Pasteur's Tweezers revisited: On the mechanism of attrition-enhanced deracemization and resolution of chiral conglomerate solids

Article abstract of DOI:10.1021/ja303566g

Insights into the mechanism of attrition-enhanced deracemization and resolution of solid enantiomorphic chiral compounds are obtained by crystal size and solubility measurements and by isotopic labeling experiments. Together these results help to deconvolute the various chemical and physical rate processes contributing to the phenomenon. Crystal size measurements highlight a distinct correlation between the stochastic, transient growth of crystals and the emergence of a single solid enantiomorph under attrition conditions. The rapid mass transfer of molecules between the solution and solid phases under attrition is demonstrated, and the concept of a crystal-size-induced solubility driving force is exploited to overcome the stochastic nature of the crystal growth and dissolution processes. Extension to non-racemizing conditions provides a novel methodology for chiral resolution. Implications both for practical chiral separations and for the origin of biological homochirality are discussed.

Full text of DOI:10.1021/ja303566g

Article
pubs.acs.org/JACS
Pasteur’s Tweezers Revisited: On the Mechanism of Attrition-
Enhanced Deracemization and Resolution of Chiral Conglomerate
Solids
Jason E. Hein,†,‡ Blessing Huynh Cao,‡ Cristobal Viedma,§ Richard M. Kellogg,⊥
and Donna G. Blackmond*,†
†Department of Chemistry, The Scripps Research Institute, La Jolla, California 92037, United States
‡Department of Chemistry, University of California, Merced, California 95344, United States
§Departamento Crystalografia-Mineralogia, Facultad Geologia, Universidad Complutense, 28040 Madrid, Spain
⊥Syncom BV, Kadijk 3, 9747 AT Groningen, The Netherlands
S
* Supporting Information
ABSTRACT: Insights into the mechanism of attrition-enhanced
deracemization and resolution of solid enantiomorphic chiral compounds
are obtained by crystal size and solubility measurements and by isotopic
labeling experiments. Together these results help to deconvolute the
various chemical and physical rate processes contributing to the
phenomenon. Crystal size measurements highlight a distinct correlation
between the stochastic, transient growth of crystals and the emergence of a
single solid enantiomorph under attrition conditions. The rapid mass
transfer of molecules between the solution and solid phases under attrition
is demonstrated, and the concept of a crystal-size-induced solubility driving force is exploited to overcome the stochastic nature
of the crystal growth and dissolution processes. Extension to non-racemizing conditions provides a novel methodology for chiral
resolution. Implications both for practical chiral separations and for the origin of biological homochirality are discussed.
The potential for development of an efficient separation/
deracemization methodology in a pharmaceutical context has
also been noted.8 For example, slurry mixtures of compound 1,
a conglomerate imine derivative of 2-Cl-phenylglycine, the (S)
enantiomer of which makes up the chiral component of the
blockbuster drug Clopidogrel (Plavix), undergoes interconver-
sion between the two enantiomorphic solids with vigorous
stirring under attrition or sonication in the presence of an
organic base (Scheme 1). This system has been studied
extensively in detailed mechanistic and scale-up studies of the
attrition-enhanced deracemization process.11
INTRODUCTION
■
Ever since Viedma’s1 striking demonstration that attrition of a
racemic slurry of enantiomorphic solids promotes an inexorable
emergence of solid-phase homochirality, this phenomenon has
been the subject of intense interest from both practical and
fundamental perspectives. The phenomenon was initially
applied to molecules such as NaClO3 that are achiral in
solution but exist as two mirror-image solids, or enantiomorphs,
and whose phase behavior has intrigued scientists from the time
of van’t Hoff.2,3 The idea that this concept could be extended to
intrinsically chiral molecules that can be made to interconvert
in solution (Scheme 1) was recently realized,4,5 and
implications for the evolution of biological homochirality
from a racemic prebiotic mixture of enantiomers have been
discussed.4−10
Scheme 1. Evolution of Solid-Phase Homochirality from a
Mixture of Racemizable Chiral Conglomerate Crystals
At the same time, ongoing research seeks a better
understanding of the physical and chemical processes under-
lying the phenomenon. In addition to further experimental
studies on a variety of systems,11,12 several types of
mathematical models have been presented, including Monte
Carlo simulations of the process13 and kinetic modeling of
Received: April 13, 2012
© XXXX American Chemical Society
A
dx.doi.org/10.1021/ja303566g | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
reaction-type networks.14 A recent report15 based on
population balance modeling helps to overcome limitations of
these earlier efforts by providing a more accurate mathematical
description of the processes thought to be involved in the
deracemization, including solution racemization, crystal attri-
tion and agglomeration, and crystal growth and dissolution
caused by crystal-size dependence of solubility.
