On the oxidation state of iron in iron-mediated C-C couplings

Article abstract of DOI:10.1016/j.jorganchem.2013.04.024

The nature of the active catalyst in iron-catalyzed C-C couplings has been under debate. In here, we study the couplings with aryl Grignard reagents, and clearly show that the active catalyst is an Fe(I) species. The Grignard alone can reduce the pre-catalyst to the Fe(I) state, and no further, as shown by quantification of product formation. Addition of the electrophile results in complete cross-coupling, validating the nature of the active catalyst. A computational study reveals that the active iron catalyst has a spin state of S = 3/2, high spin for Fe(I) but intermediate spin for Fe(III) complexes, even though the Fe(III) precatalyst salts have a high spin state (S = 5/2). The spin change occurs after the first transmetallation, when the strong ligand field of the aryl group raises the energy of one d-orbital, inducing an electron pairing event. All steps in the formation of an active cross-coupling catalyst are facile and strongly exergonic.

Full text of DOI:10.1016/j.jorganchem.2013.04.024

Journal of Organometallic Chemistry xxx (2013) 1e5
Contents lists available at SciVerse ScienceDirect
Journal of Organometallic Chemistry
journal homepage: www.elsevier.com/locate/jorganchem
On the oxidation state of iron in iron-mediated CeC couplings
*
Anna Hedström, Erik Lindstedt, Per-Ola Norrby
University of Gothenburg, Department of Chemistry and Molecular Biology, Kemigården 4, SE-412 96 Göteborg, Sweden
a r t i c l e i n f o
a b s t r a c t
Article history:
The nature of the active catalyst in iron-catalyzed CeC couplings has been under debate. In here, we
study the couplings with aryl Grignard reagents, and clearly show that the active catalyst is an Fe(I)
species. The Grignard alone can reduce the pre-catalyst to the Fe(I) state, and no further, as shown by
quantification of product formation. Addition of the electrophile results in complete cross-coupling,
validating the nature of the active catalyst. A computational study reveals that the active iron catalyst
has a spin state of S ¼ 3/2, high spin for Fe(I) but intermediate spin for Fe(III) complexes, even though the
Fe(III) precatalyst salts have a high spin state (S ¼ 5/2). The spin change occurs after the first trans-
metallation, when the strong ligand field of the aryl group raises the energy of one d-orbital, inducing an
electron pairing event. All steps in the formation of an active cross-coupling catalyst are facile and
strongly exergonic.
Received 3 March 2013
Received in revised form
13 April 2013
Accepted 16 April 2013
Keywords:
Iron
CeC bond formation
Reaction mechanism
Density functional theory
Cross-coupling reactions
Ó 2013 Elsevier B.V. All rights reserved.
1. Introduction
studies, it is hard to distinguish between full conversion of into Fe(I)
complexes, and partial conversion into Fe in lower oxidation states.
The interest in iron catalyzed coupling reactions is continuously
growing; iron salts are environmentally benign, cheap, and with
relatively low toxicity. Iron can exist in various oxidation states,
from Fe(ꢀII) to Fe(þVIII) [1], allowing the metal to participate in
electron transfer and redox processes, including CeC coupling
processes [2]. Several groups have proposed catalytic cycles for the
iron catalyzed coupling reaction between an alkyl Grignard reagent
and an aryl halide, based on the well-established mechanisms for
Ni- and Pd-catalyzed cross-couplings [3] (Scheme 1).
