Studies of low-coordinate iron dinitrogen complexes

Article abstract of DOI:10.1021/ja052707x

Understanding the interaction of N₂ with iron is relevant to the iron catalyst used in the Haber process and to possible roles of the FeMoco active site of nitrogenase. The work reported here uses synthetic compounds to evaluate the extent of NN weakening in low-coordinate iron complexes with an FeNNFe core. The steric effects, oxidation level, presence of alkali metals, and coordination number of the iron atoms are varied, to gain insight into the factors that weaken the NN bond. Diiron complexes with a bridging N₂ ligand, L^RFeNNFeL^R (L^R = β-diketiminate; R = Me, ^tBu), result from reduction of [L^RFeCl]_n under a dinitrogen atmosphere, and an iron(I) precursor of an N₂ complex can be observed. X-ray crystallographic and resonance Raman data for L^RFeNNFeL^R show a reduction in the N-N bond order, and calculations (density functional and multireference) indicate that the bond weakening arises from cooperative back-bonding into the N₂ π* orbitals. Increasing the coordination number of iron from three to four through binding of pyridines gives compounds with comparable N-N weakening, and both are substantially weakened relative to five-coordinate iron-N ₂ complexes, even those with a lower oxidation state. Treatment of L^RFeNNFeL^R with KC₈ gives K₂L ^RFeNNFeL^R, and calculations indicate that reduction of the iron and alkali metal coordination cooperatively weaken the N-N bond. The complexes L^RFeNNFeL^R react as iron(I) fragments, losing N₂ to yield iron(I) phosphine, CO, and benzene complexes. They also reduce ketones and aldehydes to give the products of pinacol coupling. The K₂L^RFeNNFeL^R compounds can be alkylated at iron, with loss of N₂.

Full text of DOI:10.1021/ja052707x

Published on Web 12/31/2005
Studies of Low-Coordinate Iron Dinitrogen Complexes
Jeremy M. Smith,†,‡ Azwana R. Sadique,† Thomas R. Cundari,*,§
Kenton R. Rodgers,*,| Gudrun Lukat-Rodgers,| Rene J. Lachicotte,†
Christine J. Flaschenriem,† Javier Vela,† and Patrick L. Holland*,†
Contribution from the Department of Chemistry, UniVersity of Rochester, Rochester, New York
14267, Department of Chemistry, UniVersity of North Texas, Denton, Texas 76203, and
Department of Chemistry and Molecular Biology, North Dakota State UniVersity,
Fargo, North Dakota 58105
Received April 26, 2005; E-mail: holland@chem.rochester.edu
Abstract: Understanding the interaction of N2 with iron is relevant to the iron catalyst used in the Haber
process and to possible roles of the FeMoco active site of nitrogenase. The work reported here uses synthetic
compounds to evaluate the extent of NN weakening in low-coordinate iron complexes with an FeNNFe
core. The steric effects, oxidation level, presence of alkali metals, and coordination number of the iron
atoms are varied, to gain insight into the factors that weaken the NN bond. Diiron complexes with a bridging
t
N2 ligand, LRFeNNFeLR (LR ) â-diketiminate; R ) Me, Bu), result from reduction of [LRFeCl]n under a
dinitrogen atmosphere, and an iron(I) precursor of an N2 complex can be observed. X-ray crystallographic
and resonance Raman data for LRFeNNFeLR show a reduction in the N-N bond order, and calculations
(density functional and multireference) indicate that the bond weakening arises from cooperative back-
bonding into the N2 π* orbitals. Increasing the coordination number of iron from three to four through binding
of pyridines gives compounds with comparable N-N weakening, and both are substantially weakened
relative to five-coordinate iron-N2 complexes, even those with a lower oxidation state. Treatment of
LRFeNNFeLR with KC8 gives K2LRFeNNFeLR, and calculations indicate that reduction of the iron and alkali
metal coordination cooperatively weaken the N-N bond. The complexes LRFeNNFeLR react as iron(I)
fragments, losing N2 to yield iron(I) phosphine, CO, and benzene complexes. They also reduce ketones
and aldehydes to give the products of pinacol coupling. The K2LRFeNNFeLR compounds can be alkylated
at iron, with loss of N2.
Introduction
used in the biosynthesis of nitrogen-containing molecules. Iron
is the only transition metal present in all nitrogenase enzymes,3,4
Atmospheric N2 is an abundant and cheap source of nitrogen
for nitrogen-containing compounds. However, the N2 molecule
is difficult to manipulate due to both thermodynamic factors
(NtN bond strength of 944 kJ mol-1; negative electron affinity;
high ionization energy) and kinetic factors (poor electrophilicity
and nucleophilicity; no permanent dipole).1 Catalysts are used
in nature and industry to activate N2, and the largest-scale
processes use iron.
but iron-molybdenum nitrogenase is the best characterized. In
iron-molybdenum nitrogenase, the site of substrate binding and
reduction is the iron-molybdenum cofactor or “FeMoco”
(Figure 1). The most recent X-ray crystallographic study of the
enzyme (1.16 Å resolution) shows the FeMoco in the native
state to be a MoFe7S9X cluster,5 and Mo¨ssbauer data indicate
an (Fe3+)3(Fe2+)4(Mo4+) oxidation level.6 The six iron atoms
in the center are bridged by a light atom X (C, N, or O).5,6b,7
In nature, the nitrogenase enzymes2 catalyze the reduction
(3) Iron-vanadium nitrogenase: (a) Robson, R. L.; Eady, R. R.; Richardson,
T. H.; Miller, R. W.; Hawkins, M.; Postgate, J. R. Nature 1986, 322, 388-
390. (b) Arber, J. M.; Dobson, B. R.; Eady, R. R.; Stevens, P.; Hasnain, S.
S.; Garner, C. D.; Smith, B. E. Nature 1987, 325, 372-374. (c) Morningstar,
J. E.; Hales, B. J. J. Am. Chem. Soc. 1987, 109, 6854-6855. (d)
Morningstar, J. E.; Johnson, M. K.; Case, E. E.; Hales, B. J. Biochemistry
1987, 26, 1795-1800. (e) Arber, J. M.; Dobson, B. R.; Eady, R. R.;
Hasnain, S. S.; Garner, C. D.; Matsushita, T.; Nomura, M.; Smith, B. E.
Biochem. J. 1989, 258, 733-737. (f) Eady, R. R. Chem. ReV. 1996, 96,
3013-3030. (g) Rehder, D. J. Inorg. Biochem. 2000, 80, 133-136. (h)
Eady, R. R. Coord. Chem. ReV. 2003, 237, 23-30.
of atmospheric dinitrogen to ammonium salts, which are in turn
† University of Rochester.
‡ Current address: Department of Chemistry and Biochemistry, New
Mexico State University, Las Cruces, NM 88003.
§ University of North Texas.
| North Dakota State University.
(1) Hidai, M.; Mizobe, Y. Chem. ReV. 1995, 95, 1115-1133. (b) Fryzuk, M.
D.; Johnson, S. A. Coord. Chem. ReV. 2000, 200-202, 379-409. (c)
Gambarotta, S.; Scott, J. Angew. Chem., Int. Ed. 2004, 43, 5298-5308.
(d) MacKay, B. A.; Fryzuk, M. D. Chem. ReV. 2004, 104, 385-401.
(2) Burgess, B. K. Chem. ReV. 1990, 90, 1377-1406. (b) Burgess, B. K.; Lowe,
D. J. Chem. ReV. 1996, 96, 2983-3011. (c) Howard, J. B.; Rees, D. C.
Chem. ReV. 1996, 96, 2965-2982. (d) Rees, D. C.; Howard, J. B. Curr.
Opin. Chem. Biol. 2000, 4, 559-566. (e) Barrie`re, F. Coord. Chem. ReV.
2003, 236, 71-89. (f) Holland, P. L., “Nitrogen Fixation,” in Comprehen-
siVe Coordination Chemistry II, McCleverty, J.; Meyer, T. J., Eds.
Elsevier: Oxford, 2004; Vol. 8, pp 569-599.
(4) Iron-only nitrogenase: (a) Mu¨ller, A.; Schneider, K.; Knu¨ttel, K.; Hagen,
W. R. FEBS Lett. 1992, 303, 36-40. (b) Schneider, K.; Gollan, U.;
Dro¨ttboom, M.; Selsemeier-Voigt, S.; Mu¨ller, A. Eur. J. Biochem. 1997,
244, 789-800. (c) Siemann, S.; Schneider, K.; Drottboom, M.; Muller, A.
Eur. J. Biochem. 2002, 269, 1650-1661. (d) Krahn, E.; Weiss, B. J. R.;
Krockel, M.; Groppe, J.; Henkel, G.; Cramer, S. P.; Trautwein, A. X.;
Schneider, K.; Muller, A. J. Biol. Inorg. Chem. 2002, 7, 37-45.
(5) Einsle, O.; Tezcan, F. A.; Andrade, S. L. A.; Schmid, B.; Yoshida, M.;
Howard, J. B.; Rees, D. C. Science 2002, 297, 1696-1700.
9
756
J. AM. CHEM. SOC. 2006, 128, 756-769
10.1021/ja052707x CCC: $33.50 © 2006 American Chemical Society

Iron Dinitrogen Complexes
A R T I C L E S
manner reminiscent of proposals by Sellmann.13 A very recent
study shows a single N environment for an apparent adduct of
N2 (or reduced form thereof) on the FeMoco, which is consistent
with a symmetrical binding mode.14
In addition to nitrogenase, there are heterogeneous iron
systems that catalytically reduce dinitrogen. Some iron oxide15
and iron sulfide16 surfaces produce ammonia from N2. Most
industrially relevant is the Haber-Bosch process, in which
ammonia is synthesized from nitrogen and hydrogen over an
iron catalyst.17 Single-crystal iron surfaces reduce N2, and the
(111) face of iron is most active.18 LEED experiments suggest
that N2 bound to Fe(111) is strongly inclined rather than
perpendicular to the surface.19 Therefore, N2 bridged between
iron atoms is again a potential binding mode.
Figure 1. FeMo cofactor “FeMoco” of iron-molybdenum nitrogenase in
the native form. X ) C, N or O.
8
Spectroscopic investigations with 15N2 and the spectroscopic
similarity of active and inactive forms of the enzyme2 argue
against the interstitial atom resulting from splitting of N2,
suggesting that its role may be structural.
Despite the importance of these iron catalysts, the current
understanding of N2 reduction chemistry in synthetic iron
compounds is rudimentary. Some solution Fe-N2 systems have
been reported to give hydrazine or ammonia upon proto-
nation,20-22 but these systems are not understood at a mecha-
nistic level. For example, uncharacterized mixtures of iron(III)
chloride and strong reducing agents are reported to give
hydrazine upon protonation.20 Treatment of some iron-phos-
phine-N2 complexes with excess acid gives limited amounts
of ammonia.23 In a well-characterized recent study, the reduced
The picture in Figure 1 is not strictly relevant to N2 binding
by the FeMoco, because the crystallographically characterized
native state of the FeMoco does not bind substrates. Several
reducing equivalents are required before N2 binds.9 ENDOR
studies of mutant enzymes are most consistent with binding of
products at the central iron atoms.10 These results are most easily
interpreted within a model where reduction and substrate binding
are coupled to cleavage and formation of bonds between iron
and X.11,12 Recent computational studies support this idea,
6b,7d,e,10g
dinitrogen complex Fe(PhBPiPr3)(N2)MgCl(THF)2 (PhBPiPr
)
and one has even located low-energy transition states
3
for each proposed step of N2 binding and activation.7d Interest-
ingly, in this mechanism, the FeMoco breaks open at one
bridging sulfide to bind N2 as an Fe-NN-Fe intermediate, and
this intermediate is subsequently protonated by a thiol, in a
PhB(CH2PiPr2)3-) reacts with electrophiles to give a diazenido
ligand derived from dinitrogen.24
One important difference between the catalytic iron sites and
synthetic iron compounds is the coordination number at iron,
which is typically 5 or 6 for synthetic compounds but possibly
lower in the active forms of the catalysts. Because the bulky
â-diketiminate ligands, LMe and LtBu (Figure 2) stabilize
synthetic complexes with three-coordinate and four-coordinate
iron centers,12 it is possible to evaluate the importance of iron
coordination number. Here, we show that when N2 binds to low-
coordinate iron, the N-N bond becomes much weaker than in
other iron-N2 complexes. Further, we evaluate the structural
(6) (a) Yoo, S. J.; Angove, H. C.; Papaefthymiou, V.; Burgess, B. K.; Mu¨nck,
E. J. Am. Chem. Soc. 2000, 122, 4926-4936. (b) Vrajmasu, V.; Mu¨nck,
E.; Bominaar, E. L. Inorg. Chem. 2003, 42, 5974-5988.
(7) Calculations support the idea that X is a nitrogen atom: (a) Dance, I. Chem.
Commun. 2003, 324-325. (b) Hinnemann, B.; Norskov, J. K. J. Am. Chem.
Soc. 2003, 125, 1466-1467. (c) Lovell, T.; Liu, T.; Case, D. A.;
Noodleman, L. J. Am. Chem. Soc. 2003, 125, 8377-8383. (d) Schimpl, J.;
Petrilli, H. M.; Blo¨chl, P. E. J. Am. Chem. Soc. 2003, 125, 15772-15778.
(e) Huniar, U.; Ahlrichs, R.; Coucouvanis, D. J. Am. Chem. Soc. 2004,
126, 2588-2601. (f) Hinnemann, B.; Norskov, J. K J. Am. Chem. Soc.
2004, 126, 3920-3927. However, recent ENDOR and ESEEM studies
argue strongly against nitrogen as X: Yang, T.-C.; Maeser, N. K.;
Laryukhin, M.; Lee, H.-I.; Dean, D. R.; Seefeldt, L. C.; Hoffman, B. M. J.
Am. Chem. Soc. 2005, 127, 12804-12805.
(13) Sellmann, D.; Sutter, J. Acc. Chem. Res. 1997, 30, 460-469.
(14) Barney, B. M.; Yang, T.-C.; Igarashi, R. Y.; Dos Santos, P. C.; Laryukhin,
M.; Lee, H.-I.; Hoffman, B. M.; Dean, D. R.; Seefeldt, L. C. J. Am. Chem.
Soc. 2005, 127, 14960-14961.
(8) (a) Thomann, H.; Morgan, T. V.; Jin, H.; Burgmayer, S. J. N.; Bare, R. E.;
Stiefel, E. I. J. Am. Chem. Soc. 1987, 109, 7913-7914. (b) Lee, H.-I.;
Benton, P. M. C.; Laryukhin, M.; Igarashi, R. Y.; Dean, D. R.; Seefeldt,
L. C.; Dean, D. R. J. Am. Chem. Soc. 2003, 125, 5604-5605.
(15) Brandes, J. A.; Boctor, N. Z.; Cody, G. D.; Cooper, B. A.; Hazen, R. M.;
Yoder, H. S. Nature 1998, 395, 365-367.
(9) (a) Lowe, D. J.; Thorneley, R. N. F. Biochem. J. 1984, 224, 877-886. (b)
Thorneley, R. N. F.; Lowe, D. J. Biochem. J. 1984, 224, 887-894. (c)
Lowe, D. J.; Thorneley, R. N. F. Biochem. J. 1984, 224, 895-901. (d)
Thorneley, R. N. F.; Lowe, D. J. Biochem. J. 1984, 224, 903-909.
(10) (a) Christiansen, J.; Dean, D. R.; Seefeldt, L. C. Annu. ReV. Plant Physiol.
Plant Mol. Biol. 2001, 52, 269-295. (b) Benton, P. M. C.; Mayer, S. M.;
Shao, J.; Hoffman, B. M.; Dean, D. R.; Seefeldt, L. C. Biochemistry 2001,
40, 13816-13825. (c) Benton, P. M. C.; Christiansen, J.; Dean, D. R.;
Seefeldt, L. C. J. Am. Chem. Soc. 2001, 123, 1822-1827. (d) Mayer, S.
M.; Niehaus, W. G.; Dean, D. R. J. Chem. Soc., Dalton Trans. 2002, 802-
807. (e) Benton, P. M. C.; Laryukhin, M.; Mayer, S. M.; Hoffman, B. M.;
Dean, D. R.; Seefeldt, L. C. Biochemistry 2003, 42, 9102-9109. (f) Lee,
H.-I.; Igarashi, R. Y.; Laryukhin, M.; Doan, P. E.; Dos Santos, P. C.; Dean,
D. R.; Seefeldt, L. C.; Hoffman, B. M. J. Am. Chem. Soc. 2004, 126, 9563-
9569. (g) Igarashi, R. Y.; Dos Santos, P. C.; Niehaus, W. G.; Dance, I. G.;
Dean, D. R.; Seefeldt, L.C. J. Biol. Chem. 2004, 279, 34770-34775. (h)
Barney, B. M.; Igarashi, R.Y.; Dos Santos, P. C.; Dean, D. R.; Seefeldt, L.
C. J. Biol. Chem. 2004, 279, 53621-53624. (i) Dos Santos, P. C.; Igarashi,
R. Y.; Lee, H.-I.; Hoffman, B. M.; Seefeldt, L. C.; Dean, D. R. Acc. Chem.
Res. 2005, 38, 208-214. (j) Barney, B. M.; Laryukhin, M.; Igarashi, R.
Y.; Lee, H.-I.; Dos Santos, P. C.; Yang, T.-C.; Hoffman, B. M.; Dean, D.
R.; Seefeldt, L. C. Biochemistry 2005, 44, 8030-8037. (k) Igarashi, R.
Y.; Laryukhin, M.; Dos Santos, P. C.; Lee, H.-I.; Dean, D. R.; Seefeldt, L.
C.; Hoffman, B. M. J. Am. Chem. Soc. 2005, 127, 6231-6241.
(16) Do¨rr, M.; Ka¨bbohrer, J.; Grunert, R.; Kreisel, G.; Brand, W. A.; Werner,
R. A.; Geilmann, H.; Apfel, C.; Robl, C.; Weigand, W. Angew. Chem.,
Int. Ed. 2003, 42, 1540-1543.
(17) Jennings, J. R. Catalytic Ammonia Synthesis: Principles and Practice;
Plenum: New York, 1991.
(18) (a) Dumesic, J. A.; Topsøe, H.; Boudart, M. J. Catal. 1975, 37, 513-522.
(b) Spencer, N. D.; Schoonmaker, R. C.; Somorjai, G. J. Catal. 1982, 74,
129-135. (c) Falicov, L. M.; Somorjai, G. A. Proc. Natl. Acad. Sci. U.S.A.
1985, 82, 2207-2211.
(19) (a) Grunze, M.; Golze, M.; Hirschwald, W.; Freund, H. J.; Pulm, H.; Seip,
U.; Tsai, M. C.; Ertl, G.; Kueppers, J. Phys. ReV. Lett. 1984, 53, 850-
853. (b) Freund, H. J.; Bartos, B.; Messmer, R. P.; Grunze, M.; Kuhlenbeck,
H.; Neumann, M. Surf. Sci. 1987, 185, 187-202.
(20) (a) Borodko, Y. G.; Broitman, M. O.; Kachapina, L. M.; Shilov, A. E.;
Ukhin, L. Y. J. Chem. Soc. D, Chem. Commun. 1971, 1185-1186. (b)
Bazhenova, T. A.; Ivleva, I. N.; Kachapina, L. M.; Shilova, A. K.; Shilov,
A. E.; Tchoubar, B. J. Organomet. Chem. 1985, 296, 95-101. (c)
Bazhenova, T. A.; Kachapina, L. M.; Shilov, A. E.; Antipin, M. Y.;
Struchkov, Y. T. J. Organomet. Chem. 1992, 428, 107-123.
(21) Yamamoto, A.; Miura, Y.; Ito, T.; Chen, H.-L.; Iri, K.; Ozawa, F.; Miki,
K.; Sei, T.; Tanaka, N.; Kasai, N. Organometallics 1983, 2, 1429-1436.
(22) (a) Hughes, D. L.; Leigh, G. J.; Jimenez-Tenorio, M.; Rowley, A. T. J.
Chem. Soc., Dalton Trans. 1993, 75-82. (b) Leigh, G. J.; Jimenez-Tenorio,
M. J. Am. Chem. Soc. 1991, 113, 5862-5863. (c) Hall, D. A.; Leigh, G.
J. J. Chem. Soc., Dalton Trans. 1996, 3539-3541.
(11) Seefeldt, L. C.; Dance, I. G.; Dean, D. R. Biochemistry 2004, 43, 1401-
1409.
(23) (a) George, T. A.; Rose, D. J.; Chang, Y.; Chen, Q.; Zubieta, J. Inorg.
Chem. 1995, 34, 1295-1298. (b) Gilbertson, J. D.; Szymczak, N. K.; Tyler,
D. R. J. Am. Chem. Soc. 2005, 127, 10184-10185.
(12) The role of X may be to reversibly coordinate to iron in the native form
but release low-coordinate iron in reduced forms of the cofactor: Holland,
P. L. Can. J. Chem. 2005, 83, 296-301.
(24) Betley, T. A.; Peters, J. C. J. Am. Chem. Soc. 2003, 125, 10782-10783.
9
J. AM. CHEM. SOC. VOL. 128, NO. 3, 2006 757

A R T I C L E S
Smith et al.
Figure 2. â-Diketiminate ligands used in this study. These are abbreviated
LMe for R ) CH3 and LtBu for R ) C(CH3)3.
Scheme 1
and spectroscopic effects of changes in oxidation level, coor-
dination number, and steric bulk. The unusual N2 complexes
are characterized through structural, spectroscopic, and theoreti-
cal methods, as well as their characteristic reactivity patterns.
Figure 3. X-ray crystal structures of LMeFeNNFeLMe. Top: pentane solvate
in C2/m (1a). Bottom: solvent-free crystal in P21/n (1b). Thermal ellipsoids
shown at 50% probability, hydrogen atoms omitted for clarity. Selected
bond lengths (Å) and angles (°): Fe-N(N2) 1.745(3) in 1a, 1.775(2) in
1b; Fe-N(diketiminate) 1.949(2) in 1a, 1.945(2) (N11) and 1.984(2) (N21)
in 1b; N-N 1.186(7) in 1a, 1.172(5) in 1b; N(N2)-Fe-N(diketiminate)
130.75(6) in 1a, 155.5(1) (N11) and 109.5(1) (N21) in 1b; N(diketiminate)-
Fe-N(diketiminate) 97.6(1) in 1a, 94.8(1) in 1b.
Results
Synthesis of LRFeNNFeLR. We previously reported that red-
purple LtBuFeNNFeLtBu could be prepared from LtBuFeCl by
reduction with sodium naphthalenide.25 Potassium/graphite
(KC8) has proven to be a more effective reducing agent for
LtBuFeCl and [LMeFe(µ-Cl)]2, leading to LtBuFeNNFeLtBu and
LMeFeNNFeLMe in about 70% yield (Scheme 1).26 LtBuFeNN-
FeLtBu is also produced by photolysis (4 d, RT, high-pressure
structure of LtBuFeNNFeLtBu (the angle between the diketiminate
planes is 92.6°).25 We ascribe this difference to steric interactions
between the two diketiminate ligands, because molecular
mechanics calculations27 based on the crystallographic geom-
etries of 1a and of LtBuFeNNFeLtBu suggest that the parallel
geometry is more stable by 2 kcal/mol in the LMe complex, but
less stable by 4 kcal/mol in the LtBu complex. Interestingly, 1a
and 1b differ by a surprising 20-25° shift of the N2 ligand off
of the line that bisects the diketiminate ligand. Apparently, the
highly symmetric core found in 1a may be distorted by the small
energy of crystal packing forces. The solution 1H NMR spectrum
of 1a in cyclohexane-d12 is consistent with averaged D2h or D2d
symmetry. The complex has a large magnetic moment in
solution, µeff ) 7.9(3) µB, which compares well with its LtBu
analogue (8.4 µB).
46
mercury lamp) of [LtBuFeH]2 under an N2 atmosphere.
We have reported the X-ray crystal structure of
LtBuFeNNFeLtBu 25
Only seven resonances are observed in its
.
1H NMR spectrum (C6D6), suggesting that the diketiminate
ligand contains two mirror planes, and the molecule has
averaged D2h or D2d symmetry in solution. The 1H NMR
spectrum can be assigned on the basis of integration, chemical
shift and relaxation time (T2 from width of peaks). Although
highly sensitive to air and moisture, solutions of LtBuFeNNFeLtBu
show no evidence of decomposition after several hours of
heating in aromatic solvents at 100 °C. LMeFeNNFeLMe, on the
other hand, reacts with aromatic solvents (see below).
We have obtained two different X-ray crystal structures of
LMeFeNNFeLMe (Figure 3). In one crystal (1a), a disordered
pentane molecule lies in the unit cell, and the other crystal (1b)
is free of solvent. The two crystal structures give similar bond
distances, which are similar to those in LtBuFeNNFeLtBu (Table
1). The N-N bond lengths in LMeFeNNFeLMe are 1.18 ( 0.01
Å, indicative of substantial N-N bond weakening relative to
free N2 (1.098 Å). The iron-diketiminate ligand planes in
The weakening of the N-N bond is also observed by
resonance Raman spectroscopy. A band at 1810 cm-1 in
LMeFeNNFeLMe shifts to 1739 cm-1 in LMeFe15N15NFeLMe
(Figure 4A and 4B). Similar N-N bond weakening is observed
in LtBuFeNNFeLtBu, for which νN-N occurs at 1778 cm-1 25
. Both
values are substantially decreased from free N2 (2331 cm-1).
Computations: Ground-State Multiplicities. Our compu-
tational models for evaluation of iron-dinitrogen complexes with
multiconfiguration (MCSCF) and density functional theory
LMeFeNNFeLMe are coplanar in each form of LMeFeNNFeLMe
,
in contrast to the nearly perpendicular ligand planes in the crystal
(25) 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-
9223.
(26) Vela, J.; Stoian, S.; Flaschenriem, C. J.; Mu¨nck, E.; Holland, P. L. J. Am.
Chem. Soc. 2004, 126, 4522-4523.
(27) MM calculations were carried out with the MMFF94 force field (Halgren,
T. A. J. Comput. Chem. 1996, 17, 616-641 and references therein) as
implemented within the Spartan package (Spartan, Wavefunction, Inc.,
18401 Von Karman Ave., Ste. 370, Irvine, CA 92612; http://www.
wavefun.com).
9
758 J. AM. CHEM. SOC. VOL. 128, NO. 3, 2006

