Theoretical, thermodynamic, spectroscopic, and structural studies of the consequences of one-electron oxidation on the Fe-X bonds in 17- and 18-electron Cp*Fe(dppe)X complexes (X=F, C1, Br, I, H, CH3)

Article abstract of DOI:10.1021/ja0106927

The compounds Cp*Fe(dppe)X ([Fe*]X)and the corresponding cation radicals [Fe*]X⁺ are available for the series X=F, C1, Br, I, H, CH₃. This has allowed for a detailed investigation of the dependence of the nature of Fe-X bonding on th

Full text of DOI:10.1021/ja0106927

9984
J. Am. Chem. Soc. 2001, 123, 9984-10000
Theoretical, Thermodynamic, Spectroscopic, and Structural Studies of
the Consequences of One-Electron Oxidation on the Fe-X Bonds in
17- and 18-Electron Cp*Fe(dppe)X Complexes (X ) F, Cl, Br, I, H,
CH3)
Mats Tilset,*,1a,b Irene Fjeldahl,1b Jean-Rene´ Hamon,*,1c Paul Hamon,1c Lo1ıc Toupet,1d
Jean-Yves Saillard,*,1e Karine Costuas,1e and Anthony Haynes1f
Contribution from the Department of Chemistry, UniVersity of Oslo, P.O. Box 1033, Blindern,
N-0315 Oslo, Norway, UMR CNRS 6509, Organome´talliques et Catalyse: Chimie et Electrochimie
Mole´culaires, Institut de Chimie de Rennes, UniVersite´ de Rennes 1, Campus de Beaulieu,
F-35042 Rennes Cedex, France, UMR CNRS 6626, Groupe Matie`re Condense´e et Mate´riaux,
UniVersite´ de Rennes 1, Campus de Beaulieu, F-35042 Rennes Cedex, France, UMR CNRS 6511,
Chimie du Solide et Inorganique Mole´culaire, Institut de Chimie de Rennes, UniVersite´ de Rennes 1,
Campus de Beaulieu, F-35042 Rennes Cedex, France, and Department of Chemistry, UniVersity of
Sheffield, Sheffield S3 7HF, England
ReceiVed March 15, 2001
Abstract: The compounds Cp*Fe(dppe)X ([Fe*]X) and the corresponding cation radicals [Fe*]X•+ are available
for the series X ) F, Cl, Br, I, H, CH3. This has allowed for a detailed investigation of the dependence of the
nature of Fe-X bonding on the identity of X and the oxidation state (charge) of the complex. Cyclic voltammetry
demonstrates that the electrode potentials for the [Fe*]X0/+ couples decrease in the order I > Br > Cl > H >
F > CH3. An “inverse halide order” is seen, in which the most electronegative X leads to the most easily
oxidized complex. This suggests that F is the best donor among the halides. The halide trend is also reflected
in NMR spectroscopic data. Mo¨ssbauer spectroscopy data also suggest that the F ligand is a strong donor
(relative to H and CH3) in [Fe*]X•+. DFT calculations on CpFe(dpe)X ([Fe]X) model complexes nicely reproduce
the trend in the electrode potentials for the [Fe*]X0/+ couples. Analysis of the theoretical data within the
halogen series indicates that the energy of the [Fe]X HOMO does not correlate with the extent of its Fe(dπ)-
X(pπ) antibonding character, which varies in the order I > Br > Cl > F, but rather depends on the destabilizing
electrostatic effect caused by X. This effect varies in the order F > Cl > Br > I. A thermochemical cycle that
incorporates the [Fe*]X0/+ and [Fe*]0/+ electrode potentials was used to investigate the effect of the oxidation
state of the complex on the homolytic bond dissociation energy (BDEhom), defined for the processes Fe-X f
Fe• + X• and Fe-X•+ f Fe•+ + X•. For all X, it was found that a one-electron oxidation leads to a weakening
of the Fe-X bond. This trend was reproduced by the DFT calculations. On the other hand, IR νFe-X spectroscopy
data showed an increase in the stretching frequencies for X ) H and Cl upon oxidation. X-ray crystallographic
data showed a shortening of the Fe-Cl bond upon oxidation. The trends in IR and Fe-Cl bond distances
were reproduced in the DFT calculations. The combined data therefore suggest that oxidation leads to weaker,
but shorter, Fe-X bonds. A second thermochemical cycle was applied to investigate the effect of the one-
electron oxidation on the heterolytic bond dissociation energies (BDEhet), defined for the processes Fe-X f
Fe+ + X- and Fe-X•+ f Fe2+ + X-. In this case, the oxidation led to bond strengthening in all cases. The
computed BDE values have been analyzed within Ziegler’s transition state methodology and decomposed into
two components, one electrostatic and one covalent, describing the interaction between the unrelaxed fragments.
In all the computed BDEhom and BDEhet values of the [Fe]X models the electrostatic component is important.
This helps to understand their respective variations upon oxidation.
Introduction
Recently, the nature of the bonding between organotransition-
metal centers and electronegative σ-bonded ligands X such as
halide, alkoxide, and amido groups has received considerable
attention.3 In addition to forming covalent M-X bonds, these
ligands are capable of acting as π donors toward the metal. A
(2) (a) Simo˜es, J. A. M.; Beauchamp, J. L. Chem. ReV. 1990, 90, 629.
(b) Simo˜es, J. A. M., Ed. Energetics of Organometallic Species; Kluwer
Academic: Dordrecht, The Netherlands, 1992. (c) Marks, T. J., Ed. Bonding
Energetics in Organometallic Compounds; ACS Symp. Ser. No. 428;
American Chemical Society: Washington, DC, 1990. (d) Halpern, J. Inorg.
Chim. Acta 1985, 100, 41. (e) Minas da Piedade, M. E., Ed. Energetics of
Stable Molecules and ReactiVe Intermediates; Kluwer Academic: Dordrecht,
The Netherlands, 1999.
Knowledge of the nature and energetics of metal-ligand
bonding in organotransition-metal complexes is crucial to the
understanding of organometallic reactions and catalysis.2 Insight
into the often complex reaction mechanisms based on a
quantitative understanding of strengths of bonds being formed
and broken in the reaction steps that are involved may ultimately
help in the design of new and improved processes.
(1) (a) Corresponding author. E-mail: mats.tilset@kjemi.uio.no. (b)
University of Oslo. (c) UMR CNRS 6509, Universite´ de Rennes 1. (d) UMR
CNRS 6626, Universite´ de Rennes 1. (e) UMR CNRS 6511, Universite´ de
Rennes 1. (f) University of Sheffield.
10.1021/ja0106927 CCC: $20.00 © 2001 American Chemical Society
Published on Web 09/21/2001

Fe-X Bonds in 17- and 18-Electron Cp*Fe(Dppe)X Complexes
J. Am. Chem. Soc., Vol. 123, No. 41, 2001 9985
ligand pπ-metal dπ electron-pair interaction generally serves
to destabilize electronically saturated complexes via filled-filled
repulsive interactions, whereas coordinatively unsaturated spe-
cies may achieve considerable stabilization through partial π
bond formation.4 A substantial line of evidence,3 including IR
Scheme 1
ν
CO spectroscopy data,5,6 electrode potentials,7 chemical reactiv-
ity,5,8 and theoretical calculations,4b,5a,9 suggests that among the
halides, it is the fluoride ligand that appears to be the most
efficient electron-pair donor toward the metal. The nature of
this interaction, at least in saturated complexes, remains highly
controversial, however: Holland et al.10 have provided an
alternative explanation based on Drago’s E-C bonding scheme.11
Here, M-X bonding is primarily explained in terms of
electrostatic (E) and covalent (C) contributions. Destabilizing
π interactions were considered not to be the major source of
M-X bonding preferences in some late metal complexes.10
The presence of significant π bonding from halide to metal
should be reflected in bond dissociation energy (BDE) changes
when this bonding is “switched on”4a by the generation of
coordinative unsaturation. The unsaturation can conceptually
be generated by ligand dissociation from 18-electron precursors
or by two-electron oxidations of the 18-electron species. One
might even expect “partial unsaturation” to be achieved by one-
electron oxidations. With this in mind, we have investigated
the effects of one- and two-electron oxidation processes on
M-X bonding through the use of thermochemical cycles that
incorporate electrode potential data. Such techniques have
proven powerful for extracting bond energy data that are
frequently not directly available by other experimental methods.
Transient electrochemical techniques, of which cyclic voltam-
metry is most commonly used, offer the advantage that relatively
short-lived species can be investigated. The methodology was
pioneered in organic chemistry by Breslow and co-workers12
and was later adapted by numerous research groups, as
manifested by a number of relevant reviews.13
organotransition-metal hydrides.14,15 Absolute homolytic M-H
bond dissociation energies (henceforth to be denoted BDEhom
)
in solution have been determined, and relative M-H acidity
(pKa) and BDEhom data for 18-electron neutral complexes and
their 17-electron cation radicals have been estimated. The
method that was applied to evaluate the M-H BDE differences
between 18- and 17-electron species14c,d can be straightforwardly
modified to accommodate any M-X bond. Scheme 1 depicts
the resulting thermochemical cycle that can be used to evaluate
the effect of a one-electron oxidation on the homolytic M-X
bond dissociation energy through eq 1 . (The subscript “G” in
∆BDEhom ) BDEhom(MX•+) - BDEhom(MX) )
F[E°ox(M•) - E°ox(MX)] (1)
the scheme signifies that the quantities are Gibbs free energy
changes; enthalpy values may be obtained if the entropy
contributions from the two redox couples cancel. This may be
assumed to be the case for large molecules of similar shapes
and charges,16 as will be the case in this work). Thus, according
to eq 1, BDE changes between 18-electron M-X and 17-
electron M-X•+ can be determined if electrode potentials for
the M-X/M-X+ and M•/M+ couples are available.
Data pertaining to the consequences of one-electron oxida-
tions on M-X bonding in organotransition-metal complexes
are scarce. Past results from our laboratories14c,d have established
that one-electron oxidations of low-oxidation state metal car-
bonyl hydrides (CpCr(CO)2(L)H, where L ) P(OMe)3, PPh3,
PEt3; Cp*Cr(CO)3H; and TpM(CO)3H and Tp*M(CO)3H, where
M ) Cr, Mo, W)17 consistently cause M-H BDE values to be
reduced by ca. 30 kJ/mol. Contrasting these results, the IR νFe-H
During the past decade, analogous thermochemical cycles
have been adapted to investigate M-H bonding energetics in
(3) For recent reviews, see: (a) Caulton, K. G. New J. Chem. 1994, 18,
25. (b) Doherty, N. M.; Hoffman, N. W. Chem. ReV. 1991, 91, 553.
(4) (a) Poulton, J. T.; Folting, K.; Streib, W. E.; Caulton, K. G. Inorg.
Chem. 1992, 31, 3190. (b) Johnson, T. J.; Folting, K.; Streib, W. E.;
Eisenstein, O.; Caulton, K. G. Inorg. Chem. 1995, 34, 488. (c) Bickford,
C. C.; Johnson, T. J.; Davidson, E. R.; Caulton, K. G. Inorg. Chem. 1994,
33, 1080. (d) Barthazy, P.; Stoop, R. M.; Wo¨rle, M.; Togni, A.; Mezzetti,
A. Organometallics 2000, 19, 2844.
stretching frequency was raised from 1869 to 1886 cm-1
,
indicative of a strengthened Fe-X bond, when Cp*Fe(dppe)H
was oxidized to its stable cation radical.18,19 Shifts to higher
energies are also observed in the Cp′Fe(dippe)H0/+ series (Cp′
(5) (a) Poulton, J. T.; Sigalas, M. P.; Folting, K.; Streib, W. E.; Eisenstein,
O.; Caulton, K. G. Inorg. Chem. 1994, 33, 1476. (b) Huang, D. J.; Caulton
K. G. J. Am. Chem. Soc. 1997, 119, 3185.
(6) Vaska, L.; Peone, J. J. Chem. Soc., Chem. Commun. 1971, 418.
(7) Schiavon, G.; Zecchin, S.; Pilloni, G.; Martelli, M. Inorg. Nucl. Chem.
1977, 39, 115.
(8) Darensbourg, D. J.; Klausmeyer, K. K.; Reibenspies, J. H. Inorg.
Chem. 1995, 34, 4933.
(9) Abu-Hassayan, F.; Goldman, A. S.; Krogh-Jespersen, K. Inorg. Chem.
1994, 33, 5122.
(10) (a) Holland, P. L.; Andersen, R. A.; Bergman, R. G.; Huang, J.;
Nolan, S. P. J. Am. Chem. Soc. 1997, 119, 12800. (b) Holland, P. L.;
Andersen, R. A.; Bergman, R. G. Comments Inorg. Chem. 1999, 21, 115.
(11) (a) Drago, R. S. Applications of Electrostatic-CoValent Models in
Chemistry; Surfside: Gainesville, FL, 1994. (b) Drago, R. S.; Wong, N.
M.; Ferris, D. C. J. Am. Chem. Soc. 1992, 114, 91.
(12) (a) Breslow, R.; Chu, W. J. Am. Chem. Soc. 1973, 95, 411. (b)
Breslow, R.; Grant, J. J. Am. Chem. Soc. 1977, 99, 7745. (c) Jaun, B.;
Schwarz, J.; Breslow, R. J. Am. Chem. Soc. 1980, 102, 5741.
(13) (a) Wayner, D. D. M.; Parker, V. D. Acc. Chem. Res. 1993, 26,
287. (b) Arnett, E. M.; Flowers, R. A.; Ludwig, R. T.; Meckhof, A.; Walek,
S. Pure Appl. Chem. 1995, 67, 729. (c) Bordwell, F. G.; Satish, A. V.;
Zhang, S.; Zhang, X.-M. Pure Appl. Chem. 1995, 67, 735. (d) Arnett, E.
M.; Flowers, R. A. Chem. Soc. ReV. 1993, 22, 9. (e) Bordwell, F. G.; Zhang,
X.-M. Acc. Chem. Res. 1993, 26, 510. (f) Astruc, D. Acc. Chem. Res. 2000,
33, 287.
(14) (a) Tilset, M.; Parker, V. D. J. Am. Chem. Soc. 1989, 111, 6711;
1990, 112, 2843 (corrigendum). (b) Parker, V. D.; Handoo, K. L.; Roness,
F.; Tilset, M. J. Am. Chem. Soc. 1991, 113, 7493. (c) Tilset, M. J. Am.
Chem. Soc. 1992, 114, 2740. (d) Skagestad, V.; Tilset, M. J. Am. Chem.
Soc. 1993, 115, 5077. (e) Pedersen, A.; Skagestad, V.; Tilset, M. Acta Chem.
Scand. 1995, 49, 632.
(15) (a) Berning, D. E.; Noll, B. C.; DuBois, D. L. J. Am. Chem. Soc.
1999, 121, 11432. (b) Berning, D. E.; Miedaner, A.; Curtis, C. J.; Noll, B.
C.; DuBois, M. C. R.; DuBois, D. L. Organometallics 2001, 20, 1832.
(16) The assumption has been found to hold well for organic radicals R•
vs hydrocarbons R-H: Wayner, D. D. M.; McPhee, D. J.; Griller, D. J.
Am. Chem. Soc. 1988, 110, 132.
(17) Abbreviations: Cp ) η5-C5H5; Cp* ) η5-C5Me5; dppe ) η2-Ph2-
PCH2CH2PPh2; dpe ) η2-H2PCH2CH2PH2; dippe ) η2-iPr2PCH2CH2PiPr2;
BArf- ) [3, 5-(CF3)2C6H3]4B-; Tp ) HB(C3H3N2)3; Tp* ) HB(3,5-Me2C3-
HN2)3; [Fe*] ) Cp*Fe(dppe); [Fe] ) CpFe(dpe).
(18) The previous report (ref 19) of a lower-energy shift from 1869 to
1860 cm-1 upon oxidation was based on erroneous data for the cation.
Reproducible data have now been obtained from different batches of 5+PF6
-
(see Experimental Section).
(19) (a) Roger, C.; Hamon, P.; Toupet, L.; Rabaaˆ, H.; Saillard, J.-Y.;
Hamon, J.-R.; Lapinte, C. Organometallics 1991, 10, 1045. (b) Hamon, P.;
Toupet, L.; Hamon, J.-R.; Lapinte, C. Organometallics 1992, 11, 1429.

