Accepted Manuscript Title: Generic chiral method development in supercritical fluid chromatography and ultra-performance supercritical fluid chromatography Author: Katrijn De Klerck Yvan Vander Heyden Debby Mangelings PII: DOI: Reference:

S0021-9673(14)00892-9 http://dx.doi.org/doi:10.1016/j.chroma.2014.06.011 CHROMA 355487

To appear in:

Journal of Chromatography A

Received date: Revised date: Accepted date:

12-4-2014 2-6-2014 3-6-2014

Please cite this article as: K. De Klerck, Y.V. Heyden, D. Mangelings, Generic chiral method development in supercritical fluid chromatography and ultraperformance supercritical fluid chromatography, Journal of Chromatography A (2014), http://dx.doi.org/10.1016/j.chroma.2014.06.011 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

GENERIC CHIRAL METHOD DEVELOPMENT IN SUPERCRITICAL FLUID

2

CHROMATOGRAPHY AND ULTRA-PERFORMANCE SUPERCRITICAL

3

FLUID CHROMATOGRAPHY

an

us

cr

ip t

1

Katrijn De Klerck, Yvan Vander Heyden and Debby Mangelings*

5

Department of Analytical Chemistry and Pharmaceutical Technology, Center for

6

Pharmaceutical Research (CePhaR), Vrije Universiteit Brussel - VUB, Laarbeeklaan

7

103, B-1090 Brussels, Belgium

8

* Corresponding author:

9

Debby Mangelings

Ac ce p

te

d

M

4

10

Vrije Universiteit Brussel (VUB)

11

Center for Pharmaceutical Research (CePhaR)

12

Department of Analytical Chemistry and Pharmaceutical Technology (FABI)

13

Laarbeeklaan 103

14

B-1090 Brussels, Belgium

15

Tel: (+32) 2 477 43 29, Fax: (+32) 2 477 47 35

16

E-mail: [email protected] 1 Page 1 of 39

Abstract

18

The development of chiral separation methods in pharmaceutical industry is often a very

19

tedious, labour intensive and expensive process. A trial-and-error approach remains

20

frequently used, given the unpredictable nature of enantioselectivity. To speed-up this

21

process and to maximize the efficiency of method development, a generic chiral

22

separation strategy for SFC is proposed in this study. To define such strategy, the effect

23

of different chromatographic parameters on the enantioselectivity is investigated and

24

evaluated. Subsequently, optimization steps are defined to improve a chiral separation in

25

terms of resolution, analysis time, etc. or to induce separation when initially not obtained.

26

The defined strategy proved its applicability and efficiency with the successful separation

27

of a novel 20-compound test set. In a second stage, the method transfer from a

28

conventional to an ultra-performance SFC system is investigated for the screening step of

29

the separation strategy. The method transfer proved to be very easy and straightforward.

30

Similar enantioresolution values, but slightly shorter analysis times were obtained on the

31

ultra-performance equipment. Nevertheless, even more benefit may be expected in ultra-

32

performance SFC when customized sub-2µm chiral stationary phases will become

33

available.

34

Keywords

Ac ce p

te

d

M

an

us

cr

ip t

17

35

SFC method development

36

Chiral separation strategy

37

Polysaccharide-based stationary phases

38

Ultra-performance SFC

39

Method transfer

40

Highlights 2 Page 2 of 39

Investigation of the effect of different parameters in chiral SFC with polysaccharide-

42

based selectors

43

Definition of a chiral separation strategy for SFC

44

Evaluation of the strategy with a novel 20-compound test set

45

Method transfer from a conventional to an ultra-performance SFC system

Ac ce p

te

d

M

an

us

cr

ip t

41

3 Page 3 of 39

46 47

1. Introduction Over the past years, much attention has been paid to sub- and supercritical fluid

49

chromatography (SFC) in the context of chiral separations [1-4]. By exploiting the

50

benefits of sub- and supercritical fluids, fast and efficient enantioseparations can be

51

obtained in SFC. Simply returning to ambient conditions evaporates the primary eluent,

52

carbon dioxide (CO2), from the mobile phase after analysis. Hence, SFC can deliver a

53

significant reduction in waste generation and - disposal compared to conventional high-

54

performance liquid chromatography (HPLC) [5]. The higher flow rates, that can be

55

applied in SFC, allow higher productivities relative to HPLC, which is an important asset

56

in a pharmaceutical industrial environment to accelerate the drug development process

57

[6,7]. Given these properties, SFC has become a predominant technique for (preparative)

58

enantioresolutions [2,3,6-8].

M

an

us

cr

ip t

48

As for all separation techniques, chiral method development is also in SFC quite

60

labour intensive. Enantioselectivity remains unpredictable and the best way to achieve

61

appropriate separation conditions is by experimental trial-and-error. To make method

62

development more efficient and faster, generic separation strategies can be utilized [9-

63

12]. These strategies screen a chiral compound on a limited number of complementary

64

chromatographic systems (stationary + mobile phase combinations) in order to find the

65

most suitable system, showing the best enantioselectivity. Depending on the outcome of

66

this screening, optimization steps guide the user further to obtain the desired separation.

67

In case the desired separation could not be achieved, one is referred to screen in another

68

separation technique.

Ac ce p

te

d

59

69

A generic screening approach in SFC, that allows a fast selection of an appropriate

70

chromatographic chiral separation system for diverse chiral mixtures, was proposed

71

earlier [13]. Polysaccharide-based chiral stationary phases (CSPs) were used in this

72

screening because of their broad enantiorecognition capabilities and easy availabilities

73

[3,14]. However, after executing this screening, one might not have achieved the desired

74

separation yet. In that context, further method optimization steps can be defined. These

75

aim to optimize resolution, selectivity, analysis time, and in relevant cases also the peak

76

shape. A first part of this paper focuses on the influence of different parameters on a 4 Page 4 of 39

77

chiral SFC separation. Based on this information, appropriate optimization steps are

78

derived to complete the entire generic separation strategy. To evaluate the performance of

79

this strategy, a novel 20-racemates set is tested. To catch up with the state-of-the-art technology found in the field of HPLC, SFC

81

equipment is becoming better adapted, more robust and more reliable to achieve

82

chromatographic separations with acceptable repeatability and reproducibility. In

83

particular, the mobile phase density can be controlled much stricter, which is a crucial

84

aspect in SFC since the density has a direct impact on the mobile-phase strength.

85

Following the trend in HPLC, SFC is undergoing an evolution to ultra-high performance

86

SFC (UHP-SFC) [15,16]. With minimal void volumes and maximal sensitivity, fast

87

separations can be achieved with high efficiencies. Because certain parameters are

88

different between the different systems, (enantio)separations might be impacted when

89

transferred. A second part of this research therefore focuses on the method transfer from

90

conventional SFC to UHP-SFC.

91

2. Experimental

92

2.1 Chromatographic equipment

d

M

an

us

cr

ip t

80

The analytical SFC method station from Thar® (Pittsburgh, PA, USA, a Waters®

94

company) equipped with a Waters® 2998-DAD detector (Milford, MA, USA) was used

95

for the first part of the experiments (definition of the separation strategy). The

96

autosampler was equipped with a 10μl loop. For all analyses partial loop injections of 5μl

97

were done. Data acquisition and processing were performed using Chromscope® V1.10

98

software (2011) from Waters®.