It is generally accepted that the deracemization mechanism
requires (i) chemical racemization in solution, (ii) crystal
growth, and (iii) mechanical or thermal energy input to
instigate crystal dissolution, typically attrition by grinding. The
latter two are physical processes that influence the solubility of
the system as dictated by the Gibbs−Thomson rule (larger
crystals exhibit lower solubility than smaller crystals). Crystal
growth occurs by Ostwald ripening (growth of larger crystals at
the expense of smaller crystals), and mechanical grinding or
sonication enhances crystal attrition. When enantiomorphic
crystals of a chiral molecule are mixed together as a slurry in the
presence of a racemization catalyst, these chemical and physical
processes combine to allow the system to evolve to a single
solid enantiomorph. A central mechanistic question that has
been posed is whether attrition-enhanced deracemization may
be rationalized by consideration of solubility and crystal growth
stemming from tenets of the Gibbs−Thomson rule, as was
originally proposed by Viedma7 and by Blackmond,8 or
whether additional driving forces are required to explain the
temporal evolution of solid-phase enantiomeric excess (ee).
In this report we present experimental results that help in
deconvolution of the various chemical and physical rate
processes that occur during attrition-enhanced deracemization.
Careful solubility and crystal size measurements offer evidence
for the role of crystal-size-induced solubility differences over the
course of the deracemization process. Isotopic labeling aids in
the deconvolution of the physical mass transfer from the net
conversion of molecules from one enantiomorph to the other.
This work provides further insights into the mechanism of
attrition-enhanced deracemization and resolution of chiral
conglomerate solids and proposes a novel modification of the
protocol for practical and reproducible chiral separations.
Figure 1. Evolution of solid-phase enantiomeric excess (%CEE) of rac-
1 over time during 24 attrition-enhanced deracemization experiments
carried out under identical conditions. Dotted lines are given as a
guide to the eye.
present work, the nearly equal number of conversions to
opposite enantiomorphs suggests that any influence from chiral
impurities is small.17 It is important to note that the outcome of
the mathematical model is sensitive to the specific parameter
values chosen and to the particular rules employed for how
crystals break and how crystals grow. In addition, the model
does not account for any variability in how these physical
processes occur under actual experimental conditions. Thus, it
is reasonable to assume not only that identical initial conditions
of mass and crystal size distribution may be difficult to achieve
experimentally, but also that, under macroscopic experimental
attrition conditions, not all crystal-breaking events for any given
crystal size occur in an identical manner, as they do in the
mathematical model. This factor may help account for
differences in the temporal profiles for evolution of solid-
phase homochirality as well as the apparent randomness in the
sense of the outcome shown in Figure 1.
Three of the experiments from Figure 1 are analyzed in
further detail in Figures 2 and 3. Two of the three experiments
achieved solid-phase homochiralityin opposite sensesin an
accelerated manner within 5 h, while the third sample remained
racemic even after 10 h (Figure 1). Crystal size distributions
were measured for all three experiments at times designated by
arrows in Figure 2, namely at the outset of the deracemization
experiment and after 3, 5, and 8 h.18 Although the three
samples exhibit similar, non-Gaussian crystal size distributions
at the outset, intriguingly, it was found that a broadening of the
distribution commenced at the onset of the evolution of solid-
phase CEE at 3 h and endured through the emergence of
homochirality at ca. 5 h for series b and series c (Figure 2). By
contrast, the sample that remained racemic (series a) retained a
narrow size distribution and smaller average size throughout the
grinding/sonication process. In addition, Figure 2 shows that,
after the system converts to a single solid enantiomorph, further
attrition returns the system to its original crystal size
distribution. Thus, a transient growth in crystal size is
implicated in the accelerating profile for the evolution of
solid-phase homochirality. The crystals are irregularly shaped,
as shown in Figure 3, and their size is calculated as the
rectangular crystal area as defined by the two-dimensional
perimeter, as is common practice in measurements of biological
samples.19 This deviation from ideal spherical crystals may also
rationalize differences between experimental %CEE profiles and
that predicted for ideal spherical crystals in mathematical
models such as ref 15.
