Redox chemistry of the triplet complex (PNP)Co

Published on Web 03/11/2008

Redox Chemistry of the Triplet Complex (PNP)Co^I

Michael J. Ingleson, Maren Pink, Hongjun Fan, and Kenneth G. Caulton*

Department of Chemistry, Indiana UniVersity, Bloomington, Indiana 47405

Received July 6, 2007; E-mail: caulton@indiana.edu

Abstract: Reaction of PNPCo, where PNP is (^tBu₂PCH₂SiMe₂)₂N^-, with the persistent radical galvinoxyl,

G, gives PNPCo^IIG, a nonplanar S ) ³/₂species. Reaction with PhCH₂Cl or with 0.5 mol I₂gives PNPCoX

(X ) Cl or I, respectively), but additional I₂, seeking Co^III, gives instead oxidation at phosphorus: (^tBu₂P(I)CH₂-

SiMe₂NSiMe₂CH₂P^tBu₂)CoI₂. Hydrogen-atom transfer reagents fail to give PNPCoH, but H₂gives instead

PNPCo(H)₂, a result rationalized thermodynamically based on DFT calculations. Multiple equiv of PhSiH₃

give a product of Co(V), where N/SiPh and P/Si bonds have formed. N₂CH(SiMe₃) gives a 1:1 adduct of

PNPCo, whose metric parameters suggest partial oxidation above Co^I; N₂CHPh gives a 1:1 adduct but

with very different spectroscopic features. PhN₃reacts fast, via several intermediates detected below 0

°C, to finally release N₂and form a Co^Iproduct where one phosphorus has been oxidized, PN(PdNPh)Co.

Whereas PNPCo(N₃) resists loss of N₂on heating, one electron oxidation gives a rapid loss of N₂, and the

remaining nitride nitrogen is quickly incorporated into the chelate ligand, giving [^tBu₂PCH₂SiMe₂NSiMe₂-

NP(^tBu₂)dCH₂Co]. O₂or PhIdO generally gives products where one or both phosphorus centers are

converted to its oxide, bonded to cobalt.

Introduction

even if it is far short of an 18 valence electron count, still has

the potential to act as a reducing agent. Here, we probe the

We have reported¹the synthesis and characterization of

(PNP)Co, where PNP is the monoanion (^tBu₂PCH₂SiMe₂)₂N^-,

as a three-coordinate T-shaped species devoid of agostic inter-

actions. The somewhat-low oxidation state of this metal qualifies

it as a potential reducing agent but it is unclear whether the

final oxidation state will be +2 or +3. Moreover, because d⁸

(PNP)Co has a spin triplet ground state, all of its redox reactivity

involves a change of spin state.^2-5An earlier report¹on the re-

action of (PNP)Co with traditional Lewis bases, L, showed some

evidence for diminished adduct formation enthalpies, which can

originate from the spin pairing energy penalty in forming singlet

(PNP)CoL. We now report on the reactivity of (PNP)Co toward

potential one- and two-electron oxidants, including some uncon-

ventional ones: O₂, PhN₃, N₃^-, RSiH₃, I₂, and RR′CN₂. These

latter are potential precursors to MdCR₂, MdNR, MdO,

M≡CR, and MtN ligands. These results more fully define how

the reducing power of (PNP)Co, already established to be greater

than its rhodium analog by CO stretching frequencies, is not

limited to occur exclusively at the metal center. Some of these

results have been communicated previously.^6,7

limits of this expectation and show it to be true but with certain

subtle variations.

Reaction with One-Electron Oxidants. We first looked at

the propensity of 1 to react with radical one-electron donors

and reagents that are sources of A^•(A ) halogen or hydrogen).

The reaction of equimolar quantities of 1 and the free radical

galvinoxyl resulted in a rapid color change from green to red.

1

The H NMR spectrum now exhibited eight paramagnetically

shifted resonances, consistent with the addition of galvinoxyl

to form a single new product. Mass spectroscopy confirmed

this to be PNPCoGalvinoxyl, 2. The solution magnetic moment

(3.96µ_B) supports a formal oxidation of the cobalt center to a

high-spin (distorted-tetrahedral) Co(II) and the concomitant

formation of an aryloxide anionic ligand. Crystallization by

cooling a saturated pentane solution and subsequent analysis

by X-ray diffraction further confirmed the formation of 2 (Figure

1) formed by a rapid radical combination reaction between 1

and galvinoxyl.

The structure around Co^IIis indeed distorted tetrahedral,

though there are significant distortions compared¹to the solid-

state structures of PNPCoI, PNPCo(OTf), and even when

compared to PNPCo(anilide),¹due to change in ortho substit-

uents from methyls in PNPCo(anilide), to significantly larger

^tBu groups in 2. These effects include, (1) a significant

elongation of the P-Co bond lengths in 2 (P1-Co ) 2.5732-

(7)Å, P2-Co ) 2.5851(7)Å, average in PNPCoI ) 2.44Å, in

anilide, 2.51 Å and in PNPCo(OTf), ) 2.41Å); (2) an opening

of the N-Co-X angle (in 2 N1-Co-O1 ) 128.39(7)°,

Results and Discussion

The pincer ligand chosen in this work, (^tBu₂PCH₂SiMe₂)₂N^-1

,

is a sufficiently potent donor that three-coordinate (PNP)Co, 1,

(1) Ingleson, M. J.; Pink, M.; Fan, H.; Caulton, K. G. Inorg. Chem. 2007, 46,

10321.

(2) Poli, R. J. Organomet. Chem. 2004, 689, 4291.

(3) Cacelli, I.; Poli, R.; Quadrelli, E. A.; Rizzo, A.; Smith, K. M. Inorg. Chem.

2000, 39, 517.

(4) Poli, R. Acc. Chem. Res. 1997, 30, 494.

(5) Poli, R. Chem. ReV. 1996, 96, 2135.

(6) Ingleson, M. J.; Pink, M.; Caulton, K. G. J. Am. Chem. Soc. 2006, 128,

4248.

(7) Ingleson, M.; Fan, H.; Pink, M.; Tomaszewski, J.; Caulton, K. G. J. Am.

Chem. Soc. 2006, 128, 1804.

9

4262

J. AM. CHEM. SOC. 2008, 130, 4262-4276

Redox Chemistry of the Triplet Complex (PNP)Co^I

A R T I C L E S

Figure 1. Left, The oxidation of cobalt in 1 by reaction with the free radical galvinoxyl. Right, ORTEP view (50% probabilities) of 2 (^tBu carbons and

hydrogens omitted for clarity). Selected bond lengths (Å) and angles (°): Co-N1, 1.9663(18); Co-P1, 2.5732(7); Co-P2, 2.5851(7); Co-O1, 1.8936(15);

P1-Co-P2, 137.44(2), N1-Co-O1, 128.39(7); P1-Co-O1, 109.46(5); P2-Co-O1, 102.22(5); P1-Co-N1, 92.70(6); P2-Co-N1, 89.14(6).

Figure 2. Reactivity of PNPCo and PNPCoX with halides/benzyls halides and phenyl halides.

whereas in PNPCoI, 117.45(4)° and in PNPCo(anilide), 118.95-

(11)°); (3) a compression of the N1-Co-P angles (in 2 92.70-

(6)° and 89.14(6)° compared to an average in PNPCoI of 93.51°,

in anilide 90.4° and in PNPCo(OTf) of 95.69°). All of these

Bu₂)CoI₂is reduced by two electrons. These results are

summarized in Figure 2.

Following the successful oxidation of 1 with halogen atom

donors, we targeted the synthesis of PNPCoH by the reaction

of 1 with molecules with weak X-H bonds as potential

hydrogen-atom donors. Radical metal complexes are well

documented to abstract a hydrogen atom from silanes, stannanes,

and even homolytically split H₂(e.g., the reaction of TpCoN₂

with 0.5 equiv of H₂yields⁹TpCoH). However, there is no

t

structural changes are required to accommodate the ortho Bu

groups of galvinoxyl, thereby allowing for the close approach

of oxygen to cobalt. The galvinoxyl phenoxy containing O1 is

twisted to be orthogonal to the P-Co-P plane, thereby

minimizing steric repulsion with PNP. The Co-O1 distance

(1.8936(15)Å) in 2 is fully consistent with other high-spin four-

coordinate Co(II) aryloxide complexes;⁸however on compari-

son, the severe steric encumbrance in 2 does manifest itself

further by inducing a considerable opening of the Co-O1-

C23 angle (158.16(15)°). There is a significant twisting around

C29, preventing any conjugation between aryl rings and

precluding any further ligand reduction (e.g., oxidation to Co^III

would result in a ligand-based radical that could be delocalized

through both aryl rings).

n

reaction of 1 with Bu₃Sn-H (BDE Sn-H ) 74 kcal‚mol^-1).

To eliminate the possibility that the steric encapsulation that

enshrouds the cobalt center in 1 may be preventing any reaction

from taking place, the smaller H^•atom source 1,4 cyclohexa-

diene (BDE C-H ) 77 kcal‚mol^-1) was reacted with 1, again

without success (18h at 25 °C). Therefore, steric bulk is not

the insurmountable factor; instead it is reasonable that the failure

stems from the Co-H bond in the hypothetical PNPCoH being

weaker than the X-H bonds of these reagents (entirely

consistent with the proposed¹weak Co-R bond to be the origin⁶

of reductive alkylation of 1 rather than simple metathesis). Thus,

1 can be readily oxidized by one-electron, provided the resultant

Co-X bond is sufficiently strong to provide a driving force

Following the ease of oxidation of 1 with the free radical

galvinoxyl, we next probed the reactivity of 1 with respect to

potential X atom donors (X ) Cl or I). Reaction of 1 with

PhCH₂Cl rapidly forms PNPCoCl and half of an equiv of

PhCH₂CH₂Ph. There is no further reaction between PNPCoCl

and excess PhCH₂Cl. 1 also readily reacts with iodobenzene at

25 °C, generating PNPCoI quantitatively. The more-challenging

chlorobenzene required heating to 60 °C to activate the C-Cl

bond (at 25 °C 1 is recovered unchanged after 18 h in the

presence of 5 equiv of PhCl), producing PNPCoCl quantitatively

(by ¹H NMR). 1 also reacts with 0.5 equiv of I₂, predominantly

forming PNPCoI.¹However, small quantities of [κ²-(^tBu₂P(I)CH₂-

SiMe₂NSiMe₂CH₂P^tBu₂)CoI₂and 1 are also present due to the

secondary reaction that occurs between PNPCoI and I₂. The

reverse reaction, [κ²-(^tBu₂P(I)CH₂SiMe₂NSiMe₂CH₂P^tBu₂)CoI₂

returning to complex PNPCoI, is enabled by the addition of 2

equiv of 1. This redox reaction simply involves two molecules

of 1 undergoing one-electron oxidations at cobalt while a single

phosphonium cation in [κ²-(^tBu₂P(I)CH₂SiMe₂NSiMe₂CH₂P^t-

n

(the BDE Sn-H in Bu₃SnH at 74 kcal‚mol^-1is comparable

to that of C-Cl for benzyl chloride, 72 kcal‚mol^-1). Therefore,

as we have previously discussed,⁶the reducing power of PNPCo

is strongly dependent on not only the redox potential of cobalt

but also, equally importantly, on the energy recouped in the

formation of new Co-X bonds.

Oxidative Addition Reactions of H-H and Si-H. We

investigated the simple reaction of 1 with H₂, envisaging a

number of possible outcomes involving oxidative addition or

heterolytic activation of dihydrogen. Any homolytic cleavage

of H₂might be discounted due to the high H-H bond

dissociation energy (104.2 kcal‚mol^-1); furthermore a simple

σ complex from an intact H₂moiety was doubtful due to weak

donors displaying no propensity to bond to cobalt in 1 because

of the spin pairing penalty discussed earlier.¹Exposure of a

(9) Jewson, J. D.; Liable-Sands, L. M.; Yap, G. P. A.; Rheingold, A. L.;

(8) Jenkins, D. M.; Peters, J. C. J. Am. Chem. Soc. 2005, 127, 7148.

