Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
Repeated experiments showed that the reaction conversion
and yield were very sensitive to slight changes in the reaction
conditions, particularly the water content. Thus, when 5
equivalents of water were added to the reaction mixture,
quantitative conversion was observed (rather than 74%), but
the yield was lower (53% vs. 55%). Moreover, depending on
the exact temperature of the reaction and the reaction time, the
water elimination can occur to a larger or smaller extent and is
therefore able to significantly influence the reaction yield.
Accordingly, we observed lower-than-usual reproducibility
for experiments with almost similar reaction conditions, and
the conversion and yield data contain a significant experi-
mental error. We must emphasize that the differences (up to
12%) were observed mostly between different series of experi-
ments, while within the same series of experiments the
reproducibility of the conversion and yield was much better.
Furthermore, the enantioselectivity seems less sensitive to the
variation in conditions and depends mostly on the dendron
generation. Thus, for the reaction of benzaldehyde with
acetone, catalyzed by G1(O-Pro) under our standard condi-
tions, an enantiomeric excess of 61% ꢀ 2% was always
observed. Only a significant raise of the reaction temperature
(75 1C) or the decrease in the amount of acetone (5 equivalents
instead of 27) led to small but notable changes in the ee
(55% in the first case and 71% in the latter).
Scheme 4 Synthesis of the proline ester catalysts. Reagents and conditions:
(i) DIC, DMAP, DCM, rt, 6 h; (ii) piperidine, DMF, rt, 5 min.
clear ‘‘positive’’ or ‘‘negative’’ trend. We were slightly encour-
aged by the positive influence of the dendritic interface on the
conversions and yields of the reactions (particularly in the case
of nitrobenzaldehyde).
Since we already had at hand the hydroxyl-terminated
dendronized resins Gn(OH) [n = 1–3] (as well as the commer-
cial Wang resin for comparison),16 we decided to examine a
simple esterification mode of immobilization and prepared a
series of proline-decorated resins Gn(O-Pro) [n = 0–3] with
proline units immobilized as esters (Fig. 1b). The potentially
catalytic resins were prepared cleanly and quantitatively from
the parent OH-terminated resins (Scheme 4). A characteristic
splitting of the diastereotopic benzylic hydrogen signals into
1
an AB pattern was observed in the H NMR of the cleaved
proline-decorated dendrons, because of the proximal chiral
a-carbon center of proline.
The testing of the new catalytic series in the model aldol
reactions demonstrated a significant positive dendritic effect, with
the resins of the first to second generation being the optimal
catalysts. The effect was observed on conversion, yield and
enantioselectivity and was particularly prominent for the
nitrobenzaldehyde substrate (Table 2). Along with the aldol
product the achiral water elimination product 2 was usually
observed, particularly in the reaction of benzaldehyde. In
addition, the product 3 of Robinson annulations of acetone with
the enone byproduct 2 was found in the mixtures of the reactions
involving benzaldehyde (but not nitrobenzaldehyde).17
One of our major concerns, upon discovering that polymer-
bound proline esters catalyze the aldol reaction, was the
possibility of proline hydrolytic detachment from the support
and subsequent catalysis by the generated L-proline in
solution. This hypothesis was considered as an alternative to
the multivalency-derived dendritic effect of presumably
cooperative origin, similar to that previously observed by us
for polymer-supported proline catalysis.8f,g The influence of
added water (vide supra) and the relative lability of the ester
bond (vs. the amide bond, for instance) seemed to support the
hydrolysis-involving explanation of the effect. Moreover,
when we analyzed the resin G1(O-Pro) recovered from a
catalytic experiment, cleaving the dendrons from the resin
acidolytically, the 1H NMR demonstrated that ca. 20% of
the terminal sites of the dendrons had ‘‘lost’’ the proline unit,
which was attached to them prior to the catalytic run.
L-Proline by itself was not observed in this cleavage solution
by NMR. Thus the possibility that the ‘‘missing’’ proline units
were hydrolized by the acid in the cleavage solution can be
ruled out. Likewise, this observation implies that these units,
being cleaved during the catalytic reaction, were released into
the solution component of the reaction mixture, rather than
trapped in the polymer matrix.18
Table 2 The aldol reaction with Gn(O-Pro) supported catalystsa
Entry Catalyst
R
Conversionb (%) Yieldb (%) eec (%)
1
2
3
4
5
6
7
8
G0(O-Pro)
G1(O-Pro)
G2(O-Pro)
G3(O-Pro)
G0(O-Pro) NO2
G1(O-Pro) NO2
H
H
H
H
26
74
100
100
49
26
55
40
39
49
67
81
62
48
61
58
57
8
44
60
54
In spite of these facts, implicating L-proline released from
the support being an active catalyst, other observations did
not support this scenario. Firstly, in the case of the ‘‘released
proline’’ being a primary catalyst, we must assume a positive
dendritic effect on the proline hydrolysis (i.e. that the proline
release from G2(O-Pro) is faster than from the lower genera-
tion catalysts), rather than on the catalysis itself. Moreover, we
still must assume catalysis by bound proline esters, but with a
lower activity and selectivity. In order to address these far
going assumptions of the hydrolysis-involving explanation of
the effect, we monitored the reaction between benzaldehyde
67
G2(O-Pro) NO2 100
G3(O-Pro) NO2 100
a
Reaction conditions: 1 equiv. of aldehyde, 27 equiv. of acetone and
0.3 equiv. of catalyst in DMSO for 4 days at room temperature.
b
c
Conversions and yields determined by NMR. ee was determined by
HPLC, using Chiralpak AD (R = H) or Chiralcel OJ (R = NO2)
columns.
c
396 New J. Chem., 2012, 36, 394–401
This journal is The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2012
Kehat, Tzofit
