Iron complex-catalyzed ammonia-borane dehydrogenation. A potential route toward B-N-containing polymer motifs using earth-abundant metal catalysts

Article abstract of DOI:10.1021/ja210542r

Ammonia-borane (NH₃BH₃, AB) has garnered interest as a hydrogen storage material due to its high weight percent hydrogen content and ease of H₂ release relative to metal hydrides. As a consequence of dehydrogenation, B-N-containing oligomeric/polymeric materials are formed. The ability to control this process and dictate the identity of the generated polymer opens up the possibility of the targeted synthesis of new materials. While precious metals have been used in this regard, the ability to construct such materials using earth-abundant metals such as Fe presents a more economical approach. Four Fe complexes containing amido and phosphine supporting ligands were synthesized, and their reactivity with AB was examined. Three-coordinate Fe(PCy₃)[N(SiMe₃)₂]₂ (1) and four-coordinate Fe(DEPE)[N(SiMe₃)₂]₂ (2) yield a mixture of (NH₂BH₂)_n and (NHBH)_n products with up to 1.7 equiv of H₂ released per AB but cannot be recycled (DEPE = 1,2-bis(diethylphosphino)ethane). In contrast, Fe supported by a bidentate P-N ligand (4) can be used in a second cycle to afford a similar product mixture. Intriguingly, the symmetric analogue of 4 (Fe(N-N)(P-P), 3), only generates (NH₂BH₂)_n and does so in minutes at room temperature. This marked difference in reactivity may be the result of the chemistry of Fe(II) vs Fe(0).

Full text of DOI:10.1021/ja210542r

Article
pubs.acs.org/JACS
Iron Complex-Catalyzed Ammonia−Borane Dehydrogenation.
A Potential Route toward B−N-Containing Polymer Motifs Using
Earth-Abundant Metal Catalysts
R. Tom Baker,*,† John C. Gordon,*,‡ Charles W. Hamilton,‡,∥ Neil J. Henson,§ Po-Heng Lin,†
Steven Maguire,† Muralee Murugesu,† Brian L. Scott,⊥ and Nathan C. Smythe‡
†Department of Chemistry and Centre for Catalysis Research and Innovation, University of Ottawa, Ottawa, Ontario Canada
K1N 6N5
§
⊥
‡Chemistry Division, Theoretical Division, and Materials Physics and Applications Division, Los Alamos National Laboratory,
Los Alamos, New Mexico 87545, United States
S
* Supporting Information
ABSTRACT: Ammonia−borane (NH3BH3, AB) has garnered interest as a
hydrogen storage material due to its high weight percent hydrogen content and
ease of H2 release relative to metal hydrides. As a consequence of
dehydrogenation, B−N-containing oligomeric/polymeric materials are formed.
The ability to control this process and dictate the identity of the generated
polymer opens up the possibility of the targeted synthesis of new materials. While precious metals have been used in this regard,
the ability to construct such materials using earth-abundant metals such as Fe presents a more economical approach. Four Fe
complexes containing amido and phosphine supporting ligands were synthesized, and their reactivity with AB was examined.
Three-coordinate Fe(PCy3)[N(SiMe3)2]2 (1) and four-coordinate Fe(DEPE)[N(SiMe3)2]2 (2) yield a mixture of (NH2BH2)n
and (NHBH)n products with up to 1.7 equiv of H2 released per AB but cannot be recycled (DEPE = 1,2-bis(diethyl-
phosphino)ethane). In contrast, Fe supported by a bidentate P−N ligand (4) can be used in a second cycle to afford a similar
product mixture. Intriguingly, the symmetric analogue of 4 (Fe(N−N)(P−P), 3), only generates (NH2BH2)n and does so in
minutes at room temperature. This marked difference in reactivity may be the result of the chemistry of Fe(II) vs Fe(0).
characterized, the identity and controlled synthesis of higher
order oligomers and polymers is more nebulous.19 This is in
stark contrast to the mature field of polyolefin chemistry, where
the emphasis is on fine-tuning control over parameters such as
tacticity and polydispersity. Manners and co-workers have
recently demonstrated the utility of employing metal-catalyzed
dehydrocoupling for the synthesis of B−N and B−P materials
from amine− and phosphine−boranes,20−22 and this area would
also benefit by the development of efficient earth-abundant metal
catalysts.
Metal-based catalysts offer the potential to tune both rate and
extent of hydrogen (H2) release from AB. Systems employing
transition metals have provided both the fastest rates (in the
case of Ir,23 Ru,24,25 and Pd26) and the largest extent of H2 loss
(in the case of Ni27 and Ru15,16) yet reported for AB
dehydrogenation under reasonable reaction conditions. Further
tunability has been reported through the use of ionic liquids as
the reaction medium.28 Unfortunately, many of these systems
use relatively expensive metals (Rh,29 Ir, Ru, and Pd) or suffer
from catalyst instability under the reaction conditions (Ni, Pd).
We have recently investigated the synthesis, characterization,
and modes of reactivity of Fe-based catalysts, as these have the
potential to be much more cost-effective toward scale-up on a
INTRODUCTION
■
Hydrogen (H2) has been proposed as an alternative power
source, both as a transportation fuel and, through its use in a
PEM fuel cell, as a replacement for batteries for portable
electricity.1 In the former case, the U.S. Department of Energy
(DOE) has set performance targets such that a vehicle could
travel approximately 300 miles on a single charge.2,3
The details of the full spectrum of approaches funded by the
DOE have been covered elsewhere,4−14 but one approach
investigated within the DOE Chemical Hydrogen Storage Center
of Excellence is the use of ammonia−borane (NH3BH3, AB) as a
hydrogen carrier. AB contains 6.5, 13.1, and 19.6 wt % H2 for the
loss of the first, second, and third equivalents of H2, respectively.
As in all chemical storage approaches, however, the ultimate
utility of AB depends upon finding ways to efficiently remove15,16
and replace H2.17 With an arena as large as the transportation
sector, this also requires materials that are both cheap and
abundant.
Another important application related to dehydrogenation of
amine−boranes is the development of new BN-containing
polymeric materials. For example, (H2NBH2)n can be thought
of as an inorganic analogue of polyethylene and (HNBH)n as
an inorganic analogue of polyacetylene. While cyclic com-
pounds such as (HNBH)3 (borazine, a benzene analogue) and
(H2NBH2)3 (cyclotriborazane, CTB, a cyclohexane analogue)
Received: November 17, 2011
Published: March 19, 2012
18
as well as the n-butane analogue H3NBH2NH2BH3 are well
© 2012 American Chemical Society
5598
dx.doi.org/10.1021/ja210542r | J. Am. Chem. Soc. 2012, 134, 5598−5609

Journal of the American Chemical Society
Article
30
global scale. FeH(CH2PMe2)(PMe3)3 has been shown to
Scheme 2. AB Dehydrogenation Using a Bifunctional
exhibit mild reactivity at 20 °C, while Manners et al. recently
reported the formation of borazine and B-(cyclodiborazanyl)-
aminoborohydride (BCDB) from the photolytic activation of
[CpFe(CO)2]2.31 We describe herein some of our recent
results toward developing efficient Fe-based AB dehydrogena-
tion catalysts and their potential utility for aminoborane poly-
mer synthesis.
Catalyst
Bifunctional Catalysis for Ammonia−Borane Dehy-
drogenation. Developing effective catalysts for AB dehydro-
genation requires control over two fundamental reaction
pathways (Scheme 1). The initial dehydrogenation of AB and
other reaction pathways are operating. In contrast, recent work
by Williams et al.16 has identified an efficient bifunctional Ru
catalyst that affords exclusively borazine and PB at 80 °C.
We now report the preparation and characterization of amido
iron phosphine complexes and their evaluation as potential
bifunctional catalysts for AB dehydrogenation.
Scheme 1. Divergent Reaction Pathways Lead to Different
AB Dehydrogenation Products32
RESULTS
■
Generation of Amido Iron Complexes and Prelimi-
nary Reactivity with Ammonia−Borane. As might be
expected, treatment of Fe[N(SiMe3)2]2 (A)41 with a THF
solution of AB resulted in immediate formation of a dark
precipitate, presumably containing iron metal. Addition of PCy3
to A as a stabilizing agent yielded crude material (later shown
to contain 1, Scheme 3) that was capable of generating greater
than 1 equiv of H2 per AB, but the reaction required heating
which again led to formation of black precipitate during the
course of the reaction. In an attempt to improve this stability
further, we employed a chelating phosphine in the form of 1,2-
bis(diethylphosphino)ethane (DEPE) (2, Scheme 3). In this
case, there was again significant AB dehydrogenation at 60 °C,
and darkening of the reaction solution was accompanied by
formation of the bis(borane) adduct of DEPE and the
protonated amido ligand NH(SiMe3)2, likely a poor ligand
for iron. This deficiency was addressed by treatment of A with
N,N′-diphenyl-1,2-ethylenediamine [(PhHNCH2)2], which af-
forded a blue paramagnetic material (B). While B and excess
AB exhibited initial gas evolution, formation of a black
precipitate limited conversion to less than two turnovers.
Reaction of B with 1,2-bis(dicyclohexylphosphino)ethane
(DCPE), gave a red complex, later identified as 3 (Scheme 3).
Complex 3 exhibited rapid dehydrogenation of AB at room
temperature, but was not able to be recycled with a second
batch of AB due to decomposition through ligand loss (vide
infra). In order to protect against ligand (phosphine) loss from
3, its asymmetrically bridged cousin was developed (4, Scheme 3).
While 4 is more robust than 1, its reactivity is markedly different
despite the structural similarities (vide infra).
the release of 1 equiv of H2 forms aminoborane (H2NBH2)
which is unstable and reacts with further AB in two reported
ways depending on the nature of the catalyst employed.32 In
some cases, such as the aforementioned Ir catalyst, exclusive
formation of insoluble (H2NBH2)n corresponds to rapid release
of a single equivalent of H2 per molecule of AB at room
temperature (lower pathway, Scheme 1).21,23 Several catalysts,
based on Rh and Ni, as well as Williams’s bifunctional Ru
catalyst, however, afford greater than 2 equiv of H2 per mole-
cule of AB.16,27,29 The Ni-based catalyst is reported to yield
almost exclusively cross-linked borazine, i.e. polyborazylene
(PB), and greater than 2.5 equiv of H2 per molecule of AB.
These reactions are typically slower than those with the Ir- and
other Ru-based15,16,24,25 catalysts and require heating and/
or longer reaction times. Attempts to further dehydrogenate the
colorless, insoluble, (H2NBH2)n product were unsuccessful
under reasonable conditions with any of the above catalysts,
confirming that divergent reaction pathways are in operation.
While the intimate details of the different mechanisms of
dehydro-oligomerization are not yet fully understood, a key
step that differentiates between these pathways was recently
identified via the use of an aminoborane trapping agent,
cyclohexene.32,33 Trapping of aminoborane in the form of
Cy2BNH2 (Cy = cyclohexyl) occurs predominantly in catalyzed
reactions exhibiting greater extents of H2 release, suggesting
that efficient release of aminoborane from the metal center
(presumably forming a hydrogen or dihydride complex; upper
pathway, Scheme 1), affords borazine and PB, while complexes
that bind aminoborane form singly dehydrogenated (H2NBH2)n
(lower pathway, Scheme 1). Consistent with this hypothesis, use
of the Ir catalyst at higher temperatures afforded some borazine
and PB presumably due to some dissociation of aminoborane
from the metal center. Moreover, computational studies34−36
indicated that the key BN-ethylcyclobutane intermediate, BCDB,
results from intermolecular oligomerization of aminoborane via
the novel H2NBHNH2BH3 intermediate, accounting for the
significant rate decrease on dilution.37
Synthesis of Complexes 1−4. X-ray crystallographic data
for compounds 1−4 are summarized in Table 1.
Complex 1. The most straightforward synthesis of 1 was
accomplished by addition of FeCl2 to a THF solution of PCy3
(presumably forming FeCl2(PCy3) in situ, which has some
solubility in THF). Subsequent addition of KN(SiMe3)2
resulted in darkening of the reaction mixture. The pentane-
soluble components of this mixture yielded a dark solution
from which 1 could be crystallized. The use of other solvents,
other iron halides, or shorter times between PCy3 and base
addition generally resulted in lower yields or no product
formation (with a concomitant increase in the formation of black
precipitate). When isolated cleanly (a process that generally
These observations point to the importance of bifunctional
catalysts38−40 in metal-catalyzed AB dehydrogenation as
aminoborane is precluded from binding to the metal (Scheme 2).
While initial bifunctional systems afforded active Ru catalysts,24,25
predominant formation of insoluble (BH2NH2)n suggests that
5599
dx.doi.org/10.1021/ja210542r | J. Am. Chem. Soc. 2012, 134, 5598−5609

Journal of the American Chemical Society
Article
Scheme 3. Optimized Syntheses of Complexes 1−4 and Their AB Dehydrogenation Selectivity
Table 1. X-ray Crystallographic Structural Parameters for Complexes 1−4
142
242
3
4
formula
fw (g/mol)
T (K)
C30H69FeN2PSi4
657.05
C22H60FeN2P2Si4
582.87
C44H70FeN2OP2
760.81
C40H62FeN2P2
688.71
120(1)
200(2)
120(1)
140(1)
crystal system
space group
a (Å)
monoclinic
P21/n
tetragonal
P41
monoclinic
P21/c
monoclinic
P21/n
12.0279(9)
15.7960(12)
20.3153(16)
90
11.9184(3)
11.9184(3)
24.2844(8)
90
11.677(3)
22.210(5)
16.123(4)
90
11.443(13)
20.06(2)
16.304(19)
90
b (Å)
c (Å)
α (°)
β (°)
γ (°)
V (Å3)
90.3290(10)
90
90
100.869(3)
90
99.697(19)
90
90
3859.7(5)
4
3449.56(17)
4
4106.4(16)
4
3690(7)
4
Z
GoF
1.124
1.042
1.089
0.941
R [I > 2σ(I)] (%)
R1 = 3.9
wR2 = 8.8
R1 = 6.1
wR2 = 9.6
R1 = 2.9
wR2 = 5.9
R1 = 3.7
wR2 = 6.1
R1 = 7.8
wR2 = 18.1
R1 = 16.9
wR2 = 22.0
R1 = 7.3
wR2 = 12.8
R1 = 22.1
wR2 = 18.4
R (all data) (%)
requires 3−4 crystallizations from cold pentane), 1 is essentially
colorless with only a very slight greenish-blue hue. While it is
unclear if this slight coloring is the result of a small amount of
impurity or an inherent property of the molecule, we tend to
favor the former explanation as multiple crystallizations decrease
the intensity of the color. 1 is highly air-sensitive and will
scavenge trace oxygen to form an orange solid. The relatively low
isolated yield of 1 is a result of its high solubility in pentane and
the need for multiple crystallization steps.
The molecular structure of 142 (Figure 1) determined by
single-crystal X-ray diffraction reveals a 3-coordinate Fe center
with close to trigonal planar geometry. This is shown by the
N(2)−N(1)−P(1)−Fe(1) dihedral angle of ca. 1.6°, which
puts Fe at 0.03 Å out of the N(1)−N(2)−P(1) plane. At
128.46(7)°, the N(1)−Fe(1)−N(2) angle is slightly greater
than 120°, creating a small distortion as seen by the N(1)−
Fe(1)−P(1) and N(2)−Fe(1)−P(1) angles of 115.16(5)° and
116.31(5)°, respectively. The Fe−N bonds are 1.9496(16) and
Figure 1. Ball-and-stick representation of the crystallographically
determined molecular structure of 1.42 Fe (orange), N (blue), Si
(pink), and P (orchid) with hydrogen atoms omitted.
5600
dx.doi.org/10.1021/ja210542r | J. Am. Chem. Soc. 2012, 134, 5598−5609

Journal of the American Chemical Society
Article
1.9503(16) Å for Fe−N(1) and Fe−N(2), and the Fe−P bond
is 2.5167(6) Å. Complex 1 is paramagnetic, with a χT of
5.16 cm3·K/mol (300 K), indicating that it is high spin, although
this is much larger than the expected spin-only value. Our expla-
nation for the source of this behavior is reported elsewhere.42
Complex 2. 2 was synthesized by in situ formation of
Fe[N(SiMe3)2]2 from FeBr2 and KN(SiMe3)2. The brownish
solid that was collected after subsequent addition of DEPE was
extracted with hexanes and crystallized to give the complex
illustrated in Figure 2.42 The Fe−N(1) and Fe−N(2) bond
Figure 3. Crystallographically determined molecular structure of 3.
Hydrogen atoms and solvent are omitted for clarity, and thermal
ellipsoids are shown at 50% probability level.
Scheme 4. Preparation of the HNP Ligand Used in the
Synthesis of 4
Figure 2. Ball-and-stick representation of the crystallographically
determined molecular structure of 2.42 Fe (orange), N (blue), Si
(pink), and P (orchid) with hydrogen atoms omitted.
lengths of 1.9949(15) and 1.9886(15) Å are slightly longer than
the corresponding bonds in 1, as are the Fe−P(1) and Fe−P(2)
lengths of 2.5965(5) and 2.5878(5) Å. The bond angles result
in a distorted tetrahedron with the P(1)−Fe−P(2) angle
(78.454(17)°) significantly less than the N(1)−Fe−N(2) angle
(120.93(6)°). Compared to 1, N(1)−Fe−N(2) is slightly more
acute. Significant twisting of the tetrahedron is shown by the
angle pairs of N(1)−Fe−P(1) and N(2)−Fe−P(2) (129.06(5)°
and 126.22(5)°) compared to N(1)−Fe−P(2) and N(2)−Fe−
P(1) (99.24(5)° and 98.33(5)°). 2 is high spin with a χT value
(300 K) of 3.20 cm3·K/mol.
Complex 3. The direct synthesis of 3 was accomplished by
addition of [(PhHNCH2)2] to a solution of FeCl2(DCPE) in
toluene or THF to form a yellow solution, followed by
deprotonation with KN(SiMe3)2. The resultant red solution
yielded an air-sensitive, dark red crystalline product upon
isolation by addition of pentane and collection of the resulting
precipitate. While 3 is quite soluble in toluene and ethereal
solvents, it exhibits limited solubility in aliphatic solvents. 3 is
paramagnetic with a χT value (300 K) of 2.71 cm3·K/mol, close
to the spin-only value for a high-spin, tetrahedral d6 ion. The
molecular structure (Figure 3) reveals it to be a pseudo-
tetrahedral complex with an average N(x)−Fe−P(x) angle of ca.
121.3°, but N(1)−Fe−N(2) and P(1)−Fe−P(2) angles of
91.13(18)° and 84.30(6)°, respectively. The Fe−N(1) and Fe−
N(2) bond lengths are 1.923(4) and 1.938(5) Å, while the Fe−
P(1) and Fe−P(2) bond lengths are 2.4015(17) and
2.3974(17) Å.
reaction of N-(2-chloroethyl)-benzenamine with LiPCy2 in
THF yielded HNP on a small scale, larger scale reactions
showed significant byproduct formation as a result of the
unprotected N−H bond. Attempts at protecting this position
using Me3SiCl were unsuccessful as the reaction did not go to
completion; however, substitution with Me3SiBr in this step
gave essentially quantitative conversion. The HNP ligand is a
colorless solid that is soluble in common organic solvents. A 2
equiv portion of HNP was used to solubilize FeCl2 in THF,
followed by deprotonation with KN(SiMe3)2. The resulting
product could be isolated as an air-sensitive orange solid (4)
from toluene. Complex 4 is paramagnetic with a χT of
3.88 cm3·K/mol (300 K) significantly larger than 3, despite the
gross similarity of the ligand environments (Figure 4). The Fe−
N(1) and Fe−N(2) bond lengths are 1.959(5) and 1.961(6) Å,
and the corresponding Fe−P bonds are 2.422(3) and 2.423(3) Å.
These are slightly longer than what is seen in 3. The internal
N(1)−Fe−P(1) and N(2)−Fe−P(2) angles of 85.85(18)° and
84.02(18)° are closest to the P−Fe−P angle in 3. Of the other
angles in the pseudotetrahedron, the N(2)−Fe−N(1) angle at
136.1(2)° stands out as significantly different than the N(2)−
Fe−P(1), N(1)−Fe−P(2), and P(1)−Fe−P(2) angles of
119.95(17)°, 117.53(17)°, and 116.55(10)°. This represents a
spread among the external angles (difference between the
largest and smallest) of 19.6° vs 1.6° in 3.
Reactivity of Complexes 1−4 with AB. Addition of AB
to a THF or diglyme solution of complex 1 resulted in an
immediate darkening and some gas release. However, AB was
Complex 4. The synthetic route to the HNP ligand required
for the synthesis of 4 is summarized in Scheme 4. While direct
5601
dx.doi.org/10.1021/ja210542r | J. Am. Chem. Soc. 2012, 134, 5598−5609

Journal of the American Chemical Society
Article
Figure 6. 11B NMR spectra of AB dehydrogenation using 5 mol % 2 at
60 °C for 4 h with added NaBPh4 (5) as an internal standard (1H
decoupled spectrum on top. Products include borazine (1), PB (3),
BDCB (4), CTB (6), and phosphine−borane (7). The resonance
assigned as 2 and shoulder on resonance assigned to 4 may be due to
H2NBHNH2BH3.35
Figure 4. Crystallographically determined molecular structure of 4
with hydrogen atoms omitted for clarity. Thermal ellipsoids are shown
at 50% probability level.
HN(SiMe3)2, significant darkening of the solution, and for-
mation of uncharacterized diamagnetic Fe phosphine complexes
(Figure S10).
not completely consumed unless heat (60 °C) was applied, and
additional PCy3 was added. Without additional PCy3, the active
species decomposed into a black precipitate before all the AB
was consumed. In this regard, by adding 0, 1, or 2 equiv or an
excess (i.e., 1 equiv per mole of AB) of PCy3, it was observed
that an additional 2 equiv of PCy3 per Fe center (3 total P/Fe)
was the optimal amount; 1 equiv did not prevent decom-
position, and excess PCy3 yielded a greater quantity of the
corresponding phosphine−borane. It is readily apparent from
the 11B NMR data that higher order dehydrogenation products
(i.e., polyborazylene) were formed (Scheme 3 and Figure 5).
The addition of ca. 20 equiv of AB to a THF or diglyme
solution of 3 resulted in an immediate darkening in color. After
a short induction period (ca. 2 min, concentration dependent),
rapid effervescence was observed. By 11B NMR spectroscopy,
the AB was completely consumed in 15 min. If the concen-
tration of 1 was lowered to 2 mol %, complete consumption of
AB was observed after 2 h. An insoluble white precipitate
formed during the course of these reactions, and filtration of
the reaction mixture after completion gives a solution that does
not contain any boron containing species (beyond trace
materials in the baseline) by 11B NMR spectroscopy (Figure 7).
Figure 5. 11B NMR spectrum (baseline corrected) of the dehydroge-
nation of AB using 1 plus 2 equiv of PCy3 at 60 °C with added
NaBPh4 as an internal standard shows PB (ca. 20−35 ppm), NaBPh4
(−7.29 ppm), and PCy3BH3 (ca. −44 ppm).
Figure 7. 11B NMR spectrum of AB dehydrogenation using 3. This
shows (added) NaBPh4 as essentially the only soluble boron-
containing species remaining after AB consumption.
Gas buret measurements carried out at 60 °C indicate that up
to 1.7 equiv of H2 were released over a period of hours (Figure S8)
which is consistent with calibrated (internal NaBPh4) 11B NMR
spectral data, suggesting that ca. 30−50% insoluble PAB is also
formed in this reaction.
AB dehydrogenation results using 5 mol % 2 were similar to
those of 1 in terms of both rate and the array of dehydro-
genation products formed. After 4 h at 60 °C the 11B NMR
spectrum shows a typical mixture of soluble dehydrogenation
products (Figure 6). At 60 °C, the reaction is complete in 20 h
with formation of PB, PAB, and phosphine−borane (Figure S9).
Decomposition of complex 2 is evidenced by formation of
This indicates selective generation of poly(aminoborane) (PAB),
which was confirmed by increasing the scale of the reaction and
isolation of the off-white precipitate. Based on consistency with
the FTIR spectroscopic data (Figure S7) of [NH2BH2]n
obtained from IrH2(POCOP) by Goldberg et al. and recently
elucidated by Manners et al.,21 this product is assigned as PAB. 3
decomposes during the course of polymerization as can be
seen by the appearance of multiple 31P NMR peaks (Figure 8),
1
a suspected hydride peak in the H NMR at ca. −18 ppm, and
that the reaction mixture is unreactive to further AB addition.
5602
dx.doi.org/10.1021/ja210542r | J. Am. Chem. Soc. 2012, 134, 5598−5609

Journal of the American Chemical Society
Article
Figure 9. Space-filling models of 1 (left), 2 (middle), and 3 (right).
supported this interpretation, and Schneider reported kinetic
isotope effects that were consistent with bifunctional activation.
Nonetheless, formation of insoluble polyaminoborane suggests
that initially formed aminoborane remains bound to the metal, and
Schneider even reported such a complex in which the boron is
coordinated to the amido nitrogen atom (Figure 10).50,51 In
Figure 8. 31P NMR spectra of 3 + AB in THF showing the
decomposition of 3 over time: 0 min (top), 10 min (middle), and ca.
20 h (bottom).
Addition of AB (20 equiv) to a diglyme or THF solution of 4
at room temperature causes a darkening of the solution and shows
a mixture of AB along with singly and doubly dehydrogenated
products (approximately 0.8 equiv of H2 total with minimal
phosphine−borane formation) after ca. 20 h. Increasing the
temperature to 60 °C gives a mixture of insoluble PAB and soluble
PB and singly dehydrogenated products along with unreacted AB
and phosphine−borane after ca. 3 h. Increasing the reaction time
to 20 h allowed for the complete consumption of AB, with only
PB and phosphine−borane present in solution (Figure S12). The
amount of H2 produced was ca. 1.3 equiv. While some decom-
position was observed per the formation of black precipitate over
the course of dehydrogenation and evidenced by the presence of
phosphine−borane, the catalytically active species remained active
for a second cycle. When an additional 20 equiv of AB was added
and the reaction mixture was heated to 60 °C, the AB 11B NMR
signal disappeared after 24 h with a concomitant increase in the
quantity of PB (ca. 1.5 equiv of H2, 46% PB, Figure S13). This is
in contrast to 3 where reactivity is lost after the first cycle, and 1
where complete decomposition to bulk iron is eventually
observed. Thus, despite its structural similarity to 3, the reactivity
of 4 with AB is more akin to that of 1 and 2. Some possible
reasons for this are discussed in the next section.
Figure 10. Structure of Ru−N-coordinated aminoborane reported by
Schneider et al.50
short, a necessary implication of the formation of PAB is that the
NH2BH2 remains bound to the metal center during propagation.
Similar conclusions have recently been reached in a compre-
hensive mechanistic study of dimethylamine−borane dehydroge-
nation catalyzed by bis(phosphine) Rh cations.52
b. Possible Reaction Pathways for 1 and 2. The reactivity
of 1 toward AB is indicative of a precatalyst that eventually
undergoes decomposition. We believe that darkening of color
upon AB addition to 1 is the result of its reduction to a
phosphine-solubilized Fe(0) species (vide infra) and loss of the
poor ligand HN(SiMe3)2. The importance of the presence of
PCy3 is illustrated by the effect of adding additional phosphine
to the reaction between 1 and ca. 20 equiv of AB. At 60 °C and
after long reaction time (overnight) to ensure the reaction is
over, unreacted AB remains in solution. Adding 1 equiv of PCy3
(relative to 1) decreases the amount of residual AB, while
addition of 2 equiv (relative to 1) of PCy3 results in complete
consumption (Figure 5). However, addition of excess of PCy3
(1:1 AB:PCy3) decreased PB yield and increased the quantity
of PCy3BH3 by 1.5−2 times that of the case when 2 equiv of
PCy3 per Fe center are used while decreasing the formation of
black precipitate. The implication of this is that the excess
phosphine is stabilizing the catalytically active species.
DISCUSSION
■
a. Selectivity Differences in Iron Complex-Catalyzed
AB Dehydrogenation. Considerations for Initiation and
Propagation. Compared to the Ru complexes described by
Fagnou et al.24 and Schneider et al.,25 the PNNP-Fe transfer
hydrogenation catalysts described by Morris et al.,43−47 and the
PNNP-Ir complexes described by Gao et al.,48,49 1−4 are much
more sterically hindered as shown by their space-filling models
(Figure 9) and supported by initial DFT calculations carried
out on 1 that showed that there was no viable approach for AB
without the ligand sphere around 1 relaxing in some manner.
Thus, any explanation for the reactivity of these molecules with
AB requires accounting for this fact.
Upon sitting for weeks, the stoichiometric reaction between
1 and AB at room temperature yielded dark crystals for which
an X-ray structure indicates an Fe9 cluster (Figure 11). While the
data suffered from significant disorder and attempts to resynthesize
the material for a more detailed analysis were unsuccessful, we
believe that this structure provides some insight into what is
happening upon AB addition. The cluster itself consists of nine
Fe(0) atoms arranged as alternating triangles capped at the ends
with PCy3 molecules. There is insufficient room for any more than
six PCy3 ligands, and although the X-ray data cannot rule out the
presence of hydrides, their presence has not been observed upon
The AB dehydrogenation activity of the potentially bifunc-
tional Ru complexes reported by Fagnou and Schneider were
proposed to proceed in a manner analogous to that of Noyori
transfer hydrogenation (i.e., bifunctional activation of AB,
Scheme 2). DFT calculations reported by Fagnou et al.
1
AB addition to 1 by H NMR.
5603
dx.doi.org/10.1021/ja210542r | J. Am. Chem. Soc. 2012, 134, 5598−5609

Journal of the American Chemical Society
Article
Table 2. Calculated and Experimental Bond Lengths for
Compounds 1 and 2 Showing the Effect of Spin State and
Change in Oxidation State of the Iron Center (Spin States:
Fe(II) low spin, Singlet; Fe(II) High Spin, Quintet; Fe(0)
High Spin, Triplet)
Fe−P (Å)
Fe−N (Å)
compound 1
experiment
2.62
2.58
2.32
2.32
2.60
2.63
2.26
2.33
1.93
1.95
1.85
2.10
1.99
1.98
1.99
2.05
Fe(II), high spin
Fe(II), low spin
Fe(0) high spin
experiment
compound 2
Fe(II), high spin
Fe(II), low spin
Fe(0), high spin
much stronger ligand than HN(SiMe3)2. The induction period
observed upon AB addition to a solution of 3 would also be
consistent with the latter’s role as a precatalyst. The active
species formed from 3 decomposes during the course of
dehydrogenation (Figure 8), presumably at least partially into
iron metal based on the formation of black precipitate. This
decomposition is also reflected by the lack of observed de-
hydrogenation when a second 20 equiv of AB are added. While
this makes it impossible to attempt to isolate the active form of
3, 31P NMR data indicate that loss of the chelating phosphine is
occurring, while 11B NMR shows that this does not result in
formation of phosphine−borane.56 Thus it is possible that
reaction of 3 with AB is preceded by dissociation of one of the
chelating phosphine arms. The activity of 3 toward AB is
consistent with the divergent pathway proposal of Baker et al.
(vide supra) as addition of excess cyclohexene to the reaction
mixture does not lead to any trapping of NH2BH2 (Figure S6).
In an attempt to shed additional light upon the operation of
catalyst 3, we investigated its reactivity with 10 equiv of
dimethylamine−borane (DMAB). After 12 h at 20 °C, most of
the DMAB was converted into a mixture of the cyclic
dimethylaminoborane dimer, the diaminoborane HB(NMe2)2,
and the linear product NHMe2BH2NMe2BH3 (Figure S14).
Close inspection of the resulting 11B NMR spectra, however,
also revealed formation of the diazaborane compound 5, which
Figure 11. Connectivity of the Fe9 cluster.
Identification of this Fe9(PCy3)6 cluster compound lends
some credence to the proposal that the first step in the reaction
of 1 with AB is reduction of divalent iron and the formation of
soluble Fe(0) species. The dehydrogenation activity observed is
not just due to bulk Fe(0) since a commercial sample of iron
powder exhibited negligible activity with or without the presence
of PCy3. In order to investigate the reduction mechanism further,
simple quantum chemical calculations were carried out using
density functional theory. Calculations were performed on
models of 1 and 2 with both Fe(II) and Fe(0) centers using
the Gaussian09 software (Revision B.01)53 involving full
geometry optimization at the PBE-h level54 with the LANL08
basis set on iron (LANL2DZ+5s5p2d+f),55 and a 6-31+G* basis
set for the ligand orbitals. A LANL2DZ effective core potential
was used for iron. Models were constructed using the crystal-
lographically derived structural data elucidated in the earlier
section, and the Fe(0) complexes were treated as the dianions.
Although it is not being suggested that the Fe(0) models
represent actual intermediate structures in the reduction pathway
to the Fe9 cluster, the calculations do provide an indication of the
type of electronic and structural changes taking place as a
consequence of the two-electron reduction process. The first
finding from the calculations is that the high-spin model is
predicted to be the most stable for both 1 and 2, by 295.4 and
265.3 kJ mol−1, respectively. A summary of the structural
parameters for the minimized structures is given in Table 2. The
calculated bond lengths also show closer agreement to exper-
iment for the high-spin models. The plausibility of the proposed
reduction mechanism is further supported by the predicted
decrease in Fe−P bond length on going from the calculated
optimized structures between Fe(II) and Fe(0) models. This
concept of the active catalyst being a solubilized Fe(0) aggregate
is also consistent with the observation that addition of excess
PCy3 minimizes the formation of black precipitate. Due to the
similarities in the product distribution between 1 and 2 and their
initial ligand environments, it is likely that they follow similar
reaction pathway(s).
was obtained independently from reaction of (NHPhCH2)2
with borane−THF (Figures S15, S16). This then points to an
additional catalyst decomposition pathway, namely borane
complexation to the basic amido- or amine arm of the N−N
chelate, similar to the situation observed by Schneider et al.50
discussed above. In any case, it is clear that the ultimate fate of
3 is decomposition and loss of activity, as it cannot be used to
dehydrogenate a second batch of AB. The unsymmetrical NP
ligand was therefore utilized in order to isolate a more robust
version of 3 with similar steric and electronic properties such
that if phosphine dissociation occurred in 4, one atom per
ligand would still remain bound to the iron center, as compared
to complete dissociation of the chelating phosphine in 3.
Surprisingly, 4 behaves more similarly to 1 and 2 with respect
c. Reaction Pathways of 3 and 4. Complex 3 is unique in
both the increased AB dehydrogenation reaction rate
(competent at 25 °C) and its selectivity in forming PAB. It
also differs from 1 and 2 in that the parent amine of its amido
ligand (PhHNCH2)2, as demonstrated by B, vide supra) is a
5604
dx.doi.org/10.1021/ja210542r | J. Am. Chem. Soc. 2012, 134, 5598−5609

Journal of the American Chemical Society
Article
to AB dehydrogenation. The following section attempts to
address why this is the case.
of Fe−H at the end of catalysis, decomposition renders this
inconclusive.
Interestingly, Fagnou’s calculations on a Ru-NP′ complex
(NP′ = NH2CH2CH2PMe2) also point to the feasibility of end-
chain propagation and proton transfer from substrate to ligand
(Figure 13, complex H).24 While there are insufficient data to
draw any definitive conclusions with respect to the role of
species such as G, H, and I in this particular case, end-chain
propagation would also explain the formation of PAB. On the
other hand, our previous report of selectivity control of AB
dehydrogenation in ionic liquids,28 in which better coordinating
anions discouraged formation of PAB, is more consistent with a
coordination polymerization mechanism.51
Complex 4: Comparisons with Complex 3. Both of the
mechanisms shown in Schemes 6 and 8i illustrate pathways by
which 4 could be more susceptible than 3 toward ligand loss
(Scheme 9). Our original motivation for synthesizing the mixed
NP donor ligand was to prevent ligand loss by tethering the
bulky P-donor atom to an amide group, rather than to another
phosphorus atom. However, a side effect of this is that the acute
N−Fe−P angle draws the tethered N- and P-substituents closer
together while simultaneously increasing N(1)−P(2) and
N(2)−P(1) (i.e., trans atom) distances.
If proton transfer upon AB binding is indeed a mechanistic
step, the closest proton-acceptor atom in proximity to the
ligated AB moiety is the amide nitrogen contained within the
now monodentate NP ligand (i.e., J in Scheme 9). It is expected
that the (Me3Si)2N− ligands within 1 and 2 are readily prone to
protonation and subsequent dissociation, leading to metal-
based reduction (based on the observed precipitate formation).
Thus, it appears to be the case that complex 4 also undergoes
ligand dissociation and metal-based reduction in an analogous
fashion to 1 and 2. In contrast, however, in the case of complex
3, chain initiation and propagation to form PAB is faster than
competing protonation/ligand dissociation chemistry which
eventually leads to catalyst decomposition. Additionally, phos-
phine−borane is not observed in the polymerization of AB by 3
due to the lower temperature employed compared to 1, 2, and 4.
d. Formation of Poly(aminoborane): Some Mechanistic
Observations. Due to the selective and rapid formation of PAB
with 3, this complex intrigued us as a possible route to the
controlled polymerization of B−N materials if catalyst robust-
ness issues could be addressed. A first step for PAB formation is
likely via coordination of the BH3 moiety, either via bridging
hydrides or through a σ-type B−H bond interaction. These are
well-established binding motifs within metal complexes
containing Rh, Ir, and Ru.57−64 In the case of an Ir complex,
Weller et al. have observed coordination of a BH3 unit to Ir,
with no evidence for formation of a terminal metal hydride
containing the H−Ir−BH2 unit (Figure 12).64 This group also
Figure 12. NMeH2BH3 binding to an Ir complex.64
demonstrated the dimerization of BH3NH2Me, however the
proposed mechanism in this case cannot account for higher
order linear oligomers as it depends upon proximity of the
terminal NH group to the metal center in order to release H2
and form the dimer (Scheme 5).
Scheme 5. Dimerization of BH3NMeH2 As Proposed by
Weller and Co-Workers64
Complex 3: Pathway A. The induction period in the
reaction of 3 with AB may signify the reversible dissociation of
one arm of the phosphine chelate followed by ligation of AB via
the Lewis basic B−H bond to give C (Scheme 6). Protonation
of the amido nitrogen would then afford the amidoborane
complex D followed by dissociation of the amino arm, allowing
room for ligation of a second equivalent of AB to give E.
Subsequent rapid dehydrogenative insertion steps could then
build the polymer chain in an analogous fashion to that of
Ziegler−Natta polymerization of olefins. While amidoborane
complexes like D have only been isolated for electrophilic
metals such as Mg, Ca, and Zr,65 β-H elimination to give Fe-
coordinated η2-H2BNH2 intermediates (F), as characterized for
Ru by Sabo-Etienne et al. and Rh by Weller et al., could also be
an important step in the chain propagation.
Complex 3: Pathway B. Alternatively, coordination of the
NH2BH2 fragment to Fe via the BH2 group of aminoborane
(e.g., F) could also result in end-chain propagation (Scheme 7).
Such a species could also be accessed if, after loss of coordination
of one of the phosphine arms, AB could coordinate in a manner
similar to that observed in the Rh, Ir, and Ru systems (Figure 12).
If an Fe−H moiety is formed (Scheme 8, pathway i) and the
chelating amide is protonated, the result would also be a bound
NH2BH2 unit. Alternatively, the coordinated AB could dehydro-
genate directly (Scheme 8, pathway ii), which has been observed
with BH3NMe2H and Ir.63 While 1H NMR indicates the presence
CONCLUSIONS
■
While AB possesses attractive properties for use as a
transportation-based H2 fuel carrier, its thermal decomposition
characteristics are unsuitable for practical deployment. Thus,
the catalyzed dehydrogenation of AB is a promising method-
ology to circumvent this, although care must also be taken that
any “spent fuel” form is also amenable to regeneration under
realistic conditions.17 Due to the scale of the transportation
sector, any catalyst must also be comprised of elements that are
globally abundant. To this end, we have explored the use of Fe-
based complexes as catalytically active templates for AB
dehydrogenation.
Herein we have demonstrated AB dehydrogenation using
four Fe-based complexes supported by mixed amido/phosphine
ligands. While the identity of the catalytically active species is
unclear in each case, it is likely that complexes 1−4 are in fact
precatalysts that react with AB to form the active catalytic
species. The nature of precatalyst activation and its subsequent
effect on coordination and dehydrogenation of AB appears to
be a key factor in understanding the marked reactivity dif-
ferences between 3 (increased rate) and 1, 2, and 4 (increased
extent). Another influence is whether or not reduction of
Fe(II) to soluble Fe(0) species is required in order to form
higher order dehydrogenation products. Furthermore, the
5605
dx.doi.org/10.1021/ja210542r | J. Am. Chem. Soc. 2012, 134, 5598−5609