RESULTS AND DISCUSSION
■
Effect of Crystal Size on Deracemization. Attrition-
enhanced deracemization was carried out by sonication of
racemic mixtures of compound 1 in MeCN in the presence of
glass beads, using catalytic 1,8-diazabicyclo[5.4.0]undec-7-ene
(DBU) to induce solution-phase racemization. Crystal
enantiomeric excess (%CEE) was monitored over time for 24
experiments monitored for up to 22 h. Figure 1 shows the
results of experiments carried out under apparently identical
conditions and underscores the variability of the outcome: the
24 runs produced nine cases of homochiral R-1, eight of
homochiral S-1, and seven that remained racemic. The time
required for the evolution of solid-phase ee and the form of the
profile differed even for seemingly identical initial conditions.16
The apparent randomness of the deracemization process for
equal initial masses of two enantiomorphic crystals has been
reported previously.12 By contrast, however, the mathematical
model reported by Iggland and Mazzotti15 demonstrated that,
in a system exhibiting perfect initial symmetry, symmetry will
be conserved. These authors suggested that the observation of
apparent randomness instead may indicate either the existence
of undetectable initial asymmetry in crystal size or CEE, or
alternatively the presence of trace chiral impurities. In the
The link between an increasing crystal size and the evolution
of solid-phase homochirality is in accord with the original
B
dx.doi.org/10.1021/ja303566g | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 3. Optical microscopy showing crystals from series a (left) and
series b (right) after 5 h of enhanced attrition/deracemization, from
which the crystal size distributions shown in the middle of Figure 2
were determined.16
solubility was measured in slurries of crystals in the larger size
range (shown in Figure 2 after 5 h) compared to those with the
initial narrower distribution, as is predicted by the Gibbs−
Thomson rule. For series b and series c in Figure 2, the smaller
crystals dissolve and feed the growing larger crystals as
predicted by Ostwald ripening. In these cases the movement
from small to large crystals is accompanied by a net movement
of mass from one enantiomorph to the other via the conduit of
solution-phase racemization. The experimental results do not
reveal why crystal growth occurred in these two cases nor in the
third, but these observations imply that the deracemization
process is instigated by what might be considered an “internal
seeding” event whereby larger crystals are stochastically formed
and temporarily sustained during a period of rapid mass
movement from one enantiomorph to the other estimated at
ca. 1 mg/min during this critical period.18 This suggests that
the accelerated deracemization rate is limited by the rate at
which molecules are able to dissolve from the enantiomorph
being depleted. A net movement of molecules from small to
large crystals is set up by this temporary perturbation of the
reversible physical equilibrium of the dissolution process. The
driving force of net dissolution for small crystals together with
the thermodynamic driving force of crystal growth provide self-
reinforcing effects that rationalize the autoinductive profile of
crystal ee vs time.
Such a perturbation of the equilibrium between crystal
dissolution and growth may be engineered by introducing a
small quantity of large crystals of one hand to a racemic slurry
of crystals under deracemization conditions. This is shown in
Figure 4. When a small quantity of large crystals of R-1 was
Figure 2. %CEE curves of three attrition-enhanced deracemization
experiments (top) and crystal area distributions (bottom) measured
for solid-phase samples at the time points noted by arrows (top).
Crystal size distribution (CSD) is measured as the two-dimensional
surface area of the non-spherical crystals.16
Figure 4. Normalized solution concentration of 1 as a function of time
monitored by in situ FTIR spectroscopy for rac-1 with DBU under
enhanced attrition conditions. Seed crystals of R-1 added at time
shown to give solid-phase ee (%CEE) = 7%. Transient drop in
solution concentration and evolution of CEE are shown.16
proposals of the Gibbs−Thomson rule for crystal size-
dependent solubility and Ostwald ripening as the driving
mechanism behind the deracemization process. A ca. 10% lower
C
dx.doi.org/10.1021/ja303566g | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
introduced into a racemic system under attrition enhanced
conditions (ca. 2 wt %, to give an overall CEE of ca. 7% R-1),
the solution concentration of 1 dropped immediately and
rapidly by ca. 8%, followed by a slower recovery over time.
Figure 4 shows that, concomitant with the drop in solution
concentration, this seeding event resulted in a rapid evolution
of solid-phase ee toward R-1, attaining 57% CEE measured 1 h
after the seed crystals were added.16
The stochastic nature of the process is intriguing, and the
parameters that control or trigger the crystal growth/broad-
ening and the concomitant rapid evolution of solid-phase ee
shown in Figures 1−3 are not well understood. Experiments
carried out under what appear to be identical experimental
conditions, resulting in homochirality in opposite senses in
some cases and not at all in others, imply that whether crystals
grow, and which enantiomorph will grow, depends both on
possible minute initial asymmetry in mass or crystal size as well
as on variability in the attrition process and is difficult to predict
quantitatively for any particular attrition experiment.
Isotopic Labeling Studies. As a further probe of how
attrition influences the reversible processes of molecule
dissolution from and attachment to crystals, we designed
experiments aimed at “tagging” molecules to monitor and
deconvolute these mass transfer processes from the overall
deracemization. Enantiopure compound 1 of both hands was
synthesized with 15N label at the amide nitrogen, as described
in the Supporting Information, in order to monitor the physical
movement of molecules between the solution and solid phases
and in their conversion between the two enantiomorphic solids.
In a first set of experiments, saturated CH3CN solutions of
14N-rac-1 were mixed with solid 15N-rac-1 to form a slurry. The
reverse mixture was also prepared (isotopically 15N-labeled 1 in
solution phase). The mixtures were prepared such that the total
mass of 1 in solution at the outset is approximately equal to the
mass of 1 in the solid phase, such that the overall system
contains an equal number of 14N and 15N molecules.
Monitoring the isotopic label in each phase allows tracking of
the net movement of molecules over time. Thus, upon
complete mixing, the system will exhibit a ca. 50:50 mixture
of 14N/15N in both solution and solid phases. The system was
allowed to mix over time under non-racemizing conditions both
with attrition by glass beads and without attrition, as illustrated
in Figure 5 (top). The plot in Figure 5 (bottom) shows the
physical movement of mass between solution and solid phases
in the absence of the chemical influence of solution-phase
racemization.
Solution- and solid-phase samples were acquired over time,
and 14N/15N ratios were measured by high-resolution LC/MS
integrating the ion patterns of the M+H and M+Na peaks.16
Figure 5 shows that the mixing between phases is much more
rapid under attrition conditions, with the labeled molecules
being almost fully distributed between the two phases within 6
h. However, complete exchange of solution- and solid-phase
molecules occurs over time under either mixing protocol.
These studies demonstrate that for systems undergoing
attrition with glass beads, the complete exchange of mass
between solid and solution phases under non-racemizing
conditions occurs over a relatively short period of time. This
suggests that crystals have been ablated to such an extent that
all solid-phase molecules, even those at the very interior of
crystals at the outset, have been exposed to the solution phase.
This net movement of mass occurs on approximately the same
time scale as the net conversion of one enantiomorph to the
Figure 5. Monitoring 15N content to follow mass transfer between
solution (blue) and solid (gold) phases under nonracemizing
conditions from the experiment shown (top) using isotopically labeled
rac-1. Starting conditions: equal masses of 14N-rac-1 in solution and
15N-rac-1 in solid phase at outset. Graph shows 15N content of rac-1
molecules as they appear in the solution (blue symbols) and disappear
from the solid (gold symbols) over time. Open symbols, no attrition;
filled symbols, with attrition.16 Dotted lines are given as a guide to the
eye.
other under conditions where a racemization catalyst is present
in the solution phase. Thus it is probable that each solid-phase
molecule may have exchanged even multiple times with
solution-phase molecules over the course of deracemization
experiments such as those shown in Figure 1.
We next explored this movement of mass between solid and
solution phases when it is accompanied by the other rate
processes critical to attrition-enhanced deracemization, which
include the chemical rate of interconversion between the
enantiomers in solution and the net rate of conversion from
one solid enantiomorph to the other. Experiments were carried
out starting with racemic mixtures of enantiomorphic solids in
saturated aqueous solution, with conditions as described in
Figure 6 (top). In this set of experiments, one solid
enantiomorph is comprised of molecules with the 15N label
and the other with 14N. Thus, in the saturated solution at the
outset of the experiment, all solution molecules of one
enantiomer are labeled 14N and the other enantiomer 15N, as
is the case for the solids. The vessel was subjected to sonication
in the presence of glass beads, and the organic base DBU was
then added, providing conditions identical to those of the
experiments that were shown in Figure 1, where the
racemization half-life in solution has been measured to be on
the order of minutes. The solid phase was sampled over time
and subjected to high-resolution chiral LC/MS to determine
the 15N content of each enantiomer.16
D
dx.doi.org/10.1021/ja303566g | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
from one solid enantiomorph to the other in the deracemiza-
tion process. The results of Figures 1−4 show that the
evolution of CEE commences concomitant with a transient
crystal-size-induced solubility difference between the two
enantiomorphs. The results in Figures 5 and 6 show that
while the net solution−solid physical movement of molecules
and the solution−solution chemical interconversion of
enantiomers are facile processes, the stochastic manner in
which an attrition-induced solubility difference is established
makes it difficult to control or predict if and when such a crystal
growth/solubility driving force will be sustained long enough
for the evolution of solid-phase ee to commence.
Crystal-Size-Induced Solubility Driving Force. We
reasoned that the process triggering the evolution of solid-
phase ee would be reliably reproducible if a means could be
found to maintain the crystal-size-induced solubility gradient.
Sustaining a population of larger crystals of one of the
enantiomorphs requires a reliable means of preventing their
attrition, which is difficult to control in the presence of crystals
of the other enantiomorph. This led to the idea of creating a
physical separation between two solid-phase environments:
one, a racemic slurry of crystals under attrition-enhanced
solubility conditions, and two, a mildly stirred, lower solubility
enantiopure “seed” population of single handedness in a
separate vessel. The two vessels communicate without
movement of solids, solely by circulation of the liquid phase
between the vessels. A deracemization protocol based on this
concept is illustrated in Figure 7.
In step 1 of Figure 7, a racemic mixture of crystals of
compound 1 is stirred vigorously in the presence of glass beads
in the left vessel, while a second pot containing a saturated
solution of a small amount of S-1 seed crystals is mildly stirred.
The entire system is maintained at constant temperature, with
solution-phase racemization and crystal-size-induced solubility
differences being the sole driving forces for the net conversion
of one solid enantiomorph to the other. Over time (ca. 16 h)
and with circulation of only the liquid phase between the two
vessels, the entire mass of rac-1 is moved from the left-side
vessel and converted to solid S-1 in the right-side vessel, leaving
only the glass beads and a saturated racemic solution behind in
the left-side vessel.
In step 2 of Figure 7, glass beads are added to the vessel
containing the mass of newly converted S-1, which is stirred
vigorously with liquid circulation connecting it to a new vessel
containing seeds of R-1 that are mildly stirred in the absence of
glass beads. In ca. 16 h this system is in turn fully converted
from solid S-1 in the right-hand vessel to solid R-1 in the left-
hand vessel. The procedure is repeated in step 3, where solid R-
1 is converted back to solid rac-1. The entire process is
isothermal, with only dissolved molecules and no solids moving
between the vessels. The interconversion between solid
enantiomorphs relies solely on the differential enantiomer
solubility induced by crystal size differences promoted by
attrition in one vessel and mild stirring in the other vessel, as
dictated by the Gibbs−Thomson rule, with solution race-
mization serving as the conduit between the two enantiomor-
phic solids.
Figure 6. Percentage 15N in R or S solid enantiomorph (open circles
and squares) and solid-phase ee (filled circles and squares) as a
function of time for two attrition-enhanced deracemization experi-
ments carried out using initially racemic solids under conditions
identical to those shown in Figure 1. Experiment 1 (circles): S-solid
initially labeled with 15N. Experiment 2 (squares): R-solid initially
labeled with 15N.16
The open symbols in the plot of Figure 6 (bottom, left axis)
show that under the combined processes of solution
racemization and sonication-attrition, the exchange of mole-
cules between the two hands of the molecule and between the
two solid enantiomorphs is rapid, reaching complete
homogeneity within 3−4 h. In four separate experiments
(two shown in Figure 6; all four shown in Figure 1), 15N
molecules from one enantiomorph dissolve into solution,
convert to the opposite enantiomer, and attach to the other
enantiomorph in a rapid, reversible, and reproducible manner.20
Interestingly, however, Figure 6 also shows that, while mass
transfer occurs in a consistent manner, the net movement
converting one solid enantiomorph to the other remains a
random process, as indicated by the solid-phase ee.
Homochirality emerged in one case while the other remained
racemic. Of a total of four runs carried out using isotopically
labeled racemic solids under these conditions, two achieved
homochirality toward S-1 and the other two remained
racemic.16 Thus, the physical and chemical exchanges of
molecules in solution and solid phases cannot be correlated in a
simple manner with the net movement of enantiomeric
molecules from one solid enantiomorph to the other.
The overall three-step process thereby converts rac-1 to rac-1
in 86% yield, with losses being due to sampling and to solid
retention on the filters.16 While the conversion of rac-1 to rac-1
is not a practically useful result, this experiment serves to
demonstrate the remarkable ease with which a predictable
crystal-size-induced solubility difference can drive the attrition-
Taken together, these results demonstrate the ease with
which molecules move between solution and solid phases under
attrition-enhanced stirring or sonication, and at the same time
they highlight the randomness of the overall net conversion
E
dx.doi.org/10.1021/ja303566g | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 7. Schematic of iterative process of attrition-enhanced deracemization under solution-phase racemizing conditions (0.1 M DBU) in a two-
vessel setup held under isothermal conditions at 20 °C. Solution phase is circulated between the left and right vessels at ca. 1 mL/min while
movement of solids is prevented by 0.2 μm filters. Step 1: left vessel containing 200 mg of solid rac-1 and glass beads in saturated CH3CN solution is
rapidly stirred, while right vessel containing 10 mg of seed crystals of S-1 is mildly stirred. After ca. 24 h, rac-1 solid was completely converted to S-1
solid and was transferred to the right vessel via the solution phase, leaving only solution phase and glass beads in the vessel on the left. Step 2: glass
beads are added to vessel containing S-1 from step 1, which is connected to a new, mildly stirred vessel on the left containing 10 mg of seed crystals
of R-1. After ca. 8 h, all solid was transferred to the left vessel via the solution phase and was completely converted to solid R-1. Step 3: glass beads
are added to vessel containing R-1 from step 2, which is connected to a new, mildly stirred vessel on the right containing 10 mg of seed crystals of
rac-1. After ca. 24 h, all solid was transferred to the left vessel via the solution phase and was completely converted back to solid rac-1. The 86%
overall solid-phase yield includes solid losses due to sampling and adherence to the filters.16
enhanced deracemization process in a controlled manner. This
is in stark contrast to the stochastic nature of the process
observed in the single-vessel protocol, where establishment of a
transient crystal-size-induced solubility difference between
enantiomorphs is correlated with the evolution of solid-phase
ee. The protocol shown in Figure 6 utilizes this solubility
driving force while eliminating the randomness observed in the
one-pot deracemization.
Attrition-enhanced deracemization has typically been carried
out using a small excess of one enantiomorph in order to
provide a bias to help overcome the stochastic nature of the
process. However, results showing that a slight bias in CEE can
be overcome by a bias toward larger crystals of the opposite
enantiomorph highlight the sensitive interplay between the
driving forces presented by a population advantage and by a
size advantage: a larger number of smaller crystals of one
enantiomorph can be made to evolve to the opposite
enantiomorph, which initially contained a smaller total number
of molecules exhibiting a larger average crystal size.
Experimental results to date demonstrating the success of
“size over numbers” are limited to cases with a small initial bias,
however, and certainly the conversion of an overwhelming
population of one enantiomorph to what was initially only a
few crystals of the otherthe transformation shown in step 2
of Figure 7is unprecedented in attrition-enhanced deracem-
ization results reported to date.
enantiomers in such cases. Figure 8 demonstrates this protocol
for an initially racemic mixture of enantiomorphic crystals of
Pasteur’s tartrate, 2.
A flask containing equal amounts of solid sodium ammonium
L- and D-tartrate crystals in a saturated aqueous solution
together with 3 mm glass beads was connected to a second
flask, with liquid circulation between the two vessels at ca. 1
mL/min, while the movement of solids was prevented by filters
(Figure 8). To the second flask was added 50 mg of larger D-
tartrate crystals. The flask containing crystals and beads was
stirred vigorously while the second vessel was stirred only
mildly. The system was held isothermally at 13 °C. Over time
an increase in the amount of solid in the seed crystal vessel
became apparent. The experiment was terminated at 60 h, and
the solid phase in each vessel was collected, weighed, and
dissolved in water for optical rotation measurements. The
sample from the vessel originally containing an equal mixture of
L and D crystals measured >98% ee L-tartrate, and the seed
vessel, now containing more than an order of magnitude greater
mass of crystals than at the outset, measured >99% ee D-
tartrate. Solid mass recovery was ca. 70%, with losses due to
sampling and adherence to the filters.16
Crystals from the attrition flask (Figure 8, left) exhibit much
smaller average size than those collected from the seed flask
(Figure 8, right), but both show the distinct hemihedral shape
reported by Pasteur.16 Mass transfer and growth of larger D
crystals on the right occurs at the expense of the smaller, more
readily dissolvable crystals undergoing attrition on the left. In
the absence of primary nucleation of new L crystals in the right
flask, only D molecules are able to add to the solid phase in the
right flask. The success of the experiment indicates that primary
nucleation is hindered by control of the level of supersaturation
in the right flask. This net movement of mass via the liquid
phase results ultimately in the quantitative physical separation
From Deracemization to Resolution. The majority of
chiral molecules that form enantiomorphic solids do not
undergo solution racemization in a facile manner. For example,
sodium ammonium tartrate 2, the conglomerate system
separated manually by Pasteur using tweezers,20 does not
racemize in the presence of acid or base. We envisioned that
the concept of crystal-size-induced solubility differences might
be extended to the separation of, rather than conversion between,
F
dx.doi.org/10.1021/ja303566g | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 9. Extension of crystal-size-induced solubility driving force to
the separation of non-racemizing conglomerate crystals, employing a
temperature gradient and a continuous three-pot protocol.16
slurry of the racemic D and L solids under attrition conditions at
35 °C. The mass of each pure enantiomorphic solid builds up
in the respective mildly stirred vessels while the racemic solid is
depleted in the third vessel. Over the course of 48 h, rac-3 was
added to the central flask, allowing the process to operate in
continuous fashion. Eight hours after stopping the central feed,
no solids remained in the central vessel. Enantiopure L-3 and D-
3 were isolated from the left and right receiver flasks in 76%
and 79% yield, respectively, demonstrating a practical and
efficient resolution of the enantiomorphic solids.16
In addition to the clear practical application to chiral
separations demonstrated here, the concept of a crystal-size-
induced solubility driving force may have implications for
studies aimed at understanding the origin of biological
homochirality. The concept of separation of conglomerate
crystals has been discussed in this context since Pasteur’s time.
Welch24 introduced the idea of “stochastic sorting”, in which
simple physical processes (wind and waves, coupled with rain
and evaporation) might randomly partition crystals to create
local regions of single chirality, even in a globally racemic
environment. Upon dissolution of crystals from such regions,
local pools of enantioenriched molecules might be formed,
which could then undergo chemical reactions leading to
molecules of higher complexity and sustainable enantioenrich-
ment. However, in the absence of further driving forces to
promote enantiomer separation, competing processes of re-
equilibration to a locally racemic environment may be sufficient
to hinder enantioenrichment. The crystal-size-induced solubil-
ity gradient acting as a separation driving force, as described in
the current work, could enhance these simple stochastic sorting
processes toward the establishment of regions exhibiting a
single solid enantiomorph.
