Accepted Manuscript Title: Adsorption of aqueous organic mixtures on a chiral stationary phase with bound antibiotic eremomycin Author: Yuliya K. Nikitina Imran Ali Leonid D. Asnin PII: DOI: Reference:

S0021-9673(14)01327-2 http://dx.doi.org/doi:10.1016/j.chroma.2014.08.062 CHROMA 355753

To appear in:

Journal of Chromatography A

Received date: Revised date: Accepted date:

17-5-2014 23-7-2014 20-8-2014

Please cite this article as: Y.K. Nikitina, I. Ali, L.D. Asnin, Adsorption of aqueous organic mixtures on a chiral stationary phase with bound antibiotic eremomycin, Journal of Chromatography A (2014), http://dx.doi.org/10.1016/j.chroma.2014.08.062 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.

1

Adsorption of aqueous organic mixtures on a chiral stationary phase with bound antibiotic

2

eremomycin

3 Yuliya K. Nikitina1, Imran Ali2, Leonid D. Asnin1,*

4 5

1

Perm National Research Polytechnic University, 29 Komsomolski Al., Perm 614990, Russia

6

2

Jamia Millia Islamia (Central University), Jamia Nagar, New Delhi 110025, India

7

ip t

8

The adsorption of two typical hydro-organic mobile phases, with methanol and

10

acetonitrile as the organic component, on an antibiotic based chiral stationary phase (CSP)

11

Nautilus-E was studied by the minor perturbation method. In both cases, the excess adsorption of

12

water was positive over a wide range of concentrations from 0 to ~75 or 90 mol % for MeOH or

13

MeCN containing mobile phases, respectively. Such hydrophilic properties of the CSP were

14

attributed to multiple polar functional groups of the chiral ligand and to the residual silanol

15

groups of the silica support. The adsorbed phase was found to be thinner for H2O-MeOH (~ 1.1

16

Å) and thicker for H2O-MeCN (9.4 Å). The measurements of the column hold-up volume by

17

different methods allowed us to suggest a model of the adsorbed phase consisting of the volume

18

between bound chiral selectors inaccessible to large size molecules and of the stagnant layer of

19

the mobile phase adsorbed on the external surface of the chiral selectors.

ed

M

an

us

cr

9

20

pt

Key words: chiral stationary phase, eremomycin, excess adsorption, adsorption thermodynamics

22 23 24 25 26

*

Ac ce

21

Corresponding author: Tel./fax: +7(342)2391-511. E-mail address: [email protected]

Page 1 of 31

2 26

1. Introduction

27 Adsorption of binary solutions on solids was intensively studied in the 1960s through the

29

1980s [1–3] in particular to explain properties of HPLC packings in contact with

30

multicomponent mobile phases [4,5]. A solid theoretical basis for these works was provided by

31

Schay and co-workers [1] and Everett [6,7]. This period of “Storm and Stress” had been

32

followed by a peaceful decade of abating interest, not in the last part because the experimental

33

protocol used at that time demanded to keep constant the mass of the solution that is not

34

convenient in chromatography where the volume of solution is controlled. A revival of attention

35

to this problem was due to a series of publications by Kazakevich et al. [8–10], who not only

36

adopted from earlier works [11,12] a convenient technique to measure excess adsorption

37

isotherms of binary liquid mixtures, the minor perturbation method, but, more importantly,

38

demonstrated the usefulness of the method to study the structure of the adsorbed layer on

39

stationary phases.

an

us

cr

ip t

28

This approach has successfully been used by some authors to understand retention

41

mechanisms in reversed phase (RP) liquid chromatography [13,14] and in order to characterize

42

RP stationary phases [15]. It also proved to be informative in chiral chromatography. So,

43

Cavazzini and co-workers studied the adsorption of binary eluents on polar chiral stationary

44

phases (CSPs) to clarify the effect of strong mobile phase additive on enantioseparation [16,17].

