Nitrogen-ligated iron complexes: Photolytic approach to the FeN+ moiety

Article abstract of DOI:10.1021/ic702279b

The synthesis of (PNP)FeCl, (PNP)Fe[NH(xylyl)], and (PNP)FeN₃ are reported(PNP = (^tBu₂PCH₂SiMe ₂)₂N^-). While the azide is thermally stable, it is photosensitive to lose N₂ and form [(PNP=N)Fe]₂,in which the nitride ligand has formed a double bond to one phosphorus, and this N bridges to a second iron to form a 2-fold symmetric dimer. The reaction energy to form the (undetected) monomeric [η³-^tBu ₂PCH₂SiMe₂NSiMe₂CH₂P ^tBu₂=N]Fe is -15.9 kcal/mol, so this P^III → P^V oxidation is favorable. The η² version of this same species is less stable by 23.7 kcal/mol, which shows that the loss of one P→ Fe bond is caused by dimerization, and therefore, it does not precede and cause dimerization. A comparison is made to Ru analogs.

Full text of DOI:10.1021/ic702279b

Inorg. Chem. 2008, 47, 5129-5135
Nitrogen-Ligated Iron Complexes: Photolytic Approach to the FeN+
Moiety
Drew Buschhorn, Maren Pink, Hongjun Fan, and Kenneth G. Caulton*
Department of Chemistry, Indiana UniVersity, Bloomington, Indiana 47405
Received November 21, 2007
The synthesis of (PNP)FeCl, (PNP)Fe[NH(xylyl)], and (PNP)FeN3 are reported(PNP ) (tBu2PCH2SiMe2)2N-). While
the azide is thermally stable, it is photosensitive to lose N2 and form [(PNPdN)Fe]2,in which the nitride ligand has
formed a double bond to one phosphorus, and this N bridges to a second iron to form a 2-fold symmetric dimer.
3 t
The reaction energy to form the (undetected) monomeric [η -Bu2PCH2SiMe2NSiMe2CH2PtBu2dN]Fe is -15.9 kcal/
2
mol, so this PIII f PV oxidation is favorable. The η version of this same species is less stable by 23.7 kcal/mol,
which shows that the loss of one Pf Fe bond is caused by dimerization, and therefore, it does not precede and
cause dimerization. A comparison is made to Ru analogs.
Introduction
these last results were established with a C3 symmetric
anionic tripodal tris-phosphine borate ligand. Of special
interest to us in this class of compounds is to establish both
the intrinsic Fe/E bond character (single, double, electro-
philic, nucleophilic), as well as its reactivity, for E ) N
versus E ) O. Oxidative utilization of high-valent FeOq+ is
of both biological and catalytic relevance.11,12 If an FeN unit
can oxidize H2, this would be one step in the hydrogenation13
of N2: nitrogen fixation. Our FeN+ results, reported here,
are set in the above 2-fold-symmetric PNP environment
whose π donor strength is established14 to be considerable.
We reported earlier1 that reactions designed to produce
(PNP)Ru(N3), PNP ) (tBu2PCH2SiMe2)2N-, proceed with
rapid loss of N2 to deliver (PNP)RuN, a (slightly) nonplanar
molecule with a multiple Ru/N bond containing d4-RuIV. This
molecule is persistent at 25 °C for days, both in benzene
solution and in the solid state. We report here results that
enable evaluation of periodic trends up Group 8B by
describing the analogous iron chemistry. Iron chemistry of
imide and nitride ligands has experienced considerable recent
study,2–9 including one puzzling report,10 showing how an
FeN+ unit is thermodynamically fated to form FeNNFe2+;
Results
Synthesis of (PNP)FeCl. The reaction of equimolar
* To whom correspondence should be addressed. E-mail: caulton@
indiana.edu.
(1) Walstrom, A.; Pink, M.; Yang, X.; Tomaszewski, J.; Baik, M.-H.;
Caulton, K. G. J. Am. Chem. Soc. 2005, 127, 5330.
(2) Sadique, A. R.; Gregory, E. A.; Brennessel, W. W.; Holland, P. L.
J. Am. Chem. Soc. 2007, 129, 8112.
(PNP)MgCl with anhydrous FeCl2 in THF resulted in a clean
1
conversion to (PNP)FeCl. The H NMR spectrum (Figure
1) shows that (PNP)FeCl appears to be C2V symmetric in
solution, with three broad signals for the CH2, Si-Me, and
P-tbutyl peaks. A 31P NMR spectrum shows no signal in a
wide window suggesting that the phosphorus atoms are
directly bound to a paramagnetic iron center. A magnetic
moment determination (Evans method) was performed and
confirmed the complex to be high spin (µeff ) 4.6 µB) with
four unpaired electrons. A crystal structure determination
(3) Eckert, N. A.; Vaddadi, S.; Stoian, S.; Lachicotte, R. J.; Cundari, T. R.;
Holland, P. L. Angew. Chem., Int. Ed. 2006, 45, 6868.
