane Me₂NH· BH₃ (3) to afford the cyclodiborazane [Me₂N-BH ₂]₂ (4), upon UV photoirradiation at ambient temperature. In situ analysis of the reaction mixtures by ¹¹B NMR spectroscopy indicated that different two-step mechanisms operate in each case. Thus, precatalyst 1 dehydrocoupled 3 via the aminoborane Me₂N=BH ₂ (5) which then cyclodimerized to give 4 via an off-metal process. In contrast, the reaction with precatalyst 2 proceeded via Me ₂NH-BH₂-NMe₂-BH₃ (6) as the key intermediate, affording 4 as the final product after a second metal-mediated step. The related complex Cp₂Fe₂(CO)₃(MeCN) (7), formed by photoirradiation of 1 in MeCN, was found to be a substantially more active dehydrocoupling catalyst and not to require photoactivation, but otherwise operated via a two-step mechanism analogous to that for 1. Significantly, detailed mechanistic studies indicated that the active catalyst generated from precatalyst 7 was heterogeneous in nature and consisted of small iron nanoparticles (≤10 nm). Although more difficult to study, a similar process is highly likely to operate for precatalyst 1 under photoirradiation conditions. In contrast to the cases of 7 and 1, analogous experimental studies for the case of photoactivated Fe precatalyst 2 suggested that the active catalyst formed in this case was homogeneous. Experimental and computational DFT studies were used to explore the catalytic cycle which appears to involve amine-borane ligated [CpFe(CO)]⁺ as a key intermediate.

Full text of DOI:10.1021/ja410129j

Article
pubs.acs.org/JACS
Iron-Catalyzed Dehydrocoupling/Dehydrogenation of Amine−
Boranes
James R. Vance, Andre Schafer, Alasdair P. M. Robertson, Kajin Lee, Joshua Turner, George R. Whittell,
́
̈
and Ian Manners*
School of Chemistry, University of Bristol, Cantock’s Close, Bristol BS8 1TS, U.K.
S
* Supporting Information
ABSTRACT: The readily available iron carbonyl complexes,
[CpFe(CO)2]2 (1) and CpFe(CO)2I (2) (Cp = η-C5H5), were
found to be efficient precatalysts for the dehydrocoupling/
dehydrogenation of the amine−borane Me2NH·BH3 (3) to
afford the cyclodiborazane [Me2N−BH2]2 (4), upon UV
photoirradiation at ambient temperature. In situ analysis of the
reaction mixtures by 11B NMR spectroscopy indicated that
different two-step mechanisms operate in each case. Thus,
precatalyst 1 dehydrocoupled 3 via the aminoborane Me2N
BH2 (5) which then cyclodimerized to give 4 via an off-metal
process. In contrast, the reaction with precatalyst 2 proceeded
via Me2NH−BH2−NMe2−BH3 (6) as the key intermediate,
affording 4 as the final product after a second metal-mediated
step. The related complex Cp2Fe2(CO)3(MeCN) (7), formed by photoirradiation of 1 in MeCN, was found to be a substantially
more active dehydrocoupling catalyst and not to require photoactivation, but otherwise operated via a two-step mechanism
analogous to that for 1. Significantly, detailed mechanistic studies indicated that the active catalyst generated from precatalyst 7
was heterogeneous in nature and consisted of small iron nanoparticles (≤10 nm). Although more difficult to study, a similar
process is highly likely to operate for precatalyst 1 under photoirradiation conditions. In contrast to the cases of 7 and 1,
analogous experimental studies for the case of photoactivated Fe precatalyst 2 suggested that the active catalyst formed in this
case was homogeneous. Experimental and computational DFT studies were used to explore the catalytic cycle which appears to
involve amine−borane ligated [CpFe(CO)]+ as a key intermediate.
hydrogen.21−24 In addition, primary amine−boranes, RNH2·
BH3 (R = Me, nBu) (Scheme 1(i)) and 9, have been shown to
function as precursors to soluble polyaminoboranes33−35,45 and
also to “white graphene”, which consists of a monolayer film of
INTRODUCTION
■
Transition-metal complexes catalyze a broad spectrum of
important organic transformations, and detailed mechanistic
studies have revealed the importance of both homogeneous and
heterogeneous processes.1−6 A notable feature of recent work
in this field has been a drive to replace the more prevalent
“precious” metal catalysts with cheaper, more earth-abundant
metals such as Ti, Ni, Co, and Fe.7−10 By comparison with
organic reactions, the corresponding use of metal-catalyzed
transformations for main group substrates is in its relative
infancy despite many significant advances.11−14
Scheme 1. Metal-Catalyzed Dehydrocoupling/
Dehydrogenation Pathways for (i) Primary and (ii)
Secondary Amine−Boranes
Amine−boranes, RxNH3−x·BH3 (R = H, alkyl, aryl), are
readily (and in some cases commercially) available examples of
main group coordination compounds and are widely used as
either reducing15−18 or hydroboration reagents.19,20 In recent
years, however, interest in amine−boranes and their deriva-
tives21−69 has attracted rapidly expanding attention due to
potential uses in chemical hydrogen storage,21−25 transfer
hydrogenations and reductions of organic substrates,26−32 and
as precursors to polymeric and solid-state materials.33−37,41 For
example, NH3·BH3 (9), which is an air- and moisture-stable
solid of ∼20% gravimetric hydrogen content, has been the
subject of intense interest worldwide as a portable source of
Received: October 2, 2013
Published: February 14, 2014
© 2014 American Chemical Society
3048
dx.doi.org/10.1021/ja410129j | J. Am. Chem. Soc. 2014, 136, 3048−3064

Journal of the American Chemical Society
Article
Scheme 2. Dehydrogenation of Me2NH·BH3 (3) via (i) Me2NBH2 (5) (Off-Metal) or (ii) Me2NH−BH2−NMe2−BH3 (6)
(On-Metal) with Regard to the Final Step to form [Me2N−BH2]2 (4)
hexagonal boron nitride.36 Of crucial relevance to the
aforementioned applications are efficient protocols for the
dehydrogenation of amine−boranes,38 and substantial progress
has been made in the field of metal-catalyzed reactions
(Scheme 1).39−66
high yield, allowing them to be treated as simple model
substrates for other amine−boranes. Even for amine−borane 3,
however, at least two metal-mediated mechanisms of H2 release
have been identified. Dehydrogenation is believed to occur
either via the monomeric aminoborane, Me2NBH2 (5), or
the linear diborazane, Me2NH−BH2−NMe2−BH3 (6), invok-
ing initial (i) intra- and (ii) intermolecular H2 release steps,
respectively (Scheme 2). The mechanism of dehydropolyme-
rization of the primary amine−borane MeNH2·BH3 (10) has
also been studied, and although this will not be discussed in
detail here, it is believed to involve a metal-centered
dehydrogenation to form MeNHBH2 (11) followed by a
second “on-metal” coordination polymerization step.35,74
In this paper, we report full details of our studies on
photoactivated Fe complex 1 and other readily accessible and
related iron carbonyl species as amine−borane dehydrogen-
ation catalysts, focusing on a mechanistic investigation of the
catalytic dehydrocoupling of 3. In addition, we also report the
development of an Fe precatalyst (7) that does not require
photoactivation and that exhibits even higher activity. Our
studies provide clear evidence that subtle structural differences
in the precatalyst can lead to either heterogeneous or to
homogeneous dehydrocoupling mechanisms that involve active
catalysts that operate under mild conditions.
Amine−borane dehydrogenation catalysts that are based on
metals from much of the periodic table have now been
reported, but notably, if not unsurprisingly, almost all are 4d or
5d elements and many of these are “precious met-
als”.41,42,46,50,52,62,67,68 The development of catalysts based
upon less expensive metals is therefore of major interest, and
several examples based on earth abundant elements have now
been published.39,40,43,44,48,51,53−57,69 Systems based on Fe are
of particular interest, and the first such example, FeH-
(PMe2CH2)(PMe3)3, was described briefly by Baker in 2007.
In this case, high conversion of 9 to various dehydrocoupled
oligomers was observed using 10 mol % catalyst after 4 days in
THF at 20 °C.47 In 2011, we reported our preliminary results
on the use of photoactivated [CpFe(CO)2]2 (1) (5 mol %, 10
mol % Fe) in the catalytic dehydrogenation of 9 at 20 °C over 4
h in THF.53 We also found that 1 is a rare example of a
dehydrocoupling catalyst with significant activity toward a range
of amine−boranes. For example, treatment of the primary
amine−borane MeNH2·BH3 (10) and the secondary amine−
borane 3 with catalytic quantities of 1 also led to dehydrogen-
ation to form various oligomeric and polymeric products.
Recently, Baker et al. have also reported details on a series of
mixed N−P donor ligated Fe(II) precatalysts with significant
activity for the dehydrogenation of 9 with high conversion
using 5 mol % catalyst in THF at 20−60 °C.48,70 Furthermore,
Morris et al. have recently demonstrated the use of Fe
complexes such as trans-[Fe(NCMe)CO(PPh2C6H4CH
NCHPh)2][BF4]2 as precatalysts for the dehydrocoupling of
both 3 and 9 in THF using 2.5 mol % catalyst in 1 h at 22 °C.49
This report focused on the identification of the active catalyst
which was proposed to be heterogeneous based on electron
microscopy imaging of reaction mixtures and the similar
behavior detected to that for previously studied transfer
hydrogenation systems. Nevertheless, the acquisition of further
evidence to support dehydrogenation by Fe nanoparticles was
hindered by the rapid initial rates which obstructed any in
operando studies.49
RESULTS
■
1. Dehydrocoupling of Me2NH·BH3 (3) with Fe
Complexes. With a view to extending the range of known
Fe precatalysts for amine−borane dehydrogenation, we
investigated the activity of various photoactive carbonyl
complexes, beginning our studies with [CpFe(CO)2]2 (1) for
which we have briefly reported preliminary results.53 Addition
of 1 (5 mol %, 10 mol % Fe) to the model amine−borane 3 in
THF under photoirradiation led to rapid catalytic turnover,
with quantitative conversion to the cyclic diborazane [Me2N−
BH2]2 (4) over 4 h at 20 °C. It is noteworthy that the relative
activity of 1 was high when compared to the other potential
precatalysts investigated. Furthermore, only CpFe(CO)2X [X =
I (2) or Cl (12)] also produced a respectable rate of
dehydrogenation under photoirradiation, with both precatalysts
reaching ca. 60% conversion of 3 in toluene after 4 h,
conditions under which the analogous use of 1 led to reaction
completion (20 °C, 5 mol %) (Table 1).75
As part of the drive to improve fundamental understanding
and catalyst efficiency, various experimental and computational
studies have attempted to elucidate the mechanism(s) of metal-
catalyzed amine-borane dehydrogenation.39,40,44,50,52,55,71−73
Most of these studies have focused on the secondary amine−
The respectable rates of dehydrogenation of 3 in the cases of
photoirradiated precatalysts 1 and 2 encouraged us to initially
focus on detailed studies solely of these two systems. Although
a respectable rate of dehydrogenation was also observed using
i
boranes Me2NH·BH3 (3) (Scheme 1 (ii)) and Pr2NH·BH3 as
substrates due to the formation of a single final product in very
3049
dx.doi.org/10.1021/ja410129j | J. Am. Chem. Soc. 2014, 136, 3048−3064

