Full Paper
a significant amount of benzaldehyde 9a after workup, where-
as addition of allyltributylstannane and Sc(OTf)3 to zirconacycle
6a provided the allylated product 12a in 36% yield, along
with secondary alcohol 13a in 34% yield (Table 1, entries 1
and 2). This result indicated that addition of Sc(OTf)3 to zirco-
nacycle 6a accelerated, not only the formation of iminium ion
7a, but also the hydrolysis to benzaldehyde 9a, which under-
went acid-mediated allylation to give byproduct 13a. There-
fore, we surveyed various acids to form iminium ion 7a selec-
tively. Use of BF3·Et2O improved both the yield and selectivity
and gave 12a in 56% yield (Table 1, entry 3). After extensive
investigation of acid additives, trifluoroacetic acid (TFA) was
found to be the best: 12a was isolated in 75% yield (Table 1,
entry 4). Finally, we found that addition of MeCN as a co-sol-
vent prior to the allylation completely suppressed the forma-
tion of 13a and provided 12a in 86% yield (Table 1, entry 5).[14]
The optimized conditions were then tested in the presence
of methyl esters, which are generally more electrophilic than
amides (Scheme 3). As we expected, aromatic methyl esters
were not affected in our reductive allylation of amide carbonyl
groups (12b–c: 72–79%). A more reactive aliphatic methyl
ester was reasonably tolerated (12d: 67%). The developed
conditions were found to be compatible with, not only esters,
but also a variety of sensitive functional groups (Scheme 3).
Electrophilic functionalities including nitro, nitrile, and benzyl
chloride were all tolerated (12e–g: 61–67%). Aryl bromide,
which can be used subsequently in metal-catalyzed coupling
reactions, did not interfere with the nucleophilic addition
(12h: 70%). Carbamate groups are common nitrogen protect-
ing groups and were challenging substrates due to the pres-
ence of the relatively acidic proton and an electrophilicity simi-
lar to that of the amide carbonyl center. However, these func-
tional groups were well differentiated from the amide group
(12i–j: 74–78%). Sulfonamides, with a more acidic proton, did
not disturb the reaction (12k–l: 76–78%). One limitation was
low compatibility with functional groups that contained pro-
tons derived from oxygen atoms such as alcohols and carbox-
ylic acids (12m: 0%; 12n: 0%). No desired product was ob-
served, likely due to the high oxophilicity of zirconium. Termi-
nal alkenes were well tolerated despite potential hydrozircona-
tion of the unsaturated carbon bonds. Substrates that con-
tained a cinnamoyl group disturbed the second allylation step
(12p: 0% versus 12o: 77%). The reaction of a terminal alkyne
resulted in competition with hydrozirconation: 12q was ob-
tained in 46% yield, along with alkene 12o in 28% yield. Acid-
labile functional groups, such as tert-butoxycarbonyl (Boc) and
acetal groups, were not affected despite the use of TFA (12i:
74%; 12r: 68%; 12s: 83%).
Scheme 3. Chemoselective reductive allylation of tertiary amides. Reaction
conditions: 5 (1 equiv), [Cp2ZrHCl] (1.2–2.6 equiv) in THF, CH2Cl2 (0.1m), RT,
10 min; CH2=CHCH2SnBu3 (3 equiv), TFA (1.5 equiv), MeCN, RT, 12 h. Yield of
isolated product after purification by column chromatography. Boc=tert-bu-
toxycarbonyl, Cbz=benzyloxycarbonyl, Ns=o-nitrobenzenesulfonyl,
MOM=methoxymethyl, THP=2-tetrahydropyranyl, Ts=4-toluenesulfonyl.
directly installed at the amide carbonyl center to give 20 by
a Mukaiyama-type vinylogous Mannich reaction with a siloxy-
furan (Table 2, entry 7).
With the chemoselective reductive nucleophilic addition to
tertiary amides established, we turned our attention to secon-
dary amides (Table 3). The reductive allylation was initially eval-
uated with N-benzylbenzamide (1a). Exposure of 1a to a sus-
pension of freshly prepared Schwartz reagent initiated reduc-
tion to imine 3. A variety of allylation reagents were then
added to the resulting imine 3 in a one-pot sequence. The se-
lectivity between secondary amine 21a and secondary alcohol
13a was largely influenced by the choice of allylation reagent.
Secondary amine 21a was not formed when an allylboronic
acid was used; secondary alcohol 13a was obtained exclusive-
ly (Table 3, entry 1). Use of allylindium bromide afforded 21a
(56%), along with 13a (7%; Table 3, entry 2). Complete selec-
The reaction allowed us to employ a variety of mild nucleo-
philes without deterioration of the high chemoselectivity
(Table 2). Reductive propargylation and Strecker reaction of ter-
tiary amide 5b proceeded in good yields, whereas intermolec-
ular Pictet–Spengler reaction with N-methylindole gave only
trace product (Table 2, entries 1–3). The reductive Mannich re-
action was possible under Mukaiyama-type conditions with
silyl ketene acetals, but silyl enol ethers provided none of the
desired product (Table 2, entries 4–6). A butenolide moiety was
Chem. Eur. J. 2014, 20, 1 – 8
3
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