J. E. Jackson et al. / Tetrahedron Letters xxx (2015) xxx–xxx
3
yields. Variations in yield may in part reflect numbers and distribu-
tion of the residual Al-OH groups that surprisingly remain in the
material (see below) and serve as proton sources.
0.9 mmol of reducible Al-OH groups per gram of Al2O3, enough
to protonate all reduced intermediates. Thus, the residual Al-OH
groups on the surface of the alumina nanocrystals, and therefore
near the encapsulated sodium, could provide the proton source.
The variable density of such sites from sample to sample could also
be responsible for the variations of desired versus overreduced
product ratios from run to run. It had been thought that Al-OH
groups would be eliminated by hydrogen production during nor-
mal preparation (by molten Na treatment at 160 °C) of Na-AG from
alumina that had been calcined at 600 °C. However, careful moni-
toring of the gas phase during Na-AG preparation revealed that
hydrogen is not formed during this incorporation of sodium.
The above findings prompted several attempts in D2O under
hydrothermal conditions to pre-deuterate the alumina used for
preparation of the Na-AG. Despite the use of rigorous glassware
drying and drybox techniques, no replacement of H by D was seen,
either by FT-IR of the dried, calcined AG, nor in either of the over-
reduction products from Na-AG treatment of 7 or 8.
Though competing reductions of functional groups within a
compound do occur, the absence of products in which only the
C–Cl bond is cleaved (7) or only the pi bond is reduced (8) suggests
preferential N–OCH3 cleavage in advance of, or to initiate further
reduction. Thus, when both reductions occur, it is likely that a sin-
gle encounter of the Weinreb amide with the reducing site is
involved. Reaction does not appear to depend on special sites;
explicit attempts to quench the most reactive sites of Na-AG by
pre-treatment with ethanol at 10–30% of the Na content do not
alter selectivity. Since each of the reactive sites on a nanoparticle
of sodium has many adjacent sodium atoms that can exchange
rapidly with a reducing site,18 multiple reduction steps are possi-
ble. If a molecule with two reducible sites stays adsorbed long
enough, a second reduction could occur. The resulting reduced
but unprotonated products would remain bound to the alumina
gel until protonation released them.
The rearranged byproduct from compound 8 (Scheme 3) can
most simply be understood in terms of base- or radical promoted
elimination across the HCH2–ON bond of the Weinreb amide to
form formaldehyde. This electrophile, trapped in the pore in prox-
imity to the reduced cinnamate anion evidently undergoes aldol
condensation with the amide enolate, forming a hydroxymethyl
group vicinal to the amide. Such base-promoted cleavage of Wein-
reb amides has been observed previously under treatment with
strong bases such as LDA.24 N-Hydroxymethylated amides have
been produced in one case,25 albeit in a Lewis acid context, but
formaldehyde adducts directly analogous to the present case have
not been reported from such reductions, even under conditions
where enolate formation was verified via H/D exchange. Interest-
ingly, in the present case, only the cinnamide 8 yields this product;
its saturated analogue 3 yields no hydroxymethylation. We specu-
late in analogy to previous suggestions24,25 (Scheme 4) that the
radical anion formed upon addition of an electron to the cinnamide
pi system serves as a base to abstract one of the methoxy –OCH2-H
hydrogen atoms, forming the formaldehyde in close proximity to
the enolate anion and leaving an unpaired electron on the amide
N, which then rapidly accepts an additional electron.
In summary, this work shows that Na-AG and Na-SG easily and
cleanly effect WA reduction in simple, batch reactions under ambi-
ent conditions and in modest reaction times. These reactions are
not highly selective, but in situations with simple substrates or
where global reduction is desired, they are quite effective. Cases
where over-reduction occurs show how powerful the reducing
agents Na-AG and Na-SG are, and demonstrate unambiguously
that even after 600° calcination and treatment with molten
sodium, the alumina in Na-AG retains a nontrivial load of –OH
sites. In the absence of secondary reducible sites, this method
can be usefully applied to WAs to form secondary amines under
Since protonation of intermediates seems required to release
products, a number of runs were made with H2O or D2O or alcohols
added to the THF as possible protonating agents or used to quench
reactions which had been run in dry THF. The percentage of desired
product increased slightly, but no conditions completely prevented
over-reduction. When 5–10% H2O or D2O and 2 equiv of Na-AG
were used at 25 °C, five reductions of compound 8 yielded
37 9% desired product, compared with 19 3% for six runs in
dry THF.
Although the formation of over-reduced compounds could not
be eliminated, the percent desired reduction at the N–O site alone
is influenced by several factors. Complete over-reduction of 8
occurs in 2 h at 25 °C when 4 equiv of Na-AG are used, but about
25% simple N–O cleavage product is produced with 2 equiv. Lower-
ing the temperature also yields more of the desired product
although higher reagent amounts and/or longer reaction times
are required.
Consistent with the idea that both N–O reduction and over-re-
duction processes happen in one encounter, sampling at time
intervals through the reaction showed essentially no selectivity
variation; except for a slight preference for N–O reduction early
in the reaction, the relative ratios of reduced to over-reduced prod-
ucts were essentially independent of reaction progress. If protona-
tion of the amide’s N(À) site and release to solution occurred
before reduction at the secondary site, the proportion of over-re-
duced intermediates formed in a second reduction step should
vary, and reduced products should undergo further reduction; nei-
ther of these outcomes are seen.
The over-reduced product of the 4-chloro WA (compound 7),
has the Cl replaced by H. Similarly, the reduction of the double
bond in compound 8 requires the addition of two hydrogen atoms.
In an attempt to determine the source of this hydrogen, the normal
reduction in dry THF, followed by quenching with saturated NH4Cl
was modified (in separate experimental studies) by (1) reduction
in deuterated THF, (2) quenching with D2O and (3) reaction in
THF containing small amounts of H2O, D2O, or deuterated alcohols.
Product analyses (1H NMR and GC–MS) in all cases showed at most
small amounts of deuterium in the over-reduced products from
either of these experiments (see Supporting information). The
absence of deuterated over-reduced products suggests that the
strongly basic amide anion and carbanion intermediates formed
are bound to the alumina gel until protonation, which must occur
at or near the site of reduction before the products migrate into the
solution. Thus, proton donor sites must remain in the AG. Indeed,
even after alumina calcination at 600 °C, –OH bands could be seen
in FT-IR spectra of the AG, and the formation of hydrogen upon
reaction with liquid Na2K alloy verified the presence of at least
Na+
O
H2C
Na+
O
O
H
O
CH3
Na
CH3
CH3
N
O
N
N
N
O
H3C
C
H2
8
Na+
O
Na+
O
OH
O
O
O
2H+
Na
CH3
CH3
CH3
N
N
Na+
H
Scheme 4.
Jackson, James E.
