Journal of the American Chemical Society
Article
The kinetic isotopic effect experiment for parallel reactions
failed to be performed because of the extremely fast reaction
rate.
MECHANISTIC STUDIES
■
This ligand-controlled iron-catalyzed regiodivergent hydro-
silylation provides an ideal model for understanding the
behavior of iron catalysts. To this end, we performed some
control experiments and DFT calculations. We began with
deuterium-labeling experiments (Scheme 5A). When hydro-
To obtain detailed information about this unusual divergent
regioselectivity, we carried out DFT calculations on the
transition-state of ligand to ligand hydrogen transfer (LLHT,
Ts-1a) at all three possible spin states (singlet, triplet, and
quintet) of the iron catalysts (Figure 1a). Gaussian 09 was used
for calculations at the (dispersion-corrected) unrestricted
ωB97XD/6-31G*/TZVP level of theory with CPCM-
UωB97XD/def2-TZVPP single-point energy calculations.
Our calculation revealed that the LLHT at triplet state has
the lowest energy (3Ts-1a, Figure 1a). Next, we calculated the
Scheme 5. Control Experiments
3
energies of Ts-1 with iron catalysts C1e and C1i, which gave
opposite regioselectivities under the identical conditions
(Figure 1b and Table S7). When C1e was used, the energy
of anti-Markovnikov addition transition state Ts-1a was about
3.5 kcal mol−1 lower than that of Markovnikov transition state
Ts-1a′, which is consistent with the experimentally observed
high anti-Markovnikov selectivity. In contrast, the calculations
indicated that anti-Markovnikov addition transition state Ts-
1b was 1.8 kcal mol−1 higher in energy than Markovnikov
addition transition state Ts-1b’ when C1i was used as the
catalyst, which is again consistent with the observed high
Markovnikov selectivity. The calculation results exhibited good
accordance with the experimental outcomes of C1b and C1d
To gain a better understanding of the parameters
responsible for the regiodivergence of the reaction, we
performed additional calculations by using 1,10-phenanthro-
line-modified iron catalyst C1a as a reference (Figure 1b).
When this catalyst was used, the energy difference between
transition states Ts-1c and Ts-1c′, which lead to anti-
Markovnikov and Markovnikov selectivity, respectively, was
very small (0.13 kcal mol−1), indicating poor regioselectivity.
Because coordination of the triple bond with the iron center
was almost completely unaffected by the ligand, the overlap
between the alkyne and iron center orbitals in this transition
state was larger than that in any of the other transition states.
Because of the directionality of molecular orbitals, addition of
substituents to the 2- and 9-positions of the 1,10-phenanthro-
line ligand resulted in steric hindrance between the
substituents, which decreased the overlap between the alkyne
and iron orbitals and thus increased the transition state energy.
To quantify the degree of orbital overlap, we chose the plane
defined by the iron atom and the nitrogen atoms of the 1,10-
phenanthroline ligand (Fe−N−N) as the reference and the
dihedral angle (θ) between this plane and the plane defined by
the iron atom and the carbon atoms of the alkyne triple bond
(Fe−C−C) as an index to measure the orbital overlap. These
calculations showed that the difference in dihedral angle (Δθ)
between anti-Markovnikov transition state Ts-1a and reference
transition state Ts-1c (Δθ = 76.75° − 77.64° = −0.89°) was
considerably lower than the difference for Markovnikov
transition state Ts-1a′ and Ts-1c′ (Δθ = 78.40° − 80.92° =
−2.52°); these results are consistent with the experimentally
observed anti-Markovnikov selectivity exhibited by C1e. In
contrast, when catalyst C1i, which has 3,5-disubstituted aryl
groups, was used, the difference in dihedral angle between anti-
Markovnikov transition state Ts-1b and reference transition
state Ts-1c (Δθ = 81.51° − 77.64° = 3.87°) is greater than the
difference between Markovnikov transition state Ts-1b′ and
Ts-1c′ (Δθ = 83.10° − 80.92° = 2.18°), which is consistent
silylation reactions of 1-octyne with a 1:1 mixture of PhSiD3
(98% D) and C12H25SiH3 were carried out with catalysis by
C1g (condition A) or C1j (condition B), only homoaddition
products (3ab-d and 3df with C1g; 4ab-d and 4df with C1j)
were obtained; the fact that heteroaddition products 3ab′,
4ab′, 3df′, and 4df′ were not generated indicates that the H/D
atom and the silyl group that added to the 1-octyne came from
a single silane molecule. The results of the mixed-silane
experiments exclude a catalytic cycle that starts from an Fe−H
(Scheme 5, Path a) or Fe−Si (Scheme 5, Path b) species but
support a Chalk−Harrod-type catalytic cycle15 (Scheme 5,
Path c) because the redox-neutral process involving Fe−Si or
Fe−H species would inevitably lead to heteroaddition
products. According to the literature, the precursor of the
iron catalyst can be reduced to active Fe(0) species by
EtMgBr.16 A kinetic isotope effect experiment involving
competitive hydrosilylation reactions between 1-octyne and
phenylsilane/PhSiD3 with catalysis by C1g (condition A) or
C1j (condition B) showed first-order kinetic isotope effects
(kH/kD = 4.0 and 3.8, respectively; Scheme 5B). These data
indicate that hydrogen transfer might be a rate-limiting step.17
F
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX