Aluminum hydrides, once a simple class of stoichiometric reductants, are now emerging as powerful catalysts for organic transformations such as the hydroboration or hydrogenation of unsaturated bonds. The coordination chemistry of aluminum hydrides supported by P donors is relatively underexplored. Here, we report aluminum dihydride and dimethyl complexes supported by amidophosphine ligands and study their coordination behavior in solution and in the solid state. All complexes exist as κ2-N,P complexes in the solid state. However, we find that for amidophosphine ligands bearing bulky aminophosphine donors, aluminum dihydride and dimethyl complexes undergo a "ligand-slip" rearrangement in solution to generate κ2-N,N complexes. Thus, importantly for catalytic activity, we find that the coordination behavior of the P donor can be modulated by controlling its steric bulk. We show that the reported aluminum hydrides catalyze the hydroboration of alkynes by HBPin and that the variable coordination mode exhibited by the amidophosphine ligand modulates the catalytic activity.
Aluminum hydrides, once a simple class of stoichiometric reductants, are now emerging as powerful catalysts for organic transformations such as the hydroboration or hydrogenation of unsaturated bonds. The coordination chemistry of aluminum hydrides supported by P donors is relatively underexplored. Here, we report aluminum dihydride and dimethylcomplexes supported by amidophosphine ligands and study their coordination behavior in solution and in the solid state. All complexes exist as κ2-N,Pcomplexes in the solid state. However, we find that for amidophosphine ligands bearing bulky aminophosphinedonors, aluminum dihydride and dimethylcomplexes undergo a "ligand-slip" rearrangement in solution to generate κ2-N,N complexes. Thus, importantly for catalytic activity, we find that the coordination behavior of the P donorcan be modulated by controlling its steric bulk. We show that the reported aluminum hydridescatalyze the hydroboration of alkynes by HBPin and that the variable coordination mode exhibited by the amidophosphine ligand modulates the catalytic activity.
Aluminum hydrides such
as LiAlH4, sodium bis(2-methoxyethoxy)aluminum hydride
(RedAl), and AlH3are ubiquitous in syntheticchemistry
for their use as reducing agents.[1] Recently,
the scope of the reactivity of these simple aluminum hydrides has
been expanded into catalytic hydroboration of alkenes and alkynes,
a development of significant environmental and economic importance
because of the high abundance and relatively low toxicity of aluminumcompared to platinum group metals.[2,3] Numerous other
uncomplicated aluminum hydridecompounds are also capable of hydroboration
or even hydrogenation of unsaturated polar bonds such as aldehydes,
ketones, or imines.[4,5] Aluminum hydridecompounds with
more complex ligands have also been investigated. For example, N-heterocyclicimine-coordinated aluminum hydridescatalyze carbonyl hydroboration[6] while the β-diketiminate-stabilized aluminumdihydride I (Figure ) also catalyzes the hydroboration of alkynes.[7] The dihydride I is also a precursor
to β-diketiminate-stabilized aluminum(I) species (at least within
the coordination sphere of a transition metal).[8]
Figure 1
Literature examples of aluminum dihydride and dimethyl complexes
stabilized by N-based ligands (I–III) or mixed donor ligands (IV–VI).[9−11,15,20,21] (I and IV have
Ar = 2,6-C6H3iPr2, and VI has R = R′ = Ph, iPr or R = Ph and R′
= iPr).
Literature examples of aluminum dihydride and n>an class="Chemical">dimethylcomplexes
stabilized by N-based ligands (I–III) or mixed donor ligands (IV–VI).[9−11,15,20,21] (I and IV have
Ar = 2,6-C6H3iPr2, and VI has R = R′ = Ph, iPr or R = Ph and R′
= iPr).
Reported aluminum dihydriden>an class="Chemical">complexes overwhelmingly use N-donor ligands (e.g., I–III; Figure ).[9−13] Typically, these ligands are also multidentate (to stabilize the
intrinsically electron-poor Alcenter) and sterically hindered, in
order to prevent dimerization or oligimerization by bridging interactions.
In coordination chemistry, ligands greatly influence the chemistry
at the metalcenter. Thus, the investigation and development of aluminumhydridechemistry using a diverse array of ligand classes is essential
for the expansion of aluminum hydridechemistry and catalysis.
