A synthetic and mechanistic investigation into the cobalt(i) catalyzed amination of aryl halides

Article abstract of DOI:10.1039/c4sc01257g

Employing first-row transition metals in catalytic two-electron transformations remains a synthetic challenge. In order to overcome the common and often deleterious single-electron reactivity, an electron rich ligand was targeted on cobalt. Herein, we report the Co(i) catalyzed amination of aryl halides with lithium hexamethyldisilazide. This transformation features (PPh₃)₃CoCl (1) as the catalyst and affords structurally diverse and electronically varied primary arylamines in good chemical yields, with the scope of the reaction featuring arylamines that cannot be synthesized via traditional metal-catalyzed amination routes, including 4-aminophenylboronic acid pinacol ester. Stoichiometric reactivity revealed that (PPh₃)₂CoN(SiMe₃)₂ (2) is likely generated within the catalytic cycle and could be independently synthesized from the reaction of (PPh₃)₃CoCl with LiN(SiMe₃)₂. Catalytic reactivity featuring the Co-amide complex, (PPh₃)₂CoN(SiMe₃)₂, showed that it is a competent catalyst, implying that the (PPh₃)₃CoCl may be serving as a pre-catalyst in the reaction. Both stoichiometric and kinetic studies support the catalytic cycle involving a Co(i) complex. Catalytic reactions featuring Co(ii) complexes resulted in undesired biaryl formation, a product that is not observed under standard catalytic conditions and any productive catalytic reactivity likely arises from an in situ reduction of Co(ii) to Co(i). A Hammett study was carried out to differentiate between a closed-shell or radical mechanism, the results of which are consistent with the proposed closed-shell mechanism. Initial studies indicate that this reactivity may be expanded to other bulky nucleophiles. This journal is

Full text of DOI:10.1039/c4sc01257g

Chemical
Science
View Article Online
View Journal
EDGE ARTICLE
A synthetic and mechanistic investigation into the
cobalt(I) catalyzed amination of aryl halides†
Cite this: DOI: 10.1039/c4sc01257g
*
Marshall R. Brennan, Dongyoung Kim and Alison R. Fout
Employing first-row transition metals in catalytic two-electron transformations remains a synthetic
challenge. In order to overcome the common and often deleterious single-electron reactivity, an
electron rich ligand was targeted on cobalt. Herein, we report the Co(I) catalyzed amination of aryl
halides with lithium hexamethyldisilazide. This transformation features (PPh3)3CoCl (1) as the catalyst and
affords structurally diverse and electronically varied primary arylamines in good chemical yields, with the
scope of the reaction featuring arylamines that cannot be synthesized via traditional metal-catalyzed
amination routes, including 4-aminophenylboronic acid pinacol ester. Stoichiometric reactivity revealed
that (PPh3)2CoN(SiMe3)2 (2) is likely generated within the catalytic cycle and could be independently
synthesized from the reaction of (PPh3)3CoCl with LiN(SiMe3)2. Catalytic reactivity featuring the Co–
amide complex, (PPh3)2CoN(SiMe3)2, showed that it is
a competent catalyst, implying that the
(PPh3)3CoCl may be serving as a pre-catalyst in the reaction. Both stoichiometric and kinetic studies
support the catalytic cycle involving a Co(I) complex. Catalytic reactions featuring Co(II) complexes
resulted in undesired biaryl formation, a product that is not observed under standard catalytic conditions
and any productive catalytic reactivity likely arises from an in situ reduction of Co(II) to Co(I). A Hammett
study was carried out to differentiate between a closed-shell or radical mechanism, the results of which
are consistent with the proposed closed-shell mechanism. Initial studies indicate that this reactivity may
be expanded to other bulky nucleophiles.
Received 1st May 2014
Accepted 29th July 2014
DOI: 10.1039/c4sc01257g
www.rsc.org/chemicalscience
Carbon–nitrogen bonds are abundant in organic molecules and limited to two cobalt(II) catalyzed processes; the rst example
methods for their direct construction in amination reactions couples pyridyl-chlorides with piperidine7 and the second is an
from aryl halides are important in small molecule synthesis.1 intramolecular C–N bond coupling.8 Both of these reactions
The direct synthesis of primary anilines has been carried out suffer from dramatically limited substrate scopes compared
using palladium catalysts, ammonia, ammonia surrogates and with the palladium-mediated methods.
aryl halides.2 In carbon–nitrogen bond formation it is oen the
The intermolecular amination of unactivated aryl halides
large substrate scope afforded by the palladium catalysts that is with cobalt has, to the best of our knowledge, not been
ideal, however the costly metals and ligands necessary to demonstrated, whereas palladium-mediated processes have
perform these transformations are not. To this end, base metals been quite successful.2 A key example by Hartwig and co-
have become attractive targets in amination reactions typically workers uses a Pd(0) catalyst in the reaction of lithium hexam-
mediated by their heavier congeners in part due to the cost and ethyldisilazide (LiN(SiMe3)2) with aryl halides.2e–g In these
environmental advantages; but more excitingly, due to the reactions, LiN(SiMe3)2 can be cross-coupled with a variety of aryl
potential of these catalysts to facilitate new reactivity.3 Amina- bromides and chlorides, resulting in the formation of
tion reactions featuring copper catalysts (Ullmann coupling) ArN(SiMe3)2 in good yields. These N-aryldisilazides can subse-
have been studied extensively in the coupling of aryl iodides, quently be deprotected under acidic conditions to yield the
and to a lesser extent, aryl bromides with nitrogen nucleophiles; primary aniline.2e–g Ortho-substituted aryl halides failed to react
however, the scope of reactivity is limited.4 Nickel catalyzed under these conditions, which the authors attribute to the steric
amination has also been explored.5 Although cobalt mediated bulk of the nucleophile.2e Buchwald and co-workers subse-
cross-coupling reactions are known,6 C–N bond coupling is quently conrmed this hypothesis by demonstrating that a less-
encumbered primary silylamine, Ph3SiNH2, could successfully
yield the corresponding ortho-substituted aniline following
deprotection.2f Since coupling a-branched amine nucleophiles
is challenging, these reactions were targeted in an effort to
evaluate the potential of cobalt-mediated catalysis in C–N cross-
coupling reactivity.