9986 J. Am. Chem. Soc., Vol. 123, No. 41, 2001
Tilset et al.
) Cp, 1911 vs 1939 cm-1; Cp′ ) Cp*, 1901 vs 1915 cm-1).20
Poli and co-workers21 recently reported that the IR νM-H
stretching frequencies of Cp*M(dppe)H3•+ (M ) Mo, W) were
10-20 cm-1 higher than those for the neutral counterparts.
These IR results indicate an oxidative strengthening of the M-H
bonds also in the relatively high oxidation state complexes, and
Scheme 2
this conclusion was supported by DFT calculations for the
0/+
Cp*W(dppe)H3
system. Finally, IR νM-H data have been
reported for redox couples of non-Cp hydridometal phosphine
complexes of Co22 and W;23 oxidation-induced frequency shifts
occur to higher and lower frequencies in these systems.
It has been demonstrated that the sterically crowded and
electron-rich Cp*Fe(dppe) moiety24 (to be abbreviated as [Fe*])
supports metal complexes in a great number of oxidation states,
and compounds have been isolated with electron counts ranging
from 16 to 19.19,25 A persistent 15-electron species has even
been generated in solution.25e The Cp*Fe(dppe) derivatives
therefore are particularly well suited for the application of
thermochemical cycles. In this contribution, we probe the details
of Fe-X bonding in a range of Cp*Fe(dppe)X complexes (X
) F, Cl, Br, I, H, CH3) using a variety of techniques, including
electrochemistry, thermochemical cycles, X-ray structural stud-
ies, NMR, IR, ESR, and Mo¨ssbauer spectroscopy, as well as
theoretical (DFT) calculations. Parts of the experimental results
have been previously communicated.26
Table 1. Selected NMR Chemical Shift Data for Neutral
Cp*Fe(dppe)X Complexesa
1H
(C5Me5)
13C{1H}
(C5Me5)
13C{1H}
(C5Me5)
31P
(dppe)
compound
Cp*Fe(dppe)F (1)
Cp*Fe(dppe)C1 (2)
Cp*Fe(dppe)Br (3)
Cp*Fe(dppe)I (4)
Cp*Fe(dppe)H (5)
Cp*Fe(dppe)Me (6)
1.33
1.40
1.53
1.61
1.62
1.47
10.0
10.3
10.8
11.5
11.5
10.5
82.9
83.6
83.6
83.6
85.2
85.5
89.4
91.6
93.2
95.1
107.9
106.5
a Chemical shifts (δ) recorded in benzene-d6 at ambient temperature.
leading always to decomposition to unidentified products.
Consequently, neither satisfactory elemental analysis nor crystals
suitable for an X-ray structural determination could be obtained.
Analogous relative instabilities of low oxidation state, coordi-
natively saturated Fe28a and Mo28b fluoro complexes have been
recently reported, and also in those cases precluded their
isolation. Analytically pure CpRe(PPh3)(NO)F can be isolated
but decomposes on the time scale of hours in solution.29 Despite
Results and Discussion
I. Preparation of Complexes and Structural and Spec-
troscopic Studies. The preparative work was aimed at comple-
menting a family of related complexes of the type Cp*Fe(dppe)-
Xn+ [X ) F (1), Cl (2), Br (3), I (4), H (5), Me (6)] with a
variety of σ-bonded X ligands, both as 18-electron neutral
species (n ) 0) and as 17-electron radical cations (n ) 1). This
has facilitated a detailed investigation on the effect of X on
spectroscopic and electrochemical properties and on Fe-X
bonding energetics in this series of complexes.
this instability in solution, complex 1 features good H, 13C,
1
31P, and 19F NMR spectra (see Experimental Section and Table
1) that were obtained from a sample that was analyzed by NMR
immediately after its preparation. The presence of the fluoride
as the sixth ligand in 1 is clearly authenticated by a doublet at
δ 89.4 in the 31P NMR spectrum and a triplet at δ -44.4 in the
19F NMR spectrum (2JPF ) 43 Hz). Similarly, large couplings
between P and F bonded to the same metal center are commonly
observed for metal phosphine fluorides.3b,4d,29-31 The cyclic
voltammogram of complex 1 (vide infra) exhibits two reversible,
monoelectronic waves attributable to a single redox-active
species.
Synthesis of the Iron(II) Halide Complexes. The novel
fluoro complex Cp*Fe(dppe)F (1) is prepared in almost
quantitative isolated yield (86%) by treating the paramagnetic,
-
red 16-electron Cp*Fe(dppe)+PF6 derivative25e with 1 equiv
The neutral Fe(II) bromo complex Cp*Fe(dppe)Br (3) is
readily prepared by chloride-bromide ion exchange between
the known Cp*Fe(dppe)Cl (2) and KBr in dichloromethane,
following the same procedure as that previously described for
the iodo derivative Cp*Fe(dppe)I (4).19a The complex 3 was
isolated in 90% yield after slow crystallization, as air and
thermally stable, dark-brown crystals that provided satisfactory
elemental analysis.
Table 1 summarizes some key NMR data, recorded under
identical conditions, for the series 1-6. In several respects,
fluoro complex 1 stands out from the other halides. For example,
the 1H signal of the Cp* ligand at δ 1.33 and the dppe 31P NMR
signal at δ 89.4 appear substantially upfield shifted relative to
the others (δ 1.40-1.61 and 91.6-95.1, respectively). In fact,
there is a general trend in the data in Table 1 that the chemical
shift decreases with increasing halide electronegativity. This
trend is opposite to that reported by Gladysz et al. for the CpRe-
(PPh3)(NO)X series,29 and also counters the trend that might
of dry CsF in THF (Scheme 2). Complex 1 is isolated as a
thermally stable, yellowish-green powder, and can be stored for
several weeks under argon without any signs of decomposition.
This first isolated organometallic Fe(II) fluoride compound
bearing a cyclic hydrocarbon ligand27 proved to be rather
sensitive in solution and all attempts at recrystallizing it failed,
(20) Jimenez-Tenorio, M.; Puerta, M. C.; Valerga, P. Organometallics
1994, 13, 3330.
(21) Pleune, B.; Morales, D.; Meunier-Prest, R.; Richard, P.; Collange,
E.; Fettinger, J. C.; Poli, R. J. Am. Chem. Soc. 1999, 121, 2209.
(22) (a) Pilloni, G.; Schiavon, G.; Zotti, G.; Zecchin, S. J. Organomet.
Chem. 1977, 134, 305. (b) Bianchini, C.; Innocenti, P.; Meli, A.; Peruzzini,
M.; Zanobini, F.; Zanello, P. Organometallics 1990, 9, 2514. (c) Kruse,
W.; Atalla, J. H. Chem. Commun. 1968, 921. (d) Sacco, A.; Ugo, R. J.
Chem. Soc. 1964, 3274. (e) Ghilardi, C. A.; Midollini, S.; Sacconi, L. Inorg.
Chem. 1975, 14, 1790.
(23) Sharp, P. R.; Frank, K. G. Inorg. Chem. 1985, 24, 1808.
(24) Paul, F.; Lapinte, C. Coord. Chem. ReV. 1998, 178-180, 427.
(25) (a) Roger, C.; Marseille, P.; Salus, C.; Hamon, J.-R.; Lapinte, C. J.
Organomet. Chem. 1987, 336, C13. (b) Morrow, J.; Catheline, D.; Desbois,
M.-H.; Manriquez, J. M.; Ruiz, J.; Astruc, D. Organometallics 1987, 6,
2605. (c) Morrow, J. R.; Astruc, D. Bull. Soc. Chim. Fr. 1992, 129, 319.
(d) Hamon, P.; Hamon, J.-R.; Lapinte, C. J. Chem. Soc., Chem. Commun.
1992, 1602. (e) Hamon, P.; Toupet, L.; Hamon, J.-R.; Lapinte, C.
Organometallics 1996, 15, 10.
(28) (a) Landau, S. E.; Morris, R. H.; Lough, A. J. Inorg. Chem. 1999,
38, 6060. (b) Aston, G. M.; Badriya, S.; Farley, R. D.; Grime, R. W.; Ledger,
S. J.; Mabbs, F. E.; McInnes, E. J. L.; Morris, H. W.; Ricalton, A.;
Rowlands, C. C.; Wagner, K.; Whiteley, M. W. J. Chem. Soc., Dalton Trans.
1999, 4379.
(29) Agbossou, S. K.; Roger, C.; Igau, A.; Gladysz, J. A. Inorg. Chem.
1992, 31, 419.
(26) Tilset, M.; Hamon, J.-R.; Hamon, P. Chem. Commun. 1998, 765.
(27) Murphy, E. F.; Murugavel, R.; Roesky, H. W. Chem. ReV. 1997,
97, 3425.

Fe-X Bonds in 17- and 18-Electron Cp*Fe(Dppe)X Complexes
J. Am. Chem. Soc., Vol. 123, No. 41, 2001 9987
be anticipated on the basis of the relative electronegativities of
the halide ligands. However, the trend is in good agreement
with the electrochemical data (vide infra): The more electron
rich complex exhibits the most upfield shifted NMR signals.
The instability of complex 1 may not be too surprising if
one considers the generalized perturbation theory of donor-
acceptor interactions, which does not favor the formation of a
stable covalent Fe-F bond.32 As predicted by the “hard and
soft acid and base” (HSAB) concept,33 the interaction between
F-, the hardest base known, and soft acids, such as the low-
valent Fe center in Cp*Fe(dppe)+, should be weak. Soft acids
are expected to exhibit considerably higher affinity for the
heavier halide ions. However, this is not always true. For
instance, several authors have clearly demonstrated that the order
F- > Cl- > Br- > I- is characteristic of anion affinity for
cationic, coordinatively unsaturated Rh(I),34 Pd(II),35 and W(II)36
moieties in media of low polarity.
Figure 1. X-band ESR spectrum of Cp*Fe(dppe)F+PF6 (1+PF6
)
-
-
recorded on a 9:1 THF/pentane frozen solution (glass) at 77 K.
Chemical Oxidations: Synthesis of the 17-Electron Iron-
(III) Halide Complexes. Cyclic voltammetric investigations
(vide infra) revealed that 1, 3, and 4 all undergo reversible one-
electron oxidations to the corresponding 17-electron radical
cations. This is analogous to the behavior of 2, 5, and 6 that
has been previously reported.19 These oxidations occur at
potentials that suggest the use of Cp2Fe+ salts as chemical
oxidants. Accordingly, PF6- salts of the 17-electron complexes
Cp*Fe(dppe)X+ [X ) F (1+), Br (3+), I (4+)] were synthesized
(dppe)Cl+PF6-, is also believed to arise via a 15-electron
intermediate CpMo(dppe)Cl+.31
The structures of these three new 17-electron Fe(III) halides
were confirmed by elemental analysis, magnetic susceptibility
measurements, and an X-ray crystal structure determination of
-
1+PF6 (vide infra). The complexes 1+PF6-, 3+PF6-, and
4+PF6- exhibit magnetic moments µeff ) 1.9, 2.7, and 2.3 µB,
respectively, as determined in solution by Evans’ method.37
These values correspond reasonably well to a d5 low-spin
configuration. As already noted for related 17-electron Fe(III)
half-sandwich compounds, these magnetic moments are some-
what greater than the calculated spin-only value (1.73 µB) for
one unpaired electron, and this has been attributed to orbital
contributions.19a
by one-electron oxidations with 0.9 equiv of Cp2Fe+PF6 in
-
THF and were isolated as dark red (1+) or black (3+ and 4+)
microcrystals in 87, 95, and 90% yield, respectively. These air
and thermally stable compounds exhibit cyclic voltammetry
waves identical with those of their Fe(II) precursors. Accord-
ingly, the cations can be chemically reduced with 1 equiv of
Cp2Co to quantitatively regenerate the respective 18-electron
complexes.
-
A crystalline sample of the fluoro derivative 1+PF6 was
subjected to zero-field Mo¨ssbauer spectrometry at 80 K. The
spectrum exhibits a single Mo¨ssbauer doublet with typical
isomeric shift (IS) and quadrupole splitting (QS) parameters
diagnostic of a pure d5 low-spin Fe(III) species.19,24,25 In
addition, both IS and QS parameters can provide insight into
the nature of the bond to iron.38 Moreover, in such half-sandwich
17-electron Fe(III) complexes, it has been noted that the IS value
increases with the electron releasing power of the X group.39
The high IS value of 0.426 mm/s vs Fe recorded for
Originally, the novel Fe(III) fluoro derivative 1+PF6 was
-
unexpectedly isolated in 81% yield as analytically pure dark
red crystals from the reaction of the 16-electron cation
Cp*Fe(dppe)+PF6- with Cp2Fe+PF6- (Scheme 2). Based on the
redox potentials determined by cyclic voltammetry (vide infra),
the first step of the reaction is believed to be a one-electron
oxidation to generate the dicationic 15-electron intermediate
-
Cp*Fe(dppe)2+ that reacts via F- abstraction from PF6 to
provide the 17-electron fluoro complex 1+PF6-. PF5 was not
detected in this reaction but its formation was implicated by
the observed polymerization of the THF solvent. The abstraction
-
Cp*Fe(dppe)F+PF6 would therefore suggest that the fluoride
acts as an apparently strong σ-donor ligand. The IS value is
-
indeed greater than those measured for Cp*Fe(dppe)H+PF6
-
of fluoride from the PF6 anion in the above reaction is
-
(0.260 mm/s) and Cp*Fe(dppe)Me+PF6 (0.35 mm/s).19 It is
presumably a result of the potent Lewis acidity of the 5-coor-
dinated metal dication intermediate Cp*Fe(dppe)2+. The half-
sandwich complex CpMo(dppe)(MeCN)ClF+PF6-, resulting
from thermal decarbonylation of the 17-electron CpMo(CO)-
also noteworthy that the QS value of 0.915 mm/s is also large
compared to those of 5+ (0.84) and of 6+ (0.76).
The X-band ESR spectrum (Figure 1) of 1+PF6- was recorded
on a 9:1 THF/pentane frozen solution (glass) at 77 K. The
spectrum displays three well-separated signals corresponding
to the three g-tensor components, appropriate for a pseudooc-
tahedral environment around the Fe-centered radical. The
extrapolated values (gx ) 2.419, gy ) 2.018, gz ) 1.998) confirm
the metal-centered radical nature of the 17-electron cationic
species.19,24 The gy and gz values are close to the free-electron
g value (g ) 2.0023), whereas the gx component (gx ) 2.419)
is much larger, as usually noted for such monomeric Fe(III)
(30) Veltheer, J. E.; Burger, P.; Bergman, R. G. J. Am. Chem. Soc. 1995,
117, 12478.
(31) Fettinger, J. C.; Keogh, D. W.; Poli, R. J. Am. Chem. Soc. 1996,
118, 3617.
(32) Klopman, G. In Chemical ReactiVity and Reaction Paths; Klopman,
G., Ed.; Wiley: New York, 1974; p 55.
(33) Pearson, R. G. In Hard and Soft Acids and Bases; Dowden,
Hutchison and Ross: Stroudsburg, PA, 1973.
(34) (a) Branan, D. M.; Hoffman, N. W.; McElroy, E. A.; Miller, N. C.;
Ramage, D. L.; Schott, A. F.; Young, S. H. Inorg. Chem. 1987, 26, 2915.
(b) Araghizadeh, F.; Branan, D. M.; Hoffman, N. W.; Jones, J. H.; McElroy,
E. A.; Miller, N. C.; Ramage, D. L.; Salazar, A. B.; Young, S. H. Inorg.
Chem. 1987, 26, 3752.
(35) Flemming, J. P.; Pilon, M. C.; Borbulevitch, O. Y.; Antipin, M.
Y.; Grushin, V. V. Inorg. Chim. Acta 1998, 280, 87.
(36) Bartlett, I. M.; Carlton, S.; Connelly, N. G.; Harding, D. J.; Hayward,
O. D.; Orpen, A. G.; Ray, C. D.; Rieger, P. H. Chem. Commun. 1999,
2403.
(37) (a) Evans, D. F. J. Chem. Soc. 1959, 2003. (b) Crawford, T. H.;
Swanson, J. J. Chem. Educ. 1971, 48, 382.
(38) Guillaume, V.; Thominot, P.; Coat, F.; Mari, A.; Lapinte, C. J.
Organomet. Chem. 1998, 565, 75.
(39) Paul, F.; Meyer, W. E.; Toupet, L.; Jiao, H.; Gladysz, J. A.; Lapinte,
C. J. Am. Chem. Soc. 2000, 122, 9405.

9988 J. Am. Chem. Soc., Vol. 123, No. 41, 2001
Tilset et al.
low-spin configuration compounds, indicating strong interactions
between the single electron and the electrons of lower lying
orbitals.40 However, the signal at high field is weaker than
usually observed.24 Indeed, the simulation of the spectrum
indicates that this signal corresponds to one line of the doublet
resulting from the coupling of the unpaired electron to the 19F
nucleus (AF ) 110 G). The second line of this doublet overlaps
with the gy signal, for which a smaller coupling constant AF )
60 G is estimated. These ESR data agree well with those
reported for the d3 tungsten derivative Tp*W(CO)(MeCCMe)-
F•+, where the three hyperfine AF coupling constants (A1 ) 108
G, A2 ) 22 G, A3 ) 18 G) have been estimated.36 Hyperfine
coupling with the phosphorus nuclei in 1+ was not observed.
-
Infrared spectra of 5 and 5+PF6 (Nujol mulls) revealed
absorptions that are assigned to the νFe-H stretching modes at
1869 and 1886 cm-1, respectively.18 Attempts were also made
at locating the IR νFe-X stretching vibrations for the halides.
Spectra were run on both a dispersive instrument (see Experi-
mental Section for details) in the region 600-200 cm-1 and an
FTIR instrument with a cutoff at 400 cm-1. Spectra were
recorded for the whole series of neutral and cationic complexes
except the unstable Cp*Fe(dppe)F complex. Absorptions that
were common to all complexes within the neutral series were
rejected on the grounds that they would likely originate from
the Cp*Fe(dppe) moiety, rather than the Fe-X bond vibration,
and similarly for the cationic series (which also displayed a
common band at 557 cm-1 for the PF6 counterion). Absorp-
-
tions that could then with some confidence be attributed to the
Fe-X stretching mode were found only for the Cp*Fe(dppe)Cl0/+
(20/+) couple; unfortunately, the other halides did not provide
useful data for our purpose. A band at 337 cm-1 assigned to
the Fe-Cl stretching mode of 2 exhibited a shift to 367 cm-1
in 2+.41 A very similar shift (from 349 to 373 cm-1) was
reported for the νFe-Cl modes of an Fe(diarsine)2Cl20/+ couple.42
The measured data for 2 and 2+ agree quite well with DFT
calculated values of 279 and 307 cm-1, respectively, for the
model compounds CpFe(dpe)Cl and CpFe(dpe)Cl+ (vide infra).
X-ray Crystal Structure Determinations of 1+PF6-, 2+PF6-,
-
and 6+BF4-. The crystal structures of compounds 1+PF6
,
2+PF6-, and 6+BF4 have been determined as outlined in the
Experimental Section. The molecular structures of the three
cationic organometallic moieties 1+, 2+, and 6+ are presented
in similar perspectives for the sake of comparison in Figure 2.
Relevant interatomic distances and angles for the three com-
plexes are listed in Table 2, together with key data19a for the
related neutral chloro compound. In each species, the iron atom
shows a pseudooctahedral coordination with the three-legged
piano-stool geometry, and the three compounds exhibit general
features that compare well with previous structures in this
mononuclear Fe(III) σ-bonded Cp*Fe(dppe)X+ series.24,43,44
-
Figure 2. ORTEP drawings of the cationic moieties of the X-ray
structures of 1+PF6- (top), 2+PF6- (center), and 6+BF4- (bottom). See
Table 2 for selected bond lengths and angles and Table 11 for crystal
data, data collection, and refinement parameters.
respect, the Fe-F distance deserves particular attention. The
Fe-F bond length in 1+ might be compared with the Fe-Cl
bond distance of 2.237(2) Å (Table 2) in the analogous
compound 2+PF6-. This Fe-Cl bond distance and the difference
Although some X-ray characterized mononuclear coordina-
tion45,46 and purely inorganic47 iron fluorides have been reported,
to the best of our knowledge, the crystal structure of the fluoro
compound 1+PF6- is the first one to be reported for a molecular
organometallic iron complex containing a fluoro ligand. In this
(45) Hexacoordinated high-spin Fe(III) fluorides of the type Fe(porphy-
rin)F have been structurally characterized. (a) Anzai, K.; Hatano, K.; Lee,
Y. J.; Scheidt, W. R. Inorg. Chem. 1981, 20, 2337. (b) Scheidt, W. R.;
Lee, Y. J.; Tamai, S.; Hatano, K. J. Am. Chem. Soc. 1983, 105, 778. (c)
Lee, S. C.; Holm, R. H. Inorg. Chem. 1993, 32, 4745.
(40) Rieger, P. H. Coord. Chem. ReV. 1994, 135-136, 203.
(41) IR bands observed at 394, 426, and 436 cm-1 for 2 also exhibited
shifts to 448, 475, and 483 cm-1, respectively, for 2+. These bands are
assigned to νFe-ligand modes (predominantly νFe-P), but we cannot rule out
some mixing with the νFe-Cl vibration.
(42) Lewis, J.; Nyholm, R. S.; Rodley, G. A. J. Chem. Soc. 1965, 1483.
(43) Roger, C.; Toupet, L.; Lapinte, C. J. Chem. Soc., Chem. Commun.
1988, 713.
-
(46) The difluoro iron(III) derivative trans-Fe(Im)4F2+BF4 (Im )
methylimidazole) has also been structurally characterized: Christie, S.;
Subramanian, S.; Wang, L.; Zaworotko, M. J. Inorg. Chem. 1993, 32, 5415.
(47) (a) Rother, G.; Worzala, H.; Bentrup, U. Z. Anorg. Allg. Chem. 1996,
622, 1991. (b) Fourquet, J. L.; Plet, F.; Calage, Y.; de Pape, R. J. Solid
State Chem. 1987, 69, 76. (c) Griebler, W.-D.; Babel, D. Liebigs Ann. Chem.
1980, 1549. (d) For a comparison with bond distances observed in various
structural types of Fe(III) fluorides, see: Leblanc, M.; Pannetier, J.; Ferey,
G.; de Pape, R. ReV. Chim. Min. 1985, 22, 107.
(44) Denis, R.; Toupet, L.; Paul, F.; Lapinte, C. Organometallics 2000,
19, 4240.

Fe-X Bonds in 17- and 18-Electron Cp*Fe(Dppe)X Complexes
Table 2. Selected Bond Lengths (Å) and Angles (deg) for
J. Am. Chem. Soc., Vol. 123, No. 41, 2001 9989
Complexes 1+PF6-, 2, 2+PF6-, and 6+BF4
-
1+
2a
2+
6+
Fe(l)-Cp(centroid)
Fe-X
1.766(4)
1.856(2)
1.768(4) 1.780(9) 1.786(7)
2.346(1) 2.237(2) 2.109(7)
Fe(1)-P(1)
2.3058(16) 2.197(1) 2.308(2) 2.275(2)
2.2899(14) 2.210(1) 2.276(2) 2.262(2)
Fe(l)-P(2)
Fe(l)-C(l)
2.146(4)
2.131(4)
2.125(4)
2.150(4)
2.148(4)
83.94(5)
87.99(9)
84.47(8)
2.101(4) 2.133(9) 2.130(7)
2.114(4) 2.160(9) 2.191(8)
2.128(4) 2.147(8) 2.188(7)
2.134(4) 2.148(9) 2.156(7)
2.082(4) 2.134(9) 2.126(7)
84.98(5) 81.75(8) 84.77(7)
86.03(4) 90.32(10) 91.76(17)
87.23(5) 91.07(9) 83.56(17)
Fe(1)-C(2)
Fe(1)-C(3)
Fe(1)-C(4)
Fe(1)-C(5)
P(1)-Fe(1)-P(2)
P(1)-Fe(1)-X
P(2)-Fe(1)-X
Cp(centroid)-Fe(1)-X 122.4(2)
120.0
119.9(3) 120.5(3)
a From ref 19a.
between the covalent radii48 of Cl and F (0.99-0.64 ) 0.35 Å)
suggest that the Fe-F covalent bond length should be around
1.89 Å, slightly longer than the experimental value of 1.856(2)
Å obtained for complex 1+PF6-. This difference between the
observed and expected values indicates an increase in the Fe-F
bond strength, suggesting that the apparent π contribution to
the bonding is stronger in the fluoro than in the chloro
compound. However, the Fe-F bond length is in agreement
with that measured for octahedral Fe(III) porphyrinato com-
plexes with Fe-F distances ranging from 1.792(3) to 1.966(2)
Figure 3. Cyclic voltammograms of Cp*Fe(dppe)+PF6-, Cp*Fe-
3-
Å,45,46 and for the hexafluorometalate FeF6 ions47a-c (Fe-F
(dppe)F+PF6-, and Cp*Fe(dppe)Cl in THF/0.2 M Bu4N+PF6 at T )
-
) 1.93 Å).
25 °C and a voltage sweep rate ν ) 1.0 V s-1 at a Pt disk electrode (d
) 0.4 mm).
For the chloro complexes, X-ray crystal structures have now
been determined for both the neutral19a and oxidized forms. The
similarities in bond distances and angles in the two compounds
(Table 2) suggest that the structural reorganization associated
with the electron transfer is rather weak.24,25e,44 Upon mono-
electronic oxidation, a shortening (0.107 Å) of the Fe-Cl bond
was observed with concomitant slight lengthening of the Fe-
Cp* centroid and the Fe-P distances of 0.012 and ca. 0.09 Å,
respectively. This is in accord with the computational findings,
and consistent with a decrease of the Fe(dπ)-Cl(pπ) electron
repulsion (see below). A similar trend was also noted for the
Cp*Fe(dppe) frameworks in the Fe(I/II) Cp*Fe(dppe)0/+ and
Fe(II/III) Cp*Fe(dppe)(CtC-1,4-C6H4NO2)0/+ pairs, for which
the crystal structures of both members of each couple have been
determined.25e,44 Oxidatively induced M-Cl bond shortenings
were also seen for Tp*W(CO)Cl(MeCCMe)0/+ (0.09 Å),36
the computational findings (see below and Table 5), the Fe-
C(alkyl) distances of these two 17-electron Fe(III) complexes
are shorter than the Fe-C(alkyl) bond length (2.154(4) Å)
measured for the neutral Fe(II) benzyl derivative CpFe(dppe)-
(CH2Ph).51 This contrasts with the small elongation (0.016 Å)
of the Fe-C(alkynyl) distance observed in the p-nitrophenyla-
lkynyl derivative Cp*Fe(dppe)(CtC-1,4-C6H4NO2) upon one-
electron oxidation.44
The synthesis and spectroscopic characterization of the other
Cp*Fe(dppe)Xn+ complexes discussed in this paper [X ) Cl
(2/2+), I (4), H (5/5+), Me (6/6+)] as well as the crystal
-
structures of 2 and 5+PF6 have been described previously.19
As a result of the present preparative work, both members of
each of the six redox couples under inVestigation haVe been
isolated in high yields and fully characterized by spectroscopic
methods and, in part, by X-ray crystal structure determinations.
II. Electrochemical Studies and Thermochemical Results.
0/+
0/+
ReCl(CNtBu)3(PCy3)2 (0.18 Å),49 and CpMoCl2(PMe3)2
(0.1 Å).50
The molecular structure of the methyl complex 6+BF4 is
quite reminiscent of that of its methoxymethyl congener
Cp*Fe(dppe)(CH2OMe)+PF6-, with most bond lengths and
angles falling into the same range.43 The most remarkable
structural difference is the shortening (0.106 Å) of the Fe-
C(alkyl) bond, with a concomitant lengthening (0.016 Å) of
the Fe-Cp* centroid distance when going from 6+ to the
methoxymethyl derivative. This lengthening/shortening interplay
is presumably due to steric constraints imposed by the two bulky
ancillary ligands when accommodating the alkyl group in the
coordination sphere of the metal. In agreement with what has
been observed for the Fe-Cl bond distances in 20/+ and with
-
Cyclic Voltammetry Studies. The electrochemical investigation
of the complexes was performed in THF/0.2 M Bu4N+PF6
.
-
The solvent THF offers good solubility of all species that were
investigated, and is rather inert (at least on the experimental
time scale) toward even the most reactive of the species,
including the Cp*Fe(dppe)•/+ couple. Possible consequences of
solvent effects and/or ion pairing with the supporting electrolyte
will be addressed later.
Figure 3 shows cyclic voltammograms for Cp*Fe(dppe)+PF6 ,
-
-
Cp*Fe(dppe)F, and Cp*Fe(dppe)Cl in THF/0.2 M Bu4N+PF6
(T ) 20 °C, ν ) 1.0 V/s, d ) 0.4 mm Pt disk electrode). As
can be seen, the 16-electron cation Cp*Fe(dppe)+ is reversibly
reduced to the 17-electron radical Cp*Fe(dppe)• and reversibly
(48) Pauling, L. The Nature of the Chemical Bond; Cornell University
Press: New York, 1960.
(49) Heinekey, D. M.; Voges, M. H.; Barnhart, D. M. J. Am. Chem.
Soc. 1996, 118, 10792.
(50) Krueger, S. T.; Poli, R.; Rheingold, A. L.; Staley, D. L. Inorg. Chem.
1989, 28, 4599.
(51) Hill, D. H.; Parvez, M. A.; Sen, A. J. Am. Chem. Soc. 1994, 116,
2889.

