Angewandte
Communications
Chemie
the rate of quenching of the *[Ir] photocatalyst. At this stage,
the rate of deprotonation which cooperates to minimise
we propose this effect might be perceived as an attenuated
version of what usually observed in proton-coupled electron
transfers (PCETs) between amides and phosphate bases.[11]
We were therefore intrigued by the possibility that this pre-
association upon facilitating SEToxidation could also be used
to synergistically aid a following deprotonation that, circum-
venting BET, would provide access to the a-anilinoalkyl
radical D.
BET.[8] This effect is also operating in the demethylations
leading to 23–25 which were quantitative. The chemistry was
amenable to a less activated dichloro aniline (26) and found
compatible to other aromatics like N-methylnapthalen-1-
amine (27), several aminopyridines (28–30) and 5-amino-
benzothiophene (31). In terms of limitations, we did not
succeed in engaging heteroaromatic anilines located on
electron rich 5-membered ring systems (e.g. 32).
In approaching this blueprint for a-anilinoalkyl radical
generation, we envisaged its utilization to target the still
elusive demethylation of N-methyl anilines (Scheme 2B).
Specifically, we proposed the development of a dual photo-
redox-cobalt[12] catalytic system where a visible light-excited
[IrIII] photocatalyst would, helped by the amine, generate D
by oxidation and deprotonation of 2 while minimizing BET.[13]
As D is a nucleophilic radical (local electrophilicity index,[14]
w+rc = 0.36), we speculated it could react with a [CoII] co-
catalyst and undergo dehydrogenation to give a [CoIII]-H
species and imine 4. A following hydrolysis would provide the
desired demethylated product 3. The pathway for dehydro-
genation can be rationalised considering the following two
mechanistic options. (1) Radical capture of D by [CoII] could
generate an alkyl-[CoIII] species followed by b-hydride-type
Having developed the first catalytic approach for the
demethylation of secondary anilines, we were keen to under-
stand if this dual photoredox-cobalt approach could be used
in a general sense as a N-dealkylation platform. In this case, to
evaluate the impact of the alkyl chain we initially looked at
a series of N-Ph derivatives. Pleasingly, the process displayed
broad scope enabling the removal of a benzyl group (33) as
well as challenging primary (34 and 35) and secondary (36 and
37) alkyl chains. As an a-anilinoalkyl radical cannot be
formed in the case of a t-Bu substituent, 38 resulted in
complete starting material recovery which agrees with our
mechanistic analysis. The chemistry was also expanded to the
removal of functionalised alkyl chains containing inductively
electron withdrawing substituents (39 and 40) and could also
be used to fully dealkylate an ethane-1,2-diamine derivative
(41). In line with our mechanistic analysis, dealkylation of N-
Me-N-Bn-amine 42 was not possible and this substrate was
recovered at the end of the reaction. In this case, Stern–
Volmer and cyclic voltammetry analysis revealed no inter-
action with Et3N or piperidine, which does not enable to
overcome BET. The approach was also demonstrated on
a series of more complex derivatives (43–45) which included
the blockbuster drug primaquine. We then benchmarked the
strategy on the demethylation of 1. While all previous
approaches stop after the removal of the first methyl
group,[3,4] ours enables, for the first time, complete deal-
kylation to 3 in good yield by simply extending the reaction
time. Finally, we were able to demonstrate the orthogonality
that this strategy might offer with respect to standard
hydrogenation protocols. Taking N-Me-N-Bn-aniline 46, all
reported deprotections enable debenzylation (46 ! 2) while
this approach provides a switch in selectivity and targets the
removal of the Me-group (46 ! 33).[22]
elimination[15] across the N H bond. (2) Since D is also an
ꢀ
electron rich radical (IP = 6.1 V), an alternative process
would involve direct SET between D and [CoII] to give, via
a [CoI],[16] 4 and [CoIII]-H. While it is difficult at this stage to
distinguish between the two mechanistic options, both of
them ultimately result in the formation of the [CoIII]-H. This
species would then evolve H2 by reaction with a protic
source,[17] with the resulting electron poor [CoIII] closing both
photoredox and cobalt cycles by SET with [IrII].[18]
To put this proposal into practice, we first evaluated the
demethylation of 2 using both Et3N and piperidine (Sche-
me 2C). While in their absence no reactivity could be
observed,[10] a survey of reaction conditions revealed an
optimal process that involved the use of the Ir(dtbbpy)-
[19]
(ppy)2PF6
photocatalyst and the cobaloxime co-catalyst
Co(dmgH)(dmgH2)Cl2 in CH3CN solvent under blue light
irradiation.[20] Under these mild conditions 3 was formed in
high yields. Crude reaction analysis enabled the detection of
formaldehyde, indicative of the generation of 4, as well as H2,
which supports the interplay of a [CoII/III] cycle.[10]
The chemistry described here demonstrates that a-anili-
noalkyl radicals can be conveniently accessed under mild
conditions from the corresponding secondary anilines. Since
previous exploitation of these open shell intermediates in
synthetic chemistry required the preparation and use of N-
The scope of the demethylation process was initially
evaluated by using mono-substituted N-methyl anilines. para-
Substituted aryl groups displaying electron-rich (5, 6), elec-
tron-deficient (7, 8) and synthetically useful functionalities
(9–11) provided the desired products in good to high yields.
The ability to engage substrates containing electron poor
arenes is remarkable since BET from their corresponding
aminium radicals is accelerated.[8,21] meta-Substitution was
evaluated next and electronically different functionalities
were tolerated albeit in lower yield in the case of electron
poor systems (12–15). We then trialled ortho-substituted
aromatics and also in this case very high yields were obtained
across a broad range of derivatives (16–22). In this case, the
steric hinderance provided by the substituent decreases the
stabilisation of the aminium radical and therefore enhances
aryl glycines or a-silyl derivatives (see also Scheme 1B),[23]
3
ꢀ
this approach has the potential to facilitate divergent sp C H
functionalization.
A relevant application would involve application in
radical-polar crossover[24] reactions involving acrylates for
the construction of poly-substituted pyrrolidones (Sche-
me 3A).[25] This would exploit the intrinsic vicinal radical-
ionic di-nucleophilic nature of D and constitute a novel [3+2]-
like retrosynthetic disconnection alternative to more classical
approaches based on azomethine ylide chemistry.[26] Pleas-
ingly, the realization of this cascade process was achieved by
employing conditions similar to the ones presented above but,
Angew. Chem. Int. Ed. 2021, 60, 7669 –7674
ꢀ 2021 The Authors. Angewandte Chemie International Edition published by Wiley-VCH GmbH www.angewandte.org
7671