Lamellar racemic twinning as an obstacle for the resolution of enantiomers by crystallization: The case of Me(All)N+(CH2Ph)Ph X- (X = Br, I) salts

Article abstract of DOI:10.1021/jp035588l

Ammonium salts Me(All)N⁺(CH₂Ph)Ph X^- (X = Br, 1; X = I, 2) (Asymmetric nitrogen, communication 89, prev. see Kostyanovsky, R. G.; Kostyanovsky, V. R.; Kadorkina, G. K.; Lyssenko, K. A. Mendeleev Commun. 2003, 111) were historically the first compounds with a chiral nitrogen center obtained in optically pure forms via diastereoisomers. We were astonished by the fact that W. J. Pope, who first carried out the resolution (1899), mentioned in his articles that salts have features for conglomerate formation but nowhere did he note the registration of optical activity and manual sorting of left- and right-handed single crystals. We reinvestigated these salts with the help of X-ray diffraction and found that the space groups of both them are P2₁2₁2₁. However, despite the chiral space group, the enantiomeric composition of the majority of the crystals obtained by crystallization from solutions of racemates was found to be near racemic. Such crystal growth behavior, called lamellar twinning between enantiomers, represents a serious obstacle to the methods of enantiomeric resolution by straightforward crystallization that are used in the chiral industry. The phenomenon probably results from a high concentration in solution of heterochiral growth clusters, which promote the formation of the intergrowth region of twins. To glance experimentally at what might possibly be the first crystal growth units, electrospray ionization FT-ICR mass spectrometry was applied. We were able to register polyionic clusters containing up to five chiral cations. The relative stability of the simplest dimeric clusters (two cations linked by an anion) of 1 with homo- and heterochiral composition was estimated with a B3LYP/SDD calculation, and heterochiral one was found to be more stable by 0.883 kcal/mol.

Full text of DOI:10.1021/jp035588l

J. Phys. Chem. B 2003, 107, 13523-13531
13523
Lamellar Racemic Twinning as an Obstacle for the Resolution of Enantiomers by
Crystallization: The Case of Me(All)N+(CH2Ph)Ph X- (X ) Br, I) Salts
Vladimir Yu. Torbeev,*,† Konstantin A. Lyssenko,‡ Oleg N. Kharybin,§
Mikhail Yu. Antipin,‡ and Remir G. Kostyanovsky*,†
N. N. SemenoV Institute of Chemical Physics, A. N. NesmeyanoV Institute of Organoelement Compounds, and
Institute of Energy Problems of Chemical Physics, Russian Academy of Sciences, 119991 Moscow, Russia
ReceiVed: June 5, 2003; In Final Form: August 30, 2003
Ammonium salts Me(All)N+(CH2Ph)Ph X- (X ) Br, 1; X ) I, 2) (Asymmetric nitrogen, communication 89,
prev. see Kostyanovsky, R. G.; Kostyanovsky, V. R.; Kadorkina, G. K.; Lyssenko, K. A. MendeleeV Commun.
2003, 111) were historically the first compounds with a chiral nitrogen center obtained in optically pure
forms via diastereoisomers. We were astonished by the fact that W. J. Pope, who first carried out the resolution
(1899), mentioned in his articles that salts have features for conglomerate formation but nowhere did he note
the registration of optical activity and manual sorting of left- and right-handed single crystals. We reinvestigated
these salts with the help of X-ray diffraction and found that the space groups of both of them are P212121.
However, despite the chiral space group, the enantiomeric composition of the majority of the crystals obtained
by crystallization from solutions of racemates was found to be near racemic. Such crystal growth behavior,
called lamellar twinning between enantiomers, represents a serious obstacle to the methods of enantiomeric
resolution by straightforward crystallization that are used in the chiral industry. The phenomenon probably
results from a high concentration in solution of heterochiral growth clusters, which promote the formation of
the intergrowth region of twins. To glance experimentally at what might possibly be the first crystal growth
units, electrospray ionization FT-ICR mass spectrometry was applied. We were able to register polyionic
clusters containing up to five chiral cations. The relative stability of the simplest dimeric clusters (two cations
linked by an anion) of 1 with homo- and heterochiral composition was estimated with a B3LYP/SDD
calculation, and heterochiral one was found to be more stable by 0.883 kcal/mol.
1. Introduction
(space group P212121).10 An orthogonal orientation of crystal-
lographic axes appears to favor the formation of racemic twins
Crystallization methodologies are widely used in industry2
for large-scale enantiomeric resolution. Among them, the
methods of direct crystallization of enantiomers3,4 are the most
attractive because of their simplicity and low cost. However,
these procedures can be applied only to the minority of systems
in which the racemate crystallizes to give a mixture of
enantiopure crystals (racemic conglomerate).3 Although the
frequency of occurrence of conglomerates is rather low (perhaps
10% of all racemates), several promising approaches for the
directed prediction and design of these conglomerates were
recently proposed.5 Nevertheless, once the problem of finding
a conglomerate is solved, the resolution may be complicated
and fraught with difficulties.6 One possible difficulty is lamellar
racemic twinning, where instead of the crystal lattice containing
only one enantiomer it actually consists of many domains
(lamellae) of both enantiomers. The phenomenon has been found
and investigated for a variety of compounds: â-phenylglyceric
acid,7a helycens,7b,c triazolyl ketone,7d 2-azabicyclo[2.2.1]hept-
5-en-3-one,7e methionine, cysteine and valine hydrochlorides,7f,g
threonine,7h and 5-ethyl-5-methylhydantoin.7i,j Also, a number
of compounds reported in the literature have hints of the same
property.8,9
with a minimal number of defects.11 For a racemic twin, the
structural X-ray investigation proceeds in the same way as for
any normal single-crystal analysis; no disorder can be detected,12
and the presence of the second enantiomer can be evaluated
only from a refinement of the Flack parameter in the absolute
structure determination.13 Racemic twinning has been found to
affect optical properties; it was shown recently that this
phenomenon causes a reduction in the intensity of the second-
harmonic generation of nonlinear optical materials.10b It is also
well documented that twins usually exhibit birefringence,7d,h
which furthermore can be used to discriminate between twinned
and nontwinned species. It has also been reported that racemic
twinning affects the stereoselectivity of some solid-state
reactions.10a,14
In the present paper, we report a reinvestigation of the
compounds in which this phenomenon was probably observed
historically for the first time. At that time, it was interpreted
incorrectly and even caused the generation of an erroneous
theory of five-valent nitrogen.
2. Historical Intrigue
The common feature for the majority of compounds exhibiting
lamellar racemic twinning is orthorhombic crystal symmetry
In 1899, Edgar Wedekind summarized his work on the
stereochemistry of chiral ammonium salts.15 He believed that
nitrogen had a tetragonal-pyramidal configuration and when
bearing five different substituents (according to his model, a
covalent bond occurred between the anion and nitrogen) would
exist as three enantiomeric pairs. The preparation of each
* Corresponding authors. E-mail: V.Y.T. torbeev@polymer.chph.ras.ru.
R.G.K. kost@chph.ras.ru. Fax: 7-095-938-21-56.
† N. N. Semenov Institute of Chemical Physics.
‡ A. N. Nesmeyanov Institute of Organoelement Compounds.
§ Institute of Energy Problems of Chemical Physics.
10.1021/jp035588l CCC: $25.00 © 2003 American Chemical Society
Published on Web 11/08/2003

13524 J. Phys. Chem. B, Vol. 107, No. 48, 2003
Torbeev et al.
Figure 1. (a) Two enantiomorphous crystals of salt 1, grown by W. J. Pope and A. W. Harvey from aqueous solutions of optically pure enantiomers.17b
The faces of the left crystal are assigned with Miller indices; the right one is assigned with the original Pope symbols. (b) Twinned crystal of 1
obtained by us from an aqueous solution of racemic composition. It is interesting that oriented nucleation of the majority of crystals from a glass/
solution or an air/solution interface occurs: most of the crystals grow from one of four possible {110} faces, which finally have a square shape.
stereoisomeric pair would depend on the order of substitutents
insertion during the synthesis. When preparing methylallylben-
zylphenylammonium bromide 1, iodide 2, and later chloride16
in three different ways, he thought that he had isolated R, â,
and γ isomers that had the same elementary composition but
differed in melting points, solubilities, and crystal habits.
In the same year, compounds 1 and 2 became the targets for
British chemist W. J. Pope and his colleague S. D. Peachey,
who resolved them into enantiomers with the help of d- and
l-camphorsulfonic acid17a and for the first time proved that the
nitrogen atom can be a stereogenic center. They grew crystals
of salt 1 from aqueous solutions of optically pure D and L
enantiomers17b (Figure 1a) and compared crystallographic
parameters of these crystals and those investigated by Wedekind16a
and found them to be identical. Pope immediately recognized
that Wedekind held in his hands not a simple racemate but “a
mere mechanical mixture of the two component salts.”17b
When measuring the optical rotation of a solution of partially
resolved salt 2, Pope noticed that crystals precipitated in the
polarimetric tube. They were determined later to consist of
optically pure salt 2. The solution, meanwhile, whereas previ-
ously it had been able to rotate a plane of polarized light, now
suddenly lost its optical activity.17a In this case, Pope correctly
explained the result18 of this experiment; he understood outright
that salts 1 and 2 are conglomerates.
However, despite his knowledge of the tetrahedral model of
A. Werner, Pope interpreted the results in the context of the
“quinquevalent” nitrogen theory.19 Because he had salts 1 and
2 prepared by one of the three possible ways, he thought that
he had correspondingly one enantiomeric pair from three
possible pairs. The resolution into enantiomers only strengthened
his confidence in this theory. Another aspect of his work seems
even more mysterious and strange. Given Pasteur’s experiments
with sodium-ammonium tartrate tetrahydrate,20 Pope does not
mention in his papers the measurement of the optical activity
of single crystals of synthetic salts 1 and 2, or in the other words,
he did not perform the classical manual triage of left- and right-
handed crystals.21
mental Section). An X-ray study of monocrystals of salts 1 and
2 was performed, and the structures were found to be chiral
and isomorphous (space group P212121) (Table 1).
However, despite the chiral packing (Figure 2b), the enan-
tiomeric composition of the majority of the crystals was found
to be near-racemic, with only a few particles optically enriched.
Crystallization from absolute ethanol according to the Wedekind
procedure led to series of crystals with different habits in the
same crystallization setup (Figure 3). In this experiment, we
obtained prisms consisting of equal amount of both enantiomers
(Figure 3a) alternating with long thin needles with high
enantiomeric purities (Figure 3e). Thus, we were able to
demonstrate the racemic twinning (or polyepitaxy) phenom-
enon.22
In the case of salt 1, an X-ray investigation was performed
for a crystal of the prismatic shape and also one with needlelike
morphology (Table 1, columns A and B). We found that all
crystallographic parameters in both cases are identical. A
refinement of the Flack parameter13 led to a direct estimation
of the enantiomeric composition. In the case of prismatic crystal,
the value x was found to be 0.53 (ee ) -6%), whereas for
acircular crystal it was equal to 0.12 (ee ) 76%). Furthermore,
the stereoselective dissolution of a twinned crystal in a solution
enriched by one enantiomer clearly demonstrates its lamellar
structure (Figure 4).
The 2D nucleation of the opposite enantiomer on the growing
crystal surface of the original enantiomer requires additional
expenses of energy compared with the energy required for the
ordinary nucleation of this original enantiomer. This barrier is
equal to the difference between attachment energies of layers
with the same and opposite chiralities. Therefore, the oscillating
nucleation of two enantiomers along the specific direction should
reduce the rate of growth in this direction. Figure 5 illustrates
how a hypothetical nucleus can grow out into two morphologi-
cally diverse crystals depending on the comparative growth rates
of different faces. Morphological analysis (Figure 5 and 6)
together with measurements of the optical activity led to the
assignment of the twin plane to ab or {00l}.
Thus, the racemic twinning of salts 1 and 2 that was not
recognized by Wedekind or Pope did not allow them to observe
directly Pasteur-like spontaneous resolution and, moreover,
stimulated the development of the erroneous theory of quin-
quevalent nitrogen. Certainly, crystals with different enantio-
meric compositions have different melting points and solubilities
and, most importantly, completely different crystal habits. This
3. Lamellar Racemic Twinning: Structural and
Morphological Investigations
To provide an explanation of the activities of scientists of
that epoch and in particular the curious omission of the classical
manual triage, we again synthesized these compounds (Experi-

Lamellar Racemic Twinning
J. Phys. Chem. B, Vol. 107, No. 48, 2003 13525
TABLE 1: Crystallographic Data for Salts 1 and 2
compound
formula
1
2
C17H20BrN
C17H20IN
M (g mol-1
)
318.25
298 K
orthorhombic
P212121
4
365.24
temperature
crystal class
space group
Z
140 Ka
orthorhombic
P212121
4
radiation type
Mo KR
Mo KR
A
B
a(Å)
b(Å)
9.185(1)
12.397(1)
13.982(2)
1592.1(3)
9.190(1)
9.410(1)
12.551(2)
14.150(2)
1671.2(4)
728
1.905
19.05
12.408(2)
14.015(2)
1598.2(6)
c(Å)
V(Å3)
F(000)
656
F
calc (gcm-1
)
1.328
25.70
2.19-29.13
1.323
25.70
2.19-29.13
µ (cm-1
θ (deg)
)
2.17-30.02
reflections
measured
17073
17785
19964
independent
reflections
4201 (R(int) )
0.0421)
4220 (R(int) )
0.0479)
4875 (R(int) )
0.0263)
reflections
2775
2585
4693
[I > 2σ(I)]
Flack parameter
R(Fhkl): R1
wR2
0.117(9)
0.0246
0.0427
0.532(8)
0.0260
0.0407
0.35(2)
0.0278
0.0622
1.017
GOF
0.916
0.915
Fmax/Fmin
0.449/-0.275
0.599/-0.190
0.758/-0.798
(eÅ-3
)
a The investigation of 2 was also performed at room temperature. The structure was refined to wR2 ) 0.0992, R1 ) 0.0405, Flack parameter
0.48(4). Crystallographic parameters are a ) 9.409(2) Å, b ) 12.709(2) Å, c ) 14.280(3) Å, V ) 1707.5(5) Å3, density 1.421 g cm-1
.
Figure 2. (a) Structure of the S-(-)-Me(Ph)N+(All)CH2Ph cation. (b) Packing ([100] projection) of salts 1 and 2 in P212121. Parameters of H
bonding (C-H‚‚‚Hal-): H‚‚‚Br, 2.77-3.09 Å for 1; H‚‚‚I, 2.86-3.20 Å for 2. All C-H distances are normalized to the ideal value of 1.08 Å.
was perceived by Wedekind as direct proof of the existence of
R, â, and γ isomers.
symmetry is higher than that of the parent enantiomeric crystal.11
The whole crystal consists of alternating lamellae of left- and
right-handed crystal domains, each tens of micrometers thick
(Figure 8).
4. Structural Model for Lamellar Racemic Twinning and
Methods for Its Removal
The zone of intergrowth (Figure 7a) may be considered to
be another polymorphic modification (racemic compound) of
the parent left and right enantiomorphs.22 The occurrence of
lamellar twinning suggests a negligible energetic gap between
the enantiomorphous and racemic crystal lattice. In recent
articles,7i,j the estimation of the energies of such crystal lattices
was carried out, and the authors found that although the
enantiomorphous packing is more stabilized the difference from
On the basis of the data obtained, we modeled the twinning.
When a fragment of the crystal packing, which includes only
left-handed ammonium cations, is superimposed on a fragment
of right-handed cations, the strongest contacts H‚‚‚Hal do not
break (Figure 7a). A new symmetry element, a center of
inversion, now appears that does not belong to the original space
group. Such twins are known as mimetic twins where the twin

13526 J. Phys. Chem. B, Vol. 107, No. 48, 2003
Torbeev et al.
Figure 6. Assignment of faces for crystals of 1: (a) ee 0%; (b) ee
61%; and 2: (c) ee 0%; (d) ee 18%.
of the first enantiomer and that the impetus for the secondary
nucleation is the local supersaturation of the opposite enanti-
omer.
Figure 3. Set of crystals with different morphology and enantiomeric
composition obtained from ethanol solutions of racemic 1. Crystals:
(a) ∆ꢀ(212.5 nm) 0, Flack parameter 0.53 (S/R 1:1); (b) ∆ꢀ(212.5 nm)
+0.5; (c) ∆ꢀ(212.5 nm) -3.8; (d) ∆ꢀ(212.5 nm) +7.1; (e) ∆ꢀ(212.5
nm) -9.3, Flack parameter 0.12 (S/R 9:1).
The use of enantiomerically pure tailor-made auxiliaries,7f,g
the addition of wetting reagents,7i and the application of gentle
stirring7j have been proposed as means to remove racemic
twinning. The first way is convenient for many systems because
its application also solves the problem of enantiomeric resolu-
tion. However, it does require the often expensive synthesis and
preparation of a tailor-made additive in enantiomerically pure
form.
We anticipated that in order to avoid twinning during crystal
formation, in this case, it is necessary to suppress the growth
of the ab faces. We designed the substituent-modified methyl-
allyl(p-brombenzyl)phenylammonium bromide 3 as an additive
that is able to insert stereospecifically into the necessary faces
instead of the cations of 1. It thus blocks these faces for effective
growth (Figure 7a) and at the same time prevents the formation
of twins.
The crystallizations of 1 with the addition of racemic 3 were
performed from aqueous and ethanol solutions. In water, the
direction of fastest growth occurs along the crystallographic a
axis, whereas the direction of twinning is the c axis. At
concentrations of 1-5% (w/w) of 3, there was no noticeable
modification of crystal shape; only the addition of 3 at up to
10% (w/w) considerably reduced the rate of growth along the
a direction. However, the majority of the crystals were still
twinned. In the ethanol solutions, the direction of the most rapid
crystal growth coincides with twinning axis c and, as mentioned
before, becomes slower when twinning occurs. The addition of
3 at up to 10% (w/w) first inhibited crystallization, but finally,
after several days of slow evaporation of the solvent, crystals
appeared as asymmetric prisms, which were unusually shortened
in length along the c direction and were of homochiral
composition.
Figure 4. Crystal of 1 with racemic composition starts to dissolve in
an aqueous solution of enantiomerically enriched 1. Here the lamellar
structure of the crystal can be clearly recognized.
