Reciprocal resolutions between 1-phenylethylamine and carboxyesters of isopropylidene glycerol: Improvement of the method by replacing mono-phthalate with 3-carboxy-2-naphthoate

Article abstract of DOI:10.1016/S0957-4166(01)00439-6

A novel resolving agent, isopropylidene glycerol 3-carboxy-2-naphthoate 2, was designed on the basis of the consideration that replacement of phenyl group with a naphthyl group would improve the resolving ability of isopropylidene glycerol hydrogen phthalate 1 while also conferring more suitable physicochemical properties for such a specific use. Indeed, 1-phenylethylamine 4 was resolved by 2 more efficiently than by 1 (respective resolution efficiencies, (S) 0.88 and 0.81), while 1 and 2 were resolved by 4 with S ranging between 0.54 and 0.59. Furthermore, 2 is a solid, whereas 1 is a viscous oil, and its recovery at the end of the resolution procedure is easier than that of 1. In order to understand the chiral discrimination mechanism of the two reciprocal resolutions, the binary melting point phase diagrams of the four diastereomeric systems (S)-2·(S)-4/(S)-2·(R)-4, (S)-2·(S)-4/(R)-2·(S)-4, (S)-1·(S)-4/(S)-1·(R)-4 and (S)-1·(S)-4/(R)-1·(S)-4 were determined. The first two systems form ideal conglomerates, characterised by identical diagrams, in which the eutectic corresponds to a 0.10 molar ratio of (S)-2·(S)-4. The same behaviour was shown by the other two systems, whose eutectics, however, correspond to a 0.18 molar ratio of (S)-1·(S)-4. On the basis of the present results, which indicate an excellent resolution ability of 2 for 4, the application of this new acid to the resolution of other 1-arylalkylamines seems to have very good prospects, worthy of investigation.

Full text of DOI:10.1016/S0957-4166(01)00439-6

TETRAHEDRON:
ASYMMETRY
Pergamon
Tetrahedron: Asymmetry 12 (2001) 2489–2495
Reciprocal resolutions between 1-phenylethylamine and
carboxyesters of isopropylidene glycerol: improvement of the
method by replacing mono-phthalate with
3-carboxy-2-naphthoate
Marco Pallavicini,a,* Ermanno Valoti,a Luigi Villaa and Oreste Piccolob
aIstituto di Chimica Farmaceutica e Tossicologica, Universita` di Milano, viale Abruzzi 42, I-20131 Milan, Italy
bStudio di Consulenza Scientifica, via Borno` 5, I-20060 Sirtori (LC), Italy
Received 24 September 2001; accepted 9 October 2001
Abstract—A novel resolving agent, isopropylidene glycerol 3-carboxy-2-naphthoate 2, was designed on the basis of the
consideration that replacement of phenyl group with a naphthyl group would improve the resolving ability of isopropylidene
glycerol hydrogen phthalate 1 while also conferring more suitable physicochemical properties for such a specific use. Indeed,
1-phenylethylamine 4 was resolved by 2 more efficiently than by 1 (respective resolution efficiencies, (S) 0.88 and 0.81), while 1
and 2 were resolved by 4 with S ranging between 0.54 and 0.59. Furthermore, 2 is a solid, whereas 1 is a viscous oil, and its
recovery at the end of the resolution procedure is easier than that of 1. In order to understand the chiral discrimination
mechanism of the two reciprocal resolutions, the binary melting point phase diagrams of the four diastereomeric systems
(S)-2·(S)-4/(S)-2·(R)-4, (S)-2·(S)-4/(R)-2·(S)-4, (S)-1·(S)-4/(S)-1·(R)-4 and (S)-1·(S)-4/(R)-1·(S)-4 were determined. The first two
systems form ideal conglomerates, characterised by identical diagrams, in which the eutectic corresponds to a 0.10 molar ratio of
(S)-2·(S)-4. The same behaviour was shown by the other two systems, whose eutectics, however, correspond to a 0.18 molar ratio
of (S)-1·(S)-4. On the basis of the present results, which indicate an excellent resolution ability of 2 for 4, the application of this
new acid to the resolution of other 1-arylalkylamines seems to have very good prospects, worthy of investigation. © 2001 Elsevier
Science Ltd. All rights reserved.
1. Introduction
introduction of a fused second phenyl ring would hope-
fully result in solid hemiester derivatives of glycerol
acetonide, easier to handle, recover and purify than 1,
which is, on the contrary, a viscous oil.
The hydrogen phthalate of isopropylidene glycerol 1
has proven to be an efficient resolving agent of 1-
phenylethylamine
4 and of many other primary
amines.1–4 On the other hand, its resolution with 4 has
also been reported as a useful method to prepare both
1 and isopropylidene glycerol with very high enan-
tiomeric purity.1,5,6 Believing that replacement of the
phenyl with a naphthyl group would improve the
resolving ability of 1 and widen its range of application,
as suggested by recent investigations on mandelic acid,7
we considered the use of the carboxynaphthoates of
isopropylidene glycerol as an advantageous alternative.