Iron Dinitrogen Complexes
A R T I C L E S
Table 1. Fe-N, N-N Bond Lengths, and N-N Bond Stretching Frequencies for Crystallographically Characterized Iron Complexes with
Bridging N2
formal
oxidation
1
complex
Fe CNa
state of Fe
Fe
−N (Å)
N
−N (Å)
ν
NN (cm-
)
ref
[(PEt3)2(CO)2Fe]2N2
5
5
4
4
4
3
0
0
1
1
0.5
1
1.87(1); 1.89(2)
1.876(9)
1.816(2)
1.804(2); 1.794(2)
1.813(2)
1.745(3)
1.13(2)
1.13(1)
42
42
this work
this work
24
[(P(OMe)3)2(CO)2Fe]2N2
LMeFe(tBupy)NN(tBupy)FeLMe
LtBuFe(tBupy)NN(tBupy)FeLtBu
[(PhBPiPr3)FeNNFe(PhBPiPr3)]-
LMeFeNNFeLMe
1.151(3)
1.161(4)
1.171(4)
1.186(7)
1.172(5)
1.192(6)
1.215(6)
1.238(4)
1.241(7)
1770
1810
1778
1625, 1437
1583, 1127
1589, 1123
this work
1.775(2)
LtBuFeNNFeLtBu
3
3
3
3
1
0
0
0
1.760(6); 1.778(6)
1.741(5); 1.761(7)
1.749(3); 1.746(3)
1.773(7); 1.761(7)
25
this work
25
25
K2LMeFeNNFeLMe
Na2LtBuFeNNFeLtBu
K2LtBuFeNNFeLtBu
a Coordination number at the iron atoms.
(2B2) and sextet (6B2) are 34 and 36 kcal/mol higher, respec-
tively. MRMP2/SBKJC(d) calculations on the singly reduced
3
[L′FeNN]- indicate an A1 ground state that is 39 kcal/mol
1
below the A state and 49 kcal/mol below the lowest energy
quintet (5B1).
Single-point calculations on L′FeNNFeL′ and [L′FeNNFeL′]2-
were performed at the MCSCF level of theory. The MCSCF
wave functions were converged within the D2 subgroup of the
full D2d molecular symmetry of the diiron complexes due to
program limitations. SBKJC(d)/FORS(14,6) calculations indicate
the ground state of L′FeNNFeL′ to be a septet (7B3). This septet
state of L′FeNNFeL′ arises from ferromagnetic coupling of
quartet L′FeNN fragments. However, the calculated manifold
of states is dense: Erelative (kcal/mol) ) 0 (7B3), 1 (1A), 3 (5A),
6 (3B3). Because Mo¨ssbauer studies in progress30 and room-
temperature solution magnetic properties support a ferro-
magnetically coupled ground state, subsequent evaluation uses
7
the B3 state.
Computations: Synergism between Iron Centers. Geom-
etry optimization (B3LYP/CZDZ**++) of the neutral, quartet
L′FeNN yields Fe-N ) 1.902 Å and N-N ) 1.123 Å. This
N-N bond length is marginally longer than that in free N2
(1.098 Å). Addition of the other iron atom clearly indicates a
synergism between the Fe atoms in regards to activation of
dinitrogen: B3LYP/CZDZ**++ geometry optimization of
septet L′FeNNFeL′31 shows significant shortening of the Fe-N
bond (Fe-N ) 1.80 Å; ∆(Fe-N) ) -0.10 Å) and lengthening
of the N-N bond (N-N ) 1.19 Å; ∆(N-N) ) +0.06 Å) versus
L′FeNN, Chart 1. Note that the optimized parameters for
L′FeNNFeL′ agree well with the experimental Fe-N and N-N
distances given in Table 1. Coordination of L′Fe to the terminal
N of L′FeNN results in formation of in-phase and out-of-phase
combinations between Fe dπ orbitals and the orbital depicted
in Chart 1. This orbital overlap gives a greater infusion of
electron density into the π* orbital of N2.
Figure 4. Resonance Raman spectra of FeNNFe complexes showing 15N-
sensitive bands. Based on the isotope shifts, bands above 1500 cm-1
are attributed to modes with dominant N-N stretching character. A,
LMeFeNNFeLMe
.
B, LMeFe15N15NFeLMe
.
C, K2LMeFeNNFeLMe
.
D,
K2LMeFe15N15NFeLMe
.
E, Difference spectrum [K2LMeFeNNFeLMe
-
K2LMeFe15N15NFeLMe]. F, LMeFe(tBuPy)NN(tBuPy)FeLMe. G, LMeFe(tBuPy)-
15N15N(tBuPy)FeLMe. The spectra were obtained with 406.7 nm excitation.
Samples A, B, F, and G were prepared in pentane. Samples C and D were
prepared in toluene; toluene bands have been subtracted from the spectra.
(DFT) methods use
L′ (C3N2H5-).28 The starting geometry for calculations was
derived from the experimental geometry of LtBuFeNNFeLtBu
a
truncated diketiminate ligand
.
After replacement of 2,6-diisopropylphenyl and tert-butyl sub-
stituents with hydrogen, the geometry was slightly modified to
yield an idealized D2d geometry in L′FeNNFeL′. To evaluate
the cooperativity between iron centers, L′FeNNFeL′ was further
trimmed to yield L′FeNN and [L′FeNN]- (each with C2V
symmetry), as shown in Chart 1 on the next page.
(30) Stoian, S.; Smith, J. M.; Holland, P. L.; Bominaar, E. L.; Mu¨nck, E.,
manuscript in preparation.
4
We find a B2 ground state for L′FeNN at the MRMP2/
SBKJC(d) level of theory,29 whereas the lowest energy doublet
(31) It was not possible to obtain SCF convergence for ROB3LYP/CSDZ**++
calculations on L′FeNNFeL′. Hence, unrestricted DFT calculations were
employed. The literature suggests that spin contamination is less significant
in UDFT in relation to comparable unrestricted Hartree-Fock (UHF) wave
functions (e.g. ref 38). Geometry optimizations on quartet L′FeNN utilizing
UB3LYP and ROB3LYP methods with the same CSDZ**++ basis set
show little difference in calculated metrics (ROB3LYP/UB3LYP), e.g., FeN
) 1.902 Å/1.901 Å; NN ) 1.123 Å/1.129 Å; FeNL′ ) 1.990 Å/1.980 Å;
bite angle ) 96°/96°; the total spin expectation value S2 is 3.934 versus
the ideal value of 3.750.
(28) (a) Holland, P. L.; Cundari, T. R.; Perez, L. L.; Eckert, N. A.; Lachicotte,
R. J. J. Am. Chem. Soc. 2002, 124, 14416-14424. (b) Vela, J.; Vaddadi,
S.; Cundari, T. R.; Smith, J. M.; Gregory, E. A.; Lachicotte, R. J.;
Flaschenriem, C. J.; Holland, P. L. Organometallics 2004, 23, 5226-5239.
(29) For recent applications of this methodology to transition metal chemistry
see: (a) Aikens, C. M.; Gordon, M. S. J. Phys. Chem. A 2003, 107, 104-
114. (b) Chung, G.; Gordon, M. S. Organometallics 2003, 22, 42-46.
9
J. AM. CHEM. SOC. VOL. 128, NO. 3, 2006 759

A R T I C L E S
Chart 1
Smith et al.
ligand to iron (Table 1). However, this difference is only slightly
larger than the 3σ threshold, and not much greater than the range
seen in the two crystal structures of compound 1 above.34
Resonance Raman spectra of LMeFe(tBupy)NN(tBupy)FeLMe and
its 15N2 isotopomer (Figure 4F,G) show an isotopically sensitive
band at 1770 cm-1 that is assigned to the N-N stretching
vibration. Comparison to LMeFeNNFeLMe shows that the
frequencies for N2 bound to four-coordinate iron and three-
coordinate iron are similar, indicating a similar force constant
for the N-N bond (Table 1).
In the crystal structures of the LFe(tBupy)NNFe(tBupy)FeL
complexes, there is a substantial angle between the two
diketiminate planes (52.22(5)° in the LMe complex and 82.50-
(7)° in the LtBu complex). As a result of the twisting around the
FeNNFe axis, there is no symmetry element in the diketiminate
ligands: the two faces of the diketiminate differ in proximity
to the nearby pyridine ligand, and the two sides of the
diketiminate differ in proximity to the more distant pyridine
1
ligand. However, the room temperature H NMR spectra of
LMeFe(py-R)NNFe(py-R)LMe (py-R ) 4-tBuC5H4N and 4-NMe2-
C5H4N) in pentane-d12 show two sets of peaks for the meta aryl,
isopropyl methyl, and isopropyl methine protons and one set
of peaks for R-methyl, para aryl, and the backbone C-H of
the ligands. This suggests that at room temperature the two sides
of the diketiminate are equivalent and the two faces are not. As
the temperature was decreased to -70 °C, the singlets of the
R-methyl, para aryl, and one set of isopropyl methine and meta-
aryl protons broadened and then split into two peaks. The
observed exchange process is consistent with rotation around
the FeNNFe unit with a barrier of 9.5 ( 0.2 kcal/mol for
LMeFe(tBupy)NNFe(tBupy)LMe and 9.3 ( 0.2 kcal/mol for
LMeFe(Me2Npy)NNFe(Me2Npy)LMe. Because the exchange
process makes the two halves of the ligands equivalent but
retains the inequivalence of the two faces, and because there is
no measurable difference between the barriers for the different
pyridine adducts, we conclude that neither pyridine nor N2
dissociates on the NMR time scale.
Scheme 2
Reactions of the N2 Complexes. In evaluating the reactivity
of LRFeNNFeLR, it is important to know whether the FeNNFe
core is stable in solution. To test this possibility, resonance
Raman spectra of a pentane solution of LtBuFeNNFeLtBu have
been monitored under an atmosphere of 15N2. Very little
incorporation of 15N is observed over several days at room
temperature, suggesting that no rapid cleavage/recombination
events occur. LMeFeNNFeLMe, on the other hand, incorporates
added 15N2 gas in less than 2 d at -78 °C in pentane, showing
that the reduction of steric bulk makes the N2 ligand labile.
Consistent with this idea, LMeFeNNFeLMe reacts with benzene
to displace the N2 ligand irreversibly (Scheme 3). The rate of
this reaction in mixtures of deuterated benzene and cyclohexane
is strongly dependent on the concentration of benzene, sug-
gesting an associative mechanism for the reaction (see Sup-
porting Information for details). The X-ray crystal structure of
the product reveals that it is LMeFe(η6-C6H6) (Figure 5). All
benzene C-C and Fe-C bond distances are within 3σ of the
mean indicating a η6-binding mode. At room temperature the
1H NMR spectrum (C6D6) shows broad signals. The rhombic
Four-Coordinate Iron-N2 Complexes. Treatment of
LRFeNNFeLR with 2 molar equivalents of 4-tert-butylpyridine,
4-(dimethylamino)pyridine, or pyridine affords four-coordinate
dinitrogen complexes of iron (Scheme 2). X-ray crystallographic
analyses of the 4-tert-butylpyridine adducts, LRFe(tBupy)NN-
(tBupy)FeLR, are in the Supporting Information. The identifica-
tion of the other pyridine adducts are based on the similarity of
1
their H NMR spectra to the structurally characterized ones.
Each iron atom is four-coordinate, with a trigonal pyramidal
geometry (τ ) 0.42-0.52)32 that is reminiscent of the four
coordinate pyridine adducts of LtBuFeH,46 LtBuFeF,33 and LR-
Fe(NHR′).60b
The crystal structures of these complexes show that the N-N
bonds are shortened by ∼0.03 Å by the addition of a fourth
(32) We define τ as a normalized measure of trigonal distortion of a tetrahedral
structure, with τ ) 0 representing tetrahedral geometry and τ ) 1
representing a trigonal pyramid in which the metal lies in the plane formed
by the three basal ligands. For details, see ref 26.
(33) Vela, J.; Smith, J. M.; Yu, Y.; Ketterer, N. A.; Flaschenriem, C. J.;
Lachicotte, R. J.; Holland, P. L. J. Am. Chem. Soc. 2005, 127, 7857-
7870.
(34) Estimated standard deviations from automatic programs are generally
underestimated: Stout, G. H.; Jensen, L. H. X-ray Structure Determina-
tion: A Practical Guide; Wiley: New York, 1989; pp 418-419.
9
760 J. AM. CHEM. SOC. VOL. 128, NO. 3, 2006

Iron Dinitrogen Complexes
A R T I C L E S
Figure 5. X-ray crystal structure of LMeFe(C6H6). Thermal ellipsoids shown
at 50% probability, hydrogen atoms omitted for clarity. Bond lengths (Å)
and angles (°): Fe-N(11) 1.966(1), Fe-N(21) 1.981(1), Fe-C 2.143(2),
2.139(2), 2.151(2), 2.157(2), 2.144(2), 2.143(2); N(11)-Fe(1)-N(21)
92.20(5).
Scheme 3
Figure 6. (a) X-ray crystal structure of LMeFe(CO)3. Thermal ellipsoids
shown at 50% probability, hydrogen atoms omitted for clarity. Selected
bond lengths (Å) and angles (°): Fe(1)-N(11) 1.9827(11), Fe(1)-C(13)
1.7957(16), Fe(1)-C(14) 1.871(2); N(11)-Fe(1)-N(11)′ 90.96(6), C(13)-
Fe(1)-N(11) 90.66(6), C(13)-Fe(1)-N(11)′ 163.18(7), C(14)-Fe(1)-
N(11) 98.83(6), C(13)-Fe(1)-C(13)′ 83.08(10), C(13)-Fe(1)-C(14)
97.45(8). (b) X-ray crystal structure of cocrystallized LtBuFe(CO)2 (80%)
and LtBuFe(CO)3 (20%, dashed lines). Thermal ellipsoids shown at 50%
probability, hydrogen atoms omitted for clarity. Selected bond lengths (Å)
and angles (°) for LtBuFe(CO)2: Fe(1)-N(11) 1.984(3), Fe(1)-N(21)
1.947(3), Fe(1)-C(14B) 1.791(7), Fe(1)-C(15B) 1.825(9); N(11)-Fe(1)-
N(21) 94.20(13), C(14B)-Fe(1)-N(11) 168.3(3), C(14B)-Fe(1)-N(21)
92.29(17), C(15B)-Fe(1)-N(11) 91.9(3), C(15B)-Fe(1)-N(21) 172.2(2),
C(14B)-Fe(1)-C(15B) 82.6(3).
X-band EPR spectrum (g ) 2.20, 2.01, 1.98; Figure 7c) and
the solution magnetic moment (µeff ) 2.5 µB) indicate an S )
1/2 ground state, so we formulate this compound as a low-spin
iron(I) complex.
Reaction of LMeFeNNFeLMe with excess CO in diethyl ether
affords a novel iron(I) complex, LMeFe(CO)3, as dark green
crystals (Scheme 2). The X-ray crystal structure of this
compound is shown in Figure 6a. The geometry around the iron
center is square pyramidal, where one of the carbonyl groups
is in the axial position of the square pyramid. The bond length
of the axial Fe-C bond (Fe-C(14) 1.871(2) Å) is slightly longer
than the other two Fe-C bonds (Fe-C(13) 1.7957(16) Å). The
IR spectrum of LMeFe(CO)3 in pentane shows three bands (2042,
1971, 1960 cm-1) in the carbonyl region. The magnetic moment
is 2.0 µB, and an axial X-band EPR signal (Figure 7) is observed
at 77 K in frozen toluene, with g| ) 1.99 and g ) 2.04.
Together, these measurements indicate that LMeFe(CO)3 has a
low-spin (S ) 1/2) electronic configuration at iron.
Reaction of LtBuFeNNFeLtBu with excess CO gives a product
with more complicated characteristics. Its crystal structure
consists of a superposition of two complexes (Figure 6b and
Scheme 2): LtBuFe(CO)3 and square-planar LtBuFe(CO)2, which
differ by the loss of the axial CO ligand. Consistent with a
mixture of tricarbonyl and dicarbonyl species, the IR spectrum
of this solid shows five bands at 2036, 2000, 1967, 1953, and
1922 cm-1. The 77 K X-band EPR spectrum contains an axial
signal closely resembling that of LMeFe(CO)3 as well as a
rhombic signal (g ) 2.35, 2.13, 1.98) tentatively assigned to
LtBuFe(CO)2 (Figure 7). The EPR signature suggests that both
carbonyl complexes have low-spin electronic configurations.
Unfortunately, it has not yet been possible to separate these
two complexes for more detailed characterization of the square-
planar iron(I) complex.
Figure 7. X-band EPR spectra of (a) LMeFe(CO)3; (b) a mixture of LtBu
-
Fe(CO)3 and LtBuFe(CO)2; (c) LMeFe(C6H6). Solvent: toluene; temperature
) 77 K; frequency ) 9.42 GHz; power ) 1.0 mW; attenuation ) 30 dB.
Reaction of LMeFeNNFeLMe with 2 equiv of PPh3 results in
the formation of dark purple LMeFePPh3, as identified by X-ray
crystallography (Figure 8) and 1H NMR spectroscopy. Interest-
ingly, in the crystal structure the phosphine ligand is bent out
of the iron-â-diketiminate plane by 28.42(5)°, bringing the
phosphorus atom 1.197(5) Å out of the plane. The origin of
this bending is probably steric in nature. One of the phenyl rings
of PPh3 is wedged between two isopropyl groups of the
9
J. AM. CHEM. SOC. VOL. 128, NO. 3, 2006 761

A R T I C L E S
Smith et al.
Figure 9. X-ray crystal structure of LtBuFeClK(18-crown-6). Thermal
ellipsoids shown at 50% probability. Hydrogen atoms, minor conformer of
tert-butyl group, and toluene solvate omitted for clarity. Selected bond
lengths (Å) and angles (°): Fe-N11 1.934(6), Fe-Cl 2.235(3), K-O
2.850(9), 2.741(7), 2.952(9), 2.848(10), 2.830(9), 2.774(8), K-Cl
3.074(3); N11-Fe-N21 101.8(3), N11-Fe-Cl 134.3(2), N21-Fe-Cl
123.8(2), Fe-Cl-K 134.17(11).
Figure 8. X-ray crystal structure of LMeFePPh3. Thermal ellipsoids shown
at 50% probability, hydrogen atoms omitted for clarity. The phenyl groups
of PPh3 are shown as solid white. Selected bond lengths (Å) and angles
(°): Fe-P 2.2348(8), Fe-N11 1.953(2), Fe-N21 1.959(2); N11-Fe-N21
98.13(8), N11-Fe-P 127.25(6), N21-Fe-P 125.25(6).
Scheme 4
â-diketiminate ligand, causing the â-diketiminate aryl rings to
bend away from the phenyl ring so that the methine carbons of
the isopropyl groups are 6.798(4) Å on one side but 4.166(4)
Å on the other side. This asymmetry is evident in low-
temperature 1H NMR spectra in pentane-d12, but at higher
temperature there is free rotation and averaged C2V symmetry
(see Supporting Information). The three-coordinate environment
results in an Fe-P bond length (2.2348(8) Å) that is shorter
than the corresponding Fe-P bond length (2.2889(9) Å) in the
four-coordinate iron(I) complex, (PhB(CH2PPh2)3)FePPh3.35
Like its precursor dinitrogen complex, LMeFePPh3 is not stable
in C6D6 solution. Unlike the carbonyl and benzene adducts, the
PPh3 adduct is a high spin iron(I) complex as evidenced by the
magnetic moment (µeff ) 3.6 µB) and the absence of an X-band
EPR signal at 77 K.36
is observed and characteristic 1H NMR resonances for
LtBuFeNNFeLtBu appear. This confirms that transient iron(I)
complexes are kinetically competent in the formation of the iron-
dinitrogen complexes from LtBuFeCl, KC8, and N2.
Addition of 2 molar equivalents of acetophenone to a dark
red pentane solution of LMeFeNNFeLMe or LtBuFeNNFeLtBu at
-35 °C leads to yellow or orange solids characterized as
[LRFeOC(Me)Ph]2 (Scheme 2; crystal structures shown in
Supporting Information). Interestingly, the molecule has the rac
diastereomer of the pinacolate ligand. The loss of symmetry in
the molecule caused by the formation of the chiral centers is
The green intermediate species decomposes to an intractable
orange mixture within a few hours. However, performing the
reduction in the presence of 18-crown-6 affords dark green
crystals as part of a mixture of products. The X-ray crystal
structure of one product, LtBuFe(KCl)(18-crown-6) (Scheme 4),
is shown in Figure 9. The Fe-Cl-K angle is 134.17(11)°, and
the Fe-Cl distance is 2.235(3) Å, compared with 2.172(1) Å
in LtBuFeCl. A solvent toluene molecule has a carbon roughly
3.6 Å from the potassium atom, opposite the bridging chloride.
However, disorder of the toluene limits the precision of the
structure. This is a rare example of a complex with an alkali
halide as a ligand; to our knowledge the only related iron
complex is [(Me3Si)2N]2Fe(µ-Cl)Li(THF)3.37
1
reflected in the H NMR spectrum of [LMeFeOC(Me)Ph]2 (for
example, four sets of isopropyl CH3 groups were observed).
The 1H NMR spectra exclude the presence of the meso
diastereomer, which would show up to eight sets of isopropyl
CH3 groups.
Reduction of LtBuFeCl in the Absence of N2. Treatment of
solutions of LtBuFeCl with KC8 in diethyl ether or THF, under
an atmosphere of argon, results in the formation of forest green
solutions of a metastable compound assigned as LtBuFe(KCl)-
(solv)x (Scheme 4). When argon is removed from LtBuFe(KCl)-
(solv)x under vacuum and nitrogen is introduced to the forest
green solution, an instant color change from green to red-purple
Two-Electron Reduction of LRFeNNFeLR. Reduction of
LRFeNNFeLR or LRFe(tBupy)NNFe(tBupy)LR with additional
KC8 gives the dark blue-green complexes K2LRFeNNFeLR,
isolated in low yield (Scheme 5). They are soluble in pentane,
suggesting that the alkali metal remains tightly bound in solution.
(35) Brown, S. D.; Betley, T. A.; Peters, J. C. J. Am. Chem. Soc. 2003, 125,
322-323.
(36) We have characterized the related high-spin iron(I) compound LtBuFe-
(HCCPh) using Mo¨ssbauer and EPR spectroscopy: Stoian, S. A.; Yu, Y.;
Smith, J. M.; Holland, P. L.; Bominaar, E. L.; Mu¨nck, E. Inorg. Chem.
2005, 44, 4915-4922.
(37) Siemeling, U.; Vorfeld, U.; Neumann, B.; Stammler, H.-G. Inorg. Chem.
2000, 39, 5159-5160.
9
762 J. AM. CHEM. SOC. VOL. 128, NO. 3, 2006

Iron Dinitrogen Complexes
A R T I C L E S
Figure 11. B3LYP/CZDZ**++ optimized geometries for quartet L′FeNN,
triplet L′FeNN-, triplet Na2L′FeNN+, and triplet K2L′FeNN+.
Figure 10. X-ray crystal structure of K2LMeFeNNFeLMe. Thermal ellipsoids
shown at 50% probability, hydrogen atoms omitted for clarity. Selected
bond lengths (Å) and angles (°): Fe(1)-N(1) 1.741(5), Fe(1)-N(11)
1.915(5), Fe(1)-N(21) 1.936(5), Fe(2)-N(2) 1.736(5), Fe(2)-N(12)
1.931(5), Fe(2)-N(22) 1.928(5), N(1)-N(2) 1.215(6), K(1)-N(1)
2.778(6), K(1)-N(2) 2.748(5), K(2)-N(1) 2.800(6), K(2)-N(2) 2.796(5);
N(11)-Fe(1)-N(21) 96.2(2), N(1)-Fe(1)-N(11) 131.3(2), N(1)-Fe(1)-
N(21) 132.6(2), N(12)-Fe(2)-N(22) 95.0(2), N(2)-Fe(2)-N(12)
133.6(2), N(2)-Fe(2)-N(22) 131.0(2).
vibrational spectroscopy, the N-N frequencies are clearly lower
in the [LRFeNNFeLR]2- complexes than in their less reduced
LRFeNNFeLR counterparts (Table 1).
Computations: Role of Alkali Metals. All experimental
attempts to remove the alkali metals from these complexes have
led to decomposition or intractable products. Therefore, we
turned to calculations to distinguish the effects of reduction from
the effects of alkali metal coordination. The theoretical model
complex L′FeNN was first reduced by a single electron and
then pairs of alkali metal ions were ligated to the dinitrogen
ligand. Thus, B3LYP/CSDZ**++ geometry optimizations of
triplet [L′FeNN]-, {Na2[L′FeNN]}+ and {K2[L′FeNN]}+ were
performed under C2V symmetry using RODFT methods. The
optimized geometry of {Na2[L′FeNN]}+ is shown in Figure 11,
and the geometry of the potassium analogue is similar. The
calculated bond lengths for these species are Fe-N ) 1.755 Å
([L′FeNN]-), 1.690 Å ({Na2[L′FeNN]}+), and 1.709 Å ({K2-
[L′FeNN]}+), while the calculated N-N bond lengths are 1.159,
1.191, and 1.176 Å, respectively, for these same complexes.
These calculated values can be compared to Fe-N ) 1.902 Å
and N-N ) 1.123 Å in L′FeNN (see above). Although the
metrical parameters are not accurate due to the truncation of
the model, two points are of particular interest in the context
of N2 activation. First, comparison of the results for L′FeNN
and [L′FeNN]- indicates that reduction of iron significantly
weakens N2 (∆(N-N) ) +0.037 Å). Second, coordination of
alkali metal ions causes further dinitrogen stretching (∆(N-N)
∼ +0.02-0.03 Å). Both effects are comparable in terms of their
impact on the N-N bonding.
Scheme 5
In solution, the K2LRFeNNFeLR complexes each adopt a D2d
or D2h averaged structure, since again only seven signals are
1
observed in H NMR spectra.
The structure of K2LMeFeNNFeLMe, as determined by X-ray
crystallography (Figure 10), is analogous to that of M2LtBu
-
FeNNFeLtBu (M ) Na, K),25 where the potassium atom
coordinates to the bridging N2 and the aryl rings of the ligand.
The K-N bond lengths in K2LMeFeNNFeLMe (2.748(5)-
2.800(6) Å) are slightly longer than those of K2LtBuFeNNFeLtBu
(2.690(6)-2.696(6) Å). The Fe-N bond lengths are similar to
those in LRFeNNFeLR. As in the LtBu-supported alkali metal
complexes, the two iron-diketiminate ligand planes of K2LMe
-
FeNNFeLMe are canted at an angle of 34.4(2)°. Most interest-
ingly, the N-N bond is further lengthened beyond that in
LRFeNNFeLR, to distances of 1.215(6) (K2LMeFeNNFeLMe),
Inspection of the Kohn-Sham orbitals38 suggests that reduc-
tion of LFeNN places additional electron density in an orbital
with Fe-N π-bonding character and N-N π-antibonding
character. Also, the anion shows greater mixing of NN π* with
the Fe dπ orbitals than the corresponding (virtual) orbital in
L′FeNN. We ascribe the greater orbital mixing to a better energy
match between the N-N π* orbital and the Fe-based dπ orbitals.
Coordination of the alkali metal ions to N2 enhances these orbital
interactions and yields further N2 weakening.
Reactivity of K2LRFeNNFeLR. The complex of the larger
ligand, K2LtBuFeNNFeLtBu, is exceedingly sensitive, and upon
standing it converts to LtBuFeNNFeLtBu. We believe the major
source of decomposition to be trace water on the surface of
reaction vessels, as using silylated glassware somewhat reduces
the amount of decomposition. The fragile nature of this complex
and Na2LtBuFeNNFeLtBu has precluded further studies of their
1.238(4) (Na2LtBuFeNNFeLtBu), and 1.241(7) (K2LtBuFeNNFeLtBu
)
Å (Table 1).
Resonance Raman spectra of the alkali metal complexes
contain multiple bands with frequencies that are sensitive to
isotopic substitution in the N2 (Figure 4C,D; Figure 4E shows
the difference spectrum). Using K2LtBuFeNNFeLtBu these bands
appear at 1589 (1536/1565) and 1123 (1087) cm-1, and using
K2LMeFeNNFeLMe they are at 1625 (1569) and 1427 (1389)
cm-1 (numbers in parentheses for the 15N2 isotopomer). The
observation of multiple isotope-sensitive bands (with isotope
shifts smaller than expected for a diatomic oscillator) indicates
that there are multiple Raman active vibrational modes in which
the bridging nitrogen atoms have significant kinetic energy. We
attribute this effect to coupling between CdN/CdC bond
stretches near 1500 cm-1 and the N-N bond stretch. Although
the absence of a localized N-N stretching vibration hinders
the precise quantification of N-N bond weakening through
(38) Koch, W.; Holthausen, M. C. A Chemist’s Guide to Density Functional
Theory, Wiley-VCH: Weinheim, 2000.
9
J. AM. CHEM. SOC. VOL. 128, NO. 3, 2006 763