School of Chemical Sciences, University of Illinois at Urbana-Champaign, 600 S.
Mathews Avenue, Urbana, Illinois 61801, USA. E-mail: fout@illinois.edu
† Electronic supplementary information (ESI) available: Full experimental,
crystallographic and spectroscopic data. CCDC 1000460 and 1000461. For ESI
and crystallographic data in CIF or other electronic format see DOI:
10.1039/c4sc01257g
This journal is © The Royal Society of Chemistry 2014
Chem. Sci.

View Article Online
Chemical Science
Edge Article
ꢀ
Computational work has suggested that the greater density (<100 C) did not result in coupled product (Table S1,† entries
of states for rst-row metals should allow for two-electron redox 34–35).
processes, such as oxidative addition, to be more energetically
Since other nucleophiles (e.g. lithium amide, lithium diiso-
favourable than the analogous reactions with second-row propylamide, and lithium ditertbutylamide) did not couple
metals.9 Iron complexes featuring electron rich phosphine when substituted for LiN(SiMe3)2, the number of equivalents of
ligands have been implicated in both oxidative addition and LiN(SiMe3)2 necessary for the transformation was investigated.
reductive elimination reactions.10 More recently, Peters and co- When only one equivalent of LiN(SiMe3)2 was utilized, the
workers reported the oxidative group transfer of a Co(I) mole- formation of both aminated product and C–C coupled isomers
cule, (PhBP3)CoPMe3 with tolyl azide which afforded a cobal- of methylbiphenyl were observed from the reaction of an aryl
t(III)-imido species.11 Likewise, the oxidative addition of H2 to radical with the solvent.16 The formation of the biaryl product
Co(I) complexes has been observed.12 Bernskoetter and co- was not detected when 2.6 equivalents of nucleophile was used
workers concisely demonstrated through deuterium labeling and increasing nucleophile loading to 5.2 equivalents resulted
and trapping experiments that (PMe3)3CoI(CH3)2, formed from in the formation of aniline in 99% yield.
oxidative addition of methyl iodide to (PMe3)4CoCH3, could
Monitoring the reaction by 1H NMR spectroscopy at elevated
undergo stoichiometric reductive elimination to yield ethane temperature illustrated the loss of PPh3 under the reaction
and an equivalent of (PMe3)3CoI via a concerted two-electron conditions, suggesting that the addition of an electron rich
process.13 While the role of the alkyl ligand cannot be dis- bidentate phosphine ligand may improve overall catalysis. The
counted, it suggested that a Co(I/III) redox cycle supported by addition of a variety of chelating phosphine ligands (15 mol%)
phosphine ligands may be sufficient, under the right condi- to 1 resulted in the formation of aniline in higher yields (DPPF
tions, to allow catalysis to occur. The Chirik group has recently (97%), BINAP (98%) and DPEPhos (95%), Table S1,† entries 10–
reported several oxidative addition reactions12k,14 with a (PNP) 13). The addition of DPPE, however, resulted in a lower yield
CoCH3 complex as well as cobalt-catalyzed C–H borylation.15 (68%) than any of the other phosphines, including PPh3. DPE-
The commonality of all of the described examples is a low-val- Phos was chosen as the added phosphine on the basis of
ent, electron rich cobalt(I) center; however, mechanistic studies improved yield and overall cost effectiveness. The reaction
are necessary to differentiate between the common single conditions were optimized on the basis of these results to
electron transfer versus two-electron chemistry by base metal include 2.6 equivalents of nucleophile, 15 mol% of DPEPhos,
ꢀ
catalysts.
and 7.5 mol% of 1 at 100 C in toluene. Lowering the catalyst
Herein, we present the rst cobalt(I)-catalyzed amination of loading from 7.5 mol% to 4 mol% and 2 mol% resulted in
unactivated aryl iodides with LiN(SiMe3)2 for the formation of reduced product yields.
various anilines, including those not accessible through other
cross-coupling methods. The reaction is catalyzed by a single-
component triphenylphosphine-ligated cobalt(I) precursor.
Substrate scope
Stoichiometric investigations, trapping experiments, and linear With the optimized reaction conditions in hand, the scope of
free energy studies have been conducted and are inconsistent substrates was explored (Table 1). Under the optimized condi-
with single electron transfer featuring a Co(II) intermediate.
tions, aryl iodides are the most reactive substrates; aryl
bromides such as bromobenzene, 4-bromobiphenyl, and 2-
bromonapthalene furnish the aminated product in good yield,
albeit diminished relative to the analogous aryl iodides. Aryl
chlorides, however, are unreactive under the reaction condi-
tions, even for prolonged reaction times. Although aryl triates
Results and discussion
Optimization of cobalt-catalyzed amination
The cobalt-catalyzed amination of aryl iodides with LiN(SiMe3)2 typically tend to result in similar two-electron chemistry as aryl
was initially investigated. The reaction of one equivalent of iodides; the reactivity observed led to an intractable mixture of
iodobenzene with 2.6 equivalents of LiN(SiMe3)2 in the presence organic products. The reaction of aryl iodides featuring elec-
of 7.5 mol% (PPh3)3CoCl (1) reuxing in toluene for 12 h cleanly tron-donating groups (Me, naphthyl, Ph and OTBS) furnished
resulted in the formation of the N,N-bis(trimethylsilyl)aniline in the aminated product in excellent yields (entries 2–5) while
77% yield, which readily hydrolyzes with acid to aniline. In the electron-withdrawing groups were also tolerated in moderate
absence of 1 no coupling product was observed, suggesting that yields (entries 7–9).
this is a cobalt-mediated process. The relative importance of
The scope of this cobalt-catalyzed reaction tolerates several
solvents, temperature, equivalents of silylamide and ligands on notable substrates not amenable to the existing Pd-mediated
the reaction was evaluated using iodobenzene as a model methods. For example, a boronate ester (entry 6) is tolerated
substrate and the results are listed in Table S1.† The choice of under the reaction conditions to afford 4-aminophenylboronic
solvent was critical for success as coordinating solvents inhibit acid pinacol ester in excellent yield (91%); palladium-catalyzed
reactivity (Table S1,† entries 1–8). Among the examined methods require installation of the amine prior to the boronate
solvents, only non-coordinating solvents (toluene (77%), functionality.17 In a second example, the use of 4-chlor-
benzene (53%) and hexanes (46%)) gave rise to the coupled oiodobenzene (entry 8) as a substrate resulted in the formation
amine, with toluene resulting in the highest yield likely due to of p-chloroaniline, leaving the chloride as a functional group
the increased solubility of the catalyst. Lower temperatures handle for further elaboration by other cross-coupling methods.
Chem. Sci.
This journal is © The Royal Society of Chemistry 2014

View Article Online
Edge Article
Chemical Science
Table 1 Substrate scopea
Entry
1
Aryl halide
GC yield of RPhN(SiMe3)2
Product
Isolated yield
X = I
95%
X = I
X = Br
91%
60%
X = Br
X = Cl
X = I
65%
No RXN
99%
X = I
X = Br
96%
61%
X = Br
69%
2
3
X = I
X = Br
99%
69%
X = I
X = Br
94%
63%
X = I
X = Br
77%
45%
X = I
X = Br
74%
39%
4
5
99%
96%
90%
91%
6
7
64%
58%
8
73%
99%
37%
70%
94%
30%
9
10
11
77%
69%
a
Conditions: aryl halide (1 mmol), LiN(SiMe3)2 (2.6 mmol), 1 (0.075 mmol), DPEPhos (0.15 mmol), toluene (5 mL) 100 ꢀC, 12 h. Yields reported as
the average of three runs.
Conversely, the reaction with 4-bromoiodobenzene resulted in is used as a substrate, 2,5-diaminotoluene could be isolated in
the formation of p-phenylenediamine (entry 9) in 94% yield 74% yield, indicating that the addition of the electron-donating
aer acidic workup.
group resulting from the rst coupling enhanced the reactivity,
Although this cross-coupling reaction can tolerate ortho- ultimately overcoming the steric constraints of the o-methyl.
substitution (entry 10), the yields are diminished dramatically
with respect to the meta- and para-substituted counterparts.
Previously reported palladium-catalyzed reactions failed in the
case of ortho-substituted aryl halides and bis(trimethylsilyl)
amide likely due to the signicant bulk of the two trimethylsilyl
groups.2g Interestingly, when 2-bromo-5-iodotoluene (entry 11)
Mechanistic considerations
Based on the previously established literature precedent of 1 to
participate in radical chemistry6 we sought to understand this
catalytic process in more detail to delineate between a two-
electron pathway and a radical or caged radical mechanism. As
This journal is © The Royal Society of Chemistry 2014
Chem. Sci.

View Article Online
Chemical Science
Edge Article
such, both stoichiometric and catalytic reactions were explored reported in the radical reaction with the solvent (o : m : p ¼
with iodobenzene as the model substrate. In palladium 1.6 : 1.0 : 2.0).16a
coupling, benzyne intermediates have been proposed, and the
Stoichiometric reactions between 1 and iodobenzene were
lower yields observed by electron decient substrates are also conducted in the presence of LiN(SiMe3)2. If 1 and iodo-
ꢀ
consistent with this mechanistic pathway. However, the expec- benzene are stirred together at 100 C as previously described
ted regioisomers from benzyne formation are never observed in before the addition of LiN(SiMe3)2, the only product obtained
our reaction.2e,18 Furthermore, replacing toluene with durene as aer 12 hours is a mixture of biaryls. However, when
the solvent in the catalytic transformation did not result in LiN(SiMe3)2 is added before the reaction is heated, the ami-
formation of the Diels–Alder adduct, which has been shown to nated product is isolated in 70% yield; the same result is
trap the putative benzyne intermediate.19 In a recently reported obtained when 1 and LiN(SiMe3)2 are stirred at 100 C prior to
ꢀ
Ullmann coupling, C–N bond formation in carbazole proceeds the addition of iodobenzene. Chlorobenzene is not observed in
via a photo-induced single electron transfer.20 To determine any reaction in which the nucleophile is added before or
whether the observed coupling proceeds through a photoin- concurrently with iodobenzene. These results suggest that a
duced aminyl radical generated from LiN(SiMe3)2, the reaction non-productive halide metathesis can occur between the cobalt
was conducted in the absence of light under standard catalytic complex and iodobenzene, irreversibly converting the reactive
conditions. Running the reaction in the dark did not diminish aryl iodide into an inert aryl chloride, which then may be
the yield of aniline formation (96%) (Table S1,† entry 16).
consumed by a Co(II) species arising from the known dispro-
To investigate the possibility of radical intermediates during portionation pathway of the unstable (PPh3)2CoI.22 The pres-
catalysis, various radical traps were employed including 1,4- ence of nucleophile prevents this equilibration and results in
cyclohexadiene, BHT (2,6-di-tert-butyl-4-methylphenol) and 1,1- the formation of the desired arylamine product.
diphenylethene. Under standard catalytic conditions the
primary aryl amine was formed as the sole product and the
Cobalt(II) as an intermediate
unconsumed additive was detected by GC-MS. Mechanistic
studies of the elimination of ethane from (PMe3)3Co(CH3)2I by
Bernskoetter and co-workers suggested that these results are
evidence against an uncaged radical intermediate; however, a
caged radical could be present.13 The addition of other radical
traps such as TEMPO and ClCPh3, not surprisingly, reacted with
1 or the nucleophile prior to being exposed to the catalytic
conditions, resulting in a myriad of products.
Although evidence for radical formation was not detected with
various radical trapping experiments (vide supra), the potential
for the formation of a Co(II) intermediate within the catalytic
cycle is plausible. In this regard, (DPEPhos)CoCl2 was
substituted for 1 in the catalytic reaction. Surprisingly, the
reaction formed both the desired arylamine product and biaryl
isomers in a 1 : 1 mixture (Table 2). This is in stark contrast to
the reaction with 1, which forms only arylamine under standard
catalytic conditions. Furthermore, (DPEPhos)Co(N(SiMe3)2)2
was prepared23 in situ from the reaction of two equivalents of
DPEPhos with Co2(N(SiMe3)2)4 and submitted to the reaction
conditions, yielding an 85 : 15 mixture of arylamine and biaryl
isomers. Increasing the amount of (DPEPhos)Co(N(SiMe3)2)2 to
Stoichiometric reactivity
In order to assess the role of single- versus two-electron transfer
reactions, we sought to develop an understanding of the
behavior of 1 in a stoichiometric regime. Exposure of 1 to one
equivalent of iodobenzene at room temperature in toluene did
1
Table 2 Divergent reactivity of Co(I) and Co(II)
not result in a reaction as determined by GC-MS and H NMR
spectroscopy. Monitoring the heated mixture (100 ꢀC) by GC-MS
resulted in quantitative conversion of the iodobenzene to
chlorobenzene in the rst four hours of the reaction (Scheme 1).
Additional heating (12 h) converted the chlorobenzene quanti-
tatively to biaryl. The organic products were then removed by
extraction with toluene, from which (PPh3)2CoI2 was isolated.21
The ratio of biaryl isomers observed is consistent with that
Entry
X
[Co]
Aniline
Biaryl
1
2
3
4
5
6
7
8
I
I
I
Cl
Cl
Cl
I
I
I
(PPh3)3CoCl (1)
(DPEPhos)CoCl2
(DPEPhos)Co(N(SiMe3)2)2
(PPh3)3CoCl (1)
(DPEPhos)CoCl2
(DPEPhos)Co(N(SiMe3)2
(PPh3)3CoCl (1) + Zn
(DPEPhos)CoCl2 + Zn
(DPEPhos)Co(N(SiMe3)2)2 + Zn
(PPh3)3CoCl (1) + Zn
(DPEPhos)CoCl2 + Zn
(DPEPhos)Co(N(SiMe3)2)2 + Zn
77%
50%
85%
0%
0%
0%
65%
68%
69%
0%
0%
0%
0%
50%
15%
0%
95%
98%
0%
0%
0%
0%
0%
9
10
11
12
Cl
Cl
Cl
0%
Scheme 1 Stoichiometric reaction of 1 with iodobenzene.
Chem. Sci.
This journal is © The Royal Society of Chemistry 2014

View Article Online
Edge Article
Chemical Science
15 mol% complex increased the ratio of arylamine to biaryls Compound 1 initially reacts with an equivalent of LiN(SiMe3)2 to
(70 : 30). produce a new Co(I) species, (PPh3)2CoN(SiMe3)2 (2) which is
Given the distinct reactivity of 1 and the cobalt(II) catalysts, we proposed to be the active catalyst. Next, oxidative addition of the
sought to reconcile the role of each. As previously described, 1 did aryl halide results in the formation of a Co(III) species,
not react with chlorobenzene under the standard reaction (PPh3)2CoPhI(N(SiMe3)2 (3). Given the need for multiple equiv-
conditions and offered an opportunity to understand the diver- alents of nucleophile, this species likely undergoes a second
gent reactivity proles of the two oxidation states. As described in substitution with nucleophile prior to reductive elimination of
Table 2, the reactions of (DPEPhos)CoCl2 or (DPEPhos) the product. Based on this, we sought several avenues to isolate
Co(N(SiMe3)2)2 with chlorobenzene as a substrate differentiate proposed intermediates within the catalytic cycle.
themselves from the reactions of both Co(I) and Co(II) with iodo-
The amide species, (PPh3)2Co(N(SiMe3)2) (2) can be isolated
benzene. While (DPEPhos)CoCl2 and (DPEPhos)Co(N(SiMe3)2)2 in 93% yield from a reaction between (PPh3)3CoCl and
form arylamine from iodobenzene, the sole isolated product from LiN(SiMe3)2 in toluene at room temperature (Fig. 1A).25 X-ray
the reaction with chlorobenzene is biaryl isomers, suggesting that quality crystals of 2 were grown from a cold concentrated
a Co(II) species is likely responsible for the formation of the biaryl solution of Et2O (Fig. 1). Compound 2 displays a three-coordi-
byproducts. If a Co(II) species was accessed in our catalytic cycle, nate Co(I) ion in a trigonal planar geometry with P–Co–N bond
then reacting 1 (under standard conditions) with a 50 : 50 mixture angles of 129.51(8)ꢀ and 123.78(8)ꢀ and a Co–N distance of
˚
of iodo- and chlorobenzene should result in the isolation of biaryl 1.918(2) A. Importantly, this complex is both catalytically
products. Instead these reaction conditions resulted in no biaryl (Fig. 1B) and stoichiometrically competent (Fig. 1C) for the
products and a mixture of 50% of aniline and unconsumed formation of product in yields mildly better than that of 1. This
chlorobenzene (50% via GC-MS).
suggests that 1 may function as a pre-catalyst, whereby initial
To understand this reaction further, zinc metal was added to reaction with LiN(SiMe3)2 yields the catalytically active
several reaction mixtures as described in Table 2. The addition species, 2.