Aluminum dihydrides or related spn>ecies with P-based ligands are much
rarer. A few examples of dimethylaluminumcomplexes with mixed-donor
ligands are known, in which bidentate ligands having one N donoralso
contain a “soft” donor, such as S or P (IV and V; Figure ).[14−18] The likely more labile Al–P interaction offers the possibility
of hemilability, which can be useful in the stabilization of catalytic
transition or resting states.[19] Indeed,
Fryzuk et al. used NMR spectroscopy to demonstrate the fluxionalcoordination
of P-donor atoms in V, resulting in an equilibrium between
four- and five-coordinate Alcenters.[20]In compn>n>an class="Chemical">arison to mixed-donor methyl complexes, mixed-donoraluminum dihydridecomplexes are scarce, with only a single example.[21] Most P-coordinated aluminum hydridesare limited
to simple adducts between phosphines and alane, with the exception
of VI (Figure ), reported by Liang et al. in 2009, which was synthesized
via the reduction of the corresponding aluminum dichloride using LiAlH4.[21,22] Hemilability of the P donors was not found
in this example, likely because of the rigidity of the ligand backbone.
Herein, we describe novel n>an class="Chemical">aluminum dimethyl and dihydride species
stabilized by mixed N,P-donor ligands that display flexible coordination
modes based on a “ligand-slip” phenomenon.
Results and Discussion
The amidophosphine ligands 1a–1n>an class="Chemical">c[23] (Figure ) have previously been used to prepare nickel and palladiumcomplexes, as well as to support reactive silicon(II)compounds.[24−27] The steric bulk around both the N and P centers of 1a–1c has not only enabled the isolation of reactive
species such as silicon(II) hydrides but also modulates reversible
SiII/SiIV oxidative additions/reductive eliminations.
At the P donor in particular, both steric bulk and electron-donating
ability are readily tunable. We were interested in whether this class
of ligands could be employed to support Alcenters and whether they
could be used to modulate their structure and reactivity.
Synthesis
and Solid-State Structures of Aluminum Dimethyl Complexes
Dimethylaluminumn>an class="Chemical">complexes are a broad class of compounds that have
been reported as catalysts or cocatalysts in alkene polymerization.[28−30] Complexes of dimethylaluminum stabilized by many N- or mixed-donor
ligands have been reported, rendering this class of compounds ideal
for benchmarking the coordination abilities of ligands 1a–1c. We decided to first investigate the coordination
of ligand 1 to dimethylaluminum moieties.
The coordination
of 1b and 1n>an class="Chemical">c to SiIV centers
has been reported and was achieved by deprotonation before treatment
with the appropriate silicon halide.[27] Accordingly,
ligands 1a–1c were deprotonated with
nBuLi at −78 °C to afford yellow solutions of 2a–2c (Scheme ). A characteristic resonance is observed in the 31P{1H} NMR spectra of these solutions in the form
of a 1:1:1:1 quartet upfield compared to the free ligand. The 1:1:1:1
multiplicity indicates coordination to Li (e.g., 2a, 31P{1H} NMR δ 10.9, JPLi = 54 Hz). Similarly, in the 7Li NMR spectra,
doublets are observed because of coupling with P (e.g., 2a, 7Li NMR δ 1.3, JLiP = 54 Hz).
Scheme 1
Lithiation of Ligand 1 To Form 2 Followed by Reaction with Dimethylaluminum Chloride To Form
Dimethylaluminum Complexes 3a–3c
The dimethylaluminumn>an class="Chemical">complexes 3a–3c were obtained by reaction of the in situ
generated lithiated ligand 2 with 1 equiv of dimethylaluminumchloride. Extraction of the products in pentane, followed by filtration
and evaporation of the solvent afforded 3a–3c as yellow air-sensitive solids. Complexes 3a and 3ccould be isolated as analytically pure solids
by crystallization, while 3b was clearly identified but
resisted purification attempts. All three complexes 3a–3c were extremely sensitive to air and moisture.
