S. Shibata, Y. Masui and M. Onaka
Tetrahedron Letters 67 (2021) 152840
proton-exchanged Y-type zeolite (Si/Al = 2.75: abbreviated as H-Y)
gave the best result.
For the results of the H-Y zeolite, the number n in H-Y(n) indi-
cates that n molecules of 1a per supercage of H-Y were used. As the
n increased, the yield of 2a clearly decreased (entries 6–9): H-Y (1)
was the H-Y zeolite of choice. Hereafter, H-Y stands for H-Y (1) in
this paper. H-Beta zeolite also gave a high yield, but H- Mor and H-
ZSM-5 zeolite were not suitable for the reaction (entries 10–12).
Among a series of the H-zeolite, this trend in the yields does not
simply correlate with the solid acidity: H-Mor > H-ZSM > H-
Beta > H-Y,[22] indicating that the effective formation of 2a is
not directly dependent on the solid acidity.
Among the metal-ion exchanged zeolite, such as potassium A-
type zeolite (MS3A), sodium X- and Y-type zeolite (Na-X, Na-Y)
and silver Y-type zeolite (Ag-Y), Na-Y and Ag-Y[23]gave 50 and
59% yields of 2a, respectively (entries 3, 5), while the yield dropped
to 9% together with a low recovery of 1a (10%) on Na-X which has
the same zeolite framework as Na-Y (entry 2). MS3A has very small
cavities of 0.3 nm, which are too small for 1a to be allowed to react
with NH3, and 1a was completely recovered (entry 1). Interest-
ingly, co-adding the same amount of MS3A as Na-Y caused an
increase in the yield of 2a by ca. 20% (entry 4). This is probably
due to the powerful dehydration effect of MS3A on the
condensation.
By using the neutral and acidic silica, 2a was obtained in 80%
and 73% yields, respectively, with some 1a remaining (entries
13,14). In addition, CARiACT Q-3, which is a commercially- avail-
able mesoporous silica with a large surface area, showed a better
yield of 86% than the normal silica (entry 15). Neutral as well as
acidic alumina also promoted the reaction in good yields of 86%
and 77%, respectively (entries 16, 17). The reaction using K10,
which is a commercially-available acid-treated clay mineral, pro-
ceeded in only 40% yield (entry 18). With an excess amount of
ammonium chloride, no reaction took place (entry 19).
Fig. 1. 13C NMR spectrum of 1a in CDCl3 (a), and 13C DD/MAS NMR spectrum of
1a@H-Y (b).
In the solution NMR, the carbonyl carbon of 1a appeared at
196.9 ppm. On the other hand, in the solid-state NMR, that of
1a@H-Y was observed at 203 ppm, which was shifted by 6 ppm
lower than that in CDCl3 together with no peak of 1a. From this,
it can be said that the solid ketone material is completely occluded
into the H-Y pores only by simple mechanical mixing for such a
short period of time.
We previously observed that the carbonyl carbons of formalde-
hyde, acrolein, propargyl aldehyde and diphenylketene adsorbed
on the sodium ion sites in Na-Y showed the lower magnetic field
shifts of 13C DD/MAS NMR by 5 to 6 ppm than those in solution.
[18,23] It can be assessed that 1a is also coordinated to the proton
in the supercage of H-Y, demonstrating the same down-field shift
of the carbonyl carbon as shown in Fig. 1. It is apparent that the
protonated 1a in the H-Y was so activated that the addition of
ammonia and the successive dehydration to the NAH ketimine
2a were efficiently promoted. It should be also considered that
the zeolite’s high dehydration ability[24]and stabilization effects
[18b]on the labile ketimines caused the chemical equilibrium
shift to the ketimine formation even though it is intrinsically
unfavored.
When the intact 1a remains together with 2a, for example,
2a/1a = 8/2, in the reaction mixture, it is necessary to separate
2a from 1a. As the boiling points of 1a and 2a are close to each
other, it was difficult to purify 2a by simple distillation. Our trial
to purify the mixture on an open column using neutral silica gel
and an eluent system of hexane/AcOEt/Et3N (40/1/1) failed because
part of 2a underwent hydrolysis during passing through the col-
umn. Fortunately, 2a was reported to be purified by using flash col-
umn chromatography.[13] In the case of a solid ketimine product
like 2c, it is able to easily obtain the pure 2c by recrystallization.
Next, for the reaction of 1a with ammonia using the best pro-
moter, H-Y, the reaction temperature and time were optimized
(Table 2). When the reaction time was set at 16 h, and the temper-
ature was changed from RT to 60 ℃, it was found that the reaction
was completed even at RT (entries 1–3). When the reaction tem-
perature was fixed at 40 ℃, 2a was quantitatively obtained in ca.
3 h (entries 4, 5).
Note that 1a is a solid material at RT. In order to confirm that 1a
was completely incorporated into the zeolite pores simply by
mechanically mixing the solid materials of H-Y and 1a at RT for
30 min in the air, we compared the solution NMR spectrum of 1a
in CDCl3 (Fig. 1(a)) with the 13C DD/MAS NMR spectrum (Fig.
1
(b)) of the solid mixture, where one molecule of 1a was added to
one supercage of H-Y, which is expressed here as 1a@H-Y.
Table 2
Optimization of reaction conditionsa
Entry
Temp. (℃)
Time (h)
Yield of 2a (%)b
Recovery of 1a (%)b
1
2
3
4
5
60
40
r.t.
40
40
16
16
16
3
98
quant.
98
quant.
96
n.d.
n.d.
1
n.d.
2
1
a
1.7 mmol of 1a, 90 mmol of NH3 gas and 2.5 g of HY were employed.
1H NMR yields based on the ratios of 1a to 2a.
b
3