Journal of the American Chemical Society
Article
that the presence of the amine−borane was essential in the
formation of the active catalyst.
Table 1. Dehydrocoupling Activity of Various Fe Complexes
toward Me2NH·BH3 (3) at 20 °C
a
Monitoring the reaction of 3 with precatalyst 1 (5 mol %,
THF) by 11B NMR spectroscopy also provided mechanistic
insight into the dehydrocoupling process. During the course of
the reaction, in addition to the growing peak at 4.1 ppm
associated with the final dimeric product 4, a new resonance
was detected at δ 36.5. This was assigned to the monomeric
aminoborane 5, based on the chemical shift and the triplet
multiplicity (1JBH = 127 Hz) in the 11B NMR spectrum (Figure
1).72
precatalyst
(5 mol %)
time
(h)
solvent
yield of [Me2N−BH2]2 (4)
b,c
(%)
[CpFe(CO)2]2 (1)
4
THF
THF
0
e
[CpFe(CO)2]2 (1) 4 (hν)
100 (80 )
d
(Tol )
THF
THF
f
[CpFe(CO)2]2 (1) 4 (hν)
55
[Cp*Fe(CO)2]2
(1*)
[Cp*Fe(CO)2]2
(1*)
4 (hν)
0
24 (hν) THF
30
Fe2(CO)9 (8)
Fe3(CO)12
4 (hν)
4 (hν)
24
THF
0
THF
0
CpFe(CO)2I (2)
CpFe(CO)2I (2)
Tol
0
4 (hν)
THF (Tol)
Tol
30 (55)
CpFe(CO)2Cl (12) 24
0
CpFe(CO)2Cl (12) 4 (hν)
THF (Tol)
THF
0 (65)
FeCl2
4
4
0
0
FeCl3
THF
a
b
Concentration of 3: 0.11 M. Yield based on analysis by 11B NMR
c
spectroscopy. Small amounts of HB(NMe2)2 (<5%) were also
d
detected by 11B NMR spectroscopy. Reaction performed in toluene.
e
Although analysis by 11B NMR spectroscopy showed only 80%
conversion to 4, there was no starting material remaining in the
f
reaction mixture as it was also converted to HB(NMe2)2.76 The
reaction was repeated using 2.5 mol % of precatalyst 1 (therefore 5
mol % of Fe) to allow a direct comparison with the mononuclear Fe
precatalysts.77
Figure 1. 11B{1H} NMR spectrum of the reaction mixture derived
from photoirradiated [CpFe(CO)2]2 (1) (5 mol %) and Me2NH·BH3
(3) (30 min, THF, 20 °C). Inset illustrates triplet multiplicity of the
peak at δ 36.5 in the corresponding 11B NMR spectrum.
precatalyst 12, this species is structurally very similar to
precatalyst 2, which, unlike 12, is commercially available.
2. Observation of Intermediates in the Catalytic
Dehydrocoupling of Me2NH·BH3 (3) with Photoacti-
vated [CpFe(CO)2]2 (1). Treatment of amine−borane 3 with
Fe complex 1 (5 mol %) in THF at 20 °C in the absence of
photoirradiation yielded no detectable dehydrocoupling over 4
h, and minimal dehydrocoupling over 24 h, based on analysis
by 11B NMR spectroscopy. However, rapid virtually quantita-
tive dehydrocoupling to cyclic diborazane 4 (δB 4.1 ppm, t, JBH
= 113 Hz)42 was observed by 11B NMR spectroscopy over 4 h
under photoirradiation conditions, where the escape of the
gaseous byproducts (presumably H2 and CO, Scheme 3) was
Significantly, the linear diborazane 6 [δB −13.7 (q, 1JBH = 94
Hz, BH3), 1.4 (t, 1JBH = 108 Hz, BH2)],78 noted earlier as a key
intermediate in many other proposed catalytic cycles (Scheme
2), was not detected in significant amounts by 11B NMR
spectroscopy during the reaction of 3 with photoirradiated 1. In
addition, separate experiments showed that 1 was virtually
inactive (<5% conversion to 4) toward dehydrocoupling of
independently synthesized 6, even upon photoirradiation for 21
h. The absence of linear diborazane 6 as an intermediate was
further demonstrated upon treatment of an equimolar mixture
of Me2NH·BD3 (3-d3) and 6 with 1 (5 mol %) under
photoirradiation (4 h, THF, 20 °C). Analysis of the resulting
mixture by 11B NMR spectroscopy indicated that while 3-d3
was converted to [Me2N-BD2]2 (4-d4), 6 was left unreacted
(Scheme 4, Figure 2), further underlining the lack of reactivity
of this substrate toward photoactivated 1.
Scheme 3. Dehydrogenation of Me2NH·BH3 (3) with
Photoactivated [CpFe(CO)2]2 (1) (5 mol %) To Form
[Me2N−BH2]2 (4)
3. Observation of Intermediates in the Catalytic
Dehydrocoupling of Me2NH·BH3 (3) with Photoacti-
vated CpFe(CO)2I (2). Following the investigation of
precatalyst 1, the reactivity of Fe precatalyst 2 toward 3 was
studied in detail. Treatment of 3 with precatalyst 2 (5 mol %,
toluene, 20 °C) under photoirradiation led to virtually
quantitative dehydrogenation over 9 h to form the cyclic
diborazane 4, as evidenced by 11B NMR spectroscopy.
Remarkably, however, monitoring the reaction indicated very
different mechanistic behavior compared to that exhibited by
the reaction of precatalyst 1 with the same substrate. In the case
of 2, and in contrast to that with 1, the linear diborazane 6 was
detected in significant quantity by 11B NMR spectroscopy [δB
allowed. A reduced rate (80% conversion to 4 over 6 h) was
observed when the reaction was performed in a closed system,
with the rate likely being limited by competitive coordination of
released H2 and/or CO to the photogenerated catalyst. All
subsequent reactions were therefore performed in open
systems, under nitrogen on a Schlenk line. Interestingly,
photoirradiation of precatalyst 1 for 1 h in THF before
subsequent addition of amine−borane 3 led to no conversion
after 3 h when analyzed by 11B NMR spectroscopy, indicating
3050
dx.doi.org/10.1021/ja410129j | J. Am. Chem. Soc. 2014, 136, 3048−3064

Journal of the American Chemical Society
Article
Scheme 4. Attempted Dehydrogenation of a Mixture of Me2N−HBH2−NMe2BH3 (6) and Me2NH·BD3 (3-d3) with
Photoirradiated [CpFe(CO)2]2 (1) (5 mol %)
Figure 2. 11B NMR spectrum of the reaction of photoactivated
[CpFe(CO)2]2 (1) (5 mol %) with Me2NH-BH2−NMe2-BH3 (6) and
Figure 3. 11B{1H} NMR spectrum of the reaction of photoactivated
CpFe(CO)2I (2) (5 mol %) and Me2NH·BH3 (3) (5 h, toluene, 20
°C) affording [Me2N-BH2]2 (4) as the final product, via Me2NH−
BH2−NMe2-BH3 (6). The signal labeled with * is assigned to
HB(NMe2)2.
Me2NH·BD3 (3-d3) (4 h, THF, 20 °C) affording [Me2N-BD2]2 (4-d4)
as the final product where 6 is left unreacted.
−13.7 (q, JBH = 94 Hz, BH3), 1.4 (t, JBH = 108 Hz, BH2)]78
(Scheme 5 and Figure 3), and the monomeric aminoborane 5
(δB 36.5 ppm) was virtually absent. These results suggested that
an alternative mechanism was operational when photoactivated
2 was employed as the precatalyst.
1
1
photoirradiating both species to induce catalytic activity
somewhat hampers a detailed experimental investigation of
the contrasting mechanisms. Nevertheless, it can be assumed
that in each case photoirradiation leads to the cleavage of one
or more bonds at the respective Fe center leading to
coordinative unsaturation, thus facilitating the formation of a
catalyst that permits amine−borane coordination and entry into
the catalytic manifold.
(a). [CpFe(CO)2]2 (1) as Precatalyst. Previous work has
shown that, in the case of precatalyst 1,79 photoirradiation leads
to two distinct bond activation processes. These involve
carbonyl dissociation to form the coordinatively unsaturated
dinuclear species [Cp2Fe2(CO)3] and homolytic cleavage of the
central Fe−Fe bond to form two [CpFe(CO)2] radicals.80−82
The formation of both products has been reported to occur
under a variety of conditions. Moreover, the former species has
been shown to form adducts with simple 2-electron donors,
e.g., PPh3, and the latter found to be capable of atom
abstraction reactions. Indeed, both of the photogenerated iron
species could react directly with amine−boranes, which have
been shown to bind to coordinately unsaturated metal centers
via the hydride substituents at boron59,67,83−86 and also to
undergo hydride abstraction to form boron-centered cations87
and to participate in radical reactions.87−89 Although it was not
To further probe the mechanism of the dehydrogenation in
this case, the catalytic reaction of linear diborazane 6 with 2 was
also explored. Treatment of a toluene solution of 6 with 2 (5
mol %) under photoirradiation led to 47% conversion to 4 over
4 h at 20 °C based on analysis by 11B NMR spectroscopy. The
rate of dehydrocyclization of isolated linear diborazane 6 with
photoactivated 2 is consistent with the former species
functioning as a key intermediate in the catalytic dehydrogen-
ation of 3 to afford 4 which takes 9 h to reach completion. This
observation was in marked contrast to the behavior of the
related Fe complex 1, which was shown to dehydrocouple 3 via
the monomeric aminoborane 5 as the key intermediate and to
be virtually inactive toward 6. The reactions with 3 and 6 were
also repeated with CpFe(CO)2Cl (12), which demonstrated
very similar reactivity in toluene to that observed in the case of
2.
4. Nature of the Active Catalysts in the Catalytic
Dehydrogenation of 3 Using [CpFe(CO)2]2 (1) and
CpFe(CO)2I (2) as Precatalysts. The mechanisms of
dehydrogenation of amine−borane 3 by complexes 1 and 2
appear to be fundamentally different as distinct intermediates
are detected in each case. However, the necessity of
Scheme 5. Dehydrogenation of Me2NH·BH3 (3) with Photoactivated CpFe(CO)2I (2) (5 mol %) To Form [Me2N−BH2]2 (4)
3051
dx.doi.org/10.1021/ja410129j | J. Am. Chem. Soc. 2014, 136, 3048−3064

Journal of the American Chemical Society
Article
Figure 4. (a) DLS particle size distribution by volume. (b) TEM micrograph showing evidence for the colloidal nature of the reaction of Me2NH·
BH3 (3) and [CpFe(CO)2]2 (1) (5 mol %) after photoirradiation (insert: higher magnification image). (c) Energy dispersive X-ray spectroscopic
(EDX) analysis of the particles (probe size diameter: 35 nm). The Cu in the EDX spectrum arises from the TEM grid on which the sample was
analyzed, while we propose that the Si arises from silicone grease present in the reaction mixture.
possible to directly distinguish between the two possible
reactive species in the reaction of 3 with photoactivated 1, it is
noteworthy that [Cp2Fe2(CO)3], the product of CO
dissociation, is known to have a greater solution lifetime than
that of [CpFe(CO)2].79,90
With respect to the mechanism for precatalyst 1, it was
conceivable that photoirradiation might only be required to
initiate catalysis, with the reaction then continuing after
irradiation has been halted. To probe this possibility, a solution
of 3 in THF was irradiated in the presence of 5 mol % 1 over 1
h, before analysis by 11B NMR spectroscopy. At this point,
∼50% conversion of 3 was apparent. The mixture was then
stirred at ambient temperature without further photoirradiation
for 3 h, while monitoring by 11B NMR spectroscopy, during
which time the reaction reached completion. Therefore, it was
clear that photoirradiation was not required throughout the
reaction to achieve catalytic turnover.91
Various methods exist for differentiating between homo- and
heterogeneous catalysis, with seminal contributions in the area
from Finke and co-workers.92−94 Two of the most important
tests used to distinguish between these catalysts are (a) kinetic
studies to investigate the presence of an induction period (vide
supra) and (b) monitoring the effects of selective poisoning and
filtration. With respect to kinetic measurements, the generation
of the active catalyst under photolysis conditions precluded
accurate monitoring of the initial stages of the reaction by 11B
NMR spectroscopy, although later stages of the reaction could
be monitored. Photoirradiation also made selective poisoning
experiments non-trivial to perform. For these experiments,
most commonly a phosphine or related donor is employed as a
potential poison for the active catalytic sites, with the
stoichiometry required to completely halt the reaction being
key to distinguishing between a heterogeneous or homoge-
neous process. Thus, for a homogeneous process, assuming a
single active site per complex, a stoichiometric quantity of
phosphine relative to catalyst would be needed to completely
curtail catalysis. In contrast, in the case of a heterogeneous
system, a substoichiometric quantity is sufficient due to the
relatively small fraction of metal centers actually available for
catalysis at the metal surface.
To explore selective poisoning, amine−borane 3 was treated
with precatalyst 1 (5 mol %) at 20 °C in THF under
photoirradiation for 1 h. Analysis by 11B NMR spectroscopy
indicated ∼50% conversion to 4 after the allotted time. The
reaction mixture was then divided in two, with one sample (i)
treated with 0.1 equiv of PMe3 (relative to 1) and the other (ii)
left unchanged. After a further 3 h with no photoirradiation, the
two samples were again analyzed by 11B NMR spectroscopy,
with (i) showing no further conversion indicting complete
suppression of catalysis and (ii) showing 100% conversion to 4.
These results are consistent with the presence of a
heterogeneous, colloidal Fe nanoparticle catalyst rather than
an active homogeneous species.
To further probe the nature of the active catalyst, dynamic
light scattering (DLS) studies were performed. DLS is a
technique used for determining the size of particles in a
3052
dx.doi.org/10.1021/ja410129j | J. Am. Chem. Soc. 2014, 136, 3048−3064