of a reductant to standard catalytic conditions with 1 as the
Efforts to isolate a Co(III) species from the reaction of 2 with
catalyst should have no effect on catalysis if a cobalt(I) species is aryl halides were hindered by the reactivity of the products; at
necessary. However, if under catalytic conditions the Co(I) the temperatures required for a productive reaction to occur, all
species is oxidized to Co(II) and that is the productive catalyst, intermediates proceeded to arylamine before a complex could
then catalysis should be inhibited. Zinc metal has not been be isolated. The increased yield upon addition of a bidentate
shown to be a strong enough reductant to access Co(0).24 The phosphine is consistent with reductive elimination from the
addition of zinc (Table 2, entry 7) resulted in a slightly decreased ve-coordinate species, 4. Additional equivalents of triphenyl-
yield of the desired aniline product, but no biaryl species were phosphine or DPEPhos did not hamper the overall reaction.
formed (the diminished yield is possibly due to the heteroge- Furthermore, monitoring a catalytic reaction by 19F NMR
neity of the reaction mixture under these conditions; a similar spectroscopy with the addition of excess tris(4-uorophenyl)
decrease in yield is observed when the reaction is run without phosphine did not result in new 19F NMR signals. The disso-
stirring).
ciation of a phosphine ligand is not likely and reductive elimi-
If Co(I) is the active species, the addition of zinc to the nation to the three-coordinate species, 2, is not rate
reaction mixture should produce arylamine exclusively, determining.
regardless of whether the starting complex is Co(I) or Co(II) due
to in situ reduction of Co(II) species present in the reaction
mixture. Conversely, if arylamine is not observed under these
conditions, it may speak to the involvement of a Co(II) inter-
mediate in the catalytic cycle. Gratifyingly, when either (DPE-
Phos)CoCl2 or (DPEPhos)Co(N(SiMe3)2)2 are submitted to the
standard catalytic conditions in the presence of Zn, the desired
aniline product is the sole isolated product in 68% yield. These
results are consistent with reduction of the Co(II) species to
Co(I). To test this hypothesis further, replacing iodobenzene
with chlorobenzene under standard catalytic conditions should
result in no reaction, since chlorobenzene is not a productive
substrate for the amination reaction. As listed in Table 2, entries
11 and 12, no aniline or biaryl was detected and chlorobenzene
could be reisolated from the reaction mixture.
Catalytic cycle
The previously described stoichiometric and catalytic studies
support Co(I) as the active catalyst. Based on these results the
following Co(I)/Co(III) catalytic cycle is proposed (Scheme 2).
Scheme 2 Proposed catalytic cycle.
This journal is © The Royal Society of Chemistry 2014
Chem. Sci.

View Article Online
Chemical Science
Edge Article
Fig. 1 Synthesis (A) and reactivity (B and C) of 2. Solid state structure of (PPh3)2CoN(SiMe3)2 (2) with thermal ellipsoids drawn at 50% (right).
Hydrogen atoms have been omitted for clarity. Selected bond lengths (A˚) and angles (ꢀ): Co–N 1.918(2), Co–P 2.2545(9) and 2.2425(9); P–Co–N
129.51(8) and 123.78(8), P–Co–P 106.35(3).
transition state. This result is consistent with an oxidative
addition-type mechanism, whereby a late, largely polarized
transition state, is stabilized by inductive donation from the
substrate. The line made up of the para-uoro-, chloro-, and
triuoromethyl-substituted substrates (s > 0) yields a r value of
0.36, indicating a build-up of negative charge in the transition
state. Consequently, this can be characterized as a reductive
elimination mechanism, wherein the small r value suggests an
earlier, less polar transition state. These results are in agree-
ment with similar studies carried out on stoichiometric oxida-
tive addition and reductive elimination reactions with
palladium.28 Unlike palladium, however, the reductive elimi-
nation pathway observed with cobalt appears to be less sensitive
to electronic effects, suggesting that the h2-arene complex oen
invoked for the reductive elimination of C–X bonds from aryl-
palladium amido complexes28 does not form with cobalt. It is
worth noting that, despite the change in mechanism, the rela-
tive rates for electron decient substrates are lower than those
for the electron rich substrates, and may suggest that the poorer
yields observed for those substrates are a result of their diffi-
culty in reductively eliminating the arylamine product.
Hammett study
Given the difficulty in isolating a Co(III) intermediate, a Ham-
mett study was performed to differentiate between closed-shell
or radical mechanisms by comparing the ts of s and sꢁ
values,26 respectively, as previously demonstrated by Norrby and
co-workers, as well as others.27 Since the correlation parameters
and r values for oxidative addition and reductive elimination
are known from stoichiometric studies with palladium
complexes,28 the substituent effect on the relative rate of the
reaction is a powerful tool for understanding this cobalt-medi-
ated amination reaction.
The six competition reactions between unsubstituted and
various para-substituted iodobenzenes were monitored by GC
using the appearance of products versus an internal standard of
mesitylene. To simplify the reaction set-up and analysis, the
reaction was run without DPEPhos; instead 5.2 equivalents of
LiN(SiMe3)2 were employed. We propose that the reaction order
is the same for all substrates studied; under this assumption,
the krel values could be obtained in a manner previously
described (see eqn S1† for more details).27 The kinetic studies
all yielded straight lines (r2 > 0.96, Fig. S1†).
The relative rates and different s values used for the Ham-
mett plot are listed in Tables S2 and S3,† respectively. Fig. 2
shows the plots of log(krel) vs. sꢁ values (depicted by A) or s
values (depicted by -). Regardless of which open-shell
parameter, sꢁ, is used, the total scatter is obtained (sꢁ values
described by Jiang26d are depicted in Fig. S3, others in S4†). The
poor t of the sꢁ values is inconsistent with a single-electron or
radical mechanism. The resulting concave plot for the best t of
standard closed-shell s constants (r2 ¼ 0.96 for s < 0 and 0.98 for
s > 0) is in good agreement with the catalytic coupling of
LiN(SiMe3)2 and aryl halides proceeding via a closed-shell
mechanism.
The existence of two lines in the Hammett correlation tted
with s values suggests that a change in mechanism or a change
in the transition state occurs when more electron decient
substrates are used. The line consisting of the para-methoxy-,
methyl-, and phenyl-substituted substrates (s < 0) yields a r
value of ꢂ1.65, indicative of a build-up of positive charge in the
Alternative nucleophiles
Another important conclusion that can be drawn from the
shape of the plot in Fig. S4† is that at least one intermediate
connects the starting material to the product in a stepwise
mechanism, rather than a concerted process, such as s-bond
metathesis. A corollary of this observation is that such an
intermediate might be intercepted by nucleophiles other than
LiN(SiMe3)2. Unfortunately, attempts to cross-couple other
nucleophiles, such as arylmagnesium halides and even other
lithium amides, using 1 as a catalyst were unsuccessful.
However, the coupling of lithium piperidide with iodo-
benzene could be accomplished, albeit in very low yields (12%),
when LiN(SiMe3)2 was also present in the reaction mixture
(Table S1,† entry 42). Importantly, no cross-coupling of
LiN(SiMe3)2 to form N,N-bis(trimethylsilyl)aniline was observed
under these conditions. Given that (PPh3)2CoN(SiMe3)2 is the
likely resting state of the catalyst, this suggested that the role of
Chem. Sci.
This journal is © The Royal Society of Chemistry 2014

View Article Online
Edge Article
Chemical Science
Fig. 2 Hammett correlation diagrams. On the left, Hammett plot of log(krel) vs. s (closed-shell); on the right, Hammett plot of log(krel) vs. sꢁ
(Jiang, open-shell). Plots of other s values can be found in the ESI.† The relative rate constants, krel, are calculated using a plot of ln([X0]/[X]) ¼ krel
ꢃ ln([H0]/[H]), where X0 and H0 refer to the initial concentrations of functionalized and unfunctionalized iodoarenes, respectively.
LiN(SiMe3)2 in the reaction may also be as a spectator ligand
increasing the electron richness of the metal center. Under such
Table 3 Coupling of alternative nucleophiles with 2a
an assumption, the initial substitution of the cobalt(I) chloride
proceeds to yield (PPh3)2CoN(SiMe3)2, which then undergoes
oxidative addition to yield an intermediate poised to react with
another equivalent of nucleophile. Reductive elimination from
this intermediate may furnish multiple products resulting from
reaction with the incoming nucleophile, as well as LiN(SiMe3)2.
To test this hypothesis, 2 was used as the catalyst for the
cross-coupling of other nucleophiles. Complex 2 was initially
used as the catalyst to ensure that the stoichiometry and asso-
ciation of N(SiMe3)2 to the metal was exactly 1 : 1. The reaction
of lithium piperidine and iodobenzene with 2 as the catalyst
(Table 3, entry 1) was accomplished in moderately better yields
than with 1. Aer 12 hours, unreacted iodobenzene was present
but increased reactions times did not improve the overall yield.
Gratifyingly, lithium N-trimethylsilylaniline (entry 2), generated
in situ from the parent amine and butyllithium, and mesi-
tylmagnesium bromide (entry 6) cross-coupled cleanly and
selectively under the standard reaction conditions with 2 as the
catalyst. Furthermore, protected alkylamines, including those
bearing bulky a–substituents, cross-coupled in good to excellent
yields (Table 3, entries 3–5). This result is intriguing since
comparable palladium-mediated methods suffer from severe
limitations in cross-coupling bulky alkylamines, suggesting that
further optimization of this method may simplify the synthesis
of highly encumbered arylamines. The cross-coupling of tert-
butyltritylcarbamate (Table 3, Entry 5) represents the only
reported example of cross-coupling involving a tritylamine
synthon. The bis(trimethylsilyl)amide ligand is essential in the
coupling of these bulkier nucleophiles and the investigation
into its role is currently underway in our laboratories in addi-
tion to the expansion of this reaction scope.
Entry
1
Nucleophile
Product
Isolated yield
20%
2
3
15%b
93%
4
5
45%b
58%
31%
6
Conclusions
a
Conditions: iodobenzene (1 mmol), nucleophile (4.5 mmol), 2 (0.075
mmol), toluene (5 mL) 100 ꢀC, 12 h. Yields reported as the average of
three runs. b Deprotected with 1 N HCl prior to isolation.
The identication of a model complex through which cobalt-
mediated two-electron redox catalysis could be studied has been
This journal is © The Royal Society of Chemistry 2014
Chem. Sci.

View Article Online
Chemical Science
Edge Article
presented, providing a key entry into the library of rst-row
transition metal complexes capable of catalytic two-electron
transformations. Both stoichiometric and kinetic studies
support the assignment of a closed-shell mechanism for the
catalytic amination of aryl halides with (PPh3)3CoCl as the
catalyst. Reactions with various trapping agents disfavored
mechanisms invoking uncaged radicals or aryne intermediates.
Differentiating between a caged radical intermediate and a two-
electron pathway was less straightforward, and was investigated
in two ways.
Chem. Soc., 1997, 119, 6054; (c) N. H. Park, G. Teverovskiy
and S. L. Buchwald, Org. Lett., 2014, 16, 220; (d) E. Brenner
and Y. Fort, Tetrahedron Lett., 1998, 39, 5359; (e)
E. Brenner, R. Schneider and Y. Fort, Tetrahedron, 1999, 55,
12829; (f) E. Brenner, R. Schneider and Y. Fort, Tetrahedron
Lett., 2000, 41, 2881; (g) C. Desmartes, R. Schneider and
Y. Fort, Tetrahedron Lett., 2000, 41, 2875; (h) C. Desmartes,
R. Schneider and Y. Fort, Tetrahedron Lett., 2001, 42, 247;
(i) C. Desmartes, R. Schneider and Y. Fort, J. Org. Chem.,
2002, 67, 3029.
Stoichiometric reactivity was examined by assessing the
reactivity of Co(II) intermediates. In these cases, the product
distributions included biaryl side products, which arise from a
known radical pathway. Any productive catalytic reactivity likely
arises from in situ reduction of a Co(II) intermediate to Co(I). The
addition of a reductant such as Zn prevented the formation of
biaryl, implying that a Co(II) intermediate is unlikely to be
operative in the amination reaction. Additional support for this
came via a Hammett study, which clearly correlates a multi-step,
closed-shell mechanism. While work is still ongoing to isolate
the key Co(III) intermediate, the collection of results support a
closed-shell mechanism catalyzed by a Co(I) species. In conclu-
6 (a) C. Gosmini, J.-M. Begouin and A. Moncomble, Chem.
Commun., 2008, 3221–3233, and references therein; (b)
M. Movassaghi and M. A. Schmidt, Angew. Chem., Int. Ed.,
2007, 46, 3725.
7 G. Toma, K.-I. Fujita and R. Yamaguchi, Eur. J. Org. Chem.,
2009, 2009, 4586.
8 P. Saha, M. A. Ali, P. Ghosh and T. Punniyamurthy, Org.
Biomol. Chem., 2010, 8, 5692.
9 (a) E. C. Volpe, P. T. Wolczanski and E. B. Lobkovsky,
Organometallics, 2010, 29, 364; (b) K. F. Hirsekorn,
E. B. Hulley, P. T. Wolczanski and T. R. Cundari, J. Am.
Chem. Soc., 2008, 130, 1183.
sion, this work demonstrates the rst example of cobalt-medi- 10 (a) C. A. Tolman, S. D. Ittel, A. D. English and J. P. Jesson, J.
ated intermolecular amination of unactivated aryl halides.
Am. Chem. Soc., 1978, 100, 4080; (b) C. A. Tolman, S. D. Ittel,
A. D. English and J. P. Jesson, J. Am. Chem. Soc., 1978, 100,
7577; (c) C. A. Tolman, S. D. Ittel, A. D. English and
J. P. Jesson, J. Am. Chem. Soc., 1979, 101, 1742; (d)
M. V. Baker and L. D. Field, J. Am. Chem. Soc., 1986, 108,
2825; (e) M. V. Baker and L. D. Field, J. Am. Chem. Soc.,
1987, 109, 7433; (f) L. D. Field, A. V. George and
B. A. Messerle, J. Chem. Soc., Chem. Commun., 1991, 19, 1325.
11 D. M. Jenkins, T. A. Betley and J. C. Peters, J. Am. Chem. Soc.,
2002, 124, 11238.
Acknowledgements
The authors thank the University of Illinois at Urbana-Cham-
paign for nancial support, Dr Danielle Gray for crystallo-
graphic assistance and Prof. Per-Ola Norrby for insightful
discussions.
Notes and references
12 (a) A. Misono, Y. Uchida, T. Saito and K. M. Song, Chem.
Commun., 1967, 419; (b) M. Hidai, T. Kuse, T. Hikita,
Y. Uchida and A. Misono, Tetrahedron Lett., 1970, 20, 1715;
(c) J. L. Hendrikse and J. W. E. Coenen, J. Catal., 1973, 30,
72; (d) D. L. Dubois and D. W. Meek, Inorg. Chim. Acta,
1976, 19, L29; (e) J. R. Winkler and H. B. Gray, Proc. Natl.
Acad. Sci. U. S. A., 2012, 109, 15127; (f) M. T. Mock,
R. G. Potter, M. J. O'Hagan, D. M. Camaioni,
W. G. Dougherty, W. S. Kassel and D. L. DuBois, Inorg.
Chem., 2011, 50, 11914; (g) M. C. Rakowski and
E. L. Meutterties, J. Am. Chem. Soc., 1977, 99, 739; (h)
M. Ingleson, H. Fan, M. Pink, J. Tomaszewski and
K. G. Caulton, J. Am. Chem. Soc., 2006, 128, 1804; (i)
T.-P. Lin and J. C. Peters, J. Am. Chem. Soc., 2013, 135,
15310; (j) S. S. Rozenel, R. Padilla, C. Camp and J. Arnold,
Chem. Commun., 2014, 50, 2612; (k) S. P. Semproni,
C. C. H. Atienza and P. J. Chirik, Chem. Sci., 2014, 5, 1956.
1 (a) A. Ricci, Modern Amination Methods, Wiley-VCH Verlag
¨
GmbH, Weinheim, 2000; (b) T. Kahl, K.-W. Schroder,
¨
¨
F. R. Lawrence, W. J. Marshall, H. Hoke and R. Jackh,
Aniline, Ullmann's Encyclopedia of Industrial Chemistry,
Wiley-VCH Verlag GmbH, Weinheim, 2000; (c) A. O. King
and N. Yasuda, Top. Organomet. Chem., 2004, 6, 205; (d)