Kochi originally favored a cycle where initially formed Fe(I)
undergoes oxidative addition yielding an Fe(III) species, with sub-
sequent transmetallation and reductive elimination returning to
the catalytically active Fe(I) species [4e8]. However, Kochi could
not clearly distinguish between this process and one where iron
cycles between oxidation states 0 and II [8,9]. In the 1990’s, the
observation that isolated Fe(ꢀII) complexes are very efficient pre-
catalysts for the coupling reaction led to the proposal of an Fe(ꢀII)/
Fe(0) cycle [10e15]. Other catalyst states, like Fe(ꢀI), have also been
proposed [16]. Our own mechanistic investigations (a combination
of experimental and computational studies) [17,18] indicate that
the most likely catalyst is an Fe(I) complex, in good agreement with
the Kochi proposal [9]. However, in our case, as well as in Kochi’s
Fe(I) complexes are rare [19], but a few well-characterized exam-
ples are reported in the literature [20e24], and in particular recent
work by Bedford and coworkers support the involvement of their
isolated Fe(I)-phosphine complexes in the catalytic CeC coupling
[25]. However, the most common protocols for iron-catalyzed
couplings include only ligands with weaker ligand fields, like ha-
lides and N-methyl-2-pyrrolidone (NMP) [26]. Since the nature of
the active catalyst under these more weakly coordinating condi-
tions is still in doubt, we have here undertaken a study of the
oxidation state of the active catalyst under these conditions.
Two major versions of the iron-catalyzed CeC coupling are in
common use: coupling of alkyl Grignard reagents with aryl chlorides
or triflates [27,28], and coupling of aryl Grignards with alkyl bro-
mides [29,30], primarily secondary bromides. The former reaction
requires the presence of stabilizers like NMP for optimum perfor-
mance [26], and frequently leads to extensive catalyst breakdown
near the end of the reaction due to the strong reducing power of the
alkyl Grignard [17]. The latter reaction, due to the lower reducing
power of aryl Grignard reagents, can be performed in diethyl ether
without added stabilizers [29]. In fact, we have recently shown that
this system generates a live catalyst that stays active even after full
consumption of the electrophile [31]. Thus, the active catalyst can be
generated separately by reaction of aryl Grignard with simple iron
salts with concomitant formation of biaryl [32], and will enter the
coupling cycle on subsequent addition of alkyl bromides. The iso-
lated reaction of aryl Grignard reagents with iron salts thus forms a
* Corresponding author. Tel.: þ46 31 7869034.
E-mail address: pon@chem.gu.se (P.-O. Norrby).
0022-328X/$ e see front matter Ó 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.jorganchem.2013.04.024
Please cite this article in press as: A. Hedström, et al., Journal of Organometallic Chemistry (2013), http://dx.doi.org/10.1016/
j.jorganchem.2013.04.024

2
A. Hedström et al. / Journal of Organometallic Chemistry xxx (2013) 1e5
0.6
0.5
0.4
0.3
0.2
Fe(III)Br
Fe(II)Br
Scheme 1. Generic metal-catalyzed CeC coupling with Grignard reagents.
0.1
0
pleasingly simple test system for elucidating the nature of the active
catalyst in the iron-catalyzed CeC coupling.
0
0.2
0.4
0.6
0.8
1
iron (mmol)
2. Results and discussion
Fig. 1. Biphenyl formation on titration with iron salts.
2.1. Stoichiometric iron reduction
tested the importance of contaminant by including a sample of high
purity FeCl3. The results were virtually the same as with the stan-
dard quality salt, indicating that trace metals are not responsible for
the homocoupling. In the current case, this was expected, since iron
has been shown to perform CeC couplings at dry ice temperatures
[18], where no other metal is a competent catalyst.
The Fe(II) complex shows lower reproducibility (Table 1). This is
not entirely unexpected, since Fe(II) salts can suffer from rapid
oxidation upon air contact, resulting in a higher slope. Still, the data
gives slopes close to the expected 0.5 eq. biphenyl formation,
consistent with one-electron reduction of Fe(II). We note that
diaryl-Fe(II) complexes cannot yield Fe(I) by the normal reductive
elimination, since that must necessarily reduce the metal by two
electrons. However, we have earlier shown that diaryl-Fe(II) in a
bimetallic complex with another Fe(II) unit can undergo a variant of
the reductive elimination with a reasonable, albeit slightly higher
barrier, yielding two Fe(I) moieties [17].
There is a possibility that our titration experiment does not
yield a single Fe(I) species, but instead a mixture of iron complexes
with an average oxidation state of þ1. However, the linearity of the
plots, as well as the close correspondence between the Fe(III) and
Fe(II) precatalysts makes this scenario unlikely. An equilibrating
mixture should demonstrate a dependence on concentration, and
therefore induce noticeable curvature in the Fig. 1 plots. A stable
bimetallic Fe(0)eFe(II) complex cannot be excluded, but in the
absence of specific supporting ligands, we find it unlikely that such
a mixed-valence complex would be favored in all concentration
regimes. In the presence of the more strongly reducing alkyl
Grignard reagents, we have previously found strong concentration
effects and have argued that these are due to the formation of iron
oligomers [18], but even then, our computational study indicated
that reductive elimination was limited to forming Fe(I), and that
the role of bimetallic complexes primarily is to allow the afore-
mentioned bimetallic reduction of two Fe(II) moieties to two Fe(I)
complexes [17].
The stoichiometry of the reduction of iron using aryl Grignard
(Scheme 2) was investigated in a titration experiment under inert
conditions. An iron salt was dissolved in tetrahydrofuran (THF), and
added in batches to a solution of phenyl magnesium bromide in
diethyl ether (DEE) with an internal standard (dodecane), at con-
stant intervals of 5e15 min. Before each addition, a sample was
withdrawn, subjected to an inert workup, and then analyzed by GC.
The inert workup was crucial to reproducibility. In non-inert
workup, the reaction mixture was oxidized by oxygen in the air,
possibly through reaction with low-valent aryl iron species, so-
called “Kharasch”-complexes [32]. However, under inert condi-
tions, the concentration of biphenyl increased smoothly, as illus-
trated in Fig. 1.
The mass balance in the titration experiment was checked by
also monitoring the formation of benzene. No other phenyl-
containing species were detected. As can be seen in Fig. 1, the re-
action was followed until the Grignard reagent was fully consumed,
at which point the concentration of biphenyl no longer increases.
The slope, corresponding to the stoichiometry of the reaction, was
measured on the initial linear region, determined by comparative
F-test analysis.
The titration experiment was performed with standard FeBr3
and FeCl3 of at least 98% purity, as well as with highly purified FeCl3
(99.99%). FeBr2 was selected as a representative Fe(II) salt. We
attempted to use FeCl2, but solubility became an issue. The
convenient non-hygroscopic Fe(acac)3 (acac ¼ acetylacetonate) was
tested, but was found to yield multiple products due to addition of
Grignard to the acac moiety, and was therefore excluded from the
current study. The results are shown in Table 1.
The data for Fe(III) complexes in Table 1 clearly show a repro-
ducible stoichiometry very close to 1. Since formation of biphenyl
from Grignard reagents is a two-electron process, this gives strong
indication that iron is reduced to Fe(I), but not further, even in the
presence of excess aryl Grignard reagent.
Since our lab has recently found that apparent iron-catalyzed
processes were instead due to trace metal contaminants [33], we
Table 1
Slope and linearity of biphenyl formation for different iron salts, Scheme 2.
Iron salt
Purity
Slope
r2
FeBr3
FeCl3
98%
98%
99.99%
99.99%
0.95
0.997
1.0
0.596
0.603
0.492
0.999
0.996
0.977
0.993
1.00
FeBr2
Scheme 2. Reaction of phenyl magnesium bromide with iron salts yields biphenyl
while forming a reduced iron species.
0.987
Please cite this article in press as: A. Hedström, et al., Journal of Organometallic Chemistry (2013), http://dx.doi.org/10.1016/
j.jorganchem.2013.04.024

A. Hedström et al. / Journal of Organometallic Chemistry xxx (2013) 1e5
3
0.3
2.2. Active catalyst
0.25
To verify that the Fe(I) complex produced in the titration
experiment is indeed an active, live catalyst for the cross-coupling
reaction, we performed a variation of the titration experiment
described in Section 2.1. Iron additions were alternated with ad-
ditions of a model electrophile, cyclohexyl bromide, with sample
aliquots withdrawn before addition of each reagent (Scheme 3). In
the GC analysis, we now also monitored appearance of the cross-
coupling product, phenyl cyclohexane. The results of the moni-
toring seen in Fig. 2; even-numbered points correspond to sam-
pling after FeCl3 addition, odd-numbered points to cyclohexyl
bromide addition.
