Journal of the American Chemical Society
Article
theoretical study, three types of oxidative mechanisms20
involving inner-sphere single electron transfer (ISET), outer-
sphere single electron transfer (OSET) and concerted
oxidative addition (COA) were examined for the reaction of
9b or 2b with different Ni species. When 2b was used as
oxidant to react with Ni(0) complex v, the calculated free
energy barrier for the ISET process was 13.1 kcal/mol via an
ISET-assisted homolytic C−I bond cleavage transition state ts-
7 leading to the formation of a Ni(I)-iodide species int-6 and
an ethyl radical. Subsequent recombination with ethyl radical
furnishes Ni(II)-alkyl intermediate int-7. Although intermedi-
ate int-7 could also be generated from a COA process via
three-membered ring transition state ts-11, such a pathway
could be excluded by the relatively higher free energy barrier of
18.9 kcal/mol.
Alternatively, when 9b undergoes reaction with v, DFT
calculations revealed that the free energy barrier for the ISET
pathway (homolytic N−O bond dissociation) is only 10.5
kcal/mol via transition state ts-1. In this process, the
coordination of phthalimide nitrogen to the Ni center
promotes N−O bond cleavage in 9b, generating a Ni(I)-
phthalimide species int-1 with concomitant release of a
phenylacetate radical 9″. Facile decarboxylation of 9″ ejects
a benzyl radical, which can further oxidize int-1 via transition
state ts-3 to afford Ni(II)-benzyl intermediate int-2. In
contrast, the free energy barrier for the COA pathway was
found to be much higher (20.7 kcal/mol) than that of the
ISET process. On the other hand, 9b may serve as a single-
electron acceptor and participate in a OSET process as
commonly proposed in the literature.12 The calculated
activation free energy for this pathway by using approximation
of modified Marcus theory21 is 12.1 kcal/mol to deliver a
cationic Ni(I) intermediate int-4 that is 11.3 kcal/mol
endergonic. However, after the ensuing phthalimide anion
transfer to form int-1 via ts-5, the calculated activation free
energy of the overall process is as high as 22.6 kcal/mol. Thus,
the OSET pathway is likely to be unfavorable.
On the basis of the above calculations, N-(acyloxy)-
phthalimide 9b is found to chemoselectively react with
Ni(0) complex v through stepwise ISET-type oxidation,
contrary to previously established pathways, to afford Ni(II)-
benzyl intermediate int-2. Following intramolecular 1,2-alkene
insertion into the Ni−C(benzyl) bond and single-electron
reduction by Mn, a Ni(I)-alkyl species ix-b is formed, which
can subsequently react with either 9b or iodoethane 2b.
Further DFT studies indicated that ix-b readily reacts with 2b
through a ISET process via transition state ts-12 with an
activation free energy of only 5.4 kcal/mol. The resulting
Ni(II)-iodide int-9 then captures the ethyl radical to furnish
Ni(III)-dialkyl intermediate int-10, which is susceptible to
reductive elimination to yield the desired dialkylation product.
For comparison, we also evaluated the reactivity profile of ix-
b with N-(acyloxy)phthalimide 9b. The calculated free energy
barrier for the ISET pathway is 14.6 kcal/mol via transition
state ts-16, which is 9.2 kcal/mol higher than that with 2b via
ts-12. Furthermore, the corresponding OSET and COA
pathways were also found to be energetically less favorable
compared to the calculated ISET pathway with 2b. Therefore,
the excellent site selectivity that was observed in Scheme 3a
primarily arose from the orthogonal reactivity of N-(acyloxy)-
phthalimide 9b and iodoethane 2b with different organonickel
species (i.e., v and ix-b) generated in the catalytic system.
The origin of the high regioselectivity cannot be simply
rationalized by conventional electronic and/or steric argu-
ments involving the two electrophiles and the in situ generated
Ni complexes v and ix-b. To gain further insights, careful
distortion/interaction analysis22 of the four ISET-type
transition states ts-1, ts-7, ts-12, and ts-16 was carried out,
where the total activation energy (ΔE‡act) was separated into
the distortion energy of two reacting partners (ΔE‡dist) and the
interaction energy between those two distorted partners
(ΔE‡int). As shown in Scheme 3c, when N-(acyloxy)-
phthalimide 9b reacts with Ni(0) species v, a large interaction
energy of −40.2 kcal/mol was observed in transition state ts-1.
This can be explained by the strong association of the
phthalimide nitrogen with the Ni center. The distortion energy
of ts-1 largely arises from dissociation of the 8-aminoquinoline
amide nitrogen, which may be nullified by the coordination of
9b. On the contrary, the interaction between iodoethane 2b
and v in transition state ts-7 is significantly lower. Hence,
ΔE‡ becomes more favorable in ts-1 (vs ts-7), and v is
act
predicted to preferentially undergo oxidation with 9b (instead
of 2b). Intriguingly, the interaction between 2b and the more
electron-deficient Ni(I) species ix-b is dramatically enhanced
in transition state ts-12, whereas the distortion energy remains
almost unchanged (vs ts-7). On the other hand, the association
of ix-b with 9b causes greater distortion within ts-16 (D,
dihedral angle; DN1−Ni−N2‑C = 133.0° vs reaction with 2b via ts-
12, DN1−Ni−N2‑C = 167.5°). Therefore, the second oxidation
involving ix-b would selectively occur in the presence of 2b
(instead of 9b).
Our DFT results are supported by control experiments in
which both the N-(acyloxy)phthalimide and alkyl halide
reagent likely reacted through ISET-type radical processes
(Scheme 4a,c). With 9ag, calculations show that radical
cyclization is more favored than trapping of the initially
generated tertiary radical (Scheme 4b). Although the use of
(iodomethyl)cyclopropane 2s delivered the expected product
5bf with no trace of ring rupturing, computations showed that
the ring-opening rate of the cyclopropylmethyl radical
intermediate (generated from iodine abstraction/radical
recombination) is comparatively slower than the rate of radical
association with the catalytic organonickel species (Scheme
4c). This result is in contrast to a previous dialkylation
disclosure in which only the ring-cleaved product was
detected.9
Consolidating the information derived from experimental
and theoretical studies, we propose a possible catalytic cycle
(Scheme 3d) that begins with initial formation of a putative
Ni(0) species iv (possibly from reduction of the Ni(II)
precatalyst by Mn),10a10g that undergoes ligand exchange with
substrate 4a to generate Ni(0) complex v. Instead of engaging
with the alkyl halide 2, v chemoselectively reacts with the N-
(acyloxy)phthalimide 9 through an ISET process to afford a
Ni(I)-phthalimide intermediate vi and a carboxylate radical.
Facile extrusion of CO2 furnishes an alkyl radical that
recombines with vi to give Ni(II)-alkyl intermediate vii. An
intramolecular site-selective alkylnickelation across the che-
lated π bond to form viii followed by reduction of the Ni
center by Mn could proceed to generate Ni(I)-alkyl complex
ix. At this stage, ix preferentially reacts with 2 through another
ISET pathway to deliver a new Ni(III)-alkyl species xi (via x)
that reductively eliminates to release the desired dialkylation
adduct 5, before an ensuing reduction process and ligand
exchange regenerate v.
D
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX