Metal-alane adducts with zero-valent nickel, cobalt, and iron

Article abstract of DOI:10.1021/ja2099744

Coordination complexes that pair a zero-valent transition metal (Ni, Co, Fe) and an aluminum(III) center have been prepared. They add to the few examples of structurally characterized metal alanes and are the first reported metallalumatranes. To understan

Full text of DOI:10.1021/ja2099744

Communication
pubs.acs.org/JACS
Metal−Alane Adducts with Zero-Valent Nickel, Cobalt, and Iron
P. Alex Rudd, Shengsi Liu, Laura Gagliardi, Victor G. Young, Jr., and Connie C. Lu*
Department of Chemistry and Supercomputing Institute, University of Minnesota, 207 Pleasant Street SE, Minneapolis, Minnesota
55455, United States
S
* Supporting Information
“buttresses” to stabilize Au→Ga, Pd→In, and Au→In bonds,8
ABSTRACT: Coordination complexes that pair a zero-
valent transition metal (Ni, Co, Fe) and an aluminum(III)
center have been prepared. They add to the few examples
of structurally characterized metal alanes and are the first
reported metallalumatranes. To understand the M−Al
interaction and gauge the effect of varying the late metal,
the complexes were characterized by X-ray crystallography,
electrochemistry, UV−Vis−NIR and NMR spectroscopies,
and theoretical calculations. The M−Al bond strength
decreases with varying M in the order Ni > Co > Fe.
but the corresponding Au→Al species was unstable and
rearranged to a nonbonding Au+Al− zwitterion.9 Building on
this strategy, we report the first examples of metallalumatrane
complexes. The M→Al interactions are compared for M = Ni0,
Co0, and Fe0.
To facilitate the synthesis of metallalumatranes, we introduce
the dinucleating heptadentate ligand 1 (Scheme 1). To
Scheme 1. Syntheses of Proligand 1, AltraPhos (2) and the
Metallalumatrane Coordination Complexes and Structure of
2
he adsorption of late transition metals on an oxide
T
support forms the basis of many heterogeneous catalysts.
The catalytic activity and selectivity are both highly dependent
on the choice of the metal, the choice of the oxide, and the
extent of metal−support interaction. Notably, strong metal−
support bonds have been proposed as the cause of enhanced
catalyst performance as well as surface deactivation.1 Research
efforts remain focused on understanding the structure−activity
relationships of these metal−support interactions.2
Dinuclear metal−alane coordination complexes are the
simplest representations of the interface of late transition
metals adsorbed on alumina.3 A systematic study of the metal−
alane bonding for a wide range of late transition metals and
oxides may provide crucial insights for catalyst design. Despite
the plethora of metal−borane complexes,4 the corresponding
metal−alane complexes are unusual. The few structurally
characterized examples are shown in Figure 1.5
generate 1, the key step is the triple dehydration reaction of
N(o-NH2C6H4)3 with 3 equiv of diisopropylphosphinome-
10
thanol. In the next step, 1 is triply deprotonated and reacted
with AlCl3 to obtain the alumatrane−phosphine complex 2
(abbreviated as AltraPhos) as a white powder.11 AltraPhos is
characterized by a singlet at 8.5 ppm in the 31P NMR spectrum,
which is slightly shifted from the 3.0 ppm peak for 1 (in C6D6).
In the 1H NMR spectrum of 2, only one resonance per proton
type was observed, suggesting threefold symmetry. Further
structural information was provided by an X-ray diffraction
(XRD) study of a single crystal of 2 grown from acetonitrile.
The solid-state structure reveals a trigonal-bipyramidal
aluminum center with a bound CH3CN molecule in the apical
pocket and dangling phosphine arms (Scheme 1 inset).
Figure 1. Structurally characterized metal−alane adducts.
Bonds between transition metals and group 13 metals are
fundamentally interesting because of the potential reversal of
the metal and ligand roles in the M→L dative bond: the metal
acts as the Lewis base, donating its electron density to the
Lewis acidic group 13 ion.6 Incorporating the borane unit into
the ligand is an effective strategy for stabilizing metal-
laboratranes.7 The ligand arms bridge the metal center and
the boron atom, serving as auxiliary supports for the M→B
bond. Bourissou and co-workers have used these ligand
With its available phosphine donors, AltraPhos readily
coordinates a variety of late transition metals. Reaction of 2
Received: October 23, 2011
Published: November 28, 2011
© 2011 American Chemical Society
20724
dx.doi.org/10.1021/ja2099744 | J. Am. Chem.Soc. 2011, 133, 20724−20727