0.2
0.15
0.1
Added FeCl3
Measured Ph-Ph
Added CyHexBr
Measured Ph-Cy
0.05
0
1
2
3
4
5
6
7
8
9
10 11
As can be seen in Fig. 2, the reaction is very reliable; iron
addition gives clean production of biphenyl, whereas cyclohexyl
bromide addition only increases the yield of cross-coupling prod-
uct. The addition before sample 10 consumes the last of the
Grignard reagent, and no further product formation is seen after
this point. From this point onwards, we can also detect unreacted
cyclohexyl bromide in the reaction mixture.
Fig. 2. Product monitoring upon alternating additions, Scheme 3.
generally only allow low or high spin solution. However, in a non-
symmetric field with only one strong ligand, only one orbital is
increased in energy. With exactly one d-orbital empty, two d-
electrons will pair, yielding 3 unpaired electrons and thus S ¼ 3/2
(Fig. 4).
2.3. Proposed reaction mechanism
The first transmetallation can occur entirely on the high spin
surface, delivering a mono-aryl in a spin-excited state from a
strongly exergonic reaction. In a case like this, there is no need to
calculate spin surface crossing points. The spin crossing does not
have to be part of a barrier, but can occur in a stable species through
collisions with surrounding molecules. The calculated energy dif-
ference between the spin states may not be very reliable with a DFT
method, but should be in the range of molecular vibrations, that is,
less than ca. 50 kJ/mol.
The second transmetallation would be expected to occur on the
intermediate spin surface. The two first transmetallations are both
strongly exergonic. The diaryl Fe(III) species can choose between
two paths; either association of a Grignard reagent resulting in a
third transmetallation (path i), or reductive elimination yielding
the observed biphenyl product (path ii). The latter reaction has a
potential energy barrier of only ca. 1 kJ/mol, with a free energy of
the transition state that is actually lower than that of the reactant.
This is of course a computational artifact, a result of the fact that
one normal mode less contributes to the ZPE correction in the TS.
However, with such a low potential energy barrier, a single mo-
lecular vibration is sufficient to bring the reactant across the barrier,
with the result that no possible diffusion process can compete. As
discussed by Harvey et al. [35], a bimolecular reaction will always
have a free energy barrier of at least ca. 20 kJ/mol at ambient
temperature, even when no barrier can be found on the potential
energy surface, that is, when the reaction is under diffusion control.
We can therefore state with confidence that the third trans-
metallation cannot occur before reductive elimination. The result-
ing high spin (S ¼ 3/2) Fe(I) complex can undergo a virtually
isoergic transmetallation, in good agreement with earlier studies
using alkyl Grignard species [17]. Neither FeBr nor FePh prefers a
Starting from Fe(III), the following reaction mechanism is sug-
gested for the homocoupling: two consecutive transmetallations
(TM) give a diarylated Fe(III) complex which can either undergo a
third TM to a poly-arylated species followed by a reductive elimi-
nation (RE) to yield biphenyl and aryl-Fe(I) (path i) or a RE to yield
biphenyl and Fe(I)-bromide (path ii). The FeBr can then be trans-
metallated to PhFe. The complexes in Scheme 4 are stabilized by
solvent molecules (omitted for clarity). To verify the plausibility of
the proposed mechanism, and to elucidate details about the degree
of solvation and spin states, the proposed mechanism in Scheme 4
was further investigated by DFT calculations (Section 2.4).
2.4. Computational studies
To check the postulated reaction mechanism, and to differen-
tiate between the two suggested paths, we investigated the
mechanisms in Scheme 4 using dispersion-corrected DFT methods.