45

The authors of [18] derived information about the composition and thickness of the surface

46

liquid layer on a Whelk-O1 CSP from the excess adsorption isotherms of binary solvents. The

47

purpose of the present work is to apply the method in question to the investigation of surface

48

properties of a silica-based CSP with grafted antibiotic eremomycin (Fig. 1) in contact with

49

typical chromatographic mobile phases. This chiral adsorbent is a recent addition to the family of

50

macrocyclic antibiotic CSPs, demonstrating good separation ability for enantiomers of amino

51

acids [19–21] and 2-arylpropionic acids [21,22]. This CSP was shown to behave similarly to the

52

representative antibiotic CSP Chirobiotic T (with teicoplanin) in the chromatography of amino

53

acids [20]. Interaction of mixed solvents with the surface of antibiotic-based CSPs including the

54

CSP in question has never been studied before, leaving researchers without proper knowledge of

55

the distribution of the solvent constituents between the stationary and mobile phases.

Ac ce

pt

ed

M

40

56 57

2. Theoretical

58 59

2.1. Excess adsorption

Page 2 of 31

3 Consider an adsorption equilibrium on a liquid/solid interface, the liquid being a binary

61

mixture. The adsorption measured in both batch and chromatographic experiments is in fact

62

excess adsorption understood in the Gibbsian sense [11,12]. It is defined as the excess amount of

63

a solute present in a system over the amount contained in a hypothetical reference system in

64

which the solute concentration remains uniform throughout the whole liquid phase up to the

65

solid surface and is equal to the bulk solute concentration in the real system. The choice of the

66

reference system is arbitrary but in practice limited by the experimental setup. A natural

67

reference system in the batch method is one containing the same total number of moles n0 of the

68

liquid phase as in the real adsorption system. Such a selection of the reference system is called

69

the n convention [23]. The corresponding adsorption excess of component i per unit area of the

70

solid surface is given by

us

cr

ip t

60

71 i

72





n0 x 0l ,i  xil A

 , i = 1, 2

(1)

an

n 

73

where A is the total surface area of an adsorbent, x0l ,i and xil the initial (before contact of a liquid

75

with a solid) and equilibrium molar fraction of component i, respectively. The notation in 

76

symbolizes the use of the n convention. Summing the adsorption excesses for both components

77

of the liquid phase and taking into account that x1l  x 2l  1 , we obtain a displacement rule

ed

M

74

pt

78

80

1n   2n 

(2)

Ac ce

79

81

meaning that positive adsorption (accumulation) of one component of a binary mixture always

82

results in ousting from the surface layer the molecules of the other component and vice versa.

83

Another useful choice of a reference system, the V convention [23], assumes that (i) the

84

volume of the liquid phase in the reference system is equal to that in the real system and (ii) that

85

the volumes of the liquid phase before and after contact with the adsorbent are the same. The

86

corresponding specific adsorption excess of component i is expressed through the volume of the

87

liquid phase V0 and the molar concentrations of i before (c0,i) and after (ci) equilibrium is

88

established:

89 90

iv  



V0 c 0 , i  c i A

 , i = 1, 2

(3)

91 Page 3 of 31

4 92

The V convention is natural to chromatographic experiment, in which the volume of the

93

liquid phase remains constant, limited by the column walls. Guggenheim and Adam [23] have

94

shown that the specific excess adsorption values based on different conventions relate one to

95

another as

96



in   iv   x il 1v   2v 

97



ip t

98

(4)

This transformation is necessary in order to make use of the elaborate theory of

100

adsorption equilibrium of the batch experiment [6,7,24] for the analysis of experimental data

101

obtained by means of chromatography. There have been disputes, still unresolved, about

102

thermodynamic consistency of Eq. (4) [24–26], which is undoubtful only for ideal systems. A

103

current approach consists in using this expression and neglecting a probable inconsistency in

104

results, which is not larger than a few percent, as estimated from the volume contraction effect

105

for the mobile phases studied.

an

us

cr

99

106

2.2. Excess adsorption, total adsorbed amount, and thickness of the adsorbed layer

108

The concept of surface excess, although directly relating to experimentally measured

109

quantities, has an inherent drawback that impedes its use in studying the adsorption equilibrium..