Journal of the American Chemical Society
Article
colloidal solution. In the context of this research, DLS is
particularly useful as it can be used to confirm the heterogeneity
of a catalytic reaction by the detection of metal colloid particles.
Furthermore, the technique can be applied to samples under an
inert atmosphere and in a broad range of common solvents. A
solution of precatalyst 1 (5 mol %) and amine−borane 3 in
THF was photoirradiated for 1 h to ensure the formation of the
catalytically active species and was then analyzed by DLS. The
resulting volume distribution, derived from a distribution fit of
the correlation function (see the Supporting Information,
Figures S1 and S2), demonstrated the presence of colloidal
species in solution. Specifically, a population of particles was
detected centered at a hydrodynamic radius, Rh, of 19 nm with
a minor contribution at larger size (Rh = ca. 95 nm) (Figure
4a). An aliquot was subsequently removed and dropped onto a
carbon-coated copper grid for analysis by transmission electron
microscopy (TEM). The images obtained (e.g., Figure 4b)
indicated the presence of particles with an average diameter of
7 nm which were confirmed to contain Fe by energy-dispersive
X-ray spectroscopic (EDX) analysis (Figure 4c).95 This result is
broadly consistent with that obtained by DLS when both errors
associated with surface coatings and solvation, which may not
be evident by TEM, and the assumptions required to obtain Rh
from the diffusion coefficient are considered.96
In order to exclude the possibility that the Fe nanoparticles
were formed on exposure of the Fe complex to the electron
beam during the TEM experiment, rather than being already
present in the reaction mixture, a solution of 1 in THF was
prepared and an aliquot drop-cast on a carbon-coated copper
grid. Imaging by TEM revealed the presence of some particles
(Figure S4(a), Supporting Information), albeit with consid-
erably lower surface density on the grid than observed for the
reaction mixture. Analysis of the same solution of 1 by DLS,
however, afforded no evidence for the presence of particles
within the experimentally detectable size range (Rh = 0.3 nm
−10 μm) (see Figure S3, Supporting Information). This
suggested that the nanoparticles observed at low surface density
in the TEM image from the control solution (Figure S4(a),
Supporting Information) were a product of the technique itself
and were not inherent to the sample.
Another useful analytical technique for investigating the
presence of a heterogeneous catalyst is UV−vis spectroscopy.
This method can be used to detect the surface plasmon
resonances associated with metallic nanoparticles, which appear
as broad absorptions through the visible spectrum down to
short wavelengths (below 400 nm).97,98 However, the UV−vis
spectrum obtained upon analysis of a reaction mixture
containing precatalyst 1 (5 mol %) and amine−borane 3 in
THF following photoirradiation (1 h) showed little change
from that of a control sample of 1 in THF. This contrasts with
the results of poisoning studies and TEM analysis which
strongly indicate that the catalyst is heterogeneous. It is,
however, likely that only a very small amount of the initial
precatalyst 1 may be photoactivated and/or reduced to
nanoparticles, which may lead to a minimal change in the
UV−vis spectrum.
(b). CpFe(CO)2I (2) as Precatalyst. As for the case of
precatalyst 1, the photochemistry of 2 has been extensively
studied, with two major processes identified. These involve CO
dissociation to form the neutral 16-electron complex
[CpFe(CO)I]99,100 and heterolytic Fe−I bond cleavage to
form the cationic 16-electron [CpFe(CO)2]+.101,102 We
therefore performed a number of experiments to investigate
the differences between the active catalytic species formed by 1
and 2 under the reaction conditions.
As discussed above (section 4 (a)), we found that
photoirradiation of 3 with 1 was only necessary to initiate
catalysis with further dehydrogenation observed after photolysis
had been halted. We performed analogous experiments to
investigate whether precatalyst 2 exhibited similar behavior.
Reaction of 3 with 2 (5 mol %) in toluene with photo-
irradiation for 4 h led to 60% conversion to 4 upon analysis by
11B NMR spectroscopy. The reaction mixture was then stirred
for 18 h in the absence of photoirradiation before further
analysis by 11B NMR spectroscopy. In this case and in contrast
to that for 1, no further conversion was detected demonstrating
that continuous photoirradiation was necessary to achieve
catalysis when using 2.
As previously noted, the presence of a different intermediate
compared to that observed with precatalyst 1 suggested the
operation of a fundamentally different mechanism, and this is
supported by the loss of catalytic activity after the cessation of
photoirradiation. The same tests that indicated the active
catalyst formed from 1 was heterogeneous were then repeated
to probe the nature of the active species formed during the
reaction of 3 with precatalyst 2 under photoirradiation. Thus,
following partial (ca. 60% conversion) conversion of 3 to 4
mediated by 5 mol % of 2 in toluene after photoirradiation for 4
h, addition of 0.4 equiv of PMe3 relative to catalyst followed by
further photoirradiation led to continued, but slower,
dehydrocoupling to yield 4 (ca. 80% total conversion after a
further 3 h). When the reaction was repeated with 1 equiv of
PMe3, complete suppression of catalytic activity was observed.
These results are in line with those expected for a
homogeneous catalyst and contrast with those for 1, where
complete suppression of activity was detected with substoichio-
metric quantities of PMe3 (e.g., 0.1 equiv). Furthermore, both
TEM and DLS analysis of reaction solutions derived from the
reaction of 3 with precatalyst 2 (5 mol %) in toluene also failed
to show evidence for the presence of a significant concentration
of Fe nanoparticles. For example, DLS analysis of the resulting
mixture after 1 h (Figure S1, Supporting Information)
suggested that the reaction mixture did not contain particles
within the experimentally observable size range (Rh = 0.3 nm to
10 μm). Once again, the results for precatalyst 2 contrast with
those obtained for precatalyst 1 and are strongly indicative of a
homogeneous process in this instance.
To provide further insight into the nature of the active
homogeneous catalyst and the potential role of CO and I−
ligand loss, we also explored the behavior of several related Fe
carbonyl complexes as precatalysts. We already noted that the
chloro analogue of 2, complex 12, showed similar activity for
the dehydrogenation of 3 under photoirradiation (Table 1).
Similarly, treatment of 3 with 5 mol % of [CpFe(CO)2(THF)]-
[X] ([X]− = [BF4]− 13a, [SbF6]− 13b) and CpFe(CO)2(OTf)
(OTf = OSO2CF3) (14) in the presence of UV light in either
THF or toluene also gave a conversion to 4 by 11B NMR
spectroscopy similar to that detected with 2 as the precatalyst
(Table S2, Supporting Information). In addition, as observed
with 2 as precatalyst, no catalytic activity was detected in the
absence of photoirradiation, and the reactions with 12, 13a and
13b, and 14 (Scheme 6) were all found to proceed via the
linear diborazane 6 as an intermediate. Moreover, catalysis
ceased when UV irradiation was halted in the cases of 12 and
14 (5 mol % precatalyst, toluene, 20 °C). In contrast, in the
cases of 13a and 13b, which possess the weakly coordinating
3053
dx.doi.org/10.1021/ja410129j | J. Am. Chem. Soc. 2014, 136, 3048−3064

Journal of the American Chemical Society
Article
much less than the ca. 60% conversion otherwise anticipated.
The relative suppression of catalysis upon addition of CO or I−
provides further support for the need for photoinduced
dissociation of both CO and I− dissociation from 2 to generate
the active catalyst. In addition, the observation that, once
initiated, catalysis continues in the absence of further irradiation
in the case of 13a and 13b but not for 2 may be a consequence
of recoordination of the iodide anions at the metal center in the
latter case. The reactions with added CO and I− were also
repeated with precatalyst 13b. Similarly, reaction of 3 with 13b
(5 mol %) under an atmosphere of CO in toluene for 4 h with
photoirradiation at 20 °C led to a suppression of catalytic
activity based on analysis by 11B NMR spectroscopy (5%
conversion).104 Upon replacing the atmosphere of CO with N2
and photoirradiating for a further 18 h, 60% conversion of 3 to
4 was detected. Also, treatment of 3 and an equimolar amount
of [n-Bu4N]I with 13b (5 mol %) in THF at 20 °C under
photoirradiation for 4 h led to minimal conversion to 4 (10%)
upon analysis by 11B NMR spectroscopy.
To provide further insight into the likely intermediates in the
catalytic cycle for 2, DFT calculations were performed on the
chloro analogue 12 which behaves similarly.105 Initially, we
attempted to calculate possible intermediates based on the
generation of the 16-electron species [CpFe(CO)Cl] or
[CpFe(CO)2]+, arising from either CO or halide loss from
12, respectively. However, as these possible intermediates
possess only a single vacant coordination site we found that
complexes such as [CpFe(CO)Cl(H3BNMe2H)2] or [CpFe-
(CO)2(H3BNMe2H)2]+, upon optimization, always led to the
loss of one molecule of coordinated 3. We, therefore,
investigated species based on [CpFe(CO)]+, the product
from both CO and halide dissociation, as the key intermediate
which is presumably formed in amine−borane ligand-stabilized
form (15) (Scheme 7). This 14-electron complex offers two
vacant coordination sites, and we examined a series of plausible
intermediates that might arise from this species (Figure 5).106
The first coordination of 1 equiv of 3 to the [CpFe(CO)2]+
cation was found to be highly exergonic by −136.0 kJ·mol−1,
and the subsequent loss of one molecule of CO was calculated
to be slightly exergonic by −7.1 kJ·mol−1. The coordination of a
second equivalent of 3 was found to be exergonic as well by
Scheme 6. CpFe(CO)2I (2) and Related Complexes
[BF4]− and [SbF6]− anions, respectively, slower but significant
further conversion of 3 to 4 was detected (20% conversion after
2 h with UV light, with 60% further conversion after an
additional 18 h when photoirradiation was halted in the case of
13a and 40% further conversion in the case of 13b).
The aforementioned observations with structurally related
complexes suggest that loss of both CO and I− from 2 may be
necessary in order to generate the active homogeneous
catalyst.103 For example, the activity under UV irradiation of
13a and 13b, in which the organoiron cations already lack a
halide ligand, is also strongly indicative of the need for CO loss.
To explore this further, we investigated the effect of CO and I−
on the dehydrogenation of 3 with 2 as precatalyst. A THF
solution of amine−borane 3 and 2 (5 mol %) was exposed to
an atmosphere of CO, and the reaction mixture was
photoirradiated for 4 h. Analysis by 11B NMR spectroscopy
showed only 5% conversion to 4 which was significantly less
than observed under usual conditions with an N2 atmosphere
(60% conversion in 4 h). Subsequent replacement of CO by an
atmosphere of N2 together with photoirradiation for 16 h
resulted in further high conversion of 3 to 4 (70%). To
investigate the effect of I− on the conversion rate, an equimolar
quantity of [n-Bu4N]I and 3 was treated with 2 (5 mol %) in
THF at 20 °C under photoirradiation for 4 h. This led to only
10% conversion to 4 upon analysis by 11B NMR spectroscopy,
Scheme 7. Possible Processes in the Formation of the Active Catalyst from CpFe(CO)2I (2) in the Reaction with Me2NH·BH3
(3) under Photoirradiation Conditions in an Open System
3054
dx.doi.org/10.1021/ja410129j | J. Am. Chem. Soc. 2014, 136, 3048−3064