A R T I C L E S
Smith et al.
reactivities. K2LMeFeNNFeLMe is slightly more robust. However,
reactions with electrophiles give mixtures in which no product
deriving from transformation of the N2 group is evident.
Reaction with CH3OTf is cleaner, with quantitative formation
28a
1
of LMeFeCH3 by H NMR spectroscopy.
Discussion
Given the presence of iron in large-scale natural and industrial
N2 fixation processes, it is surprising that until very recently
all isolated complexes between iron and N2 had very little
ground-state weakening of the N-N bond, with N-N distances
less than 1.14 Å and N-N stretching frequencies greater than
1950 cm-1 39
As a result, synthetic iron-dinitrogen complexes
.
were categorized as “unactivated.”40 However, the coordination
number of iron in these complexes is five or greater. This
contrasts with the iron atoms of the FeMoco, which are four-
coordinate in the isolated MN form, and may reach an even lower
coordination number in the reduced, activated form. On the
surface of metallic iron, the (111) face of iron is thought to be
most active, and it is perhaps significant that this surface has
the greatest number of highly exposed iron atoms.18 Our ability
to make three- and four-coordinate complexes offers the
opportunity to evaluate the effect of low coordination number
of iron on coordinated N2.
Figure 12. Key frontier orbitals from the MCSCF calculations on
L′FeNNFeL′. Orbital (a) has 48% Fe and 48% N character, (b) has 96% Fe
and 2% N character, and (c) has 60% Fe and 35% N character. NOON )
natural orbital occupation number.
The effect of coordination number was evaluated through the
use of 4-tert-butylpyridine, which coordinates to each iron atom
to give trigonal pyramidal coordination. The frequencies of the
N-N stretching vibrations indicate similar extents of bond
weakening between three- and four-coordinate complexes, and
these are deemed more trustworthy than the minimal changes
seen in N-N distances (Table 1). Peters has recently reported
four-coordinate iron-N2 complexes using a tris(phosphino)borate
ligand, which give N-N bonds with comparable lengths (e.g.
1.171(4) Å) and frequencies (e.g., 1884 cm-1).24 Therefore, both
three- and four-coordinate iron complexes are not substantially
different from each other, but both give greater N-N weakening
than five- or six-coordinate complexes.
Stretching of the N-N Bond in Iron Dinitrogen Com-
plexes. The extent of N2 reduction is much greater in
LRFeNNFeLR than in five-coordinate complexes with an FeN-
NFe core (Table 1). The bond lengths (1.18 ( 0.01 Å) and
stretching frequencies (1790 ( 20 cm-1) for the NN bonds are
characteristic of a bond order between two (MeNdNMe 1.25
Å; 1575 cm-1 41 and three (NtN 1.098 Å; 2331 cm-1).1 In
)
other crystallographically characterized iron dinitrogen com-
plexes, terminal N-N bond lengths range from 1.07(1) Å -
1.14(1) Å,39 while bridging N-N have been observed at
1.13(2) Å.42 Likewise, the N-N stretching frequencies of most
other terminal iron dinitrogen complexes are also consistent with
minimal activation of the N2 ligand (νNN ) 1955-2112 cm-1).
The exceptions are the heterobimetallic bridging complexes
Fe(PhBPiPr3)(N2)MgCl(ether)x (ether ) THF, x ) 2, ν(NN) )
Computations and Electronic Structure. To tie these
observations to specific orbital interactions, and to deconvolute
the different contributions to N-N bond weakening, MCSCF
calculations were performed on L′FeNNFeL′, where L′ again
represents a truncated â-diketiminate ligand. Multiconfiguration
SCF computations of the type used here can offer a more
accurate picture of electronic structure than DFT, but are often
difficult to interpret as the natural orbitals derived from an
MCSCF calculation are not limited to integral electron occupa-
tion numbers.44 In this research, the natural orbitals (i.e., those
orbitals obtained through diagonalization of the first-order
density matrix) are analyzed, and from these are obtained the
natural orbital occupation numbers (NOONs). The NOONs
range from 0 to 2 e- and give a measure of the multireference
character. Schmidt and Gordon describe NOONs between 0.1
and 1.9 as indicative of “considerable multireference character.44
The key frontier orbitals (Figure 12) of L′FeNNFeL′ have
π-interactions between iron and nitrogen atoms, and are doubly
degenerate because the diketiminate planes are perpendicular
to one another. Figure 12c shows substantial population (NOON
) 1.58 e- per member of the E set) of a pair of natural MO’s
that have bonding interactions between Fe and the N of the
bridging N2, and antibonding interactions between the two
nitrogen atoms. This pair is strongly correlated with a pair of
lower-occupation natural orbitals (NOON ) 0.42 e- per member
for this E set) of higher energy (Figure 12a). Thus, the
calculations indicate that in these â-diketiminate complexes,
1830 cm-1; ether ) 18-crown-6, x ) 0.5, νNN ) 1884 cm-1 24
and Fe[(N3N)Mo-NdN]3 (d(NN) ) 1.20(3) Å - 1.27(2) Å,
νNN ) 1703 cm-1 43
complexes, where the iron atoms are also
low coordinate.
)
)
The N-N distances and N-N stretching frequencies are
similar between complexes of LMe and LtBu, indicating that the
size of the â-diketiminate ligand plays little role in weakening
the N-N bond. Additionally, the long N-N bond is reproduced
in density-functional and multireference calculations on the
truncated L′FeNNFeL′, where all substituents are replaced by
hydrogen. So, although the large ligands are essential for
enforcing the low coordination number, they do not play a direct
role in lengthening the N-N bond.
(39) A full list of these complexes may be found in the Supporting Information.
(40) Tuczek, F.; Lehnert, N. Angew. Chem., Int. Ed. 1998, 37, 2636-2638.
(41) Pearce, R. A. R.; Levin, I. W.; Harris, W. C. J. Chem. Phys. 1973, 59,
1209-1216.
(42) (a) Berke, H.; Bankhardt, W.; Huttner, G.; Von Seyerl, J.; Zsolnai, L. Chem.
Ber. 1981, 114, 2754-2768. (b) Kandler, H.; Gauss, C.; Bidell, W.;
Rosenberger, S.; Buergi, T.; Eremenko, I. L.; Veghini, D.; Orama, O.;
Burger, P.; Berke, H. Chem. Eur. J. 1995, 1, 541-548.
(43) (a) O’Donoghue, M. B.; Zanetti, N. C.; Davis, W. M.; Schrock, R. R. J.
Am. Chem. Soc. 1997, 119, 2753-2754. (b) O’Donoghue, M. B.; Davis,
W. M.; Schrock, R. R.; Reiff, W. M. Inorg. Chem. 1999, 38, 243-252.
(44) Schmidt, M. W.; Gordon, M. S. Annu. ReV. Phys. Chem. 1998, 49, 233-
266.
9
764 J. AM. CHEM. SOC. VOL. 128, NO. 3, 2006

Iron Dinitrogen Complexes
A R T I C L E S
π-back-bonding from iron into the N-N π* orbital is a major
contributor to the weakened N-N bond.
of TM back-bonding into the coordinated N2.50 It is perhaps
relevant that small amounts of potassium are beneficial in the
iron(0) catalyst in the Haber-Bosch process,51 and that LEED
experiments suggest that N2 bound to Fe(111) is not perpen-
dicular to the surface, but is strongly inclined,52 suggesting a
possible side-on binding mode.
The N-N bond lengths (and less dramatically, the Fe-N
bond lengths) are dependent on the nature of the â-diketiminate
ligand in the alkali metal containing complexes. In the less
congested K2LMeFeNNFeLMe complex, the N-N bond length
is only ∼0.03 Å longer than in LMeFeNNFeLMe, while in
K2LtBuFeNNFeLtBu, the N-N bond length increases by about
∼0.05 Å from LtBuFeNNFeLtBu. The Fe-N bond lengths
in K2LMeFeNNFeLMe are also slightly shorter than in
K2LtBuFeNNFeLtBu. The most likely reason is that the smaller
binding pocket in K2LtBuFeNNFeLtBu reduces the space for
binding the alkali metals, and the only way to accommodate
them is for the iron atoms to move away from each other. The
greater apparent strain in K2LtBuFeNNFeLtBu correlates with its
lower stability (see below).
Reactivity of Reduced Iron â-Diketiminate Dinitrogen
Complexes. In general, LRFeNNFeLR complexes react as low-
coordinate Fe(I) sources by loss of N2. Thus far, we have found
LtBuFeNNFeLtBu to be less reactive than LMeFeNNFeLMe, most
likely due to greater protection of the N2 ligand. Thus, dissolving
LMeFeNNFeLMe in benzene leads to quantitative formation of
LMeFe(η6-C6H6), but LtBuFeNNFeLtBu is stable in aromatic
solvents at temperatures of up to 100 °C. While reaction of
LMeFeNNFeLMe with PPh3 to quantitatively form LMeFePPh3
is rapid at room temperature, LtBuFeNNFeLtBu reacts with PPh3
only at elevated temperatures, giving an intractable mixture of
products.
These displacement reactions likely take place through an
associative mechanism, a conclusion that is supported by (a)
the dependence of the rate constant for reaction with benzene-
d6 on [C6D6], (b) the isolation of pyridine adducts of
LMeFeNNFeLMe, and (c) the aforementioned steric dependence
of the substitution reactions. It is notable that the high-spin N2
complexes react with CO very rapidly to form the low-spin
carbonyl complexes, indicating that the necessary spin flip
accompanying ligand binding is very rapid.
Evaluation of Potential Pathways to the N2 Complexes.
Reduction of an iron chloride complex to the dinitrogen
complexes requires (a) reduction of iron and (b) replacement
of Cl- with N2, but the order of these steps was unknown. We
find that reduction of LtBuFeCl under Ar gives LtBuFeCl(K)-
(solvent)x, which in turn reacts with N2 to give LtBuFeNNFeLtBu
.
The latter reaction occurs more rapidly than reaction of
LtBuFeCl with KC8 under N2. Therefore, the iron(I)-KCl complex
is kinetically competent to be an intermediate during the
reduction of LtBuFeCl under N2.45
We have also observed that LtBuFeNNFeLtBu is formed by
photolysis of the hydride complex [LtBuFeH]2 in benzene
46
solution with a high-pressure mercury lamp, and during at-
tempted resonance Raman experiments on the same hydride
complex. As pointed out by Fryzuk, the accomplishment of
homogeneous catalytic N2 reduction is more feasible when
strong reducing agents are not required to create the reduced
metal center.47 Therefore, there is much current interest in
polyhydride complexes that lead to dinitrogen complexes. Tyler
has recently reported formation of an iron-N2 complex from
dihydrogen reduction under N2.23b
Two-Electron Reduction of the FeNNFe Core. Three
complexes of the type (alkali)2LRNNFeLR have been character-
ized. Two alkali metal cations are trapped within the complex,
bound to the N2 ligand and to the arene rings of the diketimi-
nates. There is an increase in the N-N distance (0.03-0.05
Å). Resonance Raman spectra of these complexes give bands
near 1600 cm-1 that are sensitive to 15N2 substitution in the
bridging ligand. The N-N bond lengthening and stretching
frequency reduction lead to a self-consistent picture in which
the N-N bond is weakened by two-electron reduction.
According to the MCSCF description of L′FeNNFeL′ pre-
sented above, two-electron reduction of L′FeNNFeL′ unit places
additional electron density (ca. 0.25 e- according to the NOONs)
in the MO that is bonding between Fe and the N of bridging
N2, and antibonding interactions between the two nitrogen atoms
(Figure 12a). Therefore, reduction leads to greater population
of the N2 π* orbital. This model neatly accommodates the
observation that reduction leads to greater N-N weakening and
a reduction in the magnetic moment.
Attempts to Functionalize Coordinated N2. The classical
method of producing NH3 from an N2 complex is by adding
DFT calculations on simplified L′FeNN models also show
that alkali metal coordination causes a similar amount of
stretching as reduction (∼0.03 Å). The stretching caused by
side-on coordination of an alkali metal cation may seem unusual,
because in the literature examples where N2 is bound side-on
to alkali metals, there is minimal disruption of N-N bond-
ing.48,49 However, in combination with end-on binding to a
transition metal (TM), alkali metal coordination increases
polarization of the TM-N2 unit, enhancing the effectiveness
(49) However, lanthanides are capable of N-N bond weakening. Representative
examples: (a) Evans, W. J.; Ulibarri, T. A.; Ziller, J. W. J. Am. Chem.
Soc. 1988, 110, 6877-6879. (b) Jubb, J.; Gambarotta, S. J. Am. Chem.
Soc. 1994, 116, 4477-4478. (c) Roussel, P.; Scott, P. J. Am. Chem. Soc.
1998, 120, 1070-1071. (d) Ganesan, M.; Gambarotta, S.; Yap, G. P. A.
Angew. Chem., Int. Ed. Engl. 2001, 40, 766-769. (e) Cloke, F. G. N.;
Hitchcock, P. B. J. Am. Chem. Soc. 2002, 124, 9352. (f) Evans, W. J.;
Lee, D. S.; Lie, C.; Ziller, J. W. Angew. Chem., Int. Ed. 2004, 43, 5517-
5519.
(50) There are a few other synthetic complexes known with end-on binding to
a late transition metal and side-on binding to an alkali metal: (a) Jonas, K.
Angew. Chem. 1973, 85, 1050. (b) Jonas, K.; Brauer, D. J.; Kru¨ger, C.;
Roberts, P. J.; Tsay, Y. H. J. Am. Chem. Soc. 1976, 98, 74. (c) Klein, H.
F.; Hammer, R.; Wenninger, J.; Friedrich, P.; Huttner, G. Z. Naturforsch.,
B. 1978, 33, 1267. (d) Jubb, J.; Gambarotta, S. J. Am. Chem. Soc. 1994,
116, 4477-4478. (e) Caselli, A.; Solari, E.; Scopelliti, R.; Floriani, C.;
Re, N.; Rizzoli, C.; Chiesi-Villa, A. J. Am. Chem. Soc. 2000, 122, 3652-
3670. (f) Bonomo, L.; Stern, C.; Solari, E.; Scopelliti, R.; Floriani, C.
Angew. Chem., Int. Ed. Engl. 2001, 40, 1449-1452.
(45) This contrasts with the nitrogen-free reduction of [P2N2]ZrCl2 and [P2N2]-
NbCl (P2N2 ) PhP(CH2SiMe2NSiMe2CH2)2PPh), which leads to complexes
that do not react with dinitrogen. See: Fryzuk, M. D.; Kozak, C. M.;
Mehrkhodavandi, P.; Morello, L.; Patrick, B. O.; Rettig, S. J. J. Am. Chem.
Soc. 2002, 124, 516-517.
(46) Smith, J. M.; Lachicotte, R. J.; Holland, P. L. J. Am. Chem. Soc. 2003,
125, 15752-15753.
(47) Fryzuk, M. D.; Love, J. B.; Rettig, S. J.; Young, V. G. Science 1997, 275,
1445-1447.
(51) Whitman, L. J.; Bartosch, C. E.; Ho, W.; Strasser, G.; Grunze, M. Phys.
ReV. Lett. 1986, 56, 1984-1987 and references therein.
(52) (a) Grunze, M.; Golze, M.; Hirschwald, W.; Freund, H. J.; Pulm, H.; Seip,
U.; Tsai, M. C.; Ertl, G.; Kueppers, J. Phys. ReV. Lett. 1984, 53, 850-
853. (b) Freund, H. J.; Bartos, B.; Messmer, R. P.; Grunze, M.; Kuhlenbeck,
H.; Neumann, M. Surf. Sci. 1987, 185, 187-202.
(48) (a) Ho, J.; Drake, R. J.; Stephan, D. W. J. Am. Chem. Soc. 1993, 115,
3792-3793. (b) Bazhenova, T. A.; Kulikov, A. V.; Shestakov, A. F.; Shilov,
A. E.; Antipin, M. Y.; Lyssenko, K. A.; Struchkov, Y. T.; Makhaev, V. D.
J. Am. Chem. Soc. 1995, 117, 12176-12180.
9
J. AM. CHEM. SOC. VOL. 128, NO. 3, 2006 765

A R T I C L E S
Smith et al.
excess acid to release ammonia and/or hydrazine.53 Addition
of HCl, thiols, alcohols, or H(OEt2)2+ BArF-54 to Et2O solutions
of LtBuFeNNFeLtBu does not give detectable amounts (<5%)
of ammonia or hydrazine. However, acid addition protocols are
harsh, and it has been observed that the products of protonation
may also depend on the acid used,53 casting some doubt on the
generality of this method as a means of characterization.
(diketiminate)Fe is an exceptionally good π-donor because of
the low-coordinate environment, in contrast to the usual
assumption that low-coordinate complexes have predominantly
electrophilic character.
The dinitrogen complexes are also competent synthetic
precursors for low-coordinate iron(I) complexes through loss
of N2. Attempted protonation and alkylation of N2 have led to
attack at the metal or at the diketiminate ligand. Strategic ligand
modification may lead to bridging N2 complexes with higher
reactivity at coordinated N2.
Bridging N2 complexes, particularly those of oxophilic early
metal complexes, can react with aldehydes and ketones to give
azines.55 However, treatment of LRFeNNFeLR with acetophe-
none leads to the pinacol coupled complexes [LRFeOC(Me)-
Ph]2. Transition metal mediated pinacol coupling is typically
the province of early transition metal complexes,56 making this
an unusual transformation for an iron complex. The observation
of pinacol coupling suggests that the formally iron(I) metal
center acts as a one-electron reducing agent, forming a carbon-
based radical that undergoes coupling to give the pinacolate
ligand of the product. In another example of LMeFeNNFeLMe
acting as a reducing agent, it reacts with elemental sulfur to
Experimental Section
General Procedures. All manipulations were performed under a
nitrogen atmosphere by standard Schlenk techniques or in an M. Braun
glovebox maintained at or below 1 ppm of O2 and H2O. Glassware
was dried at 150 °C overnight. 1H NMR data were recorded on a Bruker
1
Avance 400 spectrometer (400 MHz) at 22 °C. All peaks in the H
NMR spectra are referenced to residual protiated solvent. All peaks
are singlets. In parentheses are listed: T2 values in ms calculated as
(π∆ν1/2) ,
-1 57 integrations and assignments. In some cases, overlapping
peaks prevented T2 determinations. For certain complexes, not all signals
could be assigned. IR spectra were recorded on a Mattson Instruments
6020 Galaxy Series FTIR using solution cells with CsF windows. UV-
vis spectra were measured on a Cary 50 spectrophotometer, using screw-
cap cuvettes, and are listed as: λ in nm (ꢀ in mM-1cm-1). Solution
magnetic susceptibilities were determined by the Evans method.58
Elemental analyses were determined by Desert Analytics, Tucson, AZ.
give the diiron(II) complex LMeFeSFeLMe 26
.
Peters and co-workers have recently shown that iron-
coordinated dinitrogen Fe(PhBPiPr3)(N2)MgCl(THF)2 reacts with
+
CH3 sources to give an iron complex of the methyldiazenido
ligand.48 Theirs is the first well-characterized iron example of
the electrophilic attack on coordinated nitrogen that has been
long observed for dinitrogen complexes of early transition
metals.1a Seeking to accomplish an analogous transformation,
we treated the most reduced iron diketiminate dinitrogen
complex, K2LMeFeNNFeLMe, with CH3OTf. This reaction
quantitatively results in the formation of LMeFeCH3. Therefore,
the doubly reduced diketiminate-iron-dinitrogen complexes
act as sources of Fe(0), and N2 is again a leaving group. Efforts
to create ligand geometries that are more conducive to N2
functionalization are underway.
Pentane, diethyl ether (Et2O), and tetrahydrofuran (THF) were
purified by passage through activated alumina and “deoxygenizer”
columns from Glass Contour Co. (Laguna Beach, CA). Benzene,
benzene-d6, cyclohexane-d12, and THF-d8 were first dried over CaH2,
then over Na/benzophenone, and then vacuum transferred into a storage
container. Before use, an aliquot of each solvent was tested with a drop
of sodium benzophenone ketyl in THF solution. Diatomaceous earth
was dried overnight at 200 °C under vacuum. Triphenylphosphine was
recrystallized from pentane at -35 °C, pyridines were dried over
molecular sieves, and acetophenone was crystallized from pentane and
stored in the dark at -35 °C. The iron(II) â-diketiminate complexes
LtBuFeCl59 and [LMeFe(µ-Cl)]260b were prepared by literature procedures.
KC8 was prepared by heating potassium and graphite at 150 °C under
an argon atmosphere. 15N2 (98% isotopic purity) was obtained from
Cambridge Isotopes, and 15N-labeled compounds were handled under
an atmosphere of argon. Photolysis experiments used a 200 W Hg/Xe
lamp.
Conclusions
Complexes in which N2 bridges multiple iron atoms are
important models of potential intermediates in nitrogenase. The
compounds reported here are promising initial models of
FeNNFe intermediates that could result from binding N2 to low-
coordinate iron sites generated upon reduction of FeMoco. Using
a combination of synthetic, structural, spectroscopic, and
computational techniques, we have demonstrated that lowering
coordination number at iron to three or four leads to exceptional
weakening of a coordinated N2 ligand. In three-coordinate iron
complexes, the fundamental orbital interaction involved is back-
bonding into the π* orbital of N2. Interestingly, the effect on
back-bonding of lowering the coordination number is greater
than that from lowering the oxidation state, demonstrating that
the coordination geometry of the metal atom plays a dominant
role in controlling the delocalization of electrons between the
metal and ligands. In particular, the formally iron(I) fragment
LMeFeNNFeLMe (1). A yellow slurry of LMeFe(µ-Cl)2Li(THF)2
60a
(2.28 g, 3.27 mmol) in toluene (40 mL) was stirred at 80 °C overnight.
The toluene was removed under reduced pressure, and pentane (60 mL)
and KC8 (530 mg, 3.92 mmol) were added to the resultant orange solid.
The reaction mixture was stirred overnight at room temperature, and it
slowly turned red with the formation of black graphite. After settling
for 2 h, the supernatant was filtered through diatomaceous earth to
give a dark red solution. The solution was concentrated to ca. 10 mL
and stored at -35 °C. A very dark red solid was obtained in three
crops (1.10 g, 69%). LMeFe15N15NFeLMe was synthesized in a similar
1
fashion under an atmosphere of 15N2 gas and handled under Ar. H
NMR (C6D12) δ 53 (12H, 0.6, CH3), -19 (24H, 2, CH(CH3)2), -20
(8H, 2, m-H), -98 (4H, 2, p-H), -112 (24H, 0.8, CH(CH3)2), -123
(8H, 2, CH(CH3)2), -280 (2H, C-H). µeff (Evans, C6D12) 7.9(3) µeff
Anal. Calcd for C58H82N6Fe2‚C5H12: C 72.26, H 9.04, N 8.03. Found
(53) Leigh, G. J. Acc. Chem. Res. 1992, 25, 177-181.
(54) Brookhart, M.; Grant, B.; Volpe, A. F., Jr. Organometallics 1992, 11, 3920-
3922.
(55) (a) Turner, H. W.; Fellmann, J. D.; Rocklage, S. M.; Schrock, R. R.;
Churchill, M. R.; Wasserman, H. J. J. Am. Chem. Soc. 1980, 102, 7809-
7811. (b) Rocklage, S. M.; Turner, H. W.; Fellmann, J. D.; Schrock, R. R.
Organometallics 1982, 1, 703-707. (c) Schrock, R. R.; Wesolek, M.; Liu,
A. H.; Wallace, K. C.; Dewan, J. C. Inorg. Chem. 1988, 27, 2050-2054.
(56) Wirth, T. Angew. Chem., Int. Ed. 1996, 35, 61-63.
(57) Ming, L.-J. In Physical Methods in Bioinorganic Chemistry; Que, L., Jr.,
Ed.; University Science Books: Sausalito, 2000; pp 375-464.
(58) Baker, M. V.; Field, L. D.; Hambley, T. W. Inorg. Chem. 1988, 27, 7.
(59) Smith, J. M.; Lachicotte, R. J.; Holland, P. L. Organometallics 2002, 21,
4808.
9
766 J. AM. CHEM. SOC. VOL. 128, NO. 3, 2006

Iron Dinitrogen Complexes
A R T I C L E S
C, 72.17 H, 8.71, N 8.07. UV-vis (pentane): 911 (9.1), 499 (20). The
crystals used for the X-ray crystal structures were grown from pentane
at -35 °C (1a) or ambient temperature (1b).
p-H), 15 (4H, CH(CH3)2), 14 (12H, N(CH3)2C5H4N), (4H, m-N(CH3)2-
C5H4N), 6 (24H, 3, CH(CH3)2), -4 (36H, 3, C(CH3)3), -10 (4H,
CH(CH3)2), -24 (24H, 0.7, CH(CH3)2), -69 (4H, 2, o-N(CH3)2C5H4N).
UV-vis (pentane): 717 (4.1), 613 (5.4), 430 (14), 385 (23).
LMeFe(Py)NN(Py)FeLMe (7). Similarly to the preparation of 3,
LMeFeNNFeLMe (86 mg, 88 µmol) was reacted with pyridine (15 µL,
185 µmol) in Et2O (6 mL) to afford 7 as dark blue crystals from pentane
at -35 °C (76 mg, 76%). 1H NMR (C6D12) δ 135 (2H, 0.6, p-C5H4N),
45 (4H, 0.5, o-C5H4N), 25 (4H, m-C5H4N), (4H, m-H), 22 (4H, 0.8,
m-H), 7 (24H, 0.6, CH(CH3)2), 0.9 (24H, CH(CH3)2), -1 (4H, 0.6,
CH(CH3)2), -8 (4H, 0.5, CH(CH3)2), -39 (4H, 0.9, p-H), -118 (2H,
0.2, backbone C-H), -166 (12H, 0.3, CH3). UV-vis (pentane): 913
(7.2), 720 (6.6), 583 (8.3).
LtBuFeNNFeLtBu (2). KC8 (532 mg, 3.94 mmol) was added in
portions to a red slurry of LtBuFeCl59 (2.0 g, 3.37 mmol) in Et2O (40
mL). There was immediate formation of black graphite and a very dark
red solution. Occasionally, the solution was observed to become deep
green, and changed to dark red within a few minutes. The mixture was
stirred overnight at room temperature, and the solvent removed in vacuo.
The residue was extracted with pentane and filtered through diatoma-
ceous earth to give a red-purple solution. The solution was concentrated
(20 mL) and placed in the freezer to give a very dark red-purple solid
in two crops (1.29 g, 70%). LtBuFe15N15NFeLtBu was synthesized
1
similarly under an atmosphere of 15N2 gas and handled under Ar. H
LMeFe(η6-C6H6) (8). A 20 mL scintillation vial was charged with
LMeFeNNFeLMe (100 mg, 103 µmol) and pentane (6 mL) to give a
dark red solution. Benzene (0.5 mL) was added, and the solution was
stirred at room temperature overnight. The solution was pumped down
(2 mL), and the residue was dissolved in hexamethyldisiloxane (2 mL)
and cooled to -35 °C to give dark red crystals (86 mg, 76%). 1H NMR
(C5D12, -60 °C) δ 118 (2H, 0.2, p-H), 11 (12H, 1, CH(CH3)2), -28
(12H, 0.1, CH(CH3)2), -42 (6H, 0.4, CH3), -57 (4H, 0.6, m-H), -158
(1H, 0.5, backbone C-H), -205 (4H, 0.1, CH(CH3)2). µeff (Evans, C6D12)
2.5 µB. Anal. Calcd for C35H47N2Fe: C 76.21, H 8.59, N 5.08. Found
C 76.29, H 8.01, N 5.28. UV-vis (pentane): 917 (0.4), 496 (1.3).
LMeFe(PPh3) (9). A 20 mL scintillation vial was charged with
LMeFeNNFeLMe (100 mg, 103 µmol) and pentane (6 mL) to give a
dark red solution. A solution of PPh3 (54 mg, 206 µmol) in pentane (2
mL) was added to give a dark purple solution. The solution was stirred
at room temperature for 30 min, and the solvent was pumped down (2
mL). The residue was redissolved in hexamethyldisiloxane (2 mL) and
cooled to -35 °C to give two crops of dark purple crystals (110 mg,
72%). LMeFe(PPh3) is soluble in pentane and Et2O, reacts with aromatic
solvents, and is thermally unstable in solution. 1H NMR (C6D12) δ 56
(1H, 0.7, backbone C-H), 37 (4H, 2, m-H), 8 (12H, 2, CH(CH3)2), 7
(6H, 3, m-Ph), 5 (3H, 3, p-Ph), 0.2 (6H, o-Ph) -14 (12H, 0.9, CH-
(CH3)2), -24 (2H, 2, p-H), -29 (2H, 0.3, CH(CH3)2), -100 (2H, 0.8,
CH(CH3)2), -122 (6H, 1, CH3). µeff (Evans, C6D12) 3.6(3) µeff. Anal.
Calcd for C47H56N2FeP: C 76.72, H 7.67, N 3.80. Found C 75.40, H
7.59, N 3.98. UV-vis (pentane): 583 (2.2).
NMR (C6D6) δ 165 (2H, 0.4, backbone C-H), 62 (36H, 0.8, C(CH3)3),
-2 (8H, 3, m-H), -22 (24H, 2, CH(CH3)2), -81 (8H, 0.1, CH(CH3)2),
-97 (24H, 0.4, CH(CH3)2), -120 (4H, 2, p-H). µeff (Evans, C6D6)
8.4(3) µeff. Anal. Calcd. for C70H106N6Fe2‚C5H12: C 74.11, H 9.78, N
6.91. Found C 73.58, H 9.95, N 6.79. UV-vis (pentane): 941 (4.2),
520 (12), 384 (24).
LMeFe(tBupy)NN(tBupy)FeLMe (3). A 20 mL scintillation vial was
charged with LMeFeNNFeLMe (100 mg, 103 µmol) and pentane (6 mL)
to give a red-purple solution. 4-tert-Butylpyridine (30 µL, 206 µmol)
was added via syringe, immediately giving a dark blue solution. After
stirring at room temperature for 30 min, the solution was concentrated
to 2 mL and cooled to -35 °C using vapor diffusion with hexameth-
yldisiloxane. This gave dark blue crystals (100 mg, 78%). The complex
is soluble in pentane and Et2O. The 15N labeled analogue of 3 was
synthesized using LMeFe15N15NFeLMe and 4-tert-butylpyridine. 1H NMR
(C6D12) δ 45 (4H, 0.7, o-C5H4NC(CH3)3), 25 (4H, 1, m-H), 23 (4H, 1,
m-H), 22 (4H, m- C5H4NC(CH3)3), 7 (24H, 0.8, CH(CH3)2), 1 (24H, 5,
CH(CH3)2), -1 (4H, 0.7, CH(CH3)2), -5 (18H, 2, C5H4NC(CH3)3),
-8 (4H, 0.6, CH(CH3)2), -39 (4H, 1, p-H), -112 (2H, 0.2, backbone
C-H), -162 (12H, 0.3, CH3). µeff (Evans, C6D12) 5.9(3) µeff. Anal. Calcd.
for C76H108N8Fe2: C 73.29, H 8.74, N 8.99. Found C 73.10, H 8.51, N
8.74. UV-vis (pentane): 912 (11), 688 (12), 586 (16).
LtBuFe(tBupy)NN(tBupy)FeLtBu (4). A 20 mL scintillation vial was
charged with LtBuFeNNFeLtBu (150 mg, 131 µmol) and Et2O (8 mL)
to give a red-purple solution. 4-tert-Butylpyridine (40 µL, 270 µmol)
was added via syringe to form a dark green solution. After stirring at
room temperature for 30 min, the solvent was removed in vacuo. The
residue was redissolved in warm pentane (4 mL) and cooled to -35
°C to give dark green crystals grown in two crops (169 mg, 91%). 1H
NMR (C6D6) δ 51 (4H, 0.6, p-H), 3.5 (18H, 4, C5H4NC(CH3)3), 3.7
(24H, CH(CH3)2), (4H, o-C5H4NC(CH3)3), 1 (12H, 4, CH(CH3)2), 0.9
(12H, 5, CH(CH3)2), -2 (18H, 2, CH(CH3)3), -5 (4H, 0.8, CH(CH3)2),
-49 (8H, m-H), (4H, m-C5H4NC(CH3)3), -68 (4H, 0.4, CH(CH3)2),
-268 (2H, 0.3, backbone C-H). µeff (Evans, C6D6) 8.5(3) µeff. Anal.
Calcd for C88H132N8Fe2: C 74.76, H 9.41, N 7.92. Found C 74.57, H
9.48, N 6.78. UV-vis (pentane): 953 (7.5), 630 (2.2), 560 (2.5) 486
(4.0), 424 (7.5).
LMeFe(CO)3 (10).
A resealable flask was charged with
LMeFeNNFeLMe (100 mg, 103 µmol) and Et2O (5 mL) to give a dark
red solution. The flask was connected to a vacuum line, and the solution
frozen at -196 °C. The headspace was evacuated, and the solution
thawed. The flask was backfilled with CO to approximately 1 atm,
leading to the immediate formation of a dark green solution. The
solution was stirred for 30 min at room temperature and the volatile
materials were removed in vacuo. The residue was extracted with
pentane and filtered through diatomaceous earth. The solution was
concentrated to 4 mL and cooled to -35 °C to give dark green crystals
(67 mg, 58%). µeff (Evans, C6D6) 2.0 µB. IR (C5H12) νCO 2042, 1971,
1960. Anal. Calcd. for C32H41N2O3Fe: C 68.94, H 7.41, N 5.02. Found
C 69.89, H 7.14, N 5.08. UV-vis (pentane): 642 (0.94), 422 (1.7),
502 (1.1).
LMeFe(Me2Npy)NN(Me2Npy)FeLMe (5). Similarly to the preparation
of 3, LMeFeNNFeLMe (100 mg, 103 µmol) was treated with 4-(di-
methylamino)pyridine (25.2 mg, 206 µmol) in Et2O (6 mL) to afford
LtBuFe(CO)n (11). Similarly to the preparation of 10, LtBuFeNNFeLtBu
(100 mg, 87 µmol) in Et2O (5 mL) was treated with CO. Brown-green
crystals of the mixture 11 were obtained from pentane at -35 °C. IR
1
5 as dark blue crystals from pentane at -35 °C (86 mg, 69%). H
NMR (C6D12) δ 45 (4H, 0.5, o-N(CH3)2C5H4N), 27 (4H, 0.5,
m-N(CH3)2C5H4N), 23 (4H, 0.9, m-H), 21 (4H, 0.9, m-H), 7 (24H, 0.6,
CH(CH3)2), 0.9 (24H, CH(CH3)2), -2 (4H, 0.6, CH(CH3)2), -10 (4H,
0.4, CH(CH3)2), -15 (12H, 1, N(CH3)2C5H4N), -39 (4H, 0.9, p-H),
-115 (2H, 0.2, backbone C-H), -177 (12H, 0.3, CH3). UV-vis
(pentane): 908 (6.6), 616 (11.5), 428 (0.2).
LtBuFe(Me2Npy)NN(Me2Npy)FeLtBu (6). Similarly to the preparation
of 4, LtBuFeNNFeLtBu (102 mg, 89 µmol) was reacted with 4-(di-
methylamino)pyridine (22 mg, 180 µmol) in Et2O (8 mL). Dark green
crystals were obtained from pentane at -35 °C (98 mg, 79%). 1H NMR
(C6D6) δ 116 (2H, 0.2, backbone C-H), 62 (8H, 2, m-H), 22 (4H, 3,
(C5H12) νCO 2036, 2000, 1967, 1953, 1922 cm-1
.
LMeFeOC(Me)(Ph)C(Me)(Ph)COFeLMe (12). A 20 mL scintillation
vial was charged with LMeFeNNFeLMe (100 mg, 103 µmol) and pentane
(8 mL) to give a dark red solution. The solution was cooled at -35 °C
for 30 min, and then acetophenone (25 mg, 205 µmol) in pentane (1
mL) was added to give a dark green solution. The reaction was warmed
to room temperature and stirred overnight to give a dark yellow solution,
with the precipitation of a yellow solid. The volatile materials were
removed under reduced pressure, and the residue washed with hex-
amethyldisiloxane (1 mL) to remove dark colored impurities. The
9
J. AM. CHEM. SOC. VOL. 128, NO. 3, 2006 767