All energies used here are free energies in solvent, calculated as
detailed in Section 5. For transmetallation steps, we assumed that
association of Grignard reagents to iron complexes, and dissocia-
tion of MgBr2 from the aryl iron products, would have negligible
barriers [17], and therefore only calculated the structures and en-
ergies of the bimetallic FeemePheMg complexes. In all cases, the
association to a bimetallic complex required dissociation of one
solvent molecule from each metal, at a slight enthalpic cost but
entropic gain. As can be seen in Fig. 3, the first transmetallation
occurs on the high spin energy surface, but the strong ligand field of
the aryl is sufficient to favor an intermediate spin (S ¼ 3/2) of all aryl
Fe(III) species.
The intermediate spin is very rarely seen [34]. Classical solutions
in ligand field theory are generally based on symmetric ligand
fields, yielding degenerate d-orbital energies, and thus will
Scheme 4. Suggested reaction mechanisms for the homocoupling of aryl Grignard
Scheme 3. Alternating catalyst production and cross-coupling.
reagent.
Please cite this article in press as: A. Hedström, et al., Journal of Organometallic Chemistry (2013), http://dx.doi.org/10.1016/
j.jorganchem.2013.04.024

4
A. Hedström et al. / Journal of Organometallic Chemistry xxx (2013) 1e5
Fig. 3. Free energies surface (FES) of the homocoupling reaction. Vertical lines connect different spin states of the same complex. Curves represent associations with barriers of at
least 20 kJ/mol, dashed lines correspond to dissociations. The number of explicit solvent molecules (s) has been optimized for each complex.
low spin state, irrespective of the number of explicit solvents co-
ordinated to iron. However, Bedford has shown that addition of
bidentate phosphines with strong ligand fields can result in for-
mation of low spin Fe(I) complexes [25].
transmetallation occurring on the high spin surface, whereas sub-
sequent steps where the iron is aryl-substituted prefers an inter-
mediate spin. After the second transmetallation, a reductive
elimination occurs virtually without a barrier, yielding biphenyl
and Fe(I)Br. The latter is in equilibrium with Fe(I)Ph through
reversible transmetallation with the Grignard reagent. It has earlier
been shown that both types of Fe(I) species are competent in-
termediates in the cross-coupling reaction [17].
3. Conclusions
Reaction monitoring of iron-mediated coupling of aryl Grignard
with iron salts clearly indicates that the lowest oxidation state of an
iron that can be reached under these conditions is Fe(I). The
experimental result is in good agreement with our earlier compu-
tational studies, where we showed that reductive elimination to
Fe(0) has a prohibitively high barrier [18], and that lower oxidation
states are strongly disfavored not only kinetically but also ther-
modynamically [17]. The resulting Fe(I) complex is stable in diethyl
ether without additives, and is shown to be a living catalyst for the
CeC coupling of phenyl magnesium bromide and cyclohexyl
bromide.
4. Experimental
4.1. Typical procedure for monitoring of aryl homocoupling
THF and DEE were distilled from benzophenone and sodium,
dodecane was distilled from calcium hydride, pentane was
degassed prior to use, phenyl magnesium bromide was bought
from SigmaeAldrich and titrated [36] before use.
An oven-dried 100 ml round bottomed flask was sealed with a
rubber septum, then evacuated and refilled with nitrogen three
Computational studies indicate that the homo-coupling fol-
lowed in the monitoring studies is a facile process, with the first
times. The flask was charged with DEE (60 ml), dodecane (225 ml,
1 mmol) and phenyl magnesium bromide (1.2 mmol, in DEE solu-
tion). An aliquot (0.5 ml) was taken from the mixture and quenched
by filtering it through a pentane-saturated short silica plug under
nitrogen. The silica plug consisted of a Pasteur pipette with a glass
wool plug, a short layer of silica and top-sealed with a rubber
septum with a nitrogen inlet. The silica was pre-flushed with
degassed pentane before the addition of any sample. The sample
was eluted through the silica plug with pentane into a degassed sat.