(4) Stoian, S. A.; Vela, J.; Smith, J. M.; Sadique, A. R.; Holland, P. L.;
Muenck, E.; Bominaar, E. L. J. Am. Chem. Soc. 2006, 128, 10181.
(5) Smith, J. M.; Lachicotte, R. J.; Pittard, K. A.; Cundari, T. R.; Lukat-
Rodgers, G.; Rodgers, K. R.; Holland, P. L. J. Am. Chem. Soc. 2001,
123, 9222.
(6) Mankad, N. P.; Whited, M. T.; Peters, J. C. Angew. Chem., Int. Ed.
2007, 46, 5768.
(7) Rohde, J.-U.; Betley, T. A.; Jackson, T. A.; Saouma, C. T.; Peters,
J. C.; Que, L., Jr. Inorg. Chem., 2007, 46, 5720.
(11) Costas, M.; Mehn, M. P.; Jensen, M. P.; Que, L., Jr. Chem. ReV.,
2004, 104, 939.
(8) Hendrich, M. P.; Gunderson, W.; Behan, R. K.; Green, M. T.; Mehn,
M. P.; Betley, T. A.; Lu, C. C.; Peters, J. C. Proc. Natl. Acad. Sci.
U. S. A. 2006, 103, 17107.
(12) Shaik, S.; Hirao, H.; Kumar, D. Acc. Chem. Res. 2007, 40, 532.
(13) Brown, S. D.; Mehn, M. P.; Peters, J. C. J. Am. Chem. Soc. 2005,
127, 13146.
(9) Thomas, C. M.; Mankad, N. P.; Peters, J. C. J. Am. Chem. Soc. 2006,
128, 4956.
(14) Ingleson, M. J.; Pink, M.; Caulton, K. G. J. Am. Chem. Soc. 2006,
128, 4248.
(10) Betley, T. A.; Peters, J. C. J. Am. Chem. Soc. 2004, 126, 6252.
10.1021/ic702279b CCC: $40.75  2008 American Chemical Society
Inorganic Chemistry, Vol. 47, No. 12, 2008 5129
Published on Web 05/15/2008

Buschhorn et al.
1
t
Figure 1. H NMR spectrum (300 MHz) of [(tBu2PCH2SiMe2)2N]FeCl in C6D6(trace C6D5H at 7.15 ppm) at 20 °C. Bu: -0.8 ppm. SiMe: 24 ppm.
the PNPPh ligand contains phenyl substituents on P. These
are both S ) 2 systems and, thus, are nonplanar.16
(PNP)FeCl shows no reaction in benzene in the presence
of (equimolar) N2, H2, or CO.
Synthesis of (PNP)Fe(N3). Reaction of (PNP)FeCl with
excess NaN3 in THF gives conversion to (PNP)Fe(N3),
1
characterized by H NMR and also by infrared spectrosco-
py(azide band at 2073 cm-1). While the solid is nearly
1
colorless, benzene solutions are dark yellow. The H NMR
spectrum of (PNP)Fe(N3) shows time-averaged C2V sym-
t
metry, with one signal each for Bu and for SiMe groups.
Its chemical shifts are readily distinguished from those of
the chloride, so that any chloride impurity can be readily
detected in the azide.
Figure 2. ORTEP drawing (50% probabilities) of the non-hydrogen atoms
of [(tBu2PCH2SiMe2)2N]FeCl. Unlabeled atoms are carbon. Selected
structural parameters: Fe1-N1 ) 1.9824(18) Å, Fe1-Cl1 ) 2.2708(7) Å,
Fe1-P1 ) 2.4943(6) Å, Fe1-P2 ) 2.5824(6) Å, N1-Fe1-Cl1 )
139.98(6)°, N1-Fe1-P1 ) 92.50(5)°, Cl1-Fe1-P1 ) 103.59(2)°,
N1-Fe1-P2 ) 88.22(5)°, Cl1-Fe1-P2 ) 103.01(2)°, P1-Fe1-P2
)137.07(2)°.
Crystals grown from pentane show (Figure 3) a nonplanar
coordination geometry with transoid angles from 131° to
136° and a bent bonding of azide to Fe (Fe-N2-N3 )
137°). The two Fe/P distances differ only modestly, by 0.06
Å, perhaps because of the different conformations of the two
fused 5-membered rings of the pincer (angles P-Fe-N1 )
89° and 94°). The Fe/N distances differ negligibly. The FeN3
group is oriented to minimize any contacts with the four tBu
groups.
