Crystallization of chiral compounds 3. 3-Phenoxypropane-1,2-diol and 3-(2-halophenoxy)propane-1,2-diols

Article abstract of DOI:10.1007/s11172-006-0243-x

Specific features of melting of crystalline samples of 3-(2-R-phenoxy) propane-1,2-diols with different enantiomeric compositions were studied by differential scanning calorimetry. The melting points and enthalpies of melting for the racemate and individual stereoisomers were determined. Binary phase diagrams were constructed. The entropy of mixing of individual enantiomers in the liquid phase and the free energy of formation of the racemic compound were calculated. The thermochemical data indicate that the racemates are formed upon the crystallization of phenoxy-and 2-fluorophenoxy-containing compounds, while crystallization of the chloro-, bromo-, and iodo-substituted analogs would form racemic conglomerates.

Full text of DOI:10.1007/s11172-006-0243-x

Russian Chemical Bulletin, International Edition, Vol. 55, No. 2, pp. 230—237, February, 2006
230
Crystallization of chiral compounds
3.* 3ꢀPhenoxypropaneꢀ1,2ꢀdiol and 3ꢀ(2ꢀhalophenoxy)propaneꢀ1,2ꢀdiols
D. V. Zakharychev, S. N. Lazarev, Z. A. Bredikhina, and A. A. Bredikhinꢀ
A. E. Arbuzov Institute of Organic and Physical Chemistry, Kazan Research Center, Russian Academy of Sciences,
8 ul. Akad. Arbuzova, 420088 Kazan, Russian Federation.
Fax: +7 (843) 273 2253. Eꢀmail: baa@iopc.knc.ru
Specific features of melting of crystalline samples of 3ꢀ(2ꢀRꢀphenoxy)propaneꢀ1,2ꢀdiols
with different enantiomeric compositions were studied by differential scanning calorimetry.
The melting points and enthalpies of melting for the racemate and individual stereoisomers
were determined. Binary phase diagrams were constructed. The entropy of mixing of individual
enantiomers in the liquid phase and the free energy of formation of the racemic compound
were calculated. The thermochemical data indicate that the racemates are formed upon the
crystallization of phenoxyꢀ and 2ꢀfluorophenoxyꢀcontaining compounds, while crystallization
of the chloroꢀ, bromoꢀ, and iodoꢀsubstituted analogs would form racemic conglomerates.
Key words: 3ꢀ(2ꢀhalophenoxy)propaneꢀ1,2ꢀdiols, crystallization, racemic conglomerates,
thermal analysis, differential scanning calorimetry, phase diagrams, IR spectroscopy.
In the case of chiral compounds, the properties of
neous resolution of enantiomers during crystallization. In
the structural respect, halogen atoms (including a hydroꢀ
gen atom) in the role of substituents are similar, being
spheres with the successively increasing van der Waals
radii (1.2 (H), 1.35 (F), 1.80 (Cl), 1.95 (Br), 2.15 Å (I)).
In the present work, we studied specific features of crysꢀ
tallization of of phenyl ethers of glycerol 1a—e containing
the aboveꢀlisted substituents in the orthoꢀposition of the
aromatic ring. All substances in the racemic and scalemic
forms were synthesized via the reaction of the correspondꢀ
ing phenol with racemic or scalemic glycidol (Scheme 1).
substances and related materials depend substantially on
the enantiomeric composition. Useful properties are usuꢀ
ally manifested themselves only when samples with high
enantiomeric purity are used. Despite of success achieved
in enantioselective synthesis, biotechnology, and preparaꢀ
tive chromatography using chiral phases, a considerable
(more than 50%) amount of enantiomerically pure subꢀ
stances is produced in industrial scales by the resolution
of racemates into individual enantiomers. In this case,
methods based on fractional crystallization are prevailing.
Among the known methods for racemate resolution, methꢀ
ods based on spontaneous separation during crystallizaꢀ
tion (spontaneous resolution) play a principal role. This
phenomenon2 provides racemate separation without auxꢀ
iliary resolving reagents, chiral phases, etc. and/or expenꢀ
sive specialized equipment.3,4 A necessary prerequisite
for the use of this approach is crystallization of a subꢀ
stance from solution or melt as a racemic conglomerate,
i.e., a mechanical mixture of single crystals, each of which
is formed by molecules of one configuration only.
Scheme 1
The phenomenon of spontaneous resolution has been
known for more than 150 years but remains poorly studꢀ
ied, which is caused, first of all, by an insufficient number
of the objects investigated. This especially concerns the
series of structurally similar molecules in which a small
number of variables changes uniformly with substituent
replacement. We have recently5 found that 3ꢀ(2ꢀchloroꢀ
phenoxy)propaneꢀ1,2ꢀdiol exhibits a property of spontaꢀ
R = H (a), F (b), Cl (c), Br (d), I (e)
The main method of investigation in this work is difꢀ
ferential scanning calorimetry (DSC). The use of this
method makes it possible to substantially restrict arbiꢀ
trariness of measurements of the temperature characterisꢀ
tics of sample melting and to determine quantitatively the
* For Part 2, see Ref. 1.
Published in Russian in Izvestiya Akademii Nauk. Seriya Khimicheskaya, No. 2, pp. 225—232, February, 2006.
1066ꢀ5285/06/5502ꢀ0230 © 2006 Springer Science+Business Media, Inc.

Crystallization of chiral aryloxypropanediols
Russ.Chem.Bull., Int.Ed., Vol. 55, No. 2, February, 2006
231
talline mixture. A precipitate that formed was filtered off and
multiply recrystallized from an appropriate solvent. The physiꢀ
cochemical characteristics of the isolated diols are given below.
racꢀ3ꢀPhenoxypropaneꢀ1,2ꢀdiol (1a). The yield was 92%,
m.p. 48—52 °C (from ether). A sample of racꢀ1a was prepared
by the successive recrystallization of the material from ether and
CCl4 followed by the slow evaporation of the solvent from a diol
solution in CCl4 for 1 month, m.p. 56—60 °C. A sample with
m.p. 60—74 °C was obtained by the further recrystallization of
enthalpies of thermally initiated processes. Based on this
quantitative information, one can compare binary phase
diagrams of melting constructed experimentally and calꢀ
culated in different theoretical approximations. Finally,
the use of DSC allows one to analyze the thermodynamic
characteristics, which are not directly determined in exꢀ
periments.
1
racꢀ1a from a hexane—ether (1 : 1) mixture. H NMR, δ: 7.29
Experimental
(td, 2 H, H(3´), H(5´), J = 7.4 Hz, J = 1.4 Hz); 6.98 (t, 1 H,
H(4´), J = 7.4 Hz); 6.91 (dd, 2 H, H(2´), H(6´), J = 8.4 Hz, J =
1.0 Hz); 4.15—4.08 (m, 1 H, CHOH); 4.04 (s, 1 H, CH2); 4.03
(d, 1 H, CH2, J = 2.1 Hz); 3.84 (dd, 1 H, CH2OH, J = 11.3 Hz,
J = 3.8 Hz); 3.75 (dd, 1 H, CH2OH, J = 11.7 Hz, J = 5.5 Hz);
2.55 (br.s, 2 H, 2 OH) (see Ref. 9). 13C NMR, δ: 158.55
1
H and 13C NMR spectra were recorded on Bruker MSLꢀ400
spectrometer (1H, 400 MHz; 13C, 100.6 MHz) using CDCl3 as
solvent, and Me4Si or signals of the solvent served as the internal
standard. Optical rotation was measured on a Perkin—Elmer 341
polarimeter. Specific rotation is given in deg mL g–1 dm–1, and
the concentration of solutions appears in g (100 mL)–1
For usual purposes, melting points were determined on the
Boetius heating stage with visual monitoring.
1
2
(s, C(1´)); 129.70 (dd, C(2´), C(6´), JC,H = 158.4 Hz, JC,H
=
.
1
2
8.7 Hz); 121.48 (dt, C(4´), JC,H = 161.3 Hz, JC,H = 8.0 Hz);
114.71 (dm, C(3´), C(5´), 1JC,H = 155.5 Hz); 70.57 (d, C(2), J =
1
145.3 Hz); 69.28 (t, C(3), JC,H = 144.6 Hz); 63.83 (t, C(1),
Samples for DSC analysis were prepared by multiple recrysꢀ
tallization and dried in a Fischer apparatus in a vacuum of an oil
pump at temperatures by ~5 °C lower than the melting point.
Chemical purity of samples was monitored by TLC (silica gel,
Silufol plates) and GCꢀMS (Finnigan MAT212 GCꢀMS specꢀ
trometer, column (50 m) with the SEꢀ54 phase, injector temꢀ
perature 240 °C, thermostat temperature 100—240 °C, heating
rate 6 °C min–1). IR spectra of racemic and enantiopure crystalꢀ
line samples in KBr pellets were recorded on a Bruker Vector 22
spectrophotometer.
Melting curves of samples of 3ꢀ(2ꢀRꢀphenoxy)propaneꢀ1,2ꢀ
diols (~2 mg) were obtained on a Setaram DSC111 upgraded
calorimeter. The calorimeter was equipped with unique systems
for controlling a heater and compensating the baseline on the
basis of precision analogꢀtoꢀdigital converters,6 which increased
the effective sensitivity and dynamic range of the instrument.
The heating rate was 1 °C min–1. Temperature and thermal flux
measurements were calibrated by the data for corundum, pheꢀ
nol, and naphthalene. Thermograms were processed by numeriꢀ
cal methods using the Mathcad program.* This made it possible
to write the processing procedures as standard mathematical
formulas, document all procedures with experimental data, and
verify and correct experimental data. Discontinuous arrays of
experimental data were interpolated by cubic splines and further
treated as continuous differentiated and integrated functions,
which substantially simplified the procedure of subtraction of
the baseline, multiplication by the calibration function, deꢀ
convolution, and integration.
1JC,H = 143.9 Hz).
(R)ꢀ3ꢀPhenoxypropaneꢀ1,2ꢀdiol ((R)ꢀ1a). The yield was 70%,
20
m.p. 67—69 °C (ether—hexane (1.3 : 1)), [α]D –9.7 (c 1,
20
EtOH), [α]D –8.2 (c 0.5, MeOH) (cf. Ref. 10: m.p.
25
62.5—64.5 °C, [α]D –9.5 (c 0.5, MeOH), ee 98.9%, Ref. 11:
20
m.p. 56.5—57 °C (EtOH, ee 88%), Ref. 12: [α]D –10.8 (c 1,
20
EtOH), ee 98%; for (S)ꢀ1a [α]D +10.2 (c 1, EtOH), ee 91%).
The 1H and 13C NMR spectra are similar to those published
earlier.10
racꢀ3ꢀ(2ꢀFluorophenoxy)propaneꢀ1,2ꢀdiol (racꢀ1b). The yield
was 70%, m.p. 46—56 °C (ether; ether—hexane) (cf. Ref. 13:
m.p. 56—57 °C (C6H6—petroleum ether)). 1H NMR, δ:
7.09—6.80 (m, 4 H, Ar); 4.15—4.07 (m, 1 H); 4.04, 4.03—3.95
(s and m, 3 H); 3.81 (dd, 1 H, CH2OH, J = 11.5 Hz, J =
3.3 Hz); 3.72 (dd, 1 H, CH2OH, J = 11.5 Hz, J = 5.7 Hz);
1
3.66—3.25 (m, 1 H). 13C NMR, δ: 152.7 (d, C(2´), JC,F
=
245.3 Hz); 146.6 (d, C(1´), 1JC,F = 10.6 Hz); 124.3 (ddd, C(5´),
1JC,H = 162.6 Hz, JC,H = 8.8 Hz, JC,F = 3.5 Hz); 121.6 (ddd,
C(4´), 1JC,H = 164.3 Hz, 2JC,H = 8.2 Hz, 3JC,F = 7.0 Hz); 116.2
(ddd, C(3´), 1JC,H = 162.9 Hz, 2JC,H = 8.2 Hz, 2JC,F = 18.8 Hz);
2
4
1
2
115.3 (dd, C(6´), JC,H = 160.2 Hz, JC,H = 8.8 Hz); 70.6 (t,
C(3), 1JC,H = 145.0 Hz); 70.4 (d, C(2), 1JC,H = 143.2 Hz); 63.46
(t, C(1), 1JC,H = 142.6 Hz).
(S)ꢀ3ꢀ(2ꢀFluorophenoxy)propaneꢀ1,2ꢀdiol ((S)ꢀ1b). The yield
was 72%, m.p. 45—57 °C (CHCl3). After several successive
recrystallizations from CCl4 and C6H6 and double recrystalliꢀ
zation from a C6H6—hexane (1 : 1) mixture, a sample of
20
diol (S)ꢀ1b with m.p. 54—60 °C was obtained, [α]D +6.4
(R)ꢀ and (S)ꢀGlycidols (ee 91%) were synthesized by the
Sharpless enantioselective epoxidation7 of allyl alcohol.
Racemic and scalemic diols 1a—e were synthesized by analꢀ
ogy to a published procedure8 from racemic and scalemic
glycidols, respectively. (S)ꢀDiol was obtained from (S)ꢀglycidol,
and (R)ꢀglycidol gave (R)ꢀdiol. A mixture of glycidol (0.015 mol,
1.14 g), the corresponding phenol (0.015 mol), triethylamine
(0.12 mL, 0.86 mmol), and anhydrous ethanol (3.5 mL) was
refluxed for 6 h. After the mixture was cooled down, the solvent
was removed, as a rule, under reduced pressure to leave a crysꢀ
(c 1, EtOH).
racꢀ3ꢀ(2ꢀChlorophenoxy)propaneꢀ1,2ꢀdiol (racꢀ1c). The
yield was 65%, m.p. 70—80 °C (CCl4; CH2Cl2—hexane (1 : 1))
(cf. Ref. 14: m.p. 71—72 °C (petroleum ether—C6H6)).
1H NMR, δ: 7.35 (dd, 1 H, H(6), J = 7.2 Hz, J = 1.4 Hz); 7.20
(td, 1 H, H(8), J = 7.8 Hz, J = 1.2 Hz); 6.92 (d, 1 H, H(9), J =
7.9 Hz); 6.92 (t, 1 H, H(7), J = 7.2 Hz); 4.17—4.05 (m, 3 H,
CH2, CH); 3.90—3.77 (m, 2 H, CH2OH); 3.20 (d, 1 H, CHOH,
J = 4.3 Hz); 2.66 (t, 1 H, CH2OH, J = 5.7 Hz) (cf. Ref. 9).
13C NMR, δ: 153.90 (m, C(1´)); 130.23 (dd, C(3´), JC,H
=
1
164.3 Hz, 2JC,H = 8.7 Hz); 127.82 (dd, C(5´), 1JC,H = 161.2 Hz,
1
* ©Mathsoft Engineering and Education Inc., http://
www.mathsoft.com.
2JC,H = 9.7 Hz); 122.95 (m, C(2´)); 122.04 (dd, C(4´), JC,H
=
163.3 Hz, 2JC,H = 7.6 Hz); 113.76 (dd, C(6´), 1JC,H = 159.7 Hz,

232
Russ.Chem.Bull., Int.Ed., Vol. 55, No. 2, February, 2006
Zakharychev et al.
1
2JC,H = 8.7 Hz); 70.71 (t, C(3), JC,H = 145.2 Hz); 70.10 (d,
similar pairs of aromatic glycerol ethers 1a—e. To subꢀ
stantiate the comparison, the spectra were subjected to a
procedure of normalization and baseline correction. For
this purpose, coefficients that minimize the difference
As – [a0 + a1ν + Ar(a2 + a3ν)], where As and Ar are the
molar absorption coefficients of the scalemic and racemic
samples, respectively; ν is the IR radiation frequency corꢀ
responding to A, and an are the desired regression coeffiꢀ
cients, were selected by the leastꢀsquares method. It is
reasonable to introduce the regression terms a1ν and a3Arν
to correct spectral differences caused by the nonspecific
(not related to particular absorption bands) interaction of
IR radiation with matter (probably, by radiation scatter
on heterogeneities of the sample). Note that the use of
polynomials of higher powers (quadratic and cubic) for
the generation of differential spectra does not improve the
statistical parameters characterizing regression.