Journal of the American Chemical Society
Article
Figure 5. Calculated relative free enthalpies at 298 K (in kJ·mol−1) for relevant intermediates of the reaction of [CpFe(CO)2]+ + 3 (all calculations
performed for the formula C11H25B2FeN2O2, respectively). Calculations were performed at the M06-2X/6-311G(d,p)(C, H, B, N, O);SDD(Fe)
level of theory.
−19.6 kJ·mol−1. In all cases, coordination to Fe through B−H
Scheme 8. Synthesis of Cp2Fe2(CO)3(MeCN) (7)
groups was preferred over the N−H groups. The B−H/N−H
bond cleavage and the B−N bond formation process is most
likely the rate-determining step for the first part of the catalytic
cycle in which the linear diborazane is generated. The overall
reaction of [CpFe(CO)2]+ with two molecules of 3 to give the
linear diborazane 6 coordinated to the metal is exergonic (by
−43.9 kJ·mol−1), while the subsequent step to form the
coordinated cyclic diborazane 4 is endergonic (+77.9 kJ·mol−1)
apparent in most solvents presumably instigated by MeCN
and the formation of free 4 from 6 is exergonic (by −57.0 kJ·
mol−1). This is in agreement with the experimental observation
that amine−borane 3 is first transformed to 6 and that only
after most of the starting material has been consumed,
conversion to the cyclic diborazane 4 as the final product is
observed. Furthermore, these calculations, coupled with the
aforementioned experimental results, are consistent with the
loss of CO and halide to form the active catalytic species from
2.
dissociation. In contrast, precatalyst 7 was shown to be
indefinitely stable in MeCN solution, presumably caused by
inhibition of ligand dissociation due to the large excess of
MeCN present.
To probe the reactivity of 7 as a precatalyst, amine−borane 3
was treated with 5 mol % of 7 at 20 °C in toluene under
conditions analogous to those employed for complexes 1 and 2
except that photoirradiation was not used. The reaction mixture
was subsequently monitored by 11B NMR spectroscopy.
Quantitative conversion to cyclic diborazane 4 was observed
in only 20 min, confirming the very high activity of 7 as an
amine−borane dehydrogenation precatalyst (Scheme 9). When
5. Studies of Cp2Fe2(CO)3(MeCN) (7): A Fe Precatalyst
for Amine−Borane Dehydrogenation That Does Not
Require Photoirradiation. Although compounds 1 and 2 are
commercially available and relatively efficient precatalysts for
the dehydrogenation of 3,107 the necessity for photoirradiation
increased the complexity of the experimental setup and also
limited the scope of the mechanistic studies. It was, therefore,
highly desirable to identify related Fe complexes, for which
photoirradiation was not a prerequisite for catalytic activity.
It has previously been demonstrated that the photo-
irradiation of Fe complex 1 in the presence of various two
electron donors can lead to the formation of new species of the
form Cp2Fe2(CO)3L [L = MeCN, PR3, P(OR)3].108,109 If the
new ligand is particularly weakly coordinating, under certain
conditions it may dissociate to yield a coordinatively
unsaturated species Cp2Fe2(CO)3, a likely intermediate formed
during the photoirradiation of precatalyst 1 in the presence of 3
[section 4 (a)]. The targeted complex Cp2Fe2(CO)3(MeCN)
(7) was synthesized via a slight modification of the method
reported by Labinger et al. through photolysis of 1 in MeCN
solution (Scheme 8).109 The compound was isolated as a green,
highly air-sensitive solid and was subsequently stored at −40 °C
under N2. Analysis by 1H, 13C NMR spectroscopy and UV−vis
spectroscopy confirmed the identity of the compound based on
the previous reports with relatively rapid decomposition
Scheme 9. Dehydrogenation of Me2NH·BH3 (3) with
Cp2Fe2(CO)3(MeCN) (7) (5 mol %) Forming [Me2N−
BH2]2 (4)
the activity of this complex was compared with other known
dehydrocoupling catalysts for 3, it was found that the rate of
dehydrogenation was among the most rapid known for this
transformation.87,110 Interestingly, catalysis was found to be
completely suppressed upon repeating the reaction in neat
MeCN (16 h, 20 °C by 11B NMR spectroscopy), presumably
due to inhibition of the ligand loss necessary for the generation
of the active catalytically species.111
Despite the rapid rate of dehydrogenation of amine−borane
3 with 7 as a precatalyst relative to that observed with complex
1, the monomeric aminoborane 5 (δB 36.5 ppm) was again
detected as an intermediate by 11B NMR spectroscopy during
3055
dx.doi.org/10.1021/ja410129j | J. Am. Chem. Soc. 2014, 136, 3048−3064

Journal of the American Chemical Society
Article
the course of the reaction (Figure 6). Furthermore, as with the
case of 1, linear diborazane 6 was not observed in significant
added. This was completely dehydrocoupled in a further 10
min. This process was repeated an additional three times before
any significant decrease in rate was observed.
6. Mechanistic Insight into the Dehydrocoupling of
Me2NH·BH3 (3) Using Cp2Fe2(CO)3(MeCN) (7) as a
Precatalyst. As noted previously, the need to photoirradiate
precatalysts 1 and 2 to achieve turnover complicated the
detailed analysis of the reactions. This was especially true in the
case of 2 as continual photoirradiation was required. Precatalyst
7 therefore provided an important opportunity to probe the
mechanism of dehydrogenation in more detail as catalysis was
achieved without photoirradiation. Furthermore, it is likely that
the initial unsaturated species formed from precatalyst 7 is
identical to that for 1 and further insight into the detailed
mechanism for the latter species would be accessible.
The reaction of precatalyst 7 with amine−borane 3 in
toluene was monitored by 11B NMR spectroscopy, using a
reduced catalyst loading of 0.5 mol % to achieve a reaction time
scale appropriate for accurate monitoring (∼110 min). Analysis
of the reaction every 2 min over 0.5 h provided clear evidence
for a significant induction period prior to the observation of
catalytic activity (Figure 7, plot a). In addition, the reaction
solution was observed to change from light green to light
brown in color with significant bubbling, concomitant with the
detection of 4 and 5 by 11B NMR spectroscopy. Repeat
experiments under the same conditions demonstrated reprodu-
cibility in these observations, with an average induction period
of 22 min over five runs.
The observation of an induction period and a color change
upon the onset of active catalysis strongly suggested, as
anticipated, that the initial MeCN adduct is not the active
catalyst in the dehydrogenation reaction. It is probable that the
initial process in solution would involve MeCN dissociation.
Evidence in support of this assertion was given by the
aforementioned suppression of catalysis when the dehydrogen-
ation of 3 using 7 is attempted in MeCN, where ligand
dissociation is presumably prevented (see section 5).
Figure 6. 11B {1H} NMR spectrum of the reaction mixture (toluene,
20 °C) derived from Me2NH·BH3 (3) with Cp2Fe2(CO)3(MeCN)
(7) (5 mol %) forming [Me2N−BH2]2 (4) via Me2NBH2 (5) after
10 min.
amounts. In line with these results, separate experiments also
showed that 7 was virtually inactive toward the dehydrogen-
ation of 6 with less than 5% conversion to cyclic diborazane 4
detected by 11B NMR spectroscopy after 21 h at 20 °C in
toluene. The experimental evidence therefore suggested that 3
may be dehydrocoupled by precatalyts 1 and 7 by similar or
related mechanisms, both based on an initial intramolecular
dehydrogenation process.
The ability of the in situ generated Fe catalyst to
dehydrocouple multiple batches of 3 was also explored.
Following the conversion of 3 to 4 in 20 min with 5 mol %
of precatalyst 7 in toluene, another equivalent of 3 was then
Figure 7. Conversion of Me2NH·BH3 (3) to [Me2N−BH2]2 (4) with time for a 0.5 mol % loading of Cp2Fe2(CO)3(MeCN) (7) in toluene at 20 °C.
Conversions based on analysis of 11B NMR spectra. Plot a shows a standard catalytic run in which 3 and 7 are added to solution at the same time.
Plot b shows the catalytic run in which 7 was stirred with an excess of Me3N·BH3 (16) for 30 min before addition of 3.
3056
dx.doi.org/10.1021/ja410129j | J. Am. Chem. Soc. 2014, 136, 3048−3064

Journal of the American Chemical Society
Article
Figure 8. (a) DLS particle size distribution by volume. (b) TEM micrograph showing evidence for the colloidal nature of the reaction of Me2NH·
BH3 (3) and Cp2Fe2(CO)3(MeCN) (7) (5 mol %) (insert: high magnification image). (c) Energy-dispersive X-ray spectroscopic (EDX) analysis of
the particles (probe size diameter: 35 nm). The Cu in the EDX spectrum arises from the TEM grid on which the sample was analyzed.
The results suggest that it is highly likely that the active
catalyst formed by 7 is heterogeneous in nature and might be
formed simply via reduction of the initial metal complex by the
amine−borane substrate, a process for which there is
substantial literature precedent.87,112 As noted above, nano-
particles based on various metals have been shown to
dehydrocouple amine−boranes under hydrolytic69,113and non-
hydrolytic conditions,49 which lends more support to the
proposal that Fe nanoparticles are the active catalyst formed via
in situ reduction of precatalyst 7. To further investigate this
hypothesis, an attempt was made to reduce 7 to form Fe
nanoparticles prior to the addition of 3. Precatalyst 7 (0.5 mol
%) was stirred in toluene with an excess of Me3N·BH3 (16),
which was used as the reducing agent due to its solubility in
toluene and also due to the lack of a feasible dehydrocoupling
pathway. After 30 min, which was comparable to the induction
period observed in the reaction of amine−borane 3 with
precatalyst 7, amine−borane 3 was added and the reaction was
analyzed every 2 min by 11B NMR spectroscopy. In this case, a
far shorter induction period (between 0 and 6 min over three
runs) and an improved reaction rate were observed (Figure 7,
plot b). This further supported the hypothesis that Fe
nanoparticles formed in situ by the reduction of 7 by 3 were
the true active catalyst.
complete conversion to 4 as expected. Next, 0.1 equiv of
PMe3 (with regard to the total iron content in the reaction
mixture) was then added before the subsequent addition of a
further equivalent of 3. Analysis by 11B NMR spectroscopy
showed no further dehydrogenation over 120 min indicating
that a substoichiometric quantity of PMe3 completely sup-
pressed the activity of the catalyst. Repeating the experiment
using CS2 as the catalyst poison yielded a similar result.
Furthermore, as a control experiment, the reaction was
repeated with omission of the poisoning agent. In this case,
the second equivalent of amine−borane 3 was rapidly
dehydrocoupled by the active catalyst generated in situ from
7 affording cyclic diborazane 4 as the final product within 10
min. The reaction was then repeated with addition of a
substoichiometric quantity of PMe3 to 7 before the addition of
3. Precatalyst 7 (5 mol %) and 0.1 equiv of PMe3 were added to
toluene and stirred for 10 min before addition of amine−
borane 3. Analysis of the reaction mixture after 30 min by 11B
NMR spectroscopy indicated no dehydrocoupling had
occurred.
The reaction mixture containing 7 (5 mol %) and amine−
borane 3 in toluene was also analyzed by DLS as previously
described for the reaction with precatalyst 1. As for the case of
1, this indicated the existence of a colloidal solution, albeit
containing particles of slightly larger size (Rh = 148 nm, Figure
8a). Broadly consistent with this result was the observation by
TEM of iron-containing114 particles ranging in diameter from
ca. 10−30 nm with the average size calculated as 10 nm (Figure
8b,c), which are slightly larger values than those found in the
reaction with 1.115,116
As discussed previously in section 4, various tests to
differentiate homo- and heterogeneous catalytic processes
have been devised.92,93 The ability of PMe3 to selectively
poison the catalyst derived from 7 was also investigated.
Amine−borane 3 was reacted with 7 (5 mol %) in toluene for
30 min, by which time 11B NMR spectroscopy showed
3057
dx.doi.org/10.1021/ja410129j | J. Am. Chem. Soc. 2014, 136, 3048−3064

