Structural Insights into the Nature of Fe0 and FeI Low-Valent Species Obtained upon the Reduction of Iron Salts by Aryl Grignard Reagents

Article abstract of DOI:10.1021/acs.inorgchem.6b02616

Mechanistic studies of the reduction of Fe^III and Fe^II salts by aryl Grignard reagents in toluene/tetrahydrofuran mixtures in the absence of a supporting ligand, as well as structural insights regarding the nature of the low-valent iron species obtained at the end of this reduction process, are reported. It is shown that several reduction pathways can be followed, depending on the starting iron precursor. We demonstrate, moreover, that these pathways lead to a mixture of Fe⁰ and Fe^I complexes regardless of the nature of the precursor. M?ssbauer and ¹H NMR spectroscopies suggest that diamagnetic 16-electron bisarene complexes such as (η⁴-C₆H₅Me)₂Fe⁰ can be formed as major species (85% of the overall iron quantity). The formation of a η⁶-arene-ligated low-spin Fe^I complex as a minor species (accounting for ca. 15% of the overall iron quantity) is attested by M?ssbauer spectroscopy, as well as by continuous-wave electron paramagnetic resonance (EPR) and pulsed-EPR (HYSCORE) spectroscopies. The nature of the Fe^I coordination sphere is discussed by means of isotopic labeling experiments and density functional theory calculations. It is shown that the most likely low-spin Fe^I candidate obtained in these systems is a diphenylarene-stabilized species [(η⁶-C₆H₅Me)Fe^IPh₂]^- exhibiting an idealized C_2v topology. This enlightens the nature of the lowest valence states accommodated by iron during the reduction of Fe^III and Fe^II salts by aryl Grignard reagents in the absence of any additional coligand, which so far remained rather unknown. The reactivity of these low-valent Fe^I and Fe⁰ complexes in aryl-heteroaryl Kumada cross-coupling conditions has also been investigated, and it is shown that the zerovalent Fe⁰ species can be used efficiently as a precursor in this reaction, whereas the Fe^I oxidation state does not exhibit any reactivity.

Full text of DOI:10.1021/acs.inorgchem.6b02616

Article
pubs.acs.org/IC
Structural Insights into the Nature of Fe0 and FeI Low-Valent Species
Obtained upon the Reduction of Iron Salts by Aryl Grignard
Reagents
Martin Clemancey,† Thibault Cantat,‡ Genevieve Blondin,† Jean-Marc Latour,*,† Pierre Dorlet,*,§
́
̀
and Guillaume Lefevre*,‡
̀
†Universite
́
Grenoble Alpes, LCBM/PMB and CEA, BIG/CBM/PMB and CNRS, LCBM UMR 5249, PMB, 38000 Grenoble, France
‡NIMBE, CEA, CNRS, Universite
́
Paris-Saclay, Gif-sur-Yvette, France
§Institute for Integrative Biology of the Cell (I2BC), CEA, CNRS, Univ. Paris-Sud, Universite
́
Paris-Saclay, Gif-sur-Yvette, France
S
* Supporting Information
ABSTRACT: Mechanistic studies of the reduction of FeIII and
FeII salts by aryl Grignard reagents in toluene/tetrahydrofuran
mixtures in the absence of a supporting ligand, as well as
structural insights regarding the nature of the low-valent iron
species obtained at the end of this reduction process, are
reported. It is shown that several reduction pathways can be
followed, depending on the starting iron precursor. We
demonstrate, moreover, that these pathways lead to a mixture of Fe0 and FeI complexes regardless of the nature of the
1
precursor. Mossbauer and H NMR spectroscopies suggest that diamagnetic 16-electron bisarene complexes such as (η4-
̈
C6H5Me)2Fe0 can be formed as major species (85% of the overall iron quantity). The formation of a η6-arene-ligated low-spin FeI
complex as a minor species (accounting for ca. 15% of the overall iron quantity) is attested by Mossbauer spectroscopy, as well as
̈
by continuous-wave electron paramagnetic resonance (EPR) and pulsed-EPR (HYSCORE) spectroscopies. The nature of the FeI
coordination sphere is discussed by means of isotopic labeling experiments and density functional theory calculations. It is shown
that the most likely low-spin FeI candidate obtained in these systems is a diphenylarene-stabilized species [(η6-C6H5Me)FeIPh2]−
exhibiting an idealized C2v topology. This enlightens the nature of the lowest valence states accommodated by iron during the
reduction of FeIII and FeII salts by aryl Grignard reagents in the absence of any additional coligand, which so far remained rather
unknown. The reactivity of these low-valent FeI and Fe0 complexes in aryl−heteroaryl Kumada cross-coupling conditions has also
been investigated, and it is shown that the zerovalent Fe0 species can be used efficiently as a precursor in this reaction, whereas
the FeI oxidation state does not exhibit any reactivity.
developed with simple iron catalysts, e.g., Fe(β-diketonate)3
and FeCl2 or FeCl3, without any stabilizing coligand. Iron salts
combined with N-heterocyclic carbenes (NHCs) and fluoride
anions proved to circumvent the homocoupling of the Grignard
reagent, which was often observed as a side reaction.3
INTRODUCTION
■
To reach a broad scope of cross-coupling reactions based on
iron catalysis is an appealing alternative to the noble-metal-
based cross-coupling catalysis because iron is a cost-efficient,
abundant, and nontoxic metal.1 The first iron-catalyzed cross-
coupling reactions were reported by Kochi in the early 1970s,
and they involved vinylic halides and alkyl Grignard reagents
(Scheme 1a).2a,b Several decades later, the methodology was
successfully extended to heteroaryl halides and aryl Grignard
From a mechanistic standpoint, several low-valent iron
species were suggested as catalytically relevant (e.g., Fe−II, Fe0,
FeI, and FeII species).4 Furstner et al. reported in 2005 that
̈
homoleptic “super-ate FeII-complexes” such as [(Me4FeII)-
(MeLi)][Li(OEt2)]2 could exhibit activity toward organic
electrophiles in catalytic and stoichiometric cross-coupling
processes.5 Later, the Bedford group demonstrated that
bis(diphosphine)iron(I) monoaryl complexes were catalytically
active in Negishi-type couplings.6 The reactivity of in situ
generated FeI species with organic halides (e.g., aryl,7a benzyl,7b
or alkyl derivatives7c) has been extensively investigated ever
since, with electron paramagnetic resonance (EPR) spectros-
copy being often used as an experimental characterization tool
reagents by Furstner et al. (Scheme 1b) and was further
̈
explored by other groups.2c−n This coupling reaction was first
Scheme 1. Representative Milestones in Iron-Catalyzed C−
C Cross-Couplings
Received: October 27, 2016
© XXXX American Chemical Society
A
DOI: 10.1021/acs.inorgchem.6b02616
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
to support the involvement of FeI d7 complexes. Several reports
focused on the possibility of promoting C−C bond formation
reactions using tetramethylethylenediamine-7d and NHC-
stabilized7e−g FeII well-defined species were also published
recently. In 2015, the Neidig group reported that the catalytic
activity observed in aryl−alkyl Kumada and Suzuki−Miyaura
cross-couplings catalyzed by SciOPP-stabilized systems was due
to (SciOPP)FeII(Ph)n species [SciOPP = C6H4-1,2-(P(C6H3-
3,5-tBu2)2)2; n = 1 and 2]. They also demonstrated that the
bisphenylated species (SciOPP)FeII(Ph)2 was less selective and
led to a significant amount of β-elimination products when
secondary alkyl halides are employed as electrophiles.
Importantly, it was also shown that FeI intermediates were
obtained as minor reduced byproducts as well as catalytically
nonrelevant Fe0 species.8 The authors demonstrated in this
report the importance of the spin titration of EPR-detected
species because iron oxidation states with an integer electronic
spin (more specifically FeII and Fe0) cannot be detected by this
technique. The matter of identification of the catalytically active
species is also crucial and is still unanswered for systems based
on the use of coligand-free iron salts and aryl Grignard reagents
as reductants, as well as the exact nature of the metal oxidation
states reached upon reduction under these conditions. Such
coligand-free systems are more difficult to investigate because
of the absence of stabilizing ligands, since the low-valent iron
species have consequently shorter lifetimes and tend to lead to
aggregates. A very representative case is the recent advance-
ment made in the reduction of iron salts with methyl
complexes featuring bisarene coordination of the metal. The
presence of Fe0-arene ligation is also confirmed by low-
temperature 1H NMR spectroscopy. Fe0 molecular species then
evolve to the formation of metallic aggregates. These results are
supported by DFT calculations, which were performed to
investigate the electronic structure of the above-mentioned FeI
and Fe0 species. Moreover, it is demonstrated at the end of the
manuscript that the zerovalent Fe0 complex can be used as an
efficient precursor in Kumada aryl−heteroaryl cross-couplings,
whereas the FeI species remains unreactive in the same
conditions.
EXPERIMENTAL SECTION
■
General Considerations. All of the samples were prepared at 25
°C in a recirculating mBraun LabMaster DP inert-atmosphere (argon)
drybox and vacuum Schlenk lines. Glassware was dried overnight at 60
°C before use. NMR spectra were obtained using a Bruker DPX 200
MHz spectrometer. Chemical shifts for 1H NMR spectra were
referenced to solvent impurities (herein THF). Solvents (toluene-h8,
toluene-d8, and THF-d8) were dried over a sodium/benzophenone
mixture and distilled before use.
EPR Spectroscopy. cw-EPR spectra were recorded on a Bruker
Elexsys E500 spectrometer equipped with a continuous-flow cryostat
(Oxford). Pulsed-EPR experiments were performed on a Bruker
Elexsys E580 spectrometer at liquid-helium temperatures. Four-pulse
hyperfine sublevel correlation (HYSCORE) spectra were recorded
with the usual four-pulse sequence (π/2)y−τ−(π/2)y−t1−(π)y−t2−(π/
2)y with pulse lengths tπ/2 = tπ = 24 ns and different amplitudes for π/2
and π pulses. The size of the data was 256 × 256, and the time
intervals t1 and t2 were varied from 120 to 4200 ns in steps of 16 ns. A
four-step phase cycle was used to eliminate unwanted echoes. The
time domain spectra were baseline-corrected and zero-filled before 2D
Fourier transformation to yield the frequency maps. Pulsed-EPR data
were processed and simulations were performed by using the Easyspin
software package and routines locally written with Matlab (The
Mathworks, Inc.).
1
nucleophiles, with delineation of the structure of the S = /2
species observed by Kochi in his first reports when reducing
ferric chloride with the alkyl Grignard MeMgBr in tetrahy-
drofuran (THF).2b This structure has remained unknown for
almost 40 years, and Neidig et al. recently reported that this
species was a cluster involving eight iron centers ([Fe8Me12]−),
which could exhibit catalytic activity when associated with an
excess of MeMgBr.9 Therefore, the question of both the
identification and titration of half-integer spin iron species that
have been detected in similar systems remains crucial. In this
paper, we focused our investigations on delineation of the
nature of reduced metallic intermediates obtained upon the
reaction of simple FeIII or FeII salts with aryl Grignard reagents
(herein PhMgBr or PhMgCl) in the absence of a stabilizing
coligand. Some mechanistic insights regarding the redox events
involved in the reduction steps (bielectronic or monoelectronic
transfers) have been gained. We demonstrate that the same
low-spin FeI species can be obtained upon the reduction of salts
such as Fe(acac)n or FeCln (n = 2, 3) by PhMgBr in toluene/
THF (10:1) mixtures. As evidenced by radical-trapping
experiments, the FeI species originate from successive
monoelectronic reductions of FeII and FeIII precursors.
Disproportionation of transient FeII intermediates also provides
FeI oxidation state, associated with FeIII species. For the first
time, structural insights are gained for this FeI complex.
