Key words nickel catalysis - photocatalysis - acylation - radical relay - C–N bond activation
- ketones
Ketones are ubiquitous chemical entities in everyday life. Found abundantly in nature,
ketone groups are also present in numerous value-added pharmaceuticals and agrochemicals,
as well as in fragrances, flavors, and fine chemicals. Alkyl ketones are extremely
important scaffolds in multistep organic syntheses and offer multiple opportunities
for late-stage derivatization of complex molecules.[1 ] The invention of new methods and strategies for the introduction of acyl functionalities
into aliphatic backbones is therefore an ever-present challenge that has significant
potential in various fields of chemistry.
Over the last decade, dual visible-light photoredox and nickel catalysis has emerged
as a powerful but operationally simple strategy for achieving, under very mild reaction
conditions, challenging C(sp2 )–C(sp3 ) bond disconnections that are beyond the reach of classical cross-coupling approaches.[2 ]
[3 ] Hence, this method brings prospects for innovation in the area of catalytic acylation
reactions of C(sp3 )-hybridized substrates,[4 ] and thereby offers a new paradigm for the synthesis of structurally diverse and
highly functionalized alkyl ketones. Within this context, resonance-destabilized N -acyl imides stand amongst other carbonyl-type compounds as promising acyl electrophiles[5 ] owing to their remarkable stability and their documented capability to engage in
nickel-catalyzed cross-coupling reactions through C–N bond activation to forge C(acyl)–C(sp3 ) bonds.[6 ] For instance, in 2017, Molander and co-workers[7 ] reported the use of N -acylsuccinimides as suitable acyl-transfer reagents for photoredox nickel-catalyzed
acylation of functionalized alkyl trifluoroborate nucleophiles.[8 ] This provided a strong impetus to develop complementary methods that would exploit
other readily available inexpensive aliphatic substrates.
Taking inspiration from recent progress in cross-electrophile coupling reactions,
i.e. the direct catalytic joining of two different electrophiles that avoids the need
for preformed carbon nucleophiles,[9 ] we developed a novel dual-catalytic acylation process using nonactivated alkyl bromides
as electrophilic partners for N -acylsuccinimides. The process was designed to follow a radical relay pathway where
a silyl radical is generated photocatalytically and serves as an abstracting agent
to activate the alkyl bromide through homolytic C–Br bond cleavage (Scheme [1 ]A).[10 ] Another exciting challenge that we subsequently considered was the design of an
alternative strategy that would rely on the selective C(sp3 )–H bond activation of simple alkanes through a radical hydrogen-atom transfer (HAT)
process.[11 ]
[12 ]
Scheme 1 Ni/Photoredox alkylation of amides with alkyl bromides or simple alkanes based on
a radical relay process
The groups of Doyle and Molander independently demonstrated that organohalides could
serve as both coupling partners and sources of halide radicals that act as potent
H-abstracting agents in Ni/photoredox cross-coupling reactions with aliphatic substrates.[13 ] Inspired by these seminal reports, we questioned whether this strategy might be
applicable to the coupling of amides with alkanes in the presence of an exogenous
source of halide anions, e.g. lithium chloride (Scheme [1 ]B).[14 ] We hypothesized that oxidative addition of the amide to nickel and subsequent succinimide-to-chloride
ligand exchange might readily generate an intermediate acyl–nickel chloride complex
that could serve as source of chlorine radicals through Ni–Cl bond homolysis (Scheme
[1 ]C).[12d ]
[15 ]
In our initial exploratory studies,[16 ] we used the coupling of N -benzoylsuccinimide (1a ) with cyclohexane as a model system (Table [1 ]). We found that the desired cross-coupling product 3a was formed in optimal yields by using [Ni(dtbbpy)(H2 O)4 ]Cl2 (I ; 4 mol%) (dtbbpy = 4,4′-di-tert -butyl-2,2′-bipyridine) and Ir{[dF(CF3 )ppy]2 (dtbbpy)}PF6 (II ; 0.5 mol%) [dF(CF3 )ppy = 2-(2,4-difluorophenyl)-5-(trifluoromethyl)pyridine] complexes as a dual-catalytic
system under irradiation by a 40 W blue LED lamp. Remarkably, solvent evaluation indicated
that only benzene was suitable for delivering the desired ketone. Another critical
feature for the success of the coupling reaction was the need for an external source
of halogen atom. Lithium chloride was found to be the most effective source among
the salts tested (LiCl, NaCl, KCl, TBAC, Et3 N+ Bn Cl– , NH4 Cl, and NaBr), and increasing the amount of this additive beyond one equivalent did
not significantly increase the yield. Additionally, the reaction proceeded more efficiently
by using a combination of tripotassium phosphate and sodium tungstate as a dual-base
system (Table [1 ], entries 1–5).[12d ] Control experiments revealed that the dual-catalytic system and light irradiation
were essential for this transformation (entries 6–8). N -Benzoylglutarimide also proved to be a reactive substrate but gave 3a in a lower yield (entry 9). Notably, by employing up to five equivalents of cyclohexane,
ketone 3a was isolated in a fair 53% yield after 24 hours of reaction. These conditions (entry
10) were therefore established as our standard conditions.[17 ] Note that the use of cyclohexane as the solvent led essentially to no reaction,
possibly due to the limited solubility of the catalyst and base in this reaction medium
(entry 11).
Table 1 Reaction of N -Benzoylsuccinimide (1a) with Cyclohexane: Selected Experimentsa
Entry
Reaction conditions
Yieldb (%) of 3a
1
as shown
50
2
no K3 PO4
27c
3
no Na2 WO4 ·2H2 O
40d
4
no K3 PO4 or Na2 WO4 ·2H2 O
18
5
no LiCl
3
6
no Ir photocatalyst
0
7
no light (darkness)
0
8
no Ni catalyst
0
9
N -benzoylglutarimide instead of 1a
42
10
cyclohexane (5 equiv), 24 h
61 (53)e
11
cyclohexane as solvent
1
a Reactions were performed in benzene on a 0.6 mmol scale with irradiation by a 40
W blue LED Kessil lamp. See the Supporting Information for details of the lighting
setup.
b Determined by GC-MS with benzophenone as internal standard.
c Na2 WO4 ·2H2 O (3 equiv).
d K3 PO4 (3 equiv).
e Isolated yield.
Having a reliable acylation protocol in hand (Conditions A), we explored the reactivity
of various acyl succinimides and potential HAT substrates (Scheme [2 ]). Various electron-rich and electron-neutral aroyl succinimides underwent coupling
with cyclohexane in fair yields (3a –c ), whereas electron-deficient derivatives were found to give significantly lower yields
(3d and 3e ). The scope was expanded to acyl succinimides bearing alkyl groups, as illustrated
by 3f , thereby providing access to dialkyl ketones. Notably, other nonactivated cycloalkanes
also participated in the coupling process (3g and 3h ). We also attempted the acylation of alkyl ethers to access the corresponding α-functionalized
ketones. For instance, we examined the use of dialkyl ethers as both alkylating reagents
and solvents, as safer alternatives to benzene. In our initial experimental observations
on the coupling of 1a with tetrahydrofuran,[16 ] we found that the new procedure (conditions B) required higher catalysts loadings
(10 mol% of catalyst I ; 2 mol% of catalyst II ), but proceeded efficiently without the need for sodium tungstate as an additive.[18 ] This is illustrated by the synthesis of a series of 2-acyltetrahydrofurans 3i –l in moderate to good isolated yields. The same procedure successfully gave coupling
products containing cyclic or linear aliphatic ethers such as 1,4-dioxane, diethyl
ether, or 1,2-dimethoxyethane (3m –q ). Notably, a separable 2.3:1 mixture of two regioisomeric products 3q and 3q′ was generated by the acylation of DME.
Scheme 2 Scope of N -acylsuccinimides and alkanes. Reagents and conditions
A : 0.6 mmol scale; Ni cat I (4 mol%), Ir cat II (0.5 mol%), K3 PO4 (2 equiv), Na2 WO4 ∙2H2 O (1 equiv), LiCl (1 equiv), alkane (5 equiv), benzene (0.1 M), 40 W blue LED Kessil
lamp (455 nm), 24–48 h. Reagents and conditions
B : 0.4 mmol scale Ni cat I (10 mol%), Ir cat II (2 mol%), K3 PO4 (1.5 equiv); LiCl (1 equiv), alkane as solvent (0.125 M), 30 W blue LED lamp (450
nm), 24–48 h. Isolated yields are reported.
Interestingly, the byproducts regularly obtained from the reaction of N -acylsuccinimides with alkanes were the symmetrical dialkyl ketones deriving from
homocoupling of the alkane, together with the corresponding C(Ar)–C(sp3 ) bond-formation products. A series of experiments were carried out to gain information
on the reaction pathway and the origin of these products (Scheme [3 ]). As a prototypical example, the coupling of N -benzoylsuccinimide (1a ) with cyclohexane afforded the desired cross-coupling product (3a ), along with small amount of cyclohexylbenzene (4a ; ≤10%) and dicyclohexylmethanone (3f ; ≤20%), suggesting the occurrence of a decarbonylative pathway.[5g ]
[9d ] Accordingly, an excess of 1a (10 equiv) was found to react spontaneously with (dtbbpy)Ni(COD) (1 equiv) to afford
a mixture of complexes assigned to the corresponding Ni(II)-aryl complex III and the Ni(II)-acyl complex (IV ) in a 3:2 ratio (1 H NMR), together with the dicarbonyl complex (dtbbpy)Ni(CO)2 (V ) (Scheme [3 ]A).[16 ] By adjusting the stoichiometry of (dtbbpy)Ni(COD) (1.5 equiv) and 1a (1 equiv), the Ni(II)-aryl complex III could be obtained selectively and isolated for full characterization, including single-crystal
X-ray diffraction analysis (Scheme [3 ]A).[19 ] The molecular structure of III displays a square-planar geometry of the nickel center, and represents a unique example
of structural characterization of a Ni(II)–NSucc complex formed by a C–N bond oxidative addition–decarbonylation sequence of amides
with Ni(0).[20 ] All attempts to generate the acyl complex IV selectively or to isolate it from the reaction mixture were unsuccessful. This result
indicates that N -acylsuccinimides undergo spontaneous oxidative addition to Ni(0) followed by facile
decarbonylation, providing a plausible pathway to the symmetrical dialkyl ketones[21 ] and the decarbonylative products[.22 ] Comparatively, the reaction of benzoyl chloride with (dtbbpy)Ni(COD) led exclusively
to the corresponding acyl-Ni(II) chloride complex VI .[16 ]
[12b ]
Scheme 3 (A) Reaction of 1a with (dtbbpy)Ni(COD) and (B) the catalytic behavior of several nickel species
We then compared the catalytic activity of complex I with that of other potent nickel catalytic intermediates (Scheme [3 ]B). Surprisingly, the use of Ni(COD)2 with dtbbpy as an exogenous ligand instead of complex I resulted in a very low 8% reaction yield. Furthermore, no reaction was observed with
the Ni(II)-aryl intermediate III . On the other hand, the nickel(II)-acyl chloride complex VI was found to be catalytically competent, affording very similar results to those
with complex I . Notably, only trace amounts of the desired product were observed in the absence
of LiCl, confirming its key role in the reaction, probably through succinimide-to-chloride
exchange at nickel. At this stage, the involvement in the catalytic cycle of active
paramagnetic species generated in situ from nickel(II) complexes was envisioned. Indeed,
as recently demonstrated by the group of Hazari,[23 ] paramagnetic Ni(I) species with bipyridine ligands can participate as key intermediates
in nickel-catalyzed C(sp2 )–C(sp3 ) bond-forming reactions. We decided to examine this possibility, and we prepared
the [(dtbbpy)Ni(Cl)]2 dimer VII by the reported procedure.[16 ] Remarkably, the paramagnetic nickel species VII was found to be catalytically active and performed more efficiently than the nickel(II)-dichloro
complex I .
These preliminary experiments suggest that the nickel-catalyzed acylation of C(sp3 )–H bonds with N -acylsuccinimides does not follow the same pathway as that proposed for other acyl
substrates. In related coupling processes involving chloroformates and acyl chlorides
as substrates reported by the groups of Doyle[12d ] and Shibasaki,[12b ] the oxidative addition of the acyl substrate proceeds at Ni(0) species. With N -acylsuccinimides, although oxidative addition of the C–N bond to Ni(0) readily occurs,
our experiments suggest that the C–N bond-activation step proceeds at Ni(I) during
the catalysis. Further mechanistic studies are clearly necessary to confirm the specific
behavior of these acyl substrates, which might be exploited to design complementary
mechanistic sequences.[24 ]
In summary, we have described a new acylation reaction of C(sp3 )–H bonds by using bench-stable N -acyl imide substrates. The dual-catalytic process combines nickel-catalyzed C–N bond
activation with photocatalytic HAT, and provides operationally simple access to valuable
alkyl ketones. Notably, N -acylsuccinimides were shown to undergo ready C–N bond oxidative addition to Ni(0),
followed by decarbonylation, under mild conditions. Identification of byproducts of
the acylation reaction suggested that the development of decarbonylative cross-coupling
pathways under mild conditions by using photoredox/Ni catalysis can be envisioned.
Further investigations are underway to achieve a better understanding of the reaction
mechanism.