Journal of the American Chemical Society
Article
DLS and TEM control experiments were also performed for
precatalyst 7 in MeCN solution, as previously described for 1 in
THF. Although some particulates were observed in the TEM
images (Figure S4 (b), Supporting Information), the surface
density was much lower than that found in the reaction with 3,
and there was no evidence for a significant concentration of
such species by DLS (see Figure S3, Supporting Information).
We therefore conclude that, as for 1, the nanoparticles observed
in the reaction with 3 are native, although we cannot
completely exclude the possibility that some of those observed
by TEM arise due to reduction of the precatalyst by the
electron beam.
As with the case of 1, UV−vis spectroscopy was again used in
an attempt to detect iron nanoparticles with 7 as the
dehydrocoupling catalyst. Although significantly different
spectra were recorded for 7 and for the reaction mixture with
3 consisting of a substantial broadening of absorptions through
the visible spectrum to lower wavelengths, this is consistent
with, rather than convincing evidence for, the presence of a
surface plasmon resonance absorption of metallic nanoparticles
(see Figures S6 and S7, Supporting Information; for a UV−vis
spectrum of Fe nanoparticles see section 7 and Figure S11,
Supporting Information).97
After determining that the active catalyst for the
dehydrocoupling reactions catalyzed by 1 and 7 was
heterogeneous, it was of interest to reinvestigate the Fe
carbonyl dimer complex Fe2(CO)9 (8) for which we previously
obtained negative results as a precatalyst. The inherent
photosensitivity of this complex, which functions as a
convenient source of nanoparticulate Fe, made it very
surprising that no dehydrocoupling activity was observed
under the catalyst screening conditions.117 Upon reinvesti-
gation, 8 (5 mol %) was photoirradiated together with amine−
borane 3 in THF at 20 °C for 24 h. Analysis by 11B NMR
spectroscopy again showed minimal dehydrogenation to afford
4 (ca. 1%). The reaction mixture was then analyzed by DLS,
which indicated the formation of a colloidal solution with
relatively large particles (Rh = 550 nm), and this was further
confirmed by TEM (Figure S14, Supporting Information). It is
likely that the inactivity of 8 is therefore due to the large size of
the particles and consequently low concentration of active
surface sites.
7. Investigation of Fe Nanoparticles as an Amine−
Borane Dehydrocoupling Catalyst. To support the
assertion that Fe nanoparticles are the active catalytic species
in dehydrocoupling reactions of 3 mediated by 7 and 1, it was
important to investigate the activity of independently prepared
Fe nanoparticles toward the dehydrogenation of 3. Among the
examples of Fe nanoparticles reported in the literature, two
distinct classes are apparent, defined by the presence or lack of
ancillary ligands to prevent agglomeration. We therefore set out
to prepare representative examples of both classes as a means of
probing the general reactivity of Fe nanoparticles in this
context.
TEM, which showed the formation of large Fe aggregates (1
μm) (Figure S9, Supporting Information). The catalytic activity
of the resulting magnetic solids was then investigated through
reaction with amine−borane 3 in toluene and THF at a 5 mol
% loading. In both cases no evidence of dehydrogenation was
apparent over 4 h at 20 °C by 11B NMR spectroscopy.
In the case of ligand-supported Fe nanoparticles, both
phosphines and amines are commonly employed in the
ancillary role, as well as various polymers.119,120 We employed
the method of Amiens and co-workers, who demonstrated the
stabilization of Fe nanoparticles by poly(2,6-dimethyl-1,4-
phenylene oxide).121 It was postulated that this polyaryl ether
stabilizer would be suitably non-interacting with the amine−
borane substrate so as to not complicate the reaction. The Fe
nanoparticles were prepared via the reduction of Fe(N-
[SiMe3]2)2 by H2 at 110 °C in the presence of the stabilizer.
The Fe nanoparticles, which were isolated as a black solid, were
characterized by TEM and UV−vis spectroscopy. Fe nano-
particles were observed by TEM (ca. 30 nm, Figure S10,
Supporting Information), although these were larger than the
nanoparticles generated from 1 or 7. UV−vis spectroscopy also
clearly showed a surface plasmon resonance typically indicative
of nanoparticles (Figure S11, Supporting Information).
However, once again, no catalytic reactivity was observed on
treatment of toluene or THF solutions of 3 with a 5 mol %
loading of these nanoparticles at 20 °C after 4 h.
The last examples of ligand-stabilized nanoparticles that were
synthesized were those published by Bonnemann et al.122,123 In
̈
this case, the Fe nanoparticles were prepared via the reduction
of FeCl2 by [N(Octyl)4][BEt3H] at 20 °C in THF. The
[N(octyl)4]+ cation stabilized Fe nanoparticles (17), which
were isolated as a dark gray solid, were then characterized by
TEM, which showed that, unlike the cases of the other
methods, small nanoparticles (∼3 nm) were formed (Figure
S12, Supporting Information). Treatment of 3 with 5 mol %
loading of 17 in toluene at 20 °C over 20 h led to 40%
conversion of 3 to 4 was detected by 11B NMR spectroscopy. A
similar conversion was observed when the reaction was
repeated in THF, and although the activity is significantly less
than that detected using precatalyst 1 and 7, this may be a
consequence of the different surface ligation. Interestingly,
there were also mechanistic similarities between the dehy-
drocoupling of amine−borane 3 with 17 and for precatalysts 1
and 7. Analysis of the reaction mixture by 11B NMR
spectroscopy showed that the aminoborane 5 was formed,
whereas linear diborazane 6 was not detected (Figure S13,
Supporting Information). Further insight was given by the very
low activity of the 17 toward the dehydrocoupling of 6,
suggesting that a similar mechanism was in operation for all the
heterogeneous Fe(0) catalysts investigated.124
DISCUSSION
■
The observation of the monomeric aminoborane 5 as an
intermediate in the dehydrogenation reactions catalyzed with
[CpFe(CO)2]2 (1) and Cp2Fe2(CO)3(MeCN) (7), in contrast
to those with CpFe(CO)2X (X− = I− (2), Cl− (12), OTf−
(14)) and also [CpFe(CO)2(THF)][X] X− = [BF4]− or
[SbF6]− (13a and 13b) as a precatalyst, suggests a potential
similarity in the reaction mechanisms in the case of the former
dinuclear species. Mechanistic studies for 1 involving poisoning
experiments and TEM and DLS analysis of reaction solutions
suggested that this species was reduced on photoirradiation in
the presence of 3 to generate Fe nanoparticles as the true
The reactivity of stabilizer-free Fe nanoparticles was initially
investigated with a sample prepared via reduction of FeCl3 by
NaBH4 in aqueous media using the method of Klabunde and
co-workers.118 Due to the likelihood of metal-catalyzed
hydrolysis of amine−boranes in aqueous media, the nano-
particles were isolated by cannula filtration before washing
successively in degassed H2O, EtOH, and dry Et2O and finally
drying under high vacuum for 2 h. The stabilizer-free Fe
nanoparticles were isolated as a black solid and characterized by
3058
dx.doi.org/10.1021/ja410129j | J. Am. Chem. Soc. 2014, 136, 3048−3064

Journal of the American Chemical Society
Article
Scheme 10. Possible Routes to Forming an Unsaturated Fe Complex Using [CpFe(CO)2]2 (1) and Cp2Fe2(CO)3(MeCN) (7)
Scheme 11. Possible Mechanism for the Dehydrogenation of Me2NH·BH3 (3) To Afford Me2NBH2 (5) on a Metal
Nanoparticle Surface and Subsequent “Off-Metal” Dimerization To Yield 4
heterogeneous catalyst. The discovery of the precatalyst 7,
which does not require photoirradiation to mediate the
dehydrocoupling of 3, facilitated more detailed mechanistic
studies. Significantly, these revealed the presence of an
induction period associated with the reduction step, a key
signature for a heterogeneous catalytic process. These
observations support the assertion that 1 and 7 are each
reduced to Fe nanoparticles, the active catalytic species in the
dehydrogenation of 3, probably via the coordinatively
unsaturated complex [Cp2Fe2(CO)3] in a process that requires
the presence of 3 (Scheme 10). This process requires
photoirradiation in the case of 1 to remove CO, unlike the
case for 7. It is possible that the CO generated in the case of 1
can coordinate to the surface of the Fe nanoparticles. This
could reduce their catalytic activity and might explain the lower
rate found for 1 compared to 7 for the catalytic dehydrocou-
pling of 3. The reduction of 1 and 7 to Fe(0) may involve
intermediates similar to those noted based on studies of the
photochemistry of 1 with H2 and main group hydride species
by Bitterwolf et al.125 where E-H bond activation followed by
subsequent rearrangement afforded iron-containing complexes
such as CpFe(CO)2H.
give 4 (although it is also possible that the transformation can
also be metal-mediated).72 This mechanistic pathway has been
invoked for various catalyst systems, although in the majority of
cases the aminoborane 5 is observed only in small quantities
due to the slow generation of the species and relativity short
solution lifetime.39,56−58,72,126,127
In the second of these mechanisms, the linear diborazane 6 is
formed as the key intermediate and is subsequently
dehydrogenated in a second “on-metal” process to give the
final product. This is exemplified in the recent study from our
group and collaborators of the [Cp2Ti]-mediated dehydrogen-
ation of secondary amine−boranes, in which experimental
observations and kinetic studies strongly implicate the linear
diborazane 6 as the key intermediate.44 Diborazane 6 has also
been postulated as an intermediate in the dehydrogenation of 3
in various other catalyst systems.42,52,57,71,126,128
In the case of photoactivated 1, the observation of significant
quantities of the monomeric aminoborane 5 during the
dehydrogenation of 3, coupled with the minimal reactivity of
this catalyst toward isolated 6, makes this one of the most
convincing examples of a dehydrocoupling of 3 solely via the
monomeric aminoborane. Interestingly, a trend develops when
comparing the intermediates detected using precatalysts 1 or 7
to other amine−borane dehydrocoupling catalysts believed to
be heterogeneous in nature. In the case of skeletal Ni,51 the
major pathway for the dehydrogenation of 3 was an “off-metal”
process affording 4 via aminoborane 5, in accordance with
reactivity observed when using in situ or ex situ generated Fe
nanoparticles. Furthermore, the extensively studied [Rh(1,5-
As discussed previously, two different basic metal-mediated
mechanisms for the dehydrocoupling of amine−borane 3 have
been postulated in the literature (Scheme 2). In the first, 1
equiv of H2 is removed from a single molecule of 3 in a metal-
catalyzed intramolecular process to yield the monomeric
aminoborane 5 as a short-lived intermediate. The aminoborane
subsequently dimerizes in a spontaneous “off-metal” process to
3059
dx.doi.org/10.1021/ja410129j | J. Am. Chem. Soc. 2014, 136, 3048−3064

