Organic Letters
Letter
a,b
occur based on their triplet energies. DFT calculations
(uB3LYP/6-31G**:LANL2DZ) provided the triplet energy
(ET) value for 1 to be 2.02 eV, which was smaller than the ET
of [Ir{dF(CF3)ppy}2(dtbbpy)]PF6 (2.76 eV; calculated using
TD-uB3LYP/6-31G**:LANL2DZ) (see the SI for the details).
DFT calculations at the uB3LYP/6-31G** level were
performed to further support the proposed pathway (Scheme
2b and see the SI for the details). The dissociation energy of
the N−O bond in the excited state 1 (1*) was calculated to be
1.97 eV, which is smaller than the electronic energies of the
triplet excited states of 1 (ET, 2.02 eV), supporting the
favorable N−O bond cleavage. The ensuing decarboxylation of
the acyloxy radical is exothermic, with a low activation barrier,
and forms the phenyl radical with the release of CO2. A
concerted fragmentation of 1* through the N−O bond
homolysis, followed by the decarboxylation, can also be
anticipated.9a However, we propose a stepwise process based
on the detection of benzoic acid, albeit in a trace amount, as
Scheme 4. Reaction Scope
•
the side product. On the other hand, Ph participates in a
thermodynamically favorable hydrogen atom abstraction from
tert-butyl methyl ether A (ΔG = 18.09 kcal/mol) to produce
the corresponding •C(sp3) radical. While homocoupling of the
persistent iminyl radical is possible,14 cross-coupling between
the C(sp3) and iminyl radicals is thermodynamically favored,
which selectively furnishes 1A.
Having obtained the initial results, we optimized the
reaction conditions further using 1 and A as the model
substrates (Table 1). As expected, the reactions in the presence
of [Ru(bpy)3]Cl2 (ET, 2.01 eV)9b and rose bengal,15 which
have lower triplet energies than that of 1, did not proceed at all
(entries 1 and 2). Despite their ability for energy transfer to 1,
none of the tested photocatalysts showed better results than
those with [Ir{dF(CF3)ppy}2(dtbbpy)]PF6 (entries 3−5).
Control experiments revealed that the reaction fails in the
absence of visible light or the photocatalyst (entries 6 and 7).
With a higher photon power (365 nm high-pressure mercury
lamp, 300 W), and in the absence of the photocatalyst oxime 1
was completely consumed. However, the desired 1A was not
formed, supporting that the visible-light-mediated process can
often provide higher efficiency and selectivity than a UV-driven
process under mild reaction conditions (entry 8). The yield
improved upon reducing the concentration to 0.02 M, with the
use of A as the solvent (entries 5 and 9−12). The use of white
LEDs as the photon source resulted in lower reactivity (entry
13), and the presence of molecular oxygen completely
prevented the reaction (entry 14). Interestingly, upon
decreasing the catalyst loading to 0.25 mol %, the yield rather
increased to 83%, and a lesser amount of the iminyl
homocoupling side product was formed (entries 11, 15, and
16). A photochemical quantum yield determined using the
standard ferrioxalate actinometry was 63% for the C−N bond
formation between 1 and A under the optimized conditions
(see the SI for the details).
a
Reaction scale: 2-10 (0.4 mmol), A−H (8−20 mL), under argon
b
1
atmosphere. Isolated yields or H NMR yields using bromoform as
the internal standard.
Next, we investigated the generality of the process with
other alkyl C(sp3)−H substrates (Scheme 3). Common
organic solvents are good candidates as coupling partners
due to their low cost, easy removal, and low toxicity.
Acetonitrile (B), tetrahydrofuran (C), diethyl ether (D), and
1,4-dioxane (E) were evaluated in the transformation and were
found to show excellent regioselectivity. The highly activated
hemiaminal ether compounds could be formed and isolated
under the conditions with C and D. In addition, cyclohexanone
suggested that either reductive or oxidative electron-transfer
between the photoexcited [Ir{dF(CF3)ppy}2(dtbbpy)]PF6 and
1 would be unfavorable due to the negative thermodynamic
driving force, excluding any of the photoredox processes as an
origin for the N−O bond cleavage. On the other hand, energy
transfer from the triplet excited state of [Ir{dF(CF3)-
ppy}2(dtbbpy)]PF6 to the triplet state of 1 is predicted to
C
Org. Lett. XXXX, XXX, XXX−XXX