Figure 8. Extension of crystal-size-induced solubility driving force to
the separation of non-racemizing conglomerate crystals.16
of the enantiomorphic crystals. A crystal-size-induced solubility
gradient serves the role of Pasteur’s tweezers.
A complementary means to establish solution concentration
gradients for separation of enantiomers is to employ temper-
ature programming as is typically carried out in preferential
crystallization methods.20,21 However, in many practical
examples of non-racemizable conglomerates, the use of a
temperature gradient to effect solubility differences may be
problematic.22 The attainment of equilibrium solubility
conditions can be slow; a suitable, stable supersaturation zone
may be difficult to identify; and primary nucleation of crystals
of the wrong hand can become difficult to avoid, especially for
highly concentrated solutions of highly soluble compounds.
The crystal-size-induced solubility driving force may present a
more reliable means of controlling the process than the use of
temperature alone. A combination of temperature- and crystal-
size-induced solubility driving forces may permit the resolution
of difficult-to-resolve conglomerate systems. We demonstrated
the combination of temperature- and crystal-size-induced
solubility driving forces in this fashion for the proteinogenic
amino acid threonine, 3, which forms conglomerate solids and
is not readily racemizable. Figure 9 demonstrates this case,
together with a further practical modification of this protocol
using a three-pot system.
Two mildly stirred vessels respectively containing seeds of D
and L enantiomorphic solids of 3 at 20 °C are connected via a
common circulating solution phase to a third pot containing a
G
dx.doi.org/10.1021/ja303566g | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Noorduin, W. L.; Meekes, H.; van der Enckevort, W. J.; Kaptein, B.;
Vlieg, E.; Kellogg, R. M. Org. Process Res. Dev. 2009, 13, 1195−1198.
(12) (a) Noorduin, W. L.; Bode, A. A.; van der Meijden, M.; Meekes,
H.; van Etteger, A. F.; van Enckevort, W. J.; Christianen, P. C.;
Kaptein, B.; Kellogg, R. M.; Rasing, T.; Vlieg, E. Nature Chem. 2009, 1,
729−732. (b) Noorduin, W. L.; Meekes, H.; van Enckevort, W. J.;
Millemaggi, A.; Leeman, M.; Kaptein, B.; Kellogg, R. M.; Vlieg, E.
Angew. Chem., Int. Ed. 2008, 47, 6445−6447. (c) Kaptein, B.;
Noorduin, W. L.; Meekes, H.; van Enckevort, W. J.; Kellogg, R. M.;
Vlieg, E. Angew. Chem., Int. Ed. 2008, 47, 7226−7229. (d) Noorduin,
W. L.; van Enckevort, W. J.; Meekes, H.; Kaptein, B.; Kellogg, R. M.;
Tully, J. C.; McBride, J. M.; Vlieg, E. Angew. Chem., Int. Ed. 2010, 49,
8435−8438.
CONCLUSIONS
■
The results reported here help to deconvolute the various
chemical and physical rate processes that occur during attrition-
enhanced deracemization. Molecules move rapidly between the
solution and solid phases under attrition conditions, on a time
scale similar to that of the net conversion of one
enantiomorphic solid to the other; yet the net evolution of
crystal enantiomeric excess cannot be correlated in a simple
manner with the phase transfer of molecules. A transient
growth in crystal size is correlated with the accelerated
evolution of solid-phase homochirality, and the stochastic
nature of the process rationalizes the observed random
outcomes of the deracemization process for initially racemic
mixtures. The concept of a crystal-size-induced solubility
driving force helps to define the mechanism and is exploited
to develop a novel and reproducible approach for the
separation of enantiomorphic solids. Implications for the
emergence of single chirality in biological molecules are
discussed.
(13) Meekes, H.; Noorduin, W. L.; Bode, A. A. C.; van Enckevort, W.
J. P.; Kaptein, B.; Kellogg, R. M.; Vlieg, E. Cryst. Growth Des. 2008, 8,
1675−1681.
(14) (a) Uwaha, M. J. Phys. Soc. Jpn. 2004, 73, 2601. (b) Skrdla, P.
Cryst. Growth Des. 2011, 11, 1957−1965.
(15) Iggland, M.; Mazzotti, M. Cryst. Growth Des. 2011, 11, 4611−
4622.
(16) See Supporting Information for details.
(17) The presence of chiral impurities was in fact invoked to help
rationalize the apparent lack of randomness in results that appeared
skewed toward one enantiomorph in the first report of attrition-
enhanced deracemization of intrinsically chiral molecules (see ref 4).
