Iron(I) in Negishi cross-coupling reactions

Article abstract of DOI:10.1021/ja303250t

Herein we demonstrate both the importance of Fe(I) in Negishi cross-coupling reactions with arylzinc reagents and the isolation of catalytically competent Fe(I) intermediates. These complexes, [FeX(dpbz) ₂] [X = 4-tolyl (7), Cl (8a), Br (8b); d

Full text of DOI:10.1021/ja303250t

Communication
pubs.acs.org/JACS
Iron(I) in Negishi Cross-Coupling Reactions
Christopher J. Ada†ms,† Robin B. Bedford,*,† Emma Carter,‡ Nicholas J. Gower,† Mairi F. Haddow,†
†
§
†
†
́
Jeremy N. Harvey, Michael Huwe, M. Angeles Cartes, Stephen M. Mansell, Carla Mendoza,
Damien M. Murphy,‡ Emily C. Neeve,† and Joshua Nunn†
†School of Chemistry, University of Bristol, Cantock’s Close, Bristol BS8 1TS, U.K.
‡School of Chemistry, Cardiff University, Main Building, Park Place, Cardiff CF10 3TB, U.K.
§Instituto de Investigaciones Químicas, CSIC, Universidad de Sevilla, c/Amer
́
ico Vespucio 49, 41092 Sevilla, Spain
S
* Supporting Information
of the cross-coupled product 5a but also small amounts of
4,4′-bitolyl (6a) (Scheme 1), which was formed very early in the
ABSTRACT: Herein we demonstrate both the impor-
tance of Fe(I) in Negishi cross-coupling reactions with
arylzinc reagents and the isolation of catalytically
competent Fe(I) intermediates. These complexes, [FeX-
(dpbz)2] [X = 4-tolyl (7), Cl (8a), Br (8b); dpbz = 1,2-
bis(diphenylphosphino)benzene], were characterized by
crystallography and tested for activity in representative
reactions. The complexes are low-spin with no significant
spin density on the ligands. While complex 8b shows
performance consistent with an on-cycle intermediate, it
seems that 7 is an off-cycle species.
Scheme 1. Iron-Catalyzed Negishi Coupling of a Benzyl
Halide
reactions [see Figure S1 in the Supporting Information (SI) for
reaction progress against time].14,15 The formation of bitolyl is a
reductive process, liberating two electrons per molecule formed,
so it is tempting to conclude that the amounts of bitolyl produced
directly correspond to a reduction of the precatalyst 1a to iron
species with average oxidation states of Fe(0) and Fe(−II),
respectively. However, reversible redox couples, with the benzyl
halide acting as a reoxidant, could also lead to some bitolyl
formation, thereby making it difficult to deduce the average
oxidation state merely from the amount of bitolyl formed in these
reactions.
he explosion in interest in iron-catalyzed cross-coupling
T
reactions over the past decade1,2 means there is a growing
need to understand the basic mechanisms of these processes, not
least to identify the oxidation states of active iron intermediates.
Suggested candidates for the lowest oxidation states accessible to
iron range from Fe(II) through to Fe(−II).3,4 Kochi suggested
the potential intermediacy of Fe(I) over 40 years ago,5 and recent
indirect evidence in favor of this was provided by Norrby and co-
workers on the basis of mechanistic and computational studies.6
Conversely, Furstner and co-workers concluded that while
̈
To circumvent this problem, the reaction was repeated at
various temperatures in the absence of the electrophile 3a, as this
should give a clear indication of both the average oxidation states
accessible under the reaction conditions and, more importantly,
whether these oxidation states are accessed fast enough to be
relevant in catalysis (Figure 1). It was immediately apparent that
at all four temperatures the reduction of Fe(II) to Fe(I) was fast,
occurring before the first samples were taken. After this point,
further reduction occurred at a much lower, temperature-
dependent rate, with the reaction run at 45 °C producing
∼0.1 mmol of bitolyl after 15 min. Interestingly this corresponds
to an average iron oxidation state of Fe(−II), consistent with
Fe(I) is accessible in the allylation of ArMgX, reactions with a
model iron(I) complex are too slow to account for the activity,
and they suggested an Fe(−II)/Fe(0) manifold instead.4c,7 Here
we demonstrate that Fe(I) is the lowest kinetically reasonable
oxidation state in a representative Negishi cross-coupling
reaction and report the isolation of catalytically relevant iron(I)
complexes.