Journal of the American Chemical Society
Communication
Figure 2. Solid-state structures of 3, 4(N2), and {5}2(μ-N2) shown at the 50% probability level. H atoms and noncoordinating solvent molecules
have been omitted for clarity.
with Ni(COD)2 (COD = 1,5-cyclooctadiene) yields the
formally zero-valent nickel−alane complex 3. As expected for
a (Ni−Al)10 d-electron count, 3 is diamagnetic. The
coordination of all three phosphine arms is consistent with
the observations of a single peak at 49.4 ppm in the 31P NMR
{5}2(μ-N2) suggests that an equilibrium exists between these
two species and establishes that N2 is indeed a labile ligand.
Whether an M−Al interaction exists is not obvious, as the
two nuclei are forced into proximity by the ligand scaffold. The
ratios of the M−Al bond distances to their respective covalent
radii (r) are all near unity (Table 1), arguing for an M−Al
1
spectrum and of diastereotopic methylene protons in the H
NMR spectrum. The broad singlet in the 27Al NMR spectrum
is shifted slightly upfield from 82.2 ppm in AltraPhos to 78.6
ppm in 3, suggesting some gain in electron density at the
Table 1. Geometrical Parameters, Including Bond Lengths
(Å) and Angles (deg), for 2(CH3CN), 3, 4(N2), and {5}2(μ-
a
N2)
aluminum center, presumably via a Ni→Al dative bond.
Further evidence for Ni−Al bonding in 3 was found by
analysis of its solid-state structure (Figure 2). The Ni−Al bond
distance of 2.450(1) Å is identical to the sum of the Ni and Al
covalent radii.12 The Al−Napical bond increases from 2.060(3) Å
in AltraPhos to 2.099(2) Å in 3: one interpretation is that the
Ni→Al interaction occurs at some cost to the Napical→Al dative
bond.
2(CH3CN)
3
4(N2)
Co
{5}2(μ-N2)
M
Ni
Fe
M−Al

2.450(1)
1.00
2.6202(9)
1.06
2.809(2)
1.11
b
r

Al−Napical
N−N
∑P−M−P
∑Neq−Al−Neq
2.060(3)

2.099(2)

2.187(2)
1.107(4)
349.47(5)
351.5(2)
2.176(4)
1.146(7)
335.03(3)
351.56(9)

359.02(6)
354.5(1)
356.58(3)
a
b
AltraPhos was found to stabilize other zero-valent transition
metal centers, namely, Co and Fe, upon mixing of 2, the
corresponding metal dibromide, and 2 equiv of the reductant
KC8. Both products are paramagnetic (1H NMR), and the IR
spectra of their solid samples (KBr pellets) contain a single N−
N stretching frequency: 2081 cm−1 for the Co complex 4(N2)
Estimated standard deviations are given in parentheses. Ratio of the
M−Al bond length to the sum of the l.s. M and Al covalent radii.12
interaction in all three complexes.6a On the basis of the
opposite trend for r, the strength of the M−Al interaction
decreases in the order Ni−Al > (N2)Co−Al > (μ-N2)Fe−Al.
Also, the apical N−Al bond length increases in going from
AltraPhos to 3 (M = Ni) to 4(N2) (M = Co) and {5}2(μ-N2)
(M = Fe). The lengthening of the Al−Napical bond correlates
with the distortion of the transition-metal center from the ideal
trigonal-bipyramidal geometry to a pseudotetrahedral geome-
try. The sums of the P−M−P angles are 349.5° in 4(N2) and
335.0° in {5}2(μ-N2), indicating significant deviations from
planarity. Moret and Peters previously described a very weak
Fe−B interaction in the ferraboratrane imide complex (TPB)-
FeNPh based on the long Fe···B bond distance (r = 1.21)
and the nearly tetrahedral coordination geometry of the iron
center (∑P−Fe−P = 333.0°).
Distortions from threefold symmetry are most notable in the
solid-state structure of Co 4(N2). The Co−P bond lengths are
different [2.2408, 2.2712, and 2.2859(9) Å] as are the P−Co−P
angles [105.07, 111.90, and 132.50(3)°]. In contrast, the M−P
bond lengths in 3 and in {5}2(μ-N2) are within 0.015 Å, and
their P−M−P angles differ by less than 1°. A simple
explanation is that the (Co−Al)9 d-electron count induces a
Jahn−Teller distortion from C3 symmetry (see below).
Consistent with this conjecture, the (Ni−Al)10 and (Fe−Al)8
configurations are expected to retain C3 symmetry.
and 2010 cm−1 for the Fe complex 5(N2) [in free N2(g), νNN
=
2331 cm−1].13 The IR data are consistent with end-on
coordination of the N2 molecule at the metal center, as
represented by M(N2)(AltraPhos). The effect of exchanging
Co0 for Fe0 is to lower the N−N stretching frequency by 71
cm−1. A similar trend was observed in another system wherein
Fe(N2) and Co(N2) species are available. Specifically, Peters
and co-workers reported a difference of ∼55 cm−1 for FeI(N2)
and CoI(N2) complexes supported by tris(phosphino)silyl
ligands.14 Moret and Peters also reported an analogous B←
Fe0(N2) complex with the tris(phosphino)borane (TPB)
ligand.15 The N−N stretching frequency of this complex is
2011 cm−1, nearly identical to that of 5(N2). The N2 ligand in
the B←Fe0(N2) complex was reported to be labile, which
impeded crystallographic characterization.
XRD studies confirmed the end-on monometallic nature of
the cobalt complex, Co(N2)(AltraPhos) [4(N2)], but revealed
an end-on N2-bridged diiron complex, {Fe(AltraPhos)}2(μ-N2)
[{5}2(μ-N2)] (Figure 2). In the Co complex, the N−N bond
distance of 1.107(4) Å is slightly elongated relative to that in
free N2(g) (1.0975 Å).13 In the Fe structure, the N−N bond
distance is 1.146(7) Å; the significantly greater activation of the
N2 ligand is partly attributed to the end-on bridged versus end-
on unbridged mode.16 The observation of both 5(N2) and
To investigate the effects of the M−Al interaction on the late
transition metal, electrochemical studies of compounds 3,
20725
dx.doi.org/10.1021/ja2099744 | J. Am. Chem.Soc. 2011, 133, 20724−20727

Journal of the American Chemical Society
Communication
4(N2), and 5(N2) were conducted. The cyclic voltammograms
of the metal complexes each display only one reversible wave
(Figure 3). In the case of 3, the reversible oxidation at −0.74 V
Figure 3. Cyclic voltammograms of 3, 4(N2), and 5(N2) in 0.1 M
[nBu4N]PF6/THF at 50 mV/s. Only reversible waves are shown.
vs Fc+/Fc is assigned to the Ni0/NiI redox couple. No
reduction events are seen. In contrast, 4(N2) and 5(N2) exhibit
single reversible reductions at −0.95 and −2.08 V, respectively,
which are assigned as the formal M−I/M0 couples. The effect of
swapping Co for Fe results in a dramatic 1.12 V difference in
the M−I/M0 redox potentials, with Co being easier to reduce
than Fe. The Fe−I/Fe0 potential of 5 is quite near the value of
−2.19 V reported for (TPB)Fe(N2).15 The similarity is
consistent with the two complexes having nearly identical N−
N stretching frequencies. Their similarities also indicate that
Fe0−AlIII and Fe0−BIII interactions must be very weak and that
the formal exchange of Al for B has little to no effect on the
iron center in this instance.
The M−Al interaction in this series was also interrogated
with UV−Vis−NIR spectroscopy (Figure 4 top). Common to
all of the complexes is a band with λmax ≈ 340 nm, which is
assigned to the π → π* transition of the ligand backbone. While
AltraPhos has no signal beyond 380 nm, the rest of the
complexes have absorptions in the visible and/or NIR regions.
3 has three visible peaks with λmax = 430, 490, and 600 nm.
4(N2) and 5(N2) have NIR bands at 1400 and 890 nm,
respectively, though the former is weak (ε < 100 L mol−1
cm−1). In the visible region, 4(N2) has a few relatively weak
Figure 4. (top) UV−vis spectra of 2, 3, 4(N2), and 5(N2) in THF at
room temperature. Inset: NIR spectrum of 5(N2) (* denotes the
swapping of light sources). (bottom) TD-DFT-predicted UV−vis−
NIR spectra of 2(CH3CN), 3, 4(N2), and 5(N2). The inset shows the
NIR region.
bands, whereas 5(N2) has an intense absorption with λmax
=
411 nm. The nature of these transitions was elucidated with
time-dependent density functional theory (TD-DFT) calcu-
lations (see below).
We conducted DFT studies to obtain a better understanding
of the M−Al electronic structures beyond the general (M−Al)n
descriptor. The DFT-optimized geometries of 3, 4(N2), and
5(N2) corresponded to the experimental structures well when
the Al and M atoms were treated with SDD and SDD+2f
pseudopotential basis sets, respectively [6-311+G(2df,p) for P
and N and 6-31G(d) for C and H atoms; M06-L;17 Gaussian
09].
Figure 5. Qualitative correlation diagram for the d-orbital manifold
and the LUMOs for 3, 4(N2), and 5(N2). MO energies are not drawn
to scale.
In all cases, open-shell calculations were employed, and no
spin density (i.e., radical character) was found on any ligand
atoms, including Al and N2. The qualitative correlation diagram
based on the calculations is shown in Figure 5. As expected, the
TD-DFT calculations (B3LYP, ORCA 2.7.0b) with solvent
considerations (COSMO) were also performed to predict the
electronic absorption spectra (Figure 4 bottom). The general
agreement between the calculated and experimental spectra is
good. Ligand-to-ligand charge-transfer (LLCT) transitions
occur in the UV range, and in all three metal complexes,
dxy/dx −y and dxz/dyz pairs are degenerate for (Ni−Al)10 and
2
2
(Fe−Al)8, whereas the dxy and dx −y orbitals are nondegenerate
for (Co−Al)9. For 3, the lowest unoccupied molecular orbital
(LUMO) is primarily a Ni 4pz orbital.
2
2
20726
dx.doi.org/10.1021/ja2099744 | J. Am. Chem.Soc. 2011, 133, 20724−20727

Journal of the American Chemical Society
Communication
intense π-based LLCT bands are seen between 325 and 360 nm
(exptl ∼340). For 3, three distinct bands are predicted in the
visible range at λmax = 440, 490, and 620 nm, in excellent
agreement with the experimental spectrum of 3. These
transitions are assigned as follows: 440 nm, dxz/dyz to 4pz
spectroscopic studies. This work was made possible by support
from the University of Minnesota. Acknowledgment is made to
the donors of the ACS Petroleum Research Fund (Grant
50395-DNI3 to C.C.L.) and to the U.S. DOE (Grant DE-
SC002183 to L.G.) for partial support of this research.
2
2
2
(LUMO); 490 nm, (dz + Lπ) to 4pz; 620 nm, dxy/dx −y to 4pz.
2
REFERENCES
The more pure dz to 4pz transition is predicted at 360 nm,
■
appearing as a shoulder on the LLCT band in Figure 4. Because
several transitions involve the LUMO, the ligand-field energies
(1) (a) Tauster, S. J.; Fung, S. C.; Baker, R. T. K.; Horsley, J. A.
Science 1981, 211, 1121. (b) Tsakoumis, N. E.; Rønning, M.; Borg, Ø.;
Rytter, E.; Holmen, A. Catal. Today 2010, 154, 162.
2
can be extracted: Δ(dz , dxz/dyz) = 0.66 eV; Δ(dxz/dyz, dxy/
(2) (a) Farmer, J. A.; Campbell, C. T. Science 2010, 329, 933.
(b) Kwak, J. H.; Hu, J.; Mei, D.; Yi, C.-W.; Kim, D. H.; Peden, C. H.
F.; Allard, L. F.; Szanyi, J. Science 2009, 325, 1670.
(3) Studies proposing metal−support interactions between metal and
aluminum centers: (a) Venezia, A. M.; Bertoncello, R.; Deganello, G.
Surf. Interface Anal. 1995, 23, 239. (b) Kunimori, K.; Ikeda, Y.; Soma,
M.; Uchijima, T. J. Catal. 1983, 79, 185. (c) Bhatia, S.; Bakhshi, N. N.;
Mathews, J. F. Can. J. Chem. Eng. 1978, 56, 575.
(4) (a) Braunschweig, H.; Dewhurst, R. D.; Schneider, A. Chem. Rev.
2010, 110, 3924. (b) Braunschweig, H.; Dewhurst, R. D. Dalton Trans.
2011, 40, 549.
(5) (a) Burlitch, J. M.; Leonowicz, M. E.; Petersen, R. B.; Hughes, R.
E. Inorg. Chem. 1979, 18, 1097. (b) Golden, J. T.; Peterson, T. H.;
Holland, P. L.; Bergman, R. G.; Andersen, R. A. J. Am. Chem. Soc.
1998, 120, 223. (c) Braunschweig, H.; Gruss, K.; Radacki, K. Angew.
Chem., Int. Ed. 2007, 46, 7782. (d) Bauer, J.; Braunschweig, H.;
Brenner, P.; Kraft, K.; Radacki, K.; Schwab, K. Chem.Eur. J. 2010,
16, 11985.
2
2
2
2
d
x − y ) = 0.79 eV; Δ(dxy/dx − y , 4pz) = 2.0 eV.
Although the TD-DFT results for Co complex 4(N2) and Fe
complex 5(N2) are complicated by MO mixing in the ground
and excited states, some important information can be gleaned.
Excitations to the metal pz orbital occur near 300 nm (see the
Supporting Information for details). Far-visible bands arising
from metal d → π*(N2) transitions occur at 400 and 431 nm
2
for 4(N2) and 362 nm for 5(N2). Additional dz (Fe) → π*(N2)
peaks are red-shifted to 432 and 434 nm. Metal d−d transitions
into the singly occupied MO (SOMO) account for the
remaining bands at 557, 990, and 1133 nm for 4(N2) and at
592 and 1113 nm for 5(N2). There are two noteworthy
discrepancies between the calculated and experimental spectra.
The NIR bands for Co are predicted to have similar intensities
as the visible bands, yet experimentally only a very weak signal
is observed at 1400 nm (ε < 100 L mol−1 cm−1). More
troubling, the far-visible bands for 5(N2) are predicted to have
relatively low intensity, but experimentally, the 411 nm band is
remarkably intense (ε = 1.5 × 104 L mol−1 cm−1). In the case of
the Fe complex, we have made the simplifying assumption that
the N2-bridged species {5}2(μ-N2) is a minor species in
solution relative to 5(N2). If this is not the case, then additional
transitions are expected for {5}2(μ-N2), and they could possibly
account for the intense band at 411 nm. The solution
equilibrium between 5(N2) and {5}2(μ-N2) is currently under
study.
(6) (a) Amgoune, A.; Bourissou, D. Chem. Commun. 2011, 47, 859.
(b) Parkin, G. Organometallics 2006, 25, 4744. (c) Hill, A. F.
Organometallics 2006, 25, 4741.
(7) (a) Hill, A. F.; Owen, G. R.; White, A. J. P.; Williams, D. J. Angew.
Chem., Int. Ed. 1999, 38, 2759. (b) Mihalcik, D. J.; White, J. L.; Tanski,
J. M.; Zakharov, L. N.; Yap, G. P. A.; Incarvito, C. D.; Rheingold, A. L.;
Rabinovich, D. Dalton Trans. 2004, 1626. (c) Pang, K.; Tanski, J. M.;
Parkin, G. Chem. Commun. 2008, 1008. (d) Bontemps, S.; Bouhadir,
G.; Gu, W.; Mercy, M.; Chen, C.-H.; Foxman, B. M.; Maron, L.;
Ozerov, O. V.; Bourissou, D. Angew. Chem. 2008, 47, 1503.
(e) Sircoglou, M.; Bontemps, S. b.; Bouhadir, G.; Saffon, N.;
Miqueu, K.; Gu, W.; Mercy, M.; Chen, C.-H.; Foxman, B. M.;
Maron, L.; Ozerov, O. V.; Bourissou, D. J. Am. Chem. Soc. 2008, 130,
16729.
(8) (a) Sircoglou, M.; Mercy, M.; Saffon, N.; Coppel, Y.; Bouhadir,
G.; Maron, L.; Bourissou, D. Angew. Chem., Int. Ed. 2009, 48, 3454.
(b) Derrah, E. J.; Sircoglou, M.; Mercy, M.; Ladeira, S.; Bouhadir, G.;
Miqueu, K.; Maron, L.; Bourissou, D. Organometallics 2011, 30, 657.
(9) Sircoglou, M.; Bouhadir, G.; Saffon, N.; Miqueu, K.; Bourissou,
D. Organometallics 2008, 27, 1675.
While we have varied the late metal to probe the chemical
nature of M→Al bonds, another intriguing and complementary
study would be to probe the effect of the supporting center on
an invariant late metal center. Future efforts are focused on
using the dinucleating proligand 1 to access novel chemical
bonds between two main-group and/or transition-metal
elements.
ASSOCIATED CONTENT
(10) Jones, M. B.; MacBeth, C. E. Inorg. Chem. 2007, 46, 8117.
(11) Su, W.; Kim, Y.; Ellern, A.; Guzei, I. A.; Verkade, J. G. J. Am.
Chem. Soc. 2006, 128, 13727.
(12) Cordero, B.; Gomez, V.; Platero-Prats, A. E.; Reves, M.;
Echeverria, J.; Cremades, E.; Barragan, F.; Alvarez, S. Dalton Trans.
2008, 2832.
■
S
* Supporting Information
Experimental procedures, X-ray crystallographic data (CIF),
computational details, and physical data. This material is
available free of charge via the Internet at http://pubs.acs.org.
(13) MacKay, B. A.; Fryzuk, M. D. Chem. Rev. 2004, 104, 385.
(14) Whited, M. T.; Mankad, N. P.; Lee, Y.; Oblad, P. F.; Peters, J. C.
Inorg. Chem. 2009, 48, 2507.
(15) Moret, M.-E.; Peters, J. C. Angew. Chem., Int. Ed. 2011, 50, 2063.
(16) Fryzuk, M. D.; Haddad, T. S.; Mylvaganam, M.; McConville, D.
H.; Rettig, S. J. J. Am. Chem. Soc. 1993, 115, 2782.
AUTHOR INFORMATION
■
Corresponding Author
clu@umn.edu
ACKNOWLEDGMENTS
(17) (a) Zhao, Y.; Truhlar, D. G. Acc. Chem. Res. 2008, 41, 157.
(b) Zhao, Y.; Truhlar, D. G. J. Chem. Phys. 2006, 125, No. 194101.
■
Aubrey (Sperier) Arenivas and Abbas Mulla are acknowledged
for their preliminary experimental and computational con-
tributions, respectively. S.L. thanks Zahid Ertem for computa-
tional advice. We thank Professor Dave Blank for the generous
use of his Vis−NIR spectrophotometer. Computing support
and resources were provided by the Minnesota Supercomput-
ing Institute. Dr. Letitia Yao assisted with the NMR
20727
dx.doi.org/10.1021/ja2099744 | J. Am. Chem.Soc. 2011, 133, 20724−20727