Continuous-wave (cw-)EPR and pulsed-EPR spectroscopies
were used to investigate its coordination sphere, which proved
to feature η6-coordination of the solvent, as well as σ-
DFT Computations. DFT computations were performed using
Gaussian09, version D.01. Unless specified otherwise, all structures
were characterized as minima by frequency analysis (no imaginary
frequency) and optimized without geometrical constraint. Unrestricted
DFT methods were applied (see the SI for detailed methods).
̈
Mossbauer Spectroscopy. Mossbauer spectra were recorded at 5
̈
K on a low-field Mossbauer spectrometer equipped with a Janis CCR
̈
5K cryostat or at 4.2 or 20 K on a strong-field Mossbauer spectrometer
̈
equipped with an Oxford Instruments Spectromag 4000 cryostat
containing an 8 T split-pair superconducting magnet. Both
spectrometers were operated in a constant-acceleration mode in
transmission geometry. The isomer shifts are referenced against that of
a metallic iron foil at room temperature. Analysis of the data was
performed with the program WMOSS (WMOSS4 Mossbauer Spectral
̈
Analysis Software, www.wmosss.org, 2009−2015) and a homemade
program.10 Computational modelings of Mossbauer parameters were
̈
carried out using ORCA software (version 3.0.3) at a BP86−TZVP/
CP(PPP)(Fe) level [see the Supporting Information (SI) for detailed
methods].
Synthesis of Isotopically Enriched Products. 57FeCl2 was
synthesized as follows. Elemental 94%-enriched 57Fe was heated under
a dry in situ generated chlorine atmosphere at 350 °C for 15 min. Dry
chlorine was obtained by the dropwise addition of anhydrous sulfuric
acid into a commercial bleach solution; the resulting chlorine gas was
then dried by bubbling into sulfuric acid (98%). 13C6H5MgBr was
synthesized according to a reported microscale method using
commercial 13C6H5Br and elemental magnesium.11
−
coordination of anionic C6H5 ligands originating from
PhMgX (X = Cl, Br). However, FeI oxidation state only
represents ca. 15% of the overall iron quantity, with the
distribution of the metallic species being quickly dominated by
the formation of diamagnetic Fe0 species, as evidenced by
RESULTS AND DISCUSSION
■
Mossbauer spectroscopy. The nuclear parameters obtained for
Role of the Nature of the Precursor in the Reduction
Mechanism. The reduction of Fe(acac)3 by 1 equiv of
̈
this Fe0 species are moreover similar to already reported
B
DOI: 10.1021/acs.inorgchem.6b02616
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
PhMgBr in THF-d8 was monitored by 1H NMR. Fe(acac)3 was
Scheme 3. Computed Free Energies (kcal·mol−1) of (a)
Mono- and Bisphenyl-FeIII Intermediate Formation, (b)
Reduction of Fe(acac)3 into Fe(acac)2 by PhMgBr Involving
fully consumed, whereas Fe(acac)2, acacMgBr, and PhPh were
1
characterized by their H NMR signals [Fe(acac)2, −31.5 ppm
a
a Dimeric Process
(12H, CH3), CH not detected in THF; acacMgBr, 5.2 ppm
(1H, CH)], attesting that they were quantitatively obtained
within 5 min at room temperature, leading to a bright-yellow
solution (Scheme 2a).12
Scheme 2. Reduction of Iron Precursors by PhMgBr in the
a
Absence (a and b) or Presence (c) of a Hydrogen Donor
a
(*) indicates that the presence of more than 100% PhH after an
a
Ground spin multiplicities are superscripted (solvation THF
molecules are omitted for clarity; see the SI).
acidic quench is due to the presence of traces of benzene in the
starting material.13
In order to evidence the formation of putative transient
radical intermediates, radical-trapping experiments were
performed using tris(trimethylsilyl)silane [TTMSSH =
(Me3Si)3SiH] as a hydrogen donor. The reduction of either
Fe(acac)n or FeCln (n = 2, 3) by an excess of PhMgBr in the
presence of TTMSSH inhibits the formation of PhPh, usually
observed as the sole byproduct upon the oxidation of PhMgBr
by FeIII or FeII (Scheme 2a), and only leads to the formation of
benzene (Scheme 2c). This suggests that in the absence of
TTMSSH, phenyl radicals Ph• are produced during reduction
of the FeIII or FeII precursor, with the recombination of these
radicals leading to the formation of PhPh. In the presence of a
strong hydrogen donor such as TTMSSH, Ph• is transformed
into benzene by hydrogen-atom transfer in a process faster than
dimerization. Therefore, this shows that the monoelectronic
reduction of Fe(acac)3 by PhMgBr is preferred to a classic
bielectronic reductive elimination from a putative bisphenyl
(Ph)2[FeIII] species, which would directly yield from FeIII to FeI
in a radical-free process (with the formation of PhPh). The
possibility of a reduction involving direct single electronic
transfer proceeding by an outer sphere between PhMgBr and
Fe(acac)3 and leading to a FeII species (namely, [FeII(acac)3]−)
would be unlikely because of the high gap between the
Fe−O−Mg linkage (ΔG = −2.7 kcal·mol−1; Scheme 3b), thus
weakening the coordination of one of the acac− ligands. This
adduct prefigures the release of acacMgBr and the formation of
the C−FeIII bond (computed exothermicity of −15.8 kcal·
mol−1). The ability of PhFeIII(acac)2 to undergo homolytic C−
Fe cleavage and generate Fe(acac)2 in its ground spin state
(quintet) was investigated. A global endothermicity of 16.4
kcal·mol−1 has been obtained (Scheme 3a), which is
incompatible with the experimental conditions (reaction time
of ca. 5 min at room temperature). Moreover, the release of
free Ph• radicals in a solvent such as THF could hardly explain
the quantitative formation of PhPh as a recombination product
because carboradicals can readily abstract hydrogen atoms from
THF. However, it is known that Fe(acac)2 can accommodate
oligomeric structures thanks to the presence of vacant
coordination sites.12 An alternative pathway involving the
formation of an O(acac)-bridged dimer, leading to [Fe-
(acac)2]2, has been computed and allows the reduction to
process with a lower energy barrier (8.1 kcal·mol−1; Scheme
3b). In this dimeric process, the phenyl radical Ph• and FeIII-
−
bonded C6H5 ligand can undergo in-cage recombination to
lead to the formation of PhPh. The further reduction of
Fe(acac)2 to lower oxidation states can then proceed, most
likely via the formation of an intermediate organo-FeII complex,
similar to the FeIII to FeII reduction route. The titration of the
biphenyl obtained by the reduction of FeCl2 by 30 equiv of
PhMgBr showed that, after 30 min at room temperature and
acidic quenching, ca. 1 equiv of PhPh was obtained per 1 mol of
iron, which corresponds to an average transfer of 2 electrons
per 1 mol of FeII (Scheme 2b).13
oxidation potential of PhMgBr in THF (ca. 0.0 V vs SHE14a
)
and the reduction potential of Fe(acac)3 (−0.49 V vs
SHE14b).14c−e DFT calculations were carried out to investigate
the reduction of Fe(acac)3 by inner-sphere-type electronic
−
transfers. Coordination of C6H5 , transferred from PhMgBr
onto the iron center, leading to PhFeIII(acac)2, is thermody-
namically favored (exothermicity of 15.8 kcal·mol−1; Scheme
3a). However, a second phenyl anion transfer, leading to
[Ph2FeIII(acac)2]−, requires an overall span of 27 kcal·mol−1,
precluding its formation under the reaction conditions (−30 to
+25 °C; Schemes 1 and 2). This explains why bielectronic
reductive elimination directly connecting FeIII and FeI oxidation
states does not occur here.
In a previous contribution mainly focused on cyclic
voltammetry,14b we demonstrated that chemical reduction of
Fe(acac)3 by PhMgBr in THF led to a mixture of FeII, FeI, and
Fe0 oxidation states depending on the PhMgBr/Fe ratio. In the
presence of a moderate excess of Grignard reagent (from 9 up
to 15 equiv per 1 mol of iron), two well-defined paramagnetic
1
The magnesium-to-iron phenyl anion transfer described in
Scheme 3a can involve the formation of an adduct exhibiting a
species could be characterized by their H NMR spectra. The
addition of a larger amount of PhMgBr (>15 equiv overall per 1
C
DOI: 10.1021/acs.inorgchem.6b02616
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
mol of iron) led to a drastic alteration of the NMR spectra
because no signal could be detected. This was associated with
an evolution of the color of the sample, which switched from
pale green to dark brown, and in these conditions, the
formation of a low-spin FeI species was attested by cw-EPR
spectroscopy and the formula [PhFeI(acac)(THF)]− was
tentatively attributed. However, as highlighted by Bedford,4
DFT-based computational analysis shows that this species
exhibits a strong preference for the quartet spin state and
therefore cannot account for the low-spin FeI complex detected
by EPR spectroscopy. In order to investigate more closely the
nature of the low-valent iron complexes obtained in these
systems, we decided to examine and compare the EPR spectra
of frozen solutions containing various FeII and FeIII salts
reduced by PhMgBr in toluene/THF (10:1) mixtures. The
choice of toluene as a major cosolvent has been motivated by
the fact that arene ligands often stabilize iron low oxidation
states.15
Reduction of FeII and FeIII Salts Monitored by EPR
Spectroscopy in toluene/THF (10:1). In addition to the use
of β-diketonate salts, the reduction of iron chloride precursors
(FeCl2 and FeCl3) with PhMgBr was investigated by EPR
spectroscopy. These salts are also excellent catalysts in Kumada
coupling reactions.2 They do not benefit from the presence of
stabilizing acac− ligands and are more electron-deficient, being
thus more easily reduced by PhMgBr. Unlike the reduction of
Fe(acac)3, the reduction of FeCl3 by 1 equiv of PhMgBr
provides a mixture of a low-spin FeI complex and high-spin FeIII
unidentified species, as attested by EPR spectroscopy (Scheme
4 and the SI). These data support the formation of transient
polyphenyl-FeIII species,16 which can evolve to the formation of
FeI.
Figure 1. cw-EPR spectrum (10:1 toluene-h8/THF-h8 glass) of FeCl2
+ 30 equiv PhMgBr (frozen after 2 min at 25 °C). Inset: Spectrum
under nonsaturating conditions of the g = 2 region (FeI species; black,
exptl; red, simul.). Experimental conditions: microwave frequency 9.50
GHz; microwave power 1 mW and 0.25 mW (inset); field modulation
amplitude 1 and 0.25 mT (inset); field modulation frequency 100
kHz; T = 10 and 60 K (inset). Simulation parameters: g = 2.2059,
2.0230 and 2.0003; Hstrain (MHz) = 50.5, 15.8, and 0.0; peak-to-peak
line width 1 mT.
Table 1. Selection of EPR g Values for Low-Spin FeI
Mononuclear Complexes
Scheme 4. Reduction of FeIII into FeI by 1 equiv of PhMgBr
in Toluene/THF (10:1)
entry
complex
g values
ref
1
2
3
(p-tolyl)(1,2-C6H4(PPh2)2)2Fe
(η6-C6H6)(NacNac)Fe
2.25/2.09/2.08
2.20/2.01/1.98
6a
a
17a
17b
b
Na(trop2dae)(THF)3Fe
2.16/2.06/2.01
a
b
NacNac = HC(C[Me]N-(2,6-diisopropylphenyl))2. trop = 5H-
dibenzo[a,d]cyclohepten-5-yl; dae = NCH2CH2N.
A comparison of the X-band EPR spectra of samples
resulting from the reduction of Fe(acac)3, FeCl2, and FeCl3 by
a large excess of PhMgBr (30 equiv) carried out at room
temperature (reaction time: 2 min) shows that the same S = 1/2
species is formed, with g values of 2.21, 2.02, and 2.00 [see
Figure 1 for the case of FeCl2 and the SI for the spectra
recorded using FeCl3 and Fe(acac)3 as precursors]. Interest-
ingly, a similar EPR spectrum exhibiting very close g values
(2.20, 2.02, and 1.99) was obtained when the reduction of
Fe(acac)3 was carried out in pure THF.14b This signal is
mechanism for generation of the FeI oxidation state can be
evidenced. The reduction of FeCl2 by 30 equiv of PhMgBr
provided a mixture containing a well-defined FeI species and
high-spin FeIII complexes, as attested by EPR spectroscopy
(Figure 1). This shows that FeI and FeIII oxidation states can be
provided by a fast disproportionation process of intermediate
polyphenyl-FeII complexes.19a A similar hypothesis was
formulated by Bedford et al., who reported that the benzylic
ate-complex [FeII(benzyl)3]− yields FeIII and FeI species by a
disproportionation pathway.19b Koszinowski et al. also reported
that [Ph4FeIII]− and [Ph3FeII]− could be detected concom-
itantly by electrospray ionization mass spectroscopy (ESI-MS)
when FeII in situ generated species reacted with an excess of
PhMgCl.19c
1
characteristic of a S = /2 transition-metal-based system with
resolved g anisotropy and is consistent with the formation of a
low-spin FeI mononuclear species. Moreover, the three g values
are in good agreement with other reported EPR data for similar
low-spin FeI mononuclear systems (Table 1).6a,17
The obtention of the same spectra regardless of the nature of
the precursor implies that all of the acac− ligands are displaced
in the bielectronic reduction of Fe(acac)3. A redistribution of
the acac− and Ph− ligands is indeed observed, with the harder
ligand (acac−) being transferred from iron to magnesium and
A quantitation of the FeI species obtained by double
integration of its EPR signal and a comparison with standard
solutions of CuSO4 recorded under the same nonsaturating
conditions demonstrated that, after a reaction time of 2 min at
room temperature (Figure 1), the FeI species represents about
15% of the overall iron quantity. This indicates that the most
important part of the metal is under other forms or oxidation
states, including EPR-silent species.
−
the softer ligand (C6H5 ) being transferred from magnesium to
iron: [trans-(THF)2Mg(acac)2]μ2(O,O)FeBr2 was obtained at
−30 °C and characterized by single-crystal X-ray diffraction.18
When a FeII salt is used as a precursor, an alternative
D
DOI: 10.1021/acs.inorgchem.6b02616
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
Figure 2. (a) cw-EPR spectrum (10:1 toluene-h8/THF-h8 glass) of FeCl2 + 30 equiv PhMgBr frozen after 5 min (black; vertical scale 1/1) or 24 h
(blue; vertical scale 5/1) at 298 K. Inset: Zoom on the gmax area. Experimental conditions: microwave frequency 9.40 GHz; microwave power 0.06
mW (blue) and 0.02 mW (black); field modulation amplitude 0.5 mT (blue) and 0.2 mT (black); field modulation frequency 100 kHz; temperature
50 K. (b) Wide-field-range spectrum of the 24 h sample. Experimental conditions: microwave frequency 9.40 GHz; microwave power 1 mW; field
modulation amplitude 0.5 mT; field modulation frequency 100 kHz; temperature 10 K.
Stability of the FeI Oxidation State. It is worth noting
that the FeI complex, formed upon the reduction of FeCl2 by
PhMgBr, proved to be unstable over time. Indeed, a second
minor complex is sometimes observed after 5 min at room
temperature (Figure 2a, black, gmax = 2.22). After 24 h at room
temperature, the overall FeI quantity only represents 20% of the
behavior of a spin S = 0 system, and all spectra could be
simulated under this assumption with the following nuclear
parameters: δ = 0.44 mm·s−1 and ΔEQ = 0.93 mm·s−1. This
simulation accounts for ca. 85% of the iron present in the
sample (Figure 3, red lines). However, it does not totally
reproduce parts of the spectrum on both sides of the central
multiplet (velocity domains from −4 to −3 mm·s−1 and from
+3 to +4 mm·s−1). Moreover, these features are both field- and
temperature-dependent (increase at 20 K). This suggests the
occurrence of a minor additional species with a half-integer
spin. The whole spectrum could be simulated by considering
that the sample is a mixture of a spin S = 0 species and a spin S
starting FeI quantity detected in EPR (Figure 2a, blue, gmax
=
2.20, 2.21, and 2.22). The system evolves toward the formation
of clusters, potentially nanoparticles, as suggested by new broad
EPR resonances observed (Figure 2b) over a wide range of
magnetic field values (1 T).
The evolution of the intensity of the FeI signal in EPR
spectroscopy clearly shows that this complex is unstable over
time. Importantly, the EPR spectrum recorded after 24 h
suggests that the system evolved toward paramagnetic clusters.
In order to investigate the formation and the fate of the EPR-
silent integer spin species, the reduction of isotopically enriched
1
= /2 species (Figure 3, blue lines) using the g values deduced
from the EPR experiment and literature hyperfine values (Table
2).20 The nuclear parameters for this paramagnetic species are δ
= 0.28 mm·s−1 and ΔEQ = 2.01 mm·s−1. Owing to the low
signal-to-noise ratio of the spectrum and the intrinsic similarity
of the spectra of both components, a precise determination of
their respective contributions is not achievable; however, the
content of the S = 1/2 species can be estimated within the range
57FeII precursors was investigated by Mossbauer spectroscopy,
̈
thus allowing an overview of the whole distribution of iron
oxidation states reached in these systems.
̈
of 10−20%. Mossbauer experiments are thus fully consistent
Mossbauer Spectroscopy Investigations. Because bro-
mine atoms strongly absorb γ radiation and are thus highly
̈
with the EPR analysis pointing to the formation of a major S =
0 species (EPR-silent) and a minor S = 1/2 species. The nuclear
parameters of the major iron center are not unambiguous, as is
often the case for low-spin systems, and could be associated
with FeII,8 FeI,21 or Fe0 species.22 However, formation of the
Fe0 oxidation state as a major species is consistent with the
overall transfer of 2 electrons per iron ion (vide supra). In
addition, the nuclear parameters of the diamagnetic species
evidenced herein (δ = 0.44 mm·s−1 and ΔEQ = 0.93 mm·s−1)
are similar to those published for related arene-stabilized Fe0
species: (η6-C6H6)(η4-C6H8)Fe0 (δ = 0.51 mm·s−1 and ΔEQ =
0.94 mm·s−1) and (η6-C6H6)(η4-C6H6)Fe0 (δ = 0.64 mm·s−1
and ΔEQ = 1.04 mm·s−1).22a It, moreover, echoes the report of
Neidig, who showed that the reduction of SciOPP-stabilized
FeII precursors by PhMgBr led at room temperature to a similar
detrimental to the quality of Mossbauer spectra, we investigated
̈
the reduction of isotopically enriched 57FeCl2 with an excess
(30 equiv) of PhMgCl in toluene/THF (10:1) for 3 min, inside
an inert-atmosphere box. Mossbauer spectra of the frozen
̈
sample (Figure 3) were recorded at low temperatures (4.2−20
K) in the absence (Figure 3a) or presence (Figure 3b,c) of a
high magnetic field (7 T) applied parallel to the γ-rays. In order
to circumvent as efficiently as possible the fast formation of iron
aggregates in the reduction process, spectra have been run at
low 57Fe concentration (2.5 mM).
In the absence of an applied field, the spectrum appears as a
broad quadrupole doublet, and a similar doublet was observed
at 80 K. The application of a magnetic field splits the doublet
over a small velocity range (−1 to +2 mm·s−1). This is a typical
E
DOI: 10.1021/acs.inorgchem.6b02616
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
spectroscopy, which shows that the overall concentration of
FeIII is significantly smaller than that of FeI. On this basis, it can
be suggested that the most important part of FeI species is not
formed by the disproportionation of polyphenyl-FeII inter-
mediates.
Stabilization of “elemental iron atoms” at the zero oxidation
state by arene and alkene neutral ligands [leading, e.g., to (η6-
arene)(η4-diene)Fe0 complexes] has already been reported in
the past.22−24 These molecular species can be obtained by
reducing bisarene-ligated FeII precursors with monoelectronic
reductants such as sodium amalgam23 or by cocondensing
elemental iron atoms and organic ligands onto low-temperature
frozen surfaces.24 Three different modes of coordination can be
accommodated by Fe0 centers in the presence of arene ligands
depending on their hapticity. This can lead to the formation of
20-electron (Scheme 5a), 18-electron (Scheme 5b), or 16-
electron (Scheme 5c) species.
Scheme 5. Possible Coordination Modes (η6 or η4) of Fe0
Atoms by Arene Ligands
Binding modes of the arene ligand strongly depend on the
nature of both the arene and temperature. Astruc et al.
demonstrated by Mossbauer spectroscopy that the reduction of
̈
FeII to the Fe0 oxidation state in mesitylene led to formation of
the bis-η6 corresponding adduct (η6-C6Me6)2Fe0 (Scheme 5a, R
= Me6).23 It has been shown by low-temperature metal-atom-
matrix UV−visible and IR investigations that iron atoms
condensed in benzene vapor led to formation of the 18-electron
complex (η6-C6H6)(η4-C6H6)Fe0 (Scheme 5b, R = H6) as the
sole product (in the range of T = 50−240 K).22a,24b On the
other hand, iron atoms condensed in toluene vapor could
exhibit two coordination modes. A 18-electron (η6-C6H5Me)-
(η4-C6H5Me)Fe0 complex is favored at medium temperatures
(90−240 K; Scheme 5b, R = Me), whereas a 16-electron
complex (η4-C6H5Me)2Fe0 is obtained at lower temperatures
(<77 K; Scheme 5c, R = Me).24b Both complexes lead to the
formation of clusters upon warming. In our case, it seems
unlikely that a bis-η6-toluene coordinated Fe0 complex is
obtained because the analogue species (η6-C6Me6)2Fe0 and (η6-
C6H6)2Fe0 exhibit much higher isomer shifts (respectively 1.06
and 1.22 mm·s−1) and higher quadrupolar splits (respectively
−1.45 and +1.84 mm·s−1).22a,23 More specifically, it has been
Figure 3. Mossbauer spectra of frozen solutions resulting from the
̈
reduction of 57FeCl2 in toluene/THF by PhMgCl (30 equiv; overall
57Fe concentration 2.5 mM) after 3 min at room temperature: (a) T =
5 K, no magnetic field; (b) T = 4.2 K, 7 T; (c) T = 20 K, 7 T. The low
signal-to-noise ratio is due to the diluted conditions (2.5 mM 57FeCl2)
as well as the large excess of chloride anions.
diamagnetic Fe0 arene-stabilized species (η6-biphenyl)-
Fe0(SciOPP) with a close isomer shift and a higher quadrupolar
split (δ = 0.44 mm·s−1 and ΔEQ = 1.75 mm·s−1).8 It is
interesting to note that the high-spin FeIII species detected in
EPR spectroscopy (Figure 1) were not detected in Mossbauer
demonstrated by Mossbauer spectroscopy that (η6-C Me ) Fe0
̈
̈
6
6 2
Table 2. Best-Fit Parameters Deduced from the Simulation of the Mossbauer Spectra (See Figure 3 and the SI) Taking into
̈
Account a Mixture of S = 0 (Fe0) and S = 1/2 (FeI) Species Using the Experimental EPR g Values Reported Herein and Nuclear
Hyperfine Values Consistent with the Literature Data20
entry
iron oxidation state
spin
gx,y,z
δ (mm·s−1
)
ΔEQ (mm·s−1
)
Γ (mm·s−1
)
η
Ax (T)
Ay (T)
Az (T)
%
5%
1
2
Fe0
FeI
0
0.44
0.28
0.93
2.01
0.60
0.48
0.97
0.63
85
15
1
/
2.21 (gx)
2.02 (gy)
2.00 (gz)
19
13
18
2
F
DOI: 10.1021/acs.inorgchem.6b02616
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
Table 3. Calculated and Observed Isomer Shifts (δ) for Several Bisarene- or Bisdiene-Fe0 Complexes
δ(calcd) (mm·s−1
)
δ(exptl) (mm·s−1
)
ref
−3
entry
complex
spin
ρ0(calcd) (e·a0
)
1
2
3
4
5
6
7
8
(η6-C6Me6)2Fe0
1
0
0
0
0
0
0
0
11826.87
11827.85
11827.87
11827.83
11827.88
11827.85
11828.35
11828.04
0.99
0.58
0.57
0.59
0.57
0.58
0.38
0.50
1.06
0.64
0.51
23
(η6-C6H6)(η4-C6H6)Fe0
22a
22a
(η6-C6H6)(η4-C6H8)Fe0
(η6-C6H5Me)(η4-C6H5Me)Fe0
(η6-C6H5Me)(η4-C6H5Me)Fe0
(η6-C6H5Me)(η4-C6H5Me)Fe0
(η4-CH2CMeCHCH2)2Fe0
(η4-C6H5Me)2Fe0
accommodated a spin triplet state (S = 1).23 This is in stark
contrast with the electronic structure of the Fe0 complex
discussed here, which is a singlet spin state (Figure 3). The
possibility of the formation of (η6,η4)- or (η4,η4)-bistoluene-
ligated Fe0 complexes has been further investigated by
theoretical modelings.
(η6-C6H5Me)(η4-C6H5Me)Fe0 complexes (entries 4−6), and
the corresponding isomer shift (0.50 mm·s−1) is in good
agreement with the experimentally observed data (δ = 0.44
mm·s−1; Figure 3). It must be stated that the two η4-
coordinated toluene ligands do not play a symmetric role: one
of them is strongly twisted, whereas the second one is almost
planar (Scheme 6). The η4-coordination onto the planar
toluene is due to a displacement of the iron atom, which is
slightly shifted from the pseudo-C6-symmetry axis of the η6
ligand.
̈
Analysis of the Mossbauer Spectra: Insights from DFT
Calculations. Mossbauer parameters can indeed be easily
̈
evaluated using straightforward DFT techniques, which allow
computation of the electronic density at the 57Fe nucleus,
denoted as ρ0 (unit: e·a0−3, where a0 is the Bohr radius). A
reliable estimation of the isomer shift (δ) can thus be obtained
with a linear extrapolation from ρ0.25a To do so, a benchmark of
Fe0 and FeI complexes25b,c has been used to calibrate the
chosen level of theory. All calculations were performed with
ORCA 3.0.3 software, using the BP86 functional in the gas
phase, associated with the TZVP basis set for all atoms except
iron, which was treated with the CP(PPP) enlarged basis set.
This choice has been motivated because it allows a good
compromise between the accuracy and computational time. It
has indeed been demonstrated that the use of the CP(PPP)
basis set on iron allowed a more accurate linear fit between ρ0
and δ for a very modest additional computational cost.25a
The level of theory used in this benchmark allowed us to
compute the isomer shift of the already reported complexes
(η6-C6Me6)2Fe0, (η6-C6H6)(η4-C6H6)Fe0, and (η6-C6H6)(η4-
C6H8)Fe0 with good accuracy (Table 3, entries 1−3). For
model complexes involving η4 toluene-coordinated species, the
position of the methyl group (borne by C1 atom) in the η4
ligand did not have a significant impact on either the calculated
isomer shift or the average Fe−Carene distances; see Table 3,
entries 4 (η4-C1C2C3C4), 5 (η4-C1C2C3C6), and 6 (η4-
C2C3C4C5). It can be noted that these species exhibit isomer
shifts very close to their bis-C6H6 analogue (Table 3, entry 2).
A noticeable trend that can be drawn from this computa-
tional analysis is that the average Fe−Carene bond is slightly
longer for η6-coordinated arene ligands than for their η4
analogues. In the case of (η6-C6H5Me)(η4-C6H5Me)Fe0
Scheme 6. Computed Structure of (η4-C6H5Me)2Fe0,
Suggested as a Model for the Diamagnetic Fe0 Species
a
Detected in Figure 3
a
Selected metric data for the η4-(C2C3C4C5) toluene moiety: Fe−C1 =
2.15 Å; Fe−C2 = 2.08 Å; Fe−C3 = 1.95 Å; Fe−C4 = 1.95 Å; Fe−C5 =
2.08 Å; Fe−C6 = 2.14 Å; Fe−Cn bonds (n = 2−5) frozen during
optimization.
Suggestion of a bisarene-ligated Fe0 species as a major
complex obtained in the conditions depicted above [reduction
of FeXn salts by 30 equiv of PhMg(Hal), 2 min in toluene] is
therefore in good agreement with the literature24b and with the
6
(Table 3, entries 4−6), the average computed Fe−Cη ‑arene
4
distance is 2.11 Å, whereas the average Fe−Cη ‑arene distance is
2.06 Å. This trend is confirmed by the optimized structure of
the model complex (η4-CH2CMeCHCH2)2Fe0 (Table 3, entry
4
7), which features an average Fe−Cη ‑diene distance close to 2.04
Mossbauer spectroscopy analysis reported herein (using
̈
Å. A smaller Fe−Csp distance leads to a higher computed
57FeCl2 as a precursor). This species evolves toward the
formation of paramagnetic aggregates, as evidenced by the
strongly distorded EPR spectrum in Figure 2b. However, the
computed isomer shift δ is extremely sensitive to the Fe−Carene
bond distances and consequently to the theoretical level applied
(δ variation of more than 30% for less than 5% of difference
between the average Fe−Carene distances in the structures
depicted in Table 3, entries 4−6 and 8). Therefore, care must
2
electronic density ρ0 and, therefore, to a smaller computed
isomer shift [0.38 mm·s−1 for (η4-CH2CMeCHCH2)2Fe0
(entry 7) versus 0.57−0.59 mm·s−1 for (η6-C6H5Me)(η4-
C6H5Me)Fe0 (entries 4−6)]. Taking this into account, analysis
of the bis-η4-arene complex (η4-C6H5Me)2Fe0 nuclear param-
eters was performed. The computed electronic density ρ0 was
significantly higher (11828.04 e·a0−3, entry 8) than that for the
G
DOI: 10.1021/acs.inorgchem.6b02616
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
be taken in the interpretation of the experimental data,
especially because of the low signal-to-noise ratio of the spectra
displayed in Figure 3. An average error of 0.05 mm·s−1 on the
simulation of the nuclear parameters should be acknowledged.
Consequently, it can be anticipated that several geometrically
close species featuring arene-ligated Fe0 centers that could fall
in this error range coexist in solution.
1H NMR Spectroscopy. The formation of transient arene-
coordinated Fe0 species was also suggested by 1H NMR
spectroscopy when Fe(acac)3 was used as a precursor in
toluene-d8. At low iron concentrations (0.7 mM) and low
temperature (−40 °C), the reduction of Fe(acac)3 by PhMgBr
(50 equiv) in the presence of PhPh as an additive (50 equiv of
which were added to ensure the presence of a large excess of
arene ligands in the bulk) led to the observation of transient
resonances in the 4−5 ppm area (δ = 5.00, 4.72, 4.45, 4.31, and
4.15 ppm; see the SI). Such resonances are usually observed in
this range for diamagnetic Fe0 arene-coordinated species, with
arene protons being upfielded by retrodonation from the metal:
Ittel et al. reported that the arene proton resonances of the
analogue complex (η6-toluene)Fe0(butadiene) were observed
between 4.66 and 4.87 ppm.24a The number of resonances
observed in our conditions (vide supra) also suggests that
multiple arene-coordinated Fe0 species are obtained in the bulk
upon chemical reduction of the precursor by PhMgX (X = Cl,
Br). Consistent with the evolution of these Fe0 complexes
toward paramagnetic aggregates, the resonances described
above quickly disappeared. The only 1H NMR signals observed
after 10 min at −40 °C or upon slow warming at room
temperature in the 4−6 ppm area belonged to the free acac
ligand (δ = 5.25 ppm, CH), which was released in the reduction
process (Scheme 2a). Concomitantly, the color of the solution
evolved from very pale-red-brown to a deep-dark-brown.
The nature of the minor FeI complex evidenced in EPR and
Figure 4. HYSCORE contour plots for FeI (FeCl2 + 30 equiv of
PhMgBr; 10:1 toluene-h8/THF-h8 glass) at gmax (315 mT), gmid (343
mT), and gmin (347 mT). Experimental parameters: microwave
frequency 9.71 MHz; τ = 120 ns; shot repetition time 5 ms; T = 10 K.
Mossbauer spectra (Figures 1 and 3) is discussed in the next
̈
section. As evidenced by spin titration in both aforementioned
techniques, this oxidation state only represents a small fraction
of the overall iron quantity (ca. 15%).
progressively formed in the bulk by oxidation of the Grignard
reagent. However, it cannot be excluded that biphenyl acts as a
η6-coordinating arene ligand in other solvents, e.g., when the
reduction is carried out in THF. The competition between the
iron coordination by toluene and biphenyl is discussed at the
end of the next section.
Nature of the FeI Complex: Insights from Pulsed-EPR
Spectroscopy. In order to investigate the nature of the FeI
coordination sphere, pulsed-EPR experiments (HYSCORE)
were realized in toluene/THF (10:1) glass. HYSCORE is a
powerful 2D pulsed-EPR technique that allows the study of
hyperfine coupling of nuclei coupled to the paramagnetic
species of interest. Figure 4 shows the HYSCORE quadrant
plots obtained for magnetic field values corresponding to the
three principal g values of the FeI complex. Here, ridges well
On the basis of cw-EPR and HYSCORE experiments, it is
thus reasonable to suggest that a η6-arene-stabilized FeI species
is formed upon reduction of the iron precursor by PhMgBr.
The nature of the other ligands of the coordination sphere was
then investigated. FeI complexes often feature strongly σ-
donating ligands, such as β-diketiminate,17a NHCs,21b aryl
groups,26a or trimethylsilylamides.26b In the presence of
1
separated from the H Larmor frequency are observed and
extend over nearly 12 MHz, indicating specific proton coupling
to the FeI ion. These ridges disappear when the same
experiments are performed in fully deuterated toluene
(associated with 10% THF-h8; see the SI), indicating that the
signals can be attributed to the protons of the toluene solvent
in close interaction with the metal ion. Therefore, the signals
observed are compatible with a toluene molecule in η6-
coordination to FeI. The strong affinity of low-valent iron ions
for arene ligands is already known (vide supra), and several
well-defined arene-coordinated FeI complexes have been
reported and, for some of them, investigated with spectroscopic
and theoretical methods.17a,26a
−
strongly coordinating species such as the C6H5 anion, it is
likely that the FeI center is coordinated by anionic phenyl
ligands. To investigate this hypothesis, the reduction of FeCl2
was carried out using isotopically enriched 13C6H5MgBr,
synthesized at a microscale from commercial 13C6H5Br. The
cw-EPR spectrum obtained for the reduction of FeCl2 by 30
equiv of 13C6H5MgBr for 2 min at room temperature is shown
in Figure 5. The resonances are centered on the same g values
as those observed when FeCl2 was reduced with nonenriched
C6H5MgBr (Figure 1) but exhibit significant broadening,
reflecting hyperfine couplings between the FeI ion and 13C
nuclei.
HYSCORE experiments performed in deuterated toluene
unambiguously show that arene coordination of the FeI species
implies ligation of a toluene molecule and not a biphenyl
Because the coupling is not resolved on the cw-EPR
spectrum, we performed HYSCORE measurements on this
H
DOI: 10.1021/acs.inorgchem.6b02616
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
quadrant do not exhibit significant changes in their position
suggesting that the hyperfine coupling of the corresponding 13
C
nuclei is essentially isotropic. On the basis of the distance
between the cross peaks, the value of the coupling was
estimated to be about 9 MHz. The fact that the coupling is
isotropic indicates that some spin density is localized on 13C
nuclei and therefore that at least one 13C6H5− anion is a ligand
to the FeI ion. In the (++) quadrant, the cross peaks are not
resolved and are close to the 13C Larmor frequency, indicating
rather weak hyperfine coupling.
On the basis of these results, it can be proposed that the
carbon bonding the FeI ion (Cipso) bears the strongest spin
−
density in the C6H5 ligands. It is therefore quite strongly
coupled to the iron and is not detectable by HYSCORE. This is
in agreement with the simulated spectrum depicted in Figure 5
that includes a parameter Aiso = 35 MHz (red lines). The two
carbon atoms in the ortho position are those responsible for the
cross peaks observed in the (+−) quadrant with a hyperfine
coupling of 9 MHz, and the carbon atoms in the meta and para
positions are very weakly coupled and detected close to the 13C
Larmor frequency in the (++) quadrant. With this in mind,
simulations were performed for the cw-EPR spectrum of the
sample. Keeping the same parameters (g values, strain, and line
width) as those for the spectrum without 13C labeling (Figure
1, inset), the best fit was obtained when considering two
Figure 5. Experimental (black) and simulated (red) cw-EPR spectrum
(10:1 toluene-h8/THF-h8 glass) of FeCl2 + 30 equiv of 13C6H5MgBr.
Experimental parameters: microwave frequency 9.38 GHz; microwave
power 0.01 mW; field modulation amplitude 0.5 mT; field modulation
frequency 100 kHz; temperature 50 K. Simulation parameters : g =
2.2059, 2.0230, and 2.0003; Hstrain (MHz) = 50.5, 15.8, and 0.0; peak-
to-peak line width 1 mT; two 13C nuclei with Aiso = 35 MHz, four 13
nuclei with A = 9.0, 8.8, and 9.2 MHz.
C
sample. The contour plots obtained at magnetic field values
corresponding to the principal g values are shown in Figure 6.
Signals were observed in both quadrants, corresponding to
strong and weak coupling cases. The cross peaks in the (+−)
−
equivalent C6H5 ligands compared to only one (see the SI).
The simulation is shown in Figure 5 with the experimental data,
and the parameters are reported in the caption. The overall
results obtained with cw- and pulsed-EPR thus allow the
suggested structure, which is represented in Scheme 7 for the S
1
= /2 FeI complex, namely, [(η6-C6H5Me)FeI(Ph)2]−.
Scheme 7. Structure Suggested for the Minor FeI Complex
Obtained by the Reduction of FeXn (X = acac, Cl; n = 2, 3)
by an Excess of PhMgBr in Toluene, on the Basis of cw- and
a
Pulsed-EPR Experiments
a
The starred atoms correspond to 13C atoms, which allowed detection
of the hyperfine FeI−13C couplings when 13C6H5MgBr was used as a
reductant.
In order to support the structure suggested above (Scheme
7), the most likely number of C6H5− ligands, and, therefore, the
most likely topology that can be accommodated by the FeI
center to afford a spin doublet arene-ligated polyphenyl-FeI
complex [(η6-C6H5Me)FeI(Ph)n](n−1)−, was then examined
using DFT calculations.
Theoretical Investigations of the FeI Electronic
Structure. The gap between the doublet and quartet spin
states was evaluated for a benchmark of representative FeI
organometallic complexes reported in the literature and
featuring for some of them η6-arene-coordinated FeI centers
(1a, 1b, 1c, and 1d in Scheme 8).27 Benzene was used as a
Figure 6. HYSCORE contour plots of FeCl2 + 30 equiv of
13C6H5MgBr (10:1 toluene-h8/THF-h8 glass) at gmax (315 mT), gmid
(343 mT), and gmin (347 mT). Experimental parameters: microwave
frequency 9.71 MHz; τ = 140 ns; shot repetition time 3 ms; T = 10 K.
I
DOI: 10.1021/acs.inorgchem.6b02616
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
This suggests that a correct description of the FeI ion
electronic structure cannot be obtained with a single electronic
configuration and is then beyond the limit of an accurate DFT
analysis (see the SI).28 Nevertheless, several trends can be
drawn from this analysis in the case of complexes 1a and 1b.
The experimentally attested stability of the doublet state can
indeed arise from the multiconfigurational character of the FeI
center, counterbalancing the slight quartet stabilization
predicted by DFT methods. Among the three examined
candidates for arene-ligated polyphenyl-FeI complexes (1d,
1e, and 1f) that could account for the doublet species reported
herein, a monophenyl species such as (η6-C6H5Me)FeIPh
seems unlikely because its structural topology is very close to
Power’s terphenyl complex (model 1d), which has a strongly
stabilized quartet ground state, as attested by EPR spectrosco-
py.26a Similar to 1d, DFT computations of (η6-C6H6)FeIPh
indeed return a strong preference for the quartet configuration
as well (18.7 kcal·mol−1 at the PBE0-D3/SDD level).29,30 The
triphenyl species [(η6-arene)FeIPh3]2− (1f) seems also unlikely
because of the strong stabilization of the quartet state (Scheme
8). However, in the case of the bisphenyl analogue 1e, the
doublet−quartet gap is small enough (−1.1 to −9 kcal·mol−1
depending on the functional used) to consider multiconfigura-
tional effects, similar to 1a and 1b. Preliminary computational
analysis run at the CCSD(T) level indeed shows that the
correlation energy is much stronger, by almost 0.1 au, in the
doublet configuration of 1e than in the quartet analogue. Such a
discrepancy between the correlation effects involved in the
doublet and quartet electronic structures can lead to
reinterpretation of the DFT-computed tendencies, mainly
dominated by exchange energies and, therefore, favoring
high-spin multiplicities. Small doublet−quartet DFT-computed
gaps (complexes 1a, 1b, and 1e; Scheme 8) could therefore be
counterbalanced by taking into account the multiconfigura-
tional effects. As highlighted by Bedford,4 an acac-ligated
monophenyl species such as [PhFeI(acac)(THF)]− exhibits a
strongly stabilized quartet configuration with a computed
doublet−quartet gap higher than 20 kcal·mol−1 [DFT level:
PBE0/6-31+G(d)/SDD basis set and pseudopotential for
iron]. This high gap could be overcome with difficulty by
multiconfigurational effects, and, consequently, on the basis of
the experimental and computational work reported herein,
[(η6-C6H5Me)FeIPh2]− seems to be a closer model for the
description of the minor FeI species obtained upon the
reduction of a FeII/III precursor with aryl Grignard reagents.
Moreover, it is consistent with the recent report of Koszinowski
et al., who reported that the [Ph2FeI]− fragment could be
detected by ESI-MS when Fe(acac)3 was reduced with PhMgCl
in THF in cross-coupling conditions (presence of iPrCl as an
electrophile).19c Determination of the structure of this FeI
complex also allowed us to compute an estimation of the
energetic barrier required to ensure disproportionation of the
trisphenylated species [Ph3FeII]− into the FeIII and FeI
oxidation states (thus affording [Ph4FeIII]− and [(η6-C6H5Me)-
FeIPh2]−), which was brought up by Koszinowski and Bedford
(vide supra). At a DFT level (see the SI), this reaction is quite
athermic (ΔE = 0.2 kcal·mol−1) and could therefore occur in
short times at room temperature.31
Scheme 8. ΔEQ−D = EQ − ED DFT Quartet−Doublet Split
(kcal·mol−1; Gas Phase; See the SI for the Assignments) for
a
1a−1f
a
Basis sets for carbon, hydrogen, nitrogen, and phosphorus: see the SI.
model for the coordinated arene solvent in each case. Different
dispersion-fitted functionals (PBE0-D3, B3LYP-D3, and M06)
as well as two pseudopotentials for the iron ion (SDD and
CEP-31G) were used. The complexes [(η6-C6H6)FeI(Ph)2]−
(1e) and [(η6-C6H6)FeI(Ph)3]2− (1f) have been used as models
of η6-arene-coordinated polyphenyl-FeI complexes exhibiting
respectively idealized C2v and C3v environments, in order to be
compared with 1d, a bulkier terphenyl analogue of which
having been experimentally characterized as a quartet spin
ground state.26a
Regarding the energetic features, the DFT methods
reproduce well the stabilization of the quartet states (1c and
1d), with a significantly high quartet−doublet gap (23.8−34.0
kcal·mol−1, 1c, and 16.0−33.4 kcal·mol−1, 1d). However, the
stabilization of doublet states is not efficiently reflected by DFT
analysis. All methods returned to a ground quartet state for
complex 1a (stabilized by 3.1−7.6 kcal·mol−1). Either a slight
preference for the doublet state for 1b (by 2.2−8.9 kcal·mol−1,
PBE0-D3 and B3LYP-D3) or a slight preference for the quartet
state (by 1.4−1.5 kcal·mol−1, M06) is returned. Moreover,
natural bond orbital (NBO) analysis shows that natural orbital
occupancy numbers (NOONs) for the d-block molecular
orbitals strongly differ from integers (0, 1, or 2) for the
computed FeI complexes (Table 4). The computed NOONs
for FeI complexes 1a, 1b, 1c, 1d, 1e, and 1f (entries 1−6) can
be compared with the computed NOONs for Fe(acac)3 (entry
7). The latter is essentially monoconfigurational (high-spin d5
complex), and its electronic structure is therefore accurately
described by DFT.
Table 4. DFT-Computed NOONs for d-Block Orbitals of
Complexes Depicted in Scheme 8 (PBE0, Gas Phase, SDD,
and Pseudopotential for Iron)
entry
complex
21a
21b
41c
41d
21e
41f
6Fe(acac)3
d-NOONs
1.73
1.47
1.27
1.14
0.97
1.14
1.14
0.81
1.20
1.29
1.57
1.73
1.08
1.26
1.16
2
1
2
3
4
5
6
7
dxy
dxy
dxy
dxy
dxy
dxy
dxy
d
xz
d
xz
d
xz
d
xz
d
xz
d
xz
d
xz
d
yz
d
yz
d
yz
d
yz
d
yz
d
yz
d
yz
d
d
d
d
d
d
d
21.79dz21.93
21.32dz21.78
21.31dz21.09
21.31dz21.64
21.78dz21.62
21.30dz21.30
21.14dz21.05
x −y
0.86
1.60
1.17
1.61
1.37
1.16
2
x −y
2
x −y
The plausibility of the bisphenylated structure [(η6-
C6H5Me)FeIPh2]− depicted above (Scheme 7) has, moreover,
2
x −y
2
x −y
been strengthened by the computation of its Mossbauer isomer
̈
2
x −y
shift following the DFT-based method mentioned in the
2
x −y
previous section (ORCA 3.0.3 software; BP86−TZVP/CP-
J
DOI: 10.1021/acs.inorgchem.6b02616
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
(PPP) for the iron atom). According to the calibration curve
using the benchmark of Fe0 and FeI complexes described
earlier,25b the computed density at the 57FeI nucleus (ρ0) for
the structure [(η6-C6H5Me)FeI(Ph)2]− corresponds to a
predicted isomer shift of 0.24 mm·s−1 (computed ρ0:
11828.69 e·a0−3). This is in good agreement with the
In the one-pot catalytic conditions described above (Scheme
10a), the first Mossbauer spectrum with 57FeCl2 as an iron
̈
source was recorded (Figure 7a).
experimental value obtained from Mossbauer spectra (Figure
̈
3), which was δ = 0.28 mm·s−1.
Competition between η6-Toluene- and η6-Biphenyl-
FeI Ligations. As discussed earlier, the coordination of a η6-
biphenyl molecule onto the FeI ion cannot be ruled out in
nonaromatic solvents such as THF. The competition between
the coordination of toluene and biphenyl (i.e., the equilibrium
between [(η6-C6H5Me)FeIPh2]− and [(η6-C6H5Ph)FeIPh2]−)
has been investigated by DFT (Scheme 9). The equilibrium is
Scheme 9. Competition between the Toluene-FeI and
Biphenyl-FeI Ligation
shifted toward the formation of the biphenyl species PhPh1e,
with a slight stabilization of ca. 4.9 kcal·mol−1 with respect to
the toluene analogue Tol1e. Therefore, it is likely that
coordination of the biphenyl molecule accumulated by the
oxidation of PhMgBr can occur in nonaromatic solvents, with
this ligand being replaced by a toluene molecule when the latter
is used as a solvent.
Very recently, the X-ray structure of the biphenyl complex
PhPh1e was obtained by Hu et al. by reducing the iron−sulfur
cluster [Fe4S4Cl4]2− with an excess of phenyllithium. The
authors, moreover, also assessed a low-spin ground state for this
complex.32
Figure 7. Mossbauer spectra at 80 K in toluene: (a) solution
̈
containing 2.5 mM 57FeCl2 and HetArCl (20 equiv), frozen after the
addition of PhMgCl (10 equiv; reaction time of 1 min at room
temperature); (b) solution containing 57FeCl2 reduced in situ by
PhMgCl (10 equiv; reaction time of 3 min at room temperature)
followed by the addition of HetArCl (20 equiv) and freezing (reaction
time of 1 min at room temperature). See the remarks regarding the
low signal-to-noise ratio in the caption of Figure 3.
Reactivity of In Situ Generated Fe0 and FeI Complexes
in Aryl−Heteroaryl Kumada Cross-Couplings. The
relevance of the complexes described in this Article as possible
active species in aryl−heteroaryl Kumada cross-couplings has
been examined. 2-Chloro-5-(trifluoromethyl)pyridine (He-
tArCl) has been used as an electrophilic partner in catalytic
conditions (Scheme 10): chloropyridines proved indeed to be
efficient cross-coupling partners in coligand-free iron-catalyzed
methodologies.2e
The spectrum displayed in Figure 7a exhibits a signal that
could be simulated with two doublets. The first one (Figure 7a,
red lines) accounts for ca. 39% of the overall iron quantity, with
the following nuclear parameters: δ = 0.53 mm·s−1 and ΔEQ =
0.98 mm·s−1. This signal is well-fitted to the Fe0 species
described in the previous section, which featured similar
parameters (Figure 3). The second doublet (Figure 7a, green
lines) accounts for ca. 51% of the iron quantity with the
following parameters: δ = 1.28 mm·s−1; ΔEQ = 2.46 mm·s−1.
This doublet corresponds to the regenerated FeII precursor
(57FeCl2) with nuclear parameters corresponding to a high-spin
FeII center (see the SI for the spectrum of 57FeCl2). These data
are gathered in Table 5 (entry 1).
Scheme 10. Aryl−Heteroaryl Kumada Cross-Coupling
between PhMgCl and HetArCl Investigated at Room
Temperature (Nonoptimized Conditions): (a) Catalytic
One-Pot Conditions; (b) Step-by-Step Conditions with
a
Generation of Fe0 Prior to the Addition of HetArCl
In agreement with the previous analysis, no transmetalated
phenyl-FeII species can be detected during catalysis (Figure 7a).
However, at this point, the origin of the regeneration of the FeII
precursor is unclear. FeII can indeed be provided by two
drastically different pathways: it could either be formed by the
reaction of the reduced Fe0 species with the electrophile
HetArCl (Scheme 11, path b) or come from the reaction of
HetArCl with a transient, short-lived, phenylated organo-FeII
intermediate formed prior to the Fe0 oxidation state (Scheme
a
The HetArPh yield was calculated with respect to HetArCl.
K
DOI: 10.1021/acs.inorgchem.6b02616
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
reactivity of transient, short-lived nonstabilized polyphenyl-
FeII intermediates [PhnFeII](n−2)− must be investigated. It was
indeed demonstrated that the SciOPP-stabilized analogues
(SciOPP)FeII(Ph)n were extremely reactive in Kumada and
Suzuki−Miyaura cross-couplings8 and that the reactivity of the
zerovalent complex (η6-biphenyl)Fe0(SciOPP) was totally
overwhelmed by the phenylated FeII complexes. Such
investigations are currently in progress, and the results will be
reported in due time.
Table 5. Best-Fit Parameters Deduced from the Simulation
of the Mossbauer Spectra Displayed in Parts a (Entry 1) and
̈
a
b (Entry 2) of Figure 7
iron oxidation
state
δ (mm· ΔEQ (mm· Γ (mm·
s−1 s−1 s−1
%
entry
1
spin
)
)
)
5%
Fe0
FeII
Fe0
FeII
0
2
0
2
0.53
1.28
0.46
1.24
0.98
2.46
0.75
2.55
0.55
0.55
0.50
0.30
39
51
24
72
2
a
An average error of 0.05 mm·s−1 must be acknowledged in the
CONCLUDING REMARKS
■
simulations because of the low signal-to-noise ratio of the experimental
spectra.
We established in this paper that several reduction mechanisms
could take place when FeII or FeIII salts were reduced by aryl
Grignard reagents in the absence of a stabilizing coligand. In
THF, Fe(acac)3 undergoes a first ligand exchange with
PhMgBr, leading to PhFe(acac)2, which is prompted to lead
to the formation of PhPh and Fe(acac)2 in a dimeric pathway,
as suggested by DFT. The formation of PhPh involves a
monoelectronic reduction of the FeIII center, which can
proceed by the homolytic cleavage of Ph−FeIII bonds. A
second phenyl transfer, which would lead from PhFe(acac)2 to
[Ph2Fe(acac)2]−, requires an overall high energetic span that
cannot be reached in the experimental conditions, making the
possibility of a classic bielectronic reductive elimination directly
connecting the FeIII and FeI oxidation states unlikely. FeI can be
obtained by further reduction of the FeII intermediates, as
attested by cw-EPR spectroscopy in 10:1 toluene/THF
mixtures. Unlike Fe(acac)3, the electron-deficient precursors
FeCl2 and FeCl3 are more easily reduced by PhMgBr to the FeI
level: FeCl3 only requires 1 equiv of PhMgBr to provide FeI
oxidation state, and FeCl2 can also provide FeI by
disproportionation of the intermediate polyphenyl-FeII species.
It was shown by EPR spectroscopy that the reduction of either
Fe(acac)n or FeCln (n = 2, 3) led to a common low-spin FeI
species regardless of the nature of the starting iron salt,
suggesting the possible formation of a polyphenyl-FeI species.
This complex has been proven to be unstable over time and
represents, moreover, a minor fraction (ca. 15%) of the overall
iron amount, as attested by EPR-based spin titration and
Scheme 11. Two Possible Pathways for the Regeneration of
FeCl2 under Catalytic Conditions
11, path a). In the second case, the bisarene-ligated Fe0 species
would be an off-cycle intermediate.33
A second Mossbauer spectrum has therefore been recorded
̈
in order to selectively examine the reactivity of the Fe0 species
(Figure 7b), in the step-by-step conditions depicted in Scheme
10b. 57FeCl2 has been reduced by an excess of PhMgCl to the
Fe0 oxidation state, and an excess of HetArCl has been added
to the resulting solution, which was frozen after 1 min at 25 °C.
It turned out that the same Fe0 and FeII species are observed in
both one-pot (Figure 7a) or step-by-step conditions (Figure 7b
and Table 5, entry 2). When Fe0 is accumulated prior to the
addition of the electrophile (conditions of Scheme 10b), the
ratio FeII/Fe0 slightly increases (3:1 vs 1.3:1 in the conditions
of Scheme 10a). Therefore, regeneration of the FeII precursor
by the reaction of Fe0 reduced species with the electrophile
(Scheme 11, path b) seems to be operative at room
temperature within a couple of minutes, and formation of the
cross-coupling product HetArPh was observed concomitantly
(Scheme 10b).
Mossbauer spectroscopy. Mossbauer analysis actually showed
̈
̈
that the distribution of the metallic species is quickly dominated
cw-EPR spectra were also recorded in the conditions
depicted in Scheme 10a, with nonenriched FeCl2 as a precursor.
These spectra showed that the FeI signal was not affected by
the presence of HetArCl. This result therefore suggests that the
minor low-spin FeI species obtained in these coligand-free
systems is actually an off-cycle species and does not contribute
to catalytic processes involving aryl Grignard reagents and
heteroaryl chlorides. In their recent report, Hu et al.
demonstrated that the analogue biphenyl complex [(η6-
C6H5Ph)FeIPh2]− was unreactive toward the addition of
chloroarenes but that stoichiometric C−C cross-coupling
reactions could be observed while using organic bromides as
electrophiles, albeit with low yield and the concomitant
formation of biphenyl.32
There is a strong similarity between the behavior of the
SciOPP-stabilized systems reported by Neidig and the coligand-
free systems investigated in this report. In both cases, the
lowest oxidation state obtained upon the reduction of FeII
precursors by PhMgX (X = Cl, Br) is a Fe0 complex, which can
be used as a precursor in Kumada-type catalytic cross-
couplings. However, before it can be stated that arene-ligated
Fe0 complexes can be on-cycle intermediates in the case of
coligand-free aryl−heteroaryl cross-coupling chemistry, the
after 3 min by diamagnetic Fe0 complexes. Computational
analysis of the experimental Mossbauer data and a comparison
̈
with the already reported Fe0 mononuclear complexes allowed
us to suggest that the major iron species detected herein
featured Fe0 centers that were stabilized by arene ligation,
specifically by toluene when the latter was used as a solvent.
The best model obtained by confronting the experimental
nuclear parameters and computed data was (η4-C6H5Me)2Fe0.
Because of the low signal-to-noise ratio of the Mossbauer
̈
spectra and the average error in the corresponding simulations,
it can however be anticipated that multiple arene-stabilized Fe0
species coexist in solution. The presence of several arene-
ligated Fe0 complexes was also confirmed by low-temperature
1H NMR spectroscopy. These complexes are prompt to
undergo aggregation and finally provide iron nanoparticles.
Structural data were also gained for the minor FeI complex.
Pulsed-EPR spectroscopy (HYSCORE) unambiguously dem-
onstrated that the FeI ion is ligated by a toluene molecule in η6-
coordination. When iron salts were reduced with isotopically
enriched 13C6H5MgBr, cw-EPR showed a significant broad-
ening of the resonances due to hyperfine FeI−13C coupling,
attesting to a ligation of the FeI ion by σ-donating anionic
L
DOI: 10.1021/acs.inorgchem.6b02616
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
−
C6H5 ligands. Analysis of the cw-EPR and HYSCORE
REFERENCES
■
hyperfine patterns suggests that the coordination sphere of
(1) Bedford, R. B.; Brenner, P. B. The development of iron catalysts
for cross-coupling reactions. In Iron Catalysis II; Bauer, E., Ed.;
Springer International: 2015.
the FeI ion involves two σ-ligated C6H5 ligands. This
−
hypothesis was strengthened by DFT calculations and NBO
analysis using different functionals (PBE0-D3, B3LYP-D3, and
M06) as well as two pseudopotentials for iron (SDD and CEP-
31G). Evaluation of the quartet−doublet energetic gap was
computed for a benchmark of low-spin and high-spin FeI
complexes reported in the literature, showing that multi-
configurational effects were predominant for low-spin FeI
species. On the basis of the experimental and computational
results, a bisphenyl structure [(η6-C6H5Me)FeIPh2]− has been
suggested. We, moreover, demonstrated that the [(η6-
C6H5Me)FeIPh2]− complex was not reactive in the aryl−
heteroaryl Kumada cross-coupling conditions because its
concentration remained roughly constant upon the addition
of an heteroaryl chloride. On the other hand, the major
bisarene-ligated Fe0 complex was able to react with HetArCl
and could be used as an efficient precursor for aryl−heteroaryl
Kumada cross-couplings. The results reported in this paper are
focused on the structural nature and reactivity of the low-valent
resting states that were observed at room temperature. The in
situ observation of transient short-lived intermediates at lower
temperatures (such as polyphenyl-FeII species) and their
relevance in similar catalytic processes are currently being
investigated.
(2) (a) Tamura, M.; Kochi, J. K. Vinylation of Grignard reagents.
Catalysis by iron. J. Am. Chem. Soc. 1971, 93, 1487−1489. (b) Smith,
R. S.; Kochi, J. K. Mechanistic studies of iron catalysis in the cross
coupling of alkenyl halides and Grignard reagents. J. Org. Chem. 1976,
41, 502−509. (c) Quintin, J.; Franck, X.; Hocquemiller, R.; Figader
̀
e,
B. Iron-catalysed arylation of heteroaryl halides by Grignard reagents.
Tetrahedron Lett. 2002, 43, 3547−3549. (d) Furstner, A.; Leitner, A.
̈
Iron-Catalyzed Cross-Coupling Reactions of Alkyl-Grignard Reagents
with Aryl Chlorides, Tosylates, and Triflates. Angew. Chem., Int. Ed.
2002, 41, 609−612. (e) Furstner, A.; Leitner, A.; Mendez, M.; Krause,
̈
H. Iron-Catalyzed Cross-Coupling Reactions. J. Am. Chem. Soc. 2002,
124, 13856−13863. (f) Martin, R.; Furstner, A. Cross-Coupling of
̈
Alkyl Halides with Aryl Grignard Reagents Catalyzed by a Low-Valent
Iron Complex. Angew. Chem., Int. Ed. 2004, 43, 3955−3957.
(g) Nakamura, M.; Matsuo, K.; Ito, S.; Nakamura, E. Iron-Catalyzed
Cross-Coupling of Primary and Secondary Alkyl Halides with Aryl
Grignard Reagents. J. Am. Chem. Soc. 2004, 126, 3686−3687.
(h) Nagano, T.; Hayashi, T. Iron-Catalyzed Grignard Cross-Coupling
with Alkyl Halides Possessing β-Hydrogens. Org. Lett. 2004, 6, 1297−
1299. (i) Sherry, B. D.; Furstner, A. The Promise and Challenge of
̈
Iron-Catalyzed Cross Coupling. Acc. Chem. Res. 2008, 41, 1500−1511.
(j) Czaplik, W. M.; Mayer, M.; Cvengros, J.; von Wangelin, A. J.
Coming of Age: Sustainable Iron-Catalyzed Cross-Coupling Reactions.
ChemSusChem 2009, 2, 396−417. (k) Kuzmina, O. M.; Steib, A. K.;
Flubacher, D.; Knochel, P. Iron-Catalyzed Cross-Coupling of N-
Heterocyclic Chlorides and Bromides with Arylmagnesium Reagents.
Org. Lett. 2012, 14, 4818−4821. (l) Kuzmina, O. M.; Steib, A. K.;
Markiewicz, J. T.; Flubacher, D.; Knochel, P. Ligand-Accelerated Iron-
and Cobalt-Catalyzed Cross-Coupling Reactions between N-Hetero-
aryl Halides and Aryl Magnesium Reagents. Angew. Chem., Int. Ed.
2013, 52, 4945−4949. (m) Nakamura, E.; Hatakeyama, T.; Ito, S.;
Ishizuka, K.; Ilies, L.; Nakamura, M. Org. React. 2013, 83, 1. (n) Bauer,
ASSOCIATED CONTENT
■
S
* Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acs.inorg-
chem.6b02616.
Experimental procedures, parameters, and software used
I.; Knolker, H.-J. Iron Catalysis in Organic Synthesis. Chem. Rev. 2015,
̈
for EPR and Mossbauer data analysis as well as Cartesian
̈
115, 3170−3387.
coordinates of optimized structures with the correspond-
ing energies and lowest frequencies (PDF)
(3) Hatakeyama, T.; Nakamura, M. Iron-Catalyzed Selective Biaryl
Coupling: Remarkable Suppression of Homocoupling by the Fluoride
Anion. J. Am. Chem. Soc. 2007, 129, 9844−9845.
(4) Bedford, R. B. How Low Does Iron Go? Chasing the Active
Species in Fe-Catalyzed Cross-Coupling Reactions. Acc. Chem. Res.
2015, 48, 1485−1493.
AUTHOR INFORMATION
■
Corresponding Authors
(5) Furstner, A.; Krause, H.; Lehmann, C. W. Unusual Structure and
̈
*E-mail: jean-marc.latour@cea.fr.
*E-mail: pierre.dorlet@cea.fr.
*E-mail: guillaume.lefevre@cea.fr.
Reactivity of a Homoleptic “Super-Ate” Complex of Iron: Implications
for Grignard Additions, Cross-Coupling Reactions, and the Kharasch
Deconjugation. Angew. Chem., Int. Ed. 2006, 45, 440−444.
ORCID
(6) (a) Adams, C. J.; Bedford, R. B.; Carter, E.; Gower, N. J.;
Haddow, M. F.; Harvey, J. N.; Huwe, M.; Cartes, M. A.; Mansell, S.
M.; Mendoza, C.; Murphy, D. M.; Neeve, E. C.; Nunn, J. Iron(I) in
Negishi Cross-Coupling Reactions. J. Am. Chem. Soc. 2012, 134,
10333−10336. (b) Bedford, R. B.; Carter, E.; Cogswell, P. M.; Gower,
N. J.; Haddow, M. F.; Harvey, J. N.; Murphy, D. M.; Neeve, E. C.;
Nunn, J. Simplifying Iron−Phosphine Catalysts for Cross-Coupling
Reactions. Angew. Chem., Int. Ed. 2013, 52, 1285−1288.
Thibault Cantat: 0000-0001-5265-8179
Pierre Dorlet: 0000-0001-7394-3374
Guillaume Lefevre: 0000-0001-9409-5861
̀
Author Contributions
The manuscript was written through contributions of all
authors. All authors have given approval to the final version of
the manuscript.
(7) (a) Kleimark, J.; Hedstrom, A.; Larsson, P.-F.; Johansson, C.;
̈
Norrby, P.-O. Mechanistic Investigation of Iron-Catalyzed Coupling
Notes
Reactions. ChemCatChem 2009, 1, 152−161. (b) Guisan
́
-Ceinos, M.;
The authors declare no competing financial interest.
Tato, F.; Bunuel, E.; Calle, P.; Cardenas, D. J. Fe-catalysed Kumada-
̀
̃
type alkyl−alkyl cross-coupling. Evidence for the intermediacy of Fe(I)
complexes. Chem. Sci. 2013, 4, 1098−1104. (c) Hedstrom, A.; Izakian,
ACKNOWLEDGMENTS
̈
■
Z.; Vreto, I.; Wallentin, C.-J.; Norrby, P.-O. On the Radical Nature of
Iron-Catalyzed Cross-Coupling Reactions. Chem. - Eur. J. 2015, 21,
5946−5953. (d) Noda, D.; Sunada, Y.; Hatakeyama, T.; Nakamura,
M.; Nagashima, H. Effect of TMEDA on Iron-Catalyzed Coupling
Reactions of ArMgX with Alkyl Halides. J. Am. Chem. Soc. 2009, 131,
6078−6079. (e) Liu, Y.; Xiao, J.; Wang, L.; Song, Y.; Deng, L.
Carbon−Carbon Bond Formation Reactivity of a Four-Coordinate
We acknowledge CEA, CNRS, ANR (Project JCJC SIROCCO-
16 to G.L.), and the French Infrastructure for Integrated
Structural Biology (Grant ANR-10-INSB-05-01 to P.D.). Dr.
́ ́
Pierre Thuery (CEA) and Dr. Christian Herrero (Universite
Paris-Sud, Orsay) are thanked for technical assistance (X-ray
diffraction and EPR spectroscopy).
M
DOI: 10.1021/acs.inorgchem.6b02616
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
NHC-Supported Iron(II) Phenyl Compound. Organometallics 2015,
34, 599−605. (f) Liu, Y.; Wang, L.; Deng, L. Three-Coordinate
Iron(II) Dialkenyl Compound with NHC Ligation: Synthesis,
Structure, and Reactivity. Organometallics 2015, 34, 4401−4407.
(g) Liu, Y.; Luo, L.; Xiao, J.; Wang, L.; Song, Y.; Qu, J.; Luo, Y.;
Deng, L. Four-Coordinate Iron(II) Diaryl Compounds with
Monodentate N-Heterocyclic Carbene Ligation: Synthesis, Character-
ization, and Their Tetrahedral-Square Planar Isomerization in
Solution. Inorg. Chem. 2015, 54, 4752−4760.
Ferrate: An Intermediate in the Reduction Pathway of Ferric Salts with
MeMgBr. J. Am. Chem. Soc. 2014, 136, 15457−15460.
(17) (a) Smith, J. M.; Sadique, A. R.; Cundari, T. R.; Rodgers, K. R.;
Lukat-Rodgers, G.; Lachicotte, R. J.; Flaschenriem, C. J.; Vela, J.;
Holland, P. L. Studies of Low-Coordinate Iron Dinitrogen Complexes.
J. Am. Chem. Soc. 2006, 128, 756−769. (b) Lichtenberg, C.; Viciu, M.;
Adelhardt, L.; Sutter, J.; Meyer, K.; de Bruin, B.; Grutzmacher, H.
̈
Low-Valent Iron(I) Amido Olefin Complexes as Promotors for
Dehydrogenation Reactions. Angew. Chem., Int. Ed. 2015, 54, 5766−
5771.
(8) Daifuku, S. L.; Kneebone, J. L.; Snyder, B. E. R.; Neidig, M. L.
Iron(II) Active Species in Iron−Bisphosphine Catalyzed Kumada and
Suzuki−Miyaura Cross-Couplings of Phenyl Nucleophiles and
Secondary Alkyl Halides. J. Am. Chem. Soc. 2015, 137, 11432−11444.
(18) The structure was an isomorph of the chloride analogue
reported in: Doring, M.; Uhlig, E.; Dahlenburg, L. Die binuklearen
̈
Produkte der Umsetzung von Grignardreagenzien mit Acetylaceto-
naten der spaten 3d-Elemente und ihre katalytische Wirkung bei der
̈
(9) Munoz, S. B., III; Daifuku, S. L.; Brennessel, W. W.; Neidig, M. L.
̃
Isolation, Characterization, and Reactivity of Fe8Me12−: Kochi’s S = 1/
2 Species in Iron-Catalyzed Cross-Couplings with MeMgBr and Ferric
Salts. J. Am. Chem. Soc. 2016, 138, 7492−7495.
Kreuzkopplung. Z. Anorg. Allg. Chem. 1989, 578, 58−68.
(19) (a) For the formation of polyphenyl-FeII species using PhLi,
see: Furstner, A.; Martin, R.; Krause, H.; Seidel, G.; Goddard, R.;
̈
Lehmann, C. W. Preparation, Structure, and Reactivity of Non-
stabilized Organoiron Compounds. Implications for Iron-Catalyzed
Cross Coupling Reactions. J. Am. Chem. Soc. 2008, 130, 8773−8787.
(b) Bedford, R. B.; Brenner, P. B.; Carter, E.; Cogswell, P. M.;
Haddow, M. F.; Harvey, J. N.; Murphy, D. M.; Nunn, J.; Woodall, C.
H. TMEDA in Iron-Catalyzed Kumada Coupling: Amine Adduct
versus Homoleptic “ate” Complex Formation. Angew. Chem., Int. Ed.
2014, 53, 1804−1808. (c) Parchomyk, T.; Koszinowski, K. Ate
Complexes in Iron-Catalyzed Cross-Coupling Reactions. Chem. - Eur.
J. 2016, 22, 15609−15613.
(10) (a) Charavay, C.; Segard, S.; Edon, F.; Clem
G. SimuMoss software; CEA/DRF/BIG, CNRS, Universite
Alpes, 2012. (b) Carboni, M.; Clemancey, M.; Molton, F.; Pec
́
ancey, M.; Blondin,
Grenoble
aut, J.;
́
́
́
Lebrun, C.; Dubois, L.; Blondin, G.; Latour, J.-M. Biologically Relevant
Heterodinuclear Iron−Manganese Complexes. Inorg. Chem. 2012, 51,
10447−10460.
(11) Williamson, K. L.; Masters, K. M. Macroscale and microscale
organic experiments, 6th ed.; Cengage Learning: 2010.
(12) Xue, Z.; Daran, J.-C.; Champouret, Y.; Poli, R. Ligand Adducts
of Bis(acetylacetonato)iron(II): A 1H NMR Study. Inorg. Chem. 2011,
50, 11543−11551.
(13) It was reported by Neidig that in situ acidic quenching at short
times (within ca. 30 s; see ref 8) could artificially lead to the formation
of biphenyl, thus leading to an overestimation of its quantity formed
by the reduction of iron. This is due to the quenching process of
transient short-lived aryl-FeII species.
(20) (a) Schilter, D.; Nilges, M. J.; Chakrabarti, M.; Lindahl, P. A.;
Rauchfuss, T. B.; Stein, M. Mixed-Valence Nickel−Iron Dithiolate
Models of the [NiFe]-Hydrogenase Active Site. Inorg. Chem. 2012, 51,
2338−2348. (b) Silakov, A.; Olsen, M. T.; Sproules, S.; Reijerse, E. J.;
Rauchfuss, T. B.; Lubitz, W. EPR/ENDOR, Mossbauer, and
̈
Quantum-Chemical Investigations of Diiron Complexes Mimicking
the Active Oxidized State of [FeFe]Hydrogenase. Inorg. Chem. 2012,
51, 8617−8628. (c) Stoian, S. A.; Hsieh, C.-H.; Singleton, M. L.;
Casuras, A. F.; Darensbourg, M. Y.; McNeely, K.; Sweely, K.; Popescu,
C. V. Hyperfine interactions and electron distribution in FeIIFeI and
FeIFeI models for the active site of the [FeFe] hydrogenases:
(14) (a) Carloni, P.; Greci, L.; Stipa, P.; Eberson, L. Electron-
Transfer Reactions. Oxidation of Grignard Reagents in the Presence of
an Aminoxyl as a Radical-Trapping Agent. J. Org. Chem. 1991, 56,
̀
4733−4737. and references cited therein (b) Lefevre, G.; Jutand, A.
Activation of Aryl and Heteroaryl Halides by an Iron(I) Complex
Generated in the Reduction of [Fe(acac)3] by PhMgBr: Electron
Transfer versus Oxidative Addition. Chem. - Eur. J. 2014, 20, 4796−
4805. (c) This is confirmed by a DFT-computed barrier of 45.1 kcal·
mol−1. Grignard reagents can act as monoelectronic reductants with
oxidants weaker than Fe(acac)3 but under harsher conditions. See:
Eberson, L.; Greci, L. Experimental Evidence for an Electron-Transfer
Mechanism in the Reaction between 2-Phenyl-3-(phenylimino)-3H-
indole and Grignard Reagents: Application of the Marcus Theory. J.
Org. Chem. 1984, 49, 2135−2139. (oxidation of alkyl Grignard
reagents by iminoindoles). (d) Uchiyama, N.; Shirakawa, E.; Hayashi,
T. Single electron transfer-induced Grignard cross-coupling involving
ion radicals as exclusive intermediates. Chem. Commun. 2013, 49,
364−366. (oxidation of PhMgBr by aryl iodides); see also ref 14a for
Mossbauer spectroscopy studies of low-spin FeI. JBIC, J. Biol. Inorg.
̈
Chem. 2013, 18, 609−622.
(21) (a) Bedford, R. B.; Brenner, P. B.; Carter, E.; Clifton, J.;
Cogswell, P. M.; Gower, N. J.; Haddow, M. F.; Harvey, J. N.; Kehl, J.
A.; Murphy, D. M.; Neeve, E. C.; Neidig, M. L.; Nunn, J.; Snyder, B. E.
N.; Taylor, J. Iron Phosphine Catalyzed Cross-Coupling of
Tetraorganoborates and related Group 13 Nucleophiles with Alkyl
Halides. Organometallics 2014, 33, 5767−5780. (b) Ouyang, Z.; Meng,
Y.; Cheng, J.; Xiao, J.; Gao, S.; Deng, L. Three- and Four-Coordinate
Homoleptic Iron(I)−NHC Complexes: Synthesis and Character-
ization. Organometallics 2016, 35, 1361−1367.
(22) (a) Parker, S. F.; Peden, C. H. F. Iron benzene reactions: a
matrix isolation Mossbauer investigation. J. Organomet. Chem. 1984,
̈
272, 411−416. (b) Blom, B.; Tan, G.; Enthaler, S.; Inoue, S.; Epping, J.
D.; Driess, M. Bis-N-Heterocyclic Carbene (NHC) Stabilized η6-Arene
Iron(0) Complexes: Synthesis, Structure, Reactivity, and Catalytic
Activity. J. Am. Chem. Soc. 2013, 135, 18108−18120.
oxidation of Grignard reagents by indole bisnitrone. (e)
A
bisarylmagnesium species such as Ph2Mg, in-situ-formed by Schlenk
equilibrium, could more likely be involved in outer-sphere
monoelectronic transfer with Fe(acac)3. Bisorganomagnesium com-
pounds are indeed more easily oxidized than their Grignard analogues.
See: Psarras, T.; Dessy, R. E. Organometallic Electrochemistry. X.
Organomagnesium reagents. J. Am. Chem. Soc. 1966, 88, 5132−5135.
(15) Transition Metal Arene π-Complexes in Organic Synthesis and
(23) Michaud, P.; Mariot, J.-P.; Varret, F.; Astruc, D. Improved
Synthesis and Electronic Structure of the 19- and 20-Electron
Complexes [Fe(η6-C6Me6)2]n+, n = 0,1. J. Chem. Soc., Chem. Commun.
1982, 1383−1385.
(24) (a) Ittel, S. D.; Tolman, C. A. Metal Vapor Synthesis of Iron η6-
Arene Complexes. Organometallics 1982, 1, 1432−1436. (b) Morand,
P. D.; Francis, C. G. Modeling Macroscale Metal Vapor Reactions. 2.
Bis(arene)iron Revisited. Organometallics 1985, 4, 1653−1659.
Catalysis; Kundig, E. P., Ed.; Springer International: 2004.
̈
(16) (a) For the formation of a well-defined tetraaryl-FeIII species,
see: Alonso, P. J.; Arauzo, A. B.; Fornies, J.; Garcia-Monforte, M. A.;
Martin, A.; Martinez, J. I.; Menjon, B.; Rillo, C.; Saiz-Garitaonandia, J.
J. A Square-Planar Organoiron(III) Compound with a Spin-Admixed
State. Angew. Chem., Int. Ed. 2006, 45, 6707−6711. (b) For the
formation of a [Me4FeIII]− anion by transmetalation of MeMgBr onto
FeCl3, see: Al-Afyouni, M. H.; Fillman, K. L.; Brennessel, W. W.;
Neidig, M. L. Isolation and Characterization of a Tetramethyliron(III)
(25) (a) Romelt, M.; Ye, S.; Neese, F. Calibration of Modern Density
̈
Functional Theory Methods for the Prediction of 57Fe Moossbauer
̈
Isomer Shifts: Meta-GGA and Double-Hybrid Functionals. Inorg.
Chem. 2009, 48, 784−785. and references cited therein. (b) Low-spin
Fe0 complexes: Fe(CO)5, (η4-C4H4)Fe(CO)3, (η6-C6H6)(η4-C6H8)Fe,
(η6-C6H6)(η4-C6H6)Fe, (η6-biphenyl)Fe(SciOPP), and Fe[bis[N-(2,6-
N
DOI: 10.1021/acs.inorgchem.6b02616
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
diisopropylphenyl)imidazole-2-ylidene]methylene](η6-C6H6). High-
spin Fe0 complex: (η6-C6Me6)2Fe. Low-spin FeI complexes: (p-tolyl)
(Ph2PCH2CH2PPh2)2Fe, NaFe(trop2dae)(THF)3 (trop = 5H-diben-
zo[a,d]cyclohepten-5-yl and dae = NCH2CH2N). See the SI for the
corresponding reported nuclear parameters and related references. (c)
Computation of the quadrupolar split was more problematic because
this nuclear parameter is highly dependent on the symmetry of the
occupied electronic levels. For several complexes in the benchmark
depicted earlier,25b the experimental value of ΔEQ could not be
reproduced with a satisfying accuracy. Therefore, minor differences
between the optimized geometries and the experimental structures can
induce strong discrepancies between the experimental and calculated
values of ΔEQ. For a report of the computational analysis of
̀ ̀
quadrupolar splits, see: Papai, M.; Vanko, G. On Predicting Mossbauer
̈
Parameters of Iron-Containing Molecules with Density-Functional
Theory. J. Chem. Theory Comput. 2013, 9, 5004−5020.
(26) (a) Ni, C.; Ellis, B. D.; Fettinger, J. C.; Long, G. J.; Power, P. P.
Univalent transition metal complexes of arenes stabilized by a bulky
terphenyl ligand: differences in the stability of Cr(I), Mn(I) or Fe(I)
complexes. Chem. Commun. 2008, 1014−1016. (b) Werncke, C. G.;
Bunting, P. C.; Duhayon, C.; Long, J. R.; Bontemps, S.; Sabo-Etienne,
S. Two-Coordinate Iron(I) Complex [Fe{N(SiMe3)2}2]−: Synthesis,
Properties, and Redox Activity. Angew. Chem., Int. Ed. 2015, 54, 245−
248.
(27) (a) 1a is a model for (η6-C6H6)FeI(HC(C[Me]N-(2,6-
diisopropylphenyl))2) and 1c for its PPh3 analogue.17a. (b) For 1b,
see ref 17b. (c) 1d is a model for (η6-C6H6)FeI(C6H-2,6-(C6H2-2,4,6-
iPr3)2-3,5-iPr2).26a
(28) Schmidt, M. W.; Gordon, M. S. The construction and
interpretation of MCSCF wavefunctions. Annu. Rev. Phys. Chem.
1998, 49, 233−266.
(29) The same conclusion was reported by Norrby regarding
PhFeI(THF) complexes: Hedstrom, A.; Lindstedt, E.; Norrby, P.-O.
̈
n
On the oxidation state of iron in iron-mediated C−C couplings. J.
Organomet. Chem. 2013, 748, 51−55.
(30) Moreover, the computed density at the 57Fe nucleus in the
monophenyl spin doublet species (η6-C6H6)FeIPh would lead to an
isomer shift of 0.67 mm·s−1 (computed ρ0 = 11827.64 e·a0−3), which is
strongly different from the experimental value.
(31) [Ph3FeII]− has been computed in the quintet spin state, which is
the usual spin state for trishydrocarbyl ate-FeII species; see ref 19b and
references cited therein. A sextet ground state has been found for
[Ph4FeIII]−; [(η6-C6H5Me)FeIPh2]− has been computed in the doublet
state.
(32) Zhurkin, F. E.; Wodrich, M. D.; Hu, X. A Monometallic Iron(I)
Organoferrate. Organometallics 2017, 36, 499−501.
(33) This second scenario is preferred for the aryl−alkyl Kumada
coupling mediated by SciOPP-FeII systems.8
O
DOI: 10.1021/acs.inorgchem.6b02616
Inorg. Chem. XXXX, XXX, XXX−XXX