(18) The experiments of Figure 2 were carried out with ca. 300 mg of
rac-1 in 3 mL of MeCN, of which ca. 250 mg remains in the solid
phase. Thus, for series b and series c, the evolution to homochirality
that takes place over 2 h between the 3 and 5 h sampling points
involves the conversion of ca. 125 mg of one enantiomorph to the
other, for an average mass conversion rate of ca. 1 mg/min.
(19) Samples were imaged on a laser scanning confocal microscope
to produce high-resolution snapshots of 5−7 randomly selected
regions (for a population size of 11 000−17 000 individual particles)
for each sample. A calibrated perimeter around each crystal defined in
2D space is created by the imaging software to generate a total count
of individual objects with area expressed in μm2. Results are sorted
into histograms reporting fraction of total crystal area as a function of
the bins of crystal area range in μm2.
ASSOCIATED CONTENT
* Supporting Information
Details of general methods for synthesis and deracemization
experiments, CSD measurements, ee measurement, and isotope
measurements. This material is available free of charge via the
Internet at http://pubs.acs.org.
■
S
AUTHOR INFORMATION
Corresponding Author
blackmond@scripps.edu
Notes
■
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
(20) The question of kinetic isotope effects in the use of 15N-1 in
tracking the movement of molecules should be addressed. In the four
experiments conducted under the conditions shown in Figure 5, the
two that evolved to single chirality in S-1 were initiated in one case
with the 15N tag present in the S-1 solid and in the other case with the
15N tag in the R-1 solid. Similarly, the two that remained racemic
commenced with the 15N tag in opposite enantiomorphs. These
results imply the absence of any kinetic isotope effects in the
deracemization process.
(21) Pasteur, L. C.R. Hebd. Seanc. Acad. Sci. Paris 1848, 26, 535−538.
́
(22) Jacques, J.; Collet, A.; Wilen, S. H. Enantiomers, Racemates and
Resolution; Krieger: Florida, 1994; Chap. 4.
(23) Levilain, G.; Coquerel, G. CrystEngComm 2010, 12, 1983−1992.
(24) Welch, C. R. Chirality 2001, 13, 425−427.
D.G.B. acknowledges funding from the NSF (CBET Award
1066608) for the work on compounds 1 and 3 and NASA
(Exobiology NNX12AD78G) for the work on Pasteur’s tartrate
2. Dr. William B. Kiosses (TSRI Core Microscopy) is
acknowledged for assistance with light microscopy imaging
and crystal population analysis.
REFERENCES
■
(1) Viedma, C. Phys. Rev. Lett. 2005, 94, 065504-1−065504-4.
(2) (a) van’t Hoff, J. H. Ber. Dtsch. Chem. Ges 1902, 35, 4252−4264.
(b) Roozeboom, H.W. B. Z. Phys. Chem. Stoech Verwandt 1899, 28,
494−517.
(3) (a) Kondepudi, D. K. Science 1990, 250, 975−976. (b) McBride,
J. M.; Carter, R. L. Angew. Chem., Int. Ed. Engl. 1991, 30, 293−95.
(4) Noorduin, W. L.; Izumi, T.; Millemaggi, A.; Leeman, M.; Meekes,
H.; Van Enckevort, W. J.; Kellogg, R. M.; Kaptein, B.; Vlieg, E.;
Blackmond, D. G. J. Am. Chem. Soc. 2008, 130, 1158−1159.
(5) Viedma, C.; Ortiz, J. E.; de Torres, T.; Izumi, T.; Blackmond, D.
G. J. Am. Chem. Soc. 2008, 130, 15274−15275.
(6) Crusats, J.; Veintemillas-Verdaguer, S.; Ribo, J. M. Chem.Eur. J.
2006, 12, 776−7781.
(7) Viedma, C. Astrobiology 2007, 7, 312−319.
(8) Blackmond, D. G. Chem.Eur. J. 2007, 13, 3290−3295.
(9) McBride, J. M.; Tully, J. C. Nature 2008, 452, 161−162.
(10) Leeman, M.; de Gooier, J. M.; Boer, K.; Zwaagstra, K.; Kaptein,
B.; Kellogg, R. M. Tetrahedron: Asymmetry 2010, 21, 1191−1193.
(11) (a) Noorduin, W. L.; van der Asdonk, P.; Bode, A. A.; Meekes,
H.; Van Enckevort, W. J.; Vlieg, E.; Kaptein, B.; van der Meijden, M.;
Kellogg, R. M.; Deroover, G. Org. Process Res. Dev. 2010, 14, 908−911.
(b) Maarten, W.; van der Meijden, M.; Leeman, M.; Gelens, E.;
H
dx.doi.org/10.1021/ja303566g | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX