Tandem crystallization strategies for resolution of 3,3,3-trifluorolactic acid [CF3CH(OH)COOH] by chiral benzylamines

Article abstract of DOI:10.1002/chir.23127

Resolution of rac-3,3,3-trifluorolactic acid by diastereomeric salt formation was reinvestigated. The use of (S)-1-phenylethylamine gives coprecipitation of two diastereomeric phases, 1 (S)-[NH₃CH(CH₃)Ph](S)-[CF₃CH(OH)COO] and 2 (S)-[NH₃CH(CH₃)Ph](R)-[CF₃CH(OH)COO]·H₂O. Pure phase 1 may be obtained using molecular sieves as desiccants. Resolution by (S,S)-2-amino-1-phenylpropan-1,3-diol gives monoclinic (S,S)-[NH₃CH(CH₂OH)CHOHPh] (R)-[CF₃CH(OH)-COO] 3 with minor (S)-3,3,3-trifluorolactate contamination, which is precluded in the recrystallized orthorhombic form 4. A new resolution using inexpensive phenylglycinol gives pure phase 5 (S)-[NH₃CH(CH₂OH)Ph] (S)-[CF₃CH(OH)COO] in 76% yield, 94% ee in a single step, in preference to its (S)-(R) diastereomer 6. Overall efficient resolution for both enantiomers of the trifluorolactic acid (each ca. 70% yield, 99% ee) may be achieved by various two-step “tandem” crystallizations, involving direct addition of either water or a second base to the filtrate from the initial reaction.

Full text of DOI:10.1002/chir.23127

Received: 29 March 2019
DOI: 10.1002/chir.23127
Revised: 14 July 2019
Accepted: 16 July 2019
R E G U L A R A R T I C L E
Tandem crystallization strategies for resolution of
3,3,3‐trifluorolactic acid [CF3CH(OH)COOH] by chiral
benzylamines
Lawrence W.‐Y. Wong1
Herman H.‐Y. Sung1
| Elena V. Vashchenko1,2
| Valerii V. Vashchenko1,2
Alexander I. Krivoshey2 | Ian D. Williams1
| Ying Zhao1 |
| Vadim Mikhailenko2 |
1 Department of Chemistry, Hong Kong
Abstract
University of Science and Technology,
Resolution of rac‐3,3,3‐trifluorolactic acid by diastereomeric salt formation was
reinvestigated. The use of (S)‐1‐phenylethylamine gives coprecipitation of two
diastereomeric phases, 1 (S)‐[NH3CH(CH3)Ph](S)‐[CF3CH(OH)COO] and 2
Kowloon, Hong Kong, China
2 Institute for Single Crystals, National
Academy of Sciences, Kharkov, Ukraine
(S)‐[NH3CH(CH3)Ph](R)‐[CF3CH(OH)COO]·H2O. Pure phase
1 may be
Correspondence
Ian D. Williams, Department of
Chemistry, Hong Kong University of
Science and Technology, Clear Water Bay,
Kowloon, Hong Kong, China.
Email: chwill@ust.hk
obtained using molecular sieves as desiccants. Resolution by (S,S)‐2‐amino‐1‐
phenylpropan‐1,3‐diol gives monoclinic (S,S)‐[NH3CH(CH2OH)CHOHPh]
(R)‐[CF3CH(OH)‐COO] 3 with minor (S)‐3,3,3‐trifluorolactate contamination,
which is precluded in the recrystallized orthorhombic form 4. A new resolution
using inexpensive phenylglycinol gives pure phase 5 (S)‐[NH3CH(CH2OH)Ph]
(S)‐[CF3CH(OH)COO] in 76% yield, 94% ee in a single step, in preference to
its (S)‐(R) diastereomer 6. Overall efficient resolution for both enantiomers of
the trifluorolactic acid (each ca. 70% yield, 99% ee) may be achieved by various
two‐step “tandem” crystallizations, involving direct addition of either water or
a second base to the filtrate from the initial reaction.
Funding information
Research Grants Council of Hong Kong,
Grant/Award Number: (RGC grant
16306515); National Academy of Sciences
of Ukraine, Grant/Award Number:
(project#0112U002186); Science and
Technology Center in Ukraine, Grant/
Award Number: (project# P654a)
KEYWORDS
diastereomeric salt formation, pyramidal disorder, resolution, solid solutions, tandem crystallization,
trifluorolactic acid
1 | INTRODUCTION
chiral synthetic or enzymatic approaches have been
employed,5-8 but these are not commercially attractive.
3,3,3‐Trifluorolactate has promise as a building block for
synthesis of organofluorine compounds for use in medicine
or optical and crystalline materials1-3 or as a chiral
derivatizing agent.4 Its widespread use is restricted by the
high cost of resolved 3,3,3‐trifluorolactic acid in excess of
US$ 600 per gram. This makes an inexpensive and efficient
resolution of the racemic compound highly desirable.
Figure 1 summarizes current approaches to enantiopure
3,3,3‐trifluorolactic acids. Path A indicates that several
Classical optical resolution using diastereomeric salt
formation (Path B) is a viable alternative,9-14 especially if
both enantiomers of the acid are desired.
A published resolution using (R,R)‐ or (S,S)‐2‐amino‐1‐
phenylpropan‐1,3‐diol appears effective with 90% ee,10 but
the resolving agents are still quite expensive. More
recent, but somewhat inconsistent, reports offered a possi-
ble resolution route via diastereomeric salt formation
using inexpensive (R)‐ or (S)‐α‐methyl‐benzylamine
Chirality. 2019;0–14.
wileyonlinelibrary.com/journal/chir
© 2019 Wiley Periodicals, Inc.
1