The solid-state structures of 3a and 3n>an class="Chemical">c were determined by X-ray crystallography (Figure ). Both compounds have a tetrahedralAlcenter
with coordinated N and P donors, forming a planar ring. The ring is
heavily skewed with (as might be expected) a substantially shorter
interaction between Al and the N donor than with the phosphine [e.g., 3a, Al1–N1 1.8985(14) Å vs Al1–P1 2.4800(6)
Å]. Both the Al–N and Al–P distances are comparable
to those previously reported, for example the N,P-coordinated dimethylaluminumcomplex IV [Al–N 1.894(6) Å; Al–P
2.477(3) Å].[15]
Figure 3
Molecular structures
of 3a (left) and 3c (right) with thermal
ellipsoids drawn at the 50% probability level. H and disordered ligand
atoms are omitted for clarity. Selected bond distances (Å) and
angles (deg) for 3a: N1–Al1 1.895(14), P1–Al1
2.4800(6), Al1–C1 1.9652(19), Al1–C2 1.970(2); N1–Al1–P1
86.67(4), N1–Al1–C1 116.12(8), N1–Al1–C2
115.25(8), P1–Al1–C1 114.53(7), P1–Al1–C2
114.00(7), C1–Al1–C2 109.00(9). Selected bond distances
(Å) and angles (deg) for 3c: N1–Al1 1.917(3),
P1–Al1 2.5304(9), Al1–C1 1.967(4), Al1–C2 1.964(4);
N1–Al1–P1 85.59(8), N1–Al1–C1 116.14(19),
N1–Al1–C2 116.19(19), P1–Al1–C1 115.98(13),
P1–Al1–C2 115.99(14), C1–Al1–C2 106.4(2).
Moleculn>an class="Chemical">ar structures
of 3a (left) and 3c (right) with thermal
ellipsoids drawn at the 50% probability level. H and disordered ligand
atoms are omitted for clarity. Selected bond distances (Å) and
angles (deg) for 3a: N1–Al1 1.895(14), P1–Al1
2.4800(6), Al1–C1 1.9652(19), Al1–C2 1.970(2); N1–Al1–P1
86.67(4), N1–Al1–C1 116.12(8), N1–Al1–C2
115.25(8), P1–Al1–C1 114.53(7), P1–Al1–C2
114.00(7), C1–Al1–C2 109.00(9). Selected bond distances
(Å) and angles (deg) for 3c: N1–Al1 1.917(3),
P1–Al1 2.5304(9), Al1–C1 1.967(4), Al1–C2 1.964(4);
N1–Al1–P1 85.59(8), N1–Al1–C1 116.14(19),
N1–Al1–C2 116.19(19), P1–Al1–C1 115.98(13),
P1–Al1–C2 115.99(14), C1–Al1–C2 106.4(2).
The Al–N bond distances of 3a and 3care indistinguishable, but the Al–P bond
length is slightly longer in the latter at 2.5304(8) Å, indicating
that P is less strongly bound to the Alcenter. The aminophosphinedonor of 3c is more electron-donating than the dialkylphosphinedonor of 3a, which would be expected to give rise to
the opposite trend.[31] The origin of the
difference is likely due to steric effects: the greater steric bulk
in 3c prevents the close approach of the phosphine to
the Alcenter. Indeed, this can be observed in the C1–Al1–C2
angle, which is smaller in the case of 3c [106.4(2)°]
than 3a [109.00(9)°] despite the similar bite angles
of the two [3a, 86.67(4)°; 3c, 85.59(8)°].
Solution Behavior of 3a–3c
Despite their similar solid-state strun>an class="Chemical">ctures, solution-phase NMR
spectroscopy revealed differences in the coordination behavior among
the dimethylaluminumcomplexes 3a–3c. No signals were observed for any of the compounds by 27Al NMR spectroscopy.
NMR spectrosn>an class="Chemical">copy of dimethylaluminumcomplexes 3a and 3b was consistent with the solid-state
structure determined for 3a. 31P{1H} NMR spectroscopy revealed a single resonance for each (3a, 1.6 ppm; 3b, 64.0 ppm) shifted upfield compared to
the respective free ligand resonances [Δ∂(3a) = −54.4 ppm; Δ∂(3b) = −83.3
ppm]. The 31P{1H} NMR resonances for 3a and 3b were also significantly broadened in comparison
to the free ligands 1a and 1b, presumably
as a result of coordination of the P to the quadrupolar (I = 5/2) Al nucleus [3a, full width
at half-maximum (Δν1/2) = 21.1 Hz; 1a, Δν1/2 = 2.7 Hz].
In the 1H NMR spn>en>an class="Chemical">ctra of 3a and 3b, resonances
corresponding to the aluminum methyl groups appear as doublets arising
from coupling to P (3a, δ −0.33 and −0.19, 2JHP = 2.5 Hz). The 1H NMR spectrum also shows that each CH3 group in the 2,6-diisopropylphenyl
(Dipp) substituent is inequivalent, indicating restricted rotation
likely because of stericconstraints.
Crystn>an class="Chemical">alline 3c was also characterized by solution-phase NMR spectroscopy. Surprisingly,
the 31P{1H} NMR spectrum contained two resonances,
at 99.9 and 49.7 ppm, in a ratio of 3:2 (the same ratio was observed
by 1H NMR spectroscopy). The resonance at 49.7 ppm is broadened
(Δν1/2 = 47.5 Hz) and downfield (Δ∂
= −40.9 ppm) from that of 1c and so is consistent
with coordination of P to the Alcenter as in 3a and 3b. Conversely, the resonance at 99.9 ppm is sharp (Δν1/2 = 5.3 Hz) and close in chemical shift to that of the free
ligand 1c (Δ∂ = +9.3 ppm), which indicates
that P in this environment is not coordinated to the Alcenter.
On the basis of the 31P NMR spn>en>an class="Chemical">ctroscopic data and by
analogy with the behavior more fully studied in the hydride analogue 5c (see below), we propose that 3c exists in
two forms in solution, in which the ligand exhibits a variable coordination
mode, having either κ2-N,P or κ2-N,N coordination (Scheme ). In the solid state, κ2-coordination is
exclusively observed. In solution, however, the two isomers are present
as a result of the flexible coordination mode of the ligand.
Scheme 2
Proposed
Structures of κ2-N,P- and κ2-N,N-3c
In the solid state, only κ2-N,P-3c is
observed, while in solution, both the κ2-N,P- and
κ2-N,N isomers are observed.
Proposed
Structures of κ2-N,P- and κ2-N,N-3c
In the solid state, only κ2-N,P-3n>an class="Chemical">c is
observed, while in solution, both the κ2-N,P- and
κ2-N,N isomers are observed.
The 1H NMR spn>en>an class="Chemical">ctrum of 3c is consistent with
both the κ2-N,P and κ2-N,N isomers
existing in solution, with two sets of resonances present in a ratio
of 57:43 (consistent with the 3:2 ratio observed by 31P
NMR). Multinuclear 2D NMR spectroscopic experiments verified that
in both isomers the ligand backbone was intact and undisturbed. The
possibility of a dimeric κ1-N isomer of 3c (with, e.g., bridging methyl ligands) was excluded based on analysis
of the 1H DOSY NMR spectrum, which indicated that both
of the observed isomers diffused at the same rate in solution. Similarly,
high-resolution mass spectrometry (HRMS) also identified the product
as 3c, with no evidence of a dimeric species observed.
Synthesis of Aluminum Dihydride Complexes
Following the
preparation of the n>an class="Chemical">dimethylaluminumcomplexes 3a–3c, we turned our attention to the preparation of aluminumdihydridecomplexes. Ligands 1a–1c do not react with Me2EtN·AlH3, in contrast
to the observed reactivity of amidine ligands, which evolve H2 and form aminidinatoaluminum dihydrides.[32] Treatment with LiAlH4also had no effect. Thus,
we used the lithiated ligands 2a–2c as precursors instead.
Treatment of 2b with a
single equivalent of Me2EtN·AlH3 resulted
in a yellow solution, the 31P NMR spectrum of which revealed
a quartet (δ 110.8, 2JPH = 34 Hz), which collapsed to a singlet in the 31P{1H} NMR spectrum. This evidence, as well as further characterization
by multinuclear NMR spectroscopy and mass spectrometry, confirmed
formation of the aluminatecomplex 4b (Scheme ).
Scheme 3
Proposed Mechanism
for the Reaction of 2 with Me2EtN·AlH3 (NR3 = NMe3 or NMe2Et) To
Form the Aluminum Dihydride 5 via the Charged Intermediate 4
The addition of a second equivpan class="Chemical">alent
of Me2EtN·AlH3 to solutions of 4b was monitored by 31P{1H} NMR spectroscopy,
which revealed complete consumption of 4b and the formation
of a new species represented by a broad singlet (61.3 ppm, Δν1/2 = 55.7 Hz), indicating P coordination to Al. Analysis of
the 27Al NMR spectrum revealed the formation of LiAlH4. On the basis of this evidence, the reaction pathway shown
in Scheme is proposed:
the reaction of 2b with Me2EtN·AlH3 proceeds by forming 4b by displacement of the
amine from Me2EtN·AlH3. The second 1 equivalent
of Me2EtN·AlH3 abstracts a hydride from 4b, generating 5b and LiAlH4 and eliminating
the amine.
When 2a was treated with 1 equiv of
Me2EtN·n>an class="Chemical">AlH3, the resulting pale-yellow
solution was revealed to contain a mixture of compounds by 31P{1H} NMR spectroscopy. In addition to residual lithiated
ligand 2a, equal quantities of the aluminate intermediate 4a (8.0 ppm) and the neutralaluminum dihydride 5a (−10.1 ppm) were observed. LiAlH4 was also observed
by 27Al NMR spectroscopy. The 2:1:1 ratio of the three
species reveals that the lithiated ligand 2a and the
intermediate aluminate 4a react at comparable rates with
Me2EtN·AlH3 to generate a statistical mixture.
This contrasts to the situation for 4b, where hydride
abstraction by Me2EtN·AlH3 is much slower
than its coordination to the lithiated ligand 2b. Upon
the addition of a second equivalent of Me2EtN·AlH3 to 4a, the reaction mixture turned colorless
and the 31P{1H} NMR spectrum showed complete
conversion to 5a (7.5 ppm).
Preparatively, the
n>an class="Chemical">dihydridecomplexes 5a–5c were obtained
in multigram quantities from treatment of the lithiated ligands 2a–2c with 2 equiv of Me3N·AlH3 or Me2EtN·AlH3. All three compounds
could be isolated as colorless solids in excellent yields of 80–90%.
Dihydrides 5b and 5ccould be further purified
by crystallization from hexane.
In the 1H NMR spn>en>an class="Chemical">ctra
of 5a and 5b, Al–H resonances are
visible as very broad singlets at 4.6 ppm (5a, Δν1/2 = 71.6 Hz; 5b, Δν1/2 = 125.3 Hz) because of the influence of the quadrupolarAl atom.
Despite the lower steric influence of the hydride ligands compared
to the methyl ligands of 3a and 3b, the
methyl groups of the Dipp substituent remain inequivalent, indicating
continued restricted rotation. Compound 5c has more complex
solution behavior that will be discussed below.
IR spectrosn>an class="Chemical">copy
of the solid-state samples of 5a–5c revealed the expected symmetric and antisymmetricAl–H stretches
(5a, 1810 and 1786 cm–1; 5b, 1831 and 1816 cm–1; 5c, 1825 and
1801 cm–1) for a four-coordinate aluminum dihydridecenter.[33,34]
Solid-State Structures of 5b and 5c
The structures of 5b and 5n>an class="Chemical">c were verified by X-ray diffraction (Figure ). Broadly, the structures are analogous
to those of 3a and 3c. The amidophosphine
ligand in each compound is κ2-N,P-coordinated, which
together with the hydride ligands (located using a difference map
and allowed to refine freely) results in a tetrahedral environment
at the Alcenter. The two structures have statistically identical
N–Al bond distances [5b, 1.8972(15) Å; 5c, 1.892(2) Å], which are essentially identical with
those observed for the dimethyl analogues 3a and 3c. A more substantial difference is observed in the P–Al
bond distances, which for the dihydride 5c is shorter
than that in the corresponding dimethylcomplex 3c [Al1–P1: 5c, 2.4791(10) Å; 3c, 2.5304(8) Å].
Contraction of this bond can be explained by the smaller size of the
hydride substituents. Similarly, a comparison between the two dihydrides 5b and 5c reveals a shorter Al1–P1 distance
for 5b as a result of reduced bulk at the P center in
comparison to 5c [5b, 2.4442(7) Å; 5c, 2.4791(10) Å]. The larger bite angles for the dihydrides 5b and 5c [5b, 87.47(5)°; 5c, 86.60(6)°] compared to those of the dimethylcompounds
are also due to the smaller hydride substituents compared to the methyl
groups.
Figure 4
Molecular structures of 5b (left) and 5c (right). The aluminum hydride atoms were located using a difference
map and allowed to refine freely. H and disordered ligand atoms are
omitted for clarity. Selected bond lengths (Å) and angles (deg)
for 5b: N1–Al1 1.8972(15), P1–Al1 2.4442(7);
N1–Al1–P1 87.47(5). Selected bond lengths (Å) and
angles (deg) for 5c: N1–Al1 1.892(2), P1–Al1
2.4790(10); N1–Al1–P1 86.60(6).
Moleculn>an class="Chemical">ar structures of 5b (left) and 5c (right). The aluminum hydride atoms were located using a difference
map and allowed to refine freely. H and disordered ligand atomsare
omitted for clarity. Selected bond lengths (Å) and angles (deg)
for 5b: N1–Al1 1.8972(15), P1–Al1 2.4442(7);
N1–Al1–P1 87.47(5). Selected bond lengths (Å) and
angles (deg) for 5c: N1–Al1 1.892(2), P1–Al1
2.4790(10); N1–Al1–P1 86.60(6).
Solution-Phase NMR Characterization of 5c
Like
its dimethyl analogue 3c, the dihydride 5c exhibits variable coordination modes in solution. Upon dissolution
of crystalline 5c, the 31P{1H}
NMR spectrum revealed the presence of two broad singlets at 96.9 ppm
(Δν1/2 = 137.9 Hz) and 47.8 ppm (Δν1/2 = 96.6 Hz) in a ratio of 1:2. By 1H NMR, two
sets of resonances were also observed for all proton environments,
including the dihydride ligands (signals at κ2-N,N-5c, 4.3 ppm, κ2-N,P-5c, 4.6
ppm; the ratio of the two species as measured by 1H NMR
in a ratio of 35:65, consistent with that observed in the 31P NMR spectrum).The two solution-phase isomers of 5c were determined to be κ2-N,n>an class="Gene">P-5c, as observed in the solid state, and a
κ2-N,N isomer in which the phosphine ligand has “slipped”
and coordinates through one of the P-bound N atoms (Scheme ). Evidence for the κ2-N,N coordination mode is as follows:
Scheme 4
Proposed Structures
of κ2-N,P- and
κ2-N,N-5c
In the solid state, only κ2-N,P-5c is
observed, while in solution, both the κ2-N,P and
κ2-N,N isomers are observed.
Proposed Structures
of κ2-N,P- and
κ2-N,N-5c
In the solid state, only κ2-N,P-5c is
observed, while in solution, both the κ2-n>an class="Chemical">N,P and
κ2-N,N isomers are observed.
(1) The two isomers are both monomerin>an class="Chemical">c species, as revealed by 1H DOSY NMR measurements, which indicate similar diffusion
coefficients. Thus, we were able to rule out the presence of a dimeric
species with bridging hydrides (consistent with solution- and solid-phase
IR spectroscopy, which did not reveal evidence of bridging hydride
ligands).
(2) In the 31P{n>an class="Chemical">1H} NMR spectrum,
the resonance at 96.9 ppm is assigned to the κ2-N,N
isomer because of its similarity to that observed for the free ligand 1c (90.6 ppm), which indicates that the P center is not coordinated
to Al. The resonance at 47.8 ppm is assigned to the κ2-N,P isomer observed in the solid state (confirmed by solid-state
NMR measurements; see below).
(3) The aluminum hydride stretn>an class="Chemical">ching
frequencies recorded for 5c in solution (1823 cm–1) and in the solid state (1825 and 1801 cm–1) are consistent with a four-coordinate aluminum dihydride species
in both phases, ruling out a κ1-N isomer in which
the phosphine is uncoordinated.
(4) Using density functionn>an class="Chemical">al
theory (DFT), we performed geometry optimization and frequency calculations
on κ2-N,P isomers of 5a–5c at the M062X/Def2SVPP and M062X/6,31G+(d,p)/Lanl2DZ levels
(Table S1). Following the lead of Crimmin
et al., we found that calculations using the split basis set were
essential to replicating experimentally observed Al–H stretching
frequencies.[33] The calculations accurately
reproduced the experimentally observed geometries and IR stretching
frequencies for 5a–5c, enabling us
to use this computational methodology to assign the identity of the
solution-phase isomer of 5c.
(5) A relaxed potential
energy surface (PES) scan of 5c in which the Al–P
distance was increased systematically starting from the κ2-N,P geometry revealed two potential minima (Figure S1), which were reoptimized at the M062X/6,31G+(d,p)/Lanl2DZ
level (Figure and Table S2). A κ1-N isomer was
found to be 22.6 kcal mol–1 higher in energy than
the κ2-N,P isomer (the calculated Al–H stretching
frequencies for this three-coordinate aluminum dihydride of 1934 and
1922 cm–1 were also inconsistent with the experimental
values). However, the κ2-N,N isomer located in the
PES scan was found to be very close in energy to κ2-N,P-5c (−0.8
kcal mol–1 more stable; DFT does not replicate the
experimentally observed order of stability, although it does correctly
place the two species very close in energy). Calculated Al–H
stretching frequencies for κ2-N,P- and κ2-N,N-5c (1863, 1845, and 1860, 1813 cm–1, respectively) are sufficiently close in order to explain the single
peak observed in the experimental solution-phase spectrum (1829 cm–1).
Figure 5
Computed energies of κ2-N,P-, κ2-N,N-, and κ1-N-5c [M062X/6,31G+(d,p)/Lanl2DZ].
Computed energies of κ2-n>an class="Chemical">N,P-, κ2-N,N-, and κ1-N-5c [M062X/6,31G+(d,p)/Lanl2DZ].
The ligand-slip rearrangement of 5n>an class="Chemical">c from κ2-N,P to κ2-N,N is likely driven by a preference
for the “hard” N-donor functionality of the diaminophosphinedonor over the “softer” P center. The increased proportion
of the κ2-N,N isomer for the dimethylcomplex 3ccompared to the dihydride 5c suggests that
the ring expansion that occurs as a consequence of isomerization from
κ2-N,P to κ2-N,N may also be favorable
as a route to relieve steric strain. The more restrained, sterically
crowded, and less basic (due to the silyl substituent) tert-butylamino groups of 3b and 5b cannot
favorably participate in the same isomerization as 3c and 5c.
Interconversion between κ2-n>an class="Chemical">N,P- and κ2-N,N-3c or -5c in solution was not observable, and we were thus unable to determine
the activation barriers for this process. Although resonances for
the coordinated and free phosphinecenters in both isomers of 5care broad, using NMR spectroscopy, we could find no evidence
for exchange between the two sites, even at elevated temperatures.
The variable coordination mode of the ligand in both 3c and 5c appears to provide them with higher reactivity
and renders them the most sensitive derivatives in these series. Indeed, 3c was found to be extremely challenging to handle because
of its high sensitivity to air and moisture.
Solid-State NMR Spectroscopy
To further confirm our assignment of n>an class="Chemical">31P resonances
for the κ2-N,P and κ2-N,N isomers
of 3c and 5c, we undertook solid-state NMR
spectroscopy because from crystallographic studies κ2-N,P-coordination is exclusively observed. The 31P{1H} MAS NMR spectra of 3c and 5care
consistent with X-ray crystallography, revealing only a single-P environment
for each compound (Figure ). In both cases, the solid-state chemical shift is almost
identical with the solution-phase signal assigned to the κ2-N,P isomers (e.g., 3c, solid phase, 47.8 ppm,
solution, 49.7 ppm; 5c, solid phase, 47.5 ppm, solution,
47.8 ppm). Furthermore, the line shapes observed in the 31P{1H} NMR spectra indicate quadrupolarcoupling between
Al and P, explaining the observed variation from the expected 1:1:1:1:1:1
sextet. No other resonances were observed in the 31P{1H} MAS NMR spectra, ruling out the presence of the κ2-N,N isomer in the solid state.
Figure 6
31P{1H} (9.4 T, 14 kHz, MAS) NMR spectra for 3c (top) and 5c (bottom).
31P{n>an class="Chemical">1H} (9.4 T, 14 kHz, MAS) NMR spectra for 3c (top) and 5c (bottom).
For 3a, 3b, 5a, and 5b, which n>an class="Chemical">all display exclusive
κ2-N,Pcoordination in solution, the observed 31P{1H} MAS NMR spectra each contain a single resonance
extremely close in chemical shift to that observed in solution (e.g., 5a, solution phase, 8.0 ppm, solid phase, 8.9 ppm). Although
we were unable to observe any resonances for any of the compounds
reported here by solution-phase 27Al NMR spectroscopy,
solid-state experiments were more successful. Details of the 27Al{1H} CPMG NMR spectra for 3a–3c and 5a–5care provided
in the Supporting Information.
Conclusions
In summary, we have synthesized n>an class="Chemical">aluminum dimethyl and dihydridecomplexes with a series of amidophosphine ligands of varying steric
bulk. The bulky bidentate ligands 1a–1c enable the isolation of reactive aluminum dihydrides, the synthesis
of which was observed to proceed through five-coordinate aluminate
intermediates (4a–4c). Evidence from
X-ray crystallography and solid-state NMR spectroscopy indicates that,
for all dimethyl and dihydridecomplexes, both N- and P-donor atoms
are bound to the Alcenters in the solid state. In solution, however,
altering the steric bulk of the ligand enables control over the coordination
mode at the Alcenter: for the bulkiest ligand employed, 1c, both the dimethyl and dihydridecomplexes 3c and 5c exist as a mixture of κ2-N,P and κ2-N,N isomers.