Ac ce p

te

93

99

For the strategy evaluation and method transfer to UHP-SFC, an Acquity

100

UltraPerformance Convergence Chromatography (UPC²) from Waters® was used. The

101

system was equipped with a binary solvent manager, a sample manager with a fixed loop

102

of 10μl, a convergence manager, an external Acquity column oven and a PDA detector.

103

For all analyses partial loop injections of 5μl were done. Empower® 3 V7.10 software

104

(2010, Waters®, Milford, MA, USA) was used for data acquisition and processing.

105 106

The chromatographic conditions were different for the analyses performed during the optimization process. For this reason they are specified further. 5 Page 5 of 39

107

2.2 Materials The columns Chiralpak® AD-H and Chiralcel® OD-H, OJ-H and OZ-H were

109

purchased from Chiral Technologies (West Chester, PA, USA). Lux® Cellulose-1, -2,

110

and -4 were purchased from Phenomenex (Utrecht, The Netherlands). To allow a fair

111

comparison, all columns had dimensions of 250 mm x 4.6 mm i.d. with 5 μm particle

112

size.

113

2.3 Chemicals

cr

ip t

108

Methanol (MeOH), ethanol (EtOH) and 2-propanol (2PrOH) were HPLC grade and

115

purchased from Fisher Chemicals (Loughborough, UK). Isopropylamine (IPA) and

116

trifluoroacetic acid (TFA) were from Aldrich (Steinheim, Germany). CO2, was used as

117

advised by the manufacturers of the individual SFC instruments. For the Thar®

118

equipment this was quality 2.7 (purity

119

for the UPC²® equipment quality 4.5 (purity

120

Leeuw, Belgium).

an

99.7%) from Linde Gas (Grimbergen, Belgium);

M

99.995%) from Messer (Sint-Pieters-

122

percentages.

123

2.4 Chiral test set

d

All percentages expressed in the context of mobile-phase composition are volume

te

121

us

114

For the definition of the optimization steps and separation strategy, a generic chiral

125

test set of 56 pharmaceuticals was used. Test solutions of these 56 racemates with a

126

concentration of 0.5 mg/ml were made in methanol. The solutions were kept at 4°C when

127

not used. The test set was composed of racemates with diverse structural, chemical, and

128

pharmacological properties. Because it was used in earlier research, we refer to these

129

papers for detailed information [17,18]. To evaluate the proposed separation strategy, a

130

novel test set composed of 20 pharmaceutical racemates is used (Table 1). These

131

racemates were also dissolved in MeOH at a concentration of 0.5mg/ml and kept at 4°C.

132

2.5 Data processing

Ac ce p

124

133

For all enantioseparations, the resolution (Rs) is calculated using the European

134

Pharmacopoeia equations applying peak widths at half heights [19]. Separations obtained

135

with a resolution higher than 1.5 are considered as baseline separated. When the 6 Page 6 of 39

resolution is between 0 and 1.5 the separations are designated as partial. The selectivity

137

(α) is calculated as the ratio of the retention factors of the last and first eluting

138

enantiomers of a pair [19]. The void time was marked as the first disturbance of the

139

baseline after injection of solvent. The retention time of the last eluting peak is taken as

140

the analysis time.

141

ip t

136

Microsoft® Excel (Microsoft® Corporation, 2010) was used for constructing the plots and graphs and for the statistical interpretation of the data (Student t-test and ANOVA).

143

3. Results and discussion

144

3.1 Screening step

us

cr

142

A generic chiral screening approach was derived from the evaluation of 12

146

polysaccharide-based chiral stationary phases in combination with eight mobile phases

147

(MP) (total of 96 chromatographic systems). The performance in terms of successful

148

enantioseparations, and the complementarity of the latter systems were taken into

149

account, to define a screening sequence (Fig. 1) [13]. The screening entails four

150

experiments, evaluating four complementary polysaccharide-based stationary phases.

151

This approach allowed the separation of all compounds from the 56-compound test set.

152

However, not every separation is optimal, e.g. Rs < 1.5 (partial separations) or excessive

153

analysis time can be obtained. In these cases further optimization imposes itself in order

154

to obtain the desired enantioseparation. Because a number of factors influence

155

enantioseparation in SFC, e.g. organic modifier, flow rate, pressure, temperature, etc., the

156

optimization is not always evident. In a first part of this work, attention will be paid to

157

these factors impacting enantioseparation. The obtained information will be used to

158

define specific optimization steps in the context of a generic chiral separation strategy.

159

3.2 Factors influencing enantioseparations in SFC

Ac ce p

te

d

M

an

145

160

Organic modifier type

161

In most cases, pure CO2 is not adequate to elute (pharmaceutical) compounds. Most

162

pharmaceutical compounds possess a structure with hydrophobic, hydrogen-bonding

163

donor and -acceptor sites. This requires the addition of an organic modifier to the mobile

164

phase to increase the solvent strength, allowing elution and analysis of these relatively

165

polar compounds [4,5]. 7 Page 7 of 39

It is well-known that the organic modifier type in the mobile phase alters the

167

enantioselectivity of a CSP towards certain racemates. The lipophilicity, polarity,

168

basicity, i.e. the properties of the organic modifier affect the interactions between the

169

solute and stationary phase [20]. Consequently, by changing the organic solvent in the

170

mobile phase, different enantioseparations can be achieved on the same CSP. In chiral

171

SFC, methanol, 2-propanol and ethanol are most often used as modifiers [6,12,21-24]. In

172

our experience, MeOH is slightly more successful on the polysaccharide-based CSPs,

173

followed by 2PrOH and EtOH. MeOH offers the additional advantage that its boiling

174

point is lower than that of 2PrOH and EtOH, making solvent evaporation after analysis

175

easier. The viscosity of MeOH is also lower and its use poses thus less stress on the

176

CSPs.

us

cr

ip t

166

In earlier research, 12 polysaccharide-based chiral stationary phases were evaluated

178

with eight MeOH- or2PrOH-containing mobile phases [13]. On eight of these twelve

179

CSPs, a MeOH-containing mobile phase provided the highest success rate. For this

180

reason we slightly favour MeOH over 2PrOH.

M

an

177

As far as enantioselectivity is concerned, it is impossible to predict which solvent will

182

provide the most favourable separation conditions for a given racemate. Earlier we

183

selected four successful and complementary chromatographic systems, using a generic

184

compound test set. We included these systems in a screening approach [13]. For most

185

compounds, executing this screening should deliver appropriate selectivity to achieve the

186

desired enantioseparation. We were able to separate (baseline or partially) the entire 56-

187

compound test set using MeOH in combination with OZ-H (or LC-2) and OD-H (or LC-

188

1); and using 2PrOH with AD-H and LC-4.