NH4Cl solution. The organic phase was diluted with DEE and
analyzed by GC (dodecane was used as internal standard, 40e
100 ꢁC, 18 ꢁC/min, 100e300 ꢁC, 20 ꢁC/min). A solution of FeCl3
(0.06 mmol, 0.05 M in THF) was added to the reaction vessel. After
stirring for 5 min, an aliquot (0.5 ml) was collected and analyzed as
described above. The procedure was repeated with addition of FeCl3
(0.05 M in THF) in portions of 0.06 mmol (a total of ten portions).
L
Weak field
high spin
Strong field
low spin
Localized field
intermediate spin
Fig. 4. Ligand field analysis of Fe(III), a d5-metal, giving the two classical solutions
based on octahedral geometries, and an expected perturbation from a single strong
field ligand.
Please cite this article in press as: A. Hedström, et al., Journal of Organometallic Chemistry (2013), http://dx.doi.org/10.1016/
j.jorganchem.2013.04.024

A. Hedström et al. / Journal of Organometallic Chemistry xxx (2013) 1e5
5
4.2. Addition of cyclohexyl bromide
References
[1] Y.D. Perfiliev, V.K. Sharma, Higher oxidation states of iron in solid state:
synthesis and their Mössbauer characterization, in: V.K. Sharma (Ed.), Fer-
rates, vol. 985, American Chemical Society, 2008, pp. 112e123.
[2] B. Plietker, Iron Catalysis in Organic Chemistry, Wiley-VCH, Weinheim, 2008.
[3] J.F. Hartwig, Organotransition Metal Chemistry: From Bonding to Catalysis,
University Science Books, Sausalito, 2009.
An oven-dried dry 100 mL round-bottomed flask was equipped
with a rubber septum and a stirrer bar, then evacuated and refilled
with nitrogen three times. The flask was charged with DEE (60 ml),
dodecane (225
ml, 1 mmol) and phenyl magnesium bromide
(1.2 mmol, 1.97 M in DEE). An aliquot (0.5 ml) was taken from the
mixture and quenched by filtering it through a wet short silica plug
with a pentane layer above the silica under nitrogen (see Section
4.1). The sample was eluted with pentane into a degassed sat. NH4Cl
solution. The organic phase was diluted with DEE and analyzed by
GC (dodecane was used as internal standard, 40e100 ꢁC 18 ꢁC/min,
100e300 ꢁC, 20 ꢁC/min). A solution of FeCl3 (0.06 mmol, 0.05 M in
THF) was added to the reaction vessel. After stirring for 5 min, an
aliquot (0.5 ml) was collected and analyzed as described above. A
solution of cyclohexyl bromide (0.03 mmol, 0.5 M in THF) was
added to the reaction vessel. After stirring for 5 min, an aliquot
(0.5 ml) was collected and analyzed as described above. The pro-
cedure was repeated with alternating addition of FeCl3 (0.05 M in
THF) and cyclohexyl bromide (0.5 M in THF).
[4] M. Tamura, J. Kochi, J. Organomet. Chem. 31 (1971) 289e309.
[5] M. Tamura, J. Kochi, Synthesis 6 (1971) 303e305.
[6] M. Tamura, J.K. Kochi, J. Am. Chem. Soc. 93 (1971) 1487e1489.
[7] M. Tamura, J.K. Kochi, Bull. Chem. Soc. Jpn. 44 (1971) 3063e3073.
[8] R.S. Smith, J.K. Kochi, J. Org. Chem. 41 (1976) 502e509.
[9] S.M. Neumann, J.K. Kochi, J. Org. Chem. 40 (1975) 599e606.
[10] L.E. Aleandri, B. Bogdanovic, P. Bons, C. Duerr, A. Gaidies, T. Hartwig,
S.C. Huckett, M. Lagarden, U. Wilczok, R.A. Brand, Chem. Mater. 7 (1995)
1153e1170.
[11] B. Bogdanovic, M. Schwickardi, Angew. Chem. Int. Ed. 39 (2000) 4610e4612.
