Organic Letters
Letter
affirm this, kinetic isotopic effect experiments were conducted.
As demonstrated by eq (a) in Scheme 2, no significant kinetic
elimination and the carbonyl insertion to be the cause of the
different chemoselectivites of the two ligands. However, the
reductive elimination barrier (TS11 and TS11a) for both
ligands is much higher than the highest transition states (TS8
and TS9a, respectively) in the pathway leading to 3a, which
explains the chemoselectivity of dppb but not of P(o-tolyl)3.
Moreover, the reductive elimination barrier (42.3 kcal/mol,
TS11a relative to IM6a) is too high to produce 4a.
Interestingly, we found that IM6a can undergo C,Br-reductive
elimination via TS12a with a feasible barrier of 22.9 kcal/mol,
leading to IM16a. Thermodynamically, the conversion of
IM15a to 4a is exergonic by 17.8 kcal/mol. To see if IM16a
could convert to 4a, we conducted an experimental study. As
we were not able to prepare separate IM16a due to its
instability, an analogue could be prepared successfully (eq (f)
in Scheme 2). Indeed, the analogue could be transformed to
the 4a analogue facilely.
Scheme 2. Mechanistic Experiments
The C,Br-reductive mechanism rationalizes the production
of 4a with P(o-tolyl)3, but the energetic difference (8.3 kcal/
mol) between TS12a and TS9a is too large for the observed
ratio (80/9) of 4a:3a. Thus, there must be a more favorable
alternative to afford 3a, encouraging us to reconsider the
pathway via a 1,4-palladium shift starting from IM6a. As
described by the green pathway in Figure 1C and Figure S1 in
the SI, the mechanism can also lead to 3a. Kinetically, the 1,4-
palladium shift mechanism is 5.8 kcal/mol more favorable than
the black pathway. Thus, the production of 3a as the minor
product with P(o-tolyl)3 actually takes place with a mechanism
completely different from that of the dppb ligand. The kinetic
preference of 2.5 kcal/mol to give 4a over to 3a explains the
major product 4a for P(o-tolyl)3.
We further crosschecked if the C,Br-reductive elimination
could take place with the dppb ligand. As compared in Figure
1A, the two reductive elimination transition states (TS12 and
TS13) are 3.7 and 18.0 kcal/mol higher than TS9, respectively.
Thus, the reaction with the dppb ligand could not produce 4a,
further supporting our mechanism on the production of ether
4a.
isotopic effect was observed, supporting our prediction of
dediazonation as the rate-determining step. To further
corroborate our mechanism, we traced the source of the
benzylic hydrogen in 3a through deuterium-labeling experi-
ments. Indeed, eq (b) demonstrates that the hydrogen
originates from the aldehyde group of 1, in line with our
mechanism. Puzzlingly, eq (c) suggests that methanol could
also be the hydrogen source, seemingly contradicting our
mechanism, because, as shown by TS7, the methanol hydroxyl
The discussions above indicate that the chemoselectivity of
the reaction is determined by how difficult it is for the C,C-
coupling intermediate (IM11 or IM6a) to undergo C,Br-
reductive elimination. Given that palladium catalysis generally
disfavors reductive elimination to form a C−heteroatom bond,
it is interesting to understand why IM6a could undergo the
reductive elimination but IM11 could not. We compared the
NBO charges of the palladium centers of IM6a and IM11. The
positive charge (0.08e) in IM6a, compared to the negative
charge (−0.12e) in IM11, agrees with that the former with a
barrier of 22.9 kcal/mol more easily undergoes reductive
elimination than the latter with a barrier of 30.9 kcal/mol. The
negatively charged palladium in IM11 could be attributed to
the stronger electron donation of the bidentate dppb ligand.
This explanation also holds true for the higher C,O-reductive
elimination barriers of IM13 (35.8 kcal/mol) than that of
IM13a (24.0 kcal/mol) because the Pd atom bears more
positive charge in IM13a (0.26e) than that in IM13 (0.09e).
On the basis of above mechanistic studies, Scheme 3
sketches the catalytic cycle of the reactions. After generation of
the activated catalyst (Pd-L), the reaction takes place via the
following steps: oxidative addition from Pd-L to A,
dediazonation from A to B which could proceed via either a
neutral or cationic mechanism, methanol deportation from B
to C, and sequential hydrogen transfers from C to D to E,
−
hydrogen is extracted by KCO3 to enter KHCO3. To
reconcile the disagreement, we conducted the control
experiments (eqs (d)−(e)). As shown by eq (d), the methanol
hydroxyl hydrogen can exchange with the benzylic hydrogen of
N-tosylhydrazone facilely, thus resulting in the scrambling in
eqs (c) and (e).
To understand the chemoselectivity of the reaction, the
reaction with L5 (i.e., P(o-tolyl3)) was also calculated. The
reaction also includes two stages. The pathway for the coupling
stage 1a and 2a′ is displayed in Figure S1 in the SI. In contrast
to the competitive two dediazonation mechanisms of dppb
(see TS2 and TS4+ in Figure 2A), the dediazonation pathway
of L5 via migratory insertion of palladium carbene is 8.7 kcal/
mol more favorable, compared to the cationic concerted
pathway. The difference of the two ligands can be ascribed to
the stronger electron-donating effect of the bidentate dppb
ligand than the monodentate P(o-tolyl)3 ligand which favors
the stabilization of the cationic species
Figure 1C displays the pathways for the C−O bond
formation stage. Analyzing IM13 and IM13a, intuitively, one
may consider the competition between C,O-reductive
D
Org. Lett. XXXX, XXX, XXX−XXX