The Negishi coupling of benzyl halides with diarylzinc
reagents catalyzed by the preformed iron(II) precursor 1a was
chosen for this study first because the reactions are robust and
highly reproducible, and thus amenable to mechanistic study,
and second because complex 1a and the related complexes 2
have shown excellent activity in a range of C−C bond-forming
reactions.8−13
Furstner’s suggestion.4 However, the reductions to oxidation
̈
states below Fe(I) appear to be far too slow for them to be
involved in the catalytic cycles of the reactions in Scheme 1:
turnover leading to at least 50% conversion of substrate was
observed in a time shorter than that needed for further reduction
in the experiments in Figure 1.16
The reactions of benzyl bromide 3a with 4-tol2Zn (4a)
catalyzed by 1a at both 45 and −10 °C gave not only high yields
Received: April 4, 2012
Published: June 13, 2012
© 2012 American Chemical Society
10333
dx.doi.org/10.1021/ja303250t | J. Am. Chem. Soc. 2012, 134, 10333−10336

Journal of the American Chemical Society
Communication
Figure 1. Formation of bitolyl from the reaction of 1a and excess 4a at
various temperatures and the corresponding average oxidation state of
iron.
Fe(I) is one of the more unusual oxidation states of iron,
particularly for mononuclear complexes, which can show unusual
reactivity, such as the cleavage and coupling of carbon dioxide17
and the catalytic N−N coupling of aryl azides.18 Accordingly,
isolating catalytic robust potential intermediates became the next
focus of our attention.
What are the likely structures of active Fe(I) intermediates in
this Negishi reaction? Two obvious candidates are [Fe(p-
tol)(dpbz)2] (7) [dpbz = 1,2-bis(diphenylphosphino)benzene]
and [FeX(dpbz)2] (8) (X = Cl, Br). The reaction of 1a with
either 4a at room temperature or 4-tolMgBr at −40 °C furnished
complex 7 in good yield (Scheme 2). The formation of 7 was
Figure 2. X-ray structures of complexes 7 (top), 8a (bottom left), and
8b (bottom right, one of two independent molecules), with the solvate
molecules and all but the Cipso atoms of the PPh2 phenyl residues
omitted for clarity. The thermal ellipsoids are set at 50% probability.14
reported structures for iron(I) halides,22 and again, none are
supported by only phosphine ligands.
The solution magnetic moment (μobs) of 7 was determined to
be 1.8 μB at room temperature,23 while the solid-state μobs of
complexes 8a and 8b were determined to be 1.9 and 1.8 μB at
1
Scheme 2. Synthesis of Iron(I) Complexes
room temperature, consistent with low-spin (S = /2) five-
coordinate iron(I) centers. The fact that organometallic Fe(I)
compounds are rather rare, coupled with the apparent privileged
position of phenylene-based bis(phosphines) in iron-catalyzed
cross-coupling reactions,8−12 led us to question whether the
unpaired electron is genuinely associated with the metal center
or is instead delocalized across the ligands as seen in recently
reported, nominally iron(I) complexes with extended conjugated
pyridine-based ligands.24
The left panel in Figure 3 shows the EPR spectrum of 7
recorded at 10 K and its simulation.14 The spin Hamiltonian
accompanied by the liberation of exactly 0.5 equiv of 6a,
consistent with reduction of the Fe(II) precursor 1a to Fe(I).
Somewhat fortuitously, we were able to isolate the halide
complexes 8a and 8b by the reaction of 1a or its bromide
counterpart 1b with excess BnMgX (X = Cl or Br, respectively).
All three of these Fe(I) complexes are highly air-sensitive red
compounds.19 Complex 7 is thermally robust: heating it in
toluene at 100 °C for 60 min in the absence of 4a gave no color
change and no formation of 6a.