2
WONG ET AL.
dichlorotrifluoroacetone purchased from ABCR. Other
reagents and solvents were obtained from Sigma‐Aldrich
or TCI chemicals at 99% reagent grade and used as received
without further purification. Ethyl acetate used was dried
with calcium hydride under N2 atmosphere.
Intensity data for single crystal structures were mea-
sured at 100 K on a Rigaku Oxford Diffraction Superova
diffractometer (Cu‐Kα radiation, Atlas detector) with
details as recently reported.16 Structures were solved
and refined using SHELXL17 embedded in the Olex2 plat-
form.18,19 Experimental powder X‐ray diffractograms
were recorded on a Panalytical Xpert Pro and compared
with those simulated using the Mercury software pack-
age.20 These may be found in Supplementary Data S1.
2.2 | Crystallization of salts 1 and 2 using
1‐phenylethylamine
A 9:1 ethyl acetate: hexane solution (2 mL) of rac‐3,3,3‐
trifluorolactic acid (0.97 mmol, Fw 144.1, 140 mg) was
added an equimolar requivalent of (S)‐1‐phenylethyl-
amine (1 mmol, Fw 121.2, 128 μL) and heated at 60°C 2
hours following previous method.12 Upon cooling the
resulting colorless crystalline solids (174 mg, ca. 70%
yield, 50% ee) were found by P‐XRD and S‐XRD to be a
mixture of 1 (S)‐[NH3CHMePh] (S)‐[CF3CH(OH)COO]
and 2 (S)‐[NH3CHMePh] (R)‐[CF3CH(OH)COO]·H2O.
Pure 1 and 2 can be isolated stepwise (see Section 2.5).
FIGURE 1 Approaches to the resolution of rac‐CF3CH(OH)
COOH
(1‐phenylethylamine),13,14 the ammonium ion of which is
a commonly used cation in such resolutions.13,15 A patent
source13 claimed that each of enantiomer of 3,3,3‐
trifluorolactic acid could be isolated by switching between
anhydrous and water‐containing solvents in the diastereo-
meric salt recrystallizations. The absolute configurations
were assigned by comparison with known samples on chi-
ral GC, and up to 95.2% ee on the first run was reported. By
contrast, a later publication from the same group14 indi-
cated only 68% ee for the (S)‐3,3,3‐trifluorolactic acid can
be obtained with a similar protocol in the first crystalliza-
tion step. No explanation for the modest % ee was given,
and no crystal structure determinations were reported for
either of the systems.10,14
2.3 | Crystallization of salts 3 and 4 using
2‐amino‐1‐phenylpropan‐1,3‐diol
Rac‐3,3,3‐trifluorolactic acid (0.97 mmol, Fw 144.1, 140
mg), (1S,2S)‐2‐amino‐1‐phenyl‐1,3‐propane‐diol (1 mmol,
Fw 167, 167 mg) heated (60°C, 2 h) in 2 mL ethyl
acetate.10 Colorless plates obtained on cooling (130 mg,
Fw 311, 86% yield, 92% ee). P‐XRD and S‐XRD reveal that
this is a mixture of polymorphic phase‐types 3 (mono-
clinic) and 4 (orthorhombic) (S,S)‐[NH3CH(CH2OH)
CHOHPh] (R)‐[CF3CH(OH)COO]. The S‐XRD of 3 shows
enantio contamination at the anion site. The product was
recrystallizated in ethyl actate/hexane 9:1 in 40°C for 10
minutes. It is cooled for 3 hours and affords pure phase
4 (75% yield) with little or no enantio contamination
(99% ee).
We therefore decided to reinvestigate these to see
whether conditions could be further optimized, deduce
the reason for the low enantioselectivity for the
1‐phenylethylamine case, and find an effective and inex-
pensive method for resolution of 3,3,3‐trifluorolactic acid.
2 | MATERIALS AND METHODS
2.1 | General
2.4 | Crystallization of salts 5 and 6 using
phenylglycinol
Rac‐3,3,3‐trifluorolactic
acid
a
was
prepared
by
Rac‐3,3,3‐trifluorolactic acid (0.97 mmol, Fw 144.0, 140
mg), (S)‐phenylglycinol (0.55 mmol, Fw 138, 75 mg)
literature method12 via
base‐hydrolysis of 3,3‐

WONG ET AL.
3
were heated (60°C, 2 h) in 2 mL ethyl acetate. Upon
cooling, colorless plates of (S)‐[NH3CH(CH2OH)Ph](S)‐
[CF3CH(OH)COO] 5 (104 mg, Fw 281, 76% yield, 95%
ee) obtained by filtration. Single recrystallization gives
overall 70% yield with 99% ee. Combustion analysis for
5 C11H14F3NO4 calc. (found) C 46.98% (46.87) H 5.02%
(5.12) N 4.98% (5.51).
2.7 | Determination of % ee by chiral
chromatography
HPLC experiments were performed with a Thermo Scien-
tific Dionex UltiMate 3000 instrument fitted with
Phenomenex Lux 5 μm Amylose‐1 LC chiral column
(250 × 4.6 mm) running with normal phase 92 hexane:
8 i‐propanol: 0.2 trifluoroacetic acid at 5°C. Injection vol-
umes of 5 μL were used and run time 40 minutes. The
typical retention times of rac‐3,3,3‐trifluorolactic acid
were 21.4 and 23.1 minutes. These peaks were close but
clearly resolved with w1/2 ave of 1.0 minute (R = 1.6).
Peaks were detected from UV absorption at 210 nm.
The quantification of the racemate gave an acceptable
ratio of 50.2: 49.8% based on peak area.
Crystals of the diastereomeric salt
6
(S)‐
[NH3CH(CH2OH)Ph](R)‐[CF3CH(OH)COO] (86 mg, Fw
281, 63% yield) can be obtained by further addition of
(S)‐phenylglycinol (0.5 mmol. 70 mg) to the filtrate
solution, crystals obtained after crystallization at room
temperature for 8 hours. The product is contaminated
ith salt 5; therefore, its enantiomeric excess was not
determined.
The % ee in salts came from (R)/(S) peak integrations
after crude 3,3,3‐trifluorolactic acid was obtained from
dried organic extract, after salts treated with 50:50 ethyl
acetate: 1 M HCl soln. to remove cation to aqueous phase.
2.5 | Tandem resolution of rac‐3,3,3‐
trifluorolactic acid using (S)‐1‐phenylethyl‐
amine
Equimolar amount of rac‐3,3,3‐trifluorolactic acid (7.0 g,
48.5 mmol) and (S)‐1‐phenylethylamine (6.4 mL, 50
mmol,) in 90 mL dry ethyl acetate (3 Å UOP molecular
sieves) were refluxed for 2 hours under N2 with flow
extractor charged with freshly calcined 3 Å UOP molecu-
lar sieves for removal the traces of water. Solution was
slowly cooled to 0°C in ice bath and left for 3 hours.
Resulting colorless block crystals of 1 were filtered off
(5.40 g, 84% yield, 92% ee).
A total of 7 mL of water was added to the filtrate,
heated to full dissolution of precipitate, and slowly
cooled down to ambient temperature. The platy white
crystals of 2 settle out after 3 hours (2.91 g) and a further
crop (2.65 g) overnight. Combined yield of salt 2
(S)‐[NH3CHMePh] (R)‐[CF3CH(OH)COO]·H2O) was
5.56 g, 81% yield (96% ee).
3 | RESULTS
3.1 | Resolution 1: Reinvestigation using
1‐phenylethylamine
In the previous report for the resolution of rac‐3,3,3‐
trifluorolactic acid, an equimolar requivalent of (S)‐1‐
phenylethylamine was added to a 9:1 ethyl acetate:
hexane solution of the acid.12 This was then heated at
60°C for 30 minutes and upon cooling afforded a solid
in 75% yield with 68% ee. Three crystallizations were
required to obtain a reasonably enantiopure sample. To
examine the system in more detail, the reaction was
duplicated on a mmol scale reaction in ethyl acetate.
Solids collected were in ca. 70% yield, and these were
ground and a powder X‐ray pattern was recorded of the
product mixture.
A
single crystal structure determination of
a
2.6 | Tandem resolution of rac‐3,3,3‐
specimen from the solid mixture showed the structure
to be the (S)‐(S) salt 1, (S)‐[NH3CHMePh]
(S)‐[CF3CH(OH)COO]. Details of this and other single
crystal structure determinations are given in Table 1.
The structure of an ion pair in 1 is shown in Figure 2.
Crystals of 1 are monoclinic, space group P21 and have
two ion pairs per cell (Z′ = 1, Z = 2) and a cell volume of
652.2Å3 at 100 K. The structure refined to acceptably low
(R)‐indices and Flack parameter with a clean final elec-
tron density map (no difference peaks or holes >0.2
e−Å−3) consistent with a clean (S)‐(S) cation‐anion stereo-
chemistry for the salt in the single crystal specimen. The
structure itself was thus consistent with a more optimal
resolution; hence, the phase purity of the solid product
trifluorolactic acid by sequential use of (S)‐
phenylglycinol and (S)‐1‐phenylethylamine
Rac‐3,3,3‐trifluorolactic acid (10.31 mmol, 1.485 g), (S)‐
phenylglycinol (5.4 mmol, 0.739 g) were heated (60°C, 2
hours) in 5.5 mL ethyl acetate. Upon cooling, colorless
plates of 5 (S)‐[NH3CH(CH2OH)Ph] (S)‐[CF3CH(OH)
COO] (1.102 g, 76% yield, 94 % ee) were obtained by fil-
tration. Then water (0.6 mL) and (S)‐1‐phenylethylamine
(5.4 mmol, 0.657 g) were added to the mother liquor
and crystals of 2 (S)‐[NH3CHMePh](R)‐[CF3CH(OH)
COO]·H2O (0.835 g, 58% yield, 95 % ee) after 5 hours
and a further crop (170 mg) collected overnight to give
a total yield of 69%.

4
WONG ET AL.
TABLE 1 Crystal data summaries for 1‐6
Compound
1
2
3
4
5
6
Abbreviated
Name
(S)‐[NH3CHMePh] (S)‐[NH3CHMePh] (S,S)‐[NH3CH
(S,S)‐[NH3CH
(CH2OH)CHOHPh]
(R)‐[CF3CH
(S)‐[NH3CH
(CH2OH)Ph]
(S)‐[CF3CH
(OH)COO]
(S)‐[NH3CH
(CH2OH)Ph]
(R)‐[CF3CH
(OH)COO]
(S)‐[CF3CH(OH)
(R)‐[CF3CH(OH)
(CH2OH)CHOHPh]
COO]
COO]·H2O
(R)‐[CF3CH
(OH)COO]
(OH)COO]
Code/CSD
number
1899879
1899880
1899881
1899882
1899883
1899884
Empirical
formula
C11H14F3NO3
C11H16F3NO4
C12H16F3NO5
C12H16F3NO5
C11H14F3NO4
C22H28F6N2O8
Formula weight 265.23
Temperature/K 104(6)
283.25
311.26
100(2)
311.26
281.23
562.46
100(2)
100(2)
100(2)
100(2)
Crystal system
Space group
a/Å
Monoclinic
Monoclinic
P21
Monoclinic
P21
Orthorhombic
P212121
5.12662(13)
9.9597(2)
26.4076(9)
90.00
Monoclinic
P21
Monoclinic
P21
P21
8.2984(3)
6.5985(3)
12.0458(5)
90.00
8.2456(4)
5.8699(3)
13.2059(9)
90.00
5.1286(2)
9.9509(3)
13.2398(5)
90.00
5.00584(13)
8.83174(18)
13.6782(4)
90.00
9.25040(18)
5.11803(11)
25.9488(5)
90.00
b/Å
c/Å
α/°
β/°
98.618(4)
90.00
91.876(6)
90.00
97.818(4)
90.00
90.00
98.511(3)
90.00
98.281(2)
90.00
γ/°
90.00
Volume/Å3
652.15(5)
2, 1
638.83(6)
2, 1
669.40(4)
2,1
1348.37(6)
4, 1
598.06(3)
2, 1
1215.71(4)
4, 2
Z, Z′
ρcalc g/cm3
1.351
1.473
1.544
1.533
1.562
1.537
Radiation,
μ/mm‐1
1.097
1.215
1.278
1.269
1.298
1.277
F(000)
276
296
324
648
292
584
Crystal size/mm3 0.10 × 0.05 × 0.02 0.10 × 0.05 × 0.01 0.12 × 0.08 × 0.02
0.20 × 0.10 × 0.02
0.10 × 0.10 × 0.02 0.13 × 0.08 ×
0.04
2Θ maximum/°
135
135
135
135
135
134
Index ranges
−8 ≤ h ≤ 9,
−7 ≤ k ≤ 7,
−14 ≤ l ≤ 13
−9 ≤ h ≤ 9,
−7 ≤ k ≤ 7,
−15 ≤ l ≤ 15
−6 ≤ h ≤ 6,
−11 ≤ k ≤ 11,
−14 ≤ l ≤ 15
−4 ≤ h ≤ 6,
−11 ≤ k ≤ 11,
−31 ≤ l ≤ 27
−5 ≤h ≤ 4,
−10 ≤ k ≤ 10,
−16 ≤ l ≤ 16
−11 ≤ h ≤ 9,
−6 ≤ k ≤ 6,
−30 ≤ l ≤ 30
Total, indep.
3430, 2298
3699
4065, 2369
3624, 2394
3165, 2093
10450, 4210
reflections (%)
Data quality
indices (%)
Rint = 0.0247,
Rsig = 0.0392
Rint = twin,
Rsig = 0.0313
Rint = 0.0234,
Rsig = 0.0327
Rint = 0.0219,
Rsig = 0.0339
Rint = 0.0294,
Rsig = 0.0387
Rint = 0.0263,
Rsig = 0.0320
Data/restraints/ 2298/1/166
parameters
3699/1/198
1.018
2369/1/214
2394/0/214
2093/1/192
4210/1/383
Goodness‐of‐
1.042
1.023
1.059
1.016
1.025
fit F2
Final R indexes R1 = 0.0360,
[I ≥ 2σ (I)] wR2 = 0.0939
R1 = 0.0305,
wR2 = 0.0756
R1 = 0.0295,
wR2 = 0.0765
R1 = 0.0277,
wR2 = 0.0677
R1 = 0.0260,
wR2 = 0.0628
R1 = 0.0284,
wR2 = 0.0689
Final R indexes R1 = 0.0378,
R1 = 0.0328,
wR2 = 0.0765
R1 = 0.0306,
wR2 = 0.0775
R1 = 0.0287,
wR2 = 0.0683
R1 = 0.0280,
wR2 = 0.0639
R1 = 0.0299,
wR2 = 0.0698
[all data]
wR2 = 0.0955
Diff. peak/
hole eÅ−3
+0.19/−0.14
+0.32/−0.14
0.34/−0.18
+0.20/−0.20
+0.24/−0.14
+0.21/−0.15
Flack parameter 0.06(16)
−0.07(8)
−0.11(10)
−0.10(8)
0.04(9)
−0.05(7)