9990 J. Am. Chem. Soc., Vol. 123, No. 41, 2001
Tilset et al.
greater than 0.2 V, in the E°([Fe*]X0/+) value for X ) F relative
to Cl and the other halides apparently supports the idea that F
acts as an exceptionally good π donor. The fluoride stands out
for the cation/dication oxidations too, although the overall halide
trend is less clear-cut in this case. We will address the fluoride
effect in more detail in the theoretical part of this paper (vide
infra).
Table 3. Cyclic Voltammetry Data for the Oxidation of
Cp*Fe(dppe) Derivativesa
b
compound FeX
E1 (FeX/FeX+)b
E2 (FeX+/FeX2+
)
Cp*Fe(dppe)c
Cp*Fe(dppe)H
Cp*Fe(dppe)CH3
Cp*Fe(dppe)Fe
Cp*Fe(dppe)Cl
Cp*Fe(dppe)Br
Cp*Fe(dppe)I
-1.272
-0.747
-0.883
-0.824
-0.618
-0.582
-0.540
-0.290
0.75d
0.56d
0.688
0.823
0.811
0.780
Consequences of Oxidation on the Homolytic [Fe*]-X
Bond Energies. Equation 1 shows that an oxidatively induced
M-X bond weakening will result if the LnM-X compound is
oxidized at more positive electrode potentials than the radical
LnM•. This situation pertains to all compounds included in Table
3, i.e., an oxidatively induced bond weakening occurs in [Fe*]-
X not only for X ) H, but also for CH3 as well as all halides.
The bond energy changes, obtained from eq 1, are summarized
in Table 4. Bond energy changes pertaining to the oxidation of
Cp*Fe(dppe)X•+ to the dications Cp*Fe(dppe)X2+ can be
similarly estimated by replacing E°ox(M•) and E°ox(MX) with
E°ox(M+) and E°ox(MX+), respectively, in eq 1, and are also
included in Table 4. Since all [Fe*]-X+ are oxidized at more
positive potentials than [Fe*]+, the result of the second oxidation
process is a further bond weakening for all X. Thus, the data
demonstrate that for all X studied, [Fe*]-X bond energies
decrease in the order [Fe*]X > [Fe*]X•+ > [Fe*]X2+. For both
oxidation processes, there is a very interesting and obvious trend
in the bond activation for the halides. The oxidatively induced
bond weakening decreases in the order I > Br > Cl > F and is
particularly less pronounced for F than for the other halides. In
particular, the difference between F and the other halides is
greater than 30 kJ/mol for the overall two-electron oxidation,
which in principle corresponds to the generation of a vacant
coordination site. This quantity may be viewed as an extra
stabilization of the unsaturated 16-electron complex Cp*Fe-
(dppe)X2+ that is provided by F, relative to the other halides.
To the extent that this picture is valid and within the context of
the involvement of halide pπ to metal dπ donation, which is
quite easy to visualize (Figure 5), this phenomenon might be
attributed to a more efficient π donation from F to the metal.
The pπ-dπ interaction may then be visualized as repulsive for
an 18-electron metal center and relatively strongly bonding for
a 16-electron metal center. An intermediate situation arises for
a 17-electron center; this bonding picture is very similar to the
simple bonding picture that has been used to describe the
interaction between a 17-electron metal radical and a two-
electron donor ligand to form 19-electron species.52
a THF/0.2 M Bu4N+PF6-, T ) 20 °C, Pt disk electrode (d ) 0.4
mm), voltage sweep rate V ) 1.0 V/s. b Oxidation potential, V vs Cp2Fe/
Cp2Fe+. The voltammograms were reversible unless otherwise stated,
and potentials are taken as the midpoints between anodic and cathodic
peaks. Values are the average of 3 separate measurements and are
reproducible to within (5 mV. c Measurements were done on
Cp*Fe(dppe)+PF6- which, contrary to Cp*Fe(dppe)•, is stable in THF
at room temperature. d Peak potential for the irreversible process.
e Measurements were performed on Cp*Fe(dppe)F•+PF6
.
-
Figure 4. Electrode potential data for the Cp*Fe(dppe)X0/+ and
Cp*Fe(dppe)X+/2+ couples (X ) H, Me, F, Cl, Br, I) in THF/0.2 M
Bu4N+PF6
.
-
oxidized to the 15-electron dication radical Cp*Fe(dppe)2+, as
described previously.25e The cyclic voltammograms of Cp*Fe-
(dppe)F and Cp*Fe(dppe)Cl exhibit reversible oxidations to the
17-electron radical cations as well as to the 16-electron dications.
It is immediately apparent from the figure that the fluoro
complex is easier to oxidize than the chloro complex. Table 3
summarizes the electrode potential data obtained by cyclic
voltammetry for the [Fe*]X0/+ and [Fe*]X+/2+ couples for X
) H, CH3, F, Cl, Br, and I. All neutral/monocation redox
processes in Table 3 were chemically reversible, near-Nernstian
processes (∆Ep ) 65-85 mV). Remarkably, except for X ) H
and CH3, the cation/dication couples were also reversible. Steric
and electronic protection of the metal center in the electron-
rich, sterically demanding Cp*Fe(dppe) moiety must be impor-
tant for the stabilization of the various oxidation states.
To the best of our knowledge, the electrochemical data for
these halides constitute the first report of reVersible oxidation
potentials for a complete organometallic LnM-X series (X )
F, Cl, Br, I), regardless of the identity of the LnM fragment.
(After our initial communication,26 other series have been
reported with trends that fully agree with our results.28b,36) The
reversible oxidation to monocation occurs most readily for the
most electronegative halide and becomes progressively more
difficult in the series F < Cl < Br < I, as visualized in Figure
4. This trend is the opposite of that predicted on the basis of
halide electronegativities alone. The particularly large jump,
Consequences of Oxidation on the Heterolytic [Fe*]-X
Bond Energies. Relative heterolytic bond dissociation energies
(to be denoted BDEhet) between two oxidation states may be
estimated from eq 2, which is derived from the thermochemical
cycle in Scheme 3. The data obtained will be Gibbs free energy
based since an assumption regarding canceling solvation
contributions will not be valid when different charges apply to
the two redox couples that are involved.
∆BDEhet ) BDEhet(MX•+) - BDEhet(MX) )
F[E°ox(M+) - E°ox(MX)] (2)
Equation 2 shows that an oxidatively induced heterolytic
M-X bond strengthening will result if the LnM-X compound
is oxidized at more negative electrode potentials than the 16-
(52) (a) Therien, M. J.; Trogler, W. C. J. Am. Chem. Soc. 1988, 110,
4942. (b) Trogler, W. C. In Metal-Ligand Interactions: From Atoms, to
Clusters, to Surfaces; Salahub, D. R., Russo, N., Eds.; Kluwer Academic
Publishers: Dordrecht, The Netherlands, 1992; p 287. (c) Tyler, D. R. Acc.
Chem. Res. 1991, 24, 325.

Fe-X Bonds in 17- and 18-Electron Cp*Fe(Dppe)X Complexes
J. Am. Chem. Soc., Vol. 123, No. 41, 2001 9991
Table 4. Relative Homolytic and Heterolytic Bond Dissociation Energies for Cp*Fe(dppe)Xn+ Complexes (kJ/mol)a
compound
∆BDEhom(MX+-MX)
∆BDEhom(MX2+-MX+)
∆BDEhom(MX2+-MX)
∆BDEhet(MX+-MX)
Cp*Fe(dppe)F (1)
Cp*Fe(dppe)Cl (2)
Cp*Fe(dppe)Br (3)
Cp*Fe(dppe)I (4)
Cp*Fe(dppe)H (5)
Cp*Fe(dppe)CH3 (6)
-43
-63
-67
-71
-51
-38
-4
-107
-106
-103
-100b
-82b
-138
-171
-173
-174
-151b
-120b
52
32
28
24
44
57
a Obtained using the data in Table 3 and eqs 1 and 3. Note that negative values signify a bond weakening and positive values a bond strengthening
upon oxidation. b Errors caused by unknown kinetic potential shift arising from irreversible electrode process are not taken into account.
solvent (0.1 M supporting electrolyte) combinations were
checked: MeCN (Bu4N+PF6 , Bu4N+BArf-); THF (Bu4N+PF6 ,
-
-
Bu4N+BF4 , Bu4N+BArf-); CH2Cl2 (Bu4N+PF6 , Bu4N+BArf-);
C6H5CF3 (Bu4N+BF4-, Bu4N+BArf-); and C6H5F (Bu4N+BF4
-
-
-
,
Bu4N+BArf-).53 The Cp*Fe(dppe)H0/+ and Cp*2Fe0/+ couples
were investigated in all these combinations, whereas the
Cp*Fe(dppe)0/+ couple was checked in the THF, C6H5CF3, and
C6H5F media only due to the limited stability of Cp*Fe(dppe)+
in MeCN and CH2Cl2. It was found that E°([Fe*]H0/+) vs
E°(Cp2Fe0/+) varied over a 0.18 V range in these media.
However, E°([Fe*]H0/+) varied over only a 0.04 V range when
referenced against E°(Cp*2Fe0/+) in the same solvent/electrolyte
systems, demonstrating that the variations for the iron hydride
relative to Cp2Fe are primarily due to medium effects on the
Cp2Fe0/+ couple, rather than on the hydride complex. (Ruiz and
Astruc54 have recently advocated the use of Cp*2Fe as an
electrochemistry standard instead of Cp2Fe because the former
is much less subject to solvent and counterion effects.) Thus,
no significant specific interactions between the solvent or
counterion and the iron hydride redox couple appeared to
complicate the electrode potential determinations. On the other
hand, E°(Cp*Fe(dppe)0/+) variations constituted less than 0.17
V, when referenced against Cp*2Fe0/+. These variations, in part
caused by less than ideal CV response for this couple, are greater
than those for Cp*Fe(dppe)H0/+, but still small enough that we
conclude that no strong specific interactions with the counter-
anion or solvent molecules exist for the Cp*Fe(dppe)0/+ couple.
Figure 5. Simplified MO diagram depicting the consequences of
M(dπ)-X(pπ) interactions in 18-, 17-, and 16-electron complexes.
Scheme 3
electron cation LnM+, which is in the same formal oxidation
state but bears a formal positive charge. This situation pertains
to all compounds included in Table 3. (The consequences of
solvent and ion pairing effects, expected to be particularly strong
for the MX+/2+ couple, will be addressed later.) The ∆BDEhet
data that are calculated from eq 2 are included in Table 4. The
data show that all bonds are strengthened in the heterolytic sense
as a consequence of the oxidation. It is not too surprising that
a bond strengthening is observedsafter all, the heterolytic
cleavage of [Fe*]X involves the separation of singly charged
cations and anions whereas the cleavage of [Fe*]X+ involves
the separation of a singly charged anion from a doubly charged
cation. For the halide series, the bond strengthening decreases
with increasing size of X in the order F > Cl > Br > I, in
accord with important contributions from electrostatic effects.
Consequences of Solvent and Ion Pairing Effects. Poli et
al. have rightfully pointed out that great care should be exerted
when thermochemical cycles are used to estimate M-H BDEhom
variations.21 It was suggested that the 16-electron cation M+
(and to a lesser and variable extent also the 17-electron species
M• and MH•+) is highly likely to establish strong interactions
with the polar solvent or with the supporting electrolyte used
for the electrochemical measurements. If this interaction with
the medium is stronger for M+ than for MH•+, the result will
be that M• is more readily oxidized compared to MH•+ than
what would be the case in the absence of interactions with the
medium. The result would be a skewing of calculated BDEhom
changes toward a bond weakening following oxidation.
Combining the data for Cp*Fe(dppe)H0/+ and Cp*Fe(dppe)0/+
,
the difference between their E° values showed a 0.13 V variation
(translating to ca. 13 kJ/mol difference in ∆BDEhom data). We
conclude that the electrode potentials are relatively uncompli-
cated by medium effects and that the oxidatively induced
BDEhom weakening effects that are calculated for the 17-electron
cations are undoubtedly real.
Ion pairing is, however, bound to influence the data for the
dications much more strongly. For example, in THF/0.1 M
Bu4N+BArf-, the Cp*Fe(dppe)+/2+ couple exhibited a very
broad and chemically irreVersible wave, contrasting the revers-
ibility of the process in THF/0.1 M Bu4N+PF6-. Thus, the
apparent stability of the 15-electron dication Cp*Fe(dppe)2+ on
the CV time scale appears to be due to electrostatic attraction
and stabilization by the (relatively speaking) small counteranion
PF6-. Ironically, it is the very same counterion that causes its
demise in the fluoride abstraction reaction that gives 1+. There
(53) For the merits of using (C6F5)4B- or BArf- as the anion in the
supporting electrolyte, see: (a) Hill, M. G.; Lamanna, W. M.; Mann, K. R.
Inorg. Chem. 1991, 30, 4687. (b) LeSuer, R. J.; Geiger, W. E. Angew. Chem.,
Int. Ed. 2000, 39, 248. For the use of C6H5CF3 as a solvent for CV
investigations of organometallic species, see: (c) Ohrenberg, C.; Geiger,
W. E. Inorg. Chem. 2000, 39, 2948. Whereas C6H5F appears unexplored
as a solvent for CV investigations of organometallics, its use has been
recently described for the synthesis of unsaturated Ru complexes: (d)
Tenorio, M. J.; Mereiter, K.; Puerta, M. C.; Valerga, P. J. Am. Chem. Soc.
2000, 122, 11230.
To probe the possible involvement of solvent and ion pairing
effects, we performed a thorough study in which solvent and
supporting electrolyte counterions were varied. The following
(54) Ruiz, J.; Astruc, D. C. R. Acad. Sci., Ser. IIc: Chim. 1998, 1, 21.

9992 J. Am. Chem. Soc., Vol. 123, No. 41, 2001
Tilset et al.
Table 5. Selected Optimized Metrical Data of the Models CpFe(dpe)X0/+ (X ) F, Cl, Br, I, H, CH3) with Available X-ray Data of Related
Cp*Fe(dppe)X0/+ Complexes in Italics
compound [Fe] ) CpFe(dpe)
Fe-X (Å)
Fe-P (Å)
Fe-C(Cp) (Å) av (range)
P-Fe-X (deg)
‘Cp’-Fe-Xa (deg)
[Fe]F
1.927
1.816
1.856
2.343
2.346
2.208
2.239
2.503
2.370
2.739
2.584
1.515
1.519
1.513
1.510
1.55
2.18 2.17
2.26 2.25
2.31 2.29
2.18 2.17
2.21 2.20
2.26 2.25
2.31 2.27
2.19 2.18
2.26 2.25
2.18 2.18
2.26 2.25
2.15 2.15
2.15 2.15
2.23 2.22
2.25 2.20
2.21 2.20
2.16 2.15
2.25 2.24
2.27 2.26
2.11 (2.08-2.13)
2.16 (2.13-2.20)
2.14 (2.13-2.15)
2.11 (2.09-2.13)
2.11 (2.08-2.13)
2.16 (2.14-2.19)
2.14 (2.12-2.16)
2.11 (2.09-2.13)
2.16 (2.13-2.18)
2.12 (2.09-2.13)
2.16 (2.14-2.18)
2.12 (2.11-2.14)
2.13 (2.11-2.15)
2.14 (2.12-2.16)
2.16 (2.12-2.21)
2.11 (2.08-2.15)
2.13 (2.12-2.14)
2.18 (2.12-2.25)
2.16 (2.12-2.20)
90 86
91 89
88 84
89 87
87 86
93 91
90 90
89 87
93 91
89 91
92 95
87 86
82 81
75 75
100 75
77 69
91 90
93 92
91 84
124
127
122
124
120
126
119
124
126
121
126
121
120
112
121
120
122
125
121
[Fe]F+
[Fe]Cl
[Fe]Cl+
[Fe]Br
[Fe]Br+
[Fe]I
[Fe]I+
[Fe]Hb
HA
HB
[Fe]H+ b
HB+
HC+
[Fe]CH3
[Fe]CH3
2.075
2.022
2.105
+
a ‘Cp’ ) C5H5 centroid. b HA and HB, and HA+ and HC+ are the DFT-optimized geometries obtained for CpFe(dpe)H and CpFe(dpe)H+,
respectively (see text).
may also be significant ion-pairing interactions that influence
the electrode potential data for the [Fe*]X+/2+ electrode
processes. This will affect the ∆BDE data involving the
dications, as well as the heterolytic BDE data for the monoca-
tions, but not the ∆BDE data that involve the monocation vs
neutral complexes.
Calculational Studies
The electrochemical experiments described above allowed the
experimental determination of relative BDE values. A better
Figure 6. Optimized geometries of CpFe(dpe)Cl0/+ (distances are given
in Å). The X-ray distances of Cp*Fe(dppe)Cl0/+ are given in paren-
theses.
insight into the factors influencing the variation of these relative
BDE values within the studied series of compounds can be
obtained by determination of absolute BDE values. Moreover,
the knowledge of absolute BDE values allows a more complete
bonding analysis of each individual complex. This can be done
theoretically at a reasonable level of accuracy. This is why we
have carried out density functional (DFT) calculations on model
complexes of the [Fe*]X0/+ series. Because of complications
arising from large solvent/electrolyte effects in solution (which
cannot be modeled) and from their open-shell configuration,
the 16-electron [Fe*]X2+ complexes were not investigated
theoretically. Details of the calculations are given in the
Experimental Section.
Optimized Geometries. All the optimized geometries of the
models CpFe(dpe)X0/+ (X ) F, Cl, Br, I, CH3) are very close
to the Cs symmetry and adopt the pseudooctahedral three-legged
piano-stool structure. The somewhat peculiar case of X ) H
will be discussed later. Selected metrical parameters for the
models CpFe(dpe)X0/+ (X ) F, Cl, Br, I, H, CH3) optimized
by DFT calculations are given in Table 5, together with available
X-ray experimental data of their Cp*Fe(dppe)X0/+ relatives for
comparison. A good agreement is obtained between the
optimized geometries and the X-ray molecular structures. The
optimized Fe-C and Fe-P distances tend to be slightly shorter
than the experimental ones. It has been shown in previous
calculations55 that this shortening (about 0.01 and 0.05 Å for
Fe-C and Fe-P, respectively) is a consequence of the replace-
ment of Cp* by Cp and dppe by dpe in the models. The Fe-X
bond lengths are also in good agreement with the available
experimental data (deviation 0.01-0.04 Å) in the cases of X )
H, F, Cl. A larger deviation is found for X ) CH3 (0.08 Å).
The P-Fe-X and Cp(centroid)-Fe-X angles differ only by
1-8°, values barely significant at our level of theory and
modeling. In the series of the neutral 18-electron species, the
average Fe-P and Fe-C separations are roughly constant within
the halogen series. They increase in the order H ∼ CH3 <
halogens. This order is somewhat different for the series of the
17-electron cations for which the average Fe-P order is H <
CH3 ∼ halogens and the average Fe-C order is H < F ∼ Cl <
Br ∼ I ∼ CH3. The oxidation of CpFe(dpe)X leads to Fe-P
and Fe-C bond lengthening, by 0.07-0.09 and 0.03-0.05 Å,
respectively. On the other hand, the Fe-X bond distances are
somewhat shortened, by approximately 0.2% (X ) H), 3% (X
) CH3), and 5-6% (X ) halogens). Therefore, the oxidation
of the neutral species has the same structural effect on all the
complexes. The X-ray structures of both the neutral and
monocationic species are known in the case of X ) Cl as
discussed earlier, and provide a good test for probing the
agreement between theory and experiment (see Table 5 and
Figure 6).
We discuss now the X ) H case in more detail. Unlike for
the X * H species, geometry optimization of CpFe(dpe)H lead
to two minima (see Figure 7), both of them characterized by
frequency calculations. One of them, denoted HA, has a
symmetry close to Cs and adopts a three-legged piano-stool
conformation comparable to those obtained for the X * H
species. In the other one, named HB, the P-Fe-P plane is much
(55) Costuas, K.; Saillard, J.-Y. Organometallics 1999, 18, 2505.

Fe-X Bonds in 17- and 18-Electron Cp*Fe(Dppe)X Complexes
J. Am. Chem. Soc., Vol. 123, No. 41, 2001 9993
three orbitals are usually referred to as the “t2g” set.57 They have
a dominant Fe character and are nonbonding with respect to
the σ-type interactions. Their major character is sketched in
Scheme 5. The particular case of X ) H will be discussed in
more detail later. Analysis of the electronic structure of the 17-
electron cationic species indicates that the oxidation always
involves the HOMO.
The computed ionization potentials follow the order I ∼ Br
> Cl > H > F > CH3. They are consistent with the energy
order of the HOMO’s. It has been shown that DFT-computed
ionization potentials within a homogeneous series of transition-
metal complexes correlate linearly with the corresponding
experimental redox potentials.58 Figure 8 shows the calculated
first ionization potentials (in eV) of the CpFe(dpe)X complexes
plotted against the first oxidation potentials of the Cp*Fe(dppe)X
series (in mV). A nice linear correlation is obtained for the X
* H series (least-squares correlation factor ) 0.99; slope )
0.96), providing confidence in the consistency of the theoretical
and electrochemical approaches.
In the case of X ) H, two squares are found in the diagram.
They correspond to the oxidation of the HA and HB conformers,
the oxidized species being HB+ in both cases. When the
considered oxidized species is HC+, the computed ionization
potentials of HA and HB are 6.31 and 6.40 eV, respectively,
i.e., farther from the least-squares curve. It is noteworthy that
the ionization potential of the computed less stable conformer
HA lies closer to the correlation line than the more stable HB
conformer. Obviously, the hydride species behave differently
compared to the other members of the studied series.
Figure 7. Optimized geometries of CpFe(dpe)H0/+ (distances are given
in Å). The X-ray distances of Cp*Fe(dppe)D+ are given in parentheses.
Erel ) computed relative energy.
Scheme 4
Bond Dissociation Energies. The computed homolytic
(BDEhom) and heterolytic (BDEhet) Fe-X bond energies of the
[Fe]X0/+ models are given in Tables 7 and 8, respectively.
Values calculated with and without BSSE corrections are
reported. The effect of the BSSE corrections does not change
the BDE order within the series, except for X ) H.
closer to being perpendicular to the Cp plane (Figure 7). HB is
more stable than HA by 0.10 eV. The exploration of the potential
energy surface around HA and HB indicated that this surface
is rather flat. The search for other minima for CpFe(dpe)H was
unsuccessful. Since the X-ray structure of Cp*Fe(dppe)H is not
available so far, there are no experimental data to support the
theoretical results for this neutral species.
In the case of a homolytic dissociation, the BDEhom order is
computed as F > Cl > Br > H > I > CH3, for both the neutral
and the cationic series. In agreement with the electrochemical
data, all the BDEhom values of the 17-electron species are lower
than those of their 18-electron parents. However, the theoretical
∆(BDEhom) values differ significantly from the experimental
ones. This difference probably originates mainly from the fact
that the former correspond to isolated molecules, while the latter
refer to molecules in solution. It is likely that interactions with
the solvent and electrolyte molecules constitute the major effect
responsible for this discrepancy between theory and experiment.
Nevertheless, there is a nice linear correlation between both
series of data, as exemplified by the plot in Figure 9a (least-
squares correlation factor ) 0.99; slope ) 0.71). The least-
squares fit does not include the X ) H case which, like in Figure
8, makes an exception. Being significantly different from 1, the
slope of the curve depends also on solvation energies which
cannot be considered as being constant within the series. The
Y-intercept of the least-squares curve is -15.7 kJ/mol. This value
corresponds to all the intermolecular energy terms which are
not considered in the calculations and which can be considered
constant within the series. Deviations in absolute numbers will
also arise from the fact that comparisons are made between
experimentally derived enthalpy data in solution at room
Two minima, characterized by frequency calculations, were
also found for CpFe(dpe)H+ (Figure 7). The more stable by
0.13 eV (HB+) has a geometry related to that of HB in which
the P-Fe-P and Cp planes are almost perfectly perpendicular.
The other isomer (HC+) has a very unsymmetrical structure that
is different from both HA and HB. One can idealize the
geometries of HB+ and HC+ by considering that they are the
two possible positional isomers of a four-legged piano-stool
CpFe(dpe)H+ complex in which one leg is occupied by a vacant
site, as sketched in Scheme 4. These types of distortions away
from the pseudooctahedral three-legged piano-stool have been
recently described by Poli and co-workers in the case of 16-
electron molybdenum complexes.56 It turns out that the X-ray
structure19b of Cp*Fe(dppe)D+ is close to HB+ (Figure 7). The
agreement between the experimental and calculated bond angles
is particularly good (see also Table 5).
Ionization Potentials. The first diabatic ionization potentials
of the CpFe(dpe)X models are reported in Table 6, together
with the energies and localization of the LUMO and of the three
highest occupied MO’s of these 18-electron complexes. For the
pseudooctahedral three-legged piano-stool complexes, these
(57) Albright, T. A.; Burdett, J. K.; Wangbo, M.-H. Orbital Interactions
in Chemistry; John Wiley: New York, 1985.
(58) (a) Bruce, M. I.; Low, P. J.; Costuas, K.; Halet, J.-F.; Best, S. P.;
Heath, G. A. J. Am. Chem. Soc. 2000, 122, 1949. (b) Ogliaro, F.; Halet,
J.-F.; Astruc, D.; Saillard, J.-F. New J. Chem. 2000, 24, 257.
(56) Cacelli, I.; Poli, R.; Quadrelli, E. A.; Rizzo, A.; Smith, K. M. Inorg.
Chem. 2000, 39, 517.

9994 J. Am. Chem. Soc., Vol. 123, No. 41, 2001
Tilset et al.
Table 6. Energy and Localization of the Frontier Orbitals and Diabatic First Ionization Potential (IP) of CpFe(dpe)X (X ) H, CH3, F, C1, Br,
I)
compound [Fe] ) CpFe(dpe)
[Fe]F
[Fe]Cl
[Fe]Br
[Fe]I
[Fe]Ha HA/HB
[Fe]CH3
HOMO-2
% Fe - % X
E (eV)
% Fe - % X
E (eV)
% Fe - % X
E (eV)
% Fe - % X
E (eV)
73-6
69-12
-5.05
67-25
-4.67
64-22
-4.25
57-15
-2.42
69-14
-5.05
65-29
-4.72
63-24
-4.32
57-16
-2.52
67-14
-5.09
46-49
-4.77
55-53
-4.39
54-18
-2.68
84-0/85-0
-4.89/-4.84
73-1/85-0
-4.29/-4.20
83-0/74-1
-4.21/-4.17
60-0/50-0
-1.77/-1.74
81-2
-4.75
77-3
-4.33
82-3
-3.97
55-0
-1.83
-4.96
72-19
-4.43
67-17
-4.00
60-11
-2.14
HOMO-1
HOMO
LUMO
IPb (eV)
6.11
6.31
6.38
6.37
6.18/6.27
6.06
a The two sets of values correspond to the 2 DFT optimized structures of CpFe(dpe)H, HA, and HB (see Figure 7 for label). b IP calculated
considering HB+ as the oxidized species.
Scheme 5
present in the other complexes, since the Fe-H interaction is
of pure σ-type. Moreover, unlike when X * H, two conformers
were found for [Fe]H (HA and HB) and for [Fe]H+ (HB+ and
HC+). HA adopts the three-legged piano-stool geometry, while
HB0/+ and HC+ are related to the four-legged piano-stool
conformation (see Figure 7 and Scheme 4). These latter
conformations can be described as resulting from distortions
away from the pseudooctahedral three-legged piano-stool ge-
ometry. Such distortions appear not to be allowed when X is a
π donor. It is noteworthy that the optimization of the [Fe]H2+
dication in its singlet state led to a geometry of type HC2+ that
was more stable than HB2+ by 0.21 eV. In this case, the ligand
vacancy of Scheme 4 is a real coordination site. Owing to the
rather small computed energy differences between the different
conformers of [Fe]H and [Fe]H+, and on the differences between
the [Fe] model and the [Fe*] “real” systems, it is difficult to
make firm predictions on which of the conformations of [Fe*]H
and [Fe*]H+ is predominant in solution during the electro-
chemical experiments. From the curves of Figures 8 and 9, one
may tentatively suggest that the electroactive species correspond
to conformations HA and HB+.
The shortest Fe-P bond distances in the neutral [Fe]X series
correspond to X ) H. This is somewhat counterintuitive since
the σ and π covalent bonding interactions between Fe and the
other ligands are expected to be weaker when competing with
the strong σ-donor and non-π-donor hydride ligand, as compared
to the weak σ-donor and π-donor fluoride ligand, for example.
However, it has been shown recently that there is not always a
strong correlation between the π-donor ability of a ligand and
the distances between the metal and the other ligand.59 As a
matter of fact, the Fe-C(Cp) distances do not vary very much
in the [Fe]X series.
Figure 8. Calculated ionization potentials of [Fe]X models plotted
against the first oxidation potentials of the [Fe*]X series, X ) H, CH3,
F, Cl, Br, I. For X ) H, the square spots correspond to the ionization
potentials of HA and HB, the oxidized species being HB+ (cf. Figure
7).
temperature and theoretically derived energy data for a single
molecule at 0 K.
In the case of a heterolytic dissociation, the BDEhet order is
computed as H > CH3 > F > Cl > Br > I, for both the neutral
and the cationic series. In agreement with the electrochemical
data, the BDEhet values of the 17-electron species are all greater
than those of their 18-electron parents. The theoretical ∆BDEhet
values differ from the experimental ones to a larger extent than
in the case of a homolytic dissociation. This is not surprising,
owing to the larger solvent/electrolyte effect expected from the
more ionic species involved in the heterolytic cycle. Neverthe-
less, there is a nice linear correlation between the experimental
and calculated values, as shown in Figure 9b (least-squares
correlation factor ) 0.99; slope ) 0.94). Again, the X ) H
case is not included in the least-squares fit. The Y-intercept of
the least-squares curve is -386.9 kJ/mol. This value is much
larger than that corresponding to the homolytic dissociation and
is indicative of strong intermolecular interactions.
The one-electron oxidation of [Fe]H corresponds to the partial
depopulation of the highest component of the “t2g” block (see
above). One has to keep in mind that the metal-ligand π-type
interactions of complexes such as CpFe(dpe)X will affect
primarily the “t2g” orbitals,57 i.e. the metallic MO involved in
the oxidation process has mixed in a bonding way with the
π-accepting frontier orbitals of the phosphine and the cyclo-
pentadienyl ligands. Therefore, the oxidation of CpFe(dpe)H
corresponds to a loss of Fe-P and Fe-C π bonding, the
consequence of which is bond elongation (see Table 5). On the
other hand, there is no simple orbital argument to explain the
slight shortening (if considered significant) of the Fe-H bond
when going from CpFe(dpe)H to CpFe(dpe)H+. It may be
attributed to some contraction of the metal atomic radius when
going from Fe(II) to Fe(III). Such an ionic radius effect has
Analysis of the Theoretical Data. We begin the discussion
by analyzing the X ) H case, which provides a good reference
starting point to understand the Fe-X π-type effects which are
(59) Heyn, R. R.; MacGregor, S. A.; Nadasdi, T. T.; Ogasawara, M.;
Eisenstein, O.; Caulton, K. G. Inorg. Chim. Acta 1997, 259, 5.

Fe-X Bonds in 17- and 18-Electron Cp*Fe(Dppe)X Complexes
J. Am. Chem. Soc., Vol. 123, No. 41, 2001 9995
Table 7. Calculated Homolytic [Fe]-X BDEhom for the CpFe(dpe)X0/+ Seriesa
[Fe]H
[Fe]H
[Fe]H
[Fe]H
[Fe]F
[Fe]Cl [Fe]Br
[Fe]I
2.67
2.49
2.21
2.15
HA/HB+ HA/HC+ HB/HB+ HB/HC+ [Fe]CH3
BDEhom (MX) (eV)
BDEhom (MX+) (eV)
without BSSE
with BSSE
without BSSE
with BSSE
4.58
3.81
4.38
3.48
3.15
3.08
2.87
3.15
2.90
2.70
2.56
2.95
1.98
2.68
1.72
2.95
1.98
2.57
1.61
3.04
2.05
2.68
1.72
3.04
2.05
2.57
1.61
2.00
1.22
1.85
1.12
3.65
∆BDEhom (MX+/MX) (kJ/mol) with BSSE
-15.4
-27.7
-32.8
-32.8
-24.5
-35.7
-31.8
-42.4
-9.6
a ∆BDEhom ) BDEhom(MX+) - BDEhom(MX). Negative values indicate a bond weakening.
Table 8. Calculated Heterolytic [Fe]-X BDEhet for the CpFe(dpe)X0/+ Seriesa
[Fe]H
[Fe]H
[Fe]H
[Fe]H
[Fe]F
[Fe]Cl [Fe]Br
[Fe]I
HA/HB+ HA/HC+ HB/HB+ HB/HC+ [Fe]CH3
BDEhet (MX) (eV)
BDEhet (MX+) (eV)
without BSSE
with BSSE
without BSSE
with BSSE
6.78
6.65
11.96
11.16
434.9
5.70
5.78
10.68
10.11
417.0
5.48
5.58
10.40
9.85
5.15
5.30
10.08
9.57
8.15
7.90
13.15
12.45
441.1
8.15
7.90
13.04
12.39
452.2
8.06
7.99
13.15
12.45
430.5
8.06
7.99
13.04
12.39
441.6
7.96
7.74
13.19
12.33
442.4
∆BDEhet (MX+/MX) (kJ/mol) with BSSE
411.9
411.2
a ∆BDEhet ) BDEhet(MX+) - BDEhet(MX). Positive values indicate a bond strengthening.
the νFe-H stretching frequency upon oxidation. The values
computed for HA, HB, HB+, and HC+ are 1892, 1876, 1907,
and 1921 cm-1, respectively. These values agree well with the
experimental data recorded at ambient temperature in Nujol
mulls, 1869 and 1886 cm-1 for [Fe*]H and [Fe*]H+, respec-
tively.
Similarly to X ) H, when X ) CH3 or halogen, the
π-accepting properties of the phosphine and cyclopentadienyl
ligands induce a lengthening of the Fe-P and Fe-C bonds upon
oxidation of CpFe(dpe)X. However, the Fe-X bond is shortened
to a much larger extent than in the X ) H case, especially when
X ) halogen. Unlike hydrogen, these ligands have π-donor
properties which cause the “t2g” set to mix in an antibonding
way with their low-lying π-type frontier orbitals (Figure 5). The
removal of one electron from this set lowers the Fe-X π-type
antibonding character, and is consequently expected to contribute
to the shortening of this bond. Surprisingly, as for X ) H, this
shortening is associated with a lowering of the corresponding
BDE (Table 7). The νFe-X stretching frequency was computed
in the case of X ) Cl, for both the neutral and cationic
complexes. As in the X ) H case, an increase in the νFe-X
stretching frequency upon oxidation is found (from 279 cm-1
to 307 cm-1), as was also observed experimentally (vide supra).
The ionization energies of the [Fe]X series (and the oxidation
potentials of the [Fe*]X series) are ordered consistently ac-
cording to their HOMO energies, with the highest HOMO
corresponding to X ) F and the lowest one corresponding to X
) I (see Tables 3 and 6). Thus, within the halogen derivatives,
it becomes increasingly difficult to oxidize [Fe]X when X
changes from the top through the halogen column of the periodic
table. As said above, this trend is unexpected from the point of
view of the HSAB concept,33 which predicts that the halide
π-donor ability decreases in the series I > Br > Cl > F. In
other words, the heavier X is, the closer in energy are the M(dπ)
and X(pπ) orbitals in Figure 5, and consequently the stronger
is their expected interaction and the higher is the antibonding
HOMO. In favoring a larger overlap, the more diffuse valence
AOs of the heavier halogens should favor the same trend, at
least in the case of F, Cl, and Br. The so-called inverse halide
order observed in the title complexes and reproduced in the
calculations has already been noted in the literature.28b,36,50,61
Figure 9. Calculated ∆BDE ([Fe]X/[Fe]X+) plotted against the
experimental ∆BDE ([Fe*]X/[Fe*]X+), X ) H, CH3, F, Cl, Br, I: (a)
homolytic dissociation and (b) heterolytic dissociation. For X ) H,
the square spots correspond to ∆BDE(HA/HB+) and ∆BDE(HB/HB+)
(cf. Figure 7).
already been noticed in other systems.50,60 The change appears
not to be associated with a bond strengthening, since oxidation
lowers the BDE (Table 7). Another result that is counterintuitive
when confronted to the Fe-H BDE values is the increase in
(61) (a) Zietlow, T. C.; Hopkins, M. D.; Gray, H. B. J. Am. Chem. Soc.
1986, 108, 8266. (b) Hascall, T.; Rabinovich, D.; Murphy, V. J.; Beachy,
M. D.; Friesner, R. A.; Parkin, G. J. Am. Chem. Soc. 1999, 121, 11402. (c)
Krueger, S. T.; Owens, B. E.; Poli, R. Inorg. Chem. 1990, 29, 2001.
(60) (a) Fettinger, J. C.; Kraatz, H.-B.; Poli, R.; Quadrelli, E. A. J. Chem.
Soc., Dalton Trans. 1999, 497. (b) See ref 21 and references therein.

9996 J. Am. Chem. Soc., Vol. 123, No. 41, 2001
Tilset et al.
Table 9. Energy Decomposition of Fe-X BDEhom* for the
Table 6 shows that HOMO energy is not related to the π-donor
ability of X. Indeed, it is clear that the amount of mixing of the
X orbitals into the “t2g” set follows the I > Br > Cl > F order,
in perfect agreement with the expected π-donor order. This
apparent contradiction between the HOMO energies and their
X character can be explained by considering the ionic character
of the Fe-X bond. In the case of fluorine, this small and
electronegative atom acts effectively as a negative point charge
located close to the metal, and this tends to destabilize the “t2g”
electrons, as has been previously suggested in an analysis of
bonding in Vaska-type complexes.9 When going down the
halogen column, the Fe-X bond becomes less polar and longer.
Both factors contribute to weaken the point charge effect,
consequently leading to a lower lying HOMO.
CpFe(dpe)X0/+ (X ) H, CH3, F, C1, Br, I) Series
compound
Eorb+Pauli
Eelect
-BDEhom*
[Fe]F
[Fe]C1
[Fe]Br
[Fe]I
[Fe]H (HA)
[Fe]CH3
-3.073
-1.351
-0.863
-0.610
-1.655
-0.118
-2.452
-2.861
-2.931
-2.225
-2.693
-3.407
-5.525
-4.212
-3.794
-3.337
-4.348
-3.526
[Fe]F+
-3.823
-1.966
-1.549
-1.099
-1.606
-0.845
-1.291
-1.613
-1.559
-1.545
-2.205
-2.355
-5.114
-3.579
-3.107
-2.644
-3.813
-3.200
[Fe]Cl+
[Fe]Br+
[Fe]I+
[Fe]H+ (HB+)
+
[Fe]CH3
The analysis of the Fe-X BDE variation across the
CpFe(dpe)X0/+ series might have been easy if it was possible
to evaluate separately all its components (σ and π covalent,
ionic, fragment relaxation components, ...). Unfortunately, this
is not at all straightforward, especially in the case of the
calculated compounds, owing to their low symmetry. With the
ADF code it is, however, possible to carry out a partial
decomposition of the computed BDE values by using the
transition-state approach developed by Ziegler.62 Since the
fragmentation is made on the [Fe]X0/+ molecules in their
optimized geometries, this decomposition is carried out on bond
dissociation energies which do not take into account the
geometrical relaxation energy of the [Fe]0/+/2+ fragments. In
addition, they are computed within the spin-restricted formalism
and they are not corrected from BSSE. These “unrelaxed” values
are written as BDE*’s in the following. They differ somewhat
from the BDE values but they exhibit similar major trends within
the halogen series and between neutral and cationic species. In
Ziegler’s transition-state method,62 BDE* values are decom-
posed into three terms. One of them, Eelect, corresponds to the
electrostatic interaction between the unperturbed fragments. It
is calculated from the electron densities of the isolated fragments
under consideration. It contains attractive contributions associ-
ated with the interaction between the electron density of one
fragment and the nuclei of the other fragment and a repulsive
contribution associated with the interaction between the fragment
electron densities. The Pauli repulsion term, EPauli, is also
calculated from the densities of the unrelaxed fragment. It can
be roughly approximated to the sum of what is called 4-electron/
2-orbital repulsions in approximate MO theory.63 Similarly, the
orbital interaction energy term, Eorb, can be roughly ap-
proximated to the so-called 2-electron/2-orbital attractive in-
teraction. Since the qualitative meaning of EPauli and Eorb is only
approximate, we will discuss only the sum of these two terms
which can be considered as being the covalent component of
the interaction between the fragments, that is: BDE* ) -(Eelect
+ EPauli + Eorb) ) -(Eelect + EPauli+orb). The values of BDE*,
Table 10. Energy Decomposition of Fe-X BDEhet* for the
CpFe(dpe)X0/+ (X ) H, CH3, F, C1, Br, I) Series
compound
Eorb+Pauli
Eelect
-BDEhet*
[Fe]F
[Fe]C1
[Fe]Br
[Fe]I
[FeH] (HA)
[Fe]CH3
+0.491
+0.938
+1.032
+0.828
+4.448
+3.656
-7.457
-6.811
-6.677
-6.156
-12.803
-12.124
-6.967
-5.873
-5.646
-5.328
-8.355
-8.236
[Fe]F+
+0.800
+0.767
+0.552
+0.193
+5.452
+3.162
-12.943
-11.636
-11.146
-10.491
-19.034
-16.684
-12.142
-10.870
-10.594
-10.298
-13.583
-13.521
[Fe]C1+
[Fe]Br+
[Fe]I+
[Fe]H+ (HB+)
+
[Fe]CH3
repulsions between the fragments. However, BDE* is dominated
by Eelect, which is largely stabilizing due to fragment charges
([Fe]+/2+ and X-). Because of the larger cationic charge in the
case of the [Fe]X+ series, Eelect is more stabilizing. It follows
that for a given X ligand, BDEhet* is larger in the cationic than
in the neutral species. In the [Fe]X (X ) halogen) series, the
Eelect stabilization decreases with the size of X in the order F >
Cl > Br > I. This can be interpreted as resulting from the fact
that the smaller X is, the more punctual and close to the cation
is its electron density. It follows that when going from [Fe]X
to [Fe]X+, the variation of Eelect (∆Eelect), and by inference of
∆BDEhet*, decreases in the order F > Cl > Br > I.
Looking now at the homolytic BDE* values (BDEhom*), one
can notice that the EPauli+orb terms are now stabilizing. This can
be explained by the fact that the 2-electron/2-orbital interaction
associated with the building of the Fe-X bond is now more
important since one of the electrons involved originates from a
high-lying metallic hybrid, as depicted schematically in Figure
10 (compare the top parts of Figures 10a,c and 10b,d). Although
weaker than in the heterolytic process, the stabilizing Eelect terms
still dominate the BDEhom* values. Contrary to the heterolytic
case, the Eelect stabilization is weaker in the cationic series.
Indeed, the oxidation of the [Fe] fragment corresponds to the
removal of a “t2g” electron, i.e., an electron that is close to the
E
elect and EPauli+orb, computed for the homolytic and heterolytic
dissociations of the [Fe]X0/+ series are listed in Tables 9 and
10, respectively.
Let us first analyze the more simple case, namely the
heterolytic BDE* values (BDEhet*). All the EPauli+orb terms are
destabilizing, indicating that the unique 2-electron/2-orbital
interaction arising from the building of the Fe-X bond is not
large enough to counterbalance the strong 4-electron/2-orbital
X nucleus, inducing a loss of stabilizing contribution in Eelect
.
It follows that for a given X ligand, BDEhom* is lower in the
cationic than in the neutral species, despite the diminution of
the Fe-X π-type repulsion upon oxidation (see Figure 10).
Unlike the situation in the heterolytic case, the variation of Eelect
when going from the neutral to the cationic species increases
with the size of X within the halogen series. This can be
explained by the fact that the mostly metal-centered electron
that is involved in the oxidation process is more effectively
screened by the halogen electron cloud when X is smaller.
(62) Ziegler, T. In Metal-Ligand Interactions: From Atoms, to Clusters,
to Surfaces; Salahub, D. R., Russo, N., Eds.; Kluwer: Dordrecht, The
Netherlands, 1992; p 367.
(63) (a) Landrum, G. A.; Goldberg, N.; Hoffmann, R. J. Chem. Soc.,
Dalton Trans. 1997, 3605. (b) Rosa, A.; Baerends, E. J. New J. Chem.
1991, 15, 815.

Fe-X Bonds in 17- and 18-Electron Cp*Fe(Dppe)X Complexes
J. Am. Chem. Soc., Vol. 123, No. 41, 2001 9997
and the alternative one based on Drago’s ETC bonding scheme11
in which an ionic (electrostatic) component to the bonding may
be of even greater significance than the π effect. The DFT
calculations on the [Fe]X model system corroborate the “inverse
halide order” of the relative electron densities at Fe, as measured
by the HOMO compositions and energies as well as by the
ionization potentials.
The use of thermochemical cycles based on electrode potential
data establishes that all [Fe*]-X bonds under investigation are
weakened with respect to homolytic cleavage (BDEhom data) as
a consequence of one-electron oxidation processes. In consider-
ing the Fe(dπ)-X(pπ) interactions or Drago’s ETC bonding
scheme, the decomposition of the BDEhom* data into Eelect
(electrostatic) and EPauli+orb (covalent) terms seems to be most
consistent with the latter interpretation, i.e., electrostatic effects
are of greater significance to explain the trends than are Fe-
(dπ)-X(pπ) interactions, at least in this system. For the
heterolytic cleavage, in which X is cleaved as X- (BDEhet data),
an oxidatively induced bond strengthening is seen, experimen-
tally and by calculations. This effect originates primarily in
electrostatic effects.
The IR νFe-X data show a shift to higher frequencies when
Cp*Fe(dppe)H and Cp*Fe(dppe)Cl are oxidized to the respective
17-electron cation radicals. These results are confirmed by
calculations for CpFe(dpe)H and CpFe(dpe)Cl. The IR trend is
apparently at odds with the observed homolytic BDEhom
weakening that results, by experiment (electrochemistry derived
data) and theory (DFT calculations), upon one-electron oxida-
tion. Interestingly, for the Cp*W(dppe)H3 complex, theory (DFT
calculations on the CpW(dpe)H3 model) and IR spectroscopic
data suggest that one-electron oxidation causes a W-H bond
strengthening.21 Unfortunately, corresponding electrochemistry
data are not available for the W system. From simple theory, it
is commonly considered that the vibrational frequency correlates
directly with the bond energy and inversely with the bond
length.64 The data presented by us suggest that extreme caution
should be exerted when relative homolytic bond strengths are
discussed on the basis of rather small differences in IR stretching
frequencies. In this respect, we point out that the oxidatively
induced increase in IR νFe-X frequencies correlate nicely with
the increase in heterolytic bond energies (BDEhet data). Analo-
gous observations and conclusions have been made in recent
studies of M-H bonding and IR νM-H data of Ni complexes
by DuBois and co-workers.15b
Experimental Section
General. All manipulations were carried out under an argon
atmosphere using Schlenk techniques or in Vacuum Atmospheres or
Jacomex 532 dryboxes filled with nitrogen. Reagent grade ether, THF,
and pentane were dried and distilled from sodium benzophenone ketyl
prior to use. Dichloromethane was distilled from P2O5. Complexes
Cp*Fe(dppe)H, Cp*Fe(dppe)H+PF6-, Cp*Fe(dppe)Me, Cp*Fe(dppe)-
Me+BF4-, Cp*Fe(dppe)Cl, Cp*Fe(dppe)Cl+PF6-, Cp*Fe(dppe)I, and
Cp*Fe(dppe)+PF6- were prepared following published procedures,19,25e
and other chemicals were purchased from commercial sources and used
as received.
Figure 10. Major orbital interactions involved in the [Fe]X and [Fe]-
X+ homolytic and heterolytic BDE values.
Concluding Remarks
Multiple methods of investigation have established that in
the halide series, the electron density at the metal decreases in
the series F > Cl > Br > I. This is corroborated experimentally
through the oxidation potentials for the [Fe*]X complexes,
through the trends in the multinuclear NMR data, and through
the Mo¨ssbauer analysis that demonstrated that the fluoride
complex appeared to be exceptionally electron rich. Thus, the
presence of a so-called “inverse halide order” that is the opposite
of that expected on the basis of halogen electronegativities alone
is firmly established.28b,36,50,61 These experimental data appear
to be consistent with the presence of Fe(dπ)-X(pπ) interactions,
but the data cannot help distinguish between this interpretation
NMR spectra were recorded on a multinuclear Bruker DPX 200
1
spectrometer at 297 K. H NMR spectra were recorded at 200 MHz,
and all chemical shifts are reported in ppm using internal tetrameth-
ylsilane (TMS) or the residual proton resonance resulting from
incomplete deuteration of the NMR solvent as the reference. 13C NMR
spectra were recorded at 50 MHz, and all chemical shifts are reported
in ppm using the carbon on the deuterated NMR solvent as the
(64) Infrared and Raman Spectroscopy: Methods and Applications;
Schrader, B., Ed.; VCH: Weinheim, Germany, 1995; p 693.

9998 J. Am. Chem. Soc., Vol. 123, No. 41, 2001
Tilset et al.
reference. 31P{1H} NMR spectra were recorded at 81 MHz, and all
chemical shifts are reported in ppm relative to external 85% H3PO4.
19F NMR spectra were recorded at 188 MHz, and all chemical shifts
are reported in ppm relative to external CFCl3. Magnetic susceptibility
measurements were performed in solution according to Evans’ method.37
X-Band ESR spectra were recorded on a Bruker ESP-300E
spectrometer at 77 K in liquid nitrogen. The 57Fe Mo¨ssbauer spectra
were recorded with a 2.5 × 10-2 C (9.25 × 108 Bq) 57Co source using
a symmetric triangular sweep mode.65 Computer fitting of the Mo¨ss-
bauer data to Lorentzian line shapes was carried out with a previously
reported computer program.66 The isomer shift values are reported
relative to iron foil at 298 K and are not corrected for the temperature-
dependent second-order Doppler shift.
Infrared spectra were recorded using a Bruker instrument IFS28
(4000-400 cm-1), a Perkin-Elmer 684 dispersive instrument (600-
200 cm-1), or a Perkin-Elmer Paragon 1000 FTIR spectrometer (KBr
beam splitter, 400 cm-1 cutoff) with a resolution of 2 cm-1. The spectra
of 1-4 and the corresponding cation PF6- salts were obtained as Nujol
mulls between polyethylene sheets.
Cyclic voltammograms were recorded using a an EG&G-PAR model
263 potentiostat/galvanostat. The working electrode was a Pt disk
electrode (d ) 0.4 or 1.0 mm), the counter electrode was a Pt wire,
and the saturated calomel electrode (SCE) or a Ag/Ag+(MeCN)
electrode were used as reference electrodes. The Cp2Fe0/+ couple was
used as an internal calibrant for the potential measurements.67 Elemental
analyses were performed at the Center for Microanalyses of the CNRS
at Vernaison, France.
Preparation of Cp*Fe(dppe)F (1). Method 1: From Cp*Fe-
(dppe)+PF6-. A dry, solid sample of CsF (0.304 g, 2.0 mmol) was
added to an orange THF solution (40 mL) of Cp*Fe(dppe)+PF6- (1.660
g, 2.26 mmol) at ambient temperature. The reaction mixture turned
progressively yellowish green after being stirred overnight. The solvent
was removed under vacuum, and the remaining residue was extracted
with ether (3 × 30 mL). The combined extracts were evaporated to
dryness, and an 86% yield (1.050 g) of crude Cp*Fe(dppe)F was
isolated as a yellowish-green, air-sensitive powder.
change from green to red. The mixture was stirred for 15 min, and the
solvent was removed under vacuum. The red material was washed with
ether (3 × 30 mL). An 87% yield of Cp*Fe(dppe)F+PF6 (0.460 g)
-
was isolated.
Anal. Calcd for C36H39F7FeP3: C, 57.39; H, 5.22; P, 12.33. Found:
C, 57.30; H, 5.24; P, 12.75. µeff (CH2Cl2, 310 K) ) 1.88 µB. Mo¨ssbauer
data (80 K, mm s-1): I.S. ) 0.426 vs Fe; Q.S. ) 0.915. ESR (9:1
THF/pentane, 77 K): gx ) 2.419, gy ) 2.018, AF ) 60 G, gz ) 1.998,
AF ) 110 G. 1H NMR (acetone-d6, 323 K) δ 4.57 (s, C5Me5, ω1/2 ) 40
Hz), 5.64 (s, CH2, ω1/2 ) 30 Hz); 213 K δ 3.23 (bs, C5Me5, ω1/2 ) 230
Hz), 4.72 (bs, CH2, ω1/2 ) 190 Hz). 13C NMR (acetone-d6, 293 K) δ
16.2 (bs, C5Me5, ω1/2 ) 500 Hz), 28.5 (bs, CH2), 99.7 (bs, C5Me5, ω1/2
) 201 Hz), 9.7, 122.7, 124.6, 130.8, 138.3 (Ph). 31P NMR (acetone-
d6, 293 K) δ -143.4 (sept., JPF ) 709 Hz, PF6-). 19F NMR (CD2Cl2,
293 K) δ -60.0 (bs, Fe-F, ω1/2 ) 360 Hz), -73.1 (d, JPF ) 709 Hz,
PF6-).
Preparation of Cp*Fe(dppe)Br (3). A sample of KBr (0.715 g,
6.0 mmol) was added to a green solution of Cp*Fe(dppe)Cl (3.12 g,
5.0 mmol) in dichloromethane (30 mL). The reaction mixture was stirred
overnight at ambient temperature. The solution was filtered and carefully
layered with pentane (100 mL). After 2 weeks, 3.00 g (yield 90%) of
air and thermally stable dark-brown crystals of Cp*Fe(dppe)Br were
collected. Anal. Calcd for C36H39BrFeP2: C, 64.59; H, 5.87; P, 9.25.
1
Found: C, 64.36; H, 5.90; P, 9.31. H NMR (C6D6) δ 1.53 (s, 15 H,
C5Me5), 2.03 and 2.61 (2 m, 4 H, CH2), 7.10-8.11 (m, 20 H, Ph).13C
NMR (C6D6) δ 10.8 (q, C5Me5, 1JCH ) 128 Hz), 31.0 (m, CH2, 1JCH
)
136 Hz), 83.6 (s, C5Me5), 127.7-140.3 (m, Ph). 31P{1H} NMR (C6D6)
δ 93.2 (s, dppe).
Preparation of Cp*Fe(dppe)Br+PF6 (3+PF6-). A sample of
-
Cp2Fe+PF6- (0.298 g, 0.9 mmol) was added to a dark-orange solution
of Cp*Fe(dppe)Br (0.67 g, 1.0 mmol) in THF (20 mL) and the reaction
mixture was stirred for 1 h at ambient temperature. After removal of
the solvent under vacuum, the dark residue was washed with ether (5
× 50 mL). Following crystallization from THF/pentane, 0.70 g (95%
yield) of air and thermally stable black microcrystals of Cp*Fe(dppe)-
Br+PF6- were isolated. Anal. Calcd for C36H39BrF6FeP3: C, 53.10; H,
4.83; P, 11.41. Found: C, 53.60; H, 4.79; P, 12.13. µeff (CD2Cl2, 297
K) ) 2.7 µB.
Method 2: From Cp*Fe(dppe)F+PF6-. A sample of Cp2Co (0.166
g, 0.88 mmol) was quickly added under argon to a cooled (-100 °C)
red THF solution (40 mL) of Cp*Fe(dppe)F+PF6- (0.780 g, 1.03 mmol),
prepared as described below. The resulting mixture was stirred while
being allowed to warm slowly overnight to room temperature. The color
progressively turned yellowish green. Workup provided Cp*Fe(dppe)F
(0.480 g, 90% based on Cp2Co).
Chemical Reduction of Cp*Fe(dppe)Br+PF6-. At -80 °C under
an atmosphere of argon, a solid sample of Cp2Co (0.17 g, 0.9 mmol)
was quickly added to a dark-brown solution of Cp*Fe(dppe)Br+PF6
-
(0.814 g, 1.0 mmol) in THF (30 mL). The mixture was stirred for 1 h.
After heating to ambient temperature, the solvent was removed under
vacuum and the remaining residue was extracted with ether (3 × 30
mL). Crystallization from dichloromethane/pentane provided 0.50 g
(95% yield) of dark-brown microcrystals of Cp*Fe(dppe)Br, analyzed
as above.
1H NMR (C6D6) δ 1.33 (s, 15 H, C5Me5), 1.75, 1.84 (m, 4 H, CH2),
7.00-7.93 (m, 20 H, Ph). 13C NMR (C6D6) δ 10.0 (C5Me5, 1JCH ) 132
1
Hz), 29.2 (t, CH2, JCH ) 124 Hz), 82.9 (s, C5Me5), 127.1-136.6 (m,
Ph). 31P{1H} NMR (C6D6) δ 89.4 (d, dppe, JPF ) 43 Hz). 19F NMR
(C6D6) δ -44.4 (t, Fe-F, JPF ) 43 Hz).
Preparation of Cp*Fe(dppe)I+PF6 (4+PF6-). At room tempera-
-
ture, a sample of Cp2Fe+PF6 (0.298 g, 0.9 mmol) was added to an
-
Preparation of Cp*Fe(dppe)F+PF6- (1+PF6-). Method 1: From
Cp*Fe(dppe)+PF6-. To a cooled (-90 °C) orange THF solution (70
mL) of Cp*Fe(dppe)+PF6- (1.10 g, 1.5 mmol) was added a solid sample
of Cp2Fe+PF6- (0.420 g, 1.27 mmol). The color darkened immediately
and at -70 °C became dark red. Stirring was continued while the
mixture was allowed to warm slowly (overnight) to room temperature.
The THF had partly polymerized. The remaining solvent was evaporated
under vacuum and the red gummy residue was extracted with acetone
(10 × 60 mL). The extracts were combined and evaporated to dryness,
and the dark red powder was washed with ether (3 × 20 mL) to remove
the ferrocene. Crystallization from an acetone solution layered with
pentane afforded dark red, thermally and air stable crystals of
orange-brown solution of Cp*Fe(dppe)I (0.720 g, 1.0 mmol) in THF
(20 mL). The reaction mixture was stirred for 1 h and the solvent was
removed under vacuum. The remaining dark solid residue was washed
with ether (5 × 50 mL). Crystallization from THF/pentane gave 0.70
g (90% yield) of air and thermally stable black microcrystals of
Cp*Fe(dppe)I+PF6-. Anal. Calcd for C36H39F6FeIP3: C, 50.20; H, 4.56;
P, 10.79. Found: C, 51.12; H, 4.48; P, 11.23. µeff (CD2Cl2, 297 K) )
2.3 µB.
Preparation of Cp*Fe(dppe)H+PF6- (5+PF6-). This method is an
improvement over that previously reported.19b At room temperature, a
-
sample of Cp2Fe+PF6 (0.298 g, 0.9 mmol) was added to an orange
-
Cp*Fe(dppe)F+PF6 (0.780 g, 81% based on Cp2Fe+PF6-).
solution of Cp*Fe(dppe)H (0.590 g, 1.0 mmol) in THF (20 mL). The
mixture was stirred for 15 min, during which the color of the solution
turned progressively red. The solvent was removed under vacuum, and
the remaining dark solid residue was washed with diethyl ether (5 ×
30 mL). Crystallization from acetone/pentane mixture gave air and
Method 2: From Cp*Fe(dppe)F. Treatment of a THF solution (30
mL) of freshly prepared Cp*Fe(dppe)F (0.480 g, 0.79 mmol) with a
sample of Cp2Fe+PF6- (0.232 g, 0.70 mmol) caused an immediate color
thermally stable red microcrystals of Cp*Fe(dppe)H+PF6 (0.75 g,
-
(65) Varret, F.; Mariot, J.-P.; Hamon, J.-R.; Astruc, D. Hyperfine Interact.
1988, 39, 67.
95%). Anal. Calcd for C36H40F6FeP3: C, 58.79; H, 5.48; P, 12.63.
(66) (a) Boinnard, D.; Boussekssou, A.; Dworkin, A.; Savariault, J.-M.;
Varret, F.; Tuchagues, J.-P. Inorg. Chem. 1994, 33, 271. (b) Varret, F.;
Varret, F., Ed.; International Conference on Mo¨ssbauer Effects Applications;
Indian Science Academy, New Delhi, 1982: Jaipur, India, 1981.
(67) Connelly, N. G.; Geiger, W. E. Chem. ReV. 1996, 96, 877.
Found: C, 58.56; H, 5.31; P, 12.70. IR (Nujol) νFe-H 1886 cm-1
;
Mo¨ssbauer (4.2 K, mm s-1) I.S. ) 0.260, Q.S. ) 0.840. EPR (CH2-
Cl2/CH2ClCH2Cl, 77 K): g1 ) 1.9944, g2 ) 2.0430, g3 ) 2.4487. µeff
(CH2Cl2, 297 K) ) 2.40 µB.

Fe-X Bonds in 17- and 18-Electron Cp*Fe(Dppe)X Complexes
J. Am. Chem. Soc., Vol. 123, No. 41, 2001 9999
Table 11. Crystal Data, Data Collection, and Refinement Parameters for 1+PF6-, 2+PF6-, and 6+BF4
-
1+PF6
2+PF6
6+BF4
-
-
-
formula
fw
temperature (K)
crystal system
space group
a (Å)
C36H39F7FeP3
753.43
293(2)
monoclinic
P21/n
14.253(8)
15.252(2)
15.873(4)
90
90.86(3)
90
3450(2)
4
C36H39ClF6FeP3
769.88
293(2)
orthorhombic
P212121
12.763(3)
13.877(7)
19.551(7)
90
90
90
3463(2)
4
1.477
C37H42BF4FeP2
691.31
293(2)
monoclinic
P21/n
31.490(4)
9.834(4)
22.322(5)
90
92.42(2)
90
6906(3)
4
b (Å)
c (Å)
R (deg)
â (deg)
γ (deg)
V (Å-3
)
Z
D(calcd) (g cm-3
)
1.450
0.665
crystal size (mm)
F(000)
0.35 × 0.35 × 0.20
0.25 × 0.25 × 0.12
0.65 × 0.45 × 0.45
1556
1588
1444
diffractometer
radiation
CAD4
MoKR
0.640
1.85-24.98
0/16, 0/18, -18/18
6297
6040
3424
CAD4
MoKR
0.710
1.80-24.95
0/15, 0/16, 0/23
3406
3406
2423
CAD4
MoKR
0.288
1.10-24.98
-37/37, 0/10, 0/26
11859
11553
6598
abs coef (mm-1
θ range (deg)
range h ,k, l
)
total no. of reflcns
no. of unique reflcns
no. of obsd reflcns [I > 2σ(I)]
no. of restraints/no. of parameters
w ) 1/[σ2(Fo)2 + (aP)2 + bP]
(where P ) [Fo2 + Fc2]/3)
final R
0/425
0/419
0/775
a ) 0.0845
b ) 1.2816
0.0476
a ) 0. 1374
b ) 0.0000
0.0530
a ) 0.1540
b ) 15.9595
0.0757
Rw
0.1284
0.1329
0.1540
1.030
0.568
-0.417
0.1453
0.1024
0.1770
1.006
0.736
-0.535
0.2241
0.1651
0.2836
1.066
1.165
-0.775
R indices (all data)
Rw (all data)
goodness of fit/F2
largest diff peak
and hole (e Å-3
)
-
Chemical Reduction of Cp*Fe(dppe)I+PF6-. Following the same
procedure and workup as described above for the bromo derivative,
Cp*Fe(dppe)I+PF6- (0.86 g, 1.0 mmol) was treated with Cp2Co (0.17
g, 0.9 mmol) to provide 0.610 g (95%) of Cp*Fe(dppe)I after
crystallization from dichloromethane/pentane.
was used for structure determination.72 ORTEP views of 1+PF6
,
2+PF6-, and 6+BF4 were generated with PLATON98.73 All the
calculations were performed on a Pentium NT Server computer.
Computational Details. DFT calculations74 were carried out using
the Amsterdam Density Functional (ADF) program.75 The model
compounds CpFe(dpe)X,17 to be abbreviated as [Fe]X,17 were used to
reduce computational effort. The Vosko-Wilk-Nusair parametriza-
tion76 was used to treat electron correlation within the local density
approximation (LDA). The nonlocal corrections of Becke77 and of
Perdew78 were added to the exchange and correlation energies,
respectively. The numerical integration procedure applied for the
calculations was developed by te Velde et al.74 A triple-ú Slater-type
orbital (STO) basis set was used for Fe 3d and 4s, and a single-ú STO
was used for Fe 4p. Concerning X (X ) F, Cl, Br, I, CH3, H), a triple-ú
STO basis set was employed for H 1s, C 2s and 2p, F 2s and 2p, Cl 3s
and 3p, Br 3d, 4s, and 4p, and I 4d, 5s, and 5p, augmented with a
single-ú polarization function (2p for H; 3d for C, F, and Cl; and 4d
for Br). The other atoms were described by a double-ú STO basis set
for H 1s, C 2s and 2p, and P 3s and 3p, augmented with a single-ú
polarization function (2p for H and C; 3d for P). Full geometry
optimizations (assuming C1 symmetry) were carried out on each
complex, using the analytical gradient method implemented by Verluis
-
X-ray Crystallographic Studies of Cp*Fe(dppe)F+PF6-, Cp*Fe-
(dppe)Cl+PF6-, and Cp*Fe(dppe)Me+BF4-. X-ray quality crystals
of Cp*Fe(dppe)Cl+PF6 were obtained after 3 weeks from a concen-
-
trated dichloromethane solution that was layered with pentane. Crystals
of Cp*Fe(dppe)Me+BF4- were grown from a concentrated THF solution
that was layered with pentane, and crystals of Cp*Fe(dppe)F+PF6- were
grown from acetone-pentane. A summary of the crystallographic data
is given in Table 11. Complete details of the crystal data, X-ray data
collection, and structure solution are provided as Supporting Informa-
tion. Cell constants and orientation matrix for data collection were
obtained from a least-squares refinement using 25 high-θ reflections.
For all compounds, after Lorenz and polarization corrections,68 the
structures were solved with SIR-97,69 which revealed many non-
hydrogen atoms of the molecules. After anisotropic refinements, a
Fourier difference map revealed many hydrogen atoms. The whole
structures were next refined with SHELXL97,70 by the full-matrix least-
squares techniques (use of F2 magnitude; x, y, z, âij for Fe, P, Cl, F,
and C atoms, and x, y, z in riding mode for H atoms with variables
(N(var.)), observations and "ω" used as defined in Table 11). Atomic
scattering factors were taken from the literature.71 A Silicon Graphics
Indy computer with the MOLEN package (ENRAF-NONIUS, 1990)
(72) Fair, C. K. MolEN. An InteractiVe Intelligent System for Crystal
Structure Analysis; ENRAF-NONIUS: Delft, The Netherlands, 1990.
(73) Spek, A. L. PLATON-98, A Multipurpose Crystallographic Tool;
Utrecht University: Utrecht, The Netherlands, 1998.
(74) (a) Baerends, E. J.; Ellis, D. E.; Ros, P. Chem. Phys. 1973, 2, 41.
(b) Baerends, E. J.; Ros, P. Int. J. Quantum Chem. 1978, S12, 169. (c)
Boerrigter, P. M.; te Velde, G.; Baerends, E. J. Int. J. Quantum Chem.
1988, 33, 87. (d) te Velde, G.; Baerends, E. J. J. Comput. Phys. 1992, 99,
84.
(75) Amsterdam Density Functional (ADF) program, version 2.3; Vrije
Universiteit: Amsterdam, The Netherlands, 1996.
(68) Spek, A. L. HELENA. Program for the handling of CAD4-
Diffractometer Output SHELX(S/L); Utrecht University: Utrecht, The
Netherlands, 1997.
(69) Altomare, A.; Burla, M. C.; Camalli, M.; Cascarano, G.; Giacovazzo,
C.; Guagliardi, A.; Moliterni, A. G. G.; Polidori, G.; Spagna, R. J. Appl.
Chem. 1998, 31, 74.
(70) Sheldrick, G. M. SHELX97. Program for the Refinement of Crystal
Structures; University of Go¨ttingen: Go¨ttingen, Germany, 1997.
(71) International Tables for X-ray Crystallography; Kluwer Academic
Publishers: Dordrecht, The Netherlands, 1992; Vol. C.
(76) Vosko, S. D.; Wilk, L.; Nusair, M. Can. J. Chem. 1990, 58, 1200.
(77) Becke, A. D. J. Chem. Phys. 1986, 84, 4524. Becke, A. D. Phys.
ReV. A. 1988, 38, 3098.
(78) Perdew, J. P. Phys. ReV. B 1986, 33, 8882; 34, 7406.

10000 J. Am. Chem. Soc., Vol. 123, No. 41, 2001
Tilset et al.
and Ziegler.79 Spin-unrestricted calculations were performed for all
open-shell systems considered.
Acknowledgment. The authors gratefully acknowledge Dr.
C. Lapinte (Rennes) for his helpful assistance and comments
concerning the ESR and Mo¨ssbauer studies. The Centre National
de la Recherche Scientifique (CNRS), the University of Rennes
1, the University of Oslo, and the University of Sheffield are
thanked for financial support. Computing facilities were pro-
vided by the Centre de Ressources Informatiques (CRI) of
Rennes and the Institut de Developpement et de Ressources en
Informatique Scientifique du CNRS (IDRIS-CNRS).
Diabatic ionization potentials were calculated as the energy difference
between the optimized geometries of the reduced and oxidized species.
The computed BDE values correspond to the following reactions:
CpFe(dpe)X f CpFe(dpe) + X and CpFe(dpe)X+ f CpFe(dpe)+
X (homolytic dissociation, BDEhom); CpFe(dpe)X f CpFe(dpe)+
+
+
X- and CpFe(dpe)X+ f CpFe(dpe)2+ + X- (heterolytic dissociation,
BDEhet). These have been calculated by subtracting the energies of the
dissociated species from those of the corresponding CpFe(dpe)X0/+
complexes. The considered energies were those obtained after full
geometry optimizations, i.e., including the fragment geometrical
relaxation. The Basis Set Superposition Errors (BSSE) were evaluated
by using the counterpoise method.80 The organometallic fragments
CpFe(dpe), CpFe(dpe)+, and CpFe(dpe)2+ were considered in their
ground state (i.e. doublet, triplet, and doublet state, respectively).55
Supporting Information Available: X-ray crystallographic
file (CIF) with listings of crystal data, X-ray experimental
details, atomic coordinates, thermal parameters, and bond
distances and angles for 1+PF6-, 2+PF6-, and 6+BF4-. This
material is available free of charge via the Internet at
http://pubs.acs.org.
(79) Verluis, L.; Ziegler, T. J. Chem. Phys. 1988, 88, 322.
(80) (a) van Duijneveldt, F. B.; van Duijneveldt de Rijdt, J. G. C. M.;
van Lenthe, J. H. Chem. ReV. 1994, 94, 1873. (b) Boys, S. F.; Bernardi, F.
Mol. Phys. 1970, 19, 553. (c) Rosa, A.; Ehlers, A. W.; Baerends, E. J.;
Snijders, J. G.; te Velde, G. J. Phys. Chem. 1996, 100, 5690.
JA0106927