Professional Documents
Culture Documents
h i g h l i g h t s g r a p h i c a l a b s t r a c t
a r t i c l e i n f o a b s t r a c t
Article history: High throughput general chiral screening method using supercritical fluid chromatography was devel-
Received 2 September 2016 oped. This method takes an advantage of very fast gradient screening (3 min þ 1 min isocratic hold) and
Received in revised form generic enantioselectivity of the combined additive formed by 0.1% trifluoroacetic (TFA) acid and 0.1%
22 October 2016
diethylamine (DEA). The TFA/DEA combined additive was systematically added to organic modifiers
Accepted 2 November 2016
Available online 21 November 2016
methanol and isopropanol. Among five tested polysaccharide-based chiral stationary phases, amylose
tris(3,5-dimethylphenylcarbamate) and cellulose tris(3,5-dimethylphenylcarbamate) provided the best
enantioseparation success rate. Therefore, the proposed initial first-line screening includes four exper-
Keywords:
Supercritical fluid chromatography
iments using these two stationary phases and the above mentioned two combinations: CO2/methanol
SFC and CO2/isopropanol þ the combined additive. If these stationary phases fail in the screening step,
Chiral stationary phases cellulose tris(3-chloro-4-methylphenylcarbamate) and cellulose tris(3,5-dichlorophenylcarbamate) can
Organic modifier be proposed for the screening in the second line.
Combined additive For further optimization in case of insufficient resolution obtained in the screening phase fine tuning
Enantioselectivity of temperature, BPR pressure and gradient slope was tested with unsuccessful results. An improvement
Pharmaceuticals of enantioselectivity was obtained only when gradient elution was replaced by isocratic elution with
substantially lower amount of organic modifier, when changing the concentration of the additive or
when using combined organic modifier, such as methanol/acetonitrile (1:1). Finally, to enable the MS
compatibility, also volatile additives including ammonium formate and ammonium acetate were tested.
* Corresponding author.
E-mail address: nol@email.cz (L. Nov
akov
a).
http://dx.doi.org/10.1016/j.aca.2016.11.002
0003-2670/© 2016 Elsevier B.V. All rights reserved.
200 kova
L. Nova , M. Dousa / Analytica Chimica Acta 950 (2017) 199e210
The results were more encouraging than expected. Volatile buffers thus make an interesting option in
chiral SFC screening methods, however, at the cost of somewhat lower enantioselectivity.
© 2016 Elsevier B.V. All rights reserved.
PDA detector. solution was prepared in methanol. The concentrations of the stock
All injected sample solutions were stored in the autosampler at solutions were 1 mg/mL. Stock solutions were further diluted with
4 C. The partial loop with needle overfill mode was set up to inject tetrahydrofuran in order to obtain a mixture of both enantiomers at
10 mL. Methanol was used as a needle wash solvent. Gradient a concentration of 0.1 mg/mL, which was used for the injection into
elution was performed using CO2 (>99.995%, LindeGas, Hradec SFC system for SST measurements. For screening purposes, a con-
Kralove, Czech Republic) and various combinations of modifiers centration of 0.3 mg/mL was used for an active isomer for faster and
(methanol, ethanol, isopropanol and acetonitrile) and additives easier identification of the isomers. In case of doubts individual in-
(trifluoroacetic acid, diethylamine, ammonium acetate, ammonium jections of each isomer were performed to confirm the peak identity.