WONG ET AL.
5
FIGURE 2 Ion pair from the crystal structure of
[NH3CHMePh](S)‐[CF3CH(OH)COO]
1 (S)‐
FIGURE 4 Asymmetric unit from the X‐ray structure of 2 (S)‐
[NH3CHMePh](R)‐[CF3CH(OH)COO]·H2O
was suspect. A simulation of the powder X‐ray diffraction
pattern from 1 showed that corresponding peaks were
found prominently in the experimental p‐XRD pattern
of the bulk powder (Figure 3).
However, these were not exclusive, and many
additional diffraction lines could not be indexed. This
implied the presence of a second phase. After mounting
of several specimens from the batch, a crystal with
similar but distinctly different unit cell constants was
identified.
using the Olex2 X‐ray software program with a split
position for C2 with occupancies 93:7 corresponding to
the (S)‐ and (R)‐anion configurations, respectively.
This type of “pyramidal” disorder may be found where
the chiral C centre is directly bonded by H.21 Here, it is
possible within the crystal, since it allows minimal
positional disruption for the three main substituents at
the chiral centre, namely, the carboxylate, the alcohol,
and the CF3 group. Closer examination of phases 1 and
2 shows that inspite of the cell similarities, the internal
packing arrangements are quite unrelated, as shown in
Figure 5.
The formation of a solvate has been identified by
Fogassy et al as a factor that may favour one enantiomer
in diastereomeric salt resolutions.22,23 More recently, we
have shown that the resolution of 1,2‐diamonopropane
by chiral spiroborate anion was facilitated by formation
of a methanol solvate.24 However, in this case, the
possibility to form hydrate 2 apparently complicates the
precipitation process.
The X‐ray structure determination of this revealed it
be the (S)‐(R) salt 2 (S)‐[NH3CHMePh](R)‐[CF3CH(OH)
COO]·H2O which crystallized as
a
monohydrate.
Structural details are given in Table 1, and the monoclinic
unit cell metrics are indeed similar to 1, and the space
group is again P21. The unit cell volumes are identical
within two standard deviations although the asymmetric
unit now also contains a water of hydration (Figure 4).
The handedness in the structure is indicated by the
(S)‐configuration of the cation used in the reaction. This
time, the initial (R)‐value was not of high quality, the
Flack parameter was of high uncertainty, and some
residual electron density could be found near C2 the
chiral carbon of the lactate. Finally, this was modelled
It is likely that due to entropic factors,25 upon cooling,
anhydrous phase 1 is first formed at a higher tempera-
ture. Subsequently, 2, which having a smaller volume
with an extra water is more efficiently packed, is then
favoured at lower temperatures closer to ambient.
Resolutions solely favouring 1 might be possible by
ensuring a higher temperature at which nucleation first
occurs, as well as scrupulous removal of water, both of
which will disfavour hydrate formation. Conversely,
room temperature crystallizations from semi‐aqueous
solvent mixtures might afford pure 2. However, pure
phase 2 would still only offer less than 86% ee which
was found even after removing of the minor (S)‐anion.
The addition of activated 3 Å molecular sieves to the
reaction absorbed traces of water and resulted in
successful formation of pure phase 1. Subsequent
addition of water to the filtrate then allowed for facile
isolation of 2 with high enantiopurity. This showed that
a successful resolution using 1‐phenylethylamine was
indeed possible.
FIGURE
3
Powder XRD patterns indicating solid is a
coprecipiatate of phases 1 and 2

6
WONG ET AL.
FIGURE 6 Ion pair in
3
(S,S)‐[NH3CH(CH2OH)CHOHPh]
(R)‐[CF3CH(OH)COO] with R2 (10) motif (50% ellipsoids)
2
anticipated, a complex network of hydrogen bonds
holds the crystals together since there are six H‐bond
donors per ion pair.
The P‐XRD of the solid indicated reasonable phase
purity, but there were a few minor additional peaks that
could not be indexed to the unit cell of 3. Our initial
assumption was that the additional peaks might belong
to the diastereomeric salt, as was the case for the α‐
methylbenzylammonium phases 1 and 2. Successive
recrystallizations with 90% recoveries were undertaken,
but surprisingly, the resulting powder patterns
(Figure 7) indicated that the contaminant peaks were
reinforced. A crystal 4 from the recrystallized material
was mounted and its unit cell found to be related but dif-
ferent to 3. Compound 4 has an orthorhombic P212121
rather than monoclinic P21 cell, with the a‐ and b‐axes
similar in both, but doubling of the longer c‐axis from
around 13 to 26 Å. Its single crystal structure revealed
that 4 was a polymorphic form of 3 and the salt again
FIGURE 5 Packing diagrams for 1 (top) and 2 both viewed
along [010] monoclinic b‐axis, indicating the unrelatedness of
their structures
3.2 | Resolution 2: Reinvestigation using
2‐amino‐1‐phenylpropan‐1,3‐diol
Although more expensive than 1‐phenylethylamine,
Seebach et al reported several successful resolutions using
the chiral (R,R)‐ and (S,S)‐forms of 2‐amino‐1‐phenyl‐
propan‐1,3‐diol.
These
included
3‐hydroxy‐2‐
trifluoromethylpropionic acid,26 as well as 3,3,3‐
trifluorolactic acid10 for which an efficient resolution
with >90% ee in the first crystallization step was claimed.
No crystal structure of the product was determined in the
original report. We therefore decided to reinvestigate the
resolution of rac‐3,3,3‐trifluorolactic acid using the (S,S)‐
2‐amino‐1‐phenylpropan‐1,3‐diol.
A crystal structure
determination of the main product phase 3 showed the
salt to contain predominantly the (R)‐3,3,3‐trifluorolactic
acid anion (Figure 6).
A particular ion pair can be identified in this struc-
ture with two intermolecular hydrogen bonds forming
FIGURE 7 Experimental powder XRD patterns of crystallized
solids (black, red, and blue) versus simulated from structures of
polymorphs 3 and 4
a R2 (10) ring by Etter notation.27 As might be
2