110

Consideration of this phenomenon in terms of the mass action law (see Eq. (11) below) requires

111

the use of the total surface concentrations of the liquid phase components rather than the excess

112

quantities. A definition of the surface concentration must rely on a certain model of the surface

113

layer as it requires assigning a specific volume to the surface layer. The simplest model was

114

developed by Guggenheim and Adam [23] who suggested the division of the liquid phase by two

115

parts. One part, adjoining to the solid surface and extending normally to the surface at a distance

116

τ, is assumed to contain all excess amounts of the liquid phase components. This is the adsorbed

117

phase. Above this layer of thickness τ, there is the bulk phase where the component

118

concentrations are supposed to be equal to the equilibrium ones. The specific total content of

119

component i in the adsorbed phase is

Ac ce

pt

ed

M

107

120 121 122 123

q i  iv   c i

(5)

qi  i n   xil q 

(6)

or

124

Page 4 of 31

5 125

In Eq. (6), the thickness of the adsorbed phase is implicitly included in the saturation

126

capacity of the adsorbed layer, q = q1 + q2. Indeed, general logics suggests that the number t of

127

monomolecular layers constituting the adsorbed phase is given by [27]

128 t  a1 q1  a 2 q 2

129

(7)

130 The molar cross-sectional areas a i of the compounds are, in general, functions of the surface

132

concentration. Neglecting this circumstance and also assuming that t is constant, a useful

133

expression for the molar fraction x1a of component 1 in the adsorbed layer is derived from Eqs.

134

(6) and (7) [24,27]:

135

tx1l  a 2 1n  t  a 2  a1 1n 



137



(8)

an

x1a 

136

us

cr

ip t

131

The application of Eqs. (7) and (8) requires knowledge of a i . In this study, we use the

139

values recommended by Gritti and Guiochon [27]: 78,000; 130,000; 160,000 m2/mol for water,

140

methanol, and acetonitrile, respectively.

M

138

Depending on the convention chosen, both Eq. (5) and Eq. (8) may be used to evaluate

142

the thickness of the adsorbed layer, although in somewhat different terms, provided that

143

additional assumptions are made. The first of them is that this characteristic does not depend on

144

the composition of the liquid adsorbed. To determine τ from Eq. (5), one should also assume,

145

following Schay et al. [1], that the derivative (dqi/dci) is equal to zero within the concentration

146

range corresponding to the linear part of the excess adsorption isotherm. Then, in that

147

concentration interval,   div  dc i  .

148

Ac ce

pt

ed

141

Eq. (8) in combination with the thermodynamic condition for two-phase coexistence

149

x

150

the isotherm’s inflection point I [,27,28] as

a 1

x1l



T

 0 [28] allows evaluation of t via the derivative of the excess adsorption isotherm at

151 152

 d n   t   1 l   x1l a1  x 2l a 2  1n  I  a 2  a1  dx1  I



   



(9)

153

Page 5 of 31

6 154

Inequality (9) defines the lower limit of the surface phase thickness. The true number of

155

adsorption monolayers may be greater than this value yet cannot be determined within the

156

framework of the classic thermodynamics [27,28].

157 158

2.3. Adsorption equilibrium

159

Consider the following liquid/solid displacement process

ip t

160 (1)l + (2)a  (1)a + (2)l

161

cr

162

(10)

Superscripts l and a refer to the bulk liquid and adsorbed phase, respectively. The adsorption

164

equilibrium constant is given by

us

163 165

K

x1a  1a x 2l  l2 x1l  1l x 2a  a2

(11)

an

166 167

Activity coefficient γ is a measure of how far deviates the behavior of a given component

169

from its behavior in a hypothetical ideal solution, bulk or adsorbed, of the same composition. We

170

employ the usual definition of activity coefficients in the bulk phase [7], with  li → 1 when xil →

171

1. Activity coefficients for the adsorbed phase are defined, following Everett [7], in terms of the

172

difference between the free energy of the real system and that for the reference ideal system:

175 176

ed

pt

174



 

RT ln  ia   ia  a i   ia ,id   id a i



(12)

Ac ce

173

M

168

In this equation, R is the gas constant, T is temperature,  ia is the chemical potential of

177

component i in the adsorbed phase, σ is the surface tension at the liquid/solid interface, and ai is

178

the partial molar area of i. This later quantity is equal to the ratio a i t , presumed to be

179

independent of the adsorption phase composition. The superscript id refers to the ideal reference

180

state. Again, the γas are so defined that  ia → 1 when xia → 1. It is seen from Eq. (12) that γa > 1

181

when the situation for component i in the real surface phase is less energetically favorable than

182

that in the hypothetical ideal system, γa < 1 when this relation is opposite, and γa = 1 when the

183

real surface solution is not distinguishable from the ideal one.