formate) at a flow-rate of 3.0 mL/min. Gradient program started at
5% of organic modifier and was linearly increased up to 30% in 3. Results and discussion
3 min. An isocratic step was kept at 30% of organic modifier for
1 min followed by column equilibration (2 min). The influence of 3.1. Selection of stationary and mobile phases for fast SFC screening
temperature was evaluated in the range of 10e60 C. The BPR
(back-pressure regulator) pressure was set to 1500 psi and its Selection of correct stationary phase for chiral separation still
variations were evaluated in the range of 1500e2500 psi. UV remains challenging both in LC and SFC techniques. Based on
detection by means of PDA detector was performed at extracted research previously performed in SFC and LC screening approaches
wavelength corresponding to the absorption maximum of indi- [11e13,21e25,38] and based on practical experience with chiral
vidual tested compounds. separations, five the most promising stationary phases were
The separation was performed using polysaccharide based chi- selected for the initial screening and subsequent comparison. These
ral stationary phases including: Lux 3u Amylose-2: coated amylose included cellulose tris(3,5-dimethylphenylcarbamate), amylose
tris(5-chloro-2-methylphenylcarbamate), provided by Phenom- tris(3,5-dimethylphenylcarbamate) and chlorinated phases
enex (Chromservis, Prague, Czech Republic), Kromasil 3-Amycoat: including cellulose tris(3-chloro-4-methylphenylcarbamate),
coated amylose tris(3,5-dimethylphenylcarbamate), provided by amylose tris(5-chloro-2-methylphenylcarbamate) and finally cel-
Kromasil (Chromservis, Prague, Czech Republic), Lux 3u Cellulose- lulose tris(3,5-dichlorophenylcarbamate). The group of analytes
1: coated cellulose tris(3,5-dimethylphenylcarbamate), provided was selected in order to reflect the current state at the drug market
by Phenomenex (Chromservis, Prague, Czech Republic), Lux 3u including acidic, basic and neutral analytes and being both APIs and
Cellulose-2: coated cellulose tris(3-chloro-4- pharmaceutical intermediates that have to be subjected to chiral
methylphenylcarbamate), provided by Phenomenex (Chromservis, purity control. Table 1 shows basic physico-chemical properties of
Prague, Czech Republic) and Chiralpak IC-3: immobilized cellulose the tested chiral compounds. Log P values reveal, that the range of
tris(3,5-dichlorophenylcarbamate), provided by Daicel (Illkirch polarity was quite large (0.06e6.19). Most of the tested com-
Cedex, France). All columns in this study had dimensions of pounds had basic character which also corresponds to the typical
150 4.6 mm and 3 mm particles. situation at the drug market.
Following the state-of-the art of SFC separations and in agree-
2.3. Standard solutions ment with the previous findings, the first comparison was made
using CO2/methanol þ0.1% TFA for the analysis of acidic compounds
The stock standard solutions of the pharmaceuticals were pre- and CO2/methanol þ0.1% DEA for basic and neutral compounds.
pared in tetrahydrofuran due to its good compatibility with SFC Two comparative mobile phases were made by a combination of
mobile phases and overall good solubility for the selected set of 0.1% DEA and 0.1% TFA added simultaneously first to methanol/CO2
compounds. Less soluble compounds including alaptide, atom- and secondly to isopropanol/CO2. In contrast to previously
oxetine, darifenacin, epinephrine, indacaterol and cetirizine were described technical problems with the system stability described in
dissolved in a mixture of tetrahydrofuran/methanol (9:1). Tamsu- Ref. [33] when using a combination of 0.5% TFA and 0.5% IPA in
losin was not soluble at all in tetrahydrofuran, therefore the stock methanol/CO2, similar issues have never been observed through
Table 1
Physico-chemical properties of chiral pharmaceuticals used in this study.
Compound Molecular formula Molecular weight (g/mol) log P pKa acidic pKa basic Detection wavelength (nm) Acid-base characteristic Active enantiomer
the course of this study despite using also higher concentrations of was the different enantioselectivity provided by the two organic
the additives later in the method optimization phase. This may be modifiers with the combined TFA/DEA additive, which is shown in
attributed to a better solubility of TFA-DEA salt complexes Fig. 2 for the two columns providing the best results. Thus, putting
compared to TFA-IPA salt. TFA-DEA salt is formed in the ratio 1:1, the two conditions together provided very good success separation
while in case of TFA-IPA it may be 2:1 leading to worse salt solu- rate (18 enantiomeric pairs out of 20 at least partially resolved
bility in CO2/methanol mobile phases. using these four conditions). To complete this set of experiments,
All the screening experiments were performed in quick gradient further analyses were performed also with CO2/ethanol þ0.1% DEA
elution (3 min þ 1 min isocratic hold) described in Section 2.2, and 0.1% TFA on Amycoat and Cellulose 1 stationary phases to
which substantially increased the throughput of the screening verify, if there was not some interesting enantioselectivity missed.
phase compared to the isocratic elution. Longer separation time The results showed, that no additional enantioselectivity was
(6 min þ 1 min isocratic hold) was also tested in the preliminary observed compared to CO2/methanol and CO2/isopropanol with the
experiments to verify if the enantioresolution was not compro- combined TFA/DEA additive. The selectivity of this CO2/ethanol-
mised due to the short analysis times. The impact of the time on
enantioseparation was negligible, thus short gradient time of 3 min
was selected.