WONG ET AL.
7
had (S,S)+(R)− configurations. The relationship in pack-
ing between the two polymorphs is well illustrated in
Figure 8, viewed along [100] direction in each structure.
Phase 3 (top) shows packing in two cells along c‐direction
corresponding to a single doubled cell for 4 in which a
second set of ions is rotated by 180° to form the screw axis
along the doubled‐c.
Notably, the H‐bond network in the two structures is
fully conserved, but they differ in the packing regions
between Ph and CF3 groups.
FIGURE 9 (R/S)‐[CF3CH(OH)COO] anion disorder model in 3.
Since the additional diffraction peaks do not belong to
a separate diastereomeric salt phase, the question is why
is there only 90% ee in the product solids? Careful exam-
ination of the anion sites in 3 and 4 reveals that although
both structures are of high quality, in 3, there are ghost
peaks of +0.34 and +0.19eÅ−3 near to the lactate chiral
carbon suggesting the presence of the (S)‐enantiomer of
3,3,3‐trifluorolactate anion. The main extra peak is from
the OH position and the weaker one an alternative C2
chiral carbon position for this minor (S)‐component.
The configurational swap can be achieved with little
reorientation of the CF3 and COO groups and switching
of two OH─O hydrogen bonds of around 2.8 Å for the
major component to a shorter OH─O of 2.6 Å for the
minor one (Figure 9). It is of note that the monoclinic
form which is slightly more efficiently packed is replaced
by the orthorhombic one upon recrystallization, and this
may be a subtle interplay of competing entropy and
enthalpy factors in the two structures.
3.3 | Resolution 3: New studies using
phenylglycinol
The high crystallinity and tractable nature of the salts 1
and 2 led us to seek a subtle modification of the crystalliz-
ing cation, which would have different packing, but
retain the compatibility with the 3,3,3‐trifluorolactate
anions. One option was to extend the methyl group to
ethyl and the other to incorporate an alcohol group to
make it CH2OH.
Figure 9. (R/S)‐[CF3CH(OH)COO] anion disorder
model in 3
Free amine bases with both (R)‐ and (S)‐ forms for
both options are commercially available. It was decided
to investigate phenylglycinol system, since it was felt the
alcohol group will place more constraints on packing
and reduce likelihood of pyramidal disorder. That addi-
tional hydroxy groups can assist resolutions is well
established.28 Recently, we found that rac‐phenylglycinol
could be cleanly resolved by chiral spiroborate anions,
whereas attempts for rac‐1‐phenylethylamine gave (R/
S)‐pyramidal disordered sites.21 An attempt to resolve
3,3,3‐trifluorolactic acid via diastereomeric salt crystalli-
zation was set up as previously, although this time using
around 0.5 mmol (S)‐phenylglycinol in place of 1.0 mmol
(S)‐1‐phenylethylamine.
The ethyl acetate solution was heated at 60°C for 2
hours and cooled to ambient temperature. A crystalline
solid 5 was isolated in 76% yield. A single crystal structure
determination of the product showed it to be the (S)‐(S)
salt
5
(S)‐[NH3CH(CH2OH)Ph](S)‐[CF3CH(OH)COO]
(Figure 10).
This is also monoclinic P21 with some metrical simi-
larities in cell constants to 1 and 2. The addition of the
alcohol functionality in the cation introduces an extra
group that can serve as donor in forming a H‐bond to
FIGURE
8
Polymorphic (S,S)‐[NH3CH(CH2OH)‐CHOHPh]
(R)‐[CF3CH(OH)COO] 3 (monoclinic, top) and 4 (orthorhombic,
bottom)

8
WONG ET AL.
FIGURE 10 Ion pair in
5 (S)‐[NH3CH(CH2OH)Ph] (S)‐
[CF3CH(OH)COO] (50% ellipsoids)
the alcohol OH of the anion. Interestingly, the volume of
the unit cell 598Å3 is substantially contracted from that
of 1 and 2, implying a substantially improved packing
efficiency than either α‐methylbenzylammonium salt.
This feature might indicate a clean resolution could be
indicated.
Powder XRD of the isolated bulk solid showed that the
only major crystalline peaks belonged to phase 5
(Figure 11). The enantiopurity in 5 was also supported
by excellent R‐value, Flack refinement, and low residual
electron density features in the final difference map.
The structure of 5 entails four hydrogen bonds from
the (S)‐cation to four separate (S)‐3,3,3‐trifluorolactate
anions, as well as an additional OH─O from the lactate
alcohol to the cation ─OH. The results from chiral chro-
matography supported the high 92% ee from this single
crystallization step.
FIGURE 12 Asymmetric unit with Z′=2 for 6 (S)‐[NH3 CH
(CH2OH)Ph](R)‐[CF3CH(OH)COO] (50% ellipsoids)
The packing efficiency in this diastereomeric (S)‐(R)
salt is considerably lower than the (S)‐(S) form 5, the den-
sity of which is about 1.3% higher. This differentiation
helps explain the preference for the (S)‐(S) salt. The
extended H‐bond network in 5 could not readily allow
for the type of pyramidal or OH/H type disorders that
were found at the anion sites in compounds 2 and 3,
respectively.
To further investigate the effective resolution in this case,
we also attempted to crystallize the diastereomeric salt 6
(S)‐[NH3CH(CH2OH)Ph] (R)‐[CF3CH(OH)COO] by adding
the (S)‐phenylglycinol to filtrate that is enantiomeric rich
in (R)‐CF3CHOHCOOH. This crystallized with two ion
pairs per asymmetric unit, which form an R4 (10) hydrogen
3
bonded ring as shown in Figure 12.
4 | DISCUSSION
4.1 | Comparison of resolving agents for
CF3‐lactate
Resolution via diastereomeric salt formation is a classical
method22,29 which is becoming somewhat neglected
partly due to the development of new routes to efficient
asymmetric syntheses with enantiomeric control such as
organocatalysis.30-32 However, the combined use of pow-
der diffraction and modern single crystal X‐ray structure
determination can be of value in understanding the
mechanism of resolution and in identifying the issues
when such resolutions are not fully efficient.
The results from the resolution in the three systems
studied (sections 2.2‐2.4) is summarized graphically in
Figure 13.
In the case of resolution of 3,3,3‐trifluorolactic acid by
(S)‐1‐phenylethylamine, the product solids were not pure
FIGURE 11 P‐XRD patterns: Bulk powder (yellow, middle)
matches simulated from (S)‐(S) salt 5 (blue), not diastereomeric 6
(orange)