The X-ray crystal structures of these three complexes were
determined and are shown in Figure 2.14 They display distorted
trigonal-bipyramidal geometries with the unique ligand (tolyl or
halide) in the equatorial plane. The equatorial planes are tilted by
10.4−11.1° from ideal, while the Peq−Fe−Peq bond angles are
more compressed and the Peq−Fe−X angles more obtuse than
ideal.20 Isolated Fe(I)−aryl complexes are very rare,21 and none
of those reported to date are supported solely by phosphine
ligands. The Fe−C bond in complex 7 [2.048(2) Å] is in the
range of those reported for structurally characterized Fe(I)−aryl
complexes (2.029−2.048 Å).21a,b,d Similarly, there are very few
Figure 3. (left) X-band continuous-wave EPR spectrum of 7 recorded at
10 K in THF: (a) experimental; (b) simulated. (right) Calculated
SOMO of low-spin complex 7, with the phosphine ligand residues
shown in wireframe.14
parameters used in the simulation (Table S2 in the SI) are in
good agreement with the theoretical calculations based on a
low-spin (S = 1/2) Fe(I) center.14 Similarly, a density functional
theory (DFT) analysis of the ground-state structure of 7 returned
10334
dx.doi.org/10.1021/ja303250t | J. Am. Chem. Soc. 2012, 134, 10333−10336

Journal of the American Chemical Society
Communication
a geometry and electronic structure consistent with low-spin
Fe(I) character (Figure 3, right).14 The Mulliken spin density
corresponding to the unpaired electron is mostly located on Fe
(0.88), with very small contributions from the ligating P and C
atoms and the other ligand atoms. DFT calculations on 8a and 8b
gave very similar structures and singly occupied molecular orbitals
(SOMOs) (Figure S8), with 0.87 and 0.88 of the unpaired
electron located on the iron center, respectively.14
Importantly, solutions of both 7 and 8b proved to be
catalytically competent, showing productivities similar to that
of the precatalyst 1a both in representative cross-coupling
reactions of ditolylzinc with benzyl halides, benzyl phosphates,
and 2-pyridyl halides under our previously reported conditions8a
and in the reactions of cycloheptyl bromide with fluoroarylzinc
reagents as reported by Nakamura (Scheme 3).8b
AUTHOR INFORMATION
Corresponding Author
r.bedford@bristol.ac.uk
■
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
We thank the EPSRC and Pfizer for funding under the
collaborative EPSRC Programme for Synthetic Organic
Chemistry with AZ−GSK−Pfizer (E.C.N.), the EPSRC for
student support (N.J.G., J.N.), and CSIC (I3P and JAE
Programs) for predoctoral support (M.A.C.). We thank
Professor Guy Lloyd-Jones at Bristol for stimulating and useful
discussions.
REFERENCES
■
(1) For recent reviews, see: (a) Nakamura, E.; Yoshikai, N. J. Org.
Chem. 2010, 75, 6061. (b) Czaplik, W. M.; Mayer, M.; Cvengros,
Wangelin, A. J. ChemSusChem 2009, 2, 396. (c) Sherry, B. D.; Furstner,
A. Acc. Chem. Res. 2008, 41, 1500. (d) Bolm, C.; Legros, J.; Le Paih, J.;
Scheme 3. Representative Cross-Coupling Reactions
Catalyzed by 7 (Prepared in Situ) or 8b Compared with the
Same Reactions Catalyzed by 1a (Values in Parentheses)
̌
J.; von
̈
Zani, L. Chem. Rev. 2004, 104, 6217.
(2) Iron Catalysis in Organic Chemistry; Plietker, B., Ed.; Wiley-VCH:
Weinheim, Germany, 2008.
(3) For an example of proposed Fe(II) intermediates, see: Noda, D.;
Sunada, Y.; Hatakeyama, T.; Nakamura, M.; Nagashima, H. J. Am. Chem.
Soc. 2009, 131, 6078.
(4) For the suggested intermediacy of Fe(−II), see: (a) Furstner, A.;
̈
Leitner, A. Angew. Chem., Int. Ed. 2002, 41, 609. (b) Furstner, A.;
̈
Leitner, A.; Men
́
dez, M.; Krause, H. J. Am. Chem. Soc. 2002, 124, 13856.
(c) Furstner, A.; Martin, R.; Krause, H.; Seidel, G.; Goddard, R.;
̈
Lehmann, C. W. J. Am. Chem. Soc. 2008, 130, 8773.
(5) (a) Tamura, M.; Kochi, J. K. J. Am. Chem. Soc. 1971, 93, 1487.
(b) Tamura, M.; Kochi, J. J. Organomet. Chem. 1971, 31, 289.
(c) Tamura, M.; Kochi, J. K. Bull. Chem. Soc. Jpn. 1971, 44, 3063.
(6) Kleimark, J.; Hedstrom, A.; Larsson, P.-F.; Johansson, C.; Norrby,
̈
P.-O. ChemCatChem 2009, 1, 152.
(7) In this case, the model complexes contained ligands (Cp or Cp*)
that were markedly different from those in the most active precatalyst,
[Fe(ethene)4]2−, making direct comparison problematic. Under the
conditions used for the comparison (−30 °C), all of the complexes with
Cp or Cp* ligands performed poorly, irrespective of oxidation state. See
ref 4c.
(8) For use with arylzinc reagents, see ref 9a and: (a) Bedford, R. B.;
Huwe, M.; Wilkinson, M. C. Chem. Commun. 2009, 600. (b) Hatakeya-
ma, T.; Kondo, Y.; Fujiwara, Y.; Takaya, H.; Ito, S.; Nakamura, E.;
Nakamura, M. Chem. Commun. 2009, 1216.
The reaction profile of the coupling of 3a with 4a at −10 °C
catalyzed by 8b is essentially identical to that catalyzed by 1a.25
This is consistent with the existence of a common inter-
mediate, and the best explanation for this is that 1a is very rapidly
reduced to 8b in situ, with the latter being an active on-cycle
species. In contrast, 7 showed a reaction rate about 5 times lower
than that of 1a, indicating that it is most likely an off-cycle
species.26
(9) For use with arylboron reagents, see: (a) Bedford, R. B.; Hall, M.
A.; Hodges, G. R.; Huwe, M.; Wilkinson, M. C. Chem. Commun. 2009,
6430. (b) Hatakeyama, T.; Hashimoto, T.; Kondo, Y.; Fujiwara, Y.;
Seike, H.; Takaya, H.; Tamada, Y.; Ono, T.; Nakamura, M. J. Am. Chem.
Soc. 2010, 132, 10674.
(10) For Al-based nucleophiles, see: Kawamura, S.; Ishizuka, K.;
Takaya, H.; Nakamura, M. Chem. Commun. 2010, 46, 6054.
(11) For alkynylation of alkyl halides, see: Hatakeyama, T.; Okada, Y.;
Yoshimoto, Y.; Nakamura, M. Angew. Chem., Int. Ed. 2011, 50, 10973.
(12) For use with aryl Grignard reagents, see: Hatakeyama, T.;
Fujiwara, Y.-i.; Okada, Y.; Itoh, T.; Hashimoto, T.; Kawamura, S.; Ogata,
K.; Takaya, H.; Nakamura, M. Chem. Lett. 2011, 40, 1030.
(13) Complex 1a is now commercially available from Sigma-Aldrich.
(14) See the SI for details.
In summary, we have shown that when the precatalyst 1a is
used for cross-coupling reactions of benzyl halides with arylzinc
reagents, Fe(I) appears to be the lowest kinetically relevant
oxidation state. We are currently undertaking a detailed
investigation of the mechanism of the iron-catalyzed Negishi
reaction, not least to explain the unusual apparent second-order
dependence on the precatalyst,15 which implies a significantly
more complex mechanism than we previously supposed,8a,9a
and to elucidate the precise roles of the Fe(I) intermediates.
Furthermore, we are exploring the possible relevance of our
results to iron cross-coupling catalysis using other ligands and
other nucleophiles.
(15) At −10 °C with pretreatment of 1a with 4a, the reaction was
sufficiently slow to allow us to determine the empirical rate law for the
catalytic reaction: d[5a]/dt = k[1a](1.9 0.3)[3a](1.15 0.1)[4a](1.3 0.1). The
possibility that the second-order dependence in catalyst is actually due
to dissociation of the phosphine was negated by the observation that the
reaction is zeroth-order in added dpbz. Repeating the reaction without
pretreatment of 1a with 4a gave a far higher rate of reaction over the first
ASSOCIATED CONTENT
* Supporting Information
Experimental and computational details, kinetic data, EPR
spectra, crystallographic information (CIF), and reactivity of
complex 7. This material is available free of charge via the
Internet at http://pubs.acs.org.
■
S
10335
dx.doi.org/10.1021/ja303250t | J. Am. Chem. Soc. 2012, 134, 10333−10336

Journal of the American Chemical Society
Communication
2.5 min, after which time the reaction slowed enormously (see Figure
S6). Under these conditions, an accurate determination of the rate law
using sampling techniques proved difficult, but the data obtained were
broadly consistent with the empirical rate law determined above. See the
SI for kinetic data.
(16) We observed 50% conversion to product after ∼20 s and 12 min in
catalytic reactions run at 45 and −10 °C, respectively (compare Figure 1
with Figure S1).
(17) Lu, C. C.; Saouma, C. T.; Day, M. W.; Peters, J. C. J. Am. Chem.
Soc. 2007, 129, 4.
(18) Mankad, N. P.; Muller, P.; Peters, J. C. J. Am. Chem. Soc. 2010,
̈
132, 4083.
(19) The red color was the same as that observed in the vast majority of
catalytic reactions undertaken using 1a as a precatalyst here and in refs
8a and 9a. The remainder are reddish-brown.
(20) The bite angle and distortions from ideal trigonal-bipyramidal
geometry are remarkably similar to those observed in the previously
reported isoelectronic cobalt(II) complex [CoCl(dppe)2]
SnCl3·C6H5Cl. See: Stalick, J. K.; Corfield, P. W. R.; Meek, D. W.
Inorg. Chem. 1973, 12, 1668.
(21) (a) Ni, C.; Ellis, B. D.; Fettinger, J. C.; Long, G. J.; Power, P. P.
Chem. Commun. 2008, 1014. (b) Nguyen, T.; Merrill, W. A.; Ni, C.; Lei,
H.; Fettinger, J. C.; Ellis, B. D.; Long, G. J.; Brynda, M.; Power, P. P.
Angew. Chem., Int. Ed. 2008, 47, 9115. (c) Ni, C.; Fettinger, J. C.; Long,
G. J.; Brynda, M.; Power, P. P. Chem. Commun. 2008, 6045. (d) Ni, C.;
Ellis, B. D.; Stich, T. A.; Fettinger, J. C.; Long, G. J.; Britt, R. D.; Power,
P. P. Dalton Trans. 2009, 5401.
(22) Bouwkamp, M. W.; Bart, S. C.; Hawrelak, E. J.; Trovitch, R. J.;
Lobkovsky, E.; Chirik, P. J. Chem. Commun. 2005, 3406.
(23) Evans’ method was used. See: Evans, D. F. J. Chem. Soc. 1959,
2003.
(24) (a) Bart, S. C.; Chłopek, K.; Bill, E.; Bouwkamp, M. W.;
Lobkovsky, E.; Neese, F.; Wieghardt, K.; Chirik, P. J. J. Am. Chem. Soc.
2006, 128, 13901. (b) Irwin, M.; Jenkins, R. K.; Denning, M. S.; Kramer,
̈
T.; Grandjean, F.; Long, G. J.; Herchel, R.; McGrady, J. E.; Goicoechea,
J. M. Inorg. Chem. 2010, 49, 6160.
(25) See Figure S11 for the reaction profiles.
(26) The off-cycle nature of 7 was further indicated by preliminary
mechanistic studies on the basis of both its catalytic and stoichiometric
reactions with 3a, 4a, and related substrates; these are discussed in detail
in the SI.
10336
dx.doi.org/10.1021/ja303250t | J. Am. Chem. Soc. 2012, 134, 10333−10336