A R T I C L E S
Smith et al.
remaining yellow solid was redissolved in Et2O and filtered through
diatomaceous earth. The solution was concentrated to 4 mL and pentane
(2 mL) was added. A yellow powder precipitated from solution at -35
resealable NMR tube. The 1H NMR spectrum was recorded. 1H NMR
(THF-d8) δ 55 (4H, m-H), 16 (18H, C(CH3)3), 8 (2H, CH(CH3)2), 4
(O(CH2CH2)2), 1 (O(CH2CH2)2), -1 (2H, CH(CH3)2), -5 (12H, CH-
(CH3)2), -14 (6H, CH(CH3)2), -27 (6H, CH(CH3)2), -73 (2H, p-H),
-119 (1H, backbone C-H). The tube was frozen and placed under
vacuum and N2 was introduced into the tube. The green solution became
red-purple within 2 min, and signals characteristic of LtBuFeNNFeLtBu
1
°C (70 mg, 57%). H NMR (C6D6) δ 193 (2H, backbone C-H), 145
(6H, 0.1, OCH3), 71 (12H, 0.5, CH3), 37 (4H, 1, m-OC(C6H6)), 33
(2H, 2, p-OC(C6H6)), 3 (4H, 5, o-OC(C6H6)), -9 (12H, 1, CH(CH3)2),
-11 (12H, 1, CH(CH3)2), -15 (4H, 2, m-H), -16 (4H, 2, m-H), -68
(4H, 2, p-H), -90 (12H, 0.4, CH(CH3)2), -95 (12H, 0.5, CH(CH3)2),
-112 (4H, 0.1, CH(CH3)2), -158 (4H, 0.1, CH(CH3)2). µeff (Evans,
C6D6) 6.7 µB. Anal. Calcd. for C74H98N4O2Fe2: C 74.86, H 8.32, N
4.72. Found C 72.25, H 8.20, N 4.45. UV-vis (pentane): 486 (0.60).
LtBuFeOC(Me)(Ph)C(Me)(Ph)COFeLtBu (13). A 20 mL scintillation
vial was charged with LtBuFeNNFeLtBu (100 mg, 87 µmol) and pentane
(8 mL) to give a purple-red solution. The solution was cooled to -35
°C, and a solution of acetophenone (20 mg, 175 µmol) in pentane (1
mL) was added. There was no immediate change, but within 30 s the
solution developed a yellow hue, and in 10 min an orange solid started
to precipitate. The reaction was stirred at room-temperature overnight.
The volatile materials were removed under reduced pressure, and the
residue was washed with hexamethyldisiloxane (1 mL) to remove dark
colored impurities. The remaining orange solid was redissolved in warm
THF, filtered through diatomaceous earth, and cooled to -35 °C to
give an orange powder (60 mg, 51%). The complex is insoluble in
pentane and sparingly soluble in benzene. It dissolves slowly in THF.
1H NMR (C6D6) δ 145 (2H, backbone C-H), 111 (6H, OCH3), 38
(36H, 1, C(CH3)3), 34 (4H, 1, m-OC(C6H6)), 17 (2H, 2, p-OC(C6H6)),
3 (4H, 4, o-OC(C6H6)), -4 (4H, 2, m-H), -5 (4H, 2, m-H), -14 (12H,
CH(CH3)2), -15 (12H, CH(CH3)2), -45 (4H, CH(CH3)2), -55 (4H,
CH(CH3)2), -69 (4H, 1, p-H), -81 (12H, 0.3, CH(CH3)2), -88 (12H,
0.3, CH(CH3)2). Anal. Calcd. for C86H122N4O2Fe2: C 76.20, H 9.07, N
4.13. Found C 72.60, H 8.99, N 4.18. UV-vis (THF): 519 (1.1), 492
(0.88).
1
were observed in the H NMR spectrum.
Reaction of K2LMeFeNNFeLMe with CH3OTf. K2LMeFeNNFeLMe
(10 mg, 9 µmol) and C6D6 (ca. 0.5 mL) were added to a resealable
NMR tube. Methyl triflate (2.0 µL, 18 µmol) was added via syringe to
the dark blue solution. Gas evolution was observed, and the solution
became yellow orange in color with the formation of a white solid. 1H
NMR spectroscopy revealed the quantitative formation of LMeFeMe.28b
X-ray Structural Determinations. Crystalline samples of all the
complexes were grown in the glovebox from pentane solutions at -35
°C. All crystals were rapidly mounted under Paratone-8277 onto glass
fibers, and immediately placed in a cold nitrogen stream at -80 °C on
the X-ray diffractometer. The X-ray intensity data were collected on a
standard Bruker-axs SMART CCD Area Detector System equipped
with a normal focus molybdenum-target X-ray tube operated at 2.0
kW (50 kV, 40 mA). A total of 1321 frames of data (1.3 hemispheres)
were collected using a narrow frame method with scan widths of 0.3°
in ω and exposure times of 30 s/frame, using a detector-to-crystal
distance of 5.09 cm. Frames were integrated to a maximum 2θ angle
of 56.5° with the Bruker-axs SAINT program. The final unit cell
parameters (at -80 °C) were determined from the least-squares
refinement of three-dimensional centroids of >3400 reflections for each
crystal. Data were corrected for absorption with the SADABS program,
except when noted otherwise. The space groups were assigned using
XPREP, and the structures were solved by direct methods using Sir9242-
(WinGX v1.63.02) and refined employing full-matrix least-squares on
F2 (Bruker-axs, SHELXTL-NT, version 5.10). Nonhydrogen atoms were
refined with anisotropic thermal parameters, except disordered solvent
in some cases. Hydrogen atoms were included in idealized positions
with riding thermal parameters. Details are provided in Table 2 and
the Supporting Information.
K2LMeFeNNFeLMe (14).
A
flask was charged with LMeFe-
(µ-Cl)2FeLMe 60b (2.50 g, 2.46 mmol) and pentane (25 mL). KC8 (1.39
g, 10.3 mmol) was added to the yellow slurry, leading to the formation
of a very dark blue solution. The reaction mixture was stirred overnight
at room temperature and filtered through diatomaceous earth to give a
dark blue solution. The solution was concentrated to 15 mL and cooled
Computational Methods. The calculations reported herein employed
the GAMESS61 and Jaguar62 packages. GAMESS was used for
multireference calculations and Jaguar for density functional theory
(DFT) calculations. The Stevens effective core potentials and valence
basis sets were employed,63 augmented with a d polarization function
on main group elements. Hydrogen was described with the -31G basis
set. For density functional calculations the B3LYP64 hybrid functional
was used. This combination of theory level and basis sets was used in
previous calculations on iron-â-diketiminate-dinitrogen complexes.25
Multireference calculations44 were used to cross-reference the ground
state multiplicities obtained by experiment and density functional
methods given the density of low energy states in these complexes.
The active space was chosen to include all orbitals (and the electrons
contained therein) needed to describe the Fe-dinitrogen moiety. All
multireference calculations were performed within the FORS (fully
optimized reaction space) paradigm, i.e., all possible configuration state
functions, within the limits of spatial and spin symmetry, were generated
for the active space of interest. For multireference Møller-Plesset 2nd
1
to -35 °C to give dark blue crystals (1.21 g, 43%). H NMR (C6D6)
δ 25 (8H, 5, m-H), 3 (12H, 1, CH3), 1 (24H, 1, CH(CH3)2), -1 (24H,
1, CH(CH3)2), -38 (4H, 1, p-H), -82 (2H, 0.5, backbone C-H), -305
(8H, 0.2, CH(CH3)2). µeff (Evans, C6D6) 4.9 µB. Anal. Calcd for
C58H82N6Fe2K2: C 66.14, H 7.85, N 7.98. Found C 66.19, H 7.86, N
6.48. UV-vis (pentane): 722 (11.4), 486 (7.7), 449 (8.5), 369 (17).
LtBuFe(µ-Cl)K(18-crown-6) (15). In an argon-filled glovebox, a 20
mL scintillation vial was charged with LtBuFeCl (200 mg, 337 µmol),
18-crown-6 (89 mg, 337 µmol) and Et2O (10 mL) to give a red solution.
To this solution was added KC8 (50 mg, 370 µmol), resulting in the
immediate formation of a dark green solution and black graphite. The
reaction was stirred at room-temperature overnight and filtered through
diatomaceous earth to give a dark green solution. The solvent was
removed in vacuo to give a dark green solid. This solid was dissolved
in toluene (4 mL), layered with pentane (2 mL) and cooled to -35 °C
to give dark green crystals (103 mg, 35%). 1H NMR (C6D6) δ 48 (4H,
CH(CH3)2), 10 (18H, C(CH3)3), 6 (24H, (OCH2CH2)6), 3 (12H, CH-
(CH3)2), 2 (4H, m-H), -8 (12H, CH(CH3)2), -44 (2H, p-H), -61 (1H,
backbone C-H). 1H NMR (THF-d8) δ 43 (4H, m-H), 7 (18H, C(CH3)3),
5 (24H, (OCH2CH2)6), 0.02 (12H, CH(CH3)2), -2 (2H, CH(CH3)2),
-9 (14H, CH(CH3)2, CH(CH3)2), -56 (2H, p-H), -165 (1H, backbone
C-H). µeff (Evans, C6D6) 3.6(3) µeff.
(60) (a) Smith, J. M.; Lachicotte, R. J.; Holland, P. L. Chem. Commun. 2001,
1542-1543. (b) Eckert, N. A.; Smith, J. M.; Lachicotte, R. J.; Holland, P.
L. Inorg. Chem. 2004, 43, 3306-3321.
(61) Schmidt, M. W.; Baldridge, K. K.; Boatz, J. A.; Elbert, S. T.; Gordon, M.
S.; Jensen, J. H.; Koseki, S.; Matsunaga, N.; Nguyen, K. A.; Su, S.; Windus,
T. L.; Dupuis, M.; Montgomery, J. A. J. Comput. Chem. 1993, 14, 1347-
1363.
Reduction of LtBuFeCl under Argon. In an argon-filled glovebox,
a 20 mL scintillation vial was charged with LtBuFeCl (20 mg, 34 µmol)
and THF-d8 (0.5 mL) to give an orange solution. Solid KC8 (5 mg, 37
µmol) was added, resulting in the immediate formation of a dark green
solution and black graphite. The reaction was stirred at room temper-
ature for 3 h, and then filtered through diatomaceous earth into a
(62) Jaguar, version 5.5, Schro¨dinger, 1500 S. W. First Avenue, Suite 1180,
Portland, OR 97201, http://www.schrodinger.com.
(63) Stevens, W. J.; Krauss, M.; Basch, H.; Jasien, P. G. Can. J. Chem. 1992,
70, 612-613.
(64) Becke, A. D. J. Chem. Phys. 1993, 98, 5648-5652; Lee, C.; Yang, W.;
Parr, R. G. Phys. ReV. B 1988, 37, 785-789; Vosko, S. H.; Wilk, L.; Nusair,
M. Can. J. Phys. 1980, 58, 1200-1211.
9
768 J. AM. CHEM. SOC. VOL. 128, NO. 3, 2006

Iron Dinitrogen Complexes
A R T I C L E S
order perturbation theory (MRMP2) calculations, single and double
excitations from the FORS active space to the remaining virtual orbitals
were allowed.
Geometry optimization calculations utilized the B3LYP functional
and density-functional methods given the much greater computational
tractability of DFT in relation to MCSCF and MRMP2 methods. In all
cases, the ground-state multiplicities predicted by B3LYP agree with
those obtained from MCSCF and MRMP2 calculations. DFT calcula-
tions were performed within the restricted open-shell (RODFT)
paradigm to obviate spin contamination issues.
Resonance Raman Spectroscopy. Resonance Raman spectra were
obtained from 10 to 20 mM solutions contained in a 5-mm NMR tube
spinning at ∼20 Hz. Raman scattering was excited using 406.7 nm
emission from a Kr+ laser (15 to 35 mW). Spectra were recorded at
ambient temperature using 135° backscattering geometry with the laser
beam focused to a line. Scattered light was collected with an f1 lens
and filtered with a holographic notch filter to attenuate Rayleigh
scattering. The polarization of the scattered light was then scrambled
and the spot image was f-matched to a 0.63 m spectrograph fitted with
a 2400 groove/mm grating and a LN2-cooled CCD camera. The
spectrometer was calibrated using the Raman bands of toluene, pentane,
acetone, and d6-DMSO as external frequency standards.
Acknowledgment. Funding for this work was provided by
the National Science Foundation (CHE-0309811 to T.R.C. and
CHE-0112658 to P.L.H.), National Institutes of Health (GM-
065313 to P.L.H.), A. P. Sloan Foundation (Research Fellowship
to P.L.H.), USDA (ND05299 to K.R.R.), and Hermann Frasch
Foundation (446-HF97 to K.R.R.). We thank Cambridge
Isotopes for a generous gift of 15N2.
Supporting Information Available: Variable-temperature
NMR spectra and crystallographic details. This material is
available free of charge via the Internet at http://pubs.acs.org.
JA052707X
9
J. AM. CHEM. SOC. VOL. 128, NO. 3, 2006 769