WONG ET AL.
9
FIGURE 13 Summary of 3,3,3‐trifluorolactic acid resolution outcomes using (S)‐amines (1 mmol scale)
phase, and a mixture of 1 (S)‐[NH3CHMePh] (S)‐
[CF3CH(OH)COO] and (S)‐[NH3CHMePh] (R)‐
Phases 3 and 4 were found to be monoclinic and ortho-
rhombic polymorphic arrangements respectively with
preserved H‐bonding patterns. These differ from each
other by taking AAAA type translationally related molec-
ular layers in the monoclinic form 3 and changing to an
ABAB arrangement, with 180° rotation of alternate
layers, in the orthorhombic form 4. This effectively dou-
bles the unit cell c‐axis from 13 to 26 Å.
Detailed study of crystallization conditions might
allow for one or other form to be obtained pure phase,
but it is notable that again we find phase 3 has a slight
enantio contamination by careful analysis of the single
crystal structure. In this case, it is due to an H/OH posi-
tional disorder in the 3,3,3‐trifluorolactate anion. In the
predominant (R)‐ form, two OH─OH‐bonds are found
involving the lactate alcohol group as both donor and
acceptor. When the H and OH positions are effectively
switched, just one H‐bond is formed, but this is shorter
and thus energetically competitive. Further recrystalliza-
tion to give higher enantiopurity appears to favour ortho-
rhombic 4.
In the case of (S)‐phenylglycinol, a new alternative res-
olution system is presented. This benefits from having
conditions affording a single phase product in good yield
that is largely free of enantio contamination.
The cost and availability of both enantiomers of the
resolving cations may also be a consideration and ulti-
mately the use of (S)‐phenylglycinol, at least for the first
step may be the best overall approach. The detailed study
of the structures of the product phases is valuable in giv-
ing insight into the mechism for resolution and in some
cases how to improve its efficiency, but finally the
2
[CF3CH(OH)COO]·H2O was formed. The single crystal
structures of each phase were determined from speci-
mens taken from the same product batch. These revealed
that whist internally the (S)‐(S) salt was effectively
enantiopure the same was not true for the hydrated phase
2, which could be described as a diastereomeric solid
solution. Such issues were noted as a source of potential
problems in previous resolutions in which the enantio-
mers showed considerable shape similarity.21 A similar
example in which 1‐phenylethylamine was involved was
in the resolution of rac‐malic acid.33 The product salt
was found to be predominantly (R)‐[NH3CHMePh] (S)‐
[Malate] which was contaminated by the (R)‐anion in a
single solid‐solution phase.
Here, the S‐XRD studies help explain the low resolu-
tion efficiency found in the previous study14 and also
indicate that isolation of pure phase 1 would be a better
strategy than isolating pure phase 2. Fortunately,
suppression of 2 is fairly easily achieved by the introduc-
tion of 3 Å molecular sieves to thoroughly dessicate
the reaction and allows production of pure phase 1 in
acceptable yield.
The reinvestigation of the resolution of 3,3,3‐
trifluorolactic acid using 2‐amino‐1‐phenyl‐propan‐1,3‐
diol supports the earlier findings8 that an effective
resolution can be achieved after two steps. The first
crystallization afforded two phase types 3 and 4;
however, in this case, both were predominantly of
the chirality (S,S)‐[NH3CH(CH2OH)CHOHPh] (R)‐
[CF3CH(OH)COO].

10
WONG ET AL.
enantioexcess of isolated solids, or residual solutions
needs to also be proved.
and ensure its protonation. The UV detection at 210 nm
was found to be sufficiently sensitive to give quantitation
of the rac material with adequate separation (R = 1.6)
and integrated values of 50.2% and 49.8% for the two
enantiomeric peaks, which eluted at times 21.5 and 23.1
minutes (Figure 14).
The identification of the two peaks only came from the
later screening of the 3,3,3‐trifluorolactic acid from the
resolved solids and is consistent with the first eluted peak
being the (S)‐isomer. The % ee in salts came from relative
peak integrations obtained after dissolution of the salts
with 50:50 ethyl acetate: 10% HCl mixtures.
As shown in Table 2, the best strategies have been
evaluated. Entry I gives our results for the classical
method of Seebach et al10 after one (Ia) or two steps
(Ib). Entries II and III are tandem resolutions which
may be slightly superior. Entry II involves two steps with
the same (S)‐1‐phenylethylamine base which (S)‐(S) salt 1
is first isolated under anhydrous conditions followed by
the (S)‐(R) hydrate 2 under aqueous conditions. Entry
III which is also competitive is similar to II but uses (S)‐
phenylglycinol in the first step.
Optical methods such as circular dichroism are rather
problematic for 3,3,3‐trifluorolactic acid and derivatiza-
tions to afford spectroscopic handles are both time‐
consuming and have the potential for introducing either
diastereomer ratio distortion or even additional racemiza-
tion. The availability of new chiral HPLC chromatogra-
phy columns (Lux 5 μm Amylose‐1 LC Column 250 ×
4.6 mm) able to withstand direct application of acids
and bases is a useful development and herein we have
attempted to prove that separation protocols for quantify-
ing its % ee are possible.
First, the chromatography conditions for separation
and detection of the isomers of rac‐3,3,3‐trifluorolactic
acid were established using 92 hexane: 8 i‐propanol as
the eluting solvent with 0.2 eq. trifluoroacetic acid added
to provide an acidic buffer for the 3,3,3‐trifluorolactic acid
The cations are segregated into the aqueous acid layer.
After further extractions with ethyl acetate, the combined
organic fractions were dried with Na2SO4 evaporated to
dryness to give quantitative yield of the enriched 3,3,3‐
trifluorolactic acid.
4.2 | Resolution by tandem crystallization
The results from the three resolutions of 3,3,3‐
trifluorolactic acid indicate that all three benzyl‐ or aryl‐
amines could form the basis of an efficient resolution
(Figure 12). The price and availability of both enantio-
mers of the amines studied can be an issue. In general,
1‐phenylethylamine would be typically preferred, since
the cost of both enantiomers is low.
However, in our studies, we have shown that unless
some care is taken, both diastereomeric salts will precip-
itate in two separate phases. Fortunately, one of these is
a hydrate, the nucleation of which can be suppressed
using molecular sieves.
With these points in mind, several overall strategies of
resolution by tandem (step‐wise) crystallization that offer
reasonable resolution efficiency with isolation of both
enantiomers in acceptable yield and relatively simple
TABLE 2 Summary of 3,3,3‐trifluorolactic acid resolution on
gram scale
Configuration
of Salt
Yield
%
% ee
TFLA
Amine
Ia
Ib
II
3 (S,S)‐(R)a
84
92
4 (S,S)‐(R)b
73
99
1 (S)‐(S)c
70
77
92
96
2 (S)‐(R)·H2Od
III
5 (S)‐(S)a
76
69
94
95
2 (S)‐(R)·H2Od
aAfter precipitation from racemic acid.
bAfter 2nd crystallization of the salt from Entry Ia.
cFrom dry ethyl acetate.
FIGURE 14 Chiral HPLC chromatograms showing separated
eluted [CF3CHOHCOOH] peaks for (A) racemic starting material,
(B) (R)‐acid from 3, and (C) (S)‐acid from 1
dAddition of 12% water to the filtrate for crystallization.

WONG ET AL.
11
FIGURE 15 Two possible tandem strategies
work up can be devised and two of the better approaches
are detailed in Table 2 and Methods A and B (Figure 15).
ee. A single high‐yield recrystallization will produce this
phase with >99% ee. The filtrate solution may then be
readily treated with a further 5 mmol eq. of a second
resolving base. In Method B, the (S) enantiomer of 1‐
phenylethylamine can be used to give a high yield and
enantiopurity of the (S)‐(R).H2O salt 2 with the addition
of water to the mixture in the second step. In this case,
the anhydrous form of (S)‐(S) salt 1 is strongly
disfavoured as little (S)‐3,3,3‐trifluorolactate remains in
solution.
4.2.1 | Tandem Method A
This method requires the availability of just one enantio-
mer of 1‐phenylethylamine. If this is (S)‐1‐phenylethyl-
amine, then the first step should be carried out under
anhydrous conditions (eg, 3 Å molecular sieves) using
equimolar ratio of (S)‐1‐phenylethylamine to rac‐3,3,3‐
trifluorolactic acid.
This can give 84% yield of the (S)‐(S) salt 1 with 92%
ee. This then requires a single recrystallization (at around
90% yield) to afford 1 with 99% ee. In Step 2, the filtrate
can also then be readily worked up by simple addition
of water to give 86% yield of the hydrated (S)‐(R) salt 2
with high 96% ee.
The chiral 3,3,3‐trifluorolactic acids can be released
from these salts by addition of acid and organic phase
extraction into ethyl acetate. However, it is suggested this
may be carried out just prior to their further use, since the
chiral acids themselves are highly hygroscopic and
untractable solids. Hence, longer‐term storage is better
effected by leaving the enantiopure 3,3,3‐trifluorolactate
anions in their organic salt forms.
4.2.3 | Other possible tandem methods
In theory, any chiral amines such as 2‐amino‐1‐
phenylpropan‐1,3‐diol that may form salts with 3,3,3‐
trifluorolactic acid can be used in Step 2. However,
due to the high availability and low cost of
1‐phenylethylamine, our suggested tandem methods
show an obvious advantage against others.
4.2.4 | Summary of tandem resolutions
A key point in the tandem resolutions is that by switching
to a second chiral amine for Step 2, the possibility of a
lingering contamination through formation of racemic
crystals containing enantiomers of both cation and
anion is avoided. The further advantage of the
phenylglycinol system is that only 0.5 eq. need be used
in the first step (both of the other systems require 1.0
eq. of base to work effectively) allowing the other base
to be directly added to the filtrate solution in the second
step. Our work shows a tandem resolution method with
1‐phenylethylamine because of its availability and cost.
4.2.2 | Tandem Method B
Method A is relatively simple and inexpensive. Competi-
tive alternative methods starting with (S)‐phenylglycinol
in Step 1 can also be devised. Reaction of 5 mmol eq.
(S)‐phenylglycinol with rac‐3,3,3‐trifluorolactic acid
affords the (S)‐(S) salt 5 in 76% isolated yield and 94%

12
WONG ET AL.
In principle, however, any combination involving
another amines that can form diastereomeric salts with
3,3,3‐trifluorolactic acid (not limited to mentioned resolu-
tion) could also be successful.
Herman H.‐Y. Sung
0332
Valerii V. Vashchenko
1949
https://orcid.org/0000-0002-3370-
https://orcid.org/0000-0003-0831-
Ian D. Williams https://orcid.org/0000-0001-8743-401X
5 | CONCLUSION
REFERENCES
Under suitable conditions, effective resolution of 3,3,3‐
trifluorolactic acid can be carried out by all the three
benzylammonium ion systems studied. The resolution
using 1‐phenylethylamine was reinvestigated, and the
modest % ee obtained previously14 was explained by the
fact that two diastereomeric phases coprecipitate, as indi-
cated by p‐XRD patterns of the product solids. The (S)‐(S)
salt 1 predominates in the mixture but is contaminated by
the (S)‐(R) hydrate 2. This can be suppressed using dry
conditions.
1. Kubota T, Kondo Y, Oyama T, Tanaka T. Synthesis and reac-
tions of 2‐alkoxy‐1,1,3,3,3‐pentafluoropropenes as a synthon
for trifluoromethylated compounds. Nippon Kagaku Kaishi.
1989;21(9):1576‐1586.
2. Takahashi S, Jukurogi T, Katagiri T, Uneyama K. Isomorphic
supramolecular structures via one‐dimensional hydrogen bond-
ing motifs in crystals of chiral difluorolactates, trichlorolactates
and trifluorolactates. CrystEngComm. 2006;8(4):320‐326.
3. Katagiri T, Uneyama K. Chiral recognition by multicenter single
proton hydrogen bonding of trifluorolactates. Chem Lett.
2001;30(12):1330‐1331.
The more expensive (S,S)‐2‐amino‐1‐phenyl‐propan‐
1,3‐diol, use of which was originally reported by
Seebach,10 is indeed quite effective and gives solids with
90% ee. These are actually polymorphic, and the mono-
clinic form 3 allows some diastereomeric contamination
at the anion site. Recrystallization affords a cleaner ortho-
rhombic form 4. Finally, another alternative may be
offered by use of (S)‐phenylglycinol as resolving base.
This affords a single‐phase product (S)‐(S) salt 5 with no
diastereomeric contamination via solid solution. The
(S)‐(R) salt 6 prepared from the separate chiral compo-
nents is found to have less efficient crystal packing,
explaining its higher solubility, and hence the selective
precipitation of 5.
Based on effective resolutions by all the three
arylamine systems, tandem strategies for isolation of both
enantiomers of 3,3,3‐trifluorolactic acid with 99% ee and
approximately 70% isolated yields can readily be devised.
These involve either addition of water or addition of a
second aliquot of base to the filtrate from the original res-
olution and minimal work up.
4. Kubota T, Kanega J, Katagiri T. Application of (S)‐trifluorolactic
acid for chiral derivatizing agent. J Fluorine Chem. 1999;97(1‐
2):213‐221.
5. Ramachandran PV, Gong B, Brown HC. Chiral synthesis via
organoboranes. 40. A simple one‐pot synthesis of the enantio-
mers of (trifluoromethyl) oxirane. A general synthesis in high
optical purities of a‐trifluoromethyl secondary alcohols via the
ring cleavage reactions of the epoxide. J. Org. Chem.
1995;60(1):41‐46.
6. Katagiri T, Obara F, Toda S, Furunashi K. Synthesis of
trfluoroacetic acid from 1,2‐epoxy‐3,3,3‐trifluoropropane—one
pot tandem ring opening‐oxidation reaction of epoxide. Synlett.
1994;1994(07):507‐508.
7. Kuroki Y, Asada D, Sakamaki Y, Iseki K. Enantioselective syn-
thesis of 1,1,1‐trifluoroalkan‐2‐ols by ruthenium‐catalyzed
hydrogenation. Tetrahedron Lett. 2000;41(23):4603‐4607.
8. Sawai N, Ishii S, Nishii T, Ishii, A. Method for manufacturing
optically active fluorolactic acid derivative. Patent WO
2015005341 A1 20150115 (2015).
9. Darrall RA, Smith F, Tatlow JC. Organic fluorides. Part IX. The
formation and resolution of α‐hydroxy‐α‐trifluoromethyl‐
propionic acid. J. Chem. Soc. 1951;0:2329‐2332.
ACKNOWLEDGEMENTS
10. Vondembussche‐Hunnefeld C, Cescato C, Seebach D. Simple
access to (R)‐3,3,3‐trifluorolactic and (S)‐3,3,3‐trifluorolactic
The authors are grateful to the Research Grants Council
(RGC) of Hong Kong for financial support via grant
16306515, as well as funds of the NAS of Ukraine (project
# 0112U002186) and STCU (project # P654a).
acid
and
to
(R)‐trifluoromethyloxirane
and
(S)‐
trifluoromethyloxirane. Chem Ber–Receuil. 1992;125:2795‐2802.
11. Lermontov SA, Shkavrov SV, Kuryleva NV. Uncatalyzed
Meerwein–Ponndorf–Verley reduction of trifluoromethyl car-
bonyl compounds by high‐temperature secondary alcohols.
J. Fluorine Chem. 2003;121(2):223‐225.
ORCID
12. Ishii A, Kanai M, Yasumoto M, Inomiya K, Kuriyama Y,
Katsuhara Y. Synthetic application of 3,3‐dichloro‐1,1,1‐
trifluoroacetone (DCTFA) and 3,3,3‐trichloro‐1,1,1‐trifluoro‐
acetone (TCTFA) for trifluorolactic acid derivatives. J. Fluorine
Chem. 2004;125(4):567‐571.
Lawrence W.‐Y. Wong
2896
Elena V. Vashchenko
6034
https://orcid.org/0000-0003-4170-
https://orcid.org/0000-0003-2380-

WONG ET AL.
13
13. Ishii A, Yasumoto M, Tsuruta H, Inomiya N, Otsuka T, Ueda H,
Efficient optical resolution of trifluorolactic acid, JP 2006232726
A 20060907, (2006).
26. Gotzo SP, Seebach D. Preparation of (R)‐ and (S)‐3‐hydroxy‐2‐
trifluoromethylpropionic acid by resolution with (R,R) and
(S,S)‐2‐amino‐1‐phenylpropane‐1,3‐diol. Chimia. 1996;50:20‐23.
14. Yasumoto M, Ueki H, Ono T, Katagiri T, Soloshonok VA. Self‐
disproportionation of enantiomers of isopropyl 3,3,3‐(trifluoro)
lactate via sublimation: sublimation rates vs. enantiomeric com-
position. J Fluorine Chem. 2010;131(4):535‐539.
27. Etter MC, MacDonald JC, Bernstein J. Graph‐set analysis of
hydrogen‐bond patterns in organic crystals. Acta Cryst B.
1990;46(2):256‐262.
28. Kodama K, Hayashi N, Fujita M, Hirose T. Direct enantio‐
separation of 1‐(2‐hydroxyphenyl) ethylamines via diastereo-
meric salt formation: chiral recognition mechanism based on
the crystal structure. RSC Adv. 2014;4(49):25609‐25614.
15. Acs M, Vondembussche C, Seebach D. An efficient method of
preparing (R)‐4,4,4‐trifluoro‐3‐hydroxy butanoic acid and (S)‐
4,4,4‐trifluoro‐3‐hydroxybutanoic acid–resolution with (R)‐1‐
phenylethylamine or (S)‐1‐phenylethylamine. Chimia. 1990;
44:90‐92.
29. Kellogg RM. Practical stereochemistry. Acc. Chem. Res.
2017;50(4):905‐914.
16. Nisar M, Wong LW‐Y, Sung HH‐Y, Haynes RK, Williams ID.
Cocrystals of the antimalarial drug 11‐azaartemisinin with three
alkenoic acids of 1: 1 or 2: 1 stoichiometry. Acta Cryst C.
2018;74(6):742‐751.
30. Yan J, Chen M, Sung HHY, Williams ID, Sun J. An
organocatalytic asymmetric synthesis of chiral β,β‐diaryl‐α‐
amino acids via addition of azlactones to in situ generated
para‐quinone methides. Chemistry–Asian J. 2018;13(17):
2440‐2444.
17. Sheldrick GM. Crystal structure refinement with SHELXL. Acta
Cryst C. 2015;71(1):3‐8.
31. Wang Z, Sheong FK, Sung HHY, Williams ID, Lin Z, Sun J.
Catalytic enantioselective intermolecular desymm‐etrization of
azetidines. J Am Chem Soc. 2015;137(18):5895‐5898.
18. Bourhis LJ, Dolomanov OV, Gildea RJ, Howard JAK,
Puschmann H. The anatomy of a comprehensive constrained,
restrained refinement program for the modern computing
environment–Olex2 dissected. Acta Cryst A. 2015;71(1):59‐75.
32. Qian H, Zhao W, Sung HHY, Williams ID, Sun J. Stereoselective
synthesis of aminoindanols via an efficient cascade aza‐
Michael–aldol reaction. Chem Commun. 2013;49(39):4361‐4363.
19. Dolomanov OV, Bourhis LJ, Gildea RJ, Howard JAK,
Puschmann H. Olex2 a complete structure solution, refinement
and analysis program. J. Appl Cryst. 2009;42(2):339‐341.
33. Kozma D, Tomor K, Novak C, Pokol G, Fogassy E. Study of the
mechanism of optical resolutions via diastereoisomeric salt for-
mation. Thermoanalytical and X‐ray study of solid solution
formation during the resolution of racemic malic acid by (R)‐
alpha‐phenylethylamine. J Thermal Anal. 1996;46(6):1613‐1623.
20. Macrae CF, Bruno IJ, Chisholm JA, et al. Mercury CSD 2.0—
new features for the visualization and investigation of crystal
structures. J Appl Cryst. 2008;41:466‐470.
21. Wong LWY, Tam GSS, Chen X, et al. A chiral spiroborate anion
from diphenyl‐L‐tartramide [B{L‐Tar (NHPh)2}2]− applied to
some challenging resolutions. CrystEngComm. 2018;20(33):
4831‐4848.
SUPPORTING INFORMATION
Additional supporting information may be found online
in the Supporting Information section at the end of the
article.
22. Faigl F, Fogassy E, Nogradi M, Palovics E, Schindler J. Strategies
in optical resolution: a practical guide. Tetrahedron‐Asym.
2008;19(5):519‐536.
23. Kozsda‐Kovacs E, Keseru GM, Bocskel Z, et al. Role of the sol-
vent in optical resolution of tran(S)‐chrysanthemic acid via
diastereomeric salt formation. J Chem Soc, Perkin Trans.
2000;0(1):149‐153.
How to cite this article: Wong LW‐Y,
Vashchenko EV, Zhao Y, et al. Tandem
crystallization strategies for resolution of 3,3,3‐
trifluorolactic acid [CF3CH(OH)COOH] by chiral
benzylamines. Chirality. 2019;1–13. https://doi.org/
10.1002/chir.23127
24. Wong LWY, Kan JWH, Nguyen T, et al. Bis (mandelato) borate:
an effective, inexpensive spiroborate anion for chiral resolution.
Chem Commun. 2015;51(87):15760‐15763.
25. Barnes DM, Barkalow J, Chen Y, et al. Impact of entropic effects
in the classical resolution of a 2‐arylpyrrolidine. Org Proc Res
Devel. 2009;13(2):225‐229.