Journal of the American Chemical Society
Article
Scheme 6. Possible Initiation Mechanism for the Reaction of 3 with AB, Dictating the Formation of PAB
Scheme 7. End-Chain Propagation of AB Polymerization
by 3
Figure 13. Selected calculated intermediates during the dehydroge-
nation of AB by Ru-NP′ as reported by Fagnou et al.24 ΔG(298 K)
energies in THF (F is lowest energy conformer) are in kcal/mol.
Scheme 9. Possible Pathway for Enhanced Ligand Loss
from 4
observed reactivity of 3 may point toward the synthesis of
chemically robust Fe complexes that exhibit rapid, selective
polymerization of amine−boranes. Our observation of B−N
bond formation with the N−N ligand in complex 3 also points
to the need to control the basicity at N in bifunctional catalysts
for this application if they are to compete with those developed
by Williams et al. that employ a Lewis acidic main group center
that is to date only effective with Ru. Such general designed
catalytic approaches to selective formation of PB are particularly
important if robust, supported catalysts are to be developed for
practical use with hydrogen fuel cells.
Anhydrous solvents were obtained from Aldrich or Acros and stored
over 4 Å molecular sieves under an Ar atmosphere in a glovebox. NEt3
(Aldrich) was stored over 4 Å molecular sieves. Chemicals
dimethylamine−borane, borane-THF, N,N′-diphenylethylenediamine,
KN(SiMe3)2, nBuLi, Me3SiBr, HCl/Et2O, C2Cl6, Fe powder, N-(2-
hydroxyethyl)aniline and NaBPh4 (Aldrich), DCPE, FeCl2, PCy3,
HPCy2, and PPh3 (Strem) were purchased commercially and used as
received. Complexes 1 and 2 were synthesized as previously reported.42
NMR spectra were obtained on a Bruker AV400 spectrometer and
EXPERIMENTAL SECTION
■
All manipulations involving air-sensitive compounds were performed
in an Ar-filled glovebox or using standard Schlenk techniques.
Scheme 8. Possible Initiation Steps Involving Boron-Bound AB in the Case of 3: Pathway (i) Invokes Fe−H Formation;
Pathway (ii) Invokes Direct Dehydrogenation
5606
dx.doi.org/10.1021/ja210542r | J. Am. Chem. Soc. 2012, 134, 5598−5609

Journal of the American Chemical Society
Article
referenced to 85% H3PO4 in H2O (31P, 0 ppm) or NaBPh4 (11B, −7.29
ppm). Elemental analyses were performed by Atlantic Microlab
(Norcross, GA). The magnetic susceptibility measurements were obtained
using a Quantum Design SQUID magnetometer MPMS-XL7 operating
between 1.8 and 300 K for dc-applied fields ranging from −7 to 7 T. Dc
analyses were performed on polycrystalline samples of 10.2 and 20.0 mg
for 3 and 4, respectively, packed in grease and capsule under a field
ranging from 0 to 7 T between 1.8 and 300 K. Crystals of 3 and 4 were
mounted in a nylon cryoloop from Paratone-N oil under argon gas flow.
The data were collected on a Bruker D8 diffractometer, with APEX II
charge-coupled-device (CCD) detector, and Bruker Kryoflex low-
temperature device. The instrument was equipped with graphite
monochromatized Mo Kα X-ray source (λ = 0.71073 Å), and a
0.5 mm monocapillary. A hemisphere of data was collected using ω scans,
with 10-s frame exposures and 0.5° frame widths. Data collection and
initial indexing and cell refinement were handled using APEX II66
software. Frame integration, including Lorentz-polarization corrections,
and final cell parameter calculations were carried out using SAINT+67
software. The data were corrected for absorption using redundant
reflections and the SADABS68 program. Decay of reflection intensity was
not observed as monitored via analysis of redundant frames. The structure
was solved using direct methods and difference Fourier techniques. All
hydrogen atom positions were idealized and rode on the atom they were
attached to. The final refinement included anisotropic temperature factors
on all non-hydrogen atoms. For 3, two disordered carbon atoms in a
cyclohexyl ring were each modeled in two one-half occupancy positions,
and hydrogen atom positions were not included for the disordered atoms.
Structure solution, refinement, and creation of publication materials were
performed using SHELXTL.69 ORTEP diagrams were created using
ORTEP-3.70 The extent of H2 release from AB was measured either
through use of a gas buret or by integration of the 11B NMR spectrum
against the internal NaBPh4 standard. In the latter case, PB/borazine
products were treated as generating 2 equiv of H2 and PAB (inferred from
unaccounted for 11B in the product spectrum) was treated as having
released 1 equiv of H2. In cases of phosphine−borane formation, the
phosphine−borane is counted as producing 0 equiv of H2.
Synthesis of 3. FeCl2(DCPE) (1.40 g, 2.55 mmol) was added to a
20 mL scintillation vial, to which was added toluene (ca. 15 mL). N,N′-
Diphenylethylenediamine (541 mg, 2.55 mmol) was subsequently
added to give a brown mixture that was allowed to stir for ca. 1 h.
Upon addition of KN(SiMe3)2 (1.02 g, 5.11 mmol), the mixture
turned red. The mixture was filtered through Celite to give a red
solution that was then concentrated and cooled to ca. −25 °C. The
resulting solid was collected and dried in vacuo (580 mg, 33%). Anal.
C40H62FeN2P2 (calcd) C: 68.50 (69.76), H: 8.89 (9.07), N: 3.98
(4.07). χT at 300 K = 2.71 cm3·K/mol (μeff = 4.66 μB).
Synthesis of 4. The HNP ligand (1.00 g, 3.15 mmol) and FeCl2
(200 mg, 1.58 mmol) were added to a 20 mL vial followed by THF
(ca. 15 mL). After the FeCl2 was all dissolved, KN(SiMe3)2 (628 mg,
3.15 mmol) was added to yield a dark red mixture. The mixture was
filtered through Celite and washed with THF (3 × 15 mL) to give a
red solution. This solution was reduced to dryness, and the residue was
taken up in toluene and filtered again. The resulting filtrate was
concentrated, and two crops were collected of an orange solid (729 mg,
67%).
further dried over 4 Å molecular sieves before storage at ca. −25 °C
1
(7.8 g, 67%). H NMR (benzene-d6): δ 2.88 (m, 2H), 3.04 (t, 2H),
3.37 (bs, 1H), 6.29 (m, 2H), 6.73 (tt, 1H), 7.09 (m, 2H).
Synthesis of HNP. N-(2-Chloroethyl)benzenamine (6.00 g, 38.6
mmol) was added to a Schlenk flask followed by CH2Cl2 (ca. 20 mL)
and NEt3 (8.1 mL, 58.1 mmol). The reaction mixture was cooled in an
ice bath, and Me3SiBr (5.6 mL, 42.4 mmol) was added dropwise using
an addition funnel. The reaction was then allowed to warm to room
temperature overnight. The solvent was removed in vacuo, and the
resulting residue was taken up in pentane and filtered through Celite.
Removal of pentane yielded the protected aniline as a cloudy colorless
oil (7.5 g, 85%). 1H NMR (benzene-d6): δ 0.09 (s, 9H), 3.14 (t, 2H),
3.34 (t, 2H), 6.80−6.90 (m, 3H), 7.09 (m, 2H).
HPCy2 (1.00 g, 5.04 mmol) was added to a 20 mL vial followed by
THF (ca. 10 mL), and the resulting solution was cooled to ca. −25 °C.
Cold nBuLi (2.5 M in hexanes, 2.02 mL, 5.05 mmol) was added via
syringe in 3 portions with cooling of the reaction mixture back down
to ca. −25 °C in between additions. The resulting yellow mixture of
LiPCy2 was cooled to ca. −25 °C. In a separate vial, the protected
aniline (1.15 g, 5.05 mmol) in THF (ca. 2 mL) was also cooled. Once
cooled, the protected aniline was added dropwise to the LiPCy2
mixture to give an orange solution. After stirring for ca. 2 h, the
volatiles were removed in vacuo, and the residue was taken up in Et2O
and filtered. Excess HCl/Et2O (2 M) was added to the filtrate resulting
in the formation of white precipitate. The mixture was stirred
overnight. The white solid was subsequently collected, washed with
Et2O (2 × 10 mL), added to a THF/NEt3 mixture (ca. 7 mL/10 mL),
and left stirring for 5 days. After filtration through Celite, volatiles were
removed from the filtrate in vacuo to give a colorless solid (1.17 g,
60%) that was used without further purification (see Figures S18, S19
1
1
for H and 31P NMR spectra). H NMR (benzene-d6): δ 1.00−1.90
(m, 24H), 3.17 (m, 2H), 3.60 (bs, 1H), 6.56 (d, 2H), 6.76 (t, 1H),
7.19 (t, 2H). 31P{1H} NMR (benzene-d6): δ −10.16 (s).
Dehydrogenation of AB. In a typical dehydrogenation experi-
ment, AB and NaBPh4 were added to a 4 mL vial followed by ca. 1.5−
2 mL of diglyme. Once the solids dissolved, the metal complex was
added to the solution, and the vial was loosely capped to allow for the
escape of H2. For experiments that required heating, the vial was
placed in a heating block. At the end of the run, the sample was filtered
through Celite into a 5 mm NMR tube. Typical quantities used: 3,
8.2 mg, 0.012 mmol; AB, 7.3 mg, 0.237 mmol; and NaBPh4, 27.7 mg,
0.081 mmol.
Bulk Dehydrogenation of AB Using 3. Complex 3 (115 mg,
0.167 mmol) was added to a 50 mL round-bottom flask and dissolved
in THF (ca. 25 mL). While the solution was stirring, AB (105 mg,
3.40 mmol) was added, and the reaction was allowed to proceed for ca.
1 h. The resulting precipitate was collected on a glass frit, washed with
THF, and dried to yield a grayish-white solid (89 mg, 91%) (see
Supporting Information for IR spectrum).
Dehydrogenation of DMAB Using 3. Complex 3 (21 mg, 0.031
mmol) was added to a solution of 20 mg (0.34 mmol) dimethyl-
amine−borane dissolved in 2 mL of 1,2-dimethoxyethane. The red
solution slowly turned to black with some black precipitate. After 12 h
at 20 °C a 1 mL aliquot was withdrawn from the vial and combined
with 3 drops of C6D6 in a 5 mm NMR tube for 11B NMR analysis.
Generation of 5. To a solution of 21 mg (0.1 mmol) of N,N′-
diphenylethylenediamine in 1 mL of 1,2-dimethoxyethane was added
200 uL of a 1 M solution of borane−THF in THF. After 6 h at 20 °C,
the colorless sample was transferred to a 5 mm NMR tube, and 3
drops of C6D6 were added for 11B NMR analysis. Observed products
were the diamine bis(borane) and 5 in a 4:1 ratio.
Anal. C40H62FeN2P2 (calcd) C: 69.61 (69.76), H: 8.94 (9.07), N:
4.15 (4.07). χT at 300K = 3.88 cm3·K/mol (μeff = 5.57 μB).
Synthesis of N-(2-Chloroethyl)benzenamine. This compound
has been reported in the literature.71 Alternatively, N-(2-
hydroxyethyl)aniline (10.3 g, 75.1 mmol) was added to a 300 mL
round-bottom flask followed by CH2Cl2 (ca. 80 mL). The resulting
solution was cooled in an ice bath before addition of C2Cl6 (19.6 g,
82.8 mmol). To this, PPh3 (21.7 g, 82.7 mmol) was slowly added in
portions using a powder funnel. After complete addition of PPh3, the
funnel was washed with CH2Cl2 (ca. 20 mL). The reaction was sub-
sequently left to warm to room temperature overnight. The resulting
white precipitate was collected, added to H2O (ca. 150 mL), and
quenched with 3 equiv of NaHCO3. The product was extracted with
CH2Cl2 (3 × 100 mL) and dried with Na2SO4. After solvent removal
on a rotary evaporator, the very pale yellow oil was degassed and
ASSOCIATED CONTENT
■
S
* Supporting Information
Supplementary spectra, X-ray diffraction tables and CIF files for
compounds 3 and 4, and complete refs 34 and 53. This material
is available free of charge via the Internet at http://pubs.acs.org.
5607
dx.doi.org/10.1021/ja210542r | J. Am. Chem. Soc. 2012, 134, 5598−5609

Journal of the American Chemical Society
Article
(25) Kaß, M.; Friedrich, A.; Drees, M.; Schneider, S. Angew. Chem.,
Int. Ed. 2009, 48, 905−907.
(26) Kim, S. K.; Han, W. S.; Kim, T. J.; Kim, T. Y.; Nam, S. W.;
̈
AUTHOR INFORMATION
■
Corresponding Author
jgordon@lanl.gov
́
Mitoraj, M.; Piekos, Ł.; Michalak, A.; Hwang, S. J.; Kang, S.-O. J. Am.
Chem. Soc. 2010, 132, 9954−9955.
Present Address
(27) Keaton, R. J.; Blacquiere, J. M.; Baker, R. T. J. Am. Chem. Soc.
2007, 129, 1844−1845.
∥QD Vision, Inc., Watertown, MA 02472
Notes
(28) Wright, W. R. H.; Berkeley, E. R.; Alden, L. R.; Baker, R. T.;
Sneddon, L. G. Chemm. Commun. 2011, 47, 3177−3179.
(29) Jaska, C. A.; Temple, K.; Lough, A. J.; Manners, I. Chem.
Commun. 2001, 962−963. Jaska, C. A.; Temple, K.; Lough, A. J.;
Manners, I. J. Am. Chem. Soc. 2003, 125, 9424−9434. Jaska, C. A.;
Manners, I. J. Am. Chem. Soc. 2004, 126, 9776−9785.
(30) See Supporting Information in ref 27.
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
This work was funded by the U.S. Department of Energy,
Office of Energy Efficiency and Renewable Energy. J.C.G.
thanks Michael Janicke for NMR spectrometer maintenance,
and R.T.B. thanks the Canada Foundation for Innovation and
the Ontario Research Fund for instrumentation within the
CCRI and Glenn Facey for assistance with the NMR
spectrometers. The authors also thank the reviewers for their
insightful suggestions.
(31) Vance, J. R.; Robertson, A. P. M; Lee, K.; Manners, I. Chem.
Eur. J. 2011, 17, 4099−4103.
(32) Pons, V.; Baker, R. T.; Szymczak, N. K.; Heldebrant, D. J.;
Linehan, J. C.; Matus, M. H.; Grant, D. J.; Dixon, D. A. Chem.
Commun. 2008, 6597−6599.
(33) Shrestha, R. P.; Diyabalanage, H. V. K.; Semelsberger, T. A.;
Ott, K. C.; Burrell, A. K. Int. J. Hydrogen Energy 2009, 34, 2616−2621.
(34) Ott, K. C.; et al. Chemical Hydrogen Storage Research at Los
Alamos National Laboratory, DoE Hydrogen Annual Progress Report,
2007 (http://www.hydrogen.energy.gov/pdfs/progress07/iv_b_5g_
ott.pdf).
(35) Nutt, W. R.; McKee, M. L. Inorg. Chem. 2007, 46, 7633−7645.
(36) Zimmerman, P. M.; Paul, A.; Zhang, Z.; Musgrave, C. Bioinorg.
Chem. 2009, 48, 1069−1081.
(37) Shaw, W. J.; Linehan, J. C.; Szymczak, N. K.; Heldebrant, D. J.;
Yonker, C.; Camaioni, D. M.; Baker, R. T.; Autrey, T. Angew. Chem.,
Int. Ed. 2008, 47, 7493−7496.
(38) Ikariya, T.; Blacker, J. Acc. Chem. Res. 2007, 40, 1300−1308.
(39) Noyori, R. Angew. Chem., Int. Ed. 2002, 41, 2008−2022.
(40) Noyori, R.; Hashiguchi, S. Acc. Chem. Res. 1997, 30, 97−102.
(41) Andersen, R. A.; Faegri, K. Jr.; Green, J. C.; Haaland, A;
Lappert, M. F.; Leung, W. P.; Rypdal, K. Inorg. Chem. 1988, 27, 1782−
1786.
(42) Lin, P.-H.; Smythe, N. C.; Gorelsky, S. I.; Maguire, S.; Henson,
N. J.; Korobkov, I.; Scott, B. L.; Gordon, J. C.; Baker, R. T.; Murugesu,
M. J. Am. Chem. Soc. 2011, 133, 15806−15809.
REFERENCES
■
(1) Schlapbach, L.; Zuettel, A. Nature 2001, 414, 353−358.
(2) Satyapal, S.; Petrovic, J.; Read, C.; Thomas, G.; Ordaz, G. Catal.
Today 2007, 120, 246−256.
(3) Technical System Targets: On-Board Hydrogen Storage for Light-
Duty Vehicles (http://www1.eere.energy.gov/hydrogenandfuelcells/
storage/pdfs/targets_onboard_hydro_storage.pdf).
(4) Marder, T. B. Angew. Chem., Int. Ed. 2007, 46, 8116−8118.
(5) Stephens, F. H.; Pons, V.; Baker, R. T. Dalton Trans. 2007,
2613−2626.
(6) Klerke, A.; Christensen, C. H.; Nørskov, J. K.; Vegge, T. J. Mater.
Chem. 2008, 18, 2304−2310.
(7) Wang, P.; Kang, X.-D. Dalton Trans. 2008, 5400−5413.
(8) Peng, B.; Chen, J. Energy Environ. Sci. 2008, 1, 479−483.
(9) Chen, P.; Zhu, M. Mater. Today 2008, 11, 36−43.
(10) Murray, L. J.; Dinca, M.; Long, J. R. Chem. Soc. Rev. 2009, 38,
1294−1314.
(11) Hamilton, C. W.; Baker, R. T.; Staubitz, A.; Manners, I. Chem.
Soc. Rev. 2009, 38, 279−293.
(12) Eberle, U.; Felderhoff, M.; Schuth, F. Angew. Chem., Int. Ed.
2009, 48, 6608−6630.
(13) Smythe, N. C.; Gordon, J. C. Eur. J. Inorg. Chem. 2010, 509−
(43) Sues, P. E.; Lough, A. J.; Morris, R. H. Organometallics 2011, 30,
4418−4431.
(44) Mikhailine, A. A.; Morris, R. H. Inorg. Chem. 2010, 49, 11039−
11044.
̈
521.
(45) Paraskevi, O. L.; Lough, A. J.; Morris, R. H. Inorg. Chem. 2010,
(14) Staubitz, A; Robertson, A. P. M; Manners, I. Chem. Rev. 2010,
110, 4079−4124.
(15) Conley, B. L.; Williams, T. J. Chem. Commun. 2010, 46, 4815−
4817.
49, 10057−10066.
(46) Mikhailine, A.; Lough, A. J.; Morris, R. H. J. Am. Chem. Soc.
2009, 131, 1394−1395.
(47) Sui-Seng, C.; Freutel, F.; Lough, A. J.; Morris, R. H. Angew.
Chem., Int. Ed. 2008, 47, 940−943.
(16) Conley, B. L.; Guess, D.; Williams, T. J. J. Am. Chem. Soc. 2011,
133, 14212−14215.
(48) Li, Y.-Y.; Zhang, X.-Q.; Dong, Z.-R.; Shen, W.-Y.; Chen, G.;
Gao, J.-X. Org. Lett. 2006, 8, 5565−5567.
(17) Sutton, A. D.; Burrell, A. K.; Dixon, D. A.; Garner, E. B. III;
Gordon, J. C.; Nakagawa, T.; Ott, K. C.; Robinson, J. P.; Vasiliu, M.
Science 2011, 331, 1426−1429.
(49) Dong, Z.-R.; Li, Y.-Y.; Chen, J.-S.; Li, B.-Z.; Xing, Y.; Gao, J.-X.
Org. Lett. 2005, 7, 1043−1045.
(18) Chen, X.; Zhao, J. C.; Shore, S. G. J. Am. Chem. Soc. 2010, 132,
10658−10659.
(50) Friedrich, A.; Drees, M.; Schneider, S. Chem.Eur. J. 2009, 15,
10339−10342.
(19) Ewing, W. C.; Marchione, A.; Himmelberger, D. W.; Carroll,
P. J.; Sneddon, L. G. J. Am. Chem. Soc. 2011, 133, 17093−17099.
(20) Staubitz, A.; Robertson, A. P. M; Sloan, M. E.; Manners, I.
Chem. Rev. 2010, 110, 4023−4078.
(51) Recent results from Manners et al. also point to the possibility
of a solvent effect as generation of aminoborane in THF leads to
insoluble poly(aminoborane): Robertson, A. P. M; Leitao, E. M.;
Manners, I. J. Am. Chem. Soc. 2011, 133, 19322−19325.
(52) Sewell, L. J.; Lloyd-Jones, G. C.; Weller, A. S. J. Am. Chem. Soc.
2012, 134, 3598−3610.
(21) Staubitz, A.; Sloan, M. E.; Robertson, A. P. M; Friedrich, A.;
Schneider, S.; Gates, P. J.; Schmedt auf der Gunne, J.; Manners, I.
J. Am. Chem. Soc. 2010, 132, 13332−13345.
(22) Dallanegra, R.; Robertson, A. P.; Chaplin, A. B.; Manners, I.;
Weller, A. S. Chem. Commun. 2011, 47, 3763−3765.
(23) Denney, M. C.; Pons, V.; Hebden, T. J.; Heinekey, D. M.;
Goldberg, K. I. J. Am. Chem. Soc. 2006, 128, 12048−12049.
(24) Blaquiere, N.; Diallo-Garcia, S.; Gorelsky, S. I.; Black, D. A.;
Fagnou, K. J. Am. Chem. Soc. 2008, 130, 14034−14035.
̈
(53) Frisch, M. J.; et al. Gaussian 09, Revision B.01; Gaussian, Inc.:
Wallingford, CT, 2009.
(54) Ernzerhof, M.; Perdew, J. P. J. Chem. Phys. 1998, 109, 3313−
3320.
(55) Roy, L. E.; Hay, P. J.; Martin, R. L. J. Chem. Theory Comput.
2008, 4, 1029.
5608
dx.doi.org/10.1021/ja210542r | J. Am. Chem. Soc. 2012, 134, 5598−5609

Journal of the American Chemical Society
Article
(56) Formation of phosphine−borane from PCy3 and AB requires
elevated temperatures (see Figures S2, S4, and S5), which implies that
unbound phosphine is not sufficient for phosphine−borane formation.
́
(57) Rossin, A.; Caporali, M.; Gonsalvi, L.; Guerri, A.; Lledos, A.;
Peruzzini, M.; Zanobini, F. Eur. J. Inorg. Chem. 2009, 3055−3059.
(58) Douglas, T. M.; Chaplin, A. B.; Weller, A. S.; Yang, X.; Hall, M.
B. J. Am. Chem. Soc. 2009, 131, 15440−15456.
(59) Alcaraz, G.; Sabo-Etienne, S. Angew. Chem., Int. Ed. 2010, 49,
7170−7179.
(60) Alcaraz, G.; Chaplin, A. B.; Stevens, C. J.; Clot, E.; Vendier, L.;
Weller, A. S.; Sabo-Etienne, S. Organometallics 2010, 29, 5591−5595.
(61) Chaplin, A. B.; Weller, A. S. Inorg. Chem. 2010, 49, 1111−1121.
(62) Ledger, A. E. W.; Ellul, C. E.; Mahon, M. F.; Williams, J. M. J.;
Whittlesey, M. K. Chem.Eur. J. 2011, 17, 8704−8713.
(63) 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.
(64) 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, 11076−11079.
(65) Forster, T. D.; Tuononen, H. M.; Parvez, M.; Roesler, R. J. Am.
Chem. Soc. 2009, 131, 6689−6691.
(66) APEX II, 1.08; Bruker AXS, Inc., Madison, WI, 2004.
(67) SAINT+, 7.06; Bruker AXS, Inc., Madison, WI, 2003.
(68) Sheldrick, G. SADABS, 2.03; University of Gottingen, Germany,
̈
2001.
(69) SHELXTL, 5.10; Bruker AXS, Inc., Madison, WI, 1997.
(70) Farrugia, L. J. J. Appl. Crystallogr. 1997, 30, 565.
(71) Gangjee, A.; Wang, Y.; Queener, S. F.; Kisliuk, R. L. J. Heterocycl.
Chem. 2006, 43, 1523−1531.
5609
dx.doi.org/10.1021/ja210542r | J. Am. Chem. Soc. 2012, 134, 5598−5609