(Figure 2) shows a monomeric four-coordinate nonplanar
structure around iron.15 Because this is, at best, Cs symmetric,
1
the higher symmetry observed in the H NMR spectrum
Toward a Nitride Complex. (a) Synthesis and Char-
acterization. Photolysis17–19 of this azide complex in
benzene gives rapid evolution of gas, with darkening, to give
a product whose 1H NMR spectrum shows two signals each
implies that the chloride rapidly flexes above and below the
P-N-P plane in solution (inverting the geometry at iron),
resulting in the observed spectrum being time-averaged to
C2V symmetry. The molecule has significantly (by 0.1 Å)
inequivalent Fe/P distances, apparently caused by the dif-
ferent ring conformations of the two fused rings of the
coordinated pincer. All distances to iron are long enough to
be consistent with a high-spin FeII. The Fryzuk group has
reported (PNPPh)FeX with X ) Cl and CH(SiMe3)2 when
t
for Bu, SiMe, and CH2. The 31P NMR spectrum of this
(16) Fryzuk, M. D.; Leznoff, D. B.; Ma, E. S. F.; Rettig, S. J.; Young,
V. G., Jr. Organometallics, 1998, 17, 2313.
(17) Berry, J. F.; Bill, E.; Bothe, E.; George, S. D.; Mienert, B.; Neese, F.;
Wieghardt, K. Science, 2006, 312, 1937.
(18) Aliaga-Alcalde, N.; George, S. D.; Mienert, B.; Bill, E.; Wieghardt,
K.; Neese, F. Angew. Chem., Int. Ed. 2005, 44, 2908.
(19) Petrenko, T.; DeBeer George, S.; Aliaga-Alcalde, N.; Bill, E.; Mienert,
B.; Xiao, Y.; Guo, Y.; Sturhahn, W.; Cramer, S. P.; Wieghardt, K.;
Neese, F. J. Am. Chem. Soc. 2007, 129, 11053.
(15) Renkema, K. B.; Ogasawara, M.; Streib, W. E.; Huffman, J. C.;
Caulton, K. G. Inorg. Chim. Acta 1999, 291, 226.
5130 Inorganic Chemistry, Vol. 47, No. 12, 2008

Photolytic Approach to the FeN+ Moiety
unchanged at +2. The asymmetric unit of the unit cell
contains one full dimer and one-half dimer, the latter
possessing a crystallographic center of symmetry. There are
no significant differences in bond lengths or angles between
the two dimers. Even the dimer in the general position in
the asymmetric unit has an idealized C2 axis relating the two
ends of the molecule. The three coordinate geometry at iron
is distorted toward T-shaped; this probably reflects what
would be steric interference between the bridging NdPtBu2
group and the PtBu2 group, which is pendant, because that
is the largest angle around Fe. The dimer is planar (within
2°) at both metals and at both phosphiniminate nitrogens.
All three Fe/N distances to a given iron are equal to within
(0.02 Å, and the P/N distances are short, consistent with a
multiple bond.21 Both 2-fold-symmetric Fe2N2 quadrangles
have their shorter Fe/N distances to the iron to which that N
is not chelated. The Fe/Fe distance, ∼2.55 Å, is comparable
to that in other related compounds.22 The larger N-Fe-N
angle, ∼149° (vs 112°), is nevertheless apparently not
suitable for binding the bulky PtBu2 donor of the pendant
arm. The 6-membered ring formed by the coordinated pincer
arm is highly nonplanar, which twists the amide plane
unusually far from the iron coordination plane, an effect
unusual in η3-pincer structures. However, if ring torsions
have only a modest barrier; then the dimer will have a time-
averaged mirror plane of symmetry containing the Fe2N2
quadrangle, hence the observed equivalence of both tBu and
Me groups on a given P or Si. This easy flexibility is
consistent with the fact that the half-dimer has rigorous
inversion symmetry, hence a different conformation from
the whole dimer in the asymmetric unit (idealized C2
symmetry).
Figure 3. ORTEP view (50% probabilities) of the non-hydrogen atoms of
(PNP)Fe(N3). Unlabeled atoms are carbon. Selected structural parameters:
Fe1-N2 ) 1.9613(14) Å, Fe1-N1 ) 1.9691(12) Å, Fe1-P2 ) 2.4878(4)
Å, Fe1-P1 ) 2.5503(4) Å, N2-N3 ) 1.191(2) Å, N3-N4 ) 1.151(2) Å,
N2-Fe1-N1 ) 131.00(6)°, N2-Fe1-P2 ) 106.28(5)°, N1-Fe1-P2 )
94.16(4)°,N2-Fe1-P1)103.66(5)°,N1-Fe1-P1)89.46(4)°,P2-Fe1-P1
) 136.333(15)°, N3-N2-Fe1 ) 137.46(14)°, N4-N3-N2 ) 175.9(2)°.
(b) Energetics and Mechanism. Why is this product
formed? We have addressed some aspects of this question
by DFT(PBE) calculation of both an anticipated primary
product, (PNP)FeN, and also the monomeric part of the
dimer, (PNPdN)Fe, the latter with the ligand bound both
η2 and η3. We have calculated both singlet and triplet states
for all these species.
There are two general classes of mechanism, distinguished
by the timing of P/N bond formation versus dimerization
events: which occurs first? If P/N bonding occurs first, it
could either be a migration of coordinated P to N or first
dissociation to form one free P arm, then unimolecular attack
on the local N. The fact that we find very inequivalent Fe/P
bond lengths in both the chloride and the anilide reported
here (below) indicates that it might be generally possible to
dissociate one phosphine arm. If this dissociation of one P
arm begins the reaction, then this nucleophilic P could also
attack N on a different iron, to start dimerization concurrent
with P/N bond formation. However our past reasoning in
(PNP)M chemistry has generally been that the steric protec-
tion conferred by the PNP ligand prevents reactions bimo-
Figure 4. ORTEP view (50% probabilities) of the non-hydrogen atoms of
the general position dimer [(PNPdN)Fe]2. Unlabeled atoms are carbon.
Selected structural parameters: Fe1-N3 ) 1.895(4) Å, Fe1-N2 ) 1.936(4)
Å, Fe1-N1 ) 1.946(4) Å, Fe1-Fe2 ) 2.5417(10) Å, Fe2-N1 ) 1.908(4)
Å, Fe2-N4 ) 1.939(4) Å, Fe2-N3 ) 1.943(4) Å, P1-N1 ) 1.578(4) Å,
P3-N3 ) 1.587(4) Å, N3-Fe1-N2 ) 148.42(18)°, N3-Fe1-N1 )
97.26(17)°, N2-Fe1-N1 ) 112.85(17)°, N1-Fe2-N4 ) 149.19(18)°,
N1-Fe2-N3 ) 96.92(17)°, N4-Fe2-N3 ) 112.61(17)°.
product shows no signal, consistent with line broadening
from rapid paramagnetic relaxation. Together these NMR
data indicate that the product is not a singlet state and is not
C2V symmetric. Crystals grown from pentane were shown
(Figure 4) to contain the expected two nitrogens per Fe but
in a dimeric structure and with formation of one imine PdN
bond for each Fe.20 One arm of each former pincer ligand
is pendant, leaving each iron three coordinate. A nitrogen
ligand has coupled with trivalent phosphorus. Because P is
oxidized by two electrons, the iron oxidation state remains
(21) Dehnicke, K.; Straehle, J. Polyhedron 1989, 8, 707.
(22) Olmstead, M. M.; Power, P. P.; Shoner, S. C. Inorg. Chem. 1991, 30,
2547.
(20) LePichon, L.; Stephan, D. W.; Gao, X.; Wang, Q. Organometallics
2002, 21, 1362.
Inorganic Chemistry, Vol. 47, No. 12, 2008 5131

Buschhorn et al.
of the reagent nitride nitrogen and not by one or two frontier
orbitals but rather by more general reaction characteristics,
including those of the product.
Fe/P dissociation prior to oxidation of phosphorus en-
counters an energy cost to form the stationary state singlet
(η2-PNP)FeN of +23.0 kcal/mol. While not impossible, this
is probably inconsistent with the rapid rate of our reaction
(at 25 °C in a water-cooled photochemical reactor). The
reaction certainly does not proceed via an excited state
because thermalization/decay to ground electronic state
would have occurred long before this stage is reached (i.e.,
soon after N2 loss). We therefore favor a unimolecular
migration/insertion for forming the P/N bond: P/N bond
formation concurrent with Fe/P bond scission.
Synthesis of (PNP)Fe[NH(xylyl)]. Anilide ligands serve
to evaluate the interaction of a metal center with a reduced
form of nitrogen.24–26 The title molecule was rapidly formed
by reaction of equimolar LiNH(xylyl) with (PNP)FeCl in
benzene. The structure of a light brown crystal grown from
pentane (Figure 7) shows a nonplanar coordination geometry.
The hydrogen found in the difference Fourier map confirms
that the molecule is an anilide, not an imide. The nonplanarity
at iron is best seen in the two trans angles, PFeP at
138.128(17)° and NFeN at 127.08(6)°. The distance from
iron to the anilide N is only slightly (0.04 Å) shorter than
that to the silylamide N; this distance falls in the range
(1.92-1.94Å) of other 4-coordinate FeII anilides.27–29 The
Fe-N2-Cipso angle of 141.01(12)° is not characteristic of
an imide. The molecule has no agostic interactions to any
methyl (shortest distance to a tBu or SiMe carbon is 3.71 Å
and to xylyl methyl is 3.31 Å). An unusual feature of the
coordination geometry is that the two Fe/P distances differ
dramatically (by 0.1 Å) with the longer distance being to
P1, which is part of the highly nonplanar five-membered
chelate ring; this may be diagnostic of strain because the
other pincer ring is nearly planar. Without this distortion,
which leaves the two nonbonded P/N2 distances equal within
0.06 Å, one such contact might be destabilizing. The angles
from anilide N to P are identical (106°). The silylamide
nitrogen nevertheless remains planar in this pincer conforma-
tion. This ferrous complex is crystallographically isomor-
phous with its CoII analog.30 M-P and M-N(xylyl) distances
are shorter for M ) Co.
Figure 5. DFT-optimized structure of (PNPdN)Fe; only the non-hydrogen
atoms are shown.
lecular in complex.13,23 Thus, we focused our DFT(PBE)
calculation on the two unimolecular mechanisms; all quoted
energies are electronic energies.
We found minima for both singlet and triplet (PNP)FeN,
with the singlet more stable by 14.5 kcal/mol. The reaction
energies for conversion of (PNP)Fe(N3), which is calculated
to have a quintet ground state, to singlet (PNP)FeN + N2 is
-4.7 kcal/mol, thus favorable, but only modestly so.
However, the reaction energy for singlet (PNP)FeN to
isomerize to (PNPdN)Fe (Figure 5), the monomer of the
observed dimer, is -11.2 kcal/mol. The iron in this three-
coordinate product (Figure 5) is divalent, and this species is
again found to have a quintet ground state. Thus, the overall
reaction for monomeric iron species is favorable, and the
last step in particular, reduction of FeIV to FeII, as phosphorus
is oxidized from +3 to +5, drives the reaction.
This P/N bond formation might occur if the nitrogen is
oxidizing, as it is in RN3.21 Singlet (PNP)FeN was therefore
analyzed for such character at N or in the FeN unit. The
calculated Fe/N distance to the nitride is 1.52 Å and, thus,
is consistent with a triple bond; for comparison, the Fe/NSi2
distance is 1.94 Å. Three orbitals in Figure 6 comprise the
Fe-N σ and an orthogonal pair of π bonds. The three
highest-energy occupied orbitals (see Supporting Informa-
tion) are two primarily d orbitals (hence d4-FeIV) and a lone
pair mainly on the amide nitrogen(HOMO-2).
The character of the Fe/nitride bond is also revealed by
comparison to the geometric and electronic structure of the
triplet state of (PNP)FeN. As shown in Table 1, the Fe/N
distance to nitride (N2) is longer in the triplet (as are all
other metal/ligand bonds). The triplet spin densities are
mainly on iron(1.89e), with only 0.04 on amide N and 0.11
on nitride N. Two Fe/N π bonding orbitals can be identified
(Figure 6), but SOMO1 is π*FeN, thus reducing the FeN bond
order below 3. It is the two SOMOs that give the (modest)
spin density on nitride N, yet the nitride reacts as an oxidant.
Therefore, the favorable thermodynamics for PdN bond
formation is not determined simply by ground-state character
A least-squares fit31 of the structures of (PNP)FeCl to that
of (PNP)Fe(N3) shows them to be extremely similar, even
to the conformation around single bonds. The anilide is more
(24) Brown, S. D.; Peters, J. C. J. Am. Chem. Soc. 2004, 126, 4538.
(25) Betley, T. A.; Peters, J. C. J. Am. Chem. Soc. 2003, 125, 10782.
(26) Brown, S. D.; Betley, T. A.; Peters, J. C. J. Am. Chem. Soc. 2003,
125, 322.
(27) Eckert, N. A.; Smith, J. M.; Lachicotte, R. J.; Holland, P. L. Inorg.
Chem. 2004, 43, 3306.
(28) Au-Yeung, H. Y.; Lam, C. H.; Lam, C.-K.; Wong, W.-Y.; Lee, H. K.
Inorg. Chem. 2007, 46, 7695.
(29) Stokes, S. L.; Davis, W. M.; Odom, A. L.; Cummins, C. C.
Organometallics 1996, 15, 4521.
(30) Ingleson, M. J.; Pink, M.; Fan, H.; Caulton, K. G. Inorg. Chem. 2007,
46, 10321.
(23) Indeed our DFT search for bimolecular intermediates here led only to
separation of the two halves during geometry optimization.
(31) See Supporting Information.
5132 Inorganic Chemistry, Vol. 47, No. 12, 2008

Photolytic Approach to the FeN+ Moiety
Figure 6. Selected frontier orbitals of singlet (upper) and triplet (lower) (PNP)FeN.
Table 1. Selected Bond Lengths (Å) and Angles (deg) of (PNP)FeN
singlet
triplet
Fe-P1
2.271
2.302
1.938
1.524
141.4
134.5
Fe-P1
2.287
2.432
2.050
1.586
138.4
156.3
Fe-P2
Fe-P2
Fe-N1
Fe-N2
P1-Fe-P2
Fe-N1
Fe-N2
P1-Fe-P2
N1-Fe-N2
N1-Fe-N2
different from the chloride, specifically concerning the pincer
ring conformations. For all three structures, the Fe-NSi2
distances are unexceptional.16,22,32,33
The 1H NMR spectrum of (PNP)Fe[NH(xylyl)] in benzene
t
shows one signal each for Bu, SiMe, and CH2 protons,
suggesting C2V symmetry by time-averaged inversion of the
nonplanar Fe coordination geometry. However, the 2,6-xylyl
protons show inequivalent methyl signals and inequivalent
meta protons(as well as one para proton signal), suggesting
slow rotation around the N-Cipso bond, even as the iron
inverts rapidly. This shows how the planar anilide is hindered
by the four flanking tBu groups, as was the Fe/azide group.
Figure 7. ORTEP view (50% probabilities) of (PNP)Fe[NH(xylyl)]. Only
the hydrogen on the anilide nitrogen is illustrated. Unlabeled atoms are
carbon. Selected structural parameters: Fe1-N2 ) 1.9505(15) Å, Fe1-N1
) 1.9893(14) Å, Fe1-P2 ) 2.5480(5) Å, Fe1-P1 ) 2.6336(5) Å,
N2-Fe1-N1 ) 127.08(6)°, N2-Fe1-P2 ) 107.03(5) Å, N1-Fe1-P2 )
92.50(4) Å, N2-Fe1-P1 ) 106.23(5) Å, N1-Fe1-P1 ) 87.23(4) Å,
P2-Fe1-P1 ) 138.128(17) Å, C23-N2-Fe1 ) 141.01(12) Å.
Discussion
iron species reported here, and which is recovered unchanged
after 4 days reflux in benzene. Indeed, to elicit (thermal)
reactivity, it was necessary to first do one-electron outer
sphere oxidation (with ferricinium) of this CoII species, and
then (PNP)Co(N3)+ rapidly evolved N2. As with the iron case
here, the product detected at 25 °C is not a terminal nitride
(Figure 8). However, the oxidation product (PNP)CoN+, now
isoelectronic with the Fe case, showed reactivity even beyond
that shown by the isoelectronic Fe analog: “deeper” ligand
The apparent attack here of phosphorus on a nitride ligand
has some relation to chemistry we have reported34 for
(PNP)Co(N3), which itself has one more electron than the
(32) Andersen, R. A.; Faegri, K.,Jr.; Green, J. C.; Haaland, A.; Lappert,
M. F.; Leung, W. P.; Rypdal, K. Inorg. Chem. 1988, 27, 1782.
(33) Panda, A.; Stender, M.; Olmstead, M. M.; Klavins, P.; Power, P. P.
Polyhedron 2003, 22, 67.
(34) Ingleson, M. J.; Pink, M.; Fan, H.; Caulton, K. G. J. Am. Chem. Soc.
2008, 130, 4262.
Inorganic Chemistry, Vol. 47, No. 12, 2008 5133

Buschhorn et al.
Figure 8. Schematic and ORTEP view (50% probabilities) of the cationic product from outer sphere oxidation of (PNP)Co(N3).
rearrangement. Thus, although a P/N bond was formed, this
was accompanied by reversal of the Si-C and Si-N bonds,
to give two small rings. This is obviously a way to keep
t
coordination number four for P, and to minimize Bu
repulsions but avoid the entropy penalty of dimerization
which characterizes the iron analog. A feature shared by Fe
and Co is that one M/P bond is lost by double bonding to
the nitride. In sum, the isoelectronic cationic analog for cobalt
takes a different course, giving a product with wholly
different connectivity than the iron case. This difference is
perhaps a result of the higher charge in the cobalt case (more
electrophilic metal center), together with a consequent higher
need of cobalt to be four coordinate. For example, the higher
charge of CoIII may prevent loss of phosphorus donor
subsequent to PdN bond formation, hence preventing
dimerization from crowded 3-coordinate species
metal favoring more its higher oxidation state). For the PdN
bond formation, this reaction is about 37 kcal/mol more
favorable for iron, consistent with this reduction of metal
being favored for the lighter metal. Indeed the reaction energy
for formation of (PNPdN)Ru from its nitride isomer is +25.6
kcal/mol, so will not occur; this is consistent with experiment
in that (PNP)RuN is persistent for days in solution at 25 °C.
Another example of PdN bond formation has been
reported, for P on a tantalum nitride, where free phosphorus
is less likely.37 This is a case, contrasting to ours for Ru,
where even a 5d metal is reduced rather than adopt the TaV
state with N3-.
a
(PNPdN)Co+. The longer lifetime of this monomeric species
then permits intramolecular N attack on Si, leading to the
observed cobalt product. The unfavorable Coulomb factor
for dimerization of two cations will also lengthen the lifetime
of the cobalt monomer. Support for this idea comes from
t
DFT calculation of the energy for dissociation of the Bu2P
arm from (PNPdN)Fe versus (PNPdN)Co+. For iron, the
energy is 23.7 kcal/mol (both complexes are quintet ground
states), while for (PNPdN)Co+ the energy is 29.4 kcal/mol
(here both are triplets).
The expulsion of N2 is a highly selectively photochemical
event because heating a solution of (PNP)Fe(N3) for 48 h at
80 °C in toluene returned only unreacted starting material.
Indeed, the thermal stability is noteworthy and consistent
with the large number of azide complexes studied recently
for their magnetic properties.35,36 This thermal stability is
especially interesting because21 the thermal Staudinger
reaction converts free phosphine and azide R′N3 to R3P )
N3R′; the fact that A does not form upon heating
(PNP)Fe(N3) indicates that binding to FeII suppresses this
reaction, even if a “phosphine arm-off” species is ther-
mally accessible.
It is also interesting that (PNP)Fe[NH(xylyl)] is not a
singlet ground state(based on NMR spectra), given that there
is π donation from both silylamide and anilide lone pairs,
which might have forced spin pairing. Apparently neither
of these is a sufficiently strong π-donor to separate the d
orbitals enough to overcome the spin pairing energy in this
low coordination number and quasi-tetrahedral ligand field.
Hence two strong π bonds are needed to cause spin pairing.
The work described shows that intramolecular capture of
a presumed FeN+ unit is the favored reaction channel. While
this represents loss of the reactive monatomic ligand for
bimolecular use, it does serve to define its high reactivity.
Intramolecular capture is increasingly the fate of later 3d
metalloligand multiple bonds in high oxidation states. In
general, ligand vulnerabilities (here phosphorus is more
reducing than iron) set limits on this metal/O or metal/N
chemistry, but every ligand has its specific vulnerability,37–41
Comparison to the ruthenium analog1 is informative of
periodic trends based on the DFT calculations. Why are the
products different for iron versus ruthenium? Loss of N2 from
(PNP)M(N3) is calculated to be about 16 kcal/mol more
favorable for M ) Ru than it is for iron, consistent with the
heavier divalent analog being more reducing (i.e., the heavier
(37) Spencer, L. P.; MacKay, B. A.; Patrick, B. O.; Fryzuk, M. D. Proc.
Natl. Acad. Sci. U. S. A. 2006, 103, 17094.
(35) Ribas, J.; Escuer, A.; Monfort, M.; Vicente, R.; Cortes, R.; Lezama,
L.; Rojo, T. Coord. Chem. ReV. 1999, 1027.
(38) Morello, L.; Yu, P.; Carmichael, C. D.; Patrick, B. O.; Fryzuk, M. D.
J. Am. Chem. Soc. 2005, 127, 12796.
(36) Escuer, A.; Aromi, G. Eur. J. Inorg. Chem. 2006, 472.
5134 Inorganic Chemistry, Vol. 47, No. 12, 2008

Photolytic Approach to the FeN+ Moiety
concentrated under vacuum to ∼5 mL and placed into a -40 °C
freezer; after two days no crystals were observed. The sample was
evacuated to dryness, dissolved in pentane, and filtered. The filtrate
was concentrated under vacuum to ∼3 mL and placed into a -40
°C freezer; after two weeks crystals formed. Yield: 0.02 g(66%).
The colorless and brown crystals formed were used for X-ray
crystallography. 1H NMR (300 MHz, C6D6): δ 23.9 (br s, Si-CH3),
and diminishing the oxidizability of the ligand donor atom
is required based on the results reported here.
Experimental Section
General Considerations: Standard techniques for inert atmo-
sphere (argon) conditions were used for air-sensitive manipulations.
All solvents, including deuterated NMR solvents, were dried and
t
2.61 (br s, butyl); CH2 groups were not observed. IR (pentane
1
stored under argon. H and 31P NMR spectra were recorded on
solution): 2073 cm-1. MS (positive ion methane CI): (PNP)FeN3
(C22H52FeN4P2Si2), calcd 546.647 g/mol. No parent ion observed
apparently because of loss of azide under the energetic conditions.
[(PNPdN)Fe]2. The procedure described for the synthesis of
(PNP)FeN3 was used to synthesize an aliquot of (PNP)FeN3. After
the filtration from the toluene step described above, the sample was
evacuated to dryness and dissolved in C6D6 and filtered into a
J-Young NMR tube. Reference 1H NMR and 31P NMR spectra were
taken, and the sample, held at 25 °C, was irradiated using a medium-
pressure mercury UV lamp in 10 min intervals, followed by
collection of 1H NMR spectra. After the sample had been irradiated
for a total of 20 min, 1H NMR showed complete (95%) conversion
of (PNP)FeN3 to product.
Varian spectrometers, either a Gemini XL300 or a Unity I400
instrument, with chemical shifts reported in parts per million and
referenced to each specific solvent, with the exception of 31P which
was externally referenced to H3PO4 (neat). No 31P NMR signal was
observed for any of the paramagnetic species reported here. All IR
values reported were taken in a 0.1 mm path length KBr gastight
solution cell dissolved in pentane or as solids. Because these new
compounds are very soluble and resist crystallization, product
homogeneity is supported by NMR spectra in the Supporting
Information. In multiple executions, the progress of each reaction
over time was monitored by multinuclear NMR spectroscopy, and
yields are essentially quantitative, unless indicated otherwise. Mass
spectra were determined on a Thermo Electron Corporation MAT
95XP-Trap mass spectrometer. In a nitrogen atmosphere glovebox,
a solution in pentane was applied to the tungsten filament and dried.
The filament was then quickly introduced into the instrument
vacuum chamber and heated to approximately 400 °C.
The sample was evacuated to dryness and redissolved in pentane.
The solution was placed in a vial with a lightly perforated top to
allow for slow evaporation of pentane in a glovebox. When slow
evaporation proved to be an ineffective means of crystallization,
the sample was redissolved in a small quantity of a 9:1 mixture of
pentane and toluene and placed into a -40 °C freezer. After two
weeks, the sample had formed small crystals, which were used for
X-ray data collection. 1H NMR (300 MHz, C6D6):6.0 (br s, CH2),
(PNP)FeCl. Commercial anhydrous FeCl2 (s) was heated at 120
°C for two days under vacuum and then allowed to cool overnight
under vacuum. IR analysis was performed on the FeCl2 in a Nujol
mull to verify that no water was present. FeCl2 (0.045 g, 0.35 mmol)
was reacted with 0.200 g of (PNP)MgCl ·dioxane (0.34 mmol) in
30 mL of dry THF. The heterogeneous mixture was stirred overnight
resulting in a homogeneous clear yellow-brown solution. This was
dried under vacuum, dissolved in a minimal amount of toluene,
and filtered through a coarse frit. The resulting deep yellow-brown
filtrate was dried slowly, then dissolved in 5 mL of pentane and
filtered using a fine frit filter. The clear deep yellow filtrate was
collected, concentrated to approximately 1 mL, and placed into a
-40 °C freezer overnight. After 12 h, purple and colorless crystals
totaling approximately 0.025 g (0.046 mmol) were collected for a
13% yield. The best of each color type were set aside for X-ray
analysis as detailed below. The two color sets of crystals were
shown to have the same composition, with the varying colors the
t
t
4.8 (s, Si-CH3), 3.6 (s, butyl), 1.5 (s, CH2), 1.3 (s, butyl), -3.1
(s, Si-CH3).
The full dimer in the asymmetric unit shows disorder of only
one of the pendant phosphine donor arms, in the form of two
conformers (populated 82:18) around the N-Si bond. The half-
dimer shows no disorder.
(PNP)Fe[NH(xylyl)]. (PNP)FeCl (0.022 g, 0.041 mmol) was
added to a J-Young NMR tube and dissolved in C6D6, to which
was added LiNH(2,6-Me2C6H3) (0.006 g, 0.047 mmol). On addition,
a slight color change from yellow to brown was observed. The
sample was mixed by tumbling overnight. Then, 1H NMR and 31
P
1
NMR spectra were taken of the sample, with H NMR showing
complete conversion and the formation of a paramagnetic product
(<90% yield of crude product). The sample was evacuated to
dryness, dissolved in pentane, and filtered into a J-Young tube. The
filtrate was concentrated to ∼1 mL, and placed into a -40 °C
freezer. After one week crystals had formed, and these were used
1
result of dichroism. H NMR (C6D6): δ 24 (br s, 12H, Si-CH3),
21 (br s, 4H, P-CH2-Si), -0.8 (br s, 36H, tbutyl). 31P NMR (C6D6):
no signal. Evans method magnetic susceptibility:31 µeff) 4.6 µB in
C6D6. MS (positive ion methane CI): calcd (PNP)FeCl
(C22H52ClFeNP2Si2), calcd 540.080 g/mol; obsd M+ with correct
isotopic abundances for one Cl.
1
for X-ray structure determination. H NMR (300 MHz, C6D6): δ
68.4 (br s, CH3-Ar), 24.1 (br s, P-CH2-Si), 20.3 (br s, Si-CH3),
t
13.8 (br s, H-Ar) 1.60 (br s, butyl), -3.41 (br s, H-Ar), -10.6
(PNP)FeN3. (PNP)FeCl (0.030 g, 0.056 mmol) was loaded into
a Schlenk flask and dissolved in 20 mL of THF. To this solution
was added a 20-fold molar excess of NaN3 (0.08 g, 1 mmol). The
pale yellow solution was allowed to stir overnight, turning to an
opaque gray by the next day. The solution was evacuated to dryness,
dissolved in toluene, and filtered. NMR at this point showed
complete conversion. The filtrate, a dark yellow color, was
(br s, H-Ar) -80.1 (br s, CH3-Ar).
Acknowledgment. This work was supported by the NSF
(CHE-0544829). We thank Michael Ingleson for enlightening
discussions.
Supporting Information Available: Magnetic susceptibility
determinations, graphics of 1H NMR spectra, crystallographic
narratives, and full details of the DFT calculations. This material
is available free of charge via the Internet at http://pubs.acs.org.
(39) Barrett, A. G. M.; Crimmin, M. R.; Hill, M. S.; Hitchcock, P. B.;
Procopiou, P. A. Angew. Chem., Int. Ed. 2007, 46, 6339.
(40) Kazi, A. B.; Jones, G. D.; Vicic, D. A. Organometallics 2005, 24,
6051.
(41) Conroy, K. D.; Hayes, P. G.; Piers, W. E.; Parvez, M. Organometallics
2007, 26, 4464.
IC702279B
Inorganic Chemistry, Vol. 47, No. 12, 2008 5135