We speculated that a rigid and large naphthyl group
would produce more stable crystal structures for the
less soluble salts, accentuating the solubility difference
between the diastereoisomers. Last but not least, the
O
O
O
O
O
O
O
O
COOH
COOH
1
2
O
NH2
O
O
O
COOH
3
4
* Corresponding author. E-mail: marco.pallavicini@unimi.it
0957-4166/01/$ - see front matter © 2001 Elsevier Science Ltd. All rights reserved.
PII: S0957-4166(01)00439-6

2490
M. Palla6icini et al. / Tetrahedron: Asymmetry 12 (2001) 2489–2495
Herein, we describe the preparation of isopropylidene
glycerol 3-carboxy-2-naphthoate and 1-carboxy-2-naph-
thoate, 2 and 3, respectively, and the results of the
successful reciprocal resolution between 2 and 4. These
results are compared with those of the previous recipro-
cal resolution between 1 and 41,5 and both are evaluated
on the basis of the respective binary melting point phase
diagrams. The latter were determined in order to identify
the type of diastereomeric salt mixtures occurring in the
two reciprocal resolutions and to assess the degree of
optimisation of the procedures.
Before undertaking resolution of 4 with 2, samples of
diastereomeric salts (S)-2·(S)-4 and (S)-2·(R)-4 were
prepared in order to determine the respective melting
points and solubilities in alcohols. The results of these
measurements are listed in Table 1, where the data
obtained for the corresponding diastereomeric phtha-
lates of 1-phenylethylamine (S)-1·(S)-4 and (S)-1·(R)-49
are also reported. Resolution of 4 with 2 was performed
in methanol, which, unique among the tested alcohols,
showed largely differentiated and, at the same time,
reasonable dissolving abilities for the two diastereomeric
salts. The resolving agent 2 was used in the (S)-form and
in equimolar amount to the racemic substrate. The salt
of (S)-4 with (S)-2 precipitated in 90% yield (relative to
the theoretical amount, i.e. half of the overall starting
salt). (S)-4 was liberated from the precipitate and isolated
by distillation with e.e. 98.3%, determined by chiral
HPLC, and in 74.4% yield (37.2% of the starting 4) (Table
2, entry 1). (S)-2·(R)-4 was obtained by concentration of
the methanolic filtrate in 108.8% yield (54.4% of the
overall starting salt) and decomposed to give crude (R)-4,
which was isolated by distillation with 80.0% e.e. and in
82.0% yield (41.0% of the starting 4). The total recovery
of the resolving agent from the two separated
diastereomeric salts amounted to 93.4% of the initially
employed quantity.
2. Results
The racemic acid 2 was obtained by reaction of 2,3-naph-
thalic anhydride with racemic isopropylidene glycerol in
pyridine. The anhydride was in turn prepared by heating
naphthalene–2,3-dicarboxylic acid in refluxing acetic
anhydride as reported in the literature.8 (S)-2 and (R)-2
were synthesised by the same method, but starting from
(R)- and (S)-isopropylidene glycerol, respectively. In any
case, the pure hemiesters were isolated as viscous oils in
near quantitative yield by simple dilute sulphuric acid/
ethyl acetate extraction, which turned into high melting
solids upon standing at room temperature overnight.
Table 1. Melting point and solubilities of the salts of (S)-2 and (S)-1 with (S)- and (R)-4
(S)-2·(S)-4
(S)-2·(R)-4
(S)-1·(S)-4
413.7
(S)-1·(R)-4
400.7
Mp (K)a
427.3
408.9
Solubility (g/100 mL solvent)b
In methanol
In ethanol
2.1
0.3
0.2
20
2.7
1.2
4.4c
20c
d
d
–
–
d
d
In 2-propanol
–
–
a Determined by DSC analysis at a scan rate of 2.5°C min−1
.
b Determined at 20°C or, in the only case of 2-propanol, at 45°C.
c Previously reported in Ref. 9.
d Not determined.
Table 2. Resolutions, performed in methanol, of 4 with (S)-2 and (S)-1 and of 2 and 1 with (R)- and (S)-4, respectively
Entry
Racemic
substrate
Resolving
agent
Precipitate
Yielda (%)
2nd cryst.
E.e.b (%)
2nd cryst.
Sc
1st cryst.d
Recovered
substrate
1st cryst.
