Communication
Table 3. Verification of the formation of intermediate nitrogen-centered
radicals by trapping experiments.[a]
entry
catalyst
yield [%][b]
1
2
3
4
5
RhO (2 mol%)/1b (20 mol%)
RhO (2 mol%)
no transition metal catalyst
IrO (2 mol%)/1b (20 mol%)
IrS (2 mol%)/1b (20 mol%)
61
6
4
77
78
[a] Reaction conditions: N-methylindole (0.20 mmol), 2a (0. 10 mmol), 2,6-
lutidine (0.17 mmol), together with the indicated amounts of 1b and cat-
alyst in MeCN/DMSO (3:1, 0.5 mL) irradiated with blue LEDs (24 W) for 2 h
under nitrogen. [b] Yields are determined with 1H NMR using an internal
standard.
Figure 2. Proposed mechanism for the visible light-activated enantioselec-
tive rhodium catalysis.
toredox-induced formation of the nitrogen-centered radical
intermediate.
cycle to regenerate the oxidized rhodium enolate (II++eÀ!II)
or directly transfers a single electron to the ODN-carbamate,
thereby propagating a chain process.
The superiority of RhO over its iridium congeners in this re-
action can rather be pinpointed to kinetic effects, namely the
requirement of a high turnover frequency of the catalytic
cycle. Amidyl radicals are known to be highly reactive electro-
philic p-type radicals[17] that add to electron-rich alkenes much
faster than carbon-centered radicals.[18,19] Such electrophilic ni-
trogen-centered radicals are also highly prone to reduction.[19]
A concomitant short lifetime and fast reaction of the nitrogen-
centered radical intermediate with the rhodium enolate II re-
quires a fast turnover frequency of the catalytic cycle to regen-
erate new II rapidly enough to be a reaction partner for the
aminyl radical. Indeed, comparison of initial rates for an ex-
change of the acetonitrile ligands against substrate in RhO
versus IrO demonstrates an increased rate constant for ligand
exchange of the rhodium complex by >1650-fold (Figure 3).
On the other hand, the inferior photophysical properties of
bis-cyclometalated rhodium over iridium complexes,[20] result-
ing in a less efficient rhodium photoredox-sensitizer, is not rel-
evant for this reaction owing to a highly efficient chain propa-
gation, as demonstrated by the observed high quantum yield.
Thus, the rhodium enolate intermediate II serves as a smart ini-
tiator,[11] which is needed to initiate and, from time to time, re-
initiate the chain reaction after chain termination.
A number of experiments strongly support this mechanism.
We obtained a crystal structure of the proposed enolate inter-
mediate II and confirmed its catalytic competence (see Sup-
porting Information). A reduced yield in the presence of
oxygen (Table 1, entry 5) as well as a suppression of product
formation (below 10%) in the presence of TEMPO (5 equiva-
lents) support a radical mechanism (see Supporting Informa-
tion). Furthermore, a calculated quantum yield of 14 for this
photochemical reaction lends support to a chain mechanism
in which the electron transfer from the ketyl intermediate III to
the ODN-carbamate dominates. Considering nonproductive
quenching and energy decay processes of the photoexcited
rhodium-enolate complex II, chain cycles of 100 or even higher
can be assumed for this reaction.[15]
Trapping reactions with N-methylindole as shown in Table 3
provide further important insights into the mechanism: 1) The
observed amination of the 2-position (leading to 4) is indica-
tive of a radical reaction through intermediate nitrogen-cen-
tered radicals; 2) RhO is an effective catalyst only in the pres-
ence of catalytic amounts of 2-acyl imidazole (Table 3, entries 1
and 2), which is consistent with our proposal that the inter-
mediate rhodium enolate complex II is responsible for the
light-activated reductive formation of the nitrogen radical;[16]
3) in the complete absence of any transition-metal catalyst,
only small amounts of product are formed, thus revealing that
MacMillan’s proposed mechanism[5] of a direct activation of the
ODN-carbamate is not operative in our system; 4) for IrO and
IrS in the presence of some 2-acyl imidazole, the indole amina-
tion product was obtained in high yields upon visible-light ac-
tivation (Table 3, entries 4 and 5). This means that the incom-
petence of the established dual-function chiral Lewis acid/pho-
toredox catalysts IrO[9] and IrS[8,10] to catalyze the enantioselec-
tive CÀN bond formation of 1+2!3 is not related to the pho-
In conclusion, we have introduced a very efficient photoacti-
vated enantioselective radical amination of 2-acyl imidazoles
catalyzed by a chiral-at-metal rhodium complex, which serves
a dual function, namely as a chiral Lewis acid to catalyze asym-
metric enolate chemistry and, furthermore, as a light-activated
smart initiator of a radical chain process. Intriguingly, under re-
lated conditions, previously developed iridium complexes fail
to work in this context. This is attributed to much faster ligand
exchange kinetics in the rhodium system, which are required
to match the high reactivity and short lifetime of the inter-
mediate nitrogen-centered radicals. The inferior photoredox
properties of the rhodium system do not play a role here due
Chem. Eur. J. 2016, 22, 9102 – 9105
9104
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim