The role of diastereomer impurity in oiling-out during the resolution of trans -4-methyl-2-piperidine carboxylic ethyl ester enantiomers by crystallization

Article abstract of DOI:10.1021/op500026z

Isolating enantiomers by crystallization via diastereomer salt formation is widely used in the pharmaceutical industry. Unfortunately, we found that, when some diastereomer impurity exists, oiling-out will take place and significantly decrease the resolution efficiency. To identify the role of impurities in oiling-out, three series of chiral resolution of trans-4-methyl-2-piperidine carboxylic ethyl ester (trans-4MPE) with different levels of diastereomer impurity were performed. Using process analytical technologies (PAT) technologies, the demixing region was identified by measuring T_L-L curves and supersolubility curves for different series. It was found that a series with a higher level of impurity has a higher T_L-L point and a wider demixing region. Besides, it seems that the resolution efficiency through oiling-out, as well as the crystallization kinetics, was significantly decreased. Here we initially put forward this particular diastereomer purification problem, where the impurity has a positive effect on oiling-out. To reduce or eliminate oiling-out, an efficient prepurification strategy must be ensured.

Full text of DOI:10.1021/op500026z

Article
pubs.acs.org/OPRD
The Role of Diastereomer Impurity in Oiling-Out during the
Resolution of trans-4-Methyl-2-piperidine Carboxylic Ethyl Ester
Enantiomers by Crystallization
Riju Ren,†,‡ Dengqiong Sun,†,‡ Tingting Wei,†,‡ Shixin Zhang,§ and Junbo Gong*,†,‡
†The National Engineering Research Center of Industrial Crystallization Technology, School of Chemical Engineering and
Technology, Tianjin University, Tianjin, 300072, P. R. China
‡Collaborative Innovation Center of Chemical Science and Chemical Engineering (Tianjin), Tianjin University, Tianjin, 300072, P. R.
China
§Tianjin Hengbida Chemical Synthesis Co. Ltd., Tianjin, 300072, P. R. China
ABSTRACT: Isolating enantiomers by crystallization via diastereomer salt formation is widely used in the pharmaceutical
industry. Unfortunately, we found that, when some diastereomer impurity exists, oiling-out will take place and significantly
decrease the resolution efficiency. To identify the role of impurities in oiling-out, three series of chiral resolution of trans-4-
methyl-2-piperidine carboxylic ethyl ester (trans-4MPE) with different levels of diastereomer impurity were performed. Using
process analytical technologies (PAT) technologies, the demixing region was identified by measuring TL−L curves and
supersolubility curves for different series. It was found that a series with a higher level of impurity has a higher TL−L point and a
wider demixing region. Besides, it seems that the resolution efficiency through oiling-out, as well as the crystallization kinetics,
was significantly decreased. Here we initially put forward this particular diastereomer purification problem, where the impurity
has a positive effect on oiling-out. To reduce or eliminate oiling-out, an efficient prepurification strategy must be ensured.
well as in protein and polymer systems.5−8 Usually, LLPS is
unexpected, as products obtained from oil droplets are usually
1. INTRODUCTION
Crystallization must surely rank as the oldest and most
of low purity and poor crystal properties, and scale-up of such
commonly used unit operation in chemical processing. It can
systems will be problematic.
not only provide crystalline products with the desired particle
size, particle distribution, and even special polymorphs with
high purity but also directly affect the extent of chiral
(2R,4R)-4MPE (4-methyl-2-piperidine carboxylic ethyl ester)
is a synthesis intermediate of an active pharmaceutical
ingredient that is used for cardiovascular diseases. It is usually
separation. However, in less than 8% of cases, enantiomer
separated from its racemic mixture by a diastereomeric salt
mixtures could be separated by direct and preferential
formation method, as raw 4MPE supplied is usually a mixture
crystallization, where each enantiomer could crystallize
separately, known as a conglomerate. In other chiral separation
processes, direct crystallization would give rise to a racemic
crystal structure and separation cannot be achieved, and then
of (2R,4R)-4MPE, (2S,4S)-4MPE, and their diastereomers,
(2R,4S)-4MPE and (2S,4R)-4MPE. A brief resolution process
for (2R,4R)-4MPE is shown in Figure 1. Distillation here is
meant to remove the cis-isomer. Then the trans-4MPE is
we need to bring about diastereomeric crystallization.
treated with L-(+)-tartaric acid (LTA) in the acetone/water
Diastereomeric crystallization is based on the reaction of
solvent mixture, where LTA works as a resolving agent. The
racemic mixture with a single isomer resolving agent to give
resulting salt form of (2R,4R)-4MPE is selectively isolated from
two diastereomeric derivatives, which are salts or complexes. By
taking advantage of the differences in their solubility,
enantiomer mixtures can be separated indirectly. A recent
study of market drugs shows that more than 65% are
solution by crystallization.9 In this system the ratios of solvent
mixture V1/V2 cannot be identified here for commercial
interests.
What we need to point out here is that the boiling points of
trans-4MPE and cis-4MPE are so close that they are difficult to
manufactured by methods involving diastereomer crystalliza-
tion.1,2
be totally separated from each other without very delicate
A large number of factors, such as supersaturation, agitation
distillation technology. Besides, a certain degree of racemization
intensity, crystallization temperature, and so forth, need to be
controlled to separate enantiomer mixtures with a high
will occur during distillation. Thus, the presence of cis-4MPE
before salt formation becomes inevitable, and the ratio of cis-
resolution efficiency. The difficulty in chiral resolution is
4MPE remained after distillation may vary from 5% to 30% in
further exacerbated when a second liquid phase is formed upon
cooling, where one phase contains the solvent(s) and the other
phase is mainly made up by the solute in the form of a very
heavy viscous oil-like phase. This phenomenon is frequently
termed oiling-out or liquid−liquid phase separation (LLPS).3,4
It is a commonly recognized issue in organic crystallization as
different batches. In that case, it is reasonable for us to regard
the undesirable cis-diastereomers as impurities for chiral
Received: January 23, 2014
© XXXX American Chemical Society
A
dx.doi.org/10.1021/op500026z | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Figure 1. A brief resolution process for (2R,4R)-4MPE.
Figure 2. Structures of trans-4MPE and cis-4MPE.
resolution. The structures of main components are given in
Figure 2.
it was used to make sure the salt-forming reaction was complete
and monitor the solute concentration during oiling-out and
subsequent crystallization.
During our resolution, the phenomenon that levels of
impurity have a dramatic impact on the miscible region has
aroused our great interest. While most of the researches have
focused their study on the levels of saturation, the selection of
solvents, cooling rate, seeding or not, etc.,10−12 we believe that
without some discussion of impurities, no discussion of oiling-
out would be complete. Oiling-out in chiral resolution is no
exception. Since neither process could guarantee a product with
an enantiomeric excess meeting regulatory requirements, but
instead provide a chiral mixture enriched with the desired
enantiomer. In this paper, with a series of experiments we
designed, the aims of the investigations discussed here are to (i)
use a range of process analytical technologies (PAT) to track
the generation and evolution of oiling-out and subsequent
crystallization in different experiments; (ii) explore the role of
diastereomer impurity in oiling-out during chiral resolution.
FBRM (model M400LF) system coupled with iC FBRM
software from Mettler-Toledo was introduced to detect both
droplets and particle dimensions during oiling-out and
subsequent crystallization processes with a 10 s duration. A
Lasentec PVM (probe model V819) with an update rate of 1
image per 30 s was employed to visualize the formation of the
oil phase and subsequent dispersion into the solution in
situ.13,14
An optical microscope (Eclipse E200, Nikon) was used for
the examination of oil droplets and final products. High-
performance liquid chromatography (HPLC), with an Eslipse
Plus C18 Agilent column (5 μm, 4.6 × 250 mm) and acetyl
glucosamine isothiocyanate (GITC) working as precolumn
derivatization reagent, was performed on samples quickly
removed from the bottom layer to determine the concentration
of 4MPE-LTA in the oil phase.
2. EXPERIMENTAL SECTION
2.3. Measurement of the Solubility of the LTA Salts of
4MPE. As indicators of relative solubility and the propensity to
separate, the solubility of the two individual diastereomer salts
of the R,R-form and S,S-form, namely (2R,4R)-4MPE-LTA and
(2S,4S)-4MPE-LTA, and their equimolar mixturea typical
raw saltwas measured gravimetrically in v1/v2 acetone/water
using the isothermal method. Since the boiling point of acetone
is about 56.5 °C, the temperature range was about from 15 to
45 °C. Excess solid was added to the solvent and kept at a
particular temperature with agitation for 24 h to reach solid−
liquid equilibrium. A portion of 10 mL of liquid samples was
extracted from each vial for three times using a preheated
filtered syringe, and then the samples were weighed and stored
in a drying chamber to reach constant weight, which was used
to determine the solubility. Each experiment was performed
three times, and the average value was taken as the final
solubility data. The average relative error was less than 2.1%.
2.4. Methodology Used To Determine Oiling-Out
Curves and Crystallization Experiments during Chiral
Resolution. Three series of chiral resolution experiments,
series I, series II, and series III, were performed, with different
2.1. Materials. Salt forms of trans-4MPE with a purity of
97.8% and (2R,4R)-4MPE-LTA and (2S,4S)-4MPE-LTA with a
purity of 99% were supplied respectively by Tianjin Hengbida
Chemical Synthesis Co., Ltd. Raw 4MPE with different ratios of
cis-4MPE (liquid state), 24%, 16%, and 8%, was obtained from
different manufacturing batches operated by Hengbida, the rest
of which are (2R,4R)-4MPE and (2S,4S)-4MPE in equimolar.
They were used in the chiral separation experiments without
further purification. L-(+)-LTA was purchased from Qingdao
Shengke Biological Technology Co. Ltd. of China. Acetone,
analytical reagent-grade was obtained from Tianjin KeWei
Chemical Co. Ltd. of China, and deionized water was generated
from a Millipore filter unit.
2.2. Equipment and Technique. A cryo-compact
circulator of the high-tech series CF41 from JULABO
Labortechnik GmbH was used to control temperature
throughout the experiment, and its temperature stability is
0.02 °C with a PID temperature control technology. ATR-
FTIR (ReactIR 15 reaction analysis system coupled with iC IR
4.2 software) was supplied by Mettler-Toledo, and in our work
B
dx.doi.org/10.1021/op500026z | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
ratios of cis-4MPE, 24%, 16%, and 8%, respectively. In each
series, solutions with different concentrations of 4MPE-LTA
were prepared through reaction crystallization at 40 °C. To be
exact, a known amount of LTA, namely, 5.6 g, was dissolved in
certain mass of mixed solvent v1/v2 varied from 100 to 200 g at
40 °C. Then equimolar amounts of 4MPE, about 6.3 g, were
added into the solution. In this way solutions with different
concentration of 4MPE-LTA were obtained. Then the solution
was gradually cooled down to 5 °C at a fixed cooling rate of 5
°C/h and stood at 5 °C for 5 h. For the experiment that oiled
out, we took down the temperature so that oil droplets
appeared as Tcloud and then raised the temperature slightly until
the solution clarified. We denoted this temperature as Tclear, the
minimum temperature at which the solution clarified. The
average of Tcloud and Tclear was taken as TL−L. During the whole
process, the FBRM and ATR-FTIR, combined with PVM, were
used to measure and identify the clouding and clarifying points.
Furthermore, the effects of impurities and the oiling-out
process on the final products were studied. When LLPS or
crystallization started, the solutions were allowed to undergo a
gravity-driven separation into two coexisting phases in
quiescent medium at about 20 °C. To allow for an easy
comparison, the solutions with a concentration of 4MPE-LTA
11.67 g/100 g solvent in three series were kept quiescent for
more than 8 h at 20 °C. Samples of the bottom phases, in the
oil phase or crystal form, were withdrawn simultaneously using
a rubber buret, followed by a free-basing step that yielded the
(2R,4R)-4MPE free base as the final drug intermediate. The
ratio of all of the four isomers was measured by HPLC.
2.5. Defining Suitable Peaks for the Salt-Forming
Reaction and the Concentration of 4MPE-LTA during
the in Situ Oiling-Out Measurement. Selection of a suitable
peak for 4MPE-LTA is very essential for in situ measurement.13
At the beginning of one of the experiments in series,
background was taken of the solvent and LTA after LTA was
totally dissolved at 40 °C. Then an equimolar amount of 4MPE
was added to the solution, at which point the salt-forming
reaction started. It allowed for the rapid identification of the
special peak to 4MPE-LTA.
Figure 3. Solubility of the R,R-form, S,S-form, and typical coarse salt in
v1/v2 acetone/water.
3. RESULTS AND DISCUSSION
Figure 4. Mid-IR spectra taken at 40 °C to identify the absorption
fingerprint of the 4MPE-LTA.
3.1. Solubility. As presented in Figure 3, the solubility of
the two diastereomer salt exhibits a conservative pattern of
increasing solubility with temperature. The desired R,R-form is
less soluble than the S,S-form, which indicates that the R,R-form
would precipitate more easily than the S,S-form. Furthermore,
to obtain a better resolution efficiency, the end point of cooling
crystallization was set at 20 °C in our chiral separation, below
which the difference in solubility is getting smaller and it
becomes difficult for resolution. No oiling-out was observed to
take place in any of the experiments. However, given the
limited materials, here only some of the solubility data was
presented.
annotation in Figure 4 points out the peak of interest in our
study at 1611 cm−1. It should be noted that IR peak height is a
reflection of absorbed energy corresponding to the vibration
and rotation of chemical bonds. Therefore, it is not a good
technology to be used to distinguish the difference between all
of these diastereomeric salts efficiently. The peak height here
actually represents for the total concentration of four kinds of
salt we studied.
The resolution of pure trans-diastereomers (the ratio of
trans-form is 97.8%) is given here as supplementation at 20 °C.
According to the report of HPLC, the purity of (2R,4R)-4MPE-
LTA could be increased from 50.3% to 92.7% after two times
crystallization. The melting point of final products is between
95 and 98.0 °C, consistent with the result of a Japanese patent.9
3.2. Defining Suitable Peaks for in Situ Measurement.
As shown in Figure 4, 4MPE-LTA produced a distinct FTIR
fingerprint, with several peaks available to trace the salt-forming
reaction and the concentration of 4MPE-LTA in solution. The
3.3. In Situ Measurement of Oiling-Out and Sub-
sequent Crystallization. When oiling-out took place, the
transparent solution would become cloudy, and oil droplets
appeared on the wall of the crystallizer (Figure 5a). However, it
was difficult to discriminate if the cloudiness of the solution was
caused by droplets or crystal particles with naked eyes, and
when the stirrer was turned off, two liquid phases separated,
and the oil phase sank to the bottom after several seconds15
(Figure 5b). Figure 5c shows a photograph of oil droplets with
a microscope.
C
dx.doi.org/10.1021/op500026z | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Figure 5. Images of oiling-out: (a) Oil droplets appeared on the wall of the crystallizer; (b) after the agitation was stopped, a visible phase separation
interface appeared gradually; (c) optical microscope images of oil droplets.
Figure 6. Time course of FBRM and FTIR measurements during the liquid−liquid phase separation and crystallization of 4MPE-LTA.
The combination of FBRM, IR, and PVM measurements16,17
was used in one of the experiments that oiled out to have a
better understanding of this phenomenon. According to those
PVM images, the dimensions of the droplets or particles
throughout the process are smaller than 150 μm. Therefore, the
statistical interval for the time course FBRM was selected as 0
to 150 μm, as shown in Figure 6.
Upon cooling the clear solution (Figure 7a) after reaction,
the specific IR peak of 4MPE-LTA was seen to have a very
sharp increase whilst the counts for small dimension (≤50 μm)
detected by FBRM had a very sudden increase in total counts at
about 3090s (point B). Examination of the PVM images
(Figure 7b) showed that this corresponded to the generation of
amorphous-like aggregates. As for the sudden increase in IR
measurement, although similar measurements were observed in
the work of Deneau and Veesler et al.,18,19 they simply ascribed
it to be due to transient oiling-out on heating or generation of
droplets upon cooling and did not give any further
interpretation. Although FTIR will not respond to the presence
of either droplets or crystals in suspension for its the detection
limit (about 2 μm), The ATR probe is sensitive to the film in
contact with the probe surface. Here we do think that the
sudden increase might be a respond to the generation of films
of oil phase, which may initially form on the vessel and probe
surface and be detected by the ATR probe. With the quantity of
amorphous growing, the visible droplets, which were beyond
the detection limit of FTIR. Then the IR probe began to detect
the concentration in the lean phase.
At about 4520 s (point C), both IR peak height and counts
for dimension ≤50 μm, as well as the total counts (summing up
all the counts in different intervals), were seen to have a sharp
decrease. It could be interpreted as the coalescence of small oil
droplets. As the visible droplets were formed (Figure 7c), the
IR probe began to detect the concentration in the lean phase.
The speculation was further affirmed by the slightly increase of
the particle counts for larger particles between 50 and 150 μm
at 4710 s.
Upon further cooling, it slowly began to generate many
needle-like crystals. At first, it seems that the particle appeared
to hold at a constant level, with a slight decline in the IR
measurements. By the examination of Figure 7d (point D, at
about 15180 s), it was found that almost all of the fine crystals
generated were confined to the oil droplets, and FBRM might
still detect the counts of droplets that contain fine crystals. As
the droplets gradually transformed into crystals, an apparent
increase in particle counts of all of the statistical intervals was
observed by FBRM at 22410 s (point E). This can be explained
by the release of crystals into the solution as shown in Figure
7e. Then those crystals began to grow at the consumption of
concentration just like the process in normal crystallization,
D
dx.doi.org/10.1021/op500026z | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Figure 7. PVM images for the evolution of oiling-out.
which was in line with the subsequent decrease of the IR peak
height. At the end of this measurement, all of the droplets
transformed into the normal crystals. Compared with Figure 7e,
there were more crystals within view, and the particle
dimension had significantly increased in Figure 7f. It indicted
the generation and the growth of more crystals outside the oil
droplets.
To summarize, five main areas could be identified for the
process in this work: (a) clear solution after reaction; (b)
generation of oil droplets or a amorphous aggregates; (c)
coalescence of small oil droplets; (d) oil droplets converting
into crystal particles; (e) crystal growth. The observations were
similar to the work made by others.18,19
The generation of some amorphous aggregates before oiling-
out was first observed in our work, and it could be best
interpreted by Ostwald’s law of stages.20 According to
Ostwald’s law, if a multitude of solid-state forms is possible,
the most unstable form will nucleate first. This can either be an
unstable crystallization form or even amorphous solid or even
oil (Figure 8). Besides, it was not difficult to find that the
nucleation and growth of crystals were significantly hindered
when LLPS took place. The final products obtained from
oiling-out were more like “fine feathers”, and other character-
izations were discussed in section 3.5.
3.4. Effect of Diastereomer Impurities on Oiling-Out
during Resolution. Generally, the solution is supersaturated,
and crystal nucleation and growth can occur in the region
above the solubility curve. However, we found that it would
undergo different crystallization process when different ratios of
impurity existed. Experiments in series III preceded a normal
crystallization without oil droplets, while, in series I and II,
oiling-out would be observed distinctly upon cooling, and it
seems that crystallization seems to be so hindered.
To have an better understanding of the reactive crystal-
lization, TL−L curves for series I and II (the pink and olive
curves), the solubility curves discussed before and expected
solubility curve for a typical raw salt above 40 °C (the green
dashed line) was also plotted in Figure 9. The region between
E
dx.doi.org/10.1021/op500026z | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
the temperature range studied, and miscibility increases with an
increase in temperature.
Oiling-out is an intermediate undercooled liquid phase and
can be regarded as a preferred kinetic route to crystallization
when the kinetic processes of nucleation and growth are so
hindered.7 According to that logic, a higher TL−L point might
indicate a lower energy barrier for a new phase, and the
presence of impurities would promote the generation of the oil
phase, since that TL−L curve not only depends on the initial
supersaturation but also the level of diastereomer impurities.
3.5. Influence of Oiling-Out on Final Products. Typical
optical microscope images of final products obtained from
oiling-out and true crystallization are shown in Figure 10 for
comparison. Figure 10a is a typical micrograph of the needle-
like crystallization product in the absence of oiling-out. Figure
10b is a typical micrograph of the products from oil droplets. It
can be seen that particles yielded from oil phase are so small
and tend to melt even at ambient temperature, which may be
attributed to the inclusion of solvents. The molecular
rearrangement required for growth seems to be so hindered
in the heavy viscous oil-like phase.
To investigate the composition change between the oil phase
and the crystals once formed, both the oil products and the
solid forms need to be obtained. However, there are many cases
that stable oil phase formed, where no nucleation is observed
and no solid products can be obtained at 20 °C, unless holding
the whole systems at 5−8 °C for more than 5 h, as mentioned
before. The composition of all the samples was measured by
HPLC.
Figure 8. Visualization of Ostwald’s law of stages.
TL−L curve and supersolubility curves was defined as the
demixing region in this work.
It can be seen that the TL−L point is much higher than the
spontaneous nucleation points in the presence of the oil phase.
For instance, the systems began to oil out from 44.51 and 35.86
°C, respectively, at the concentration of 11.67 g in 100 g
mixture solvent in series I and II, and only by holding the whole
systems at 5−8 °C for more than 5 h could the solid products
be expected to be obtained, which indicates that the kinetics of
crystal nucleation is rather slow relative to the LLPS process.
The schematic supersolubility curve for the two series was
represented by the red dashed line. However, in series III the
spontaneous nucleation points were at about 15−21 °C at the
concentrations of interest (see the dark blue curve).
Since LLPS is the concentration at which the substance
exceeds its miscibility limit with the mixed solvent, most
compounds show the expected increase of miscibility with the
solvent with the temperature increase.21 As exemplified by
4MPE-LTA in series I and II, the phase transition
concentration in this study increased with temperature over
According to the results of HPLC analysis, shown in Table 1,
products obtained from the oil phase are of lower purity. The
composition of oil phase was comparable to that of solid
products obtained at 5−8 °C, and all of the four enantiomers
tend to be involved in the final products with the poor
selectivity of oiling-out, resulting a poor separation process.
Thus, LLPS occuring in the metastable region of series I and II
can be regarded as the precursor of solid products for chiral
separation in this work. Moreover, when the purity of 4MPE
Figure 9. Phase diagram for 4MPE-LTA: TL−L curves and the supersolubility curve of series I, II, and III.
F
dx.doi.org/10.1021/op500026z | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Figure 10. Microscope images: (a) product from normal crystallization and (b) solidification product from oiling-out.
Table 1. Concentration of the ratio of four diastereomers in the bottom phase for three series with an initial composition of the
a
11.67 g/100 g solvent mixture
series (% initial ratio of trans-
form)
temperature samples
obtained (°C)
oiling-
out
layer
% R,R-form % S,S-form % R,S-form % S,S-form
note
series I (76%)
bottom (oil)
20.2
39.66
40.12
39.59
39.60
10.49
10.17
10.26
10.11
much no crystals at
20.1 °C
bottom
(crystals)
5.0−8.0
series II (84%)
bottom (oil)
20.1
47.65
49.51
71.81
76.33
42.38
43.02
23.37
22.21
4.75
3.59
2.16
0.67
5.22
3.88
2.66
0.79
less
no crystals at
20.0 °C
bottom
(crystals)
5.0−8.0
20.0
series III (92%)
bottom
(crystals)
none
none
crystallize at
21.5 °C
trans-4MPE
crystals
23.5
a
(2R,4R)-form and (2S,4S)-form are nearly equimolar in trans-4MPE; the rest are (2S,4R)-form and (2R,4S), also equimolar.
reaches a certain degree, or in other words, when the ratio of
impurity decreased to some level, the phenomenon of LLPS
will not appear, and a significant improvement on the
resolution efficiency will be achieved. Further work will be
carried out in our next study in looking for the critical purity of
4MPE-LTA.
easier to conduct. Furthermore, the critical purity can be used
to determine the least tolerable distillation results and work as
guidance for upstream distillation processes. Besides, the
identification of the role of impurity on oiling-out will draw
some hints on the building of LLPS kinetic models. Much
kinetic data need to be measured in our further study.
Here we have to confess that this is not a good system for the
measurement of the composition change between the oil phase
and the crystals, since all the crystals through oiling-out were
obtained at an extremely low temperature, which will affect the
composition greatly.
AUTHOR INFORMATION
■
Corresponding Author
*Tel.: +86-22-27405754. Fax: +86-22-27374971. E-mail:
Junbo_gong@tju.edu.cn.
Notes
4. CONCLUSIONS
The authors declare no competing financial interest.
In industrial crystallization, impurity levels often change during
process development and on scale-up. Therefore, it is necessary
to take the effect of diastereomeric impurities into consid-
eration when it oiled out during chiral resolution. With the aid
of PAT tools, the phase diagram consisting of the solubility
curves, the liquid−liquid separation curves, and the super-
solubility curves was determined experimentally for three series
of interest. We found that the system with a higher ratio of
diastereomer impurity is much easier to oil-out, with a higher
TL−L and a wider demixing region. According to the results of
HPLC, there were not many changes in the relative content of
R,R-form and S,S-form both of the oil phase and the crystals,
and the amorphous form generated at first might work as the
precursor of final solid products.
ACKNOWLEDGMENTS
■
The authors gratefully acknowledge the support of the National
Natural Science Foundation of China (No. 21376164) and the
assistance of Mettler Toledo Instruments Co., Ltd. This
material is based upon works supported by the National
Engineering Research Center of Industrial Crystallization
Technology, School of Chemical Engineering and Technology,
and the Co-Innovation Center of Chemistry and Chemical
Engineering of Tianjin, Tianjin University. It is supported by
the Program for New Century Excellent Talents in University
(NCET-12-0393).
REFERENCES
■
One area for further study is the identification of a critical
purity for the resolution of 4MPE-LTA, since it seems that
there may be a critical purity beyond which oiling-out will not
appear and the diastereomer separation will become much
(1) Kaspereit, M.; Swernath, S.; Kienle, A. Org. Process Res. Dev. 2012,
16, 353−363.
(2) Lorenz, H.; Von Langermann, J.; Sadiq, G.; Seaton, C. C.; Davey,
R. J.; Seidel-Morgenster, A. Cryst. Growth Des. 2011, 11, 1549−1556.
G
dx.doi.org/10.1021/op500026z | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
(3) Jacques, J.; Collet, A.; Wilen, S. H. Enantiomers, Racemates and
Resolutions; Krieger Publishing Company: Malabar, 1994.
(4) Codan, L.; Babler, M. U.; Mazzotti, M. Org. Process Res. Dev.
̈
2012, 16, 294−310.
(5) Tung, H.-H.; Paul, E. L.; Midler, M.; McCauley, J. A.
Crystallization of Organic Compounds, An Industrial Perspective; John
Wiley & Sons, Inc.: New York, 2008.
(6) Kiesow, K.; Tumakaka, F.; Sadowski, G. J. Cryst. Growth 2008,
310, 4163−4168.
(7) Bonnett, P. E.; Carpenter, K. J.; Dawson, S.; Davey, R. J. Chem.
Commun. 2003, 698−699.
(8) Svard, M.; Gracin, S.; Rasmuson, Å. C. J. Pharm. Sci. 2007, 96,
2390−2398.
(9) Mitsubishi Kasei Corp (MITU-C). JP Patent 2,212,473-A, 1990.
(10) Duffy, D.; Cremin, N.; Napier, M.; Robinson, S.; Barrett, M.;
et al. Chem. Eng. Sci. 2012, 77 (30), 112−121.
(11) Lu, J.; Li, Y.-P.; Wang, J.; Li, Z.; Rohani, S.; Ching, C. B. Org.
Process Res. Dev. 2012, 16, 442−446.
(12) Lafferrere, L.; Hoff, C.; Veesler, S. Eng. Life Sci. 2003, 3, 127−
131.
(13) Sistare, F.; St. Pierre Berry, L.; Mojica, C. Org. Process Res. Dev.
2005, 9, 332−336.
(14) Liotta, V.; Sabesan, V. Org. Process Res. Dev. 2004, 8, 488−494.
(15) Barrett, P.; Glennon, B. Chem. Eng. Res. Des. 2002, 80 (A7),
799−805.
(16) Birch, M.; Fussell, S. J.; Higginson, P. D.; McDowall, N.;
Marziano, I. Org. Process Res. Dev. 2005, 9 (3), 360−364.
(17) Howard, K. S.; Nagy, Z. K.; Saha, B.; Robertson, A. L. Org.
Process Res. Dev. 2009, 13, 590−597.
(18) Deneau, E.; Steele, G. Org. Process Res. Dev. 2005, 9, 943−950.
(19) Veesler, S.; Revalor, E.; Bottini, O.; Hoff, C. . Org. Process Res.
Dev. 2006, 10, 841−845.
(20) Beckmann, W., Ed. Crystallization - Basic Concepts and Industrial
Applications; Wiley-VCH: Weinheim, 2013.
(21) Lafferrere, L.; Hoff, C.; Veesler, S. Cryst. Growth Des. 2004, 4
(6), 1175−1180.
H
dx.doi.org/10.1021/op500026z | Org. Process Res. Dev. XXXX, XXX, XXX−XXX