Influence of the metal orbital occupancy and principal quantum number on organoazide (RN3) conversion to transition-metal imide complexes

Article abstract of DOI:10.1021/ic801035z

The reaction of phenyl azide with (PNP)Ni, where PNP = (^tBu ₂PCH₂SiMe₂)₂N^-, promptly evolves N₂ and forms a P=N bond in the product (PNP=NPh)Ni ^I. A similar reaction with (PNP)FeCl proceeds to form a P=N bond but without N₂ evolution, to furnish (PNP=N-N=NPh)FeCl. An analogous reaction with (PNP)RuCl occurs with a more dramatic redox change at the metal (and N₂ evolution), to give the salt composed of (PNP)Ru(NPh) ⁺ and (PNP)RuCl₃^-, together with equimolar (PNP)Ru(NPh). The contrast among these results is used to deduce what conditions favor N₂ loss and oxidative incorporation of the NPh fragment from PhN₃ into a metal complex.

Full text of DOI:10.1021/ic801035z

Inorg. Chem. 2008, 47, 9002-9009
Influence of the Metal Orbital Occupancy and Principal Quantum
Number on Organoazide (RN3) Conversion to Transition-Metal Imide
Complexes
Amy N. Walstrom, Benjamin C. Fullmer, Hongjun Fan, Maren Pink, Drew T. Buschhorn,
and Kenneth G. Caulton*
Department of Chemistry, Indiana UniVersity, Bloomington, Indiana 47405
Received June 5, 2008
The reaction of phenyl azide with (PNP)Ni, where PNP ) (tBu2PCH2SiMe2)2N-, promptly evolves N2 and forms a
PdN bond in the product (PNPdNPh)NiI. A similar reaction with (PNP)FeCl proceeds to form a PdN bond but
without N2 evolution, to furnish (PNPdNsNdNPh)FeCl. An analogous reaction with (PNP)RuCl occurs with a
more dramatic redox change at the metal (and N2 evolution), to give the salt composed of (PNP)Ru(NPh)+ and
(PNP)RuCl3-, together with equimolar (PNP)Ru(NPh). The contrast among these results is used to deduce what
conditions favor N2 loss and oxidative incorporation of the NPh fragment from PhN3 into a metal complex.
nucleophiles, especially by a coordinated phosphine (eq 2).12
Introduction
Thus,13 the reaction of PhN3 with (PNP)Co evolves N2, but
Phenyl azides, and organoazides1 in general, are frequently
the locus of oxidation by the resulting NPh fragment is at
used to form metal imide complexes (eq 1), in a two-electron
phosphorus, giving [tBu2PCH2SiMe2NSiMe2CH2 Bu2Pd
t
oxidation of the metal.2-10 This can be thought of as an
NPh]CoI, with a coordinated phosphinimine arm.
inner-sphere electron-transfer reaction, hence requiring an
intermediate adduct LnM(N3R), and when the metal lacks
LnM(PR3) + R ′ N3 f LnM r N(R′)dPR3 + N2 (2)
reducing power (e.g., Al3+, Ag+, and Cu+), adducts have
indeed been identified and are persistent.11
It is attractive to speculate that this cannot involve coordi-
nated phosphorus but that instead a preliminary dissociation
LnM + RN3 f LnM(NR) + N2
(1)
of the P f M bond is needed, to liberate a lone pair on
phosphorus. We report here results that bear on such a
hypothesis, by studying the metal dependence on the
oxidation of PIII to PV by imino ligands. This type of reaction
is important to understand fully because it represents ligand
“modification”, under the oxidizing conditions available
(RN3) and thus diverts use of the M(NR) functionality from
its intent,1,14 which is normally NR transfer to olefins and
cumulenes. The study of the effect of both vertical and
horizontal periodic trends helps to define the influence of
available d orbitals and metal-phosphine bond lability on
this reaction. The comparisons offered here include FeII, CoI,
NiII, and NiI, hence d6, d8, and d9, and then the isoelectronic
comparison of divalent iron and ruthenium, all of which show
distinctly different behavior toward phenyl azide. For
Certain later-transition-metal imides (>8 d electrons) behave
as if the imide ligand is electrophilic: they are attacked by
* To whom correspondence should be addressed. E-mail: caulton@
indiana.edu.
(1) Brase, S.; Gil, C.; Knepper, K.; Zimmermann, V. Angew. Chem., Int.
Ed. 2005, 44, 5188.
(2) MacBeth, C. E.; Thomas, J. C.; Betley, T. A.; Peters, J. C. Inorg.
Chem. 2004, 43, 4645.
(3) Puiu, S. C.; Warren, T. H. Organometallics 2003, 22, 3974.
(4) Kogut, E.; Wiencko, H. L.; Zhang, L.; Cordeau, D. E.; Warren, T. H.
J. Am. Chem. Soc. 2005, 127, 11248.
(5) Eckert, N. A.; Vaddadi, S.; Stoian, S.; Lachicotte, R. J.; Cundari, T. R.;
Holland, P. L. Angew. Chem., Int. Ed. 2006, 45, 6868.
(6) Sydora, O. L.; Kuiper, D. S.; Wolczanski, P. T.; Lobkovsky, E. B.;
Dinescu, A.; Cundari, T. R. Inorg. Chem. 2006, 45, 2008.
(7) Badiei, Y. M.; Krishnaswamy, A.; Melzer, M. M.; Warren, T. H. J. Am.
Chem. Soc. 2006, 128, 15056.
(8) Dai, X.; Kapoor, P.; Warren, T. H. J. Am. Chem. Soc. 2004, 126,
4798.
(9) Fickes, M. G.; Davis, W. M.; Cummins, C. C. J. Am. Chem. Soc.
1995, 117, 6384.
(10) Proulx, G.; Bergman, R. G. Organometallics 1996, 15, 684.
(11) Dias, H. V. R.; Polach, S. A.; Goh, S.-K.; Archibong, E. F.; Marynick,
D. S. Inorg. Chem. 2000, 39, 3894.
(12) Shah, S.; Protasiewicz, J. D. Coord. Chem. ReV. 2000, 210, 181.
(13) Ingleson, M. J.; Pink, M.; Fan, H.; Caulton, K. G. J. Am. Chem. Soc.
2008, 130, 4262.
(14) Amisial, L. D.; Dai, X.; Kinney, R. A.; Krishnaswamy, A.; Warren,
T. H. Inorg. Chem. 2004, 43, 6537.
9002 Inorganic Chemistry, Vol. 47, No. 19, 2008
10.1021/ic801035z CCC: $40.75  2008 American Chemical Society
Published on Web 08/30/2008

Organoazide (RN3) ConWersion
comparison, we have found that there is no reaction between
d7 (PNP)Co(O3SCF3) and equimolar PhN3 under comparable
conditions.
Results
NiI and NiII. The three-coordinate spin-triplet T-shaped
molecule (PNP)Co has been shown13 to form a coordinated
phosphinimine, 2 (eq 3).
Figure 1. ORTEP drawing (50% probability) of the non-hydrogen atoms
of [tBu2PCH2SiMe2NSiMe2CH2PtBu2NPh]Ni, showing selected atom label-
ing. Unlabeled atoms are carbon. Selected structural parameters: Ni1-N2,
1.9866(14) Å; Ni1-N1, 1.9929(15) Å; Ni1-P2, 2.1815(5) Å; P1-N2,
1.6153(15) Å; N2-Ni1-N1, 107.32(6)°; N2-Ni1-P2, 157.73(4)°;
N1-Ni1-P2, 94.06(5)°.
The reaction is complete at -20 °C. We now find that the
same transformation occurs when M ) Ni,15 an unusual
example of monovalent nickel.16,17 There is no detectable
31P NMR signal in [PNP(dNPh)]Ni, typical for a paramag-
netic compound. The 1H NMR spectrum at 25 °C, however,
shows four resolved paramagnetically shifted signals due to
PNP in a molecule of lower than C2V symmetry; at -30 °C,
each shows shifts due to the Curie-Weiss effect (i.e., shifted
further away from the 0-10 ppm region at lower tempera-
ture) and now five signals are resolved; we propose that no
resonances of the phenyl ring are detected because of a
combination of rapid relaxation and their low intensity.
Species (PNP)Ni(NPh) would be expected to be C2V sym-
metric, so the observed spectra indicate a lower-symmetry
product. The electrospray ionization mass spectrometry (ESI-
MS) spectrum agrees with the formula [PNP(dNPh)]Ni.
(PNPdNPh)M crystallizes in different space groups for
M ) Co and Ni (Figure 1), but their molecular structures
show no significant differences in the bond lengths and angles
involving M. The largest difference is a 0.02 Å lengthening
of M-NSi2 from Co to Ni, and angles around M sum to
359.98° (Co) and 359.11° (Ni). The three substituents are
coplanar (within 1°) at the imine N2. A best least-squares
fit of the two molecules18 shows only minor differences
between Co and Ni, originating in the conformations of the
M-N-P-C-Si-N1 ring, which leads to different P2-M-
N1-Si1 dihedral angles: 174.8° (Co) and 148.8° (Ni).
The NiI center in (PNP)Ni is expected to be a potent
reducing agent15 not only because there is a relatively low
oxidation state but also because it is a neutral molecule (vs
a cation) and because π donation occurs from the PNP amide
nitrogen. Nevertheless, the overall reaction is oxidation at
P, not Ni. We suggest that reducing power is essential to
this reaction because equimolar (PNP)NiCl and added PhN3
can be recovered unchanged after 24 h at 25 °C in benzene.
No new species is detected in this mixture, so there is no
simple adduct (PNP)NiCl(PhN3) formed. The robustness of
the P f NiII bonds and the absence of simple adduct
formation here inhibit any loss of N2 to create a PhN ligand.
The rapid reaction with (PNP)NiI thus supports the idea that
this electron-rich species has some σ*PNi character, which
makes the P f Ni bond labile.
FeII. A comparison to (PNP)FeCl is enlightening. This
molecule is a nonplanar d6 species with a quintet ground
state.15 The reaction of (PNP)FeCl15 with equimolar PhN3
in toluene at -78 °C gives, upon warming toward room
temperature, a color change from light brown to deep
maroon, and the 1H NMR spectrum of this solution shows a
product of Cs symmetry, which indicates inequivalent chelate
arms but rapid fluxionality via chloride passing through a
planar transition state. The 1H NMR chemical shifts (ranging
from +34 to -13 ppm) and the absence of 31P NMR signals
indicate paramagnetism; the temperature-dependent 1H NMR
chemical shifts deviate more from the 0-10 ppm region at
lower temperatures, consistent with the Curie-Weiss law.
Crystals were grown, following trituration with pentane, by
treatment of the red oil with Et2O/pentane.
The crystal structure determination reveals (Figure 2) that
the product has not evolved N2 but contains intact phenyl
azide, having oxidized phosphorus to PV (3; eq 4). Consistent
with Lewis structure 4 (with Fe binding both N2 and N4),
the N4-N3 distance is lengthened by about 0.1 Å over the
N2-N3 distance.
(15) Ingleson, M. J.; Fullmer, B. C.; Buschhorn, D. T.; Fan, H.; Pink, M.;
Huffman, J. C.; Caulton, K. G. Inorg. Chem. 2008, 47, 407.
(16) Eckert, N. A.; Dinescu, A.; Cundari, T. R.; Holland, P. L. Inorg. Chem.
2005, 44, 7702.
(17) Iluc, V. M.; Miller, A. J. M.; Hillhouse, G. L. Chem. Commun. 2005,
5091.
The N4-P distance is short, consistent with a double bond
(1.65 Å). The unit cell contains two crystallographically
independent molecules, and this is a rare case where the two
are not identical within 3 estimated standard deviations
(esd’s). Both molecules have iron coordinated to one P,
(18) See the Supporting Information.
Inorganic Chemistry, Vol. 47, No. 19, 2008 9003

Walstrom et al.
Figure 2. ORTEP drawing (50% probability) of the non-hydrogen atoms of [tBu2PCH2SiMe2NSiMe2CH2PtBu2N3Ph]FeCl, showing selected atom labeling
of molecules a and b. Unlabeled atoms are carbon. Selected structural parameters (molecules a and b): Fe1-N1, 2.0439(17) and 2.0359(16) Å; Fe1-N2,
2.4422(17) and 2.3283(16) Å; Fe1-N4, 2.1290(17) and 2.2099(16) Å; Fe1-Cl 1, 2.2971(6) and 2.3062(6) Å; Fe1-P2, 2.4860(6) and 2.5015(6) Å; P1-N4,
1.6538(17) and 1.6587(16) Å; N2-N3, 1.271(2) and 1.278(2) Å; N2-C23, 1.422(3) and 1.421(3) Å; N3-N4, 1.360(2) and 1.347(2) Å.
amide N, and Cl and N2 and N4 of the Staudinger adduct,
but the Fe-N4 distances differ by 0.08 Å (20 esd’s) and the
Fe-N2 distances differ by 0.11 Å (57 esd’s). Thus, while
iron in both molecules is five-coordinate and both have N4
bonded closer to iron than N2, there is a shift toward more
asymmetric bonding of the two nitrogens in molecule a. This
occurs with no change in the N-N distances of more than
0.01 Å. We interpret this unusual variability of bidentate
binding of the PhN3PR3 unit to it being fundamentally weak,
at least as a bidentate form: when one bond weakens, the
other strengthens. Such a compensation is especially effective
because the bite angle of the ligand is so small (∼56°). The
identity of this product contrasts with those from the reaction
of PhN3 with (PNP)M for M ) Co and Ni. For FeII, the
reduction of the number of d electrons by 2 (vs CoI) and 3
(vs NiI) opens up orbitals that, in effect, trap an intermediate
in phosphinimine formation and show that P-N bond
formation does not happen at the FeNPh stage but instead
even earlier. Nevertheless, this molecule is robust against
N2 loss: it can be recovered unchanged after 12 h at 90 °C
in toluene.
ruthenium analogue of the above iron reagent, (PNP)FeCl.
The addition of equimolar PhN3 to (PNP)RuCl21 in an arene
solvent shows evident gas evolution at 25 °C and even at
-78 °C, together with the deposition of a red solid. The
precipitated product shows no 31P NMR signal, consistent
with being paramagnetic, and the 1H NMR spectrum shows
six signals with intensity consistent with two different PNP-
containing compounds (the phenyl resonances are too weak
and too paramagnetically broadened to be reliably assigned).
Only the SiCH2 signals of one species lie outside the normal
1
alkyl H NMR region, but the three signals for the second
PNP species all lie outside the 0-10 ppm region. The crystal
structure of the solid (Figure 3) identifies the reason for its
low solubility in arenes: it is the salt [(PNP)Ru(NPh)][(P-
NP)RuCl3]. The monocation is planar with a linear RuNC
imide structure, and the monoanion is a mer-octahedral
species. The cation has nearly ideal planar geometry, with a
short [1.716(3) Å] Ru-N2 bond; the Ru-N distance trans
to this is somewhat lengthened at 1.994(2) Å. The Ru-N2-
C23 angle to the phenyl ipso carbon is essentially linear
[177.5(2)°]. The anion is very near to ideal octahedral
geometry, with cis angles from 84.8 to 95.0°; deviations are
mainlyduetoconstraintsofthePNPchelate.TheRu-N(amide)
distance is long, at 2.053(3) Å, because the anion is a 17-
valence-electron species even without any amide π donation;
a πRu-NSi bond order of 0.5 pertains. Nevertheless, the Ru-Cl
distance trans to the amide N is longer (by 0.05 Å) than the
two Ru-Cl distances that are mutually trans. Given this
knowledge of the composition of the solid, 1H NMR
assignments of the solid product described above become
clear: one C2V cation (d4) and one C2V anion (d5).
The conversion of a phosphine to a phosphinimine is a
known and facile reaction of a free phosphine (eq 4), and it
occurs via Staudinger intermediate 3.19,20 The lone pair of a
coordinated phosphine is, in effect, masked or protected,
which is what makes a coordinated phosphine less oxygen-
sensitive (the thermodynamic product would be R3PdO) than
is a free phosphine. Thus, “imination” of a phosphine during
the reaction of a metal phosphine complex can be interpreted
as diagnostic of phosphine dissociation, especially because
the metal is oxidized in forming LnM(NR), and such higher
oxidation states (hard Lewis acids) generally have less
affinity for a (soft base) R3P. A reviewer has suggested that
the adduct (PNP)Ni(CO) has more Ni-P antibonding
character than (PNP)Ni itself,15 hence favoring Ni-P dis-
sociation in an adduct (PNP)NiL.
b. Zinc Reduction of [(PNP)Ru(NPh)][(PNP)RuCl3].
Derivatization of the salt was sought to further establish its
composition. A solution of [(PNP)Ru(NPh)][(PNP)RuCl3]
in THF undergoes a color change from bright red to maroon
during 1 h of stirring in contact with excess zinc powder.
RuII. a. Oxidation of (PNP)RuCl with PhN3. Ruthenium
offers new insights, particularly because we study the
(20) Lin, F. L.; Hoyt, H. M.; Van Halbeek, H.; Bergman, R. G.; Bertozzi,
C. R. J. Am. Chem. Soc. 2005, 127, 2686.
(21) Watson, L. A.; Ozerov, O. V.; Pink, M.; Caulton, K. G. J. Am. Chem.
Soc. 2003, 125, 8426.
(19) Dehnicke, K.; Straehle, J. Polyhedron 1989, 8, 707.
9004 Inorganic Chemistry, Vol. 47, No. 19, 2008

Organoazide (RN3) ConWersion
flanked by two of the six lines expected for the 13 and 17%
abundant ruthenium isotopes 99 and 101 (both I ) 5/2), with
an A value of about 20 G. There is no resolved hyperfine
coupling to any other ligand atoms in the fluid medium. In
a frozen glass at -196 °C, the spectrum resolves into three
g values, consistent with the planar C2V-symmetric structure.
There is incipient ligand hyperfine coupling resolved in two
of the three g tensor signals, with all coupling being below
20 G. Coupling to the central g component was modeled by
two 31P NMR nuclei, while the coupling in the large g value
signal is attributed to one 14N. Given the frontier orbitals
calculated (see section d), it is reasonable that coupling to
phosphorus is small, and it is reasonable to assign the
nitrogen involved in the EPR spectrum as that of the
phenylimide ligand because the spin density of the PNP
amide nitrogen is very small (see section d). Finally, the g
values are significantly shifted from the free electron value,
indicating that there is spin density on ruthenium.
Figure 3. ORTEP drawing (50% probability) of the non-hydrogen atoms
of [PNP]Ru(NPh)+ and its counterion [PNP]RuCl3-, showing selected atom
labeling. Unlabeled atoms are carbon, and tBu methyls have been omitted.
Selected structural parameters: Ru1-N2, 1.716(3) Å; Ru1-N1, 1.994(2)
Å; Ru1-P2, 2.4291(9) Å; Ru1-P1, 2.4386(9) Å; C23-N2-Ru1, 177.5(2)°;
N2-Ru1-N1, 179.07(12)°; N2-Ru1-P2, 95.27(9)°; N1-Ru1-P2,
84.78(8)°; N2-Ru1-P1, 94.29(9)°; N1-Ru1-P1, 85.66(8)°; P2-Ru1-P1,
170.44(3)°; Ru2-N3, 2.053(3) Å; Ru2-Cl3, 2.3859(8) Å; Ru2-Cl1,
2.3909(8) Å; Ru2-Cl2, 2.4375(8) Å; Ru2-P4, 2.4576(8) Å; Ru2-P3,
2.4646(8)Å;N3-Ru2-Cl3,87.81(7)°;N3-Ru2-Cl1,88.87(7)°;Cl3-Ru2-Cl1,
176.67(3)°;N3-Ru2-Cl2,179.34(8)°;Cl3-Ru2-Cl2,91.53(3)°;Cl1-Ru2-Cl2,
91.79(3)°; Cl3-Ru2-P4, 94.98(3)°; Cl1-Ru2-P4, 84.87(3)°; Cl2-Ru2-P4,
92.70(3)°; Cl3-Ru2-P3, 85.09(3)°; Cl1-Ru2-P3, 94.76(3)°; Cl2-Ru2-P3,
92.45(3)°; P4-Ru2-P3, 174.85(3)°.
c. Overall Balanced Reaction of (PNP)RuCl with
PhN3. The implication of the red crystalline product is that
not all redox equivalents are accounted for with only this
product. In fact, the supernatant solution that deposits the
1
red crystalline salt is shown by H NMR spectroscopy to
contain (PNP)Ru(NPh), identical with the product of the zinc
reduction. Thus, the overall reaction is shown in eq 5. The
total of four redox equivalents in two molecules of PhN3
thus goes to oxidize three rutheniums by 2, 1, and 1 electrons,
respectively, in this reaction.
Scheme 1
-
3(PNP)RuCl + 2PhN3 f (PNP)RuNPh+ + (PNP)RuCl3
+
(PNP)Ru(NPh) (5)
The key point is that this has been redox chemistry, with
electron-rich ruthenium being oxidized, in contrast to the iron
analogue. As the NPh ligand settles onto the metal, it
apparently adopts a linear RuNC geometry, which is enough
to weaken the Ru-Cl bond. Chloride is then removed/
scavenged by a second initially 14-electron (PNP)RuCl
center, which must become more Lewis acidic upon oxida-
tion to RuIII because it again serves as the halide acceptor,
-
generating (PNP)RuCl3 . The redox balance (eq 6) shows
that two phenyl azides consume four electrons.
2PhN3 + 4e- f 2PhN2- + 2N2
(6)
The reason why the reaction does not simply produce
molecular (PNP)Ru(NPh)Cl or the salt [(PNP)Ru(NPh)]Cl
must lie in the Lewis acidity of (PNP)RuCl, or its RuIII
Produced in an equimolar ratio (1H NMR assay) are the
known (PNP)RuCl together with another paramagnetic C2V-
symmetric species, which we assign as (PNP)Ru(NPh). Zinc
thus reduces RuIV of the cation to RuIII and RuIII of the anion
to RuII. The addition of 4 equiv of PhCN per Ru to a benzene
solution of this zinc-reduced mixture converts (PNP)RuCl
completely to the known21 diamagnetic trans-(PNP)Ru-
Cl(NCPh)2, but free PhCN persists and 1H NMR signals due
to (PNP)Ru(NPh) are unchanged, indicating it to be not
Lewis acidic toward PhCN.
-
analogue, together with (PNP)RuCl3 being inert to PhN3.
The latter is consistent with the idea that PhN3 is only a
redox reagent if it first coordinates to the metal, something
-
that evidently cannot happen with octahedral (PNP)RuCl3 .
Note also that the reduction of (PNP)Ru(NPh)+ by zinc
is also relevant to the production of (PNP)Ru(NPh) in the
original reaction because the reductant there is (PNP)RuCl
itself, at least until late in the reaction, when this reductant
is depleted.
The X-band electron paramagnetic resonance (EPR)
spectrum of (PNP)Ru(NPh) in a toluene solution18 at 25 °C
A plausible mechanism for forming the products (Scheme
1) proceeds through an adduct with PhN3, and then Cl- lost
1
is a single line at g ) 1.96, correct for an S ) /2 species,
Inorganic Chemistry, Vol. 47, No. 19, 2008 9005

Walstrom et al.
Figure 4. Frontier orbitals of (PNP)Ru(NPh) showing three occupied d orbitals (HOMO-1, HOMO, and SOMO) and spin density on the NPh group, as well
as the LUMO.
from that adduct can be scavenged by (PNP)RuCl, to yield
N2, (PNP)Ru(NPh)+, and (PNP)RuCl2-. The reaction of PhN3
with this five-coordinate RuII anion would yield
(PNP)RuCl2(NPh)-, which would lose Cl- to another (PN-
P)RuCl, together with a chlorine atom transfer, to give the
other observed products, N2, (PNP)Ru(NPh), and (PN-
coordinated PhN3 fragment is capable of forming such a four-
membered ring.
Alternative identities for the two diamagnetic species
in dynamic equilibrium at low temperature include
(PNP)RuCl(N3Ph), with the phenyl azide bound η1 through
its R- or γ-nitrogen as well as bound η2 through both of
these nitrogens (i.e., species 5). These remain at low
temperature at the end of the reaction, in the presence of
excess PhN3, because there is insufficient (PNP)RuCl to
effect either chlorine atom transfer or chloride transfer.
However, by 23 °C, these species have decayed, to yield
only an array of paramagnetic signals, together with (PN-
P)Ru(NPh), which was already evident at -60 °C.
-
P)RuCl3 . This is attractive in avoiding reactions bimolecular
in bulky (PNP)Ru complexes and using the fact that an NPh
ligand will tend to expel a ligand as it becomes linear and
donates more electrons to ruthenium.
An alternative to reaction A in Scheme 1 is chlorine atom
transfer from (PNP)RuIVCl(NPh) to (PNP)RuIICl, giving
observed products (PNP)RuIII(NPh) and (PNP)RuIIICl2, which
will later scavenge Cl- to give observed (PNP)RuCl3 . This
-
d. Density Functional Theory (DFT) Analysis of
(PNP)Ru(NPh) Species. DFT (B3LYP) calculations on the
redox pair (PNP)Ru(NPh) and (PNP)Ru(NPh)+ were carried
out. For the cation, the species characterized in the salt
(Figure 3), these show a singlet ground state with the triplet
9.5 kcal/mol higher. In the triplet, the spin density is 1.07e-
on ruthenium but 0.54e- on N2 (the amide nitrogen) and
0.32e- on the phenyl ring, hence highly delocalized, with
significant oxidation at the phenylimide ligand. Any singlet-
to-triplet conversion thus has metal-to-ligand charge-transfer
character (see the Supporting Information). The spin density
on the phenyl ring thus naturally accounts for the NMR
nonobservance of the phenyl protons, due to rapid relaxation.
The highest occupied molecular orbital (HOMO) and HO-
MO-1 of the singlet cation are the dxz (x along the Ru-P
shows that there are two reactions in competition for
available (PNP)RuCl, oxidation by PhN3 and chlorine (atom
or ion) scavenging, which is what leads to the observed 3:2
(PNP)RuCl/PhN3 stoichiometry in eq 5.
Observations relevant to some of these steps are available
from low-temperature monitoring of (PNP)RuCl and PhN3
reagents combined and held at low temperature prior to low-
1
and variable-temperature H and 31P NMR monitoring (see
the Experimental Section); the salt (Figure 3) is not detected
in these studies in toluene because it is already fully formed
at -60 °C but is essentially insoluble in toluene at this
temperature. It is important to recognize that new species
detected in this soluble portion represent a small fraction of
the total yield and are under the special influence of a modest
initial excess of the PhN3 oxidant; in particular, these
represent a late stage of the reaction. These spectra show a
rapid (31P NMR time scale, hence, one temperature-depend-
ent 31P NMR chemical shift) equilibrium between two
diamagnetic molecules, both of which disappear by +20 °C.
Diamagnetism indicates that these are both even-electron
species and have a coordination number 5 or 6. The 1H NMR
spectra recorded concurrently during this study show these
2
direction) and dz orbitals; the lowest unoccupied molecular
orbital (LUMO) is the dπ orbital perpendicular to the phenyl
ligand plane (i.e., in the Ru-N1-P2 plane) in the antibond-
ing phase with that phenylimide nitrogen and carbon orbit-
als.18 Because the triplet is achieved by populating an orbital
with Ru-NPh π* character, the Ru-N distance lengthens
by 0.08 Å from singlet to triplet. In the neutral doublet
(PNP)Ru(NPh), the same two orbitals are doubly occupied,
and the singly occupied molecular orbital (SOMO) closely
resembles the LUMO of the singlet cation, as expected for
simple one-electron reduction without gross electronic
reorganization (Figure 4; see also the Supporting Informa-
tion). The neutral doublet has 0.31e- spin density on
ruthenium with 0.41e- on N2 and 0.30e- on the phenyl ring,
hence again considerably delocalized, yet still with a short
calculated Ru-N2 distance, 1.81 Å. The SOMO of (PN-
P)Ru(NPh) and both SOMOs of triplet (PNP)Ru(NPh)+ show
negligible participation by phosphorus orbitals, so it is
t
time-averaged species to have Cs symmetry (two Bu and
two SiMe signals). We suggest (Scheme 1, reaction B) that
these are (PNP)Ru(N2)Cl(NPh) and (PNP)RuCl(NPh), where
chloride loss has not happened because of depletion of the
chloride scavenger and reductant (PNP)RuCl in the late stage
of the reaction. Coordination of N2 is a natural kinetically
controlled consequence if the (unobserved) primary product
is (PNP)RuCl(N3Ph); thus, that terminal N of N2 is bonded
to ruthenium prior to N-N bond cleavage, a reaction that
may occur by a four-membered ring transition state (5 in
Scheme 1). The structure of (PNPN3Ph)FeCl shows that a
9006 Inorganic Chemistry, Vol. 47, No. 19, 2008

Organoazide (RN3) ConWersion
understandable that the EPR spectra show little or no
coupling to those nuclei.
then loses N2, to give22 Cl2L3IrIII[N(H)dCHPh]+ rather than
isomeric Cl2L3Ir(dNCH2Ph)+. This is a case where oxidation
of benzylic carbon is favored over oxidation of trivalent
iridium. In addition, the Cl2L3IrV(dNCH2Ph)+ alternative
would have a bent imide ligand, with a nitrogen lone pair
that is probably a reactive site.
The most distinctive structural difference between the
singlet and triplet (PNP)Ru(NPh)+ is the Ru-N2 distance,
1.74 Å for the singlet and 1.82 Å for the triplet. However,
because we observe no 31P NMR signal for the spe-
cies (PNP)Ru(NPh)+ in its salt with the paramagnetic anion
-
(PNP)RuCl3 , it is difficult to escape the conclusion that
(PNP)Ru(NPh)+ is indeed paramagnetic, hence a triplet,
unless the NMR features of the cation are influenced by
intimate ion pairing with the truly paramagnetic anion. What
favors the singlet attribution for (PNP)Ru(NPh)+ is that the
calculated singlet Ru-NPh distance (1.74 Å) is in better
agreement with the observed value (1.72 Å) than is the triplet.
One reviewer disagrees: “...a conclusion about the cation’s
spin state cannot yet be reached.”
Work with electron-rich (PNP)RuCl shows that the metal
is the preferred reductant, even if the thermodynamic product
might have been [PNP(dNPh)]RuCl, so the reaction may
be under kinetic control. The high rates to all observed
products indicate that electron transfer is facile when it is
thermodynamically favored.
Discussion
Perhaps the most important conclusion from the composi-
tion and structure of the FeII product reported here is to show
that the formation of the PdN bond does not necessarily
require that an electrophilic NPh ligand be present, as is
evident from the (metal-free) Staudinger reaction. The intact
azide is sufficiently electrophilic to perform two-electron
oxidation of PIII. Why does this iron complex not rapidly
evolve N2? It is clear from eq 7 that N2 elimination requires
an s-syn conformation around the single N3-N4 bond, while
the observed bidentate binding of the P-N4-N3-N2 group
imposes an s-anti configuration.
(PNP)RuIV(NPh)+ is a 16/18-electron, planar d4 species
(counting a triple Ru-NPh bond), which is the formal
product of transferring C6H5 to the nitride of known23
+
(PNP)RuN. The Ru-N distance in this nitride is 1.627(2)
Å, or 0.09 Å shorter (vs Figure 3) because the nitrogen was
made more nucleophilic but still triply bonded to ruthenium.
One point of interest is that, while d4 (PNP)RuN is
diamagnetic,23 the addition of a C6H5 electrophile to the
+
nitride nitrogen yields a triplet ground state. Why is this?
One answer might be that the addition of this electrophile
to nitride nitrogen weakens the ligand field, hence leading
to smaller orbital energy spacings and high spin behavior.
The reaction of phenyl azide with (PNP)RuCl is excep-
tionally complex because four-coordinate RuII here is not
only a reducing agent but also a Lewis acid (hence scaveng-
ing ionic chloride) and a primary product (PNP)RuCl(N3Ph)
has M-P bonds that are more kinetically inert than those of
(PNP)FeCl. The observed chemistry of PhN3 with (PNP)FeCl
shows what happens when metal-based redox chemistry
(including N2 loss) is not favorable, so another reaction
channel (i.e., mechanism) intervenes. That reaction channel
is influenced by empty orbitals in S ) 1 (PNP)RuCl, in
contrast to all orbitals at least half-filled in S ) 2 (PNP)FeCl.
The other reaction channel uses the additional available iron
orbital (even if spin pairing is required) and thus can stop
without oxidation of iron to FeIV, whereas those preferences
are reversed for (PNP)RuCl. In the case of monovalent
nickel, why is the redox reaction so fast even at -60 °C,
with no intermediate detectable even there, and likewise no
(PNP)Ni(NPh) detectable? We suggest that, finally for the
d9 configuration among the 3d configurations studied here,
a linear NiNPh moiety in (PNP)Ni(NPh) displaces one
phosphine arm to avoid a 19-electron configuration and favor
a 17-electron count (thus using all nine valence orbitals, even
Bidentate binding interferes with the cyclization to form
6 because that cyclization requires rupture of the Fe-N4
bond. This raises the barrier for N2 elimination, and so the
product [PNP(N3Ph)]FeCl is trapped. The final point is that
the decrease of the d electron count to 6, from monovalent
Co and Ni, allows the PNPN3Ph- ligand to be tetradentate
and thus is the origin of the features unique to iron.
The results reported here show a variety of outcomes for
(PNP)MXn complexes in contact with the oxidizing environ-
ment of phenyl azide. The results with 3d metals show that
attack on the PNP ligand phosphorus is favored over attack
at the methyl C-H bonds or at electron-rich amide N
(forming 7); a reducing metal is required, however, because
(22) Albertin, G.; Antoniutti, S.; Baldan, D.; Castro, J.; Garcia-Fontan, S.
Inorg. Chem. 2008, 47, 742.
(23) Walstrom, A.; Pink, M.; Yang, X.; Tomaszewski, J.; Baik, M.-H.;
Caulton, K. G. J. Am. Chem. Soc. 2005, 127, 5330.
+
(PNP)NiCl is inert to PhN3. The reaction of IrCl2L3 (L )
PR3) with PhCH2N3 gives an adduct of intact azide, which
Inorganic Chemistry, Vol. 47, No. 19, 2008 9007

Walstrom et al.
t
[(tBu2PCH2SiMe2)N(SiMe2CH2 Bu2PdNPh)]Ni. To a solution
of 15 mg (0.030 mmol) of (PNP)Ni in 0.5 mL of C6D6 was added
in a J-Young NMR tube at 25 °C 3 µL (0.027mmol) of PhN3. The
solution went from pale yellow to yellow brown, and gas bubbles
were observed. The solution was stripped to dryness in a vacuum;
the solid was then dissolved in minimal pentane and cooled to -40
°C for 24 h. Yellow crystals formed, suitable for X-ray diffraction
study. The solution was decanted, and the crystals were washed
if one is only singly occupied). The pendant phosphine is
then sufficiently nucleophilic and reducing to attack the
phenylimide nitrogen, leading to the observed product. The
complete lack of spectroscopic change for (PNP)NiCl in
the presence of PhN3 shows that any adduct between these
two has a very small binding constant for an η1 adduct
(PNP)NiCl(N3Ph) and certainly lacks enough valence orbitals
to bind PhN3 in an η2 fashion,17 which we believe is central
to breaking the N-N bond after two-electron transfer has
taken place. η1 binding is not conducive to breaking of the
NR-Nꢀ bond, even after two-electron transfer (8, which
shows that it is the Nγ-Nꢀ bond that is weakened), but
certain resonance forms of η2-RN3 are conducive to scission
of this bond (10 and 11 vs 9). The fact that phosphorus is
not oxidized in the ruthenium case probably results from the
Ru-P bonds being kinetically inert, so there is no nucleo-
philic phosphorus to attack electrophilic nitrogen. η1 binding
at NR is sterically unfavorable.
1
with cold pentane and then dried by a vacuum. H NMR (25 °C,
C6D6): δ 16.8 (vbs, 2H, CH2), 8.8 (vbs, 12H, SiMe), 4.6 (vbs, 36H,
tBu), -24.6 (vbs, CH2). 1H NMR (-30 °C, C7D8): δ 19 (vbs, 2H,
t
CH2), 10.6 (vbs, 12H, SiMe), 4.9 (vbs, 18H, Bu), 3.9 (vbs, 18H,
tBu), -24.8 (vbs, 2H, CH2). At this temperature, it is clear that the
tBu peaks resolve into two separate signals, which implies accidental
degeneracy at 25 °C. ESI MS (THF): m/z 597 and 599 ((PNPN-
Ph)NiH+) for 58Ni and 60Ni.
Reaction of (PNP)FeCl with Phenyl Azide (PhN3). To a
Teflon-sealed reaction flask containing 12.5 mg of (PNP)FeCl
(0.0231 mmol) dissolved in toluene-d8 and chilled in a dry ice/
acetone bath was added an equimolar amount of phenyl azide. Upon
warming to room temperature with stirring, a sharp color change
from the characteristic light brown of the starting material to deep
maroon occurred. The resulting solution was analyzed by NMR
[see below; conversion of (PNP)FeCl was complete, and only one
product was formed in >90% yield], and then the volatiles were
stripped in a vacuum. The resulting oil was triturated with pentane
several times, followed by a 10% ether/pentane mixture. A small
crop of deep-red orthorhombic crystals was isolated and analyzed
by X-ray diffraction. A portion of the mother liquor was analyzed
by FT-IR, with the only peak of interest found above 1000 cm-1
being ν 1591 cm-1. Upon attempted thermolysis in a J-Young NMR
tube for 12 h with a 90 °C oil bath, the sample experienced no
color or 1H NMR change. 1H NMR (400 MHz, C7D8): δ 34 (br s,
18H, PtBu2), 31 (br s, 2H, Si-CH2-P), 23 (br s, 6H, Si-Me2), 15
(br s, 6H, Si-Me2), 13 (br s, 2H, Si-CH2-P), 9.7 (br s, N3-Pharyl),
5.7 (br s, N3-Pharyl), -2.3 (br s, N3-Pharyl), -13 (br s, 18H,
Finally, attack on the ancillary ligand observed here for
the 3d metals must be seen as a general example22,24 of
“ligand degradation”, which needs to be taken into account
in seeking application for a given ligand type. Here, the
nucleophilic character of the PNP ligand is evident under
electrophilic conditions, which redirects future attention to
applications of PNP under reducing conditions. On the other
hand, the synthesis of (PNP)Ru(NPh) here suggests that
isoelectronic (PNP)RuO might be a realistic synthetic target
and that (PNP)Ru(O)Cl might be a useful oxidant.
P-tBu2). The assignment of signals to silyl Me and Bu groups
t
was done by relative integration values. The assignment of the
intensity 2 methylene protons and phenyl protons was more difficult
and therefore more tentative; the most separated, 31 and 13 ppm,
are assigned to CH2 protons, based on their proximity to the iron
center.
Experimental Section
General Considerations. Preparations from literature sources
were used to synthesize starting materials RuN(SiMe2-
CH2PtBu2)2Cl,21 (PNP)Ni,15 and (PNP)FeCl,15 and standard Schlenk
or glovebox techniques in an inert (argon) atmosphere were used
for air-sensitive manipulations. All solvents, including deuterated
NMR solvents, were dried over and distilled from Na/benzophenone
and stored in anaerobic conditions. All other reagents were degassed
and/or used as received from commercial vendors. 1H and 31P{1H}
NMR spectra were recorded on a Varian Unity I400 (400 MHz 1H
and 162 MHz 31P{1H}) instrument, with chemical shifts reported
in ppm, referenced to protio impurities in each stated solvent, and
31P{1H} spectra were externally referenced to 85% H3PO4 (0 ppm).
FT-IR spectra were recorded on a Nicolet 510P spectrophotometer.
Mass spectra (ESI-MS) were acquired on a PE-Sciex API III triple-
quadrupole spectrometer and simulated with IsoPro 3.0 isotopic
distribution software.
Reaction of (PNP)RuCl with Phenyl Azide (PhN3). Room
Temperature Experiment. To a J-Young NMR tube containing
15.3 mg (0.0257 mmol) of (PNP)RuCl in C6D6 was added 3.1 mg
(approximately 1 mol equiv) of neat PhN3. Upon separation of the
precipitated red crystalline product, comprised of pure
[(PNP)RuNPh]+[(PNP)RuCl3]-, and preparation of a saturated
THF-d8 solution, the following were obtained. 1H NMR (400 MHz,
THF-d8): δ 19.58 (4H, CH2 of (PNP)RuCl3-), 10.26 (12H, SiMe2
of (PNP)RuCl3-), 1.81 (36H, PtBu2 of (PNP)RuNPh+), 0.30 (12H,
SiMe2 of (PNP)RuNPh+), -0.92 (4H, CH2 of (PNP)RuNPh+),
-2.63 (36H, PtBu2 of (PNP)RuCl3-). Assignments of any two
signals of equal intensity were done assuming the broader, more
shifted signals are due to the anion. 31P{1H} NMR (162 MHz, THF-
d8): no signals in the δ -500 to +500 range. FT-IR (solid on a
KBr plate): ν(RudN) 1261 cm-1. ESI-MS (THF, positive ion
detection): m/z 641.2 ((PNP)RuNPh+). THF is also an acceptable
solvent for this reaction, but if the reaction is carried out in pentane
(24) Gregory, E. A.; Lachicotte, R. J.; Holland, P. L. Organometallics 2005,
24, 1803.
9008 Inorganic Chemistry, Vol. 47, No. 19, 2008

Organoazide (RN3) ConWersion
or diethyl ether, [(PNP)Ru(NPh)][(PNP)RuCl3] is not produced.
Attempts to precipitate salts containing only (PNP)Ru(NPh)+ failed
using NaBPh4 or TlPF6.
show a broad peak at δ +5.4-5.6, due to (PNP)Ru(NPh). Red crystals
persist at 22 °C at the end of this study.
Other Attempted Oxidations. For comparison, equimolar
(PNP)RuCl and PhNdNPh in THF-d8 show unreacted starting
materials (NMR assay) even after heating at 65 °C for 24 h.
For some combination of electronic and steric effects, adamantyl
azide does not react with PNPRuCl: the reagents (equimolar, in
benzene) are recovered unchanged after 20 h at 65 °C.
Variable-Temperature Experiment. To a J-Young NMR tube
containing 20.3 mg (0.0342 mmol) of (PNP)RuCl in C7D8 at -78
°C (dry ice/acetone bath) was added by a cannula a solution
containing 4.0 mg of PhN3 in C7D8. The tube was immediately
frozen in liquid N2 and kept frozen until it was inserted into a
precooled NMR probe. When the sample was placed into the
instrument, the sample was briefly shaken vigorously to ensure
homogeneity. First, spectra were obtained at -60 °C and then
repeated in 10 °C increments, waiting 15 min for thermal equilibra-
tion at each new temperature.
Zinc Reduction of [(PNP)Ru(NPh)]+[(PNP)RuCl3]-. To a
J-Young tube containing 50 mg of [(PNP)Ru(NPh)][(PNP)RuCl3]
in THF-d8 was added a 10-fold excess of activated (by HCl) zinc
powder. This mixture was allowed to rotate overnight, and after
1
15 h, the H NMR spectrum showed complete reduction to a 1:1
mixture of (PNP)Ru(NPh) and (PNP)RuCl, with the latter identified
Already at -60 °C, all (PNP)RuCl had been consumed, and the
solution color has changed to dark purple, small red crystals began
to form, and the solution phase showed a 31P{1H} NMR singlet
by comparison to an authentic sample. NMR yield: >90% (no other
1
products detected). H NMR (400 MHz, THF-d8): δ 5.36 (36H,
PtBu of (PNP)Ru(NPh)), 0.42 (12H, SiMe2 of (PNP)Ru(NPh)),
-5.01 (4H, CH2 of (PNP)Ru(NPh)). 31P{1H} NMR (162 MHz,
THF-d8): no signal in the δ -500 to +500 range.
t
1
(0.3 ppm broad), together with two Bu H NMR and two SiMe
signals consistent with diastereotopically inequivalent PtBu2 and
SiMe2 groups but mirror symmetry relating the two arms of the
PNP ligand. As the temperature is increased in 10 °C increments,
no new compounds form up to -20 °C. However, the 31P NMR
chemical shift moves dramatically as the temperature rises, from δ
+37 at -60 °C to δ +48 at -20 °C. At +20 °C, this signal is at
δ +77 but of diminished integrated intensity. The two tBu signals
also clearly decline in intensity beginning at about -20 °C and
then are very weak by +20 °C. These observations are consistent
with temperature-dependent equilibrium between two diamagnetic
species. An adduct (PNP)Ru(PhN3)Cl will make the two groups
on a PtBu2 or SiMe2 inequivalent, and this adduct would have
decreasing mole fraction as the temperature rises because of the
unfavorable entropy change to form the adduct. If the observed
31P NMR signal is a mole-fraction-weighted average of the two
equilibrating complexes, this explains the observations. Other
features bear mention: even at -60 °C, a weaker 31P{1H} NMR
signal of unidentified origin is observed, at δ +55, but this does
not change in chemical shift or grow in intensity with increased
temperature, suggesting it to be some product formed only on initial
local heating when the reagents were combined at -78 °C; this signal
is still present (sharp at 56 ppm) in the NMR spectrum at +20 °C, but
it decays completely and irreversibly after 3 h at 25 °C. A tBu apparent
Reaction of a 1:1 (PNP)Ru(NPh)/(PNP)RuCl Mixture with
PhCN. To a J-Young tube containing 10 mg of a 1:1 (PN-
P)Ru(NPh)/(PNP)RuCl mixture was added 7.5 µL (approximately
4 mol equiv per ruthenium) of PhCN. The red color of the sample
intensified immediately, and an immediate 1H NMR spectrum
showed complete conversion of (PNP)RuCl to the known molecule
(PNP)RuCl(PhCN)2 and showed no reactivity of (PNP)Ru(NPh)
toward PhCN (free PhCN was observed and (PNP)Ru(NPh)
remained unchanged). The solvent was then stripped and the
resulting residue extracted with pentane, effectively separating the
two compounds because of the extremely poor solubility of
(PNP)RuCl(PhCN)2 in pentane. ESI-MS of the neutral (PN-
P)Ru(NPh), separated from the pentane-insoluble (PNP)Ru-
Cl(PhCN)2, was obtained and showed m/z 641.25 ((PNP)RuNPh+).
Acknowledgment. This work was supported by the NSF
(Grant CHE-0544829). We thank Prof. Daniel Mindiola for
useful discussions and suggestions.
Supporting Information Available: CIF files for three mol-
ecules, plots of NMR spectra, a least-squares fit of (PNPdNPh)Ni
versus (PNPd]NPh)Co, EPR spectra of (PNP)Ru(NPh), and DFT
results for (PNP)Ru(NPh)0,+. This material is available free of
charge via the Internet at http://pubs.acs.org.
1
virtual triplet in the H NMR (∼1.48 ppm) is observed at every
temperature and with constant intensity, which we therefore assign to
the species with the δ +55 31P chemical shift; it finally decays after
3 h at 25 °C. In addition, the 1H NMR spectra from -10 to +20 °C
IC801035Z
Inorganic Chemistry, Vol. 47, No. 19, 2008 9009