ACS Catalysis
Research Article
III
Scheme 1. (a) Concept of “Umpolung” Reactivity of π-Allylmetal Complexes and (b) Conversion of Electrophilic π-AllylCo to
II
Nucleophilic π-AllylCo via SET
mixture was irradiated with a 450 nm LED lamp for 32 h and
afforded the product 2a in 58% yield (entry 1). Various solvents
such as acetonitrile, tetrahydrofuran, and dichloroethane were
tested, generally producing low yields in the 10−20% range. The
use of Hantzsch ester as an organic electron donor instead of
DIPEA resulted in no product (entry 2). Further screening of
such as indomethacin, probenecid, and lithocholic acid
derivatives, and the desired products 3a−3c were obtained in
a decent yield.
The regio- and diastereo-selectivity of the new catalysis
system was further explored using cinnamyl alcohol as the
coupling substrate. It was found that tert-butyl cinnamyl
carbonate 4 is more reactive than cinnamyl acetate for the
allylation of various aldehydes, as shown in Table 3. Only
branched products 5a−5l were obtained in favor of trans-
diastereoselectivity with moderate to good yields. The highest
diastereoselectivity (d.r. > 20:1) was achieved with cyclo-
hexanecarboxaldehyde (5h), albeit in somewhat low yield. The
trans-diastereoselectivity can be elucidated based on the
Zimmerman−Traxler transition state, where the oxygen atom
of the carbonyls coordinates with the cobalt. The aryl group of
the π-allylcobalt complexes and the R group of the aldehyde
preferentially adopt an equatorial orientation, resulting in the
generation of an anti-isomer of the corresponding homoallylic
various photocatalysts, including 4CzIPN, [Ir(dF(CF )-
3
ppy) (dtbbpy)]PF6 (Ir-2), and Ru(bpy) (PF ) (Ru-1),
2
3
6 2
revealed that they are less efficient for allylation (entries 3−5).
Different ligands such as 4,7-diphenyl-1,1-phenanthroline L1,
1
,10-phenanthroline L2, and 1,2-bis(diphenylphosphino)-
ethane L4 (entries 6−8), were tested as replacements of L3.
These ligands were less efficient for catalysis. Further testing of
various cobalt salts showed that CoSO ·H O gave the best
results and increased the yield up to 80% (entries 9−13).
Additive effects were tested but did not increase the yield
4
2
(
entries 14 and 15). Additionally, the reaction did not occur in
the absence of either light, Co, DIPEA, or photocatalyst (entries
6−19). The reaction is sensitive to low light intensity, H O,
14
1
alcohols.
2
The scalability of the reaction was determined by the
synthesis of 5a in a gram-scale under standard conditions.
Gratifyingly, product 5a was obtained in decent 83% isolated
yield. The use of earth-abundant cobalt, wide compatibility with
various sensitive functional groups, and easy to operate
conditions of this reaction are appealing features for laboratory
and industrial applications.
45
After optimizing the reaction conditions, we evaluated the
scope of aldehydes for the allylation reactions (Table 2). It was
clear that the reactions worked smoothly with lots of aromatic
and aliphatic aldehydes producing various trans-alcohols in
generally decent yields. The substituent effect of the aromatic
ring has little effect on the allylation reaction. Both the electron-
rich groups and electron-withdrawing groups, including
methoxy 2a and fluorine 2l, gave homoallylic alcohols in
reasonable yields. The reaction of sterically hindered mesitalde-
hyde 2e was also successful. Heteroatoms did not impede the
catalytic cycle, including oxygens 2g and 2u and the protected
amine 2x. Besides, various heteroarenes, which are widely used
as core structures in medicinal synthesis, were also suitable
substrates for this photoredox and cobalt catalysis, including
furans 2q and 2r and thiophene 2p. Due to the extremely mild
photocatalytic conditions, reactions showed excellent compat-
ibility with many synthetically important and valuable functional
groups that include the free alcohol 2j, alkene 2w, ether 2h,
bromobenzene 2k, and chlorobenzene 2o. Allylation of the acid-
sensitive citronellal 2w substrate was also successful without
forming a cyclization byproduct. The synthetic potential of this
dual photoredox and cobalt catalysis was demonstrated by the
late-stage functionalization of complicated structure molecules,
can readily quench the excited photocatalyst Ir-1 at the rate of
8
−1 −1
3.42 × 10 L M s . Then, a series of experiments were
performed to probe the different π-allylcobalt species in the
catalytic cycle. First, CoBr , L3, and Ir-1 were added to a
2
solution of DIPEA in DMF (Figure 1C(a)). The resulting
mixture was exposed to 450 nm LED irradiation. In the first SET
process, the color of the reaction changed from green to blue
(Figure 1C(b)). The ultraviolet−visible (UV−vis) spectro-
scopic analysis of the reaction was also performed. A broad
absorption band at 550−700 nm gradually appeared within 30
I
min, indicating the formation of Co 7 species, which is
27,46,47
consistent with previous reports (Figure 1A(b)).
Then,
the allyl acetate was added to the resulting reaction system in the
dark, and the color changed to brown (Figure 1C(c)). At the
same time, UV−vis spectroscopic analysis of the reaction
I
revealed the rapid disappearance of absorption of Co 7, and a
2
993
ACS Catal. 2021, 11, 2992−2998