Theopold, K. H. Organometallics 1999, 18, 300.

9

J. AM. CHEM. SOC. VOL. 130, NO. 13, 2008 4263

A R T I C L E S

Ingleson et al.

Dihydride 3 is formally electronically and coordinatively

unsaturated but exhibits no inclination to bind in a low steric

profile to two electron donors (e.g., PhCN). There are two

complementary factors that may be inhibiting additional ligand

binding, π donation by the amide nitrogen (acting as a 3e donor),

and the strong trans effect of hydride that in any potential

product would be located trans to the incoming ligand.

Figure 3. Reversible addition of H₂to cobalt in 1. Right, the hydride region

of the ¹H NMR spectrum of PNPCo(H)₂(298 K, C₆D₆) showing the broad

triplet resonance for Co-H.

In spite of this, 1 and 3 can be envisaged as two key resting

states on a catalytic alkene hydrogenation cycle (Figure 5);

therefore their utility as homogeneous hydrogenation catalysts

was investigated.¹⁰An initial test reaction was promising, with

rapid hydrogenation of ethene occurring at room temperature

degassed C₆D₆solution of 1 to ca. 4 atm of H₂produced an

1

immediate color change to yellow/brown. The subsequent H

1

in C₆D₆, a rare occurrence for cobalt complexes. H NMR

NMR spectrum at 25 °C revealed only a partial conversion of

1 to a single new diamagnetic product that displayed C_2V

symmetry on the NMR time scale, a broad two-hydride

resonance (this intensity is repeatable over a number of

independently prepared samples) at high field (triplet, -32.2

ppm, J_H-P) 51 Hz Figure 3), and a very broad ³¹P{¹H} NMR

resonance at 89.9 ppm. These findings are consistent with the

formation of a Co(III) dihydride complex, PNPCo(H)₂, 3, that

is in dynamic equilibrium with 1, illustrated by vacuum removal

studies on this reaction mixture revealed the expected equilib-

rium mix of 1/3. Unfortunately, attempts to hydrogenate the

substituted alkenes, 1-hexene, and cyclohexene (50 equiv) with

1 under ca. 4 atm of H₂failed at room temperature. This is

attributable to the severe steric encapsulation of the cobalt center

disfavoring the alkene approach, with the 1/3 mix formed in an

identical manner to that observed in the successful hydrogena-

1

tion of ethene (by H NMR). Heating to 60 °C did initiate

limited hydrogenation (1-hexene TOF ) 1.05mol/h, cyclohexene

TOF ) 0.70mol/h); the slower rate observed for the disubstituted

alkene further substantiates steric blocking as the retarding factor

in the catalysis. It is noteworthy that, during the hydrogenation

of 1-hexene, isomerization to internal alkenes was also observed,

implying that the rates of reductive elimination and â hydride

elimination from putative PNPCo(H)(R) are comparable.

We next attempted to determine if 1 could oxidatively add a

Si-H bond and provide an analog to 3, namely PNPCo(SiR₃)H.

1 is rapidly consumed by 4 equiv of PhSiH₃, forming a single,

low-symmetry, pale-yellow, diamagnetic product; 1 equiv of

1

of H₂, yielding pure 1 (by H NMR). This process is fully

reversible, recharging with H₂, reproducing the incomplete

conversion to 3. Cooling a solution of 1 under ca. 4 atm of H₂

1

to 205 K results in nearly complete conversion to 3. The H

and ³¹P{¹H} NMR spectra at this temperature reveal no

additional fine coupling, with the inherent broadness an intrinsic

property of atoms directly bound to quadrupolar cobalt. This

frustrated attempts at determining hydride number by collecting

a

³¹P NMR spectrum with selective coupling to hydride, with

no additional fine coupling observed. Attempts to observe any

HD coupling constant by the addition of HD to 1 were also

inconclusive, with 3 formed (by ³¹P{¹H} NMR) but the broad

hydride resonance still only a triplet; the broad nature of this

resonance (13 Hz fwhm) could easily conceal a small HD

coupling that would be consistent with a stretched dihydrogen

complex.

1

H₂and Ph₂SiH₂were observed as byproducts (by H and ²⁹Si

NMR). Identification was only forthcoming when crystals

suitable for X-ray diffraction analysis were obtained from the

slow evaporation of a 1:1 (% v/v) pentane/Me₄Si solution. This

revealed the addition of 3 equiv of PhSiH₃to the cobalt center

of 1, yielding {κ²-^tBu₂PCH₂Me₂SiNSiMe₂CH₂Bu₂P(H)Si)}-

t

Extensive DFT(PBE) calculations were therefore performed

on all possible products from the reaction between 1 and excess

H₂. Figure 4 shows all of the calculated structures with singlet

ground states and the respective electronic energies of their

formation from 1 and H₂.

Co(H)₃(SiH₂Ph)₂, 4 (Figures 6 and 7). The structure consists of

a formally Co(V) center ligated by three hydrides (freely refined

from the penultimate Fourier difference map), two primary

silanes, a phosphine, and a base-(R₃P)-stabilized silylene formed

by cleaving one Si-C bond and two Si-H bonds of one silicon.

The Co-Si bond distance of the silylene (2.1848(8) Å) is shorter

than the two pure σ bonded Co-Si(Ph)H₂groups (2.2622(9)

and 2.2477(8) Å), confirming some multiple bond character in

the Co-Si3 bond. The pure σ Co-Si bonds are comparable to

the only other Co(V)-silane complex structurally character-

ized,¹¹Cp*Co(H)₂(SiHPh₂)₂, (Co-Si ) 2.2492(21) and 2.2536-

(19) Å) and as expected (due to the reduced ionic radius of

cobalt in the +5 oxidation state) are slightly shorter than Co-

(III)-Si bonds (2.28-2.30 Å).¹¹The Si3-P2 bond length is

fully consistent with other phosphine stabilized silylenes,^12,13

confirming that it is a dative bond. The Si4-Si5 distance of

Complexes with an intact, σ-bound H₂ligand are not

stationary states, with optimization of singlet PNPCo(H₂)

converging to 3, whereas triplet PNPCo(H₂) dissociates H₂

during the geometry optimization, generating 1 and H₂. Inspec-

tion of the relative electronic energies shows (after the entropic

factors of the addition of one/two molecules of H₂are

considered) that there are only two feasible products, 3 and B.

Hypothetical B can be ruled out due to the failure to observe

an N-H resonance in the ¹H NMR spectrum and the observation

of only one hydride environment even at 205 K. The calculations

therefore support our product assignment as the classical

dihydride 3 and are moreover consistent with the observed near-

thermoneutral conversion of 1 to 3. The calculated structural

metrics of 3 revealed a H7-H8 distance (1.63 Å) that indicates

H/H bond scission and a formal cobalt +3 oxidation state. The

geometry around cobalt approximates to a trigonal bipyramid

(or Y-shaped), with the major distortion the compression of the

H7-Co-H8 angle (67.9°).

(10) Knijnenburg, Q.; Horton, A. D.; van der Heijden, H.; Kooistra, T. M.;

Hetterscheid, D. G. H.; Smits, J. M. M.; de Bruin, B.; Budzelaar, P. H.

M.; Gal, A. W. J. Mol. Catal. 2005, 232, 151.

(11) Brookhart, M.; Grant, B. E.; Lenges, C. P.; Prosenc, M. H.; White, P. S.

Angew. Chem., Int. Ed. 2000, 39, 1676.

(12) Gusev, D. G.; Fontaine, F.-G.; Lough, A. J.; Zargarian, D. Angew. Chem.,

Int. Ed. 2003, 42, 216.

(13) Kawamura, K.; Nakazawa, H.; Miyoshi, K. Organometallics 1999, 18, 1517.

9

4264 J. AM. CHEM. SOC. VOL. 130, NO. 13, 2008

Redox Chemistry of the Triplet Complex (PNP)Co^I

A R T I C L E S

Figure 4. DFT geometry optimized structures for all of the possible complexes with a singlet ground state from the reaction between PNPCo and excess

H₂. Inset, the respective electronic energies for each conversion from 1.

quadrupolar cobalt, and the second split into a doublet (³J_P-P

) 16 Hz) by coupling to phosphine. The ¹H NMR is complex,

as expected for a low-symmetry molecule; the salient points

are the observation of five Si-H resonances, consistent with

diasterotopic Si-H on both Co-SiH₂Ph groups. The fifth

resonance, the silylene Si-H is split into a doublet (²J_P-H

)

50 Hz) due to coupling to phosphorus (collapses to a singlet in

1

the H{³¹P} NMR spectrum). A broad hydride resonance of

intensity three with no observable fine coupling is present at

-11.29 ppm. The high coordination number (7) leads to rapid

fluxionality (the slow hydride exchange regime is not reached

even at -70 °C), but the fluxionality is selective for hydride

migration among inequivalent sites and does not involve site

exchange between the two SiH₂Ph ligands. Furthermore, there

is no exchange between any Si-H and a hydride on cobalt.

Formulation of 4 as a Co(V) center based on X-ray diffraction

data will always prove contentious, especially involving the

possibility of classical/nonclassical hydrides. A number of

experimental findings are relevant to the situation in 4: (i) the

observed lack of exchange between Co-H and free H₂(or D₂),

(ii) no exchange between Si-H and Co-H (precluding a Co-

(III){η²-SiH₂Ph}), (iii) the absence of H₂evolution on extended

evacuation, and (iv) no reaction (and specifically no H₂

evolution) between 4 and PhCN. All of these, combined with

the existence¹¹of a related Co(V) complex, ligated by silyl and

hydride groups, support a Co(V) center in 4.

Figure 5. Proposed catalytic hydrogenation cycle commencing from 1.

3.726 Å is definitively nonbonding, confirming two PhH₂Si^-

ligands, in contrast to a recent ambiguous case^14,15where close

Si-Si distances imply a lower oxidation state at the metal. The

three inter-hydride distances are all greater than 2.0 Å, sup-

porting a classical hydride structure. A closely related polyhy-

drido-based stabilized silylene osmium complex involving a PCP

pincer ligand has also been recently reported¹²and also forms

by the facile activation of numerous Si-H and Si-C bonds.

The NMR spectra for 4 are fully consistent with this

formulation. The ²⁹Si NMR spectrum (Figure 8) displays five

resonances of the expected multiplicity for ¹J coupling to protons

(two triplets for inequivalent PhSiH₂, one doublet for the base-

stabilized silylene, and two singlets for ligand backbone SiMe₂);

the three multiplets all collapse to singlets on proton decoupling.

The silylene chemical shift is at least ∼50 ppm shifted downfield

compared to the cobalt-silyl groups as expected. The ³¹P{¹H}

NMR displays two resonances, one significantly broadened

(concealing any fine coupling) due to direct bonding to

The formation of 4 was unexpected, and the multistep reaction

(Figure 7) occurs despite the mild conditions (25 °C). The

reaction shows Si-C bond cleavage (ultimately forming Ph₂-

SiH₂), multiple Si-H additions to cobalt, silylene formation,

followed by phosphine migration from cobalt to silylene, and

oxidation of d⁸Co(I) to d⁴Co(V). This complex reaction at

25 °C proceeds via an observable brown intermediate that can

be characterized by repeating the reaction at -78 °C. The NMR

spectra of this intermediate revealed a low-symmetry product,

with silylene formation already evident by the characteristic ³¹P-

{¹H} NMR spectrum. Thus, the formation of the silylene moiety

involving Si-C and multiple Si-H activation is extremely

(14) Chen, W.; Shimada, S.; Tanaka, M. Science 2002, 295, 308.

(15) Crabtree, R. H. Science 2002, 295, 288.

9

J. AM. CHEM. SOC. VOL. 130, NO. 13, 2008 4265

A R T I C L E S

Ingleson et al.

t

Figure 6. Left, the conversion of PNPCo to 4, right ORTEP view (50% probabilities) of 4, Bu carbons and all hydrogens aside from Co-H and silylene

Si-H are omitted for clarity. Selected bond lengths (Å) and angles (°): Co-P1, 2.267(8); Co-Si3, 2.1848(8); Co-Si4, 2.2622(9); Co-Si5, 2.2477(8);

P2-Si3, 2.3783(10), N1-Si3, 1.774(2); Co-Si3-N1, 124.83(8); Co-Si3-P2, 122.94(4); P2-Si3-N1, 92.68(7).

Figure 7. Reaction scheme displaying the reaction of PNPCo with 2 equiv of PhSiH₃to generate intermediate 5 that upon addition of further PhSiH₃leads

to complete conversion to the structurally characterized 4.

downfield corresponding to the silylene moiety), and two

singlets (for the SiMe₂ligand backbone); both doublets collapse

to singlets on proton decoupling. A structure fully consistent

with the spectroscopic data is a close analog of 4. 5 is also a

Co(V) complex and retains the Ph₂HSi fragment that is

prearranged to undergo reductive elimination of diphenylsilane.

Similar facile Si-H/M-R redistributions are well documented.¹²

Addition of further PhSiH₃to 5 does indeed produce 4 along

with 1 equiv of Ph₂SiH₂(and presumably H₂). 5 can be cleanly

synthesized by the addition (at -78 °C) of 2 equiv of PhSiH₃

to (PNP)Co. Both 4 and 5 are indefinitely stable in the solid

state under argon but decompose completely in toluene solutions

(at 25 °C) within 24 h to generate intractable mixtures. Taken

together, these observations show an extremely low barrier to

multiple silane activation and insertion into P-Co and N-Co

bonds, whereas the observed formation of 4 and 5 implies that

both contain an electrophilic silylene intermediate that is

attacked by the nucleophilic phosphine, with the formation of

the new Si-N bond providing the overall driving force for this

conversion.

Figure 8. The ²⁹Si and ²⁹Si{¹H} NMR spectra of, top, 4 (280 K, C₆D₆, *

The purity observed in the complex transformation of 1 to 4

) resonances due to Ph₂SiH₂), and bottom, 5 (263 K d₈-toluene).

(which is quantitative by NMR) is explained by the lack of

reactivity we independently observed between Ph₂SiH₂and 1.

Simply moving from a primary to a secondary silane switches

off all reactivity, with the steric repulsion now sufficient to

facile, occurring even at -78 °C. No simple oxidation product

PNPCo(H)(SiPhH₂) is observed at -78 C.

An identical intermediate is observed in the addition of 1

prevent the first step on this cascade reaction. 1 also failed to

equiv of PhSiH₃to PNPCo at -78 °C, which on warming to

react with Et₃SiH.

25 °C produces one new diamagnetic product (along with

unreacted 1). In the ¹H NMR, this product, 5, displays an integral

four-hydride resonance along with three signals in the Si-H

region (consisting of one singlet and one doublet for the silylene

The higher oxidation states observed for cobalt in 4, 5, and

PNPCo(H)₂combined with the inability to exceed the Co(II)

oxidation state with conventional oxidants reveal that hydride

and silyl ligands are integral to reaching PNPCo(III). This may

in part arise because these ligands allow for the formation of

new bonds to cobalt without significantly reducing the electron

2

proton, J_P-H) 49 Hz); the ³¹P{¹H} NMR is similar to that

observed for 4. The ²⁹Si NMR spectrum is especially informative

(Figure 8), revealing four resonances, two doublets (one shifted

9

4266 J. AM. CHEM. SOC. VOL. 130, NO. 13, 2008

Redox Chemistry of the Triplet Complex (PNP)Co^I

A R T I C L E S

Figure 9. Left, possible terminal nitrogen η¹coordination modes, A and B as a neutral ligand and C and D as a doubly negative ligand. Right, ORTEP view

(50% probabilities) of 6 (most hydrogens are omitted for clarity and disorder in the PNP ligand is not shown). Inset, key structural metrics of the diazo

functionality. Selected bond lengths (Å) and angles (°): Co-N1, 1.9148(15); Co-N2, 1.7519(16); N2-N3, 1.183(2); N3-C23, 1.331(2); Co-P1, 2.262(1);

Co-P2, 2.249(1); N1-Co-N2, 175.56(6); N1-Co-P1, 90.121(5); P1-Co-P2, 170.90(16); Co-N2-N3, 171.47(14), N2-N3-C23; 146.47(18); N3-

C23-Si3, 120.53(15).

density (i.e., higher formal oxidation state) at the metal center

due to their electropositive nature. Indeed, these ligands (and

alkyls) are pervasive in metal complexes in high formal

oxidation numbers (e.g., Cp*ReH₆, Cp*IrH₄, CpIrH₂(SiR₃)₂)^16-19

due to their low electronegativity, resulting in metal ligand bonds

with a significant covalent contribution thus rendering the

oxidation-state formalism meaningless.

discernible, suggesting an intact diazoalkane unit, a fact

supported by the observation of an intense peak in the infrared

spectrum at 2069 cm^-1that corresponds to a coordinated

1

cumulene. The H NMR spectrum revealed the formation of a

single new diamagnetic C_2Vsymmetric product that displayed

an unusual resonance for the unique methine C-H of the diazo

functionality at -3.04 ppm. The dramatic upfield shift of this

resonance can be attributed to significant carbanionic character

at carbon (structures B or C in Figure 9). It is important to note

that structures B and C differ by a change in the oxidation state

at the metal, which emphasizes the ability of diazoalkanes to

act as a σ donor and a π acid, even to the extent of accepting

two electrons and forming a metal-imido complex. In this case,

a significant contribution from structure D can be discounted

because this produces diazoalkane methine C-H resonances

in the range of 7 to 9 ppm. Unambiguous characterization of

the diazoalkane binding mode was provided by an X-ray

diffraction study (Figure 9) on crystals grown from (Me₃Si)₂O

at -40 °C, which revealed the formation of PNPCo(N₂C(H)-

SiMe₃), 6, with the diazo moiety intact and bound in an η¹

fashion through the terminal nitrogen.

Synthesis of Transient Electrophilic CodE and Co≡E (E

) CR₂, NR, N, and O). There has been significant recent

progress in isolating stable mid-to-late 3d transition metal

complexes with multiple bonds to simple ligands (e.g., MdO,

MdS, MdNR, MtN, MdCR₂, MdPR).^20-28Interest in

molecules of this nature stems from their proposed involvement

as key intermediates in a number of important catalytic atom/

group transfer processes.^29-33The ancillary ligand environment

has proven to be the key to this success, with bulky, electron-

rich ligands (two properties the PNP ligand certainly satisfies)

a prerequisite for accessing these compounds. Furthermore,

PNPCo has been shown to achieve the +3 oxidation state with

certain reagents, so it is feasible that it would be able to support

multiple bonds to these ligands. This applies particularly to

carbenes, with their low-electronegativity carbon center and their

close relationship to silylenes.

The geometry around cobalt approximates to square planar,

with unremarkable bond lengths and angles associated with the

PNP ligand. More interestingly, the distances and angles

associated with the diazo moiety do not conform to any of the

Lewis structures shown in Figure 9, with a short Co-N2 bond

(in comparison to Co-N1), short N-N and N-C bonds, an

essentially linear N3-N2-Co angle, and an angle of 146.47-

(18)° at N3. Inspection of the space-filling diagram for 6 reveals

no close contacts between the diazo SiMe₃and any of the four

ligand ^tBu groups; therefore, steric impact will have a minimal

effect on the diazoalkane binding motif; thus electronic factors,

specifically the degree of metal-to-ligand electron transfer is

the dominant factor. Comparison of the Co-N2 bond length

(1.7519(16) Å in 6) with a related complex (PhB(CH₂PPh₂)₃-

Co-N₂CPh₂),³⁹where full two-electron metal oxidation occurs

a. N₂CHR.^34-38Addition of a stoichiometric equiv of N₂C-

(H)SiMe₃to a toluene solution of 1 produced an immediate color

change from green to dark yellow. No gas evolution was

(16) Herrmann, W. A.; Okuda, J. Angew. Chem. 1986, 98, 1109.

(17) Gilbert, T. M.; Bergman, R. G. Organometallics 1983, 2, 1458.

(18) Fernandez, M. J.; Maitlis, P. M. Organometallics 1983, 2, 164.

(19) Shortland, A. J.; Wilkinson, G. J. Chem. Soc., Dalton Trans. 1973, 872.

(20) Mindiola, D. J.; Hillhouse, G. L. J. Am. Chem. Soc. 2002, 124, 9976.

(21) Vicic, D. A.; Jones, W. D. J. Am. Chem. Soc. 1999, 121, 4070.

(22) Mindiola, D. J.; Hillhouse, G. L. J. Am. Chem. Soc. 2001, 123, 4623.

(23) Melenkivitz, R.; Mindiola, D. J.; Hillhouse, G. L. J. Am. Chem. Soc. 2002,

124, 3846.

(24) MacBeth, C. E.; Thomas, J. C.; Betley, T. A.; Peters, J. C. Inorg. Chem.

2004, 43, 4645.

(25) Eckert, N. A.; Vaddadi, S.; Stoian, S.; Lachicotte, R. J.; Cundari, T. R.;

Holland, P. L. Angew. Chem., Int. Ed. 2006, 45, 6868.

(26) Gregory, E. A.; Lachicotte, R. J.; Holland, P. L. Organometallics 2005,

24, 1803.

(27) Badiei, Y. M.; Krishnaswamy, A.; Melzer, M. M.; Warren, T. H. J. Am.

Chem. Soc. 2006, 128, 15056.

(34) Dartiguenave, M.; Joelle Menu, M.; Deydier, E.; Yves, D.; Siebald, H.

Coord. Chem. ReV. 1998, 178-180, 623.

(35) Hillhouse, G. L.; Haymore, B. L. J. Am. Chem. Soc. 1982, 104, 1537.

(36) Polse, J. L.; Kaplan, A. W.; Andersen, R. A.; Bergman, R. G. J. Am. Chem.

Soc. 1998, 120, 6316.

(37) Cohen, R.; Rybtchinski, B.; Gandelman, M.; Rozenberg, H.; Martin, J. M.

L.; Milstein, D. J. Am. Chem. Soc. 2003, 125, 6532.

(38) Guillemot, G.; Solari, E.; Floriani, C.; Rizzoli, C. Organometallics 2001,

20, 607.

(28) Dai, X.; Warren, T. H. J. Am. Chem. Soc. 2004, 126, 10085.

(29) Hu, X.; Meyer, K. J. Am. Chem. Soc. 2004, 126, 16322.

(30) Zhao, P.; Krug, C.; Hartwig, J. F. J. Am. Chem. Soc. 2005, 127, 12066.

(31) Thyagarajan, S.; Shay, D. T.; Incarvito, C. D.; Rheingold, A. L.; Theopold,

K. H. J. Am. Chem. Soc. 2003, 125, 4440.

(32) Shay, D. T.; Yap, G. P. A.; Zakharov, L. N.; Rheingold, A. L.; Theopold,

K. H. Angew. Chem., Int. Ed. 2005, 44, 1508.

(33) Waterman, R.; Hillhouse, G. L. J. Am. Chem. Soc. 2003, 125, 13350.

9

J. AM. CHEM. SOC. VOL. 130, NO. 13, 2008 4267

A R T I C L E S

Ingleson et al.

Figure 10. Decomposition of 6 and 7 on heating, with two proposed, but undetected, intermediates.

(Co-N ) 1.667(2)Å), indicates an intermediate situation in 6

where there is metal-to-diazo electron transfer but short of full

two-electron transfer, with its implied oxidation of cobalt to

+3. The diazo group in 6 therefore lies somewhere on the

continuum between Lewis structures A and C, with this

intermediate diazo geometry a direct result of the reluctance of

cobalt in PNPCo to undergo full oxidation to Co(III).

rhodium-carbene complex.³⁷The kinetic barrier observed in

the reactivity of 6 and 7 is due to the requirement of a

rearrangement from the nitrogen-bound to the carbon-bound

diazoalkane isomer, which has been previously calculated to

be the key intermediate for N₂evolution and concomitant

carbene formation. It is noteworthy that Peters’ imido complex³⁹

PhB(CH₂PPh₂)₃Co-N₂CPh₂is indefinitely stable, presumably

due to a stronger Co-N bond preventing the required diazoal-

kane rearrangement, though steric inhibition of the C-bound

isomer due to the presence of two phenyl groups may also

contribute.

In an analogous reaction, the addition of 1 equiv of Ph(H)-

1

CN₂to 1 quantitatively (by H NMR) led to a single new

diamagnetic product, also possessing local C_2Vsymmetry and

a diazoalkane methine resonance shifted significantly upfield

to -0.83 ppm. The lack of any gas evolution and the observation

of a peak in the infrared spectrum corresponding to a coordinated

cumulene led to the assignment of PNPCo(N₂C(H)Ph), 7. The

only significant difference in the spectroscopic data between 6

and 7 (apart from those associated with the change in diazo R

group) is a dramatic shift in the infrared vibration for the

coordinated diazoalkane (from 2069 cm^-1in 6 to 1592 cm^-1

in 7). This must correlate to some degree of change in the

bonding within the diazoalkane moiety in 7 because the infrared

stretch of structurally characterized diazoalkane complexes only

shifts by a maximum of ca. 100 cm^-1on changing the

diazoalkane substituents significantly (i.e., from N₂CPh₂to

N₂C(CO₂Et)₂, while maintaining the same metal complex and

oxidation state). Despite repeated attempts, material suitable for

single-crystal X-ray diffraction was not obtainable, frustrating

definitive assignment of the bonding situation in the diazo

b. RN₃. It is well documented that metal complexes that react

with diazoalkanes to generate metal-imido species will in

contrast react with organic azides to liberate N₂and produce

metal-nitrene complexes.^38-46This distinct reactivity arises

from a reduction in sterics and substitution of carbon by more-

electronegative nitrogen, resulting in a greater negative charge

at the key atom and thus a lower energy for the key intermediate.

Targeting a metal-nitrene complex, PNPCo(III)NPh, a sto-

ichiometric equiv of phenyl azide was added at -78 °C to 1 in

pentane. Subsequent warming to room temperature led to gas

evolution and to the formation of a dark-yellow solution, which

on concentrating and cooling to -40 °C for 18 h yielded dark-

1

yellow crystals in moderate yield. The H NMR spectrum of

these crystals revealed the formation of a new paramagnetic

complex with local C_ssymmetry. Relevant precedents for

ground-state triplets have been reported.^47,48The solution

magnetic moment (Evans Method) at 298 K of 3.40µ_Bindicated

a triplet, S ) 1 cobalt center, consistent with either Co(I) or

Co(III). The ³¹P{¹H} NMR spectrum displayed one broad,

paramagnetically shifted resonance at 206.8 ppm, suggesting

that at least one phosphine was no longer bound to the para-

magnetic cobalt center. This was confirmed by an X-ray

diffraction study, which revealed the formation of PN(PdNPh)-

Co, 8 (Figure 11), a Co(I) complex with a P(V) phosphinimide

functionality;⁴⁹thus azide-to-nitrene transformation has occurred

but the ultimate locus of oxidation by nitrene is at phosphorus,

not cobalt.

1

fragment. The H and ¹³C{¹H} NMR resonances for the diazo

methine group (-0.83 and 41.5 ppm, respectively) eliminate

structure D as a possibility; redox innocent binding (i.e.,

structure A) can also be discounted because a related complex,³⁷

(PCP)RhN₂C(H)Ph (PCP ) 1,3-C₆H₃-(CH₂PⁱPr₂)₂), has a

1

methine resonance in H NMR at +3.76 ppm (close to free

PhC(H)N₂4.8 ppm). Thus, a structure related to C but distinct

from that observed in 6 is probable.

In solution, 6 and 7 are both stable at 25 °C for weeks;

attempts to activate N₂release by the addition of a Lewis acid

catalyst (Sm(OTf)₃)²⁰resulted in no reaction after 48 h at

25 °C. Gentle heating (Figure 10, 60 °C in C₆D₆), however,

results in the rapid consumption of the starting diazoalkane

complex (no starting material remained after 4 h), with 1 being

The geometry around cobalt in 8 is still planar (angles sum

to 360°), though the expansion of one of the five-membered

(40) Bart, S. C.; Lobkovsky, E.; Bill, E.; Chirik, P. J. J. Am. Chem. Soc. 2006,

128, 5302.

1

the major PNPCo-containing product detected (by H NMR);

(41) Betley, T. A.; Peters, J. C. J. Am. Chem. Soc. 2004, 126, 6252.

(42) Matsunaga, P. T.; Hess, C. R.; Hillhouse, G. L. J. Am. Chem. Soc. 1994,

116, 3665.

(43) Proulx, G.; Bergman, R. G. Organometallics 1996, 15, 684.

(44) Brown, S. D.; Betley, T. A.; Peters, J. C. J. Am. Chem. Soc. 2003, 125,

322.

other minor paramagnetic products are also present but have

proven to be intractable. The products derived from the

diazoalkane for both 6 and 7 are a mixture of the alkene and

1

the azine products (by H NMR); at no stage is a diamagnetic

(45) Dai, X.; Kapoor, P.; Warren, T. H. J. Am. Chem. Soc. 2004, 126, 4798.

(46) Kogut, E.; Wiencko, H. L.; Zhang, L.; Cordeau, D. E.; Warren, T. H. J.

Am. Chem. Soc. 2005, 127, 11248.

PNPCo(carbene) complex detected in either case. This is in

distinct contrast to a related 4d congener (PCP)RhN₂C(H)Ph

that on warming evolves N₂and cleanly forms a stable

(47) Bart, S. C.; Chlopek, K.; Bill, E.; Bouwkamp, M. W.; Lobkovsky, E.; Neese,

F.; Wieghardt, K.; Chirik, P. J. J. Am. Chem. Soc. 2006, 128, 13901.

(48) Rohde, J.-U.; Stubna, A.; Bominaar, E. L.; Muenck, E.; Nam, W.; Que,

L., Jr. Inorg. Chem. 2006, 45, 6435.

(39) Jenkins, D. M.; Betley, T. A.; Peters, J. C. J. Am. Chem. Soc. 2002, 124,

11238.

(49) Dehnicke, K.; Straehle, J. Polyhedron 1989, 8, 707.

9

4268 J. AM. CHEM. SOC. VOL. 130, NO. 13, 2008

Redox Chemistry of the Triplet Complex (PNP)Co^I

A R T I C L E S

Figure 11. Left, proposed mechanism for the formation of 8. Right, ORTEP view (50% probabilities) of 8 (hydrogens omitted for clarity). Selected bond

lengths (Å) and angles (°): Co-N1, 1.9763(18) Å; Co-P2, 2.1838(6); Co-N2, 1.9866(6); N2-P1, 1.6043(6); P2-Co-N2, 153.77(5); P2-Co-N1, 97.04-

(5); N2-Co-N1, 109.6(7).

rings to six-membered by nitrene insertion results in a distortion

away from the T-shaped structure observed for 1. The N1 and

N2 bond lengths are similar, confirming both as single Co-N

bonds, whereas the P1-N2 distance is consistent with a double

bond. The P2-Co bond length (2.1838(6) Å) is contracted

somewhat when compared to those in compound 1 (Co-P1,

2.2281(6); Co-P2, 2.2237(6) Å); this presumably arises due

to the reduced steric crowding around cobalt allowing for a

shorter P2-Co bond to compensate for the poorer electron

donating ability of the phosphinimide ligand. Similar nitrene

Figure 12. Left, DFT geometry optimized structure for triplet PNPCo-

(NPh), selected bond lengths (Å) and angles (°): Co1-P2, 2.384; Co1-

P3, 2.397; Co1-N4 1.855, Co1-N5, 1.830; P2-Co1-P3, 178.3; P2-Co1-

N4, 89.9; N4-Co1-N5, 177.5; P2-Co1-N5, 89.4; C6-N5-Co1, 169.3.

Right, DFT geometry optimized structure for singlet PNPCo(NPh), selected

bond lengths (Å) and angles (°): Co1-P2, 2.65; Co1-P3, 3.002; Co1-

N4, 1.879; Co1-N5, 1.666; P2-Co1-N4, 94.7; P2-Co1-N5, 109.6; N4-

Co1-N5, 144.3; C6-N5-Co1, 153.8.

insertions into ancillary ligands have been reported, one

example²⁹of particular relevance involving a Co(III) imido

complex with an electrophilic nitrene group that inserts into a

cobalt-carbene bond. A related mechanism can be envisaged

here where a nucleophilic phosphine attacks a transient cobalt-

nitrene complex (Figure 11), generating the observed Co(I)

phosphinimide complex. Support for this mechanism is forth-

coming from a low-temperature NMR study. Combination of

stoichiometric equivalents of 1 and PhN₃at -78 °C in d₈-toluene

followed by warming to -60 °C resulted in a color change from

green to brown, and the ¹H NMR spectrum revealed the

formation of a single new diamagnetic product with C_2V

symmetry and a relatively sharp phosphine resonance at 54.0

ppm in the ³¹P{¹H} NMR spectrum. This phosphorus chemical

shift is extremely close to that observed for the coordinated

diazoalkane complexes (52.3 and 48.5 ppm) and combined with

the solution C_2Vsymmetry lead us to assign this as the intact

phenyl azide complex PNPCo(N₃Ph). Warming this solution to

-40 °C results in a second color change from brown to deep

green, and NMR spectroscopy reveals the complete transforma-

tion of this diamagnetic compound to a single new paramagnetic

complex with three significantly broadened resonances in the

¹H NMR, implying local C_2Vsymmetry; no phosphorus reso-

nance is now detectable. Repeating this experiment outside the

NMR probe allowed for the visual detection of gas evolution

at this temperature, confirming N₂loss. We tentatively assign

this new paramagnetic complex as the Co(III) imido complex

PNPCo(NPh), which on further warming to -20 °C results in

the rapid conversion to 8 (the reaction is complete within 10

min at this temperature). The narrow temperature stability

window of PNPCo(NPh) has frustrated isolation of material for

an X-ray diffraction study and subsequent unambiguous char-

acterization.

below the singlet isomer; this supports the assignment of the

low-temperature paramagnetic intermediate observed as PNPCo-

(NPh). The ligand environment around cobalt in triplet PNPCo-

(NPh) is square planar (angles sum to 360°), a geometry with

good precedent for four-coordinate triplet d⁶cobalt. The most

noteworthy structural metric in triplet PNPCo(NPh) is the imido

Co-N5 bond length (1.830 Å), which when compared to the

cobalt-amide bond distance of 1.855 Å strongly suggests a bond

order of close to 1. Cobalt imido complexes have the potential

to have bond orders up to 3, involving one σ bond and two

orthogonal π bonds. Structurally characterized, stable Co^(III)

-

imido complexes consistently exhibit short Co-N bonds,

indicative of significant multiple bond character, for example,

1.658(2)Å in PhB(CH₂PPh₂)₃Co(N-p-tolyl)³⁹and 1.624(4)Å in

[â-diketiminato]CoN-Adamantyl,⁴⁵whereas even in the ther-

mally sensitive complex [TIMEN^xylCo(N-p-OMe-C₆H₄)][BPh₄]

(TIMEN ) tris-[2-(3-arylimidazol-2-ylidene)ethyl]amine, xyl

) 2,6-dimethylphenyl)²⁹the Co-N distance is 1.675(2)Å. It is

no coincidence that these cobalt-imido complexes with their

short CodN distances are all low spin, S ) 0, d⁶Co(III)

complexes; spin pairing provides an additional empty d orbital

that is available to participate in the stabilizing nitrogen-to-cobalt

π donation. Triplet PNPCo(NPh), with an additional d orbital

singly occupied (and therefore not available to participate in π

bonding with the nitrene fragment) has a drastically elongated

Co-N bond, which presumably contributes to its reactive nature.

This requirement for sufficient empty d orbitals is exemplified

by a second singlet stationary point (only 0.8 kcal‚mol^-1further

above the most stable singlet isomer); this now has a calculated

imido Co-N bond length of 1.666 Å. Thus, in mid-to-late

Density functional theory (PBE) calculations were performed

(Figure 12) using the complete ligand set to investigate the

structure and frontier orbitals of PNPCo(NPh). The ground state

is calculated to be S ) 1, triplet PNPCo(NPh), 9.7 kcal‚mol^-1

9

J. AM. CHEM. SOC. VOL. 130, NO. 13, 2008 4269

A R T I C L E S

Ingleson et al.

Figure 13. Calculated frontier molecular orbitals for singlet PNPCo(NPh).

Figure 14. Reaction scheme for the combination of galvinoxyl radicals with (PNP)Co[NH(2,6-Me₂-C₆H₃)].

transition metals low-spin ground states would appear to be a

prerequisite for the formation of stable MdE bonds. The singlet

state of PNPCo(NPh) is energetically close to the triplet and

thus may be mechanistically relevant. Further examination of

the structure and frontier orbitals reveals a number of key

points: (1) the singlet PNPCo(NPh) (Figure 12) structure

exhibits a dramatic lengthening (indicating complete Co/P

dissociation) of one P-Co bond length (Co1-P3, 3.002 Å),

resulting in a more nucleophilic phosphorus; (2) the LUMO

for the singlet (Figure 13) reveals significant imide nitrogen

character, therefore an electrophilic N(Ph). These two factors

explain the origin of the observed low-energy barrier to P/N

bond formation.

c. Other paths to an Arylimide. With the cobalt-anilide

complex (PNP)Co[NH(2,6-Me₂-C₆H₃)] easily accessible,¹we

attempted to synthesize a PNPCo(imido) species via hydrogen-

atom abstraction (stepwise, via electron, then proton transfer),

a methodology that has been previously successful in generating

nickel-imido complexes.^20,33,50Reaction of (PNP)Co[NH(2,6-

Me₂-C₆H₃)] with the free radical TEMPO at 25 °C for 24 h

and then at 60 °C for a further 24 h causes no reaction (by ¹H

NMR). However, addition of 1 equiv of galvinoxyl radicals

(O-H bond strength 70 kcal/mol⁵¹) to (PNP)Co[NH(2,6-Me₂-

C₆H₃)] in benzene at 25 °C resulted in the rapid and clean

formation of 2 (Figure 14), PNPCo(galvinoxyl) (identified by

its characteristic ¹H NMR); the galvinoxyl free radical has been

reduced to an aryloxide but no concomitant oxidation of cobalt

has occurred. Analysis of the nitrogen-containing products from

this reaction revealed that the only amide derived species was

2,6-dimethyl aniline (by ¹H NMR comparison with commercial

2,6-dimethylaniline); no hydrazine or azo compounds were

detectable. The exact one-to-one stoichiometry of the reaction

is further confirmed by the absence of any hydrogen-atom

abstraction product; namely, the alcohol of galvinoxyl or a

radical coupled product galvinoxyl-N(H)(3,5-Me₂-C₆H₃). The

reaction proceeds identically in benzene or THF, and an attempt

to identify any intermediates by a low-temperature NMR study

in d₈-toluene was uninformative with full conversion to 2 even

at -70 °C, consistent with a rapid radical process.

The formation of 2,6-dimethyl aniline in this reaction suggests

that the locus of oxidation is at the amide nitrogen, expelling

an aminyl radical, Ar(H)N^·, which then abstracts a hydrogen

atom, presumably from the solvent. Without the observation of

any intermediates and/or the identification of any solvent derived

products, any further discussion is not warranted.

d. A Terminal Nitride? Following the success in accessing

PNPCodNR complexes (albeit with a limited temperature

stability range), we next targeted a cobalt nitride complex of

the general formula PNPCo^XN; of particular interest was the

case where X ) +5 because this would then result in an

It is worth noting that similar findings have been previously

reported by Peters et al.,²⁴notably that, in the DFT optimized

1

structure of [PhBPⁱPr₃]NiN^tBu, a d⁷S ) /₂Ni(III) complex,

there is one P-Ni distance significantly elongated. Of more

relevance to this study is the calculated structure of a hypotheti-

cal d⁶Ni(IV) nitride complex, [PhBPⁱPr₃]NitN, which opti-

mizes with nitride inserted into a P-Ni bond. The major

destabilizing effect in [PhBPⁱPr₃]NitN arises from the presence

of two unpaired electrons in high-energy orbitals (a triplet

ground state is likely considering the small energy difference

between the three high-energy orbitals), limiting π bonding

between nitrido nitrogen and nickel.

A preliminary investigation into the reactivity of 8 reveals

disparate behavior when compared to the related Co(I) 1.

Attempts to oxidize the Co(I) center of 8 with excess H₂at

25 °C failed, with no reaction observed after 24 h; this confirms

what is intuitively expected, that conversion of a ^tButyl

phosphine ligand to a phosphinimide reduces the electron density

at cobalt, thereby raising its oxidation potential. An attempt to

react 8 with a second equiv of PhN₃resulted in an intractable

mixture, possibly due to the instability of Co(I) bound by two

weaker electron donor phosphinimides. The reaction of a bulkier

organic azide with 1 was attempted with the hope that unfavor-

able steric interactions would help prevent nitrene insertion into

the P-Co bond; however, the addition of adamantyl azide to 1

(50) Cowley, R. E.; Bontchev, R. P.; Sorrell, J.; Sarracino, O.; Feng, Y.; Wang,

H.; Smith, J. M. J. Am. Chem. Soc. 2007, 129, 2424.

(51) Gupta, R.; Borovik, A. S. J. Am. Chem. Soc. 2003, 125, 13234.

1

resulted in no reaction (after 18 h at 25 °C, by H NMR and

IR).

9

4270 J. AM. CHEM. SOC. VOL. 130, NO. 13, 2008

Redox Chemistry of the Triplet Complex (PNP)Co^I

A R T I C L E S

Figure 15. Left, schematic for the formation of 9 from the oxidation of PNPCoN₃by the ferrocenium cation. Right, ORTEP view (50% probabilities) of

one of the two independent units (which are effectively identical) in the asymmetric unit. Selected bond lengths (Å) and angles (°): Co-N1, 1.856(4);

Co-N2, 1.899(4); Co-C14, 2.041(6); Co-P1, 2.2927(16); P2-C14, 1.778(6); P2-N2, 1.613(5); Si2-N2, 1.720(5); Si2-N1, 1.714(5); Si1-N1, 1.718(4);

Si1-C3, 1.892(6); N1-Si2-N2, 92.8(2); N2-P2-C14, 97.6(2), N2-Co-C14, 80.7(2); N2-Co-N1, 83.0(2); C14-C0-P1, 105.27(14) P2-N2-Si2,

62.8(3).

isoelectronic compound to the recently published⁵²Ru(IV)

complex, PNPRuN. Initial attempts to install a nitride func-

tionality involved the reaction of (PNP)CoCl with a stoichio-

metric equivalent of Li(dbabh) (dbabh ) 2,3:5,6-dibenzo-7-aza

bicyclo[2.2.1]hepta-2,5-diene) in THF (a reagent successfully

used for this purpose by the Cummins and Peters groups).^41,53

However, even after long reaction times (48 h) there is no

evidence for any amide formation, with only unreacted (PNP)-

CoCl present by ¹H NMR; the lack of metathesis observed can

be attributed to the steric bulk of the amide reagent.

Figure 16. Conversion, on heating, of the dinitrogen-bridged niobium dimer

to a nitride-bridged dimer with concomitant nitrogen inserted into the ligand

As previously shown in the formation of PNPRuN, azide is

a potential source of a nitride ligand, providing there is the

backbone; silyl methyls omitted for clarity.

-

required two-electron transfer: N₃+ 2e f N^3-+ N₂.

PNPCoN₃is recovered unchanged after 4 days refluxing in

benzene, whereas Lewis acid catalysis of N₂release²⁰utilizing

Sm(OTf)₃also revealed no reaction (18 h at 25 °C). Attempts

to increase the reductive power of cobalt in PNPCoN₃by one-

electron reduction with excess magnesium powder (or 1 equiv

of sodium naphthalide) resulted in clean formation of 1 as the

only PNPCo product with no N₂evolution; in this case azide is

behaving like a pseudo-halide. Paradoxically, the oxidation of

PNPCoN₃with the ferrocenium cation (as the BAr_Fsalt, BAr_F

) [B(3,5-C₆H₃(CF₃)₂)₄]^-) in THF at 25 °C resulted in rapid

gas evolution and a subtle color change from terracotta to red.

metrical parameters; therefore, we limit our discussion to one

independent molecule. The structure reveals one azide-derived

nitrogen atom (N2) inserted into the backbone of the ligand

bonded to one phosphorus and one silicon. The environment

around cobalt consists of two amides, one phosphine, and one

carbon from a phosphorus ylide (both CH₂protons were located

in the penultimate difference map) in a distorted square planar

geometry (angles sum to 360.2°). The accompanying anion

results in an oxidation state at cobalt of +3. Both nitrogens

are strictly planar (angles sum to 359.9 and 359.7°) with the

small difference in their respective bonds to cobalt (Co-N1,

1.856(4); Co-N2, 1.899(4)) attributable to the constraints

imposed by the two fused four-membered rings. The three

Si-N distances for 9 are identical within errors. The ¹H NMR

of 9 is consistent with this structure, displaying C_ssymmetry

with five paramagnetically shifted resonances (the 2H intensity

resonance of the ylide CH₂is not detectable because of its

proximity to the paramagnetic cobalt center). The magnetic

moment for 9 of 3.0µ_Bat 298 K (Evans method, d₈-THF) is

consistent with a triplet ground state, S ) 1 Co(III)⁵⁴center,

which is a spin state with good precedent for four-coordinate,

square planar cobalt.

1

The H NMR (d₈-THF) confirmed one-electron oxidation of

PNPCoN₃, with Cp₂Fe present at 4.12 ppm, along with the

formation of a single new paramagnetic complex of C_ssym-

metry. The mass spectrum (ESI+ mode) gave a molecular ion

peak at 521.2 m/z that corresponds to a compound with the

empirical formula [PNPCoN]⁺. Unambiguous characterization

of this species was forthcoming from an X-ray diffraction study

of crystals of the [PF₆]^-analog (which exhibits identical

spectroscopic data for the cationic portion). This revealed that

the compound is not the expected Co(V) nitride but a four-

coordinate Co(III) complex (Figure 15) resulting from the

formation of new P/N, C/Co, and N/Si bonds: [^tBu₂PCH₂Si-

(Me)₂NSi(Me)₂NP(^tBu₂))CH₂Co]⁺, 9.

The mechanism leading to 9 can be expected to be complex,

requiring the breaking of Si-C and Co-N bonds while forming

new C-Co, P-N, and Si-N bonds. There is literature

precedent⁵⁵for a related nitride insertion reaction, with Fryzuk’s

group, reporting that a dinitrogen-bridged niobium dimer on

heating undergoes an insertion of nitrogen into the ligand

backbone (Figure 16); a complex multistep mechanism via an

The crystal structure shows separated cations and PF₆^-ions

together with a THF molecule filling void space but not involved

in any significant interaction (i.e., to cobalt or via hydrogen

bonding to CH₂Co hydrogens). The asymmetric unit contains

two independent formula units with effectively identical geo-

(52) Walstrom, A.; Pink, M.; Yang, X.; Tomaszewski, J.; Baik, M.-H.; Caulton,

K. G. J. Am. Chem. Soc. 2005, 127, 5330.

(53) Mindiola, D. J.; Cummins, C. C. Angew. Chem., Int. Ed. 1998, 37, 945.

(54) Doerrer, L. H.; Bautista, M. T.; Lippard, S. J. Inorg. Chem. 1997, 36, 3578.

(55) Fryzuk, M. D.; Kozak, C. M.; Bowdridge, M. R.; Patrick, B. O.; Rettig, S.

J. J. Am. Chem. Soc. 2002, 124, 8389.

9

J. AM. CHEM. SOC. VOL. 130, NO. 13, 2008 4271

A R T I C L E S

Ingleson et al.

Figure 17. Proposed reaction scheme for the conversion of [PNPCo(N₃)]⁺to 9, with calculated electronic energies (kcal‚mol^-1) of feasible intermediates.

Figure 18. Frontier molecular orbitals calculated for singlet [PNPCoN]⁺.

electrophilic niobium nitride complex and subsequent attack by

nucleophilic phosphine is proposed.

In an effort to gain insight into the mechanism of the facile

conversion of PNPCoN₃into 9, a low-temperature NMR study

in d₈-THF using Cp₂Fe[BAr_F] (for solubility reasons) was under-

Figure 19. Silyl-oxo coupling in a Rhenium oxide complex on heating at

90 °C (X ) Cl or Br).

taken. Combining PNPCoN₃with [Cp₂Fe]BAr^Fat -65 °C

4

shows (by ¹H NMR) complete consumption of the starting cobalt

azide complex and the formation of Cp₂Fe, together with new

broad peaks of a (low symmetry) paramagnetic product. At

-50 °C, those peaks decline in intensity and while signals from

two new paramagnetic compounds grow in, these begin to

transform to the final crystallographic product (9) at -30 °C.

At no temperature is any ³¹P{¹H} NMR resonance observed,

indicating the absence of any detectable diamagnetic intermedi-

ates. In sum, the bond making and breaking reactions to form

9 involve at least three detectable intermediates, all paramagnetic

in spite of all having an even number of electrons. A plausible

reaction mechanism (Figure 17) can be proposed for the

conversion of PNPCoN₃to 9 via at least three intermediates.

Oxidation of PNPCoN₃by Cp₂Fe⁺would yield [(PNP)Co(N₃)]⁺

that can then evolve N₂and form a Co(V) nitride [PNPCoN]⁺.

There is precedent, both in this work and from that of Peters

and Meyers’ groups,^24,29for intramolecular attack by a nucleo-

philic ligand on a terminal nitrogen, which in this case would

produce the Co(III) complex [(PNPdN)Co]⁺(A). This P/N bond

formation makes an initially electrophilic nitrogen more nu-

cleophilic and thus able to attack silicon. Breaking the Si-CH₂

bond then becomes possible because the P(V) center can

stabilize the emerging CH₂carbanion by ylide formation (B).

As the Si/C bond cleaves, the developing electrophilic character

at silicon can be stabilized by binding to electron-rich nitrogen

on phosphorus (B). Closing the second four-membered ring by

nucleophilic ylide carbon binding to Co (B f 9) completes the

reaction.

was investigated with DFT(PBE) calculations with geometry

optimization using the full ligand set. The calculations find all

species in Figure 17 to be minima and with energies viable for

being intermediates, whereas forming the observed product is

highly exothermic. The formation of the postulated transient

nitride [PNPCoN]⁺and N₂from [PNPCoN₃]⁺is energetically

downhill by 7.6 kcal‚mol^-1; subsequent conversion to 9 can

then proceed via two energetically viable intermediates, A and

B. [PNPCoN₃]⁺is calculated to be a triplet ground state (6.5

kcal‚mol^-1more stable than its singlet) as are Co(III) complexes

A and B; thus these are viable candidates for the three spectro-

scopically observed complexes. [PNPCoN]⁺is calculated to be

a ground-state singlet (by 9.5 kcal‚mol^-1compared to its triplet

isomer); therefore, it would have to be extremely short-lived

(undetectable by NMR) if present in the conversion of PNPCoN₃

to 9. The LUMO and LUMO + 1 of singlet [PNPCoN]⁺(Figure

18) are primarily nitride p_πorbitals and thus are consistent with

the nucleophile/electrophile character for P/N bond formation

proposed early in Figure 17. Indeed, the LUMOs provide

evidence that writing N^3-is an oversimplification, and thus

[CoN]²⁺is a unit where Co(V) is also an imperfect description;

the oxidation is borne by both nitrogen and cobalt.

From this tentative analysis, the key points that emerge are

(a) an (outer sphere) oxidative event is needed to trigger N₂

loss from (PNP)Co(N₃), (b) the terminal nitrogen ligand in

[PNPCoN]⁺is electrophilic, (c) oxidation of phosphorus to P(V)

is the key to subsequent bond making/breaking because it

enables heterolysis of the Si-CH₂bond in the PNP backbone,

and (d) the overall two-electron redox transfer again occurs at

phosphorus, not cobalt. The reaction is essentially complete by

Because of the difficulty inherent in gaining definitive infor-

mation above and beyond purity and molecule symmetry from

paramagnetic NMR spectra, this proposed reaction mechanism

9

4272 J. AM. CHEM. SOC. VOL. 130, NO. 13, 2008

Redox Chemistry of the Triplet Complex (PNP)Co^I

A R T I C L E S

Figure 20. Left, the three products from the reaction of 1 with 1 equiv of PhIdO. Right, ORTEP view (50% probabilities) of 10 (hydrogens omitted for

clarity). Selected bond lengths (Å) and angles (°): Co-I1, 2.5942(9); Co-P2, 2.3935(15); Co-N1, 1.952(4); Co-O1, 1.943(4); P1-O1, 1.492(4); P2-

Co-O1, 116.08(14); N1-Co-I1, 125.84(12); P2-Co-I1, 111.24(5), O1-Co-I1, 105.70(14).

0 °C, so all activation energies are very modest in contrast to

silyl/heteroatom coupling in Fryzuk’s niobium nitride complex⁵⁵

(requires heating to 110 °C) and PNPReX₂(O) (90 °C, Figure

19),⁵⁶which have substantially higher energetic barriers.

e. Other Oxo-Transfer Reagents. Disappointing results of

the reactivity of PNPCo with O₂are described in the Supporting

Information. Seeking a more controlled method for the addition

of one oxygen atom to 1 (seeking a cleaner reaction), the

addition of oxygen atom donors to 1 was investigated. Addition

of 1 equiv of iodosobenzene, PhIdO, to a toluene solution of

1 resulted, after 18 h, in the production of three new paramag-

netic products. The two minor components are readily identifi-

able¹as the Co(II) complexes PNPCoI and P(O)NP(O)CoI. The

major product from this reaction possessed C_ssymmetry and

was isolated spectroscopically pure by recrystallization from

minimum (Me₃Si)₂O. The solid-state structure (Figure 20)

revealed the insertion of an oxygen atom into a P-Co bond;

along with the formation of a Co-I bond; therefore, oxidation

to P(V) and Co(II) has occurred.

The structural metrics for PNP(O)CoI, 10, are unremarkable,

with P-O, O-Co and Co-I bond lengths as expected and

comparable to the analogous bonds in PNPCoI and [κ²-(^t-

Bu₂P(I)CH₂SiMe₂NSiMe₂CH₂P^tBu₂)CoI₂. There is a minor

distortion in the geometry around cobalt away from ideal

tetrahedral geometry, but this is attributable to the constraints

imposed by the fused five- and six-membered rings. A mech-

anism leading to 10 can be envisaged where initial oxygen-

atom transfer would generate PNPCodO, analogous to the

cobalt-imido complex PNPCo(NPh) that could equally undergo

nucleophilic phosphine attack at an electrophilic oxygen (with

numerous examples of electrophilic oxo functionalities reported).

This raises the phosphorus oxidation state to five and returns

cobalt to its +1 state where it then abstracts an iodide from the

phenyl iodide byproduct to generate 10. It is noteworthy that

there is no oxygen-silicon bond formation in this reaction in

direct contrast to the silyl-heteroatom coupling observed from

the nitride analog. An intramolecular conversion analogous to

that producing 9 would be expected to be kinetically faster than

the bimolecular reaction involving phenyl iodide, precluding

this explanation as the origin of the differing reactivity of nitride

versus oxide. The relatively weaker nucleophilicity of a PdO

moiety bound to cobalt when compared to the negatively

charged PdN^-may account for this distinct behavior. The

formation of P(O)NP(O)CoI in this reaction could proceed via

10 or an undetected double oxo inserted complex, P(O)NP(O)-

Co(I). The reaction of 1 with 2 equiv of PhIdO led to the

formation of P(O)NP(O)CoI¹as the sole product (by ¹H NMR).

Because of the slow nature of this heterogeneous reaction, it

was not possible to use low-temperature NMR spectroscopy to

probe for any reaction intermediates.

A number of other oxygen atom donors were employed,

targeting the Co(I)-inserted complexes analogous to 8. Addition

of 1 equiv of either N₂O or Me₃NO to 1 led to one new

paramagnetic complex with solution C_ssymmetry; 1 was still

present in significant quantity. Increasing the stoichiometry of

the oxygen atom donor to 3 equiv resulted in the complete

consumption of 1 and an identical C_ssymmetric product. The

oxidizing reagent independence suggests the addition of only

the three oxygen atoms (i.e., Me₃N and N₂are innocent

byproducts). The final location of the third oxygen atom is of

more interest, but unfortunately, despite repeated attempts,

crystals suitable for X-ray diffraction were not obtainable,

frustrating definitive characterization. The mass spectrum of this

compound was unhelpful, with a peak observed at 715.4 m/z,

which we cannot assign to any expected product. Addition of

an excess of N₂O led to an intractable mixture of paramagnetic

and diamagnetic complexes, even at -65 °C. Whereas charac-

terization of this complex has eluded us, it is again possible to

make the generic observation that oxidation of the phosphine

probably occurs in preference to the cobalt center - a common

theme in this work.

Conclusions

Attempts to reach stable Co(III) complexes ligated by PNP

met, counter-intuitively, with limited success; simple logic

dictates that the superior electron-donating phosphine with ^tBu

substituents (compared to the phenyl derivative of Fryzuk)⁵⁷

should allow greater stabilization of this higher oxidation state.

Experimentally, the opposite was ascertained, with PNPCo^IIIX₂

complexes proving elusive; they are obtainable only when

ligated with low-electronegativity ligands (hydrides/silanes),

where the oxidation state formalism has less meaning. This

behavior may in part be attributable to ligand noninnocence,

with the presence of readily oxidized ^tBu phosphines that result

in phosphorus repeatedly being the locus of oxidation. The

inability to access persistent PNPCodE complexes (where cobalt

would equally be in the + 3 oxidation state) can, in addition to

the reducing power of the electron-rich PNP ligand, be attributed

to weak cobalt/E bonds, with the calculated triplet ground states

for four-coordinate Co(III) resulting in partially filled d orbitals

(56) Ozerov, O. V.; Gerard, H. F.; Watson, L. A.; Huffman, J. C.; Caulton, K.

(57) Fryzuk, M. D.; Leznoff, D. B.; Thompson, R. C.; Rettig, S. J. J. Am. Chem.

Soc. 1998, 120, 10126.

G. Inorg. Chem. 2002, 41, 5615.

9

J. AM. CHEM. SOC. VOL. 130, NO. 13, 2008 4273

A R T I C L E S

Ingleson et al.

that prevent the significant multiple bond cobalt-E character.

Calculations and reactivity studies have shown electrophilic

character for E in these cases, rendering it susceptible to rapid

attack by the nucleophilic (and labile) phosphine. Thus, the

overall products from the addition of electrophilic reagents (e.g.,

PhIdO, RN₃) contain pentavalent phosphorus with the initial

cobalt oxidation state unchanged. An unusual oxidative trig-

gering of N₂elimination from 1e-oxidized (PNP)Co(N₃) ex-

ecutes an even more massive ligand rearrangement, with the

postulated nitride product (PNP)CoN⁺breaking the Si-CH₂

bond to form Si-N and N-P bonds. Given the large number

of thermally stable azide complexes, the oxidative triggering

of N₂elimination warrants more general evaluation as a new

synthetic route to metal nitrides. The new bonds formed and

the final metal oxidation state show that, contrary to a redox

assignment of (PNP)Co(V)(N^3-)⁺, the nitride is electrophilic.

A related silyl-heteroatom coupling is also observed in the

reaction of PNPCo with excess PhSiH₃, highlighting the fact

that the amide nitrogen in PNP is not only a potent π donor to

a metal but also it can be a reactive functionality.

The cobalt center in PNPCo was found to be extremely

electron rich (by IR spectroscopy of its monocarbonyl⁶); in

comparison to other pincer ligands, PNP builds the most

reducing power into (pincer)M, yielding dihydride PNPCo^III-

(H)₂, not dihydrogen. The redox properties of the electron-rich

PNPCo are however paradoxical; whereas one-electron oxidation

of cobalt readily occurred, there was a reluctance for full two-

electron transfer that would result in Co(III). To a degree, this

failure (and the related inability to produce PNPCoH from

reaction with hydrogen-atom donors) is due to the fact that the

reducing power of PNPCo is strongly dependent on not only

the redox potential of cobalt but also, equally importantly, on

the energy recouped in the formation of new Co-X bonds.

Because the silylene is an electrophile, the P/Si bond

formation there also exemplifies what must be Co/P bond lability

when the metal oxidation state rises. At the same time, it is the

apparent lability of these phosphorus on high-valent cobalt that

facilitates intramolecular redox chemistry between an electro-

philic ligand and P(III) because the reactions observed here are

not evident in PNP complexes of the less-labile Ru-P and

Os-P bonds. For example, isoelectronic (PNP)RuN shows⁵²

none of the chemistry found here for (PNP)CoN⁺, apparently

due to the labile P/Co(V) bond.

which either must be frustrated (4d and 5d metals) by kinetic

inertness, or accommodated, by focusing (with PNP) on

reducing reaction conditions.

Experimental Section

General Considerations. All of the manipulations were

performed using standard Schlenk techniques or in an argon-

filled glovebox. Solvents were distilled from sodium/benzophe-

none, CaH₂, or 4 Å molecular sieves, degassed prior to use,

and stored in airtight vessels. All of the reagents were used as

received from commercial vendors or synthesized by published

routes. PNPMgCl-dioxane (PNP ) [(^tBu₂PCH₂SiMe₂)₂N]^-)

and PNPCo were prepared according to literature methods.^{1 1}

H

NMR chemical shifts are reported in ppm relative to protio

impurities in the deuterated solvents. ³¹P{¹H} spectra are

referenced to external standards of 85% H₃PO₄(at 0 ppm), ²⁹

-

Si and ²⁹Si{¹H} spectra are referenced to external standards of

(CH₃)₄Si (at 0 ppm). NMR spectra were recorded with a Varian

Gemini 900 (300 MHz ¹H, 121 MHz ³¹P, 75 MHz ¹³C), a 400

MHz Varian Unity Inova (400 MHz ¹H, 162 MHz ³¹P, 101 MHz

¹³C), or a 500 MHz Varian Inova instrument (500 MHz ¹H, 99

MHz ²⁹Si). Solution magnetic moments were recorded at

25 °C (unless otherwise stated) using the Evans method.^63,64

Consistently for these paramagnetic complexes, no ³¹P{¹H} was

observable. Infrared spectra were recorded on a Nicolet 510P

FTIR spectrometer.

PNPCoGalvinoxyl (2). Method A. A Schlenk flask was

charged with PNPCoNH(2,6-Me₂-C₆H₃) (0.055 g, 0.087 mmol)

and 5 mL of toluene to yield a dark-purple solution. One equiv

of galvinoxyl free radical (0.037 g, 0.087 mmol) was dissolved

in 5 mL toluene and added dropwise over the course of 5 min.

This resulted in an immediate color change from purple to

vermillion, and the solution was stirred for a further 15 min.

The solvent was then removed in vacuo, and the red oil was

redissolved in a minimum volume of pentane, and red crystals

(0.032 g, 0.035 mmol) suitable for an X-ray diffraction study

were formed on storing at -40 °C for 18 h.

Method B. A J. Young NMR tube was charged with PNPCo

(0.010 g, 0.02 mmol) and 0.45 mL of C₆D₆, and the addition of

1 equiv of galvinoxyl free radical as a solid (0.008 g, 0.02 mmol)

resulted in a rapid color change from green to red and NMR

spectra identical to that observed from Method A, indicating

the formation of PNPCogalvinoxyl. Yield (from Method A):

1

40% (quantitative by NMR). H NMR (298 K, C₆D₆): 69.84

Overall, this work with O₂, I₂, PhN₃, and even PhSiH₃

highlights a current common theme, that of ligand noninnocence,

(broad singlet, 4H), 44.15 (sharp singlet overlapped with a broad

singlet, combined intensity of region ) 14H), 22.6 (s, 2H), 16.02

(s, 1H), 3.80 (18H, s), 2.85 (s, 18H), -13.82 (br s, 36H).

Magnetic Moment (C₆D₆): µ_eff) 3.94µ_B. IR (cm^-1, pentane)

CdO quinone stretch: 1610. Mass spec. (ESI+): Calcd. (for

C₅₁H₉₃CoNO₂P₂Si₂): 928.6, Found: 928.8.

PNPCo(H)₂(3). In a standard experiment, a J. Young NMR

tube was charged with PNPCo (0.09 g, 0.04 mmol) and

dissolved in 0.45 mL of C₆D₆. On a gas line, the green solution

was degassed three times by freeze/pump/thaw cycles and

charged with H₂at 77 K (ca. 4 atm). On thawing, this yielded

an immediate color change to a yellow/brown solution. The

integration of the hydride region was performed against the

backbone, intensity 4, CH₂resonance at 0.92 ppm, and is

t

which in the BuPNP ligand set is evident at both the labile,

oxidizable phosphine functionalities and at the amide nitrogen

with its propensity for silyl-heteroatom coupling; these ligand

vulnerabilities need to be considered when attempting to install

any further reactive functionality at the metal center in high

oxidation state complexes with electrophilic ligands (e.g., oxo,

imido, etc). Every ligand has its characteristic vulnerability,^58-62

(58) Fekl, U.; Goldberg, K. I. J. Am. Chem. Soc. 2002, 124, 6804.

(59) Kuhlman, R.; Streib, K.; Caulton, K. G. J. Am. Chem. Soc. 1993, 115,

5813.

(60) Blackmore, I. J.; Gibson, V. C.; Hitchcock, P. B.; Rees, C. W.; Williams,

D. J.; White, A. J. P. J. Am. Chem. Soc. 2005, 127, 6012.

(61) Bartos, M. J.; Gordon-Wylie, S. W.; Fox, B. G.; James Wright, L.;

Weintraub, S. T.; Kauffmann, K. E.; Munck, E.; Kostka, K. L.; Uffelman,

E. S.; Rickard, C. E. F.; Noon, K. R.; Collins, T. J. Coord. Chem. ReV.

1998, 174, 361.

(62) Reardon, D.; Conan, F.; Gambarotta, S.; Yap, G.; Wang, Q. J. Am. Chem.

Soc. 1999, 121, 9318.

(63) Evans, D. F. J. Chem. Soc. 1959, 2003.

(64) Sur, S. K. J. Magn. Reson. 1989, 82, 169.

9

4274 J. AM. CHEM. SOC. VOL. 130, NO. 13, 2008

Redox Chemistry of the Triplet Complex (PNP)Co^I

A R T I C L E S

consistently of 2H intensity (with a d1 ) 10 s). ¹H NMR (298K,

C₆D₆): 1.26 (br s, 36H), 0.93 (br s, 4H) 0.39 (br s, 12H) and

-32.2 (t, J_P-H51 Hz, 2H). ¹H{³¹P} NMR (298 K, C₆D₆): 1.25

(s, 36H), 0.91 (s, 4H), 0.40 (s, 12H), and -32.2 (s, 2H) ³¹P-

{¹H} NMR (298K, C₆D₆): 89.9 (very broad singlet).

copy. 1H NMR (298 K, d₈-toluene): 8.07-8.00 (4H, m, Ph)

7.51-7.9 (6H, m, Ph), 6.32 (1H, s, Si-H), 5.99 (d, J_P-H49

2

Hz silylene Si-H), 1.22-0.87 (40H, overlapping signals for

t

four inequivalent Bu and four inequivalent CH₂), 0.28 (3H, s,

Si-Me), 0.24 (6H, s, 2 Si-Me co-incident), 0.9 (3H, s, Si-

1

Me) and -11.46 (4H, s, Co-H). H{³¹P} NMR (298 K, d₈-

t

{K²-^tBu₂PCH₂Me₂SiNSiMe₂CH₂Bu₂P(H)Sid}Co(H)₃-

toluene): 8.06-8.00 (4H, m, Ph), 7.48-7.19 (6H, m, Ph), 6.32

(s, 1H Si-H), 6.00 (s, 1H, Silylene Si-H), 1.9-0.89 (40H,

^tBu and CH₂resonances), -11.43 (s, 4H, Co-H). ³¹P{¹H} NMR

(298 K, d₈-toluene): 92.7 (br, s) and 19.2 (d, ³J_P-P19 Hz). ²⁹Si

(SiH₂Ph)₂(4). A Schlenk flask was charged with PNPCo (0.100

g, 0.2 mmol) and dissolved in 10 mL of toluene. PhSiH₃(4.1

equiv, 0.087 g, 0.81 mmol) was then added dropwise at 25 °C

while stirring; an immediate color change to brown was

observed followed rapidly by a second color change to pale

yellow. The solvent was then removed in vacuo to yield a pale-

yellow oil. Conversion to 4 was quantitative by NMR spec-

troscopy. This oil was dissolved in a 50/50 mix of pentane and

(CH₃)₄Si, and the solvent was allowed to slowly evaporate,

yielding a number of pale-yellow crystals suitable for X-ray

diffraction. The NMR spectra (below) of these crystals (apart

from resonances associated with Ph₂SiH₂) were identical to those

observed for the isolated oil. Attempts at large-scale recrystal-

lization repeatedly failed in part due to the thermal sensitivity

of 4 (4 was not observable by NMR after 15 h in benzene at

25 °C), producing an intractable unidentified mixture. The

integration of the ¹H NMR was performed with a d1 ) 10 s to

ensure full relaxation. A number of the CH₂resonances in the

1

NMR (263 K, d₈-toluene): 22.2 (d, J_Si-H162 Hz), 10.0 (s),

1

2.7 (s), -18.9 (d, J_Si-H195 Hz). ²⁹Si{¹H} NMR (263 K, d₈-

toluene): 22.3 (s), 9.9 (s), 2.6 (s), -18.8 (s).

PNPCoN₂C(H)SiMe₃(6). A Schlenk flask was charged with

PNPCo (0.050 g, 0.1 mmol) and dissolved in 5 mL of toluene.

To this green solution was added 1 equiv of N₂C(H)SiMe₃as

a 2 M ether solution (50 µL, 0.1 mmol), resulting in an

immediate color change to dark yellow. The toluene was

removed in vacuo, and the resultant dark-yellow oil redissolved

in (Me₃Si)₂O and was stored at -40 °C for 18 h to yield dark-

yellow crystals (0.019 g, 0.03 mmol). Yield: 31% (the conver-

sion was quantitative by NMR). ¹H NMR (298 K, C₆D₆): 1.35

(36H br s), 0.68 (4H, br s), 0.6 (21H br s), -3.04 (1H, br s).

³¹P{¹H} NMR (298 K C₆D₆): 52.3 (br s). ¹³C{¹H} NMR (298

K C₆D₆): 33.4 (br s), 29.7 (br s), 8.7 (br s), 6.4 (br s), 1.1 (br

s), despite numerous attempts, the diazoalkane carbon was not

observed in range +400 to -90 ppm. IR (cm^-1, pentane): 2069.

PNPCoN₂C(H)Ph (7). Synthesized in an entirely analogous

manner to PNPCoN₂C(H)SiMe₃although solid material could

not be obtained despite repeated attempts. The conversion was

quantitative by ¹H NMR spectroscopy. ¹H NMR (298 K,

C₆D₆): 7.41 (1H, m), 6.99 (2H, m), 6.58 (2H, m), 1.26 (36H,

br s), 0.76 (4H, br s), 0.14 (12H br s), -0.83 (1H, br s). ³¹P-

{¹H} NMR (298 K C₆D₆): 48.5 (br s). ¹³C{¹H} NMR (298 K

C₆D₆): 161.9 (s), 136.6 (s), 19.7 (s), 118.7 (s), 41.5 (br s), 33.5

(s), 29.4 (s), 8.2 (s), 6.2 (s). IR (cm^-1, pentane): 1592.

1

¹H and H{³¹P} are obscured making a complete assignment

of all inequivalent proton environments impossible. Yield:

Quantitative by NMR. ¹H NMR (298 K, C₆D₆): 8.9-8.00 (m,

4H, Ph) 7.30-7.22 (m, 6H, Ph), 5.93 (d, ²J_P-H50 Hz Silylene

Si-H), 5.75 (br s, 1H, Si-H), 5.70 (br s, 1H, Si-H), 5.54 (br,

s, 1H, Si-H) 5.31 (br s, Si-H), 1.32-0.92 (4 doublets at 1.32,

3

1.24, 1.01, and 0.96 all with J_P-H) 12 Hz for the four

inequivalent ^tBu groups overlapped with unresolved multiplets

for the four inequivalent CH₂protons, combined intensity of

region ) 40H), 0.30 (s, 3H Si-Me), 0.27 (s, 3H, Si-Me), 0.26

(s, 3H, Si-Me), 0.23 (s, 3H, Si-Me), and -11.29 (s, 3H Co-

1

H). H{³¹P} NMR (298 K, C₆D₆): 8.16-8.01 (m, 4H, Ph),

7.31-7.21 (m, 6H, Ph), 5.93 (s, 1H, silylene Si-H), 5.74 (br

s, 1H, Si-H), 5.69 (br s, 1H, Si-H), 5.53 (br s, 1H, Si-H),

5.31 (br s, 1H, Si-H), 1.32-0.95 (four singlets for overlapped

with unresolved multiplets for the four inequivalent CH₂,

combined intensity of region ) 40H), 0.30 (s, 3H, Si-Me),

0.26 (s, 6H, 2 co-incident Si-Me), 0.23 (s, 3H, Si-Me), and

-11.29 (s, 3H, Co-H). ³¹P{¹H} NMR (298 K, C₆D₆): 82.3

(br s) and 25.6 (d, ³J_P-P16 Hz). ²⁹Si NMR (280 K, C₆D₆): 34.0

(d, 158 Hz), 9.1 (s), 2.6 (s), -15.2 (t, 168 Hz), -29.6 (t, 168

Hz). ²⁹Si{¹H} NMR (280 K, C₆D₆): 33.2 (s), 8.2 (s), 1.7 (s),

-16.0 (s), -30.1 (s). Ph₂SiH₂: ²⁹Si NMR (298 K C₆D₆): -34.0

PN(PdNPh)Co (8). A Schlenk flask was charged with

PNPCo (0.123 g, 0.23 mmol) and 10 mL of pentane and cooled

to -78 °C; at this temperature a solution of PhN₃(31 µL, 0.23

mmol) dissolved in 5 mL of pentane was added dropwise over

the course of 10 min. This resulted in a rapid color change from

green to brown. The solution was stirred for a further 10 min

at this temperature before being allowed to warm to 25 °C. The

pentane was reduced to a minimum volume and stored at

-40 °C for 1 day to yield dark-yellow crystals (0.063 g). After

isolation of the first crop of crystals, the supernatant was reduced

in volume further and cooled to -40 °C for 18 h to yield a

second crop (0.016 g) of crystals. A final crop (0.018 g) can be

obtained by slow evaporation of the remaining pentane solvent

followed by washing with cold pentane, to give a combined

1

(t, J_Si-H) 90 Hz). ²⁹Si{¹H} NMR (298 K C₆D₆): -34.0 (s).

¹H NMR selected (298 K, C₆D₆): 5.07 (s, Si-H).

{K²-^tBu₂PCH₂Me₂SiNSiMe₂CH₂Bu₂P(H)Sid}Co(H)₄-

t

1

quantity of 0.097 g (0.16 mmol). Yield: 70%. H NMR (298

(SiHPh₂) (5). A Schlenk flask was charged with PNPCo (0.09

g, 0.04 mmol) and dissolved in 5 mL of toluene, resulting in a

green solution. The solution was then cooled to -78 °C and 2

equiv of PhSiH₃(0.009 g, 0.08 mmol) was added by syringe

while stirring, resulting in an immediate color change to brown.

The solvent was then removed in vacuo to yield 5 quantitatively

(by NMR) as a brown oil. Attempts to produce a crystalline

solid repeatedly failed in part due to its thermal instability (no

5 remains after 6 h at 25 °C in benzene, producing an intractable

unidentified mixture). Yield: Quantitative by NMR spectros-

K, C₆D₆): 24.27 (s, 2H), 10.96 (s, 18H), 8.64 (s, 18H), 6.70 (s,

2H), 0.27 (s, 6H), -4.16 (s, 6H), -35.14 (br s, 2H), -39.27

(s, 1H), -40.30 (s, 2H). ³¹P{¹H} NMR (298 K, C₆D₆): 206.8.

¹³C{¹H} NMR (298 K, C₆D₆): 502.0 (br s), 461.7 (br s), 402.9

(br s), 338.4 (br s), 186.5 (s), 157.9 (br s), 147.7 (br s), 8.8 (s),

-31.3 (br s), -56.1 (s), -99.7 (s), -237.9 (br s). Magnetic

Moment (C₆D₆): µ_eff) 3.40µ_B.

Low-Temperature NMR Analysis on PNPCo + 0.90 equiv

of PhN₃. A Young NMR tube was charged with PNPCo (0.09

9

J. AM. CHEM. SOC. VOL. 130, NO. 13, 2008 4275

A R T I C L E S

Ingleson et al.

g, 0.04 mmol) and dissolved in 0.30 mL of d₈-toluene and then

cooled to -78 °C. A solution of PhN₃(5 µL, 0.038 mmol)

dissolved in 0.2 mL of d₈-toluene was cooled to -78 °C and

added dropwise to the cooled solution of PNPCo. The tube was

then sealed and carefully agitated, to ensure no solvent warming;

this resulted in a brown solution. ¹H NMR (213 K, C₇D₈): 1.26

(36H, br s), 0.78 (4H, br s), 0.05 (12H, br s); phenyl region

was not clearly resolved. ³¹P{¹H} NMR (213 K C₇D₈): 53.95

(s).

d₈-THF): -140.6 (septet, J_P-F) 712 Hz). ¹⁹F{¹H} NMR (298

K, d₈-THF): -67.9 (d, J_F-P712 Hz). Magnetic moment (d₈-

THF, 298 K): µ_eff) 3.0µ_B. Mass spec (ESI+, m/z): Calcd.

for C₂₂H₅₂CoN₂P₂Si₂: 521.25, Found: 521.2.

[Y] ) [BAr_F] (BAr_f) [B{3,5-C₆H₃(CF₃)₂}₄]). In a standard

reaction, a Young NMR tube was charged with PNPCoN₃(0.015

g, 0.026 mmol), Cp₂FeBAr_F(0.028 g, 0.026 mmol), and 0.45

mL of d₈-THF was added. Immediate gas evolution was

observed, and the resultant solution was deep red. The ¹H NMR

reveals one cobalt-containing product along with a singlet

attributable to Cp₂Fe. Attempts to obtain crystalline material

On warming to -40 °C the first change in the NMR spectra

was observed, with the diamagnetic signals gradually disap-

pearing over the course of 30 min; concomitantly, new

paramagnetic signals were then observed. In addition, the sample

1

of this complex repeatedly failed, even at a larger scale. H

NMR (298 K, d₈-THF): 33.36 (s, 6H), 7.73 (14H, s, overlap

of aromatic C-H with Si-Me), 7.52 (s, 4H), -1.01 (s, 18H),

-12.84 (s, 18H),-15.54 (s, 2H). ¹⁹F{¹H} (298 K, d₈-THF):

1

at this temperature was now a dark green. H NMR (233 K,

C₇D₈): 7.2 (v. br s overlapped with protio toluene signals

preventing accurate integration), 0.3 (4H, br s), -2.1 (12H, br.

s). ³¹P{¹H} NMR (233 K C₇D₈): No phosphorus resonances

was observed from +400 to -300 ppm.

1

-60.3 (s). H NMR (288 K, d₈-THF): 35.76 (s, 6H), 8.6 (s,

6H), 7.72 (s, 8H), 7.52 (s, 4H), -1.16 (s, 18H), -13.84 (s, 18H),

-15.70 (s, 2H).

At -20 °C, the C_2Vparamagnetic product decreased in

intensity with peaks attributable to PN(PdNPh)Co growing in.

At -20 °C, the conversion to PN(PdNPh)Co was complete

after 10 min and the solution now was a dark yellow.

PN(PdO)CoI (10). A Schlenk flask was charged with

PNPCo (0.040 g, 0.08 mmol) and dissolved in 5 mL of toluene;

one equivalent of PhIdO (0.06 g, 0.08 mmol) was added as a

solid and the stirred for 18 h. This resulted in the gradual

dissolution of PhIdO and a color change from green to blue.

Removal of toluene in vacuo resulted in a blue oil. Recrystal-

lization from minimum (Me₃Si)₂O at -40 °C yielded blue

crystals (0.014 g, 0.021 mmol) that were suitable for an X-ray

[PNNP)CH₂Co][Y] (9). [PNNP)CH₂Co] ) [^tBu₂PCH₂Si-

(Me)₂NSi(Me)₂NP(^tBu₂))CH₂Co]. [Y] ) [PF₆]. A Schlenk

flask was charged with PNPCoN₃(0.030 g, 0.05 mmol) and

dissolved in 5 mL of THF. Cp₂FePF₆(0.95 equiv, 0.016 g, 0.05

mmol) was then added as a solid and stirred overnight. The

solution changes from a red/brown to a red over the course of

the reaction. The solvent was then removed in vacuo, and the

resultant oil was washed with pentane (2 × 5 mL portions).

The oil was then dried in vacuo, redissolved in minimum THF,

and cooled to -40 °C overnight to yield red needles (0.021 g)

(crystals suitable for X-ray diffraction study formed spontane-

ously in a Young NMR tube charged with 0.09 g of PNPCoN₃,

0.0075 g of Cp₂FePF₆, and 0.4 mL of d₈-THF and rotated

1

diffraction study. Isolated Yield: 27%. H NMR (C₆D₆, 298

K): 43.31 (6H, br s), 23.40 (2H, br s), 18.94 (2H, br s), 4.44

(18H, br s), -8.58 (18H, br s), and -36.57 (6H, br s). Magnetic

Moment (C₆D₆): µ_eff) 4.10µ_B.

Acknowledgment. This work was supported by the NSF

(CHE-0544829). Professors Mu-Hyun Baik and Daniel Mindiola

are thanked for useful discussions.

Supporting Information Available: CIF files for all crystal

structures. This material is available free of charge via the

Internet at http://pubs.acs.org.

1

overnight). Yield: 58%. H NMR (298 K, d₈-THF): 33.14 (s,

6H), 7.62 (s, 6H), -1.11 (s, 18H), -12.71 (s, 18H), -15.47 (s,

2H). The PdCH₂protons are not observed presumably due to

their proximity to paramagnetic cobalt. ³¹P{¹H} NMR (298 K,

JA074378+

9

4276 J. AM. CHEM. SOC. VOL. 130, NO. 13, 2008

Article Doi

Redox chemistry of the triplet complex (PNP)Co

DOI: 10.1021/ja074378+

Source and publish data:

Authors:

Article abstract of DOI:10.1021/ja074378+

Full text of DOI:10.1021/ja074378+

Products guided by the article

R&D Labs maybe for 1019663-64-9

Relevant to this article

Hot Product