184 185

Larionov and Myers [3] have developed a method for computing γa (see also [27]). The





calculation algorithm starts from the calculation of ln  1a  a2 :

Page 6 of 31

7 186 a1   10 a 2    02  1a  1l ln a  ln l  ln S   t RT t RT 2 2

187

(13)

188

190

surface.

 x x , a l 2 1

 i0 is the interfacial tension between the pure liquid i and the solid

 



 



191

n  l x1l 1l    10 x1 1   d x1l  1l l l l 1 RT x1  1 x 2

192

n  l x1l 1l     02 x1 1   d x1l  1l l l 0 RT x1  1 x 2l

(14a) (14b)

us

193

ip t

where S  x1a x 2l

cr



189

The bulk activity coefficients  1l and  l2 are found either in the literature or are calculated with

195

the help of an appropriate theoretical method, lnS is calculated using the surface molar fractions

196

x1a and x 2a  1  x1a obtained by Eq. (8) from an excess adsorption isotherm 1n  x1l . The later

197

function is derived through Eq. (4) from experimentally measured isotherms 1v  x1l

198

2v  x2l .

ln  1a  x 2a ln

206 207

Ac ce

204 205

pt

201

203

 

and

The individual surface activity coefficients are then computed as follows

200

202

 

M

 

ed

199

an

194

x1a  1a  1a a  ln dx1  a2 0  a2

ln  2a   x1a ln

x1a  1a  1a a  ln dx1  2a 0  2a

(15a) (15b)

3. Experimental 3.1. Apparatus

208

All the experiments were carried out using a LC20AD-XR liquid chromatograph

209

(Shimadzu, Japan), equipped with a solvent delivery system, an automatic injector, a column

210

oven with a temperature control precision of ±0.1°С, a refractive index (RI) detector, a DAD

211

detector (used only for hold-up volume determination with 1,3,5-tri-tert-butylbenzene), and data

212

acquisition system running the Laboratory Solution software from the same manufacturer. The

213

extra-column volume measured from the autosampler to the RI detector in the system with a

214

zero-volume connector installed in place of the chromatographic column was 0.140 ml. When Page 7 of 31

8 215

the DAD detector was used instead of the RI detector, the extra-column volume was 0.054 ml.

216

All the retention data were corrected for this contribution.

217 3.2. Chemicals and column

219

The mobile phases were prepared from chemically pure grade methanol from Vekton

220

(Russia), HPLC far UV/gradient grade acetonitrile (J.T. Baker, USA), and water purified with a

221

Millipore Synergy system purchased from Millipore (Millipore, France). Methanol was

222

additionally dried over magnesium methylate followed by distillation [29]. Acetonitrile was used

223

without further treatment. 1,3,5-tri-tert-butylbenzene was from Sigma-Aldrich (USA).

224

Deuterated methanol (99.8 atom % D) used as a tracer in hold-up measurements was supplied by

225

Acros Organics (Belgium).

us

cr

ip t

218

The chromatographic column was a Nautilus-E (25 x 0.46 cm i.d.) manufactured by

227

BioChemMak S&T (Russia), packed with approximately 2.5 g of 5 µm, 100 Å pore size,

228

Kromasil-100 silica gel particles grafted with eremomycin. The specific surface area of the

229

material was 306 m2/g and the ligand bonding density was 0.2 mol/m2 according to the

230

manufacturer.

M

an

226

231 3.3. Procedures

233

3.3.1. Measurement of hold-up volume

234

The hold-up volume (V0) was estimated using three generally accepted approaches [30]:

235

by retention of an unretained tracer, by the isotopic method, and by the minor perturbation

236

method:

pt

(a) Retention of an unretained tracer. 1,3,5-tri-tert-butylbenzene (TtBB) was chosen for

Ac ce

237

ed

232

238

this purpose according to a common practice of using this compound to evaluate the V0 value for

239

chiral columns [31] as it is believed that the hydrophobic bulky side groups of the molecule

240

protect it from adsorbing on the surface of typical CSPs. The measurement was made by a

241

triplicate 1 l injection of a strongly diluted TtBB solution with pure MeOH or MeCN, or with

242

H2O:MeOH (or MeCN) (20:80, v/v) as a mobile phase. The column temperature was set at 25°С.