The variable n>an class="Chemical">coordination mode of the
ligand is encouraging as a potential route to controlling the stoichiometric
or catalytic reactivity of the aluminum dihydridecenters. For example,
preliminary results indicate that 5a–5care active catalysts for the hydroboration of alkyl- and arylalkynes
with HBPin (see the SI). The accessibility
of the κ2-N,N coordination mode for 5c has a clear effect on the reactivity. While all three dihydridescatalyze the hydroboration of phenylacetylene with HBPin, 5a and 5b are significantly more efficient, with conversions
of 79 and 83% after 2 h at 110 °Ccompared to 53% for 5c. We are now further exploring the coordination chemistry, reactivity,
and catalytic applications of the dihydrides 5a–5c (Scheme ).
Scheme 5
Catalytic Hydroboration of Phenylacetylene and 2-Cyclooctyne
Using 5a–5c
Experimental Section
General Procedures
All manipn>ulations were n>an class="Chemical">carried out under an argon atmosphere using
standard Schlenk or glovebox techniques. Reactions were carried out
in glass Schlenk tubes, which were dried for 16 h at 110 °C before
use. Solvents were obtained from an inert solvent purification system
and stored over 4 Å molecular sieves. C6D6 and tetrahydrofuran (THF)-d8 were dried
over potassium, then vacuum-distilled, and stored over 4 Å molecular
sieves.
Ligands 1b and 1c,[23] their precursors [imine[24] and chlorophosphines PCl(NtBu)2SiMe2[23] and PCl(NtBuCH2)2[35]], and [H3Al·NMe3][36] were synthesized according
to literature procedures. SiMe2(NHtBu)2 was synthesized according to a modified literature procedure (see
the SI). tert-Butylamine
was dried over calcium hydride and vacuum-distilled prior to use.
LiAlH4 was purified by extraction with diethyl ether and
filtration to afford a white solid, which was stored under an inert
atmosphere. Trimethylammonium chloride was dried under vacuum at 50
°C for 3 h prior to use. All other reagents were purchased from
commercial suppliers and used without further purification.
General
Synthesis of 2
To a solution of 1 in THFn>an class="Chemical">cooled to −78 °C was added dropwise nBuLi (2.5
M in hexanes, 1 equiv). The cold bath was removed, and the resultant
yellow solution was stirred at room temperature for 1 h. Monitoring
by 31P{1H} NMR spectroscopy revealed the presence
of the lithiated ligand 2, which was characterized in
situ.
2a. 31P{n>an class="Chemical">1H} NMR (C4H8O, 202.5 MHz, 300 K): δ 10.9 (1:1:1:1 quartet, JP–Li = 54 Hz). 7Li NMR (C4H8O, 194.4 MHz, 300 K): δ 1.3 (d, JLi–P = 54 Hz).
2b. 31P{n>an class="Chemical">1H} NMR (C4H8O, 202.5
MHz, 300 K): δ 96.4 (1:1:1:1 quartet, JP–Li = 63 Hz). 7Li NMR (C4H8O, 194.4 MHz, 300 K): δ 1.1 (d, JLi–P = 63 Hz).
2c. n>an class="Chemical">31P{1H} NMR (C4H8O, 202.5 MHz, 300 K): δ
68.6 (1:1:1:1 quartet, JP–Li =
54 Hz). 7Li NMR (C4H8O, 194.4 MHz,
300 K): δ 1.5 (d, JLi–P =
54 Hz).
Authors: David Gau; Tsuyoshi Kato; Nathalie Saffon-Merceron; Abel De Cózar; Fernando P Cossío; Antoine Baceiredo Journal: Angew Chem Int Ed Engl Date: 2010-09-03 Impact factor: 15.336
Authors: Jerzy Krysiak; Céline Lyon; Antoine Baceiredo; Heinz Gornitzka; Marian Mikolajczyk; Guy Bertrand Journal: Chemistry Date: 2004-04-19 Impact factor: 5.236
Authors: David Gau; Tsuyoshi Kato; Nathalie Saffon-Merceron; Fernando P Cossío; Antoine Baceiredo Journal: J Am Chem Soc Date: 2009-07-01 Impact factor: 15.419
Authors: Rosalyn L Falconer; Gary S Nichol; Ivan V Smolyar; Scott L Cockroft; Michael J Cowley Journal: Angew Chem Int Ed Engl Date: 2020-11-24 Impact factor: 15.336