C. Torborg and M. Beller, Adv. Synth. Catal., 2009, 351,
3027; (e) J. S. Carey, D. Laffan, C. Thomson and
M. T. Williams, Org. Biomol. Chem., 2006, 4, 2337.
2 (a) M. S. Driver and J. F. Hartwig, J. Am. Chem. Soc., 1996, 118,
7217; (b) J. P. Wolfe, J. Ahman, J. P. Sadighi, R. A. Singer and
S. L. Buchwald, Tetrahedron Lett., 1997, 38, 6367; (c) S. Jaime-
Figueroa, Y. Z. Liu, J. M. Muchowski and D. G. Putnam,
Tetrahedron Lett., 1998, 39, 1313; (d) C. W. Lim and S. Lee,
Tetrahedron, 2000, 56, 5131; (e) S. Lee, M. Jorgenson and
J. F. Hartwig, Org. Lett., 2001, 3, 2729; (f) X. H. Huang and
S. L. Buchwald, Org. Lett., 2001, 3, 3417; (g) D. Y. Lee and 13 H. Xu and W. Bernskoetter, J. Am. Chem. Soc., 2011, 133,
J. F. Hartwig, Org. Lett., 2005, 7, 1169; (h) D. S. Surry and
S. L. Buchwald, Chem. Sci., 2011, 2, 27.
3 P. J. Chirik and K. Weighardt, Science, 2010, 327, 5967.
4 D. Ma and Q. Cai, Acc. Chem. Res., 2008, 41, 1450.
5 (a) S. Ge, R. A. Green and J. F. Hartwig, J. Am. Chem. Soc.,
2014, 136, 1617; (b) J. P. Wolfe and S. L. Buchwald, J. Am.
14956.
14 S. P. Semproni, C. M. Milsmann and P. J. Chirik, J. Am. Chem.
Soc., 2014, 136, 9211.
15 J. V. Obligacion, S. P. Semproni and P. J. Chirik, J. Am. Chem.
Soc., 2014, 136, 4133.
Chem. Sci.
This journal is © The Royal Society of Chemistry 2014

View Article Online
Edge Article
Chemical Science
16 (a) C. T. To, T. L. Chan, B. Z. Li, Y. Y. Hui, T. Kwok, S. Y. Lam 23 B. D. Murray and P. P. Power, Inorg. Chem., 1984, 23, 4584.
and K. S. Chan, Tetrahedron Lett., 2011, 52, 1023; (b) 24 Q.-A. Chen, D. K. Kim and V. M. Dong, J. Am. Chem. Soc.,
T. L. Chan, C. T. To, B.-S. Liao, S.-T. Liu and K. S. Chan,
Eur. J. Inorg. Chem., 2012, 485.
17 (a) D. Qiu, L. Jin, Z. Zheng, H. Meng, F. Mo, X. Wang,
Y. Zhang and J. Wang, J. Org. Chem., 2013, 78, 1923; (b)
2014, 136, 3772.
25 Complex 2 was described in this article, but no synthetic
details or characterization were reported: D. C. Bradley and
M. H. Chisholm, Acc. Chem. Res., 1976, 9, 273.
S. L. Wheaton, S. H. Kabir, H. Zhang, C. M. Vogels, 26 (a) M. M. P. Grutters, J. I. van der Vlugt, Y. Pei, A. M. Mills,
A. Decken and S. A. Wescott, Cent. Eur. J. Chem., 2010, 8,
725; (c) M. Kumar, N. Kumar, V. Bhalla, P. R. Sharma and
Y. Qurishi, Chem. Commun., 2012, 48, 4719; (d) A. B. Patel,
R. V. Patel, P. Kumari, D. P. Rajani and K. H. Chikhalia,
Med. Chem. Res., 2012, 22, 367.
M. Lutz, A. L. Spek, C. Muller, C. Moberg and D. Vogt, Adv.
Synth. Catal., 2009, 351, 2199; (b) C. Hansch, A. Leo and
R. W. Ta, Chem. Rev., 1991, 91, 165; (c) C. Hansch and
H. Gao, Chem. Rev., 1997, 97, 2995; (d) X. Jiang and G. Ji, J.
Org.
Chem.,
1992,
57,
6051;
(e)
X.
Creary,
18 M. Beller, C. Breindl, T. H. Riermeier and A. Tillack, J. Org.
Chem., 2001, 66, 1403.
M. E. Mehrsheikh-Mohammadi and S. McDonald, J. Org.
Chem., 1987, 52, 3254.
19 (a) J. D. Roberts, D. A. Semenow, H. E. Simmons Jr and 27 (a) J. M. Dust and D. R. Arnold, J. Am. Chem. Soc., 1983, 105,
L. A. Carlsmith, J. Am. Chem. Soc., 1956, 78, 601; (b)
H. Pellissier and M. Santelli, Tetrahedron, 2003, 59, 701.
20 (a) S. E. Creutz, K. J. Lotito, G. C. Fu and J. C. Peters, Science,
2012, 338, 647; (b) A. C. Bissember, R. J. Lundgren,
S. E. Creutz, J. C. Peters and G. C. Fu, Angew. Chem., Int.
Ed., 2013, 52, 5129.
1221; (b) L. Keinicke, P. Fristrup, P.-O. Norrby and
R. Madsen, J. Am. Chem. Soc., 2005, 127, 15756; (c)
R. S. Sanchez and F. A. Zhuravlev, J. Am. Chem. Soc., 2007,
129, 5824; (d) B. S. Lane, M. A. Brown and D. Sames, J. Am.
Chem. Soc., 2005, 127, 8050; (e) P. Fristrup, L. B. Johansen
and C. H. Christensen, Catal. Lett., 2007, 120, 184; (f)
P. Fristrup, S. Le Quement, D. Tanner and P.-O. Norrby,
Organometallics, 2004, 23, 6160.
21 P. Braunstein, D. G. Kelly, Y. Dusausoy, D. Bayeul,
M. Lanfranchi and A. Tiripicchio, Inorg. Chem., 1994, 33, 233.
22 M. Aresta, M. Rossi and A. Sacco, Inorg. Chim. Acta, 1969, 3, 28 (a) J. F. Hartwig, Inorg. Chem., 2007, 46, 1936; (b)
227.
J. O. Schreck, J. Chem. Educ., 1971, 48, 103.
This journal is © The Royal Society of Chemistry 2014
Chem. Sci.