5. Understanding the Racemic Twinning Phenomenon on
the Basis of a Cluster Model of Crystal Growth
Recently, new insights that emphasize the embryonic stages
of molecular and ionic aggregations in solution were introduced
into crystal science.23,24 During the nucleation steps in particular,
the most significant events can happen, and these determine
the macroscopic crystal growth behavior. It has now been
illustrated that processes leading to the isolation of a particular
crystal phase occur as the result of the formation and self-
assembly of clusters. Hence, racemic twinning probably results
from rather high concentrations in solution (or melt) of clusters
with a heterochiral composition.
Figure 5. (a) Crystal nucleus grows into two morphologically different
crystals, depending on the relative growth rates of faces A and B. (b)
Different morphology of crystals of 1 depending on different growth
rates along the c axis.
the modeled racemate is low (∆E ) 2.5 kcal/mol). The authors
concluded7i,j that lamellar racemic twinning results from second-
ary nucleation of the opposite enantiomer on a growing crystal

Lamellar Racemic Twinning
J. Phys. Chem. B, Vol. 107, No. 48, 2003 13527
Figure 7. (a) Two fragments of the structures of S-(-)- and R-(+)-enantiomers of 1 or 2 can be “stitched” to each other across the {00l}
crystallographic plane. An auxiliary cation of salt 3 (see text) is adsorbed onto faces {00l} and inhibits their effective growth. (b) Three types of
neutral ionic dimeric clusters, which are formed in the supersaturated solution and are probably responsible for the macroscopic crystal growth
picture; CS and CR are chiral left-handed and right-handed ammonium cations, respectively, and A is an achiral bromide or iodide anion.
pyrrole (1:1) complexes. The authors found that peaks corre-
sponding to clusters with the same stoichiometry predominate
in the mass spectra, and from these results they were able to
draw conclusions about the structural motif of the growth unit.
However, the method of ionization that they used raises doubts
concerning the application to fragile noncovalent associates with
high molecular weight.
With electrospray ionization (ESI), Cooks et al.29 and another
group30 reported that serine tends to form magic number clusters
Figure 8. Picture illustrating the alternation of left and right domains
in a lamellar racemic twin.
composed of eight molecules with strong chiral discrimination.
ESI is a soft ionization device; however, the disadvantage is
that during this process a large number of transformations and
rearrangements may occur and completely distort the structure
of agglomerates existing in solution.
The growth blocks for ionic crystals are polyionic ag-
gregates.25 Thus, if the stereogenic center is a cation, then in
the simplest case, in a supersaturated solution, three types of
dimericclusterscanbeformed: {(CS‚A‚CS)+A-},{(CR‚A‚CR)+A-},
and {(CS‚A‚CR)+A-}, where CS and CR are chiral cations and
A is an achiral anion (Figure 7b). The first two can be considered
to be responsible for aggregation into larger homochiral domains
and, finally, for the formation of chiral crystals. The heterochiral
third, in fact, can be the cause for the creation of a heterochiral
structural fragment, which in twins is between the two enan-
tiomorphic lamellae of opposite configuration.
It has been shown that mass spectrometry (MS) can be used
as an analytical tool for the investigation of the aggregation of
molecules and ions in solution.26 The technique seems to have
incontestable advantages such as the simplicity of the interpreta-
tion of results and its comparability with synchrotron methods.27
Currently, with the help of MS, several attempts to investigate
the clustering in solution that are relevant to crystallization have
been reported. In the first one,28 the liquid adiabatic expansion
technique for sample input followed by electron impact ioniza-
tion was used to explain the cocrystallization of pyridine-
A more optimistic point of view was demonstrated in the
ingenious work of Beauchamp et al.31 These authors postulated
that the clusters of serine registered after ESI are not the same
as that in solution but they do reflect their original structure
and properties. With ion-mobility spectrometry accompanied
by DFT calculations, they deduced the cubic structure of the
serine octamer. However, later calculations led to confusion
because of the absence of a preference for homochirality in such
structural fragments. The preference, however, was found for
the H-bonded chain structural motif derived from the crystal
structure of L-serine. The presence in solution of such linear
homochiral associates, which may serve as precursors for the
gas-phase clusters detected by MS, explains the origin of their
homochiral composition; their transformation to the cubic motif
was proposed to happen within the ESI sample inlet. Thus, in
attempting to explain the origin of the chiral discrimination
effect in clusters in the gas phase, the authors obtained
interesting information about the crystal growth units in solution.

13528 J. Phys. Chem. B, Vol. 107, No. 48, 2003
Torbeev et al.
Figure 9. ESI FT-ICR mass spectra demonstrating the clustering of 1 and 2 (the letter N+ indicates the ammonium cation Me(Ph)N+(All)CH2Ph).
(a) Single-charged clusters containing up to five cations were registered for 1. The complex-type signals are conditioned by a combination of ions
with different isotopes, and the picture is complicated by the natural isotopic abundance of bromine (79Br 50.54%, 81Br 49.46%). (b) Clusters with
up to four cations were obtained for 2. In all cases, the average and whole values for m/z are given.
Because it was available, we tested the solutions of salts 1
and 2 with ESI Fourier transform ion-cyclotron resonance (FT-
ICR) MS and were able to detect several positively charged
ionic agglomerates along with the single ammonium cation
(Figure 9).32,33 These clusters are possibly relevant34 to polyionic
growth blocks existing in solution. The B3LYP/SDD calculation
of the homo- and heterochiral dimeric clusters of 1 (two cations
and one anion) has revealed that they have practically the same
energy with a slightly lower value (by 0.883 kcal/mol) for the
heterochiral cluster. The theoretically obtained geometries
(Figure 10) are quite close to the experimental ones from the
X-ray investigation. The Br‚‚‚H contacts vary from 2.65-2.81
and 2.69-2.79 Å for the homo- and heterochiral clusters,
respectively. For comparison, corresponding values for the
same structural fragment of the crystal structure of 1 are
2.73-2.99 Å.

Lamellar Racemic Twinning
J. Phys. Chem. B, Vol. 107, No. 48, 2003 13529
Figure 10. Geometries of (a) homo- and (b) heterochiral clusters of 1 optimized with the B3LYP/SDD calculation. The energy of the heterochiral
cluster is 0.883 kcal/mol lower than that of the homochiral alternative.
6. Conclusions
7. Experimental Section
1H and 13C NMR spectra were recorded using Bruker DRX-
500 and Bruker WM-400 spectrometers. Optical activities were
determined by polarimetry and CD methods (polarimeter
Polamat A and CD spectrometer Jasco-500A). Optical rotations
were measured for aqueous or ethanol solutions (0.03-0.04 M)
at 23-25 °C in a 1-dm tube. Optical purities were estimated
by comparing with values of resolved 1 and 2.17b Melting points
were measured with a Stuart Scientific SMP-3 apparatus. Mass
spectra were obtained using an experimental plant based on the
ion-cyclotron resonance Bruker Spectrospin CMS-47 mass
spectrometer (Superconducting magnet 4.7 T, vacuum 8 × 10-8
Torr). The settings that were used were a source voltage of 2
kV, a capillary voltage of 100 V, a capillary temperature of
125 °C, and an electrospray tip inner diameter of 89 µm.
Samples were electrosprayed using a 50:50 methanol/water
mixture with a concentration of 1 × 10-3 -5 × 10-3 M at a
flow rate of 500 nL/min from a 25-µL Hamilton syringe.
Crystallographic data and parameters of the refinement for
the crystal structure determination of salts 1 and 2 are presented
in Table 1. The data collection was carried out on a SMART
1000 CCD using ω-scan mode with a 0.3° step in ω and a 10-s
exposure per frame. All structures were solved by direct methods
and refined by full-matrix least squares against F2 in the
anisotropic (H atoms are isotropic) approximation using the
SHELXTL-97 package. The data were processed with the use
of the SAINT PLUS program package. The absorption correc-
tion and the merging of reflections were applied using the
SADABS program. The positions of hydrogen atoms were
located from the Fourier electron density synthesis and were
included in the refinement in the isotropic riding model
approximation. The absolute structure was determined on the
basis of the Flack parameter. Crystallographic data (excluding
structure factors) for the structures reported in this paper have
been deposited in the Cambridge Crystallographic Data Centre
as supplementary no. CCDC-209797 for 1 A, 209798 for 1 B,
and 209798 for 2. Copies of the data can be obtained free of
charge upon application to the CCDC (12 Union Road,
Cambridge CB2 1EZ U.K. Fax: (internat.) +44-1223/336-033;
E-mail: deposit @ccdc.cam.ac.uk). The habits of crystals
Lamellar racemic twinning of chiral ammonium salts 1 and
2 was found and investigated. Despite the chiral space group,
determined by an X-ray study of single crystals grown from
synthetic nonresolved material, the enantiomeric composition
of the crystals was racemic. Morphological study and molecular
modeling led to the identification of the twinning plane as {00l}.
Auxiliary salt 3 was designed to be stereospecifically absorbed
onto the {00l} faces, thus blocking them from effective growth
and concurrently avoiding twin formation.The crystalline frag-
ment along the twinning plane is probably formed by polyionic
growth units with a heterochiral composition existing in
supersaturated solution. To investigate clustering in solution
directly, we applied ESI FT-ICR mass spectrometry and detected
clusters containing up to five ammonium cations with four
anions. These agglomerates are possibly relevant to elementary
building blocks existing in solution.
Whereas methods for the investigation of nucleation processes
on the surface are well-established, the corresponding technique
for 3D bulk solutions is still lacking. The most promising
approach, as was shown above, is mass spectrometry. Keeping
in mind the key problem of the experimental registration of the
critical nucleus (hundreds of molecules) and leaving the
technical details for specialists, we can dream about an apparatus
that not only will have a soft ionization setup and a low-
temperature sample inlet but also will allow the delivery of large
and heavy clusters to the detector and so permit work with
concentrated (supersaturated) solutions. When working under
these conditions, we can expect little damage and maximum
retention of the structural features of solution associates
responsible for crystal growth.
In conclusion, the lamellar racemic twinning is an undesirable
obstacle to the resolution into enantiomers by crystallization
(entrainment methods, crystallization with deficiency of
conglomerator9a). The future understanding of this phenomenon
as well as the development of new and simple methods for its
prevention will noticeably increase the number of resolvable
compounds and will have an impact on an exiting area of crystal
nucleation research.

13530 J. Phys. Chem. B, Vol. 107, No. 48, 2003
Torbeev et al.
represented in Figures 5 and 6 were measured either on the
diffractometer or with an optical goniometer.
Weizmann Institute for the support that allowed him to work
at the Materials and Interfaces Department during the summer
of 2002. We acknowledge the help of Dr. Yu. A. Strelenko (N.
D. Zelinsky Institute of Organic Chemistry, Russian Academy
of Sciences) with NMR measurements. This work was supported
by the Russian Academy of Sciences (programs of academicians
O. M. Nefedov and Yu. A. Zolotov), the Russian Foundation
for Basic Research (RFBR) (grant nos. 03-03-32019, 03-03-
06526, 02-03-32311, and 03-03-04010), and INTAS (grant no.
99-00157). We also thank the two referees for their invaluable
comments on this paper.
Calculations of homo- and heterodimeric clusters were
performed with the Gaussian 98 program package35 at the
B3LYP level. Full optimization was carried out with the SDD
basis set starting from the X-ray structural data. As convergence
criteria, the extremely tight threshold limits of 2 × 10-6 and 6
× 10-6 au were applied for the maximum force and displace-
ment, respectively.
Compound 1, obtained as colorless crystals, was prepared
from 15 g (0.102 mol) of Me(All)NPh and 17.45 g (0.102 mol)
of PhCH2Br (one week at 22 °C, in the dark). The obtained
solid was first washed with benzene-hexane (1:1) for 1 day,
yield of crude product 29.85 g (92%). Then it was recrystallized
three times from water at 5 °C, yield 6.5 g (20%). Obtained
crystals were analytically pure, mp 165-166 °C (decomp).
Solubility (20 °C) in H2O: (w/w) 3.8%. Anal. Calcd for
Supporting Information Available: CIF file. This material
is available free of charge via the Internet at http://pubs.acs.org.
References and Notes
(1) Kostyanovsky, R. G.; Kostyanovsky, V. R.; Kadorkina, G. K.;
Lyssenko, K. A. MendeleeV Commun. 2003, 111.
1
C17H20NBr: C, 64.16; H, 6.33. Found: C, 64.21; H 6.28. H
(2) Chirality in Industry II. DeVelopment in the Commercial Manu-
facture and Applications of Optically ActiVe Compounds; Collins, A. N.,
Sheldrake, G. N., Crosby, J., Eds.; Wiley & Sons: Chichester, England,
1997; pp 81-180.
(3) Jacques, J.; Collet, A.; Wilen, S. H. Enantiomers, Racemates and
Resolutions; Krieger Publishing Company: Malabar, FL, 1994.
(4) (a) Coquerel, G.; Petit, M.-N.; Bouaziz, R. PCT Patent WO 95/
08522. (b) Ndzie´, E.; Cardinael, P.; Schoofs, A.-R.; Coquerel, G. Tetra-
hedron: Asymmetry 1997, 8, 2913.
(5) (a) Kinbara, K.; Tagawa, Y.; Saigo, K. Tetrahedron: Asymmetry
2001, 12, 2927. (b) Kinbara, K.; Hashimoto, Y.; Sukegawa, M.; Nohira,
H.; Saigo, K. J. Am. Chem. Soc. 1996, 118, 3441. (c) Kostyanovsky, R.
G.; Bronzova, I. A.; Lyssenko, K. A. MendeleeV Commun. 2002, 4.
(6) The resolution may be limited by the existence of an unstable
racemic compound: (a) Petit, M.-N.; Coquerel, G. MendeleeV Commun.
2003, 95. (b) Houllemare-Druot, S.; Coquerel, G. J. Chem. Soc., Perkin
Trans. 2 1998, 2211.
NMR (500.13 MHz, CD3OD): δ 3.44 (3H, s, Me), 4.54 (1H,
dd, Ha, 2Jab ) -13.4 Hz, 3Jac ) 7.7 Hz), 5.03 (1H, dd, Hb, 2Jab
3
) -13.4 Hz, Jbc ) 5.4 Hz), 5.21 (2H, dd, CH2-Ph, AB
2
spectrum, J ) -12.7 Hz, ∆ν ) 42.3 Hz), 5.68 (dd, 1H, Hd,
3
2Jde )2.0 Hz, Jdc )9.7 Hz), 5.69 (1H, m, Hc), 5.72 (1H, dd,
2
3
He, Jed ) 2.0 Hz, Jec ) 16.4 Hz), 7.08 (2H, d, o-Bn), 7.30
(2H, t, m-Bn), 7.46 (1H, t, p-Bn), 7.60 (3H, m, p,m-Ph), 7.72
(2H, m, o-Ph).
(7) (a) Furberg, S.; Hassel, O. Acta Chem. Scand. 1950, 4, 1020. (b)
Martin, R. H.; Marchant, M. J. Tetrahedron 1974, 30, 343. (c) Green, B.
S.; Knossow, M. Science 1981, 214, 795. (d) Davey, R. J.; Black, S. N.;
Williams, L. J.; McEwan, D.; Sadler, D. E. J. Cryst. Growth 1990, 102,
97. (e) Potter, G. A.; Garcia, C.; McCague, R.; Adger, B.; Collet, A. Angew.
Chem., Int. Ed. Engl. 1996, 35, 1666. (f) Berfeld, M.; Zbaida, D.;
Leiserowitz, L.; Lahav, M. AdV. Mater. 1999, 11, 328. (g) Zbaida, D.; Lahav,
M.; Drauz, K.; Knaup, G.; Kottenhahn, M. Tetrahedron 2000, 56, 6645.
(h) Addadi, L.; Weinstein, S.; Gati, E.; Weissbuch, I.; Lahav, M. J. Am.
Chem. Soc. 1982, 104, 4610. (i) Beilles, S.; Cardinael, P.; Ndzie, E.; Petit,
S.; Coquerel, G. Chem. Eng. Sci. 2001, 56, 2281. (j) Gervais, C.; Beilles,
S.; Cardinael, P.; Petit, S.; Coquerel, G. J. Phys. Chem. B 2002, 106, 646.
(8) Hager, O. H.; Llamas-Saiz, A. L.; Foces-Foces, C.; Claramunt, R.
M.; Lopez, C.; Elguero, J. HelV. Chem. Acta 1999, 82, 2213.
(9) Previously in our group, the elaboration of racemate resolution
procedures clashed with several cases of racemic twinning: (a) Kosty-
anovsky, R. G.; Torbeev, V. Yu.; Lyssenko, K. A. Tetrahedron: Asymmetry
2001, 12, 2721. (b) Kostyanovsky, R. G.; Kadorkina, G. K.; Lyssenko, K.
A.; Torbeev, V. Yu.; Kravchenko, A. N.; Lebedev, O. V.; Grintselev-
Knyazev, G. V.; Kostyanovsky, V. R. MendeleeV Commun. 2002, 6. (c)
Kostyanovsky, R. G.; Lyssenko, K. A.; Kravchenko, A. N.; Lebedev, O.
V.; Kadorkina, G. K.; Kostyanovsky, V. R. MendeleeV Commun. 2001,
134.
(10) This phenomenon was found to happen for other crystal symmetries
as well: (a) Heller, E.; Schmidt, G. M. J. Isr. J. Chem. 1971, 9, 449. (b)
Maggard, P. A.; Kopf, A. L.; Stern, C. L.; Poeppelmeier, K. R.; Ok, K.
M.; Halasyamani, P. S. Inorg. Chem. 2002, 41, 4852.
(11) Monaco, H. L.; Viterbo, D.; Scordari, F.; Giacovazzo, C. Funda-
mentals of Crystallography; Oxford University Press: New York, 1992;
pp 133-137.
(12) According to intuition, we might expect to observe the disorder
caused by the presence of two enantiomers in a crystal. We do not see it
because they are not statistically mixed but separated in large domains,
which are oriented to each other crystallographically perfectly. We like to
consider the composite planes between domains as fraction of metastable
racemic compound (see ref 22); however, the concentration of this fraction
is very low (approximately 10-3 % (w/w)), and the planes do not translate
themselves. Thus, the X-ray reflections for twinned crystals and pure
enantiomeric crystals differ only by an anomalous correction.
(13) (a) Flack, H. D. HelV. Chim. Acta 2003, 86, 905. (b) Flack, H. D.;
Bernardinelli, G. J. Appl. Crystallogr. 2000, 33, 1143. (c) Flack, H. D.;
Bernardinelli, G. Acta Crystallogr., Sect. A 1999, 55, 908. (d) Flack, H. D.
Acta Crystallogr., Sect. A 1983, 39, 876.
13C {1H} NMR (125.03 MHz, CD3OD): δ 46.81 (Me), 70.60
(-CH2-C2H3), 73.82 (CH2Ph), 123.26 (o-Ph), 125.86 (-CH)),
128.25 (ipso-Bn), 128.74 (CH2dC2H3-), 129.52 (m-Bn), 131.17
(m-Ph), 131.44 (p-Ph), 131.52 (p-Bn), 133.43 (o-Bn), 142.60
(ipso-Ph). The assignment of carbon signals to the appropriate
proton signals was performed by 2D HSQC and HMBC
correlations. UV (1.05 × 10-3 M, H2O), ꢀ(λ, nm): 2855 (203),
10 945 (220.6), 628 (262); CD spectrum of needlelike crystal
(1.1 × 10-3 M, EtOH), ∆ꢀ(λ, nm): 9.3 (212.5).
Compound 2, colorless crystals, precipitated from an aqueous
solution of 1 and an equimolar quantity of NaI with quantitative
yield, mp 143-147 °C (decomp), has the same NMR, UV, and
CD properties. Anal. Calcd for C17H20NI: C, 55.90; H, 5.52.
Found: C, 55.91; H, 5.48.
Compound 3, colorless crystals, was prepared from p-BrC6H4-
CH2Br and Me(All)NPh. The reaction was performed without
solvent for 1 week at room temperature. The obtained oily
product was solidified with dry ethyl acetate and then recrystal-
lized two times from acetonitrile/ethyl acetate (1:1), mp 99.5-
100.5 °C. Anal. Calcd for C17H19NBr2: C, 51.41; H, 4.82.
1
Found: C, 51.47; H, 4.98. H NMR (400.14 MHz, CD3OD):
δ 3.48 (3H, s, Me), 4.61 (1H, dd, Ha, 2Jab ) -13.1 Hz, 3Jac
)
2
3
6.5 Hz), 5.05 (1H, dd, Hb, Jab ) -13.1 Hz, Jbc ) 4.3 Hz),
2
5.14 (2H, dd, CH2-Ph, AB spectrum, J ) -12.8 Hz, ∆ν )
67.2 Hz), 5.52-5.67 (3H, m, Hd, Hc, He), 6.99 (2H, d, o-C6H4-
Br), 7.42 (2H, d, m-C6H4Br), 7.60-7.64 (3H, m, p,m-Ph), 7.77-
7.79 (2H, m, o-Ph).
Acknowledgment. We are grateful to Professor M. Lahav
(Weizmann Institute of Science, Rehovot, Israel) for valuable
discussions and advice. V.Y.T. gratefully acknowledges the

Lamellar Racemic Twinning
J. Phys. Chem. B, Vol. 107, No. 48, 2003 13531
(14) (a) Thomas, J. M.; Williams, J. O. Chem. Commun. 1967, 432. (b)
Williams, J. O.; Thomas, J. M. Trans. Faraday Soc. 1967, 63, 1720.
(15) Wedekind, E. Zur Stereochemie des Fu¨nfwertigen Stickstoffes: Mit
Besonderer Berucksightigung des Asymmetrischen Stickstoffes in der
Aromatischen Reihe; Verlag von Veit & Co.: Leipzig, Germany, 1899.
(16) (a) Wedekind, E. Ber. Dtsch. Chem. Ges. 1899, 32, 517. (b)
Wedekind, E. Ber. Dtsch. Chem. Ges. 1899, 32, 3561. (c) Wedekind, E.
Zur Stereochemie des Fu¨nfwertigen Stickstoffes: Mit Besonderer Beruck-
sightigung des Asymmetrischen Stickstoffes in der Aromatischen Reihe;
Verlag von Veit & Co.: Leipzig, Germany, 1899; pp 31-53.
(24) Davey, R. J.; Allen, K.; Blagden, N.; Cross, W. I.; Lieberman, H.
F.; Quayle, M. J.; Righine, S.; Seton, L.; Tiddy, G. J. T. CrystEngComm
2002, 4, 257-264.
(25) Dufour, F.; Gervais, C.; Petit, M.-N.; Perez, G.; Coquerel, G. J.
Chem. Soc., Perkin Trans. 2 2001, 2022.
(26) (a) Gaskell, S. J. J. Mass Spectrom. 1997, 32, 677. (b) Russell, K.
C.; Leize, E.; Van Dorsselaer, A.; Lehn, J.-M. Angew. Chem., Int. Ed. Engl.
1995, 34, 209. (c) Leize, E.; Van Dorsselaer, A.; Kra¨mer, R.; Lehn, J.-M.
Chem. Commun. 1993, 990. (d) Schalley, C. A.; Rivera, J. M.; Mart´ın, T.;
Santamar´ıa, J.; Siuzdak, G.; Rebek, J., Jr. Eur. J. Org. Chem. 1999, 1325,
5. (e) Schalley, C. A.; Mart´ın, T.; Obst, U.; Rebek, J., Jr. J. Am. Chem.
Soc. 1999, 121, 2133.
(17) (a) Pope, W. J.; Peachey, S. J. J. Chem Soc. 1899, 75, 1127. (b)
Pope, W. J.; Harvey, A. W. J. Chem. Soc. 1901, 79, 828.
(18) In the case of a conglomerate, the solubility of the racemic mixture
is approximately double that of the pure enantiomer (in an ideal case, the
solubility of one enantiomer is not modified by the presence of the other);
therefore, the crystals of the superfluous enantiomer first have to appear
from solution enriched by this enantiomer.
(19) Kauffman, G. B.; Bernal, I. J. Chem. Educ. 1989, 66, 293.
(20) (a) Pasteur, L. Ann. Phys. 1848, 24, 442. (b) Pasteur, L. Compt.
Rend. Acad. Sci. 1848, 26, 535. (c) Pasteur, L. Lec¸ons de Chemie Professe´s
en 1860; Soc. Chim. de Paris, Paris. (d) Pasteur, L. Ann. Chim. Phys. 1850,
28, 56. (e) Pasteur, L. Ann. Chim. Phys. 1852, 34, 30.
(27) (a) Gidaleviwtz, D.; Feidenhans’l, R.; Matlis, S.; Smilgies, D.-M.;
Christensen, M. J.; Leiserowitz, L. Angew. Chem., Int. Ed. Engl. 1997, 36,
955. (b) Davey, R. J.; Lin, W.; Quayle, M. J.; Tiddy, G. J. T. Cryst. Growth
Des. 2002, 2, 269.
(28) Yamamoto, Y.; Sato, Y.; Wakisaka, A. J. Chem. Soc., Chem.
Commun. 1994, 2727.
(29) (a) Takats, Z.; Nanita, S. C.; Cooks, R. G. Angew. Chem., Int. Ed.
2003, 42, 3521. (b) Koch, K. J.; Gozzo, F. C.; Nanita, S. C.; Takats, Z.;
Eberlin, M. N.; Cooks, R. G. Angew. Chem., Int. Ed. 2002, 41, 1721. (c)
Cooks, R. G.; Zhang, D.; Koch, K. J.; Gozzo, F. C.; Eberlin, M. N. Anal.
Chem. 2001, 73, 3646.
(30) (a) Hodyss, R.; Julian, R. R.; Beauchamp, J. L. Chirality 2001, 13,
703. (b) Counterman, A. E.; Clemmer, D. E. J. Phys. Chem. B 2001, 105,
8092.
(21) It is without a doubt that W. J. Pope clearly realized the nature of
Pasteur’s experiments, and he and his colleague F. S. Kipping made a
significant contribution to the further understanding of the spontaneous
resolution phenomenon. We refer readers to two references: (a) Kipping,
F. S.; Pope, W. J. J. Chem. Soc., Trans. 1898, 73, 606. (b) Kipping, F. S.;
Pope, W. J. J. Chem. Soc., Trans. 1897, 71, 989. The first one is dedicated
to an elegant investigation of the crystallization of sodium chlorate. The
second one is about their interpretation of the pseudoracemism phenomenon,
where there is confirmation that the authors had an idea about twinning
between enantiomers; however, they used this concept for the structural
description of pseudoracemates, which is not actually correct.
(22) In ref 7j, the authors proposed the use of the term lamellar epitaxy
instead of the term lamellar twinning for the description of cases of twinning
of opposite enantiomers of the same chemical compound. They based this
idea on the definition of J. D. Dunitz (Acta Crystallogr., Sect. B 1995, 51,
619) that enantiomeric crystals of a conglomerate and a racemate are three
different chemical substances if the enantiomerization rate is low and can
be considered to be polymorphs if the enantiomerization rate is high. It is
seems that the authors solved the confusion because even the same
enantiomer can form lamellar twinned crystals, whereas the crystal growth
of one substance on crystals of the other it is accepted as epitaxy. However,
the confusion arises again if we take into consideration the fact that under
certain conditions the enantiomerization rate can be high (for example, the
presence of a catalyst or high temperature, which can increase the
enantiomerization rate), whereas under another set of conditions the rate
will be low. Therefore, in the first case the term twinning is appropriate,
but in the second we should use the term epitaxy. We are inclined to take
another point of view. Racemate-conglomerate polymorphism takes place
when crystallization occurs from solution or in a melt of racemic
composition and molecules or ions can assemble equiprobably into racemic
crystals (racemic compound) or mechanical mixture of left- and right-handed
crystals (conglomerate). To distinguish the ordinary twinning of crystals
of the same enantiomer from the twinning of alternating lamellae of two
opposite enantiomers leading to crystals of racemic composition, we suggest
that the term lamellar racemic twinning be used. We refer readers also to
ref 13a, where the term twinning by inVersion for twinning between the
two opposite enantiomers is proposed.
(31) Julian, R. R.; Hodyss, R.; Kinnear, B.; Jarrold, M. F.; Beauchamp,
J. L. J. Phys. Chem. B 2002, 106, 1219.
(32) The references for works dealing with salt ion clusters are in Zhang,
D.; Cooks, R. G. Int. J. Mass Spectrom. 2000, 195/196, 667.
(33) Although ESI MS is widely used for the investigation of self-
assembly processes in solution,26 we think that this technique is of limited
use for crystal nucleation studies. Other types of equipment such as cold-
spray MS (ref 36) or sonic-spray MS (ref 37) are more promising.
(34) In our case, we cannot draw a direct conclusion about whether the
registered clusters are exactly the associates existing in the supersaturated
solution, but we suppose that during the electrospraying of diluted sample
solutions the particular step of the evolution of the evaporating drop exists,
where the conditions imitate those existing in the supersaturated solution.
Moreover, in our case, there were no side ions (H+, Na+, etc.), which are
usually added for the purpose of charging neutral molecules and that actually
serve as artificial promoters for clusterization.
(35) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb,
M. A.; Cheeseman, J. R.; Zakrzewski, V. G.; Montgomery, J. A., Jr.;
Stratmann, R. E.; Burant, J. C.; Dapprich, S.; Millam, J. M.; Daniels, A.
D.; Kudin, K. N.; Strain, M. C.; Farkas, O.; Tomasi, J.; Barone, V.; Cossi,
M.; Cammi, R.; Mennucci, B.; Pomelli, C.; Adamo, C.; Clifford, S.;
Ochterski, J.; Petersson, G. A.; Ayala, P. Y.; Cui, Q.; Morokuma, K.; Malick,
D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Cioslowski, J.;
Ortiz, J. V.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi,
I.; Gomperts, R.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.;
Peng, C. Y.; Nanayakkara, A.; Gonzalez, C.; Challacombe, M.; Gill, P. M.
W.; Johnson, B. G.; Chen, W.; Wong, M. W.; Andres, J. L.; Head-Gordon,
M.; Replogle, E. S.; Pople, J. A. Gaussian 98, revision A.7; Gaussian,
Inc.: Pittsburgh, PA, 1998.
(36) (a) Yamaguchi, K. J. Mass Spectrom. 2003, 38, 473. (b) Sakamoto,
S.; Fujita, M.; Kim, K.; Yamaguchi, K. Tetrahedron 2000, 56, 955. (c)
Sakamoto, S.; Nakatani, K.; Saito, I.; Yamaguchi, K. Chem. Commun. 2003,
788.
(23) Weissbuch, I.; Lahav, M.; Leiserowitz, L. Cryst. Growth Des. 2003,
3, 125.
(37) Takats, Z.; Nanita, S. C.; Cooks, R. G.; Schlosser, G.; Vekey, K.
Anal. Chem. 2003, 75, 1514.