Under visual comparison, the spectra of the racemate
and enantiomerically pure crystalline samples of the
two first members of the series 1a,b differ noticeably,
whereas for the three last members (1c—e) they coincide
in detail (Fig. 1). A similar pattern is observed for the
differential curves: in the case of compounds 1c—e, difꢀ
ferences between the spectra of the racemates and enanꢀ
tioenriched samples are at the level of instrumental
background, while they are rather substantial for comꢀ
pounds 1a,b. This agrees with an assumption that racemic
substances are formed upon crystallization of racemic
Hꢀ and Fꢀsubstituted derivatives and racemic conglomerꢀ
ates are formed by the Clꢀ, Brꢀ, and Iꢀsubstituted derivaꢀ
tives, although cannot be considered as the final proof of
this phenomenon.
Useful information on the type of melting of chiral
compounds can be obtained by comparison of melting
points of enantiomerically pure and racemic samples. The
racemic conglomerates have a much lower melting temꢀ
perature. However, the experimental melting points of
the racemic samples of the simplest aromatic glycerol
ethers differ considerably. For instances, the database21
contains 25 references to the melting points of racemic
phenoxy derivative 1a, which overlap the interval from 43
to 70 °C. Three references represent the melting points of
racemic oꢀfluorosubstituted derivative 1b from 37 to 57 °C.
Our experience, which is presented in part in Experimenꢀ
tal, completely confirms this feature. The racemic samples
of diols 1a—e with almost equal chemical purity (GCꢀMS
monitoring found no chemical contamination) melted in
variable but always broad temperature intervals (visual
monitoring). As a rule, the samples subjected to the most
thorough "purification" and consisting of larger, well
formed grains melt in a wider interval shifted to higher
temperatures. This specific feature is not characteristic of
scalemic compounds 1a—e with a high content of one of
the enantiomers, whose samples possess sharp melting
points under visual monitoring.
C(2), 1JC,H = 145.5 Hz); 63.62 (t, C(1), JC,H = 142.9 Hz).
(R)ꢀ3ꢀ(2ꢀChlorophenoxy)propaneꢀ1,2ꢀdiol ((R)ꢀ1c). The
yield was 60%, m.p. 90—92 °C (CCl4; twice from CH2Cl2),
[α]D +13.3 (c 1, hexane—EtOH (4 : 1)) (cf. Ref. 8: m.p.
91.8 °C (EtOH, determined by DSC), Ref. 15 for (S)ꢀ1c: m.p.
87—89 °C, Ref. 12: [α]D +14.0 (c 1, hexane—EtOH (4 : 1)),
ee 99%; for (S)ꢀ1c [α]D –13.4 (c 1, hexane—EtOH (4 : 1)),
ee 99%). The H NMR spectrum is similar to that presented in
1
20
20
20
1
the earlier published work.15
racꢀ3ꢀ(2ꢀBromophenoxy)propaneꢀ1,2ꢀdiol (racꢀ1d). The
yield was 70%, m.p. 84—101 °C (CH2Cl2) (cf. Ref. 16: m.p.
82—83 °C (C6H6—petroleum ether)). 1H NMR, δ: 7.53 (dd,
1 H, H(3´), J = 7.9 Hz, J = 1.7 Hz); 7.26 (td, 1 H, H(5´), J =
7.8 Hz, J = 1.1 Hz); 6.91 (dd, 1 H, H(6´), J = 8.2 Hz, J =
1.4 Hz); 6.87 (td, 1 H, H(4´), J = 7.5 Hz, J = 1.2 Hz);
4.18—4.07 (m, 3 H, CH2, CH); 3.91—3.80 (m, 2 H, CH2OH);
3.05 (d, 1 H, CHOH, J = 4.8 Hz); 2.48 (t, 1 H, CH2OH, J =
6.2 Hz). 13C NMR, δ: 154.70 (m, C(1´)); 133.28 (dd, C(3´),
1JC,H = 164.5 Hz, JC,H = 8.4 Hz); 128.60 (dd, C(5´), JC,H
=
2
1
161.2 Hz, 2JC,H = 8.7 Hz); 122.56 (dd, C(4´), 1JC,H = 163.5 Hz,
2JC,H = 7.9 Hz); 113.58 (dd, C(6´), JC,H = 159.7 Hz, JC,H
=
1
2
1
8.7 Hz); 112.28 (m, C(2´)); 70.82 (t, C(3), JC,H = 145.2 Hz);
70.03 (d, C(2), JC,H = 145.0 Hz); 63.62 (t, C(1), JC,H
1
1
=
142.7 Hz).
(R)ꢀ3ꢀ(2ꢀBromophenoxy)propaneꢀ1,2ꢀdiol ((R)ꢀ1d). The
yield was 56%, m.p. 101—102 °C (CH2Cl2), [α]D +13.8 (c 1,
20
20
hexane—EtOH (4 : 1)), [α]D –9.0 (c 1, CHCl3) (cf. Ref. 17:
22
m.p. 98 °C (PriOH), [α]D –9.7 (c 1.08, CHCl3), ee 100%).
The H and 13C NMR spectra are similar to those presented in
1
Ref. 17.
racꢀ3ꢀ(2ꢀIodophenoxy)propaneꢀ1,2ꢀdiol (racꢀ1e). The yield
was 80%, m.p. 90—92 °C (CH2Cl2). After repeated recrystalliꢀ
zation from the same solvent, the product with m.p. 90—106 °C
was obtained (cf. Ref. 18: m.p. 95 °C (CHCl3)). 1H NMR, δ:
7.77 (dd, 1 H, H(3´), J = 7.9 Hz, J = 1.4 Hz); 7.31 (td, 1 H,
H(5´), J = 7.8 Hz, J = 1.1 Hz); 6.84 (dd, 1 H, H(6´), J =
8.1 Hz, J = 1.3 Hz); 6.75 (td, 1 H, H(4´), J = 7.6 Hz, J =
1.0 Hz); 4.17—4.07 (m, 3 H, CH2); 3.90, (dd, 1 H, CH2OH, J =
11.7 Hz, J = 3.8 Hz); 3.85 (dd, 1 H, CH2OH, J = 11.7 Hz, J =
4.5 Hz); 2.26 (br.s. 2 H, 2 OH). 13C NMR, δ: 156.94 (m,
C(1´)); 139.43 (dd, C(3´), 1JC,H = 165.2 Hz, 2JC,H = 10.0 Hz);
129.66 (dd, C(5´), JC,H = 161.2 Hz, JC,H = 8.6 Hz); 123.33
(dd, C(4´), JC,H = 163.2 Hz, JC,H = 8.0 Hz); 112.65 (dd,
C(6´), JC,H = 159.9 Hz, JC,H = 8.6 Hz); 86.80 (m, C(2´));
70.95 (t, C(3), JC,H = 145.3 Hz); 70.10 (d, C(2), JC,H
143.3 Hz); 63.68 (t, C(1), JC,H = 143.3 Hz).
(S)ꢀ3ꢀ(2ꢀIodophenoxy)propaneꢀ1.2ꢀdiol ((S)ꢀ1e). The yield
was 72%, m.p. 111—113 °C (CH2Cl2), [α]D –15.9 (c 1, hexꢀ
1
2
1
2
1
2
1
1
=
1
20
ane—EtOH (4 : 1)).
Results and Discussion
A known indication for formation of a racemic subꢀ
stance by a chiral compound in the solid phase is discrepꢀ
ancy of vibrational spectra of solid enantiomerically pure
and racemic samples.19,20 We compared the IR spectra of

Crystallization of chiral aryloxypropanediols
Transmission (rel. units)
Russ.Chem.Bull., Int.Ed., Vol. 55, No. 2, February, 2006
233
Transmission (rel. units)
1.0
1.0
1a
1b
1
1
0.5
0.5
0
2
2
3
3
0
500
1000
1500 2000
4000 ν/cm–1 500
1000
1500 2000
4000 ν/cm–1
Transmission (rel. units)
Transmission (rel. units)
1.0
0.5
0
1.0
0.5
0
1c
1d
1
2
1
2
3
3
500
1000
1500 2000
4000 ν/cm–1 500
1000
1500 2000
4000 ν/cm–1
Transmission (rel. units)
1.0
0.5
0
1e
1
2
3
500
1000
1500 2000
4000 ν/cm–1
Fig. 1. IR spectra of solid samples 1a—e: racemate (1), individual enantiomer (2), and differential spectrum (3).
∆Q
Differential scanning calorimetry confirms the differꢀ
ence between the character of melting of the racemates
and enantiomerically pure samples. The experimental
melting curves for compounds 1a—e with high (> 99%)
enantiomeric excess are shown in Fig. 2.
1c
1d
1e
1a
1b
As can be seen from the data in Fig. 2, the melting of
enantiomerically pure diols is described by narrow sharp
curves with a steep linear frontal edge (specific features of
interpretation of the thermographic curves are discussed
below). The typical experimental melting curves for the
highꢀpurity samples of racemic diols 1a,b, which were
obtained directly after recrystallization, are presented in
Fig. 3 for comparison. These thermograms are far from
ideality and represent a broad flat curve without a proꢀ
nounced maximum (see Fig. 3, curve 1) or a polymodal
curve (see Fig. 3, curve 2). These thermograms do not
60
80
100
T/°C
Fig. 2. Melting curves for samples of compounds 1a—e with
high enantiomeric purity.

234
Russ.Chem.Bull., Int.Ed., Vol. 55, No. 2, February, 2006
Zakharychev et al.
∆Q
where x is the mole fraction of one of the enantiomers in
the mixture (the mole fraction of another enantiomer is
x´ = 1 – x); ∆HfА (J mol–1) and T´fA (K) are the enthalpy
of melting and melting point of the pure enantiomers; R is
the universal gas constant (R = 8.3170 J K–1 mol–1).
The phase diagram for a racemic substance is Wꢀshaped
and characterized by two eutectics with the compositions
m : n and n : m. In this case, the liquidus line in the
terminal regions is described by Eq. (1), and its central
fragment obeys the Prigogine—Defay equation,23 which
is usually used in the simplified form
3
4
1
2
40
50
60
70 T/°C
Fig. 3. Melting curves for pure samples of racemic diols: the
starting coarseꢀcrystalline samples racꢀ1a (1) and racꢀ1b (2) and
finely ground samples racꢀ1a (3) and racꢀ1b (4).
,
(2)
allow one to characterize the "melting" of the samples by
a single temperature and a reproducible enthalpy value of
the process.
where ∆H f (J mol–1) and T´f (K) are the enthalpy of
melting and melting point of a racemic compound, and
other designations are the same as in Eq. (1).
R
R
The shape of the experimental DSC curves changed
dramatically (see Fig. 3, curves 3 and 4), when the samples
of racemic diols 1a,b were thoroughly ground before placꢀ
ing into the calorimeter cell. The qualitative change in
the shape of the curves caused by mechanical grinding
suggests that the unusual behavior of racemic glycerol
ethers on melting is caused by a change in the character of
processes accompanying the destruction of the crystal latꢀ
tice upon the interaction with the liquid phase rather than
by changes in the crystal lattice. Although curves 3 and 4
in Fig. 3 retain a substantial width, they can be characterꢀ
ized by a distinct maximum (melting point) and inteꢀ
grated reproducibly (i.e., the enthalpy of the process can
be measured). Therefore, the DSC data for these samples
can be used to construct phase diagrams for melting of
chiral aryloxypropanediols 1.
The heterogeneous equilibrium between condensed
phases in binary systems of enantiomers, which is reꢀ
flected by these diagrams, can be complicated by polyꢀ
morphism (existence of different crystalline phases of the
same composition), temperatureꢀdependent polymorphic
transitions, and partial or complete miscibility of solid
phases (formation of zones of solid solutions).22 It is more
difficult to construct and analyze the phase diagrams beꢀ
cause of this behavior, which, however, is rather rare. In
the most cases, melting/crystallization of a binary mixꢀ
ture of enantiomers can be reduced to two idealized cases:
solidꢀstate formation of a racemic conglomerate or a raꢀ
cemic substance.19,20 In the first case, the binary phase
diagram of melting is Vꢀshaped with the single eutectic
corresponding to the 1 : 1 ratio of enantiomeric compoꢀ
nents. The symmetric liquidus lines of this diagrams are
described by the simplified Schröder—Van Laar equation,
which is usually used for binary systems forming no moꢀ
lecular compounds in the solid phase,19
Thus, in the idealized case, to construct the liquidus
of the phase diagram for melting of a conglomerateꢀformꢀ
ing substance, it is enough to have data on the melting
temperature and enthalpy of melting for an enantioꢀ
merically pure sample. Additional data on the enthalpy
and melting point of a racemate are needed to conꢀ
struct (in a similar approximation) the liquidus of a chiral
substance forming a racemic compound. In both cases,
the theoretical solidus line is a straight line parallel to the
abscissa and passing through the point (points) of eutecꢀ
tic melting. The latter are found as a combined soluꢀ
tion (intersection point) for different branches of the
liquidus line.
The temperature and enthalpy of the thermally initiatꢀ
ing process (in particular, melting) can be determined
experimentally by the DSC method. If melting is deꢀ
scribed by the single peak, then its enthalpy is unambiguꢀ
ously determined by the surface area (integral) of this
peak. The choice of the temperature of the process is
rather arbitrary. In our constructions, we followed the
recommendations of the German Society for Thermal
Analysis (GEFTA)24 and methodological instructions in
Ref. 25. For a pure substance, whose melting is not comꢀ
plicated by additional processes, the melting peak looks
like a narrow triangle with the virtually linear frontal and
back edges (Fig. 4, curves 1, 1´). In this case, the intersecꢀ
tion point of the baseline and frontal edge forming the
γ angle with the baseline is taken as the melting point.
Ideally, this angle is constant for a given instrument and
specified conditions of thermogram recording. In pracꢀ
tice, this constant is determined from melting curves of
standard samples (in our case, these are twice sublimed
phenol and naphthalene). The shape of the melting curve
of the sample under study can differ strongly from ideality
and be polymodal (see Fig. 4, curve 3). However, in the
case of the real unimodal curve (see Fig. 4, curve 2), its
frontal edge contains nonlinear regions and the slope angle
,
(1)

Crystallization of chiral aryloxypropanediols
Russ.Chem.Bull., Int.Ed., Vol. 55, No. 2, February, 2006
235
∆Q
of the sample, whose specific rotation value is used as the
parameter [αmax], the more the oр value approaches to the
true enantiomeric excess ее. The used combined GCꢀMS
and thermographic monitoring warrants the chemical and
enantiomeric purity of reference samples of all diols studꢀ
ied to be at least 99%.
1´
1
The liquidus lines of the phase diagrams for these comꢀ
pounds, which were calculated by Eq. (1) and experimenꢀ
tal data for enantiomerically pure diols, are presented in
Fig. 5 by dashed lines. The same figures show the experiꢀ
mentally obtained liquidus points drawn by circles. For
samples of diols 1a—e with an intermediate enantiomeric
composition, the experimental DSC thermograms someꢀ
times contain additional peaks with the almost constant
(for a certain compound) temperature Teu, which correꢀ
sponds to the melting point of the eutectics. These experiꢀ
mental characteristics are put on the plot as solid circles.
As can be seen from the data in Figs 5, a—e for the
three last members of the series, viz., oꢀClꢀ, oꢀBrꢀ, and
oꢀIꢀsubstituted derivatives, the experimental points virtuꢀ
ally fall on the line of the theoretical phase diagrams
calculated by Eq. (1) for the conglomerateꢀforming comꢀ
pound. The calculated temperatures of eutectic melting
for these compounds coincide almost exactly with the
melting point of the racemate (see Table 1). In the case of
Fꢀsubstituted derivative 1b, a discrepancy of ~7 °C is
observed, which increases to 12 °C for unsubstituted deꢀ
rivative 1a. The liquidus branches corresponding to Eq. (2)
were calculated and are shown in the figures (dotted lines)
for these two compounds. A comparison of thus obtained
theoretical phase diagrams with experiment shows unꢀ
doubtedly that phenoxypropanediol 1a forms the typical
racemic substance on crystallization. The same concerns,
most likely, diol 1b.
2
3
α
γ
40
50
60
80
T/°C
Fig. 4. Typical melting curves: 1 and 1´, standard (naphthalene)
before (1) and after the correction procedure (deconvolution, 1´);
2, unimodal curve with a distorted frontal edge (racꢀ1a);
3, polymodal curve (racꢀ2b, large crystals).
of the linear region of the edge to the baseline α does not
coincide with the γ angle (α < γ).
Following the known recommendations,24,25 we acꢀ
cepted the intersection point of the line dropped from the
maximum of the curve to the baseline at the γ angle (TL)
as the temperature of melting termination. To facilitate
the visual perception of experimental results, we took into
account the standard slope (γ) beforehand. For this purꢀ
pose, the temperature scale was corrected to the equation
t´ = t – h•cot(γ), where t is the observed temperature, t´ is
the corrected temperature, and h is the thermal flux for
the given point. After this deconvolution of the experiꢀ
mental curves, the frontal edge of an ideal sample forms
the right angle with the baseline, and the melting temꢀ
perature TL corresponds to the maximum in the curve.
These curves are shown in Figs 2 and 3. The experimental
data on the melting points and enthalpies of melting for
racemic and enantiomerically pure samples 1a—e are given
in Table 1. When plotting experimental phase diagrams
for binary mixtures of enantiomers, it is important to
have reliable data on the enantiomeric composition of
the sample under study (mole fractions xR and xS of
each enantiomer, хR + xS = 1). For this purpose, we
used the optical purity value op = [αi]/[αmax] ≈ ee =
|хR – хS|/(хR + хS). The higher is the enantiomeric purity
Knowing the melting temperature and enthalpy of
melting for racemate and individual stereoisomers makes
it possible to consider thermodynamic characteristics,
which cannot directly be determined experimentally, for
analysis of the type of crystallization. For instance, in the
ideal case, the entropy of mixing of individual enantioꢀ
mers ∆Slm in the liquid phase for the racemic conglomerꢀ
Table 1. Thermodynamic characteristics of 3ꢀ(2ꢀRꢀphenoxy)propaneꢀ1,2ꢀdiols*
R
T f
T f
∆T f
T f
∆H f
∆H f
∆G 0
f
T
R
∆Slm
A
R
A—R
eu
А
R
°C
J mol–1
/J mol–1 K–1
H
F
Cl
Br
I
68.2
60.5
90.0
101.1
110.4
58.7
45.0
71.7
80.3
89.3
9.5
46.2 (56.0)
37.6 (43.9)
71.1
80.3
89.0
28600
26040
38150
38380
37380
26330
20500
29000
33390
34220
–1116
–623
–64
4
2.30
3.39
4.90
5.51
5.42
15.5
18.3
20.8
21.1
–31
* The eutectic melting temperatures calculated for the racemic conglomerate and racemic compound (in parentheses) are
presented.

236
Russ.Chem.Bull., Int.Ed., Vol. 55, No. 2, February, 2006
Zakharychev et al.
T/°C
T/°C
T/°C
70
60
50
40
90
80
70
1a
1b
1c
60
50
0.5
1.0
x
0.5
1.0
x
0.5
1.0
x
T/°C
T/°C
100
110
1d
1e
90
80
100
90
0.5
1.0
x
0.5
1.0
x
Fig. 5. Phase diagrams of binary mixtures of enantiomers of compounds 1a—e; x is the mole fraction of the Rꢀenantiomer.
ate is R•ln2 = 5.75 J mol–1 K–1, and the theoretical free
tions from ideality when mixing individual stereoisomers
of this substance. These deviations are related to preferꢀ
ential contacts between molecules of stereoisomers with
different configurations. The thermodynamic characterꢀ
istics of Fꢀsubstituted derivative 1b also indicate that a
racemic compound is formed in the solid phase, although
this compound is less stable than the unsubstituted one.
For Clꢀ, Brꢀ, and Iꢀsubstituted compounds 1c—e, the
energy ∆G0T R for racemic compound formation is always
f
negative. These thermodynamic parameters can be calcuꢀ
lated from the experimentally measured characteristics
using Eqs (3) and (4), respectively.26
(3)
∆Slm and ∆G0
values are close for all the three subꢀ
f
T R
(4)
stances and fall in the interval of values, which are typical
of racemic conglomerates.
We estimated the error of determination of ∆Slm and
Thus, the study of the vibrational spectra and thermoꢀ
dynamic characteristics of five similar 3ꢀ(2ꢀRꢀphenoxyꢀ
propane)ꢀ1,2ꢀdiols revealed that in this series an increase
in the substituent size primarily decreases the stability of
racemic substances that are formed in the crystalline state.
After some critical size is achieved, the type of crystallizaꢀ
tion of homologous chiral compounds changes, resulting
in the formation of racemic conglomerates in the solid
phase. Note that the change in the type of crystallization
depending on the anion size has been observed earlier27
for salts of chiral amino alcohols. Further analysis of the
nature of crystallization, in the case under study, can be
carried out by investigation of the microenvironment of
chiral molecules in the crystal lattice. In addition, the
observation of new conglomerates would extent the sphere
of practical use of spontaneous resolution of racemates
into individual enantiomers. Both these aspects (theoretiꢀ
cal and applied) will be studied in the future.
∆G0
by the Monte Carlo method on the basis of the
f
T R
data on the measurement error for T f and ∆H f of standard
substances. We took 5% of the measured enthalpy value
and 0.5 °C for the melting temperature as estimates of
these errors. Using these parameters, a pseudoꢀrandom
selection of 1000 points was generated, and ∆Slm and
∆G0T R were calculated and rootꢀmeanꢀsquare deviations,
f
which were accepted as estimates of the determination
error of ∆Slm and ∆G0T R, were estimated on the basis of
f
these points. Their values were 0.17 J mol–1 K–1 and
70 J mol–1, respectively.
The calculated ∆Slm and ∆G0
values for the comꢀ
f
T R
pounds under study are also presented in Table 1. The
considerable negative value of ∆G0
propaneꢀ1,2ꢀdiol (1a) indicates that the formation of a
racemic substance in the solid phase is thermodynamiꢀ
cally preferential. The ∆Slm value shows substantial deviaꢀ
for 3ꢀphenoxyꢀ
f
T R

Crystallization of chiral aryloxypropanediols
Russ.Chem.Bull., Int.Ed., Vol. 55, No. 2, February, 2006
237
13. J. P. Lambooy, J. Am. Chem. Soc., 1951, 73, 349.
14. B. J. Ludwig, W. A. West, and W. E. Currie, J. Am. Chem.
Soc., 1952, 74, 1935.
15. T. Seki, T. Takezaki, R. Ohuchi, M. Saitoh, T. Ishimori,
and K. Yasuda, Chem. Pharm. Bull., 1995, 43, 1719.
16. GB Pat. 628497; Chem. Abstrs, 1950, 44, 3023d.
17. S. Kuwabe, K. Torraca, and S. Buchwald, J. Am. Chem.
Soc., 2001, 123, 12202.
The authors thank D. R. Sharafutdinova for performꢀ
ing GCꢀMS analysis.
This work was financially supported by the Russian
Foundation for Basic Research (Project Nos 03ꢀ03ꢀ33084
and 06ꢀ03ꢀ32508).
References
18 P. Brenans, Bull. Soc. Chim. Fr., 1913, 13, 533.
19. J. Jacques, A. Collet, and S. H. Wilen, Enantiomers, Raceꢀ
mates, and Resolutions, Krieger Publishing Co, Malabar, FL,
1994, 447 pр.
20. E. L. Eliel, S. H. Wilen, and L. N. Mander, Stereochemistry
of Organic Compounds, John Wiley and Sons, New York,
1994, 1267 pp.
21. Beilstein Database, Copyright©1988—2004, Beilstein Institut
zur Foerderung der Chemischen Wissenschaften licensed to
Beilstein GmbH and MDL Information System GmbH,
Beilstein GmbH, Frankfurt am Main, Germany.
22. G. Coquerel, Enantiomer, 2000, 5, 481.
23. I. Prigogine and R. Defay, Chemical Thermodynamics,
Longmans Green and Co, London—New York—Toronto,
1954.
1. A. A. Bredikhin, D. V. Savel´ev, Z. A. Bredikhina, A. T.
Gubaidullin, and I. A. Litvinov, Izv. Akad. Nauk, Ser. Khim.,
2003, 812 [Russ. Chem. Bull., 2003, 52, 853 (Engl. Transl.)].
2. L. Pasteur, Ann. Chim. Phys., 1848, 24, 442; L. Pasteur,
Compt. Rend. Acad. Sci., 1848, 26, 535.
3. A. Collet, M.ꢀJ. Brienne, and J. Jacques, Chem. Rev., 1980,
80, 215.
4. L. PerezꢀGarcia and D. B. Amabilino, Chem. Soc. Rev.,
2002, 31, 342.
5. A. A. Bredikhin, Z. A. Bredikhina, S. N. Lazarev, and D. V.
Savel’ev, Mendeleev Commun., 2003, 104.
6. D. V. Zakharychev, M. D. Borisover, and B. N. Solomonov,
Zh. Fiz. Khim., 1995, 69, 175 [Russ. J. Phys. Chem., 1995, 69
(Engl. Transl.)].
24. G. W. H. Höhne, H. K. Cammenga, W. Eysel, E. Gmelin,
and W. Hemminger, Thermochim. Acta, 1990, 160, 1.
25. H. E. Gallis, F. Bougrioua, H. A. J. Oonk, P. J. van Ekeren,
and J. C. van Miltenburg, Thermochim. Acta, 1996, 274, 231.
26. Z. J. Li, M. T. Zell, E. J. Munson, and D. J. W. Grant,
J. Pharm. Sci., 1999, 88, 337.
27. A. A. Bredikhin, Z. A. Bredikhina, A. T. Gubaidullin, D. B.
Krivolapov, and I. A. Litvinov, Mendeleev Commun., 2004,
14, 268.
7. Y. Gao, R. M. Hanson, J. M. Klunder, S. Y. Ko,
H. Masamune, and K. B. Sharpless, J. Am. Chem. Soc.,
1987, 109, 5765.
8. J. Chen and W. Shum, Tetrahedron Lett., 1995, 36, 2379.
9. G. Egri, A. Kolbert, J. Balint, E. Fogassy, L. Novak, and
L. Poppe, Tetrahedron Asymmetry, 1998, 9, 271.
10. K. Kitaori, Y. Furukawa, H. Yoshimoto, and J. Otera, Tetraꢀ
hedron, 1999, 55, 14381.
11. T. Toshikuni, S. Takeshi, and O. Hiromichi, Tetrahedron
Asymmetry, 2001, 12, 2001.
12. F. Theil, J. Weidner, S. Ballschuh, A. Kunath, and H. Schick,
J. Org. Chem., 1994, 59, 388.
Received November 1, 2005