Organic Letters
Letter
conversion within 2 h of irradiation, yielding very complex
product mixtures. The recombination photoproduct 5 was
isolated in a yield of 16% and dr of 1.8:1.12 New vinyl group
resonances between 6.5 and 6.0 ppm in the 1H NMR spectrum
of the crude material suggested that some of the side products
arise from disproportionation reactions. Both the low chemical
yield and the poor stereospecificity of 5 in solution are
consistent with a mechanism that involves free radical
products.
After optimizing the solid-state photodecarbonylation
reaction, we set out to explore the synthetic versatility of
photoproduct 5. Scheme 3 shows optimized reaction
Scheme 3
Solid-state photochemical experiments were carried out on
bulk powders prepared by grinding small crystals between two
microscope slides prior to irradiation (Table 1, entries 2−5).
Conversion under these conditions was less efficient as
compared to the solution phase. This is consistent with
previous reports from our group showing that accumulated
product can act as a filter, preventing additional reaction
progress along the depth of the crystals and sometimes leading
to the formation of secondary photoproducts. However, these
reactions were highly selective, yielding the desired compound
5 with consistently higher retention of configuration as
indicated by the dr = 8.0−8.6:1.13 While bulk-solid irradiation
of 4 improved the reaction yield and stereospecificity,
diminished reaction velocities with increasingly larger sample
loading (>10 mg per slide) limited material throughput.14
Therefore, we decided to explore suspending the crystalline
solids in water via a modified reprecipitation method as a
means to scale up the photochemical reactions. We have
shown that crystalline suspensions react in a more homoge-
neous manner, increasing the reaction velocity and con-
version.15−19 Thus, aqueous suspensions of 4 were prepared by
dissolving the ketone in a minimal amount of acetone, which
was added dropwise to rapidly stirring water with submicellar
surfactant concentrations. The reprecipitated solids were
collected via filtration and confirmed to be crystalline and of
the same polymorphic phase as the bulk powders (Figure S2).
Treatment of the reaction vessel glass surface with a silanizing
agent (Sigmacote) rendered the surface hydrophobic and
helped avoid material build-up on the reactor walls. It was also
found that sonication of the suspensions for 5 min after every
hour of irradiation limited the adherence of reaction material
to the glass surface.
Similar to the bulk powder reactions, irradiation of aqueous
crystalline suspensions of 4 (Table 1, entries 6−10) exhibited
higher chemoselectivity, with 5 being the only product isolated
in 30−65% yields based on recovered starting material (brsm).
The diastereoselectivity observed in these cases was similar to
that seen for reactions carried out on small amounts of bulk
powder (dr = 8.0−8.7:1). We found that stopping the reaction
after 4 h (Table 1, entry 7) led to the highest yields of 5
(brsm), and the recovered ketone could be readily crystallized
and subjected to irradiation.
Notably, greater conversion of 4 as suspended crystals
resulted in yields of 5 that were unexpectedly low (Table 1,
entries 8−10).20 We hypothesized that the observed loss of
chemical yield at high conversion was indicative of secondary
photochemical decomposition of 5. To test this hypothesis, we
conducted a control experiment (not shown) in which pure 5
was irradiated. As expected, an unidentifiable yellow material
was observed after 2 h, suggesting 5 may decompose during
irradiation. Despite the limitations of this particular substrate,
photodecarbonylation reactions in crystalline suspensions
could be carried out on scales as large as 300 mg.
conditions first carried out on model compounds and
subsequently applied to 5. Treating 5 with LiBH4 at 55 °C
proved to be selective for methyl benzoate reduction,
providing alcohol 11 in 52% yield (brsm). The use of other
reducing agents such as LiAlH4 or DIBAL-H resulted in
complex product mixtures and significant decomposition.
Epoxidation with m-CPBA gave 12 as a mixture of
diastereomers with a dr ≈ 1.7:1 after 18 h at 35 °C.
Hydrogenation of the terminal alkene using palladium on
activated carbon gave 13 in 98% yield, and saponification
afforded the corresponding acid 14, though cold temperature
was necessary to avoid decomposition. Single-crystal X-ray
diffraction unambiguously determined the structure and
relative stereochemistry of 13 (Figure S4). Interestingly,
when 5 was treated with nitrile-selective reducing conditions,
neither 15 nor any other identifiable reduced products were
observed, and only starting material was recovered from the
reaction mixture. This reactivity contrasts with the efficient
reduction of model nitrile 10 to furnish the primary amine 16
(Scheme 3b).21 While these transformations demonstrate
some of the versatility of 5, we note that some of the
challenges faced (i.e., the failed conversion to 15) were likely
due to the dense steric encumberment of the VSAQCs motif.
In conclusion, we illustrate the relatively simple preparation
of compounds containing vicinal, all-carbon substituted
quaternary stereocenters by taking advantage of the solid-
C
Org. Lett. XXXX, XXX, XXX−XXX