Ac ce p

te

d

181

189

However, in case no (satisfying) separation is obtained after this screening, it is

190

advisable to screen the same CSPs with the alternative modifier, i.e. 2PrOH (for OZ-

191

H/LC-2 and OD-H/LC-1) or MeOH (for AD-H and LC-4), since this broadens the

192

enantioselective range. Different enantioselectivities, were observed when considering

193

both modifiers. In most cases MeOH yields more separations (Table 2). AD-H seems an

194

exception to this trend, since 2PrOH is much more successful than MeOH on this CSP.

195

Nevertheless, on each CSP, a number of unique separations is provided by both

8 Page 8 of 39

196

modifiers. This explains the second step in our screening strategy, which proposes to

197

screen the selected CSPs with an alternative modifier. We also noticed a unique enantioselectivity of some stationary phases in combination

199

with EtOH. In case no enantioselectivity is obtained after screening with MeOH or

200

2PrOH, EtOH can therefore be tested as alternative modifier. However, given the lower

201

general success rate of EtOH, it would be less advisable to include this modifier in a first

202

screening attempt.

cr

ip t

198

Concentration of the organic modifier

204

In low concentrations (< 2-5%), the organic modifier competes with the analytes for

205

interaction with residual silanol groups on the stationary phase. By surrounding the active

206

silanol sites, the stationary phase becomes more uniform in terms of polarity and

207

consequently peak shapes become more symmetrical. Hence, in the lower concentration

208

range, an increase in modifier content is advantageous for the resolution of the

209

separations. Once all silanol sites are covered by modifier molecules, a further increase in

210

modifier concentration negatively influences the resolution by impacting the solvent

211

strength of the mobile phase [25]. The separation efficiency also tends to deteriorate,

212

since the analyte diffusion through the column is inhibited by the increasing mobile phase

213

viscosity [26].

te

d

M

an

us

203

These trends are clearly seen in the separation of clopidogrel on Chiralpak® AD-H

215

(Fig. 2). When the modifier content is increased from 5 to 10% the resolution increases

216

from 3.8 to 4.6. Increasing the modifier content above 10%, decreases the resolution. On

217

the other hand, the analysis time is impacted by the mobile phase strength. A decrease

218

from 7.98 to 2.75 min occurs when the modifier increases from 5 to 20%. Further

219

increasing the modifier in the mobile phase to 40% decreases the analysis time to

220

1.72 min. However, the relation between the analysis time and modifier content is not

221

linear and the observed decrease in analysis time is higher in the lower concentration

222

range (5-20%).

Ac ce p

214

223

Conclusively it can be stated that the most appropriate modifier content in the mobile

224

should be a compromise between analysis time and resolution. In our strategy we propose

225

to increase the modifier content when shorter analysis times are desired. If higher

9 Page 9 of 39

resolutions are desired we advise the opposite. As a compromise 20% modifier is used in

227

the screening.

228

Flow rate

229

Supercritical fluid chromatography is suitable for fast analyses. Because the sub- or

230

supercritical mobile phase has a low viscosity and high diffusivity, higher flow rates can

231

be used compared to HPLC. Flow rates up to 5.0 ml per min are no exception in

232

analytical SFC. Increasing the flow rate will fasten an analysis significantly, without

233

compromising the separation efficiency too drastically.

cr

ip t

226

For example, when the flow rate for the enantioseparation of cetirizine is increased

235

from 1 to 6 ml/min, the analysis time reduces with 84% (from 16.8 to 2.6 min), the Rs

236

decreases less than 50% (from 12.42 to 6.41), while α remains almost unchanged (Fig. 3).

237

The separation at 6 ml/min is still largely acceptable, and requires 14 min less than that at

238

flow rate 1 ml/min. Increasing the flow rate above 6 ml/min was not possible due to

239

pressure limitations of the CSP.

M

an

us

234

Above an example is shown which actually is valid for all chiral SFC-separations.

241

This is explained by the flatter profile of the Van Deemter curve in SFC compared to

242

HPLC, allowing analyses at higher mobile phase velocities without a substantial loss in

243

efficiency [5]. Hence, when optimizing analysis times in SFC, it is advisable to increase

244

the flow rate, since the impact on the resolution remains rather limited. The limiting

245

factors in this approach are the pressure restrictions imposed by the equipment and the

246

chromatographic column.

Ac ce p

te

d

240

247

Back pressure

248

To guarantee a constant mobile-phase density, a back-pressure regulator is employed

249

in SFC controlling the pressure. The mobile-phase density has a direct impact on the

250

mobile-phase strength, thus on the (enantio)selectivity and retention. A higher back

251

pressure means a higher mobile-phase density, and –strength, and shorter retention times.

252

As a consequence, the selectivity might also decrease.

253

However, when exploring a pressure range in the search for optimal separation

254

conditions, a user is restricted by the limitations of the polysaccharide-based column and

255

the equipment. In practice, back pressures between 125 and 250 bar are commonly used 10 Page 10 of 39

256

for chiral SFC separations. Using lower pressures harms the chromatographic results

257

significantly since the sub-critical state of the mobile phase is no longer guaranteed [27]. In this pressure range (125-250 bar), the actual impact of the back pressure on the

259

retention and selectivity is rather limited and considerably lower than that of the organic

260

modifier content. In other words, when a large change in retention or selectivity is

261

desired, the first step should be to adopt the modifier content in the MP. When fine-

262

tuning a separation, the back pressure can be changed. For shorter retention/analysis

263

times the back pressure should be increased, while decreasing is advisable when the

264

selectivity should be improved.

us

cr

ip t

258

For the separation of econazole, a doubling of the back pressure from 125 to 250 bar

266

decreases the retention of the last eluting peak from 8.5 to 6.4 min (Fig. 4). As a

267

consequence, the partial resolution is lost when the back pressure is elevated above

268

200 bar.

an

265

For screening purposes, it is proposed to set the back pressure at 150 bar as a

270

compromise between retention time and enantioselectivity. Consequently, reducing the

271

back pressure to the lower limit of 125 bar would only result in a minimal gain in

272

enantioselectivity. Therefore this step is not included in the partial separation branch of

273

the strategy (see further). On the other hand, to speed up the analysis, it is more effective

274

to increase the flow rate and/or modifier content than the back pressure. Therefore, an

275

increase in back pressure is only recommended as a third choice to reduce the analysis

276

time of baseline separations (see further).

Ac ce p

te

d

M

269

277

Temperature

278

Temperature also influences the mobile-phase density. An increase results in a

279

decrease of the mobile-phase density and has the above-mentioned consequences. It is

280

important to realize that by reducing the temperature, the chromatographic conditions

281

deviate further from the super- into the subcritical region. This does not create practical

282

issues until the subcritical state turns into a two-phase state, which would deteriorate the

283

chromatographic results significantly and prevents proper analyses. The vapour-liquid

284

curve of the pressure-temperature phase diagram separates the two-phase region from the

285

subcritical region. For (chiral) SFC separations it is thus important to remain above that

286

vapour-liquid curve, but there are no further restrictions to the chosen conditions. SFC 11 Page 11 of 39

287

separations can thus also be performed below 31°C, i.e. the critical temperature of pure

288

carbon dioxide [27]. For polysaccharide-based columns, the temperature range is limited from 5 to 40-

290

50°C, varying by column-manufacturer info. The actual impact of the temperature on the

291

retention and selectivity in this workable range is rather limited. When the temperature is

292

increased from 10 to 45°C (a 350% increase), the retention of the last eluting peak of

293

carprofen only decreases from 2.70 to 2.56 min (a decrease of 5%) (Fig. 5). The

294

resolution and selectivity of the separation are hardly affected by this temperature

295

change.

us

cr

ip t

289

Summarized, it can be stated that although the temperature has an important impact

297

on SFC separations, the workable temperature range with polysaccharide-CSPs is too

298

limited to have a significant gain in analysis time or selectivity. For this reason, a

299

temperature optimization is not included in the final separation strategy (see further). The

300

temperature was therefore set at 30°C for all experiments, based on the study of Maftouh

301

et al. [12].

M

an

296

Additives

303

In the screening, 0.1% isopropylamine (IPA) and 0.1% trifluoroacetic acid (TFA) are

304

added to the modifier, of which only 20% in used in the mobile phase. Hence, the final

305

concentration in the MP is 0.02% IPA and TFA. Nevertheless, their addition, even in

306

these low concentrations, affects the interactions between the analytes and the stationary

307

phase. Without the presence of additives in the MP, chromatographic results tend to

308

deteriorate significantly. IPA and other basic amine-additives shield silanol sites on the

309

stationary phase, decreasing the non-specific retention of analytes. They also compete

310

with the basic functional groups of analytes for interactions with specific sites on the

311

stationary phase. These additives also neutralize charged groups of basic analytes, which

312

is important for the interactions with neutral chiral selectors, such as polysaccharide-

313

derivatives [18,28]. Acidic additives, such as TFA, suppress the ionization of acidic

314

analytes.

Ac ce p

te

d

302

315

For polysaccharide-based chiral columns, these effects do not seem directly related

316

to the concentration of the additives in the MP [29]. We investigated different additive

317

concentrations in a range from 0.1 to 0.25% and saw only a minor impact on the retention 12 Page 12 of 39

or resolution. Peak shapes tend to be slightly sharper with increasing additive

319

concentrations. On the other hand, adding less than 0.1% to the modifier was not

320

sufficient to induce the desired effect; peak shapes and chromatographic results were

321

unacceptable. Hence, in the screening, the additive concentration is set at 0.1% IPA and

322

TFA in the modifier.

ip t

318

Earlier, we observed a significant difference in enantioselectivity between the

324

simultaneous use of IPA and TFA and the individual use of these additives [18]. In the

325

latter case, TFA is used for acidic compounds and IPA for all other compounds. Since the

326

success rate tended to be higher when combining the additives, we advise using this

327

approach in a screening stage [18]. Moreover, the benefit is that the screening conditions

328

are the same for all compounds, independent of their chemical properties. However, in

329

case the desired enantioseparation is not achieved, it can be useful to try only one single

330

additive in the modifier. This is therefore recommended in the partial separation branch

331

of the strategy (see further).

332

3.3 Separation strategy

M

an

us

cr

323

Based on the above information and earlier experience, a separation strategy was

334

defined (Fig. 6). This strategy was evaluated with a novel test set of 20 pharmaceutical

335

racemates (Table 1). After executing the screening experiments, 18/20 compounds were

336

separated. After applying the entire strategy, all compounds were baseline separated, with

337

the exception of fluoxetine, which was partially separated (Rs = 1.3) (Table 3).

339

te

Ac ce p

338

d

333

Analysis time for these optimized separations was in 16/20 cases below 5 min, for 19/20 below 10 min and for fluoxetine 15 min.

340

The separation strategy applied on two racemates, i.e. thioridazine and clopidogrel is

341

presented in figures 7 and 8, respectively. The chromatograms (a-d) clearly show the

342

complementarity of the chromatographic systems included in the screening step. After the

343

optimization steps, good baseline separations with satisfying peak shapes and short

344

analysis times are obtained.

345

3.4 Method transfer from conventional SFC to UHP-SFC

346

The screening conditions from the separation strategy were transferred from a

347

conventional SFC to an ultra-performance (UPC²) SFC equipment. To evaluate the 13 Page 13 of 39

transfer, the 56-compound test set used as for the definition of the screening was applied.

349

We refer to these earlier papers for more information on its composition [13,17,18]. The

350

four chromatographic systems from the screening were evaluated, i.e. OZ-H and OD-H,

351

with 20% (MeOH:IPA:TFA, 100:0.1:0.1, v:v:v), and AD-H and LC-4, with 20%

352

(2PrOH:IPA:TFA, 100:0.1:0.1, v:v:v) in the MP. The same columns and conditions were

353

used on both instruments.

ip t

348

Generally the method transfer from conventional to ultra-performance SFC seems

355

rather straightforward. Usually similar separation results are achieved when applying the

356

same chromatographic conditions in conventional and ultra-performance SFC (Fig. 9).

us

cr

354

However, the success rates on all chromatographic systems obtained with the ultra-

358

performance system are slightly lower (Fig. 10). In this context, it is important to analyse

359

the results further since the difference in success rate may originate from small

360

differences in resolution. A partial separation is any separation with a resolution higher

361

than zero, while baseline separations have Rs > 1.5. Hence, in case a separation with

362

resolution 0.2 is obtained on one instrument, a small decrease in Rs on the other may

363

result in a loss of the separation.

d

M

an

357

We thus compared the resolutions and analysis times (AT) of the 56 compounds. The

365

obtained Rs and analysis times are similar but tend to be slightly lower on the UPC² than

366

on the conventional equipment (Fig. 11). These lower resolutions are reflected in the

367

lower success rates on the UPC². To assess the significance of the difference in Rs and

368

AT between both instruments, a two-tailed paired student t-test was performed.

Ac ce p

te

364

369

Table 4 summarizes the results in terms of the calculated t- and p-values. For two

370

chromatographic systems, i.e. AD-H and LC-4 with 2-propanol in the mobile phase, the

371

resolutions were not significantly different on the conventional Thar SFC and UPC². For

372

OD-H and OZ-H with methanol, the difference was determined to be significant.

373

The analysis times were slightly lower on the UPC² than on the conventional Thar

374

instrument. This difference was determined to be significant for all chromatographic

375

systems, with the exception of OD-H with methanol in the mobile phase.

376

Hence, to conclude it can be stated that, in general, the analysis times on the UPC² are

377

shorter than on the conventional SFC system. This can be related to the minimization of

378

the void volume in this equipment, resulting in a lower void time. However, this is not 14 Page 14 of 39

379

translated into separations with higher resolutions. In most cases, the resolutions were

380

slightly lower on the ultra-performance SFC system, on two of the four systems, this

381

decrease was significant. Thus, the resolutions are rather comparable between the two

382

systems, while a gain in analysis time is obtained with the ultra-performance system. However, the maximal potential of the UHP-SFC system might not be achieved with

384

the 5μm particle columns used in this study. Reducing the particle size to sub-2μm

385

dimensions, would possibly increase the separation efficiency significantly [15,16]. So

386

far, no sub-3μm chiral polysaccharide-based stationary phases are commercially

387

available. The coating of the polysaccharide-based selector on the silica and the uniform

388

and reproducible packing of these smaller particles appears to be very tedious. Hence,

389

more potential lies in UHP-SFC for chiral separations provided that adapted CSP become

390

available.

cr

us

an

391

ip t

383

For this study, where the same columns were used, the method transfer from the conventional to the ultra-performance system was very easy and straightforward.

