Organic Letters
Letter
d
than inexpensive collidine ($29/mol), we elected to continue
optimization using collidinium salt 1h.
Scheme 3. Scope Studies
With reductive conditions, which included catalytic Ir(ppy)3,
DIPEA, and blue light, we observed complete conversion within
6 h, but the desired product was a minor product (23%) (Table
1, entry 1). While minor amounts of radical termination
products were identified (3′ and 3″), we were encouraged to see
that the majority of the mass balance appeared to be derived
from a benzyl radical that had formed the desired C−C bond
and could, if nudged in the right mechanistic direction, lead to
product. More specifically, it appeared that rather than
terminating to give the desired product, it underwent one or
two propagation steps to give products 3a′ and 3a″. Dilution of
the reaction mixture (entry 2) helped somewhat, giving a
correspondingly higher yield, but slowed the reaction. Together
these experiments suggested that controlling the rate of
termination would be vital to achieving product selectivity.
We postulated that identification of the appropriate catalyst
could facilitate reduction of the intermediate radical.58,59
Indeed, a photocatalyst screen (see the SI) showed that while
the iridium catalyst Ir[dF(CF3)ppy]2(dtbbpy)PF6 gave more
sluggish conversion (entry 3), the critical ratio of desired to
undesired products improved by an order of magnitude.
Furthermore, increasing or decreasing the photocatalyst loading
increased (entry 5) or decreased the product ratio (entry 6),
respectively.
Changing the catalyst to Ru(bpy)3 (Table 1, entry 4), which
has a similar reduction potential (E1/2(II/I) = −1.33 V vs SCE),1
gave very sluggish conversion and no detectable product
formation, suggesting that the photocatalyst plays a nuanced
role in the reaction. Attempts to use NBu3 (entry 8) instead of
DIPEA (entry 3) led to slightly faster conversion but gave
substantial amounts of a compound derived from combination
of the amine and nitrile.60,61 Speculating that the off-cycle use of
the amine retarded the reaction at higher conversions, we
investigated the use of more amine (entry 3 vs entries 9 and 10).
Indeed, moving from 2 to 4 equiv increased the conversion from
50% to 100% and decreased the reaction time from 46 to 16 h.
Importantly, as the desired reaction was able to take place
throughout the entirety of the reaction period, the product
distribution shifted in favor of the desired product. With
evidence suggesting the involvement of the photocatalyst in the
termination step, we investigated the effect of water on the
reaction (entries 11 and 12). Indeed, the inclusion of 10 equiv of
H2O further enhanced the product distribution to 29.3:1 and
accelerated the reaction (12 h), resulting in an 88% yield.
Finally, individual control studies evidenced the critical aspect of
each reaction component (entry 13).
a
b
The 4-methylpyridinium salt was used. The catalyst loading was 0.5
c
d
mol %. 19F NMR yield. Yields are of isolated products, unless
otherwise noted.
secondary benzylic substrate (3l) also gave a good yield,
highlighting the ability to rapidly and significantly modify the
carbon framework of the substrate. The mild reaction conditions
are compatible with a wide range of functional groups, such as
nitrile (3f), ester (3d), ether (3j), and bromide (3b and 3c).
Importantly, all of these substrates were engaged photocatalyti-
cally using the same conditionsa feat that would have been
challenging using the corresponding halides. The collidinium
salts offer protection to otherwise-sensitive heterocycles such as
thiophene62 (3m) and naphthalene63 (3h), which might be
expected to undergo radical addition. We expect the broad
functional group tolerance to facilitate further synthetic
elaboration. Other electron-deficient alkenes worked well in
the reaction (3n−s), with the ester substituent of acrylates
exhibiting minimal influence (3n and 3o) while methacrylate
(3p) was slightly more prone to propagation. Similarly,
cinnamate (3s) gave the product in modest yield. Cyclic enones
proved to be competent (3q, and 3r), giving the fluorobenzy-
lated products in good yields. Other alkenes also proved to be
competent (see the SI). Interestingly, the use of α-methylstyrene
resulted in the formation of product (3t) and higher-order
oligiomers. The scope suggests that different reaction
mechanisms may be operative depending on the alkene. The
use of the bench-stable crystalline collidinium salts also
facilitates workup of the reaction. Simple extraction followed
by acidic washes removes any excess DIPEA, collidine
byproduct, and (if present) any unreacted collidinium salts.
This is in stark contrast to the Katritzky salt, which produces
triphenylpyridine, which must be removed chromatographically.
Likewise, if the benzyl halide were used, any excess would also be
expected to require removal from the organic extracts.
Having identified the optimal conditions (Table 1, entry 12),
we examined the scope of collidinium salts with acrylonitrile
(Scheme 3). A broader range of collidinium salts was prepared
(see the SI). The reaction worked well for benzylic collidinium
salts with electron-withdrawing groups (3a, 3d, and 3f),
electron-neutral groups (3b, 3c, and 3g), and electron-donating
groups (3i and 3j), which would have been a challenging feat for
the corresponding halides. This strategy could be extended to
sterically demanding, ortho-flanked benzylic substrates (3e and
3k) by use of the 4-methylpyridine-derived salts. Apparently, the
bulk of the benzyl component, which made nucleophilic
substitution more challenging, also served to protect these
salts from undergoing the Minisci-type benzylation that we had
observed earlier with less sterically demanding benzylpyridi-
nium salts. Furthermore, the 4-methylpyridinium salt of a
Our working mechanism of the reaction is shown in Scheme
4. The reaction begins with absorption of a blue photon to give
2038
Org. Lett. 2021, 23, 2036−2041