formedE-enamine, equilibrating at about 50%
conversion (Fig. 3D, a), whereas (R)-2apref-
erentially gaveZ-enamine, which was equili-
brated at rather lower conversion (~5%) (Fig.
3D, b), after which overriding formation of
E-enamine was observed. We also examined
enamine formation withrac-2a, showing a
similar kinetic profile to that of (S)-2a, with
E-enamine as the major isomer (E/Z= 30:1,
equilibrated at ~35% conversion) (Fig. 3D,
c). These results suggested that (S)-2aand
(S)-1aare stereochemically matched with each
other, and their coupling led to the formation
ofE-enamine, which is favored both thermo-
dynamically and kinetically.
Moreover, enamine isomerization was inves-
tigated with an in situ irradiative NMR ex-
periment. A 405-nm laser beam was directed
into a preequilibrated reaction mixture inside
the NMR spectrometer by means of optical
fiber.E-enamine was found to isomerize into
Z-enamine rapidly, and a photostationary
state was reached within 3 min with aE/Z-
enamine ratio maintained at 1.7:1 or 4:1, de-
pending on the presence of acid additive
(Fig. 3B). A quenching study verified that
the excited photocatalyst could be selectively
quenched by enamine (figs. S25 to S27). We
also tracked the reaction progress for dera-
cemization of2a, and a photodynamic equi-
librium could be established in 1 hour with
94% ee (Fig. 3C). Similar behaviors were ob-
served with other substrates such as2f,2h,2w,
2y, and2aa(figs. S6 to S10). The enantioselec-
tivity did not change upon extending irradiation,
but turning off the light source led to immediate
racemization (figs. S28 to S30). This observation
indicated that the deracemization was driven
by photochemical enamine isomerization.
We further rationalized the mechanism with
DFT calculations. On the basis of previous
work, a weak acid-bridged proton transfer
pathway was proposed to account for the
iminium-enamine tautomerization ( 24 ). We
calculated all four possible transition states
(TS 1to 4 ) that lead toE-andZ-enamine
(Fig. 4A). The calculations predicted that
(S)-2awould formE-enamine selectively (TS1
versusTS2) withDDG 21 = 4.6 kcal/mol, whereas
(R)-2ashowed moderate preference forZ-
enamine(TS3versusTS4) withDDG 43 =
1.2 kcal/mol. These results are fully consistent
with the experimentally observed kinetic profile
of enamine formation (Fig. 3D). According to
the principle of microscopic reversibility ( 8 ),
the backward enamine protonation may pro-ceed by means ofTS1andTS4forE-enamine
andTS2andTS3forZ-enamine: Both processes
showthesameselectivitywithDDG 41 and
DDG 32 , both equal to 2.9 kcal/mol. Hence, the
protonation ofE- orZ-enamine should proceed
with high facial selectivity to produce (S)-2a
or (R)-2a, respectively. The vertical excita-
tion energies (S 0 →T 1 ) forE/Zenamine were
calculated to be 71.8 and 74.6 kcal/mol, favor-
ing excitation ofE-enamine (Fig. 4B). The pref-
erential excitation ofE-enamine overZ-enamine
could be explained by the deconjugation of the
b-enaminyl phenyl group in theZ-geometry
because of steric hindrance. Recently, a similar
deconjugation effect has also been observed
in visible light–promoted isomerization of
E-alkenes to their thermodynamically dis-
favoredZ-isomers ( 25 – 27 ).
On the basis of the mechanistic investiga-
tions above, a plausible mechanism for optical
enrichment was proposed (Fig. 4C). Under the
ground state, the stereochemically matched
enantiomer forms a dominantE-configured
enamine, which is continuously isomerized
to its disfavoredZ-isomer through photocatalytic
energy transfer. Facially selective protonation
of theZ-enamine then delivers the mismatched
enantiomer. Hence, the consumption of the872 25 FEBRUARY 2022•VOL 375 ISSUE 6583 science.orgSCIENCE
Fig. 3. Mechanistic experiments.(A) Investigation of the performance of
various photosensitizers. Eoxvalue instead ofE1/2PC+/PC.(B) Monitoring the
ratio ofE/Z-enamine by means of in situ irradiation in NMR spectrometer. The
starting material was preequilibrated to attain an equilibrium ratio ofE/Z-
enamine. Conditions were (S)-1a/HNTf 2 (0.067 M),2a(0.067 M), benzoic acid
(20 mol %), and Ir(ppy) 3 (2.5 mol %) in 0.6 mL MeCN-d 3 , preequilibrated for
10 hours then exposed to 405-nm laser irradiation inside NMR spectrometer
for 20 min. Details are in the supplementary materials (figs. S19 to S23).
(C) Time-course profile for deracemization under the standard conditions.
(D) Monitoring the enamine formation process by NMR (a) with (S)-2a; (b)
with (R)-2a; and (c) withrac-2a. Conditions were (S)-1a/HNTf 2 (0.067 M) and2a
(0.067 M) in 0.6 mL MeCN-d 3 , with PhCOOBn as the internal standard.RESEARCH | REPORTS