1
2
3
4
2
2
4
1
1
(S)-2
(R)-4
(R)-4
(S)-1
(S)-4
(S)-4
(S)-2·(S)-4
(R)-2·(R)-4
(R)-2·(R)-4
(S)-1·(S)-4
(S)-1·(S)-4
(S)-1·(S)-4
89.7
88.9
76.5
82.0
70.8
71.8
–
74.4
60.2
48.0
62.8
–
98.3
64.8
70.7
–
0.88
0.56
0.54f
0.81
0.57
0.59
60.8
48.9e
–
–
59.8
92.0
96.1
–
4g
5h
6g
\99
81.0
–
i
58.3
–
\99
a Relative to the theoretical amount, i.e. half of starting racemate.
b Enantiomeric excess (determined by chiral HPLC) of the substrate liberated from the precipitate.
c Experimental resolution efficiency calculated from S=chemical yield of the diastereomeric salt (%)×e.e. of the liberated substrate (%)/10000.
d From the solution containing equivalent amounts of racemic substrate and resolving agent.
e Two treatments of the first precipitate were performed in refluxing ethanol instead of recrystallising from methanol.
f Calculated from S=76.5×70.7/10000.
g Data previously reported in Ref. 1.
h Data previously reported in Ref. 5.
i Not determined.

M. Palla6icini et al. / Tetrahedron: Asymmetry 12 (2001) 2489–2495
2491
Figure 1. Binary melting-point phase diagrams for the diastereomeric systems (S)-1·(S)-4/(S)-1-(R)-4 (crosses), (S)-1·(S)-4/(R)-1-
(S)-4 (squares), (S)-2·(S)-4/(S)-2-(R)-4 (dashes) and (S)-2·(S)-4/(R)-2-(S)-4 (circles). The upper solid curve represents the values
calculated on the basis of the Schro¨der–van Laar equation for the mixtures containing (S)-2·(S)-4, while the lower one those
analogously calculated for the mixtures containing (S)-1·(S)-4.
The reciprocal resolution, i.e. that of 2 with 4, was
performed in methanol using the (R)-amine as a resolving
agent. The salt (R)-2·(R)-4 precipitated in 88.9% yield.
The enantiomeric excess of (R)-2, determined by chiral
HPLC after liberation from a sample of the precipitate
and methylation with diazomethane, was 64.8%. Recrys-
tallisation of the salt from methanol raised the enan-
tiomericexcessof(R)-2to92.0%loweringtheoverallyield
to 60.8%. Finally, (R)-2 was liberated and isolated in
60.2% yield (30.1% of the starting 2) (Table 2, entry 2).
Contrary to the preparation of 2, that of the acid 3
immediately appeared quite laborious due to the poor
availability of 1,2-naphthalic anhydride and also to the
expected formation of the positional isomer of 3, iso-
propylidene glycerol 2-carboxy-1-naphthoate, from the
reaction of the same anhydride with glycerol acetonide.
Nevertheless, the synthesis was undertaken. A few grams
of 1,2-naphthalic anhydride were prepared from a-bro-
mostyrene and maleic anhydride according to the proce-
dure reported in the literature.10 In order to assess the
resolving properties of 3 immediately, the preparation of
its racemate was omitted and the trial reactions of
1,2-naphthalic anhydride with isopropylidene glycerol
directly performed using the optically active alcohol.
Independentlyonthetemperature, treatmentof1,2-naph-
thalic anhydride with (R)-isopropylidene glycerol in
pyridine afforded an approximately 2:1 mixture of (S)-3
and (S)-isopropylidene glycerol 2-carboxy-1-naphthoate.
The latter could be completely removed by one crystalli-
sation of the mixture from toluene. Isolated (S)-3, which
is a high melting point white solid, was then mixed with
pure samples of (S)-4 and (R)-4. The two diastereomeric
salts, in contrast to those of (S)-2 with (S)-4 and (R)-4,
were unfortunately amorphous and precipitation of solids
did not occur from their alcoholic solutions.
A second attempt at resolving 2 with 4 was made
increasing the methanol volume and replacing the recrys-
tallisation of the first precipitate from this solvent with
two successive treatments in refluxing ethanol (Table 2,
entry 3). Furthermore, in order to obtain both (R)-2 and
(S)-2, (R)- and (S)-4 were used in sequence. At the end
of the procedure, (R)-2 and (S)-2 were recovered with 96.1
and 98.8% e.e. and in 48.0 and 48.4% yield, respectively.
Successive saponification allowed the recovery, in near
quantitative yield, of naphthalene 2,3-dicarboxylic acid
and the enantiomers of isopropylidene glycerol with
enantiomeric purities consistent with the enantiomeric
excesses of the respective parent hemiesters.
In Table 2, the results of all the above resolutions,
including the respective experimental efficiencies S (yield×
e.e.), are compared with the data previously reported for
the reciprocal resolution between 1 and 4 (entries 4–6).1,5
Finally, prompted by the ready availability and cheapness
of 1,8-naphthalic anhydride, we contemplated the prepa-
ration of isopropylidene glycerol 8-carboxy-1-naph-
thoate. However, such a project was promptly dropped
considering the intrinsic instability of the hemiesters of
1,8-naphthalic acid.
Binary melting-point phase diagrams were accurately
determined for the four diastereomeric systems (S)-2·(S)-
4/(S)-2·(R)-4, (S)-2·(S)-4/(R)-2·(S)-4, (S)-1·(S)-4/(S)-
1·(R)-4 and (S)-1·(S)-4/(R)-1·(S)-4 (see Fig. 1) using the

2492
M. Palla6icini et al. / Tetrahedron: Asymmetry 12 (2001) 2489–2495
DSC data recorded for a number of different mixtures,
which were prepared from the salts obtained by
combining the single enantiomers of the amine and the
hemiester and recrystallising the resultant precipitates
from methanol. Two clearly distinct peaks characterised
the DSC melting profiles of the 1:1 diastereomeric
mixtures and of those with composition near to such a
ratio. Therefore, on the basis of the T value corre-
sponding to the first peak, the melting points for the
eutectics (Teu) could be easily established (see Fig. 1).
Using the Schro¨der–van Laar equation, the right
branches of the diagrams, the solid curves represented
in Fig. 1, were calculated from the melting points of the
two less soluble salts, i.e. (S)-2·(S)-4 and (S)-1·(S)-4,
and from the fusion enthalpies of these latter (161.8 and
145.8 kJ mol−1, respectively).11 As can be seen, the two
diastereomeric systems (S)-2·(S)-4/(S)-2·(R)-4 and (S)-
2·(S)-4/(R)-2·(S)-4 form conglomerates, characterised
by identical diagrams with 0.10 eutectical composition
(eu). The other two systems, (S)-1·(S)-4/(S)-1·(R)-4
and (S)-1·(S)-4/(R)-1·(S)-4, also form conglomerates
with identical diagrams, but the eutectic corresponds to
a sensibly higher molar ratio of (S)-1·(S)-4 ( 0.18 eu).
a position of the carboxyl, even when ortho to the ester
function, was unfavourable to the achievement of crys-
talline structures and both the diastereomeric salts of
(S)-3 with (S)- and (R)-4 proved to be amorphous.
Comparison of (S)-2·(S)-4 salt with (S)-1·(S)-4 shows
that the former is more stable, as proved by the higher
mp and heat of fusion. Furthermore, the solubility and
mp differences between (S)-2·(S)-4 and its (S)-2·(R)-4
diastereomer are much more pronounced than between
the two respective diastereomeric salts of (S)-1 (Table
1). Finally, as shown by the binary melting-point phase
diagrams (Fig. 1), both the diastereomeric systems (S)-
2·(S)-4/(S)-2·(R)-4 and (S)-1·(S)-4/(S)-1·(R)-4 form
conglomerates, but, in the case of the former, the
eutectic composition lies sensibly closer to the end of
the diagram (0.10 versus 0.18 eu). All these data are
consistent with the fact that 2 exhibited a higher resolv-
ing ability (S=0.88) than 1 (S=0.81), as shown in
Table 2. Both the experimental resolution efficiencies,
0.88 and 0.81, are higher than the theoretical ones,
calculated from kn (solubility of the more soluble n-salt)
and the initial concentrations of the 1:1 diastereomeric
mixtures (C0), and very close to the maximum values,
0.89 and 0.78, calculated from the eutectical composi-
tion or, alternatively, from the equation S=(kn−kp)/kn,
where kp is the solubility of the less soluble p-salt (see
Table 3). Therefore, this indicates that the two experi-
mental values of S are meaningful and the two resolu-
tion procedures of 4 with 2 and 1 were carried out
under optimal conditions. Furthermore, the perfect
agreement of the maximum theoretical efficiencies cal-
culated from the eutectical compositions with those
calculated from (kn−kp)/kn can be assumed as an indi-
rect mutual validation between the measurements of the
solubility differences and the DSC data.
In Table 3, the theoretical values of resolution
efficiency, calculated from both the eutectic composi-
tions and the observed solubilities of the diastereomeric
salts in methanol, are compared to those experimentally
obtained for the two reciprocal resolutions.
3. Discussion
Replacement of the phenyl group of 1 with a naphthyl
group produced the desired effects, i.e. more suitable
characteristics for the use as a resolving agent. In fact,
both 2 and 3 are high melting solids, more efficiently
recoverable than 1 by water/organic solvent extraction
and easy to purify by crystallisation. As expected, such
properties facilitated dosing and handling through the
resolution procedures. However, of the two carboxy-
naphthoates of isopropylidene glycerol, only 2 formed
crystalline salts with 4. In the case of naphthoate 3, the
In the case of the two reciprocal resolutions, i.e. those
of 2 and 1 with 4, experimental S ranged from 0.54 to
0.59 (see Table 2), i.e. very far from the calculated
theoretical values (see Table 3). These results were not
expected, since the eutectic compositions (see Fig. 1)
Table 3. Theoretical and experimental values of S (resolution efficiency) for the reciprocal resolutions between 2 and 4 and
between 1 and 4 in methanol
Substrate
Resolving agent
Sa
Sb
Sc
Sd
4
2
4
1
(S)-2
(R)-4
(S)-1
(S)-4
0.89
0.89
0.78
0.78
0.89
0.89
0.78
0.78
0.73
0.86, 0.75
0.66
0.88e
0.58f, 0.54g
0.81h
0.71
0.57i
a Calculated from S=(1−2eu)/(1−eu).
b Calculated from S=(kn−kp)/kn, where kn and kp are the solubilities in methanol of the more soluble n-salt and the less soluble p-salt, respectively.
c Calculated from S=(0.5C0−kp)/0.5C0, when 0.5C0<kn, or from S=(kn−kp)/0.5C0, when 0.5C0>kn. C0 is the initial concentration (of the 1:1
diastereomeric mixture) adopted in the experiment in question.
d Experimental, i.e. calculated from yields and enantiomeric excesses.
e See Table 2, entry 1.
f Calculated from S=88.9×64.8/10000 (see Table 2, entry 2). The corresponding theoretical value in the adjacent column is 0.86.
g See Table 2, entry 3. The corresponding theoretical value in the adjacent column is 0.75.
h See Table 2, entry 4.
i See Table 2, entry 5.

M. Palla6icini et al. / Tetrahedron: Asymmetry 12 (2001) 2489–2495
2493
and, obviously, the solubility differences coincided with
those previously determined for the diastereomeric salts
involved in the resolutions of 4 with 2 and 1. In order
to verify that such behaviour was not attributable to
the use of crude racemic acids as substrates, trial resolu-
tions were carried out using previously recrystallised 2
or, in the case of 1, dissolving in methanol equimolar
amounts of crystallised (S)-1·(S)-4 and (R)-1·(S)-4.
This notwithstanding, the experimental efficiencies did
not improve, thus showing the existence of markedly
unbalanced reciprocities between the two pairs of pro-
cedures, invariably in favour of the resolutions of 4
over those of 2 and 1. Such results would be better
evaluated knowing the ternary phase diagrams of the
four diastereomeric systems. Actually, we can underline
that the reciprocity of the present resolutions, though
largely imperfect, is proved and well supported by the
identity of the binary diagrams, showing that reciproc-
ity does not necessarily imply identical or similar
efficiencies, since reciprocal resolutions are not mirror-
image related and thus not subjected to the Marckwald
principle.
Starting materials. Achiral and racemic reagents were
obtained from commercial suppliers and used without
further purification. (R)- and (S)-isopropylidene glyc-
erol, with 99.4 and 99.6% e.e., respectively, and their
immediate precursors (S)-1 and (R)-1 were prepared by
Chemi S.p.a. according to the resolution methods previ-
ously reported.1,5,6 (S)-4 and (R)-4 were purchased
from Aldrich Chemical Co. 1,2-Naphthalic anhydride
and 2,3-naphthalic anhydride were prepared according
to literature procedures.8,10
5.1. Isopropylidene glycerol 3-carboxy-2-naphthoate 2
A mixture of 2,3-naphthalic anhydride (29.9 g, 151
mmol) and racemic isopropylidene glicerol (19.9 g, 151
mmol) in pyridine (20 mL) was stirred at 90°C for 1 h.
After cooling, ethyl acetate and 2N H2SO4 were added.
The organic phase was separated, washed with satu-
rated aq. NaCl and then with water, dried and concen-
trated to give 2 (49.7 g, 100%) as a viscous oil, which
turned into a beige solid upon standing overnight at
room temperature: 1H NMR (DMSO-d6) l 1.27 (s, 3H),
1.32 (s, 3H), 3.79 (dd, 1H, J=8.8, 5.9 Hz), 4.07 (dd,
1H, J=8.8, 6.6 Hz), 4.21–4.38 (m, 3H), 7.67–7.72 (m,
2H), 8.0.7–8.14 (m, 2H), 8.27 (s, 1H), 8.41 (s, 1H),
13.32 (br s, 1H).
4. Conclusions
By replacing the phenyl of isopropylidene glycerol
hydrogen phthalate 1 with a naphthyl group, we have
obtained a novel resolving agent, the 3-carboxy-2-naph-
thoate 2, which resolves 1-phenylethylamine 4 even
more efficiently than 1. The new hemiester has solid
consistency and more suitable properties for such a
specific use. Its resolution with 4 takes place with
moderate efficiency, similar to that exhibited by the
same amine in resolving 1.
5.2. 3-Carboxy-2-naphthoate of (R)-isopropylidene
glycerol (S)-2
Prepared in quantitative yield from 2,3-naphthalic
anhydride and (R)-isopropylidene glycerol as described
for 2. Crystallisation of crude (S)-2 from toluene gave
(S)-2 as a white solid in 75.4% yield: mp 106.0–107.4°C;
1
[h]2D5 =−12.9 (c 1, ethanol). H NMR identical to 2.
Apart from theoretical considerations on the reciprocity
of these resolutions, we finally wish to point out the
uncommon resolving ability of 2, which is worthy to be
further explored testing other amines. On the other
hand, our previously described resolution of 1 with 4,1,5
though not more efficient than the present resolution of
2 with 4, remains preferable to the latter in order to
obtain, after saponification, (R)- and (S)-isopropyli-
dene glycerol, considering that phthalic anhydride is
much more available than 2,3-naphthalic anhydride.
5.3. 3-Carboxy-2-naphthoate of (S)-isopropylidene
glycerol (R)-2
Prepared in quantitative yield from 2,3-naphthalic
anhydride and (S)-isopropylidene glycerol as described
for 2. Crystallisation of crude (R)-2 from toluene gave
(R)-2 as a white solid in 58.2% yield: mp 105.3–
106.2°C; [h]2D5=+ 13.3 (c 1, ethanol). 1H NMR identical
to 2.
5.4. 1-Carboxy-2-naphthoate of (R)-isopropylidene
glycerol (S)-3
5. Experimental
A mixture of 1,2-naphthalic anhydride (2.7 g, 13.6
mmol) and (R)-isopropylidene glycerol (1.8 g, 13.6
mmol) in pyridine (5 mL) was stirred at 90°C for 1 h.
After working up as described for 2, an approximately
2:1 solid mixture (3.9 g) of (S)-3 and 2-carboxy-1-naph-
thoate of (R)-isopropylidene glycerol was obtained.12
Crystallisation from toluene allowed the minor compo-
nent to be completely removed from the mixture and 2
g of the prevalent isomer (S)-3 to be isolated as a white
1H NMR spectra were recorded on a Varian Gemini
300 (300 MHz) instrument. Melting points were
recorded on a Bu¨chi Melting Point B-450 apparatus
and are uncorrected. DSC curves were recorded with a
Mettler DSC 20 instrument and processed with a Met-
tler TC 10 A processor. Optical rotations were mea-
sured in a 1 dm cell of 1 mL capacity using a
Perkin–Elmer 241 polarimeter. HPLC analyses were
performed on Chiralcel and Crownpack columns from
Daicel using a Waters 510 pump, a Hitachi L-7400 UV
detector and a Hitachi D-7000 HPLC System Manager
software.
1
solid: mp 142.3–143.4°C; H NMR (CDCl3) l 1.38 (s,
3H), 1.45 (s, 3H), 3.90 (dd, 1H), 4.17 (dd, 1H), 4.50–
4.60 (m, 3H), 7.57–7.69 (m, 2H), 7.88–8.00 (m, 3H),
8.09 (d, 1H).

2494
M. Palla6icini et al. / Tetrahedron: Asymmetry 12 (2001) 2489–2495
5.5. Resolution of 1-phenylethylamine 4 with 3-carboxy-
e.e. of (R)-2 92.0% (64.8%, before recrystallisation)
[determined on the corresponding methyl ester, pre-
pared by treatment of a liberated sample of (R)-2 with
diazomethane, by HPLC on a Chiralcel OD column,
using a mixture of n-hexane and propan-2-ol (96:4) as a
2-naphthoate of (R)-isopropylidene glycerol (S)-2
Compound 4 (10.48 g, 86.5 mmol) was added to a
stirred solution of (S)-213 (28.57 g, 86.5 mmol) in
methanol (80 mL). A white precipitate formed in a few
minutes. The mixture was boiled for 5 min and then
allowed to cool to 20°C. After 90 min, the precipitate
was collected by filtration, rinsed with cold methanol
and dried yielding (S)-2·(S)-4 salt as a white crystalline
solid (17.52 g, 90% of the theoretical amount): mp
1
mobile phase (flow-rate 1.5 mL/min)]; H NMR identi-
cal to (S)-2·(S)-4. The salt was suspended in ethyl
acetate and 2N H2SO4. The organic phase was sepa-
rated, washed with water, dried over Na2SO4, and
concentrated to give (R)-2 (15.05 g, 60.2% of the theo-
retical amount) as a white solid: e.e. 92.0% (by chiral
HPLC of the methyl ester under the above conditions);
1H NMR identical to 2.
1
154.4–155.6°C; H NMR (DMSO-d6) l 1.25 (s, 3H),
1.31 (s, 3H), 1.49 (d, 3H, J=7.1 Hz), 3.82 (dd, 1H,
J=8.4, 6.2 Hz), 4.03 (pseudo t, 1H, J=8.4 Hz), 4.20 (d,
2H, J=5.5 Hz), 4.32–4.37 (m, 2H), 7.29–7.39 (m, 3H),
7.49–7.58 (m, 4H), 7.95–7.99 (m, 3H), 8.25 (s, 1H), 8.81
(br s, 3H). The salt was decomposed by treatment with
2N H2SO4 and ethyl acetate. The aqueous phase was
separated, made alkaline by addition of sodium
hydroxide and extracted with ethyl acetate. Removal of
the solvent from the extract, previously dried over
Na2SO4, gave an oily residue, which was distilled under
vacuum yielding (S)-4 (3.9 g, 74.4% of the theoretical
amount): [h]2D0=−39.7 (neat); e.e. 98.3% (by HPLC on a
Crownpack CR (+) column; pH 1.5 HClO4 aq., 0.8
mL/min at 200 nm); 1H NMR identical to that reported
in the literature.14
(R)-2 was saponified by treatment with 2.5N NaOH (60
mL) at 90°C for 1 h. After cooling to room tempera-
ture, the reaction mixture was extracted with ethyl
acetate. Removal of the solvent from the extract, previ-
ously dried over Na2SO4, gave an oily residue, which
was distilled under vacuum yielding (S)-isopropylidene
glycerol (5.44 g, 54.5% of the theoretical amount):
1
h2D0=+13.47 (neat); H NMR identical to that reported
in the literature.14 Addition of 2N H2SO4 to the
aqueous phase resulted in the immediate precipitation
of a white solid, which was isolated and dried to give
pure naphthalene 2,3-dicarboxylic acid (8.91 g, 89.6%
on the basis of the amount of (R)-2·(R)-4 salt submit-
ted to decomposition).
The methanolic filtrate resulting from the isolation of
(S)-2·(S)-4 was concentrated to give (S)-2·(R)-4 salt
(21.24 g, 108.8% of the theoretical amount). Decompo-
sition of the latter by the procedure described for the
(S)(S)-salt and successive distillation of the recovered
crude amine afforded 4.3 g (82.0% of the theoretical
amount) of (R)-4: [h]2D0=+33.4 (neat); e.e. 80.0% (by
HPLC under the conditions reported for the (S)-
isomer).
Method B: by crystallisations from methanol and from
ethanol. (R)-4 (17.36 g, 143.3 mmol) was added to a
stirred solution of 2 (47.34 g, 143.3 mmol) in methanol
(380 mL). Precipitation occurred after about 30 min.
The mixture was stirred at room temperature overnight
and then cooled to 5°C. The precipitate was collected
by filtration, rinsed with cold methanol, and dried
yielding a white solid (24.76 g), which was suspended in
ethanol (200 mL). The suspension was boiled with
stirring for 90 min, cooled to 20°C, and filtered. The
isolated solid (18 g) was submitted to the same treat-
ment again, but using 140 mL of ethanol. Filtration
allowed isolation of (R)-2·(R)-4 salt (15.81 g, 48.9% of
the theoretical amount) as a white crystalline solid:
154.7–155.4°C; e.e. of (R)-2 96.1% (70.7%, after the
crystallisation from methanol, and 90.4% after the first
treatment in refluxing ethanol) (determined as described
The two organic phases remaining after extraction of
(S)-4 and (R)-4 with 2N H2SO4 were combined, washed
with water several times, dried over Na2SO4 and con-
centrated recovering 26.68 g (93.4% of the starting
amount) of (S)-2 as a viscous oil, which solidified upon
standing. NMR spectrum of such a solid confirmed the
identity with the starting acid revealing only traces of
the deketalised derivative, which could be easily
removed by recrystallisation from toluene.
1
in method A); H NMR identical to (S)-2·(S)-4.
5.6. Resolution of isopropylidene glycerol 3-carboxy-2-
The methanolic mother liquors were concentrated to
give 38.86 g of a solid, whence 28.43 g of (S)-2 (45.6%
e.e.) were liberated upon ethyl acetate/2N H2SO4
extraction. Treatments with equimolar (S)-4 in
methanol (230 mL) and then with boiling ethanol twice
(150 and 135 mL) under the same conditions as above
led to the ultimate isolation of (S)-2·(S)-4 salt (16.0 g,
49.5% of the theoretical amount) as a white crystalline
solid: mp 153.5–154.6°C; e.e. of (S)-2 98.8% (93.4%,
after the crystallisation from methanol, and 97.2% after
the first treatment in refluxing ethanol) (determined as
described in method A).
naphthoate 2 with (R)-1-phenylethylamine (R)-4
Method A: by crystallisations from methanol. (R)-4
(18.3 g, 151 mmol) was added to a stirred solution of 2
(50 g, 151 mmol) in methanol (220 mL). A white
precipitate was formed in a few minutes. The mixture
was boiled for 10 min and then allowed to cool to
20°C. After 90 min, the precipitate was collected by
filtration, rinsed with cold methanol, and dried yielding
a white solid (30.37 g), which was suspended in
methanol (200 mL). The mixture was stirred under
reflux for 30 min, cooled to 20°C, and filtered to give
(R)-2·(R)-4 salt (20.75 g, 60.8% of the theoretical
amount) as a white crystalline solid: mp 149.6–150.2°C;
From the two resolved diastereomeric salts, (R)-2 and
(S)-2 were liberated in 48.0 and 48.4% yield, respec-

M. Palla6icini et al. / Tetrahedron: Asymmetry 12 (2001) 2489–2495
2495
tively. Saponification of (R)-2 and (S)-2 gave (S)- and
(R)-isopropylidene glycerol with +14.13 and −14.40 hD20
(neat) and in 47.2 and 43.8% yield (relative to half of
starting 2), respectively.
3. Pallavicini, M.; Valoti, E.; Villa, L.; Piccolo, O. Tetra-
hedron: Asymmetry 2000, 11, 4017.
4. Pallavicini, M.; Valoti, E.; Villa, L.; Piccolo, O. Tetra-
hedron: Asymmetry 2001, 12, 1071.
5. Pallavicini, M.; Valoti, E.; Villa, L.; Piccolo, O. Tetra-
hedron: Asymmetry 1994, 5, 5.
6. Piccolo, O.; Pallavicini, M.; Valoti, E.; Villa, L. It. Pat.
Appl. 000232a/92 1992, applicant Chemi/Italfarmaco Sud
S.p.a. (Italia).
7. Kinbara, K.; Harada, Y.; Saigo, K. J. Chem. Soc., Perkin
Trans. 2 2000, 1339.
8. Gandolfi, C.; Beggiolin, G.; Menta, E.; Palumbo, M.;
Sissi, C. J. Med. Chem. 1995, 38, 526.
From the combined mother liquors of all the crystalli-
sations, excepting the initial one, isopropylidene glyc-
erol was isolated in near racemic form (h2D0=+0.4) and
in 22.8% yield (relative to the starting 2). Naphthalene
2,3-dicarboxylic acid was recovered from the saponifi-
cation steps in 85.0% overall yield (relative to the
starting 2).
DSC analyses. Samples of 3–5 mg were run in sealed
aluminium pans with a heating rate of 2.5 K min−1. The
samples were drawn from diastereomeric mixtures of
(S)-2·(S)-4, (S)-2·(R)-4, (R)-2·(S)-4, (S)-1·(S)-4, (S)-
1·(R)-4 and (R)-1·(S)-4. These salts were prepared
combining, in equimolar amounts, the enantiomer of
the amine with that of the hydrogen ester in ethyl ether
and crystallising the resultant precipitate from
methanol.
9. Pallavicini, M.; Valoti, E.; Villa, L.; Piccolo, O.;
Marchetti, F. Tetrahedron: Asymmetry 2000, 11, 1957.
10. Newman, M. S.; Dhawan, B.; Hashem, M. M.; Khanna,
V. K.; Springer, J. M. J. Org. Chem. 1976, 41, 3925.
11. The value of 69.4 kJ mol−1, previously reported for
(S)-1·(S)-4 in Ref. 9, resulted from an improper integra-
tion of the melting peak in the DSC trace leading to
underestimation of the area.
12. Such a ratio was established on the basis of the relative
integrations of the two distinct pairs of singlets found in
the 1H NMR spectrum of the mixture for the geminal
methyl residues. The two respective identities were
assigned on the basis of the following observations: (i) the
predominant isomer exhibited a higher Rf on silica gel (el.
Acknowledgements
This work was supported by the Ministero dell’Univer-
sita` e della Ricerca Scientifica e Tecnologica of Italy.
Thanks are due to Chemi S.p.a. (Cinisello Balsamo,
MI, Italy) for providing the necessary amounts of (R)-
and (S)-isopropylidene glycerol and of (S)- and (R)-1.
1
cyclohexane/ethyl acetate 4:6); (ii) in the H NMR spec-
trum, the doublet of its proton in position 8 was found at
a slightly lower field than the corresponding proton of the
other isomer; (iii) formation of 1-carboxy-2-naphthoate
was expected to be favoured over that of 2-carboxy-1-
naphthoate.
References
13. Crude, not crystallised from toluene, i.e. resultant from
working up the reaction of 2,3-naphthalic anhydride with
(R)-isopropylidene glycerol.
1. Pallavicini, M.; Valoti, E.; Villa, L.; Piccolo, O. Tetra-
hedron: Asymmetry 1996, 7, 1117.
2. Pallavicini, M.; Valoti, E.; Villa, L.; Piccolo, O. Tetra-
14. Pouchert, C. J.; Behnke, J. The Aldrich Library of 13C and
1H FT NMR Spectra, Edition I.
hedron: Asymmetry 1997, 8, 1069.