[12] A. Fuerstner, A. Leitner, M. Mendez, H. Krause, J. Am. Chem. Soc. 124 (2002)
13856e13863.
[13] A. Fuerstner, A. Leitner, Angew. Chem. Int. Ed. 41 (2002) 609e612.
[14] A. Furstner, R. Martin, H. Krause, G. Seidel, R. Goddard, C.W. Lehmann, J. Am.
Chem. Soc. 130 (2008) 8773e8787.
[15] Q. Ren, S. Guan, F. Jiang, J. Fang, J. Phys. Chem. A 117 (2013) 756e764.
[16] K. Weber, E.-M. Schnöckelborg, R. Wolf, Chem. Cat Chem. 3 (2011) 1572e
1577.
[17] J. Kleimark, A. Hedström, P.-F. Larsson, C. Johansson, P.-O. Norrby, Chem. Cat
Chem. 1 (2009) 152e161.
[18] J. Kleimark, P.-F. Larsson, P. Emamy, A. Hedström, P.-O. Norrby, Adv. Synth.
Catal. 354 (2012) 448e456.
5. Computational details
[19] P. N Hawker, M.V. Twigg, Iron(II) and Lower States, in: G. Wilkinson (Ed.),
Comprehensive Coordination Chemistry, Pergamon, Oxford, 1987, pp. 1179e
1270.
[20] S.A. Stoian, Y. Yu, J.M. Smith, P.L. Holland, E.L. Bominaar, E. Munck, Inorg.
Chem. 44 (2005) 4915e4922.
[21] C. Ni, D. Ellis Bobby, C. Fettinger James, J. Long Gary, P. Power Philip, Chem.
Commun. (2008) 1014e1016.
[22] M.V. Rajasekharan, S. Giezynski, J.H. Ammeter, N. Oswald, P. Michaud,
J.R. Hamon, D. Astruc, J. Am. Chem. Soc. 104 (1982) 2400e2407.
[23] M. Gargano, P. Giannoccaro, M. Rossi, G. Vasapollo, A. Sacco, J. Chem. Soc.
Dalton Trans. (1975) 9e12.
[24] P. Giannoccaro, A. Sacco, S.D. Ittel, M.A. Cushing, Bis[ethyl-
enebis(diphenylphosphine)]-hydridoiron complexes, in: A.G. MacDiarmid
(Ed.), Inorganic Syntheses, vol. 17, John Wiley & Sons, Inc, Hoboken, 2007, pp.
69e72.
[25] C.J. Adams, R.B. Bedford, E. Carter, N.J. Gower, M.F. Haddow, J.N. Harvey,
M. Huwe, M.Á. Cartes, S.M. Mansell, C. Mendoza, D.M. Murphy, E.C. Neeve,
J. Nunn, J. Am. Chem. Soc. 134 (2012) 10333e10336.
[26] G. Cahiez, H. Avedissian, Synthesis 8 (1998) 1199e1205.
[27] W.M. Czaplik, M. Mayer, J. Cvengros, A. Jacobi von Wangelin, Chem. Sus Chem.
2 (2009) 396e417.
[28] B.D. Sherry, A. Fuerstner, Acc. Chem. Res. 41 (2008) 1500e1511.
[29] T. Nagano, T. Hayashi, Org. Lett. 6 (2004) 1297e1299.
[30] M. Nakamura, K. Matsuo, S. Ito, E. Nakamura, J. Am. Chem. Soc. 12 (2004)
3686e3687.
All DFT calculations were performed in Jaguar 8.0 from Schrö-
dinger [37]. We utilized the B3LYP-D3 method, which combines the
recent dispersion correction developed by Grimme and coworkers
[38] in conjunction with the B3LYP functional [39e41]. The basis set
was LACVP*, a combination of 6-31G* for light elements together
with the Hay-Wadt ECP basis for Fe and Br [42]. Geometries were
optimized in gas phase, with explicit solvent molecules modeled by
dimethyl ether (DME). Thermodynamic corrections to the free en-
ergy were obtained from frequency calculations at the optimized
geometries. Energies in solvent were calculated using the PBF im-
plicit solvation model at the optimized gas phase geometries
[43,44]. All reported energies are final free energies obtained by
addition of the thermodynamic correction (including zero point
energy correction) from the frequency calculation to the energies in
solvent calculated using PBF. The number of explicit solvent models
was optimized for each species, as judged by the calculated free
energies. We note that the use of gas phase vibrational entropies
will slightly favor dissociation of the explicit solvents; no attempt
was made to correct for this systematic error.
For all iron-containing complexes, we used the unrestricted
method to converge the open-shell wavefunctions. The <S2>
values were inspected after each calculation and, if the value was
more than a few percent above the theoretical expectation, a
restricted open shell wavefunction was calculated and used as an
initial guess in the unrestricted calculation. In all cases, this pro-
cedure yielded acceptable values of <S2>. All possible spin states
(S ¼ 1/2, 3/2, and for Fe(III) also 5/2) were calculated.
[31] A. Hedström, U. Bollmann, J. Bravidor, P.-O. Norrby, Chem. Eur. J. 17 (2011)
11991e11993.
[32] M.S. Kharasch, E.K. Fields, J. Am. Chem. Soc. 63 (1941) 2316e2320.
[33] P.-F. Larsson, A. Correa, M. Carril, P.-O. Norrby, C. Bolm, Angew. Chem. Int. Ed.
48 (2009) 5691e5693;
See also: S.L. Buchwald, C. Bolm Angew. Chem. Int. Ed. 48 (2009) 5586e5587;
I. Thomé, A. Nijs, C. Bolm, Chem. Soc. Rev. 41 (2012) 979e987.
[34] S. Mossin, B.L. Tran, D. Adhikari, M. Pink, F.W. Heinemann, J. Sutter, R.K. Szilagyi,
K. Meyer, D.J. Mindiola, J. Am. Chem. Soc. 134 (2012) 13651e13661.
[35] C.L. McMullin, J. Jover, J.N. Harvey, N. Fey, Dalton Trans. 39 (2010)
1083310836.
[36] A. Krasovskiy, P. Knochel, Synthesis 5 (2006) 0890e0891.
[37] Jaguar, Version 8.0, Schrodinger, LLC, New York, NY, 2011. For the most recent
version, see, www.schrodinger.com.
Acknowledgments
[38] S. Grimme, J. Antony, S. Ehrlich, H. Krieg, J. Chem. Phys. 132 (2010) 154104.
[39] A.D. Becke, J. Chem. Phys. 98 (1993) 5648.
[40] C.T. Lee, W.T. Yang, R.G. Parr, Phys. Rev. B 37 (1988) 785e789.
[41] P.J. Stephens, F.J. Devlin, C.F. Chabalowski, M.J. Frisch, J. Phys. Chem. 98 (1994)
11623e11627.
This work has been supported by the Swedish Research Council,
grant no. 2010-4856. We are grateful to Professor Michael L. Neidig
for illuminating discussions while revising the current manuscript.
[42] P.J. Hay, W.R. Wadt, J. Chem. Phys. 82 (1985) 299.
[43] B. Marten, K. Kim, C. Cortis, R.A. Friesner, R.B. Murphy, M.N. Ringnalda,
D. Sitkoff, B. Honig, J. Phys. Chem. 100 (1996) 11775.
Appendix A. Supplementary data
[44] D.J. Tannor, B. Marten, R. Murphy, R.A. Friesner, D. Sitkoff, A. Nicholls,
M. Ringnalda, W.A. Goddard, B. Honig, J. Am. Chem. Soc. 116 (1994)
11875.
Supplementary data related to this article can be found at
http://dx.doi.org/10.1016/j.jorganchem.2013.04.024.
Please cite this article in press as: A. Hedström, et al., Journal of Organometallic Chemistry (2013), http://dx.doi.org/10.1016/
j.jorganchem.2013.04.024