The comparison of the success rate of the enantioseparation
using single and combined additives on all five tested stationary
phases is shown in Fig. 1. In this presentation completely separated
peaks as well as partially separated peaks are taken as “successful
separation”, because method optimization is considered as
following step, if necessary. Generally, combined TFA/DEA additive
in methanol/CO2 (blue column in Fig. 1) provided always at least the
same, but generally better results compared to single additives in
methanol/CO2 (black column). This improvement could be both
improvement of partial separation to complete baseline separation
or non-resolved enantiomeric pair obtained in case of single ad-
ditives while the separation was obtained with the combined TFA/
DEA additive. The combined TFA/DEA additive added to iso-
propanol/CO2 demonstrated overall lower success rate in the
enantioseparation. However, its benefit was different and often
complementary enantioselectivity compared to methanol with the
combined TFA/DEA additive, especially on Amycoat stationary
phase. Overall, the best enantioselectivity was obtained on Amy-
coat and on Cellulose 1 columns when using CO2/methanol þ0.1%
DEA and 0.1% TFA leading to at least partial separation of 60% of
enantiomeric pairs in both cases (12 out of 20). The second best Fig. 2. Comparison of the success-rate of enantioseparation of individual compounds
enantionselectivity was observed on Amycoat column when using for Cellulose 1 (blue) and Amycoat (black/gray) stationary phases using CO2/methanol
CO2/isopropanol þ0.1% DEA and 0.1% TFA providing separation for and CO2/isopropanol with the combined TFA/DEA additive. Successful separation (full
55% of enantiomeric pairs. Interestingly, both stationary phases separations and partial separations) is represented by the columns above the x-axis
(þ), while non-successful separation is represented by the columns bellow the x-axis
contained 3,5-dimethylphenylcarbamoyl functional groups, either
(). (For interpretation of the references to colour in this figure legend, the reader is
on amylose or on cellulose. As already mentioned, a great benefit referred to the web version of this article.)
Fig. 1. Comparison of the success-rate of enantioseparation of the 20 tested compounds on the five tested polysaccharide-based stationary phases in gradient chiral screening. CO2/
methanol with the addition of 0.1%DEA was used for the analysis of basic and neutral compounds and 0.1% TFA for the analysis of acidic compounds. Comparison was made with
CO2/methanol and CO2/isopropanol with a combined additive composed of 0.1% TFA and 0.1% DEA.
kova
L. Nova , M. Dousa / Analytica Chimica Acta 950 (2017) 199e210 203
Table 2
Final analytical conditions (stationary phase, mobile phase, gradient elution (G) or isocratic elution (ISO)) for both screening phase optimal separations and finely optimized
separations of all compounds. Retention times, repeatability and resolution are also shown.
tR1 tR2 RSD tR1 [%] RSD tR2 [%] RSD A1 [%] RSD A2 [%] RS Column Elution Mobile phase
Alaptide 3.224 (R) 3.529 (S*) 0.04 0.04 0.86 0.45 3.59 Amycoat G MeOH þ 0.1% TFA þ 0.1% DEA
3.008 (R) 3.697 (S*) 0.04 0.04 0.33 0.49 3.55 Amycoat ISO 20% MeOH þ 0.1% TFA þ 0.1% DEA
3.529 (S*) 4.538 (R) 0.06 0.05 0.82 0.40 8.97 Cellulose 2 G MeOH þ 0.1% TFA þ 0.1% DEA
2.212 (S*) 3.474 (R) 0.04 0.03 0.29 0.33 8.20 Cellulose 2 ISO 25% MeOH þ 0.1% TFA þ 0.1% DEA
Aprepitant int. 1.184 1.623 0.08 0.05 0.77 0.32 3.77 Cellulose 2 ISO 15% 2-Pr-OH þ 0.1% TFA þ 0.1% DEA
1.918 2.269 0.03 0.02 0.95 0.27 3.86 Cellulose 2 G 2-Pr-OH þ DEA 0.1%
Atomoxetine 2.375 (S) 2.924 (R*) 0.02 0.01 0.15 0.23 8.55 Cellulose 1 G MeOH þ 0.1% TFA þ 0.1% DEA
1.204 (S) 1.826 (R*) 0.06 0.05 0.22 0.21 4.40 Cellulose 1 ISO 20% MeOH þ 0.1% TFA þ 0.1% DEA
3.928 (S) 4.138 (R*) 0.04 0.03 0.31 0.48 2.14 Cellulose 2 G 2-Pr-OH þ 0.1% TFA þ 0.1% DEA
3.638 (S) 4.796 (R*) 0.03 0.02 0.32 0.05 2.75 Cellulose 2 ISO 15% 2-Pr-OH þ 0.1% TFA þ 0.1% DEA
Boc-saxagliptin 2.583 (S*SSS) 3.236 (RRRR) 0.02 0.04 0.38 0.14 4.01 Amycoat G MeOH þ 0.1% TFA þ 0.1% DEA
0.911 (S*SSS) 1.312 (RRRR) 0.08 0.04 0.28 0.46 2.22 Amycoat ISO 30% MeOH þ 0.1% TFA þ 0.1% DEA
2.408 (S*SSS) 2.597 (RRRR) 0.02 0.02 0.22 0.16 2.50 Cellulose 1 G MeOH þ 0.1% TFA þ 0.1% DEA
1.772 (S*SSS) 2.180 (RRRR) 0.04 0.05 0.24 0.79 2.24 Cellulose 1 ISO 15% MeOH þ 0.1% TFA þ 0.1% DEA
Cinacalcet 2.147 (S) 2.682 (R*) 0.06 0.05 0.16 0.20 7.60 Amycoat G 2-Pr-OH þ 0.1% TFA þ 0.1% DEA
1.639 (S) 2.806 (R*) 0.05 0.06 0.24 0.36 6.49 Amycoat ISO 13% 2-Pr-OH þ 0.1% TFA þ 0.1% DEA
2.385 (R*) 2.530 (S) 0.02 0.02 0.10 0.12 2.00 Cellulose 1 G MeOH þ 0.1% TFA þ 0.1% DEA
3.080 (R*) 3.604 (S) 0.04 0.05 0.20 0.18 2.75 Cellulose 1 ISO 10% MeOH þ 0.1% TFA þ 0.1% DEA
Darifenacin 4.716 (R) 5.840 (S*) 0.03 0.05 0.41 0.44 2.46 Cellulose 1 ISO 25% 2-Pr-OH þ 0.1% TFA þ 0.1% DEA
2.637 (S*) 3.368 (R) 0.10 0.09 0.21 0.17 2.95 Amycoat ISO 25% 2-Pr-OH þ 0.1% TFA þ 0.1% DEA
3.149 (R) 5.030 (S*) 0.05 0.08 0.96 0.79 6.55 Amycoat ISO 35% 2-Pr-OH þ 0.1% TFA þ 0.5% DEA
Epinephrine 4.867 5.960 0.08 0.09 0.23 0.31 2.82 Cellulose 2 ISO 13% MeOH þ 0.1% TFA þ 0.1% DEA
Ezetimibe 3.577 (RSS*) 4.098 (SRR) 0.09 0.12 0.22 0.43 4.51 Amycoat G MeOH þ 0.1% TFA þ 0.1% DEA
2.261 (RSS*) 3.218 (SRR) 0.04 0.03 0.24 0.34 3.90 Amycoat ISO 23% MeOH þ 0.1% TFA þ 0.1% DEA
Fesoterodine 5.875 (R*) 6.792 (S) 0.08 0.07 0.42 0.46 2.27 Cellulose 1 ISO 15% 2-Pr-OH þ 0.1% TFA þ 0.1% DEA
Fesoterodine int 6.949 (R*) 8.206 (S) 0.07 0.11 0.27 0.45 2.13 Amycoat ISO 17% 2-Pr-OH þ 0.1% TFA þ 0.1% DEA
Indacaterol 3.459 5.010 0.12 0.18 0.49 0.57 2.86 Amycoat ISO 25% MeOH þ 0.1% TFA þ 0.5% DEA
Maraviroc 5.346 (S*) 6.405 (R) 0.07 0.07 0.37 0.37 2.69 Cellulose 1 ISO 15% MeOH þ 0.1% TFA þ 0.1% DEA
Maraviroc int ester 2.128 (R) 2.476 (S*) 0.05 0.03 0.37 0.29 5.82 Chiralpak IC3 G MeOH þ 0.1% TFA þ 0.1% DEA
1.916 (R) 2.035 (S*) 0.04 0.02 0.50 0.28 2.36 Cellulose 1 G MeOH þ 0.1% TFA þ 0.1% DEA
2.647 (R) 2.815 (S*) 0.03 0.02 0.59 0.53 1.45 Cellulose 2 G MeOH þ 0.1% DEA
2.506 (S*) 2.698 (R) 0.03 0.04 0.39 0.65 2.84 Cellulose 2 G MeOH þ 0.1% TFA þ 0.1% DEA
Tamsulosin 8.508 (R*) 10.094 (S) 0.06 0.10 0.34 0.53 2.33 Amycoat ISO 20% 2-Pr-OH þ 0.1% TFA þ 0.1% DEA
Tolterodine 11.109 (S) 12.365 (R) 0.31 0.29 0.56 0.85 1.25 Amycoat ISO 5% MeOH þ DEA 0.5%
Zolmitriptan 3.940 (S*) 4.765 (R) 0.03 0.03 0.33 0.52 2.49 Cellulose 2 ISO 27% MeOH þ 0.1% TFA þ 0.1% DEA
3.717 (R) 4.668 (S*) 0.02 0.04 0.10 0.10 2.51 Amycoat ISO 17% MeOH þ 0.1% TFA þ 0.1% DEA
Ambrisentan 2.615 (S,S*) 3.201 (R,R) 0.05 0.06 0.32 0.40 3.07 Cellulose 2 ISO 12% MeOH þ 0.1% TFA þ 0.1% DEA
Cetirizine 3.836 (S) 4.038 (R*) 0.01 0.01 0.10 0.31 2.22 Cellulose 1 G MeOH þ 0.1% TFA þ 0.1% DEA
4.132 (S) 4.762 (R*) 0.04 0.04 0.19 0.13 2.30 Cellulose 1 ISO 20% MeOH þ 0.1% TFA þ 0.1% DEA
Ibuprofen 5.460 (S*) 5.940 (R) 0.06 0.24 0.30 0.53 2.08 Chiralpak IC3 ISO 2.5% MeOH:ACN (1:1) þ 0.1% TFA þ 0.1% DEA
Ketoprofen 8.375 (S*) 9.109 (R) 0.12 0.11 0.45 0.43 2.25 Chiralpak IC3 ISO 8% MeOH:ACN (1:1) þ 0.1% TFA þ 0.1% DEA
204 kova
L. Nova , M. Dousa / Analytica Chimica Acta 950 (2017) 199e210
Cellulose 1 and Amycoat columns 5 compounds and for Cellulose 2 SFC was very limited also in this study. As a result, in the tested set
and Chiralpak IC3 columns 3 compounds provided complete of 20 analytes the change in temperature did not improve any
baseline separation with Rs 2 during the screening step. Some enantioseparation in the tested set of compounds, thus all the an-
examples are shown in Fig. 3 for boc-saxagliptin, cinacalcet and alyses were further performed at 40 C. Similarly, fine tuning of BPR
maraviroc intermediate ester showing two different gradient con- pressure (1500e2500 psi) and change in gradient slope did not lead
ditions providing successful enantioseparations. In total, 8 com- to any improvement of the enantioselectivity for any of tested
pounds could be completely resolved in the screening phase using enantiomeric pairs. The former conclusion is also in agreement