Journal of the American Chemical Society
Article
Scheme 12. Possible Two-Stage Mechanism for the Dehydrocoupling of Me2NH·BH3 (3) with Iron Precatalyst CpFe(CO)2I (2)
cod)Cl]2 system,41,42,73,129−134 which has been reported to
generate Rh colloids and clusters during the dehydrocoupling
of 3, also proceeds mainly via aminoborane 5, although small
amounts of linear diborazane 6 were also detected (Figure S15,
Supporting Information). It is possible that the mechanism
involving 5 as an intermediate is favored in cases where the
active catalyst is heterogeneous as dehydrogenation of a single
molecule of 3 affording aminoborane 5 (Scheme 11) may be
more likely to occur on a metal surface than two molecules of
amine−borane 3 reacting via an intermolecular dehydrocou-
pling step to give linear diborazane 6. Interestingly, the
presence of small but significant amounts of 6 in the Ni and
[Rh(1,5-cod)Cl]2 systems may suggest that a second, perhaps
homogeneous, mechanism may also operate as a minor
component.135
In contrast to the cases of 1 and 7, experimental observations
indicate the formation of linear diborazane 6 in the
dehydrocoupling of aminoborane 3 catalyzed by Fe precatalyst
2 results from a homogeneous mechanism.136 Photolysis in this
instance is continuously required to achieve catalytic turnover
and does not result in the formation of detectable Fe
nanoparticles by TEM and DLS analysis. In this case, it is
highly likely that 2 dehydrocouples 3 via an initial
intermolecular coupling event to form 6, which then
subsequently dehydrogenates in an “on-metal” step to afford
4. This process is strongly reminiscent of the extensively
studied [Cp2Ti]-mediated dehydrocoupling mechanism for
344,54 and a similar, two-stage catalytic cycle may operate
(Scheme 12). Experimental studies of the catalytic dehydro-
coupling of 3 mediated by 2, and also 13a and 13b and 14,
together DFT calculations provide tentative support for the
proposed intermediacy of photogenerated [CpFe(CO)]+ which
readily binds 3 to form 15. In the case of precatalyst 2,
photogenerated I− can presumably recombine with the active
catalyst leading to the need for continuous irradiation to
achieve catalytic turnover. This explanation is supported by the
observation that continuous irradiation is not necessary in the
cases of precatalysts 13a and 13b, which possess weakly
coordinating anions.
An important yet challenging issue to address is the
relationship between the precatalyst structure and the
dehydrocoupling mechanism detected. Why are the mecha-
nisms involving precatalysts 1 and 7 heterogeneous whereas
that of 2 is homogeneous? We speculate that the low formal
oxidation state, Fe(I), and the binuclear structures in 1 and 7,
which already possess Fe−Fe interactions, may contribute to
their ease of reduction to Fe nanoparticles in the presence of
the reductant 3. On the other hand, entry into the
homogeneous catalytic manifold proposed in this work for 2
is probably enabled by the presence of the electronegative
iodine substituent which makes heterolytic cleavage of an iron-
ligand bond possible thereby maintaining the Fe(II) oxidation
state. Although not analyzed in detail in this work, we assume
that Fe(II) complexes 12, 13a, 13b, and 14 catalyze the
dehydrocoupling of 3 by analogous mechanisms to that for 2
(Scheme 12) and that these mononuclear Fe(II) species are
also resistant to reduction for similar reasons.
Finally, one other result warrants further comment. Why is
Fe2(CO)9, an obvious source of Fe nanoparticles, inactive as a
precatalyst? It appears that this complex decomposes to Fe
nanoparticles so rapidly that large Fe nanoparticles are formed
(Figure S14, Supporting Information) with low reactivity as a
result of their relatively low surface area. In contrast, the
smaller, catalytically active nanoparticles generated from
complexes 1 and 7 possess cyclopentadienyl ligands whose
loss may constitute the rate limiting step in this process. This
3060
dx.doi.org/10.1021/ja410129j | J. Am. Chem. Soc. 2014, 136, 3048−3064

Journal of the American Chemical Society
Article
explanation is supported by the lack of reactivity observed with
most of the Fe nanoparticles independently generated by
reductive techniques in the literature (see section 7).137 Only
when the size was very small (ca. 3 nm) and [N(Octyl)4]+
cations were used as stabilizers (Figure S12, Supporting
Information) was significant activity for the dehydrocoupling
of 3 apparent. Thus, unstabilized Fe nanoparticles formed large
micrometer-scale aggregates (Figure S9, Supporting Informa-
tion) and showed no catalytic activity. These results and our in
situ DLS and TEM studies suggest that the 1 and 7 function as
sources of small iron nanoparticles (≤10 nm) that are
presumably generated during the reaction induction period,
which was detected in the case of 7. The controlled generation
of nanoparticles is likely facilitated by the presence of
cyclopentadienyl ligands which are much less labile than CO
substituents.138 Presumably, the nanoparticles formed possess
sufficient surface stabilization to prevent rapid aggregation as
their catalytic activity is maintained after irradiation is halted (in
the case of 1) and recycling is possible.
AUTHOR INFORMATION
■
Corresponding Author
ian.manners@bris.ac.uk
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
J.R.V., A.S., A.P.M.R., G.R.W., and I.M. acknowledge the
Engineering and Physical Sciences Research Council (EPSRC)
for financial support.
REFERENCES
■
(1) Baudoin, O. Chem. Soc. Rev. 2011, 40, 4902−4911.
(2) Fihri, A.; Bouhrara, M.; Nekoueishahraki, B.; Basset, J.-M.;
Polshettiwar, V. Chem. Soc. Rev. 2011, 40, 5181−5203.
(3) Trnka, T. M.; Grubbs, R. H. Acc. Chem. Res. 2000, 34, 18−29.
(4) Schrock, R. R.; Hoveyda, A. H. Angew. Chem., Int. Ed. 2003, 42,
4592−4633.
(5) Kress, S.; Blechert, S. Chem. Soc. Rev. 2012, 41, 4389−4408.
(6) Eisch, J. J. Organometallics 2012, 31, 4917−4932.
(7) (a) Bolm, C.; Legros, J.; Le Paih, J.; Zani, L. Chem. Rev. 2004,
SUMMARY AND CONCLUSIONS
■
Amine−borane dehydrogenation/dehydrocoupling precatalysts
based on the earth-abundant metal iron have been developed.
Detailed studies have demonstrated that either heterogeneous
or homogeneous mechanisms may operate depending on subtle
structural variations of the precatalyst. The iron carbonyl
complexes 1 and 2 were found to be efficient precatalysts for
the dehydrocoupling/dehydrogenation of the amine−borane 3
upon UV photoirradiation at ambient temperature. In situ
analysis of the reaction mixtures indicated that different, two-
step mechanisms operate in each case. Thus, precatalyst 1
dehydrocoupled 3 via the aminoborane 5 which then
cyclodimerized to give the cyclodiborazane 4 via an off-metal
process. In contrast, the reaction with precatalyst 2 proceeded
via the linear diborazane 6 as the key intermediate, affording 4
as the final product after a second metal-mediated step. The
complex 7, formed by photoirradiation of 1 in MeCN, was
found to be a more highly active dehydrocoupling catalyst for 3,
and photoactivation was not required. Detailed mechanistic
studies indicated that the active catalysts generated from
precatalyst 7 and also 1 were heterogeneous in nature and
consisted of small iron nanoparticles (≤10 nm).139 In contrast,
analogous experimental studies for the case of photoactivated
Fe precatalyst 2 suggested that the active catalyst formed in this
case was homogeneous. Investigations of this and analogous
species (such as 12, 13a, 13b, and 14) together with DFT
studies suggested a two-stage catalytic cycle with amine−borane
ligated [CpFe(CO)]+ (15) as a key intermediate.
104, 6217−254. (b) Furstner, A.; Martin, R. Chem. Lett. 2005, 34,
̈
624−629. (c) Bauer, E. B. Curr. Org. Chem. 2008, 12, 1341−1369.
(8) Bedford, R. B.; Betham, M.; Bruce, D. W.; Danopoulous, A. A.;
Frost, R. M.; Hird, M. J. Org. Chem. 2006, 71, 1104−1110.
(9) Hudson, R.; Hamasaka, G.; Osako, T.; Yamada, Y. M. A.; Li, C.-
J.; Uozami, Y.; Moores, A. Green. Chem. 2013, 15, 2141−2148.
(10) (a) Bullock, R. M. Science 2013, 342, 1054−1055. (b) Jagadeesh,
R. V.; Surkus, A.-E.; Junge, H.; Pohl, M.-M.; Radnik, J.; Rabeah, J.;
Huan, H.; Schunemann, V.; Bruckner, A.; Beller, M. Science 2013, 342,
̈
̈
1073−1076. (c) Friedfield, M. R.; Shevlin, M.; Hoyt, J. M.; Krska, S.
W.; Tudge, M. T.; Chirik, P. J. Science 2013, 342, 1076−1080.
(d) Zuo, W.; Lough, A. J.; Li, Y. F.; Morris, R. H. Science 2013, 342,
1080−1083.
(11) Clark, T. J.; Lee, K.; Manners, I. Chem.Eur. J. 2006, 12, 8634−
8648.
(12) Less, R. J.; Melen, R. L.; Wright, D. S. RSC Adv. 2012, 2, 2191−
2199.
(13) Waterman, R. Chem. Soc. Rev. 2013, 42, 5629−5641.
(14) Leitao, E. M.; Jurca, T.; Manners, I. Nature Chem. 2013, 5, 817−
829.
(15) Hutchins, R. O.; Learn, K.; Nazer, B.; Pytlewski, D.; Pelter, A.
Org. Prep. Proced. Int. 1984, 16, 335−372.
(16) Lane, C. F. Aldrichimica Acta 1973, 6, 51−58.
(17) Bartoli, G.; Bartolacci, M.; Giuliani, A.; Marcantoni, E.;
Massaccesi, M. Eur. J. Org. Chem. 2005, 2867−2879.
(18) Pasumansky, L.; Goralski, C. T.; Singaram, B. Org. Process. Res.
Dev. 2006, 10, 959−970.
(19) Kanth, J. V. B. Aldrichimica Acta 2002, 35, 57−66.
(20) Beachley, O. T., Jr.; Washburn, B. Inorg. Chem 1975, 14, 120−
123.
Further work will focus on an investigation of the
dehydrocoupling reactivity of the Fe precatalysts toward
primary amine−borane adducts and other substrates such as
9 that enable polyaminoborane formation. We also aim to
perform further mechanistic work and model compound
studies in order to provide further detailed support for the
homogeneous mechanism proposed in Scheme 12.
(21) Staubitz, A.; Robertson, A. P. M.; Manners, I. Chem. Rev. 2010,
110, 4079−4124.
(22) Stephens, F. H.; Pons, V.; Baker, R. T. Dalton Trans. 2007,
2613−2626.
(23) Smythe, N. C.; Gordon, J. C. Eur. J. Inorg. Chem. 2010, 509−
521.
(24) Sutton, A. D.; Burrell, A. K.; Dixon, D. A.; Garner, E. B., III;
Gordon, J. C.; Nakagawa, T.; Ott, K. C.; Robinson, P.; Vasiliu, M.
Science 2011, 331, 1426−1429.
(25) Wright, N. R. H.; Berkeley, E. R.; Alden, L. R.; Baker, R. T.;
Sneddon, L. G. Chem Commun. 2011, 47, 3177−3179.
(26) Jaska, C. A.; Manners, I. J. Am. Chem. Soc. 2004, 126, 2698−
2699.
ASSOCIATED CONTENT
■
S
* Supporting Information
Full experimental details and details of additional experiments.
This material is available free of charge via the Internet at
http://pubs.acs.org.
(27) Sloan, M. E.; Staubitz, A.; Lee, K.; Manners, I. Eur. J. Org. Chem.
2011, 2011, 672−675.
3061
dx.doi.org/10.1021/ja410129j | J. Am. Chem. Soc. 2014, 136, 3048−3064

Journal of the American Chemical Society
Article
(28) Yang, X.; Fox, T.; Berke, H. Chem. Commun. 2011, 47, 2053−
2055.
(29) Yang, X.; Fox, T.; Berke, H. Org. Biomol. Chem. 2012, 10, 852−
860.
(55) Luo, Y.; Ohno, K. Organometallics 2007, 26, 3597−3600.
(56) Pun, D.; Lobkovsky, E.; Chirik, P. J. Chem. Commun. 2007,
3297−3299.
(57) (a) Vogt, M.; de Bruin, B.; Berke, H.; Trincado, M.;
(30) Robertson, A. P. M.; Leitao, E. M.; Manners, I. J. Am. Chem. Soc.
2011, 133, 19322−19325.
́
Grutzmacher, H. Chem. Sci. 2011, 2, 723−727. (b) Garcia-Vivo, D.;
̈
Huergo, E.; Ruiz, M. A.; Travieso-Puente, R. Eur. J. Inorg. Chem. 2013,
4998−5008.
(31) Leitao, E. M.; Stubbs, N. E.; Robertson, A. P. M.; Helten, H.;
Cox, R. J.; Lloyd-Jones, G. C.; Manners, I. J. Am. Chem. Soc. 2012, 134,
16805−16816.
(58) Chapman, A. M.; Haddow, M. F.; Wass, D. F. J. Am. Chem. Soc.
2011, 133, 8826−8829.
(32) Yang, X.; Zhao, L.; Fox, T.; Wang, Z.-X.; Berke, H. Angew.
Chem., Int. Ed. 2010, 49, 2058−2062.
(59) Tang, C. Y.; Thompson, A. L.; Aldridge, S. J. Am. Chem. Soc.
2010, 132, 10578−10591.
(33) Staubitz, A.; Soto, A. P.; Manners, I. Angew. Chem., Int. Ed. 2008,
47, 6212−6215.
(60) Duman, S.; Ozkar, S. Int. J. Hydrogen Energ. 2013, 38, 10000−
10011.
(34) Dietrich, B. L.; Goldberg, K. I.; Heinekey, D. M.; Autrey, T.;
Linehan, J. C. Inorg. Chem. 2008, 47, 8583−8585.
(35) Staubitz, A.; Sloan, M. E.; Robertson, A. P. M.; Friedrich, A.;
(61) Zahmakiran, M.; Ozkar, S. Top. Catal. 2013, 56, 1171−1183.
(62) Kim, S.-K.; Han, W.-S.; Kim, T.-J.; Kim, T.-Y.; Nam, S. W.;
́
Mitoraj, M.; Piekos, L.; Michalak, A.; Hwang, S.-J.; Kang, S. O. J. Am.
Schneider, S.; Gates, P. J.; Schmedt auf der Gunne, J.; Manners, I. J.
̈
Chem. Soc. 2010, 132, 9954−9955.
Am. Chem. Soc. 2010, 132, 13332−13345.
(63) Chaplin, A. B.; Weller, A. S. Angew. Chem., Int. Ed. 2010, 49,
(36) Liu, Z.; Song, L.; Zhao, S.; Huang, J.; Ma, L.; Zhang, J.; Lou, J.;
Ajayan, P. M. Nano Lett. 2011, 11, 2032−2037.
581−584.
(64) Butera, V.; Russo, N.; Sicilia, E. Chem.Eur. J. 2011, 17,
(37) Ewing, W. C.; Marchione, A.; Himmelberger, D. W.; Carroll, P.
J.; Sneddon, L. G. J. Am. Chem. Soc. 2011, 133, 17093−17099.
(38) There are also many examples in which main group metals have
been used to dehydrocouple amine−borane adducts. See: (a) Liptrot,
D. L.; Hill, M. S.; Mahon, M. F.; MacDougall, D. J. Chem.Eur. J.
2010, 16, 8508−8515. (b) Hill, M. S.; Kocick-Koln, G.; Robinson, T.
P. Chem. Commun. 2010, 46, 7587−7589. (c) Spielmann, J.; Piesik, D.
F.-J.; Harder, S. Chem.Eur. J. 2010, 16, 8307−8318. (d) Spielmann,
J.; Harder, S. Dalton Trans. 2011, 40, 8314−8319. (e) Appelt, C.;
Slootweg, J. C.; Lammertsma, K.; Uhl, W. Angew. Chem., Int. Ed. 2013,
52, 4256−4259. (f) Hansmann, M. M.; Melen, R. L.; Wright, D. S.
Chem. Sci. 2011, 2, 1554−1559. (g) Crowley, H. J.; Holt, M. S.; Melen,
R. L.; Rawson, J. M.; Wright, D. S. Chem. Commun. 2011, 47, 2682−
2684. (h) Less, R. J.; Simmonds, H. R.; Dane, S. B. J.; Wright, D. S.
Dalton Trans. 2013, 42, 6337−6343.
14586−14592.
(65) Paul, A.; Musgrave, C. B. Angew. Chem., Int. Ed. 2007, 46, 8153−
8156.
́
(66) García-Vivo, D.; Huergo, E.; Ruiz, M. A.; Travieso-Puente, R.
Eur. J. Inorg. Chem. 2013, 28, 4998−5008.
(67) Dallanegra, R.; Chaplin, A. B.; Weller, A. S. Angew. Chem., Int.
Ed. 2009, 48, 6875−6878.
(68) Marziale, A. N.; Friedrich, A.; Klopsch, I.; Drees, M.; Celinski,
V.; Schmedt auf der Gunne, J.; Schneider, S. J. Am. Chem. Soc. 2013,
̈
135, 13342−13355.
̈ ̈
(69) Dinc,̧ M.; Metin, O.; Ozkar, S. Catal. Today 2012, 183, 10−16.
(70) Examples of iron-catalyzed dehydrogenation/dehydrocoupling
of other B−N containing substrates have been reported. For an
example of Fe-catalyzed dehydrogenation of boron−nitrogen hetero-
cycles, see: (a) Luo, W.; Campbell, P. G.; Zakharov, L. N.; Liu, S.-Y. J.
Am. Chem. Soc. 2011, 133, 19326−19329. Also, Fe nanoparticles have
been used for the hydrolytic dehydrogenation of 9 to yield borates.
See: (b) Aranishi, K.; Jiang, H.-L.; Akita, T.; Haruta, m.; Xu, Q. Nano
Res. 2011, 4, 1233−1241. (c) Qiu, F.; Li, L.; Wang, Y.; An, C.; Xu, C.;
Xu, Y.; Wang, Y.; Jiao, L.; Yuan, H. Int. J. Hydrogen Energy 2013, 38,
7291−7297. (d) Wang, H.-L.; Yan, J.-M.; Wang, Z.-L.; Jiang, Q. Int. J.
Hydrogen Energy 2012, 37, 10229−10235. (e) Qiu, F.; Li, L.; Liu, G.;
Wang, Y.; Wang, Y.; An, C.; Xu, Y.; Xu, C.; Wang, Y.; Jiao, L.; Yuan, H.
Int. J. Hydrogen Energy 2013, 38, 3241−3249.
(39) Kawano, Y.; Uruichi, M.; Shimoi, M.; Taki, S.; Kawaguchi, T.;
Kakizawa, T.; Ogino, H. J. Am. Chem. Soc. 2009, 131, 14946−14957.
(40) Kakizawa, T.; Kawano, Y.; Naganeyama, K.; Shimoi, M. Chem.
Lett. 2011, 40, 171−173.
(41) Jaska, C. A.; Temple, K.; Lough, A. J.; Manners, I. Chem.
Commun. 2001, 962−963.
(42) Jaska, C. A.; Temple, K.; Lough, A. J.; Manners, I. J. Am. Chem.
Soc. 2003, 125, 9424−9434.
(43) Clark, T. J.; Russell, C. A.; Manners, I. J. Am. Chem. Soc. 2006,
128, 9582−9583.
(71) Sewell, L. J.; Lloyd-Jones, G. C.; Weller, A. S. J. Am. Chem. Soc.
2012, 134, 3598−3610.
(44) Sloan, M. E.; Staubitz, A.; Clark, T. J.; Russell, C. A.; Lloyd-
Jones, G. C.; Manners, I. J. Am. Chem. Soc. 2010, 132, 3831−3841.
(45) Dallanegra, R.; Robertson, A. P. M.; Chaplin, A. B.; Manners, I.;
Weller, A. S. Chem. Commun. 2011, 47, 3763−3765.
(46) Denney, M. C.; Pons, V.; Hebden, T. J.; Heinekey, D. M.;
Goldberg, K. I. J. Am. Chem. Soc. 2006, 128, 12048−12049.
(47) Keaton, R. J.; Blacquiere, J. M.; Baker, R. T. J. Am. Chem. Soc.
2007, 129, 1844−1845.
(72) Stevens, C. J.; Dallanegra, R.; Chaplin, A. B.; Weller, A. S.;
MacGregor, S. A.; Ward, B.; McKay, D.; Alcaraz, G.; Sabo-Etienne, S.
Chem.Eur. J. 2011, 17, 3011−3020.
(73) Chen, Y.; Fulton, J. L.; Linehan, J. C.; Autrey, T. J. Am. Chem.
Soc. 2005, 127, 3254−3255.
(74) For other relevant studies, see: (a) Johnson, H. C; Robertson, A.
P. M.; Chaplin, A. B.; Sewell, L. J.; Thompson, A. L.; Haddow, M. F.;
Manners, I.; Weller, A. S. J. Am. Chem. Soc. 2011, 133, 11076−11078.
For a study with the related phosphine-borane adducts see
(b) Huertos, M. A.; Weller, A. S. Chem. Sci. 2013, 4, 1881−1888.
(75) It is noteworthy that pentamethylation of the Cp rings (as in
1*) led to a reduction in catalytic activity. Furthermore, it was also of
interest that Fe2(CO)9 (8) was found to be almost completely inactive.
See the Discussion section.
(48) Baker, R. T.; Gordon, J. C.; Hamilton, C. W.; Henson, N. J.; Lin,
P.-H.; Maguire, S.; Murugesu, M.; Scott, B. L.; Smythe, N. C. J. Am.
Chem. Soc. 2012, 134, 5598−5609.
(49) Sonnenberg, J. F.; Morris, R. H. ACS Catalysis 2013, 1092−
1102.
(50) Jiang, Y.; Blacque, O.; Fox, T.; Frech, C. M.; Berke, H.
Organometallics 2009, 28, 5493−5504.
(51) Robertson, A. P. M.; Suter, R.; Chabanne, L.; Whittell, G. R.;
Manners, I. Inorg. Chem. 2011, 50, 12680−12691.
(76) HB(NMe2)2 is observed during dehydrocoupling reactions, and
similar compounds (HB(NR2)2) have been characterized in other
research. For a recent paper detailing some examples, see: Helten, H.;
Robertson, A. P. M.; Staubitz, A.; Vance, J. R.; Haddow, M. F.;
Manners, I. Chem.Eur. J. 2012, 18, 4665−4680.
(52) Friedrich, A.; Drees, M.; Schneider, S. Chem.Eur. J. 2009, 15,
10339−10342.
(53) Vance, J. R.; Robertson, A. P. M.; Lee, K.; Manners, I. Chem.
Eur. J. 2011, 17, 4099−4103.
(77) Recently, Morris et al. reported that FeBr2 was found to
dehydrocouple H3N·BH3 (9) in the presence of KOtBu: see ref 49.
Sonnenberg, J. F.; Morris, R. H. ACS Catalysis 2013, 1092−1102.
(54) Helten, H.; Dutta, B.; Vance, J. R.; Sloan, M. E.; Haddow, M. F.;
Sproules, S.; Collison, D.; Whittell, G. R.; Lloyd-Jones, G. C.;
Manners, I. Angew. Chem., Int. Ed. 2013, 52, 437−440.
3062
dx.doi.org/10.1021/ja410129j | J. Am. Chem. Soc. 2014, 136, 3048−3064

Journal of the American Chemical Society
Article
(78) Noth, H.; Thomas, S. Eur. J. Inorg. Chem. 1999, 1999, 1373−
(102) Pan, X.; Philbin, C. E.; Castellani, M. P.; Tyler, D. R. Inorg.
Chem. 1988, 27, 671−676.
̈
1379.
(103) In an attempt to detect [CpFe(CO)(PMe3)2]+ by 31P NMR
spectroscopy, PMe3 was added to CpFe(CO)2I (2) under photolysis.
However, further investigation indicated that PMe3 also reacted with 2
under ambient conditions to give [CpFe(CO)(PMe3)2]+ and
[CpFe(PMe3)3]+. See the Supporting Information.
(79) Comprehensive Organometallic Chemistry, 2nd ed.; Shriver, D. F.
B., Michael, I., Ed.; Pergamon Press: New York, 1995; Vol. 7.
(80) Caspar, J. V.; Meyer, T. J. J. Am. Chem. Soc. 1980, 102, 7794−
7795.
(81) Meyer, T. J.; Caspar, J. V. Chem. Rev. 1985, 85, 187−218.
(82) Dixon, A. J.; George, M. W.; Hughes, C.; Poliakoff, M.; Turner,
J. J. J. Am. Chem. Soc. 1992, 114, 1719−1729.
(83) Douglas, T. M.; Chaplin, A. B.; Weller, A. S. J. Am. Chem. Soc.
2008, 130, 14432−14433.
(84) Douglas, T. M.; Chaplin, A. B.; Weller, A. S.; Yang, X.; Hall, M.
B. J. Am. Chem. Soc. 2009, 131, 15440−15456.
(85) Alcaraz, G.; Vendier, L.; Colt, E.; Sabo-Etienne, S. Angew. Chem.,
Int. Ed. 2010, 49, 918−920.
(86) Alcaraz, G.; Sabo-Etienne, S. Angew. Chem., Int. Ed. 2010, 49,
7170−7179.
(87) (a) Staubitz, A.; Robertson, A. P. M.; Sloan, M. E.; Manners, I.
Chem. Rev. 2010, 110, 4023−4078. (b) Stephens, F. H.; Baker, R. T.;
Matus, M. H.; Grant, D. J.; Dixon, D. A. Angew. Chem. Int. Ed. 2007,
46, 746−749.
(88) Hai-Shan, D.; Roberts, B. P. Tetrahedron Lett. 1992, 33, 6169−
6172.
(104) An atmosphere of CO will most likely suppress any loss of CO
from the Fe precatalyst, although the observed lack of conversion may
also be due to catalyst poisoning.
(105) CpFe(CO)2Cl (12) was used instead of CpFe(CO)2I (2)
because iodide is not covered by the basis set used (6-311G(d,p)).
Both complexes showed similar reactivity in toluene meaning that this
exchange was appropriate.
(106) See the Supporting Information for further details.
(107) [CpFe(CO)2]2 (1) has previously be shown to be active
dehydrocoupling catalyst towards a range of amine−borane adducts in
a preliminary publication. See ref 53.
(108) Zhang, S.; Brown, T. L. Organometallics 1992, 11, 4166−4173.
(109) Labinger, J. A.; Madhavan, S. J. Organomet. Chem. 1977, 134,
381−389.
(110) Hamilton, C. W.; Baker, R. T.; Staubitz, A.; Manners, I. Chem.
Soc. Rev. 2009, 38, 279−293.
(111) In a separate experiment, amine−borane 3 was stirred in neat
MeCN for 20 h at 20 °C and no reactivity was observed when the
reaction mixture was analyzed by 11B NMR spectroscopy.
(112) Babu Kalidindi, S.; Sanyal, U.; Jagirdar, B. R. ChemSusChem
2011, 4, 317−324.
(89) Dang, H.-S.; Roberts, B. P. J. Chem. Soc., Perkin Trans. 1993,
891−898.
(90) Bitterwolf, T. E. Coord. Chem. Rev. 2000, 206−207, 419−450.
(91) Photoirradiation of precatalyst 1 in THF, followed by addition
of amine−borane 3 led to no observable reaction over 4 h by 11B
NMR spectroscopy.
(113) (a) Yan, J.-M.; Zhang, X.-B.; Han, S.; Shioyama, H.; Xu, Q.
Angew. Chem., Int. Ed. 2008, 47, 2287−2289. (b) Clark, T. J.; Whittell,
G. R.; Manners, I. Inorg. Chem. 2007, 46, 7522−7527.
(92) Widegren, J. A.; Finke, R. G. J. Mol. Cat. A: Chem. 2003, 198,
317−341.
(114) See ref 95.
(93) Widegren, J. A.; Bennett, M. A.; Finke, R. G. J. Am. Chem. Soc.
2003, 125, 10301−10310.
(115) The distribution of particle sizes derived from DLS and TEM
may differ for a number of reasons. First, DLS is reliant on the
scattered light intensity from an ensemble of particles, which is
dominated by that arising from the larger species. TEM, however, has
no such bias and individual particles can be observed, but a large
number of measurements are needed for a meaningful average to be
obtained. Thus, particles of the size observed by TEM (10−30 nm)
may have been present in the DLS sample, albeit in a concentration
that was too small to be detected in the presence of larger species (148
nm). Conversely, as only a selection of the particles present in
colloidal solution collects on the substrate during TEM sample
preparation, larger particles may have drifted off the grid or
serendipitously gathered in regions that were not examined. Such
problems in identifying all the species present in a colloidal solution by
TEM will often be encountered when complex mixtures are present,
and this situation is exaggerated here due to the presence of additional
reactant and/or product. The evaporation of solvent may lead to the
crystallization of these compounds, which may then obscure objects of
interest, and also other drying effects, such as particle aggregation,
which prevent the images from truly representing the particles in
solution. Furthermore, the reducing nature of the electron beam and
the unavoidable exposure of the substrate to air on introduction to the
microscope serve to divorce the particles from their native environ-
ment and may result in structural changes on performing the TEM
experiment. DLS, on the other hand, can be performed under the
standard reaction conditions (excluding photoirradiation) and, there-
fore, represents a truer picture of the overall particle size distribution.
When the appropriate TEM control experiments are performed,
however, there is no doubt that both methods suggest that the reaction
mixture is heterogeneous in nature. Also see ref 96 and the Supporting
Information.
(94) Crabtree, R. H. Chem. Rev. 2012, 112, 1536−1554.
(95) Based on recent work by Amiens and co-workers, the Fe
nanoparticles are also anticipated to contain significant amounts of
boron (which is difficult to detect by EDX analysis). These workers
showed that treatment of Fe(II) salts with amine−boranes led to Fe
nanoparticles containing 15−25% B based on EXAFS and Mossbauer
̈
spectroscopy. Mechanisms of formation involving the reactions with
aminoborane intermediates or products at the nanoparticle nucleation
stage or with the Fe nanoparticles themselves were proposed. See:
Pellelier, F.; Ciuculescu, D.; Mattei, J.-G.; Lecante, P.; Casanove, M.-J.;
Schmitz-Antoniak, C.; Amiens, C. Chem.Eur. J. 2013, 19, 6021−
6026.
(96) DLS affords a translational diffusion coefficient for the
particle(s), which is then translated in to an effective hydrodynamic
size by means of the Stokes−Einstein equation. This relationship
assumes that the particles are non-interacting hard spheres diffusing
through viscous liquid.
(97) Njagi, E. C.; Huang, H.; Stafford, L.; Genuino, H.; Galindo, H.
M.; Collins, J. B.; Hoag, G. E.; Suib, S. L. Langmuir 2011, 27, 264−
271.
(98) It should be noted that the wavelength of the peak associated to
the surface plasmon resonance is directly related to the size of the
nanoparticles and the metal they contain.
(99) Alway, D. G.; Barnett, K. W. Inorg. Chem. 1978, 17, 2826−2831.
(100) Allen, D. M.; Cox, A.; Kemp, T. J.; Sultana, Q. Inorg. Chim.
Acta 1977, 21, 191−194.
(101) It should be noted that a third conceivable photoactivated
process would involve homolytic cleavage of the Fe−I bond to form a
CpFe(CO)2 radical and I•. This route is improbable, however, as
various studies have demonstrated this process to be unfavored.
Furthermore, the formation of CpFe(CO)2 radical would presumably
lead to the formation of 1, and under photoirradiation this would lead
to dehydrogenation of 3 via aminoborane 5, which is not observed in
this case. For further information on CpFe(CO)2 radicals, see refs 99
and 100.
(116) As suggested by a reviewer, filtration of the reaction mixture
prior to analysis by DLS could lead to aggregation. This could explain
the differences in sizes obtained by DLS and TEM. Further
preliminary experiments were performed to investigate the effect of
filtration on the size of Fe(0) nanoparticles observed. In the case of the
unfiltered samples, uninterpretable data was obtained. This is likely
3063
dx.doi.org/10.1021/ja410129j | J. Am. Chem. Soc. 2014, 136, 3048−3064

Journal of the American Chemical Society
Article
due to dust particles and other contaminants making it impossible to
detect the very small nanoparticles. TEM was also obtained, and
comparison between the unfiltered and filtered cases suggested that
some aggregation did take place in the latter case.
(117) Indeed, the absence of significant activity with Fe2(CO)9 (8), a
ready source of colloidal Fe, led us to suggest in our preliminary
communication that the catalytic dehydrocoupling of 3 by 1 under UV
irradiation was a homogeneous process. See ref 53, footnote 20.
(118) Glavee, G. N.; Klabunde, K. J.; Sorensen, C. M.; Hadjipanayis,
G. C. Inorg. Chem. 1995, 34, 28−35.
(119) Alexandridas, P. Chem. Eng. Technol. 2011, 34, 15−28.
(120) Virkutyte, J.; Varma, R. S. Chem. Sci. 2011, 2, 837−846.
(121) Margeat, O.; Dumestre, F.; Amiens, C.; Chaudret, B.; Lecante,
P.; Respaud, M. Prog. Solid State Chem. 2005, 33, 71−79.
(122) Boennemann, H.; Brinkmann, R.; Neiteler, P. Appl. Organo-
metal. Chem 1994, 8, 361−378.
(123) Boennemann, H.; Braun, G.; Brijoux, W.; Brinkmann, R.;
Schulze Tilling, A.; Seevogel, K.; Siepen, K. J. Organomet. Chem. 1996,
520, 143−162.
(124) It is again worth noting that treatment of 3 with anhydrous
FeCl2 (the precursor to 17) in either toluene or THF resulted in no
dehydrocoupling after 4 h when analysed by 11B NMR spectroscopy.
See Table 1.
(125) Bitterwolf, T. E.; Linehan, J. C.; Shade, J. E. Organometallics
2000, 19, 4915−4917.
(126) Robertson, A. P. M.; Suter, R.; Chabanne, L.; Whittell, G. R.;
Manners, I. Inorg. Chem. 2011, 50, 12680−12691.
(127) Hill, M. S.; Kociok-Kohn, G.; Robinson, T. P. Chem. Commun.
̈
2010, 46, 7587−7589.
(128) Beweries, T.; Hansen, S.; Kessler, M.; Klahn, M.; Rosenthal, U.
Dalton Trans. 2011, 40, 7689−7692.
(129) Jaska, C. A.; Manners, I. J. Am. Chem. Soc. 2004, 126, 9776−
9785.
(130) Jaska, C. A.; Manners, I. J. Am. Chem. Soc. 2004, 126, 1334−
1335.
(131) Fulton, J. L.; Linehan, J. C.; Autrey, T.; Balasubramanian, M.;
Chen, Y.; Szymczak, N. K. J. Am. Chem. Soc. 2007, 129, 11936−11949.
̈
̂
(132) Zahmakiran, M.; Ozkar, S. Inorg. Chem. 2009, 48, 8955−8964.
̈
̈
̂
(133) Metin, O.; Mazumder, V.; Ozkar, S.; Sun, S. J. Am. Chem. Soc.
2010, 132, 1468−1469.
̈
̈
(134) Metin, O.; Suman, S.; Dinc,̧ M.; Ozkar, S. J. Phys. Chem. C
̂
2011, 115, 10736−10743.
(135) Although the heterogeneous catalysts appear to operate
through mechanisms that form 5 rather than 6 based on the current
literature, homogeneous catalysts seem to be more flexible and can
participate in mechanisms that involve either species as an
intermediate.
(136) It should be further noted that various catalysts have been
shown to dehydrocouple amine−borane 3 by both metal-mediated
mechanisms. For example, reaction of 3 with Schneider’s homoge-
neous Ru catalyst led to dehydrogenation to afford 4. Both
aminoborane 5 and linear diborazane 6 were detected by 11B NMR
spectroscopy, although 6 was seemingly the favored intermediate. See
ref 52.
(137) We have also shown that skeletal Fe is inactive as a precatalyst
for the dehydrogenation of 3; see ref 51.
(138) Bayram, E.; Linehan, J. C.; Fulton, J. L.; Roberts, J. A. S.;
̈
Szymczak, N. K.; Smurthwaite, T. D.; Ozkar, S.; Balasubramanian, M.;
Finke, R. G. J. Am. Chem. Soc. 2011, 133, 18889−18902.
(139) See ref 95.
3064
dx.doi.org/10.1021/ja410129j | J. Am. Chem. Soc. 2014, 136, 3048−3064