393

3.5 Precision study: conventional SFC vs UPC²

M

392

To evaluate the precision of experiments on both systems, six enantioseparations;

395

bopindolol, mepindolol, methadone, mianserine, naringenin, and sotalol, were selected

396

and repeated twice over six consecutive days. The same chromatographic conditions were

397

used on both systems. Lux Cellulose-2 was used as stationary phase, with 20%

398

MeOH:IPA:TFA (100:0.1:0.1) in the mobile phase. The total flow rate was 3.0 ml/min,

399

the temperature 30°C and back pressure 150 bar. Detection was done at 220 nm. The

400

sample loop was 10μl and partial injections of 5μl were done for each sample.

Ac ce p

te

d

394

401

The inter- and intra-day variabilities and the intermediate precision (expressed in

402

variance) were estimated for each separation using ANOVA. Table 5 shows the results

403

for all separations on both systems. Two responses were considered: the resolution and

404

the analysis time (AT). The variances obtained with both systems were compared with an

405

F-test.

406

The intra-day variance on the Rs was not significantly different between the UPC²

407

and Thar for three compounds. For methadone and sotalol the variance was smaller on

408

the Thar than on the UPC², for mepindolol the opposite was seen. The inter-day 15 Page 15 of 39

variability was not significantly different for three compounds, while for mepindolol,

410

mianserine, and sotalol it was lower on the UPC². The intermediate precision was

411

significantly different for two separations: the variance for mepindolol was lower on the

412

UPC² and for methadone on the Thar system. These results indicate that there is no

413

distinct benefit of one system over the other concerning the repeatability of experiments

414

when considering the resolution as response.

ip t

409

Next, we considered the analysis time as response. Three separations yielded a

416

significantly different intra-day variability, i.e. bopindolol, methadone and naringenin.

417

The first two separations showed a lower variability on the Thar system, while the

418

opposite situation was seen for the last. The inter-day variability was significantly lower

419

on the Thar system for bopindolol and methadone and on the UPC² for mianserine,

420

sotalol and naringenin. The intermediate precision on the AT, was lower for bopindolol

421

and methadone on the Thar system and for naringenin on the UPC². Hence, the intra- and

422

inter-day variability and intermediate precision of the analysis times between both

423

systems are comparable, and no distinct advantage of one system over the other was seen.

424

Conclusively, these experiments showed that in terms of precision the performance of both systems were similar.

426

4. Conclusions

te

425

d

M

an

us

cr

415

To define a generic separation strategy, the impact of different parameters on chiral

428

SFC separations was investigated. The influence of organic modifier type and

429

- concentration, flow rate, back pressure, temperature and additives, were considered.

Ac ce p

427

430

When dissimilar enantioselectivity is sought, it is advisable to screen different

431

modifiers in the mobile phase. Methanol was favoured over 2-propanol and ethanol, since

432

this modifier tended to generate higher success rates on the polysaccharide-based CSPs,

433

although a broad complementarity exists between MeOH and 2PrOH. To extend the

434

enantioselective recognition, it is thus advisable to screen a CSP with both modifiers.

435

When higher resolutions are desired, the modifier concentration can be decreased.

436

When aiming to decrease the analysis time, the flow rate can be increased without

437

compromising the efficiency much. The back pressure and temperature only exert minor

438

influences on the resolution or analysis time of chiral SFC separations on polysaccharide-

439

derivatives. The latter information was used to define a separation strategy, which 16 Page 16 of 39

440

applicability was evaluated with a novel test set of 20 pharmaceutical racemates. All

441

racemates could be baseline separated, with the exception of fluoxetine. Analysis times

442

where below 10 min for all separated compounds. The developed approach was transferred from a conventional to an ultra-performance

444

SFC system. Similar separation results in terms of Rs were generated by both systems,

445

while the analysis times where slightly lower on the ultra-performance system. The

446

method transfer thus proved to be very easy and straightforward.

cr

ip t

443

A precision study was performed for six separations on the Thar and UPC² system.

448

Results showed no distinct advantage of one system over the other concerning the intra-,

449

and inter-day variabilities or the intermediate precision of the resolution and analysis time

450

of the separations.

an

us

447

More efficient separations could potentially be achieved using sub-2μm columns.

452

However, so far, no CSPs are commercially available with these particle dimensions.

453

Undoubtedly, there still remains a whole unexplored domain in this context for chiral

454

separations.

455

Acknowledgements

456

This work was financially supported by the Research Foundation Flanders FWO (projects

457

1.5.114.10N/1.5.093.09N.00).

458

The authors declared no conflict of interest.

Ac ce p

te

d

M

451

17 Page 17 of 39

List of figures

460 461 462 463

Figure 1: Scheme of the screening step as defined in [13]. In the top row the chiral stationary phases are presented, while the second row represents the used modifier concentration in the carbon-dioxide based mobile phase.

464 465 466 467

Figure 2: Separation results of clopidogrel on Chiralpak® AD-H with (2PrOH+ 0.1%TFA + 0.1%IPA) in the mobile phase in varying concentrations. a) obtained resolutions and b) the total analysis time in function of the percentage modifier content.

468 469 470 471 472

Figure 3: Results of the enantioseparation of cetirizine on Chiralpak AD-H with 20% (2PrOH + 0.1%IPA + 0.1%TFA) in the mobile phase as a function of the flow rate. (a) Overlay of the obtained chromatograms; (b) analysis time; (c) resolution and selectivity as a function of the flow rate.

473 474 475 476 477

Figure 4: Overlay of the chromatograms of econazole on Chiralcel® OZ-H with 20% (MeOH:IPA:TFA, 100:0.1:0.1, v/v/v) in the mobile phase. A flow rate of 3 ml/min and temperature of 30°C was used. The backpressures were 1) 125bar; 2) 150bar; 3) 175bar; 4) 200bar; 5) 225bar; and 6) 250bar. (Results generated with the UPC² system).

478 479 480 481 482

Figure 5: Chromatograms of the enantioseparation of carprofen on Chiralcel® OZ-H with 20% (2PrOH:IPA:TFA, 100:0.1:0.1, v/v/v) in the mobile phase. A total flow rate of 4 ml/min and back pressure of 150 bar was used. The temperatures were a) 10°C; b) 15°C; c) 20°C; d) 25°C; e) 30°C; f) 35°C; g) 40°C; and h) 45°C.

483 484

Figure 6: Chiral separation strategy for polysaccharide-based columns in SFC

485 486 487

Figure 7: Separation strategy applied on the racemate thioridazine. Chromatograms a-d: experiments from the screening step, e: optimized conditions.

488 489 490

Figure 8: Separation strategy applied on clopidogrel racemate. Chromatograms a-d are the results from the screening step, e is the result after optimization.

491 492 493 494 495 496

Figure 9: Transfer of the chromatographic conditions from conventional SFC (Thar equipment) to ultraperformance SFC (UPC² equipment). The separations are obtained on Lux Cellulose2, with 20% (MeOH:IPA:TFA, 100:0.1:0.1, v:v:v) in the mobile phase, flow rate 3.0 ml/min, 30°C, detection at 220nm, and a back pressure of 150 bar. a) mepindolol, b) naringenin, c) mianserine.

Ac ce p

te

d

M

an

us

cr

ip t

459

18 Page 18 of 39

Figure 10: Number of baseline (Rs > 1.5) and partial (0 < Rs < 1.5) separations achieved with the Thar SFC and UPC² systems on the four chromatographic systems of the screening.

500 501 502 503

Figure 11: Comparison of the screening results of the 56-compound test set on the ultra-performance and conventional SFC equipment. a) resolutions (Rs), b) analysis times. Straight line = line of equality.

Ac ce p

te

d

M

an

us

cr

ip t

497 498 499

19 Page 19 of 39

[1] G. Terfloth, J Chromatogr A, 906 (2001) 301.

507

[2] G. Guiochon and A. Tarafder, J Chromatogr A, 1218 (2011) 1037.

508

[3] B. Chankvetadze, J Chromatogr A, 1269 (2012) 26.

509

[4] K.W. Phinney, Anal Chem, 72 (2000) 204A.

510

[5] Z. Wang, in S. Ahuja (Editor), Chiral separation methods for pharmaceutical and

cr us

biotechnological products, Wiley, Calabash, North Carolina, US, 2011, p. 299.

an

511

Reference List

ip t

504 505 506

[6] L. Miller, J Chromatogr A, 1250 (2012) 250.

513

[7] L. Miller and M. Potter, J Chromatogr B Analyt Technol Biomed Life Sci, 875

d

(2008) 230.

te

514

M

512

[8] E.R. Francotte, J Chromatogr A, 906 (2001) 379.

516

[9] H. Ates, D. Mangelings and Y. Vander Heyden, in S. Ahuja (Editor), Chiral

517 518 519 520

Ac ce p

515

separation methods for pharmaceutical and biotechnological products, Wiley, Calabash, North Carolina, (US), 2011, p. 383.

[10] H. Ates, D. Mangelings and Y. Vander Heyden, J Pharm Biomed Anal, 48 (2008) 288.

521

[11] D. Mangelings and Y. Vander Heyden, Adv. Chromatogr, 46 (2008) 175.

522

[12] M. Maftouh, C. Granier-Loyaux, E. Chavana, J. Marini, A. Pradines, Y.Vander

523

Heyden and C. Picard, J Chromatogr A, 1088 (2005) 67. 20 Page 20 of 39

525 526

[13] K. De Klerck, D. Mangelings and Y. Vander Heyden, J. of Supercritical Fluids, 80 (2013) 50. [14] T. Ikai and Y. Okamoto, in A. Berthod (Editor), Chiral recognition in separation

ip t

524

methods - mechanisms and applications, Springer-Verlag, Berlin-Heidelberg

528

(Germany), 2010, p. 33.

cr

527

[15] T.A. Berger, Chromatographia, 72 (2010) 597.

530

[16] C. Sarazin, D. Thiebaut, P. Sassiat and J. Vial, J Sep. Sci, 34 (2011) 2773.

531

[17] K. De Klerck, G. Parewyck, D. Mangelings and Y. Vander Heyden, J Chromatogr

535 536

an

M

d

534

[18] K. De Klerck, D. Mangelings, D. Clicq, F. De Boever and Y. Vander Heyden, J Chromatogr A, 1234 (2012) 72.

te

533

A, 1269 (2012) 336.

[19] The European Pharmacopoeia, Directorate for The Quality of Medicines of the

Ac ce p

532

us

529

Council of Europe (EDQM), Cedex (France), 2005.

537

[20] G.O. Cantrell, R.W. Stringham and J.A. Blackwell, Anal Chem, 68 (1996) 3645.

538

[21] L. Miller and M. Potter, J Chromatogr B Analyt Technol Biomed Life Sci, 875

539

(2008) 230.

540

[22] L. Miller, J Chromatogr A, 1256 (2012) 261.

541

[23] C. Hamman, M. Wong, M. Hayes and P. Gibbons, J Chromatogr A, 1218 (2011)

542

3529. 21 Page 21 of 39

546 547 548

[25] D. Leyendecker, in R.M.Smith (Editor), Supercritical fluid chromatography, Royal Society of Chemistry, Loughborough (UK), 1988, p. 53.

ip t

545

(2008) 48.

[26] K.W. Phinney, in G. Subramanian (Editor), Chiral separation techniques, a

cr

544

[24] P. Franco and T. Zhang, J Chromatogr B Analyt Technol Biomed Life Sci, 875

practical approach, Wiley, Weinheim (Germany), 2001, p. 301.

us

543

[27] A. Tarafder and G. Guiochon, J Chromatogr A, 1265 (2012) 165.

550

[28] Y.K. Ye, K.G. Lynam and R.W. Stringham, J. Chromatogr. A, 1041 (2004) 211.

551

[29] K.W. Phinney and L.C. Sander, Chirality, 15 (2003) 287.

M

an

549

te

554

Ac ce p

553

d

552

22 Page 22 of 39

cr

ip t

Figure 1

Figure 1: Scheme of the screening step as defined in [13]. In the top row the chiral stationary phases are presented, while the second row

an

us

represents the used modifier concentration in the carbon-dioxide based mobile phase.

Chiralpak AD-H

Chiralcel OD-H or Lux Cellulose-1

20% MeOH**

20% 2PrOH**

20% MeOH**

Lux Cellulose-4

20% 2PrOH**

d

M

Chiralpak OZ-H or Lux Cellulose-2

Ac c

ep te

** With 0.1% isopropylamine and trifluoroacetic acid for all compounds.

Page 23 of 39

Figure 2

Figure 2: Separation results of clopidogrel on Chiralpak® AD-H with varying concentrations (2PrOH+ 0.1%TFA + 0.1%IPA) in the mobile phase a) obtained resolutions and b) the analysis time.

us

cr

ip t

a)

Ac

ce pt

ed

M

an

b)

Page 24 of 39

ip t

Figure 4

Ac c

ep te

d

M

an

us

cr

Figure 4: Overlay of the chromatograms of econazole on Chiralcel® OZ-H with 20% (MeOH:IPA:TFA, 100:0.1:0.1, v/v/v) in the mobile phase. A flow rate of 3 ml/min and temperature of 30°C was used. The backpressures were 1) 125bar; 2) 150bar; 3) 175bar; 4) 200bar; 5) 225bar; and 6) 250bar. (Results generated with the UPC² system)

Page 25 of 39

Figure 5

tR1 = 2.42min tR2 = 2.66min

Rs = 1.34 α = 1.15

tR1 = 2.40min tR2 = 2.63min

Rs = 1.39 α = 1.15

tR1 = 2.38min tR2 = 2.60min

Rs = 1.42 α = 1.15

e) 30°C

tR1 = 2.37min tR2 = 2.58min

Rs = 1.43 α = 1.14

f) 35°C

tR1 = 2.36min tR2 = 2.57min

Rs = 1.44 α = 1.14

g) 40°C

tR1 = 2.35min tR2 = 2.56min

Rs = 1.43 α = 1.14

tR1 = 2.36min tR2 = 2.56min

Rs = 1.43 α = 1.14

Ac

h) 45°C

M

d) 25°C

ed

c) 20°C

ce pt

b) 15°C

cr

Rs = 1.30 α = 1.16

an

tR1 = 2.45min tR2 = 2.70min

us

a) 10°C

ip t

Figure 5: Chromatograms of the enantioseparation of carprofen on Chiralcel® OZ-H with 20% (2PrOH:IPA:TFA, 100:0.1:0.1, v/v/v) in the mobile phase. A total flow rate of 4 ml/min and back pressure of 150 bar was used. The temperatures were a) 10°C; b) 15°C; c) 20°C; d) 25°C; e) 30°C; f) 35°C; g) 40°C; and h) 45°C. (Results generated on the UPC² system)

Page 26 of 39

Ac c

ep te

d

M

an

us

cr

Figure 6: Chiral separation strategy for polysaccharide-based columns in SFC

ip t

Figure 6

Page 27 of 39

Figure 9

Figure 9: Transfer of the chromatographic conditions from conventional SFC (Thar equipment) to ultraperformance SFC (UPC² equipment). The separations are obtained on Lux Cellulose-2, with 20% (MeOH:IPA:TFA, 100:0.1:0.1, v:v:v) in the mobile phase, flow rate 3.0 ml/min, 30°C, detection at 220nm, and a back pressure of 150 bar. a) mepindolol, b) naringenin, c) mianserine.

a)

Mepindolol Thar Rs = 6.86 tR1 = 6.10 min tR2 = 10.13 min

Naringenin

us

b)

cr

ip t

UPC² Rs = 6.76 tR1 = 5.47 min tR2 = 9.26 min

UPC² Rs = 3.16 tR1 = 5.72 min tR2 = 7.60 min

M

an

Thar Rs = 2.92 tR1 = 6.27 min tR2 = 8.12 min

ce pt

Thar Rs = 2.08 tR1 = 6.02 min tR2 = 6.79 min

ed

Mianserine

UPC² Rs = 2.18 tR1 = 5.43 min tR2 = 6.22 min

Ac

c)

Page 28 of 39

Figure 10

Figure 10: Number of baseline (Rs > 1.5) and partial (0 < Rs < 1.5) separations achieved with the Thar SFC and UPC² systems on the four chromatographic systems of the screening.

OD-H

LC-4

ip t

AD-H

Ac

ce pt

ed

M

an

us

cr

OZ-H

Page 29 of 39

Figure 11

Figure 11: Comparison of the screening results of the 56-compound test set on the ultraperformance and conventional SFC equipment. a) resolutions (Rs), b) analysis times. Straight line = line of equality.

a)

ip t

20

cr

15

us

10 5 0

b)

5

10 15 20 Rs conventional SFC

ed

30 25

ce pt

20 15 10 5

Ac

AT ultra-performance SFC (min)

25

M

0

an

Rs ultra-performance SFC

25

0

0

5 10 15 20 25 AT conventional SFC (min)

30

Page 30 of 39

ip t

Figure 7

cr

Figure 7: Separation strategy applied on thioridazine. Chromatograms a-d: screening experiments, e: optimized conditions. (Results generated with the UPC² system) a)

us

Chiralcel OZ-H Rs = 3.1 AT = 13.9 min

d)

e)

ep te

Chiralcel OD-H Rs = 0.0 AT = 3.8 min

d

c)

M

Chiralpak AD-H Rs = 0.0 AT = 14.1 min

Lux Cellulose-4 Rs = 6.4 AT = 32.1 min

OPTIMIZED CONDITIONS Chiralcel OZ-H, 35% (MeOH/IPA/TFA, 100/0.10.1, v/v/v, flow rate 4.0 ml/min) Rs = 1.8 AT = 3.4 min

Ac c

SCREENING

an

b)

Page 31 of 39

ip t

Figure 8

d)

e)

us an M

Chiralcel OD-H Rs = 0.0 AT = 2.1 min

Lux Cellulose-4 Rs = 0.6 AT = 1.9 min

d

SCREENING

c)

Chiralpak AD-H Rs = 4.2 AT = 2.9 min

ep te

b)

Chiralcel OZ-H Rs = 0.0 AT = 2.0 min

Ac c

a)

cr

Figure 8: Separation strategy applied on clopidogrel. Chromatograms a-d are the results from the screening step, e is the result after optimization. (Results generated with the UPC² system)

OPTIMIZED CONDITIONS: Chiralpak AD-H, 35% (2PrOH/IPA/TFA, 100/01/0.1, v/v/v, flow rate 4.0 ml/min) Rs = 2.5 AT = 1.4 min

Page 32 of 39

ip t

Figure 3

cr

Figure 3: Results of the enantioseparation of cetirizine on Chiralpak AD-H with 20% (2PrOH + 0.1%IPA + 0.1%TFA) in the mobile phase as a function of the flow rate. (a) Overlay of the obtained chromatograms; (b) analysis time; (c) resolution and selectivity as a function of the flow rate.

(b)

Ac c

ep te

d

M

an

us

(a)

(c)

Page 33 of 39

Table 1

TABLE 1 : Test-set compounds used to evaluate the separation strategy. Racemate Carprofen

Structure

Origin Sigma-Aldrich, Steinheim, Germany

Madaus AG, Köln, Germany

Celiprolol

Origin unknown

Ceterizine

Sigma-Aldrich, Steinheim, Germany

Clopidogrel

Origin unknown

an

us

cr

ip t

Carteolol

Gift from Phenomenex

M

Cyclopentolate

Janssen research foundation, Beerse, Belgium

ed

Econazol

Fluoxetine

Ac

Indapamide

ce pt

Felodipine

Hassle (Astra), Sweden

Sigma-Aldrich, Steinheim, Germany Sigma-Aldrich, Steinheim, Germany

Indoprofen

Sigma-Aldrich, Steinheim, Germany

Isradipine

Origin unknown

Lorazepam

Wyeth, NY, USA

Page 34 of 39

TABLE 1 : Test-set compounds used to evaluate the separation strategy. Janssen research foundation, Beerse, Belgium

D/L-Nebivolol

Janssen research foundation, Beerse, Belgium

Ondansetron

Glaxo Wellcome, Belgium

Temazepam

Origin unknown

Terazosine

Sigma-Aldrich, Steinheim, Germany

an

us

cr

ip t

Miconazol

Origin unknown

M

Thioridazine

Origin unknown

Ac

ce pt

ed

Trans-stilbene oxide

Page 35 of 39

Table 2

TABLE 2: Number of separations obtained with the 56-compound test set using 20% methanol (MeOH) or 2-propanol (2PrOH) in the mobile phase (with 0.1% isopropylamine and 0.1% trifluoroacetic acid added to the modifier). The separations that are only obtained with one modifier on a given stationary phase are considered as unique separations. Unique separations MeOH 2PrOH 15 2 3 17 7 2 9 4

ce pt

ed

M

an

us

cr

ip t

Partial separations MeOH 2PrOH 18 10 11 12 14 10 14 12

Ac

OZ-H AD-H OD-H LC-4

Baseline separations MeOH 2PrOH 27 22 12 25 27 25 28 24

Page 36 of 39

ip t

Table 3

d

M

an

us

Selected optimal separation conditions CSP Flow rate Modifier Modifier type (ml/min) (%) OZ-H 4.0 20 MeOH:IPA:TFA, 100:0.1:0.1, v:v:v OD-H 4.0 30 MeOH:IPA:TFA, 100:0.1:0.1, v:v:v AD-H 4.0 15 EtOH:IPA:TFA, 100:0.1:0.1, v:v:v AD-H 4.0 35 2PrOH:IPA:TFA, 100:0.1:0.1, v:v:v AD-H 4.0 35 2PrOH:IPA:TFA, 100:0.1:0.1, v:v:v AD-H 3.0 20 2PrOH:IPA:TFA, 100:0.1:0.1, v:v:v OZ-H 4.0 20 MeOH:IPA:TFA, 100:0.1:0.1, v:v:v AD-H 4.0 10 2PrOH:IPA:TFA, 100:0.1:0.1, v:v:v OZ-H 2.0 5 MeOH:IPA:TFA, 100:0.1:0.1, v:v:v OD-H 4.0 30 MeOH:IPA:TFA, 100:0.1:0.1, v:v:v AD-H 4.0 35 2PrOH:IPA:TFA, 100:0.1:0.1, v:v:v LC-4 3.0 10 2PrOH:IPA:TFA, 100:0.1:0.1, v:v:v OZ-H 4.0 35 MeOH:IPA:TFA, 100:0.1:0.1, v:v:v AD-H 4.0 15 2PrOH:IPA:TFA, 100:0.1:0.1, v:v:v OZ-H 4.0 25 MeOH:IPA:TFA, 100:0.1:0.1, v:v:v OD-H 4.0 40 MeOH:IPA, 100:0.1, v:v OZ-H 4.0 35 MeOH:IPA:TFA, 100:0.1:0.1, v:v:v AD-H 4.0 30 MeOH:IPA, 100:0.1, v:v OZ-H 4.0 35 MeOH:IPA:TFA, 100:0.1:0.1, v:v:v OZ-H 3.0 20 MeOH:IPA:TFA, 100:0.1:0.1, v:v:v

ep te

Ac c

Carprofen Carteolol Celiprolol Ceterizine Clopidogrel Cyclopentolate Econazole Felodipine Fluoxetine Indapamide Indoprofen Isradipine Lorazepam Miconazol D/L-Nebivolol Ondansetron Temazepam Terazosine Thioridazine Trans-stilbene oxide

Separation results Rs α AT (min) 1.6 1.2 2.6 2.6 6.5 1.3 1.5 1.3 3.8 5.5 2.4 2.2 2.5 1.5 1.4 4.8 1.7 2.8 1.6 1.1 5.3 2.0 1.2 4.9 1.3 1.1 15.0 1.5 1.3 3.6 2.7 1.2 4.5 1.6 1.1 7.2 3.0 1.4 2.9 2.0 1.2 5.4 2.2 1.5 1.9 3.4 1.4 3.0 2.0 1.2 4.2 1.7 1.2 3.7 1.8 1.2 3.4 4.4 1.6 1.9

cr

Table 3: For separation strategy: separation results and optimal separation conditions for the 20 compounds from the test set.

Page 37 of 39

Table 4

Table 4: Two-tailed paired student t-test applied on the data obtained for the 56 compounds (58 enantiomer pairs) on the Thar and UPC².

AT

t-value

Chiralpak AD-H 20% 2PrOH:IPA:TFA 1.66

Chiralcel OD-H 20% MeOH:IPA:TFA 3.10

Lux Cellulose-4 20% 2PrOH:IPA:TFA 0.41

p-value

1.46.10-3

5.15.10-2

1.50.10-3

3.42.10-1

t-value

2.06

1.71

1.50

4.43

p-value

2.20.10-2

4.62.10-2

6.93.10-2

2.14.10-5

ip t

Rs

Chiralcel OZ-H 20% MeOH:IPA:TFA 3.11

Ac

ce pt

ed

M

an

us

cr

With Rs the resolution and AT the analysis time as responses. Null hypothesis H0 : XThar = XUPC², with X a given response (Rs or AT). t57,α=0.05 = 1.67. Significant t- and p-values are marked in bold.

Page 38 of 39

Table 5

Table 5: Results of the six precision studies of two sample injections on six consecutive days on the UPC² and Thar systems, expressed in variances. Intermediate precision UPC² Thar

3.86.10-2 2.47.10-2

2.03.10-2 6.23.10-3

1.14.10-1 4.19.10-3

5.85.10-2 6.58.10-4

1.53.10-1 2.89.10-2

7.88.10-2 6.88.10-3

1.57.10-3 2.06.10-2

1.44.10-2 7.67.10-3

4.98.10-3 1.18.10-2

1.78.10-1 1.89.10-3

6.56.10-3 3.24.10-2

1.93.10-1 9.56.10-3

9.9.10-4 2.06.10-2

1.89.10-5 3.87.10-3

2.99.10-4 1.18.10-2

4.06.10-4 9.17.10-5

1.29.10-3 3.24.10-2

4.25.10-4 3.96.10-3

1.88.10-3 8.78.10-4

8.06.10-4 1.08.10-3

7.03.10-5 6.33.10-6

1.39.10-3 2.50.10-5

2.03.10-3 3.46.10-3

3.14.10-4 2.80.10-3

5.30.10-5 7.17.10-6

7.43.10-4 4.17.10-5

2.08.10-3 3.46.10-3

1.06.10-3 2.84.10-3

2.48.10-3 8.87.10-4

7.13.10-2 4.15.10-2

2.03.10-4 3.75.10-6

1.33.10-3 2.65.10-4

2.68.10-3 8.91.10-4

7.26.10-2 4.18.10-2

cr

ip t

Inter-day variability UPC² Thar

1.95.10-3 8.84.10-4

us

an

Bopindolol Rs AT Mepindolol Rs AT Methadone Rs AT Mianserine Rs AT Sotalol Rs AT Naringenin Rs AT

Intra-day variability UPC² Thar

2.20.10-3 1.11.10-3

Ac

ce pt

ed

M

With Rs the resolution and AT the analysis time. The results obtained on the UPC² and Thar are compared with an F-test, the smallest variance of both is marked in bold if the difference is calculated to be significant, F11,11;α=0.05 = 2.82.

Page 39 of 39