some of the 5 stationary phases, see Table 2. Among these 8 com- with the previously published report [44], which observed the in-
pounds 7 could be completely resolved on the two stationary fluence of the pressure on the enantioresolution to be up to 6%. It
phases proposed for the first-line screening. These compounds clearly indicates, that this parameter can't be successfully used as
belonged among less challenging compounds based on the an important parameter in method optimization.
graphical representation in Fig. 2, except for cetirizine. It is partic- Further testing included the use of partially successful gradient
ularly interesting to mention the cases, where the change of elution screening conditions (typically Amycoat or Cellulose 1 stationary
order of the two enantiomers can be observed between the two phase with CO2, both tested modifiers and combined TFA/DEA
stationary phases, such as in case of cinacalcet and maraviroc in- additive) in the isocratic mode. However, substantially lower
termediate ester. In enantiomeric purity control when the minor amount of organic modifier was applied compared to the gradient
enantiomeric impurity has to be analyzed in the presence of the elution. This way a complete baseline separation with the Rs 2 for
major enantiomer, it is more desirable to elute the minor compo- aprepitant intermediate, darifenacin, epinephrine, fesoterodine
nent before the major one. This usually allows for more sensitive and its intermediate, maraviroc, tamsulosin, zolmitriptan and
and reproducible quantification of the relevant enantiomeric im- ambrisentan were obtained. Detailed results are shown in Table 2
purity [39,40]. and selected chromatograms in Fig. 4A. Due to the success of this
The separations of enantiomeric pairs with Rs < 2 required approach, isocratic conditions were additionally applied also for
further method optimization. Fine tuning parameters, such as compounds, which were successfully separated already in the
temperature, BPR pressure, gradient slope and change in additive screening phase to show the differences and to see the potential for
concentration were considered on the stationary phases, which further possibility to decrease the analysis time and eliminate the
provided at least partial separation. The influence of the tempera- equilibration step in final QC method. Some examples are shown in
ture was tested in the range from 10 to 60 C. Increasing the tem- Fig. 4B. The same compounds were selected to show the difference
perature reduces the mobile phase density, which results in higher between gradient elution (Fig. 3) and isocratic elution. Indeed, very
retention factors. This implies that the resolution of chiral separa- fast separations, sometimes between 2 and 4 min were obtained for
tions may improve at lower temperatures. However, the tempera- some enantiomeric pairs, see Table 2. However, for some com-
ture effect is not always so straightforward, since it can also affect pounds, such as maraviroc intermediate ester, longer analysis time
the analyte affinity for the stationary phase hereby influencing the was needed in isocratic mode, thus the gradient method would be
enantioselectivity [41e43]. In agreement with some previous re- still the most convenient one also in the final QC method. Separa-
ports, the influence of the temperature on the enantioseparation in tions of more challenging compounds took typically 5e10 min after
Fig. 3. The chromatograms of the gradient SFC screening experiments for selected analytes e boc-saxagliptin, cinacalcet and maraviroc intermediate-ester and the structures of the
analytes. Right: Amycoat column - boc-saxagliptin and cinacalcet, Chiralpak IC3 - maraviroc intermediate. Left: Cellulose 1 for boc-saxagliptin and cinacalcet, Cellulose 2 for
maraviroc intermediate. The active isomers is depicted with asterisk. Detailed conditions are given in Table 2.
kova
L. Nova , M. Dousa / Analytica Chimica Acta 950 (2017) 199e210 205
Fig. 4. Chromatograms of SFC enantioseparations fine-tuned using isocratic elution: (A) maraviroc on Cellulose 1 using CO2/methanol and TFA/DEA combined additive, darifenacin
on Cellulose 1 and tamsulosin on Amycoat column using CO2/isopropanol and TFA/DEA combined additive. (B) boc-saxagliptin and cinacalcet on Cellulose 1 using CO2/methanol and
TFA/DEA combined additive and cinacalcet on Amycoat column using CO2/isopropanol and TFA/DEA combined additive. The active isomer is depicted with asterisk. Detailed
conditions are given in Table 2.
the fine method tuning (Table 2). approach has an important influence only for some analytes, while
Another option to tune the enantioselectivity is the change in in other cases the resolution is influenced only slightly as it is
the additive concentration. It is important to note, that this shown for zolmitriptan separation on Amycoat column in Fig. 5.
Fig. 5. Effect of the amount of organic modifier and the concentration of DEA on the enantioseparation of zolmitriptan on Amycoat column. (A) CO2/methanol with 0.1% TFA/DEA
combined additive using 25, 20 and 15% or organic modifier. (B) CO2/methanol with increased concentration of DEA (0.5%) in the combined TFA/DEA additive using 25, 20 and 15% of
organic modifier. The active isomer is depicted with asterisk.
206 kova
L. Nova , M. Dousa / Analytica Chimica Acta 950 (2017) 199e210
Fig. 6. Effect of the amount of organic modifier and the concentration of DEA on the enantioseparation of darifenacin on Amycoat column. (A) CO2/methanol and (C) CO2/iso-
propanol with 0.1% TFA/DEA combined additive showing the separation at 25 and 20% of organic modifier. (B) CO2/methanol and (D) CO2/isopropanol with increased concentration
of DEA (0.5%) in the combined TFA/DEA additive showing the separation at 25, 20 and 35% of organic modifier, respectively. The active isomer is depicted with asterisk.
Due to basic properties of this molecule combined TFA/DEA addi- The last three remaining compounds, tolterodine, ibuprofen and
tive in CO2/methanol at 0.1% concentration and the combined TFA/ ketoprofen belong among the most challenging compounds in this
DEA additive with increased concentration of DEA (0.5%) were set of analytes. Only in case of tolterodine, which was also the
compared. An increase in DEA concentration lead to substantially compound showing no separation in the screening phase, the strict
higher retention of both isomers and to slightly increased enan- requirement for Rs 2 was not met. Nevertheless, baseline sepa-
tioresolution (Fig. 5). However, easier way to manipulate the ration was finally obtained when using CO2/methanol with 0.5%
enantioresolution was changing the amount of the organic modi- DEA. This option was considered in the light of the above described
fier. Nevertheless, even slight change might sometimes be impor- experiments and a broadening not yet splitting peak shape
tant to meet the strict requirements of regulations. Similar observed on Amycoat column providing the clue that the two iso-
approach, i. e. increase in DEA concentration to 0.5% in a combined mers may separate. Ketoprofen and ibuprofen demonstrated partial
additive, was successfully applied for the analysis of indacaterol on separation when using CO2/methanol with TFA or combined TFA/
Amycoat column (data not shown) enabling to obtain complete DEA additive on Amycoat column and on Chiralpak IC3. This sep-
enantioresolution. Absolutely different behavior was observed in aration seemed to be quite straightforward. However, fine tuning
case of darifenacin separation on Amycoat column (Fig. 6). When through isocratic elution or change in additive concentration, both
increasing the concentration of DEA (0.5%) in the combined TFA/ using TFA and combined TFA/DEA additive, was not successful in
DEA additive in CO2/methanol, the enantioresolution was this case. Another option to change the selectivity is the use CO2/
completely lost (Fig. 6B), while it was around 1.6 at the concen- combined organic modifiers, such as mixtures of alcohols and
tration 0.1% of combined TFA/DEA additive (Fig. 6A). Even more acetonitrile, different alcohols or others. Both separations were
interesting behavior was observed when CO2/methanol was finally enabled using isocratic elution with CO2/methanol:acetoni-
replaced with CO2/isopropanol. The complete baseline enantior- trile (1:1) and the combined TFA/DEA additive. All results are
esolution with Rs 2 was obtained for both combined TFA/DEA shown in Table 2.
0.1% and 0.5% concentrations in CO2/isopropanol (Fig. 6C and D).
However, the elution order of the enantiomers reversed, which was 3.3. MS compatibility of the chiral SFC screening method
quite surprising considering quite a small concentration change.
This kind of behavior was not typical and was not found to be Coupling of separation techniques with mass spectrometry (MS)
equally effective in other cases. is important for the enhancement of specificity, sensitivity and
kova
L. Nova , M. Dousa / Analytica Chimica Acta 950 (2017) 199e210 207
Fig. 7. A comparison of enantioselectivity of CO2/methanol mobile phases with volatile additives ammonium formate (AmF), ammonium acetate (AmAc) and the combined TFA/
DEA additive on Cellulose 1 (A) and Amycoat (B) column. Successful separation (full enantioresolution or partial separations) is represented by the columns above the x-axis (þ),
while non-successful separation is represented by the columns bellow the x-axis ().
facilitation of compound identification. While it is already quite ammonium formate or ammonium acetate, in chiral SFC separa-
common in achiral SFC separations, there is limited information tions [45] and they have never been employed for chiral screening.
published so far on the use of MS friendly buffer additives, such as Therefore, MS compatible additives ammonium acetate and
Fig. 8. Changes observed in the enantioresolution when the combined 0.1% TFA/DEA additive was replaced with volatile additives in CO2/methanol mobile phases. Separation of
cinacalcet was performed on Cellulose 1 column using CO2/methanol with the addition (A) combined 0.1% TFA/DEA additive, (B) 20 mM ammonium formate and (C) 20 mM
ammonium acetate. Separation of maraviroc was performed on Amycoat column using CO2/methanol with the addition (D) combined TFA/DEA additive, (E) 20 mM ammonium
formate and (F) 20 mM ammonium acetate.
208 kova
L. Nova , M. Dousa / Analytica Chimica Acta 950 (2017) 199e210
Fig. 9. Effect of the injection solvent on the peak shape of atomoxetine (elution order S,R*) in gradient (A) and isocratic (B) elution using following injection solvents: heptane/
isopropanol (9:1, black, a reference in all presented chromatograms), acetonitrile (green), tetrahydrofuran (dark blue), methanol (red), ethanol (light blue) and isopropanol (violet).
Injection volume was 10 ml on the Cellulose 1 column using mobile phase composed of CO2 and methanol with the combined 0.1% TFA/DEA additive in the gradient mode (5e30% in
3 min) and in the isocratic mode (20% of organic modifier). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this
article.)
ammonium formate at a concentration 20 mM added to CO2/ mode. Tested solvents included methanol, isopropanol, ethanol,
methanol were further tested on Amycoat and Cellulose 1 sta- acetonitrile, tetrahydrofuran and a mixture of heptane/isopropanol
tionary phases to verify the suitability of these additives for the (9:1), which is recommended as the best solvent for SFC due to the
enantioseparation. The results in Fig. 7A demonstrate that the close properties to SFC mobile phase. However, this dilution solvent
change in enantioselectivity was very small on Cellulose 1 column. is non-polar and many pharmaceutical compounds are not well
Only the enantioresolution of cinacalcet (Fig. 8) and tamsulosine soluble in such solvent. Therefore, some compromise has to be
was lost using whatever of the two buffers, while maraviroc in- found. The choice of the solvent for the injection was found to be
termediate ester was not separated only when using ammonium more critical in the isocratic mode than in the gradient elution, as it
acetate. The additive comparison results were different on Amycoat is shown in Fig. 9 for atomoxetine. As expected, alcohol solvents
column (Fig. 7B). The loss of enantioresolution was found for compromised the peak shape with the worst tendency for meth-
darifenacin, epinephrine and indacaterol using both volatile anol. In the gradient elution a partial peak splitting was observed
buffers. On the other hand, on this stationary phase also some for the first eluted enantiomer, while only band broadening was
enantioresolution improvements were observed when replacing observed for the second enantiomer and for both enantiomers in
the combined TFA/DEA additive in CO2/methanol mobile phase the isocratic elution. Tetrahydrofuran and acetonitrile were found
with the volatile ones. This improvement was observed for cetir- to provide the best peak shapes in the gradient mode, where the
izine and maraviroc (Fig. 8) with both buffers. These experiments results were comparable to the mixture of heptane/isopropanol
reveal that volatile additives can also be successfully employed in (9:1), Fig. 9A. In isocratic elution some band broadening was
chiral screening, but generally at the cost of loss of enantionse- observed even when using these two solvents and this phenome-
lectivity for some compounds. As the results are quite encouraging non was more important for acetonitrile than for tetrahydrofuran
and to the best of our knowledge, no reports using volatile (Fig. 9B). When tetrahydrofuran was combined with alcohols, such
ammonium acetate and ammonium formate in SFC chiral screening as methanol or ethanol, some peak broadening was still observed in
have been published so far, further research is currently being both isocratic and gradient modes (data not shown). Based on these
performed on this topic in our laboratory. results, tetrahydrofuran was chosen as a good compromise in this
study and provided overall good peak shapes for all 20 tested
analytes.
3.4. Injection solvent for SFC chiral methods
[34] K. De Klerck, G. Parewyck, D. Mangelings, Y. Vander Heyden, Enantiose- J. Pharm. Biomed. Anal. 54 (2011) 687e693.
lectivity of polysaccharide-based chiral stationary phases in supercritical fluid [40] T.J. Wozniak, R.J. Bopp, E.C. Jensen, Chiral drugs: an industrial analytical
chromatography using methanol-containing carbon dioxide mobile phases, perspective, J. Pharm. Biomed. Anal. 9 (1991) 363e382.
J. Chromatogr. A 1269 (2012) 336e345. [41] P. Mourier, P. Sassiat, M. Caude, R. Rosset, Retention and selectivity in carbon
[35] D. Albas, Y. Vander Heyden, M.G. Schmid, B. Chankevadze, D. Mangelings, dioxide supercritical fluid chromatography with various stationary phases,
Chiral separations of cathinone and amphetamine-derivatives: comparative J. Chromatogr. A 353 (1986) 61e75.
study between capillary electrochromatography, supercritical fluid chroma- [42] U. Selditz, S. Copinga, C.J. Grol, J.P. Franke, R.A. de Zeeuw, H. Wikstrom,
tography and three liquid chromatographic modes, J. Pharm. Biomed. Anal. O. Gyllenhaal, Temperature effects on the chromatographic behaviour of
121 (2016) 232e243. racemic 2-amidotetralins on a Whelk-O 1 stationary phase in super- and
[36] M.M. Wong, W.B. Holzheuger, G.K. Webster, A comparison of HPLC and SFC subcritical fluid chromatography, Pharmazie 54 (1999) 183e191.
chiral method development screening approaches for compounds of phar- [43] M.J. Del Nozal, L. Toribio, J.L. Bernal, C. Alonso, J.J. Jimenez, Chiral separation of
maceutical interest, Curr. Pharm. Anal. 4 (2008) 101e105. omeprazole and several related benzimidazoles using supercritical fluid
[37] D.W. Armstrong, R.M. Woods, Z.S. Breitbach, Comparison of enantiomeric chromatography, J. Sep. Sci. 27 (2004) 1023e1029.
separations and screening protocols for chiral primary amines by SFC and [44] Ch Wang, Y. Zhang, Effects of column back pressure on supercritical fluid
HPLC, LC-GC N. Am. 32 (2014) 742e752. chromatography separations of enantiomers using binary mobile phases on
[38] C. West, G. Guenegou, Y. Zhang, L. Morin-Allory, Insights into chiral recog- 10 chiral stationary phases, J. Chromatogr. A 1281 (2013) 127e134.
nition mechanism in supercritical fluid chromatography. II. Factors contrib- [45] R. Pell, W. Lindner, Potential of chiral anion-exchange operated via various
uting to enantiomer separation on tris-(3,5-dimethylphenylcarbamte) of subcritical fluid chromatography modes for resolution of chiral acids,
amylose and cellulose stationary phases, J. Chromatogr. A 1218 (2011) J. Chromatogr. A 1245 (2012) 175e182.
2033e2057. [46] J.N. Fairchild, J.F. Hill, P.C. Iraneta, Influence of sample solvent composition for
[39] K.S. Dossou, P. Chiap, A.C. Servais, M. Fillet, J. Crommen, Development and SFC separations, LC-GC N.A. 31 (2013) 326e333.
validation of a LC method for the enantiomeric purity determination of S- [47] V. Abrahamsson, M. Sandahl, Impact of injection solvents on supercritical fluid
ropivacaine in a pharmaceutical formulation using a recently commercialized chromatography, J. Chromatogr. A 1306 (2013) 80e88.
cellulose-based chiral stationary phase and polar non-aqueous mobile phase,