243

(b) Isotopic method. This method estimates the true void volume of a column as the

244

retention volume of an isotopically labeled solvent in the same but nonlabeled solvent [32].

245

Measurements were made at temperatures of 20, 25, 30 and 40oC using pure methanol as the

246

mobile phase and deuterated methanol as the tracer. The tracer concentration was 10 vol. %, and

247

the sample size was 2 l.

Page 8 of 31

9 248

(c) Minor perturbation method. The evaluation of the hold-up volume by the minor

249

perturbation method [33] was made based on the data obtained in measuring excess adsorption

250

isotherms as described in the following section.

251 3.3.2. Excess adsorption from binary solvents

253

Excess adsorption of the binary mobile phase components was studied by means of the

254

minor perturbation methods described in detail in [8,33], accuracy of the method discussed in

255

[34]. Two typical mobile phase systems were investigated: methanol-water and acetonitrile-

256

water. The excess adsorption isotherms of water from methanol were measured at 20, 25, 30 and

257

40°С. That of water from acetonitrile was determined only at 25°С. The flow rate was kept at 1

258

ml/min in all experiments. The column was successively equilibrated with a respective mobile

259

phase (H2O-MeOH or H2O-MeCN) containing 0, 0.5, 1, 2, 5, 10, 20, 30, 40, 45, 50, 60, 70, 80,

260

90, 95, 98, 99, 99.5, 100% (v/v) of water. Some of these points could be omitted but the total

261

number of experimental points per isotherm was not less than 17. After the equilibrium state was

262

established, a minor perturbation was induced by the injection of 2 l of one of two types of

263

solutions, in which there was a 5–10% excess of either water or the organic solvent over the

264

mobile phase composition. Retention times of these positive and negative perturbations were

265

averaged, and the averaged data were used for the calculation of the excess adsorption isotherm

266

according to the following equation:

ed

M

an

us

cr

ip t

252

267

pt Ac ce

269

1 ci VR ci   VT dci A 0

 v  ci  

268

(16)

270

where VR(ci) is the retention volume of the perturbation on the plateau of concentration ci of

271

component i and VT is the thermodynamic hold-up volume. This latter quantity is found from the

272

normalization condition for Eq. (16) [33]:

273 274

VT 

1 ci0



сi0 0

V R ci dci

(17)

275 276

where с i0 is the molar concentration of the pure liquid i. Eqs. (16) and (17) taken together imply

277

that the excess adsorption of either pure component of a binary mobile phase is equal to zero. In

278

other words, it assumes that the density of either pure liquid does not change when in a close

279

proximity to the solid surface. Subject to this restriction, VT is considered to be an estimate of the

280

volume V0. Page 9 of 31

10 281

The information on density of mobile phases at different temperatures necessary for the

282

conversion of volume fractions to mole fractions and to molar concentrations was found in the

283

literature: [35] for H2O-MeOH and [36] for H2O-MeCN. The correction for thermal expansion of

284

the mobile phase from the ambient temperature in the pump (T0) to the column temperature (Tcol)

285

was taken into account by multiplying the flow rate values and dividing the molar concentration

286

values by the ratio of the mobile phase densities at temperatures T0 and Tcol, respectively.

288

ip t

287 4. Results and Discussion

289 4.1. Hold-up volume

291

Table 1 summarizes the hold-up volume values found by different methods. It is seen that

292

the values determined from the retention of TtBB ( V0TtBB ) are essentially lower than those

293

measured by the minor perturbation and isotopic methods. Moreover, the values obtained with

294

different mobile phase compositions differ. A model in Fig. 2 tries to explain these findings.

295

First, it is assumed that there is a space between bound ligand moieties that is inaccessible to

296

relatively large molecules of TtBB but accessible to solvent molecules. This assumption is

297

supported by geometrical estimations made using ChemBio3D Ultra 11.0 software

298

(CambridgeSoft, USA) that show that a gap of approximately 8 Å between two bound ligands is

299

slightly less than the van der Waals size of a TtBB molecule, which is 9.6 Å. The dependence of

300

V0TtBB on the mobile phase composition may be explained by the formation of a layer of the

301

mobile phase on the external surface of chiral selectors, preventing a direct contact of

302

hydrophobic TtBB molecules with the polar bound antibiotic entities (Fig. 2). The thickness of

303

this intermediate layer (Δτ) depends on the mobile phase composition (Table 1) for reasons that

304

will be discussed later.

us

an

M

ed

pt

Ac ce

305

cr

290

The estimates of V0 by the minor perturbation and isotopic methods coincide within

306

±2.5% (VT(H2O-MeOH) < V0(isotopic) < VT(H2O-MeCN)) that may well be accounted for by an

307

experimental error. That cannot be explained by a random error and, therefore, should be

308

attributed to a temperature dependent methodological inaccuracy is a slight temperature trend in

309

the results of the minor perturbation method. A similar observation was made by Poplewska et

310

al. [37]. This inaccuracy is small, the drift being ~1% of V0 over the entire temperature range

311

studied, and does not challenge the assumptions underlying the method in question.

312 313

4.2. Excess adsorption

Page 10 of 31

11 Fig. 3 compares VR(xl) dependencies for H2O-MeOH and H2O-MeCN solutions. The

315

curves for the methanol and acetonitrile systems demonstrate some (qualitative) similarity in the

316

water-rich domain, above 65 mol % of water, but differ profoundly in quantitative terms in the

317

domain near pure organic solvent, where the VR values for the acetonitrile eluent are more than

318

two times those for the methanol eluent. It proves a contribution of the surface Brønsted sites to

319

the adsorption of the mobile phase. The residual silanol groups and, probably, some hydrogen

320

donating/accepting moieties of the chiral selector comprise this fraction of the adsorption sites.

321

Acetonitrile lacking a hydroxyl group is not able to interact with these sites as strongly as water.

322

Therefore, when the concentration of water in a H2O-MeCN mobile phase is too low to cover all

323

such surface sites, a strong retention of the perturbation peak must be observed due to the

324

adsorption of water on the uncovered those. This phenomenon is not so profound with H2O-

325

MeOH mobile phases because methanol molecules can shield the surface Brønsted sites with

326

their hydroxyl groups.

us

cr

ip t

314

The excess adsorption isotherms of water from both mobile phases in study are depicted

328

in Fig. 4. Both isotherms belong to the class of the excess isotherms with an azeotropic point, at

329

which the relative composition of the surface layer is identical to that of the bulk liquid [24]. In

330

both cases, the azeotropic point lies in the region of a water-rich composition, approximately 75

331

mol % for H2O-MeOH and 90 mol % for H2O-MeCN. Beyond this point, the excess adsorption

332

of water is negative and at lower water concentrations its surface excess is positive. Thus,

333

Nautilus-E can be considered a relatively hydrophilic adsorbent because it preferentially adsorbs

334

water over a larger part of the concentration range. This is attributed to polar yet ionogenic

335

functional groups of the eremomycin moiety as well as to the residual silanol groups, both types

336

able to bind water via hydrogen bonding. A much higher water excess in the case of the

337

acetonitrile comparing to methanol mobile phase suggests a larger capacity of the adsorbed

338

phase [1]. The evaluation of the thickness of the surface layer supports this conclusion. The data

339

in Table 2 shows that the estimates of this characteristic in terms of τ (Eq. (4)) and t (Eq. (9)) are

340

in qualitative agreement given the molecular size of the mobile phase components is 3-4 Å [38].

341

Increase in temperature results in a slight expansion of the thickness of the adsorbed phase

342

(Table 2). This is not surprising because elevated temperatures cause larger positional

343

fluctuations of molecules, resulting in their higher hydrodynamic diameters.

Ac ce

pt

ed

M

an

327

344

Many authors investigating the adsorption of mobile phases on RP stationary phases have

345

reported the thickness of the adsorbed layer of water-methanol being less than one nominal

346

molecular layer whereas that of water-acetonitrile being of 2-4 monomolecular layers

347

[8,13,25,27,39]. Similar results were obtained with polar CSPs in [18] and in the present work.

348

The fact that the same pattern is observed with adsorbents of quite different nature suggests that Page 11 of 31

12 the thickness of the adsorbed phase does not depend so much on the solvent-solid interactions as

350

on the solvent-solvent interactions. The mixing of methanol and water at any proportions is

351

known to result in solutions with a well ordered structure consisting of clusters maintained by

352

hydrogen bonds [40]. On the contrary, water added to acetonitrile induces the disintegration of

353

the structure of the pure liquid. In water-rich water-acetonitrile solutions, the original structure of

354

water is preserved, but acetonitrile clusters are not formed. As the total effect, the entropy of the

355

mixture increases [40]. This explains differences in the structure of the surface layers formed by

356

the two mobile phases in study. Contact of a water-methanol mixture with a hydrophilic solid

357

surface does not increase the degree of ordering comparing to the bulk phase. Therefore, there

358

will be no energy benefit from an extending the adsorbed layer. Contact of the same solution

359

with a hydrophobic surface disturbs the original structure of the liquid, and it is still beneficial

360

from energy point of view for the volume of the adsorbed phase be low. A contrary situation

361

takes place for water-acetonitrile mixtures. Contact with a surface induces ordering in the surface

362

solution. A hydrophilic surface (as in this study) preferentially coordinates water molecules, a

363

hydrophobic one coordinates acetonitrile molecules. The first layer projects its structure onto the

364

next one and so on until the Brownian motion equilibrates the ordering influence of surface

365

forces.

M

an

us

cr

ip t

349

These considerations keep true for the concentration range below the azeotropic point.

367

Beyond this point, in the area of water-rich solutions, different contributions (of solvent-solvent

368

and solvent-solid interactions) to the overall free energy of adsorption seem to superpose in such

369

a way that a positive excess adsorption of the organic component, but of very low amplitude,

370

occurs. Note that despite its negative excess adsorption the concentration of water in the

371

adsorbed phase is high (Fig. 5).

pt

Ac ce

372

ed

366

It is pertinent now to recur to Table 1 to notice, first, that the expelled volume for TtBB

373

correlates with the volume of the adsorption layer. One can even speculate that τ is a more

374

accurate estimate of the adsorption phase volume then the quantities τ and t, which are based on

375

a set of assumptions. Second, a drastic difference of τ for pure acetonitrile and an 80% (v/v)

376

acetonitrile solution may be explained now by a disturbing effect of water on the structure of

377

water-acetonitrile mixtures; in pure acetonitrile there is no energy gain in forming a thick

378

adsorbed layer.

379 380

4.3. Non-ideality of the adsorbed phase

381

Fig. 6 shows the activity coefficients of the mobile phase components in the bulk liquid

382

for both sorts of the mobile phases studied. These were computed using the UNIQUAC method

383

[41,42], which had been proven to estimate a liquid phase non-ideality with fair accuracy [43]. A Page 12 of 31

13 detail protocol of the calculation is given in the Supplementary material. Figs. 7 compare activity

385

coefficients in the adsorbed phase calculated by Eqs. (15a) and (15b) as functions of adsorbed

386

phase composition. The thus determined values are associated with inaccuracies resulted from

387

violations of the assumptions made to derive Eq. (15). The magnitude of these inaccuracies can

388

be estimated from a concentration dependence of the equilibrium constant K calculated by Eq.

389

(11). By definition, this quotient must be invariable; in practice, it appeared to be a smooth

390

function of mobile phase composition, varying around a weight-average value within ±20%.

391

Since this inconsistency is caused by the ratio  1a  a2 , the uncertainty γa in either activity

392

coefficient must not exceed ±10% according to the law of propagation of errors, supposing that

393

 1a   a2 . It is important to notice that the discussed uncertainties are inevitable as resulted

394

from imperfection of the thermodynamic model accepted, which at the moment is the most

395

accurate one to the best of our knowledge.

us

cr

ip t

384

an

396 4.3.1. Water-methanol system

398

The activity coefficients of both H2O and MeOH are less than unity throughout the whole

399

concentration range except the respective extreme points where these values are equal to 1 by

400

definition. This indicates a stabilizing effect of mixing and, probably, of the adsorption

401

interactions on both sorts of the molecules in the adsorbed phase. Both xa-γa curves demonstrate

402

a leap near the azeotropic point (Fig. 7a), corresponding to an abrupt change in the chemical

403

composition of the surface solution (Fig. 5a). Such a behavior of surface activity coefficients in

404

systems with an adsorption azeotrop is not unusual. In particular, Kiselev and Khopina [44]

405

observed an uneven change in the surface activity coefficients of toluene and n-heptane adsorbed

406

on a hydrophobic solid, around the azeotropic point.

ed

pt

Ac ce

407

M

397

The γa of MeOH is close to 1 in a wide concentration range from pure methanol to the

408

azeotropic point, revealing an ideal behavior of methanol molecules on the surface. On the

409

contrary, water molecules behave noticeably non-ideally at the same concentrations. The initial

410

a a increasing part of the x H2O -  aH2O plot at x H2O  0.2 is associated with localized adsorption of

411

water on the above mentioned strong Brønstedt sites, residual silanol groups in the first place. As

412

this fraction of the surface sites is saturated,  aH2O becomes constant up to the azeotropic point,

413

a although the surface composition is changing (Fig. 5a). Above x H2O  0.74 , the water fraction

414

behaves almost ideally, meaning that the presence of methanol molecules does not disturb

415

essentially regular water-water and water-surface interactions. At the same time,  aMeOH

416

decreases from ~1 to ~ 0.4 as the water surface concentration passes the azeotropic point. One

Page 13 of 31

14 417

can speculate, not without a reason [40], that when the concentration of alcohol is low, its

418

molecules are aggregated together due to the hydrophobic interactions and are localized on

419

hydrophobic parts of the chiral selectors.

420 4.3.2. Water-acetonitrile system

422

A situation in the water-acetonitrile adsorbed phase is quite contrary to that in the water-

423

methanol system. The surface activity coefficients of both components are higher than or close to

424

unity. This agrees well with the above described model of an H2O-MeCN surface solution whose

425

formation results in an increase of entropy. Indeed, the disordering supposes a dominating

426

influence of repulsive interactions. Those, in the virtue of Eq. (12), result in γa > 1. The existence

427

a of an adsorption azeotrop reveals itself in a sharp rising of  aMeCN as a function of x H2O . Unlike

428

for the water-methanol system, the activity coefficient of water in the water-acetonitrile surface

429

layer does not change stepwise as the water percentage passes the azeotropic point. Nonetheless,

430

it does decrease, although smoothly, and approaches a value of 1 beyond that point.

an

us

cr

ip t

421

Comparison of the plots in Fig. 7a and 7b shows that the surface activity coefficients

432

behave quite differently despite the fact that the adsorbed phases are formed on the same surface.

433

On the other hand, the present data are of qualitative similarity to those obtained by Gritti and

434

Guiochon [27] on a C18-stationary phase for the methanol and acetonitrile containing mixtures,

435

respectively. Thus, the nature of non-ideality in the surface solution is determined mainly by the

436

solvent-solvent interactions. The role of the solvent-surface interactions is secondary.

437 438 440

5. Conclusion

Ac ce

439

pt

ed

M

431

A CSP with bound to silica antibiotic eremomycin exhibits hydrophilic properties,

441

preferentially adsorbing water from hydro-organic solutions over a wide range of water

442

concentration. In both water-methanol and water-acetonitrile systems, there is a short domain of

443

organic-rich compositions where the water excess is negative. Even in this part of an adsorption

444

isotherm, the absolute content of water in the adsorbed phase is essential. The surface layer has a

445

complex structure, consisting of two compartments. The first one locates between the solid

446

support surface and the bound chiral selectors. This part of the surface layer is not in contact

447

with organic analytes if those large enough comparing to a gap between two neighboring chiral

448

selectors. The second compartment includes the layer of the mobile phase adsorbed on the

449

external surface of chiral selectors. This is the part of the adsorbed phase that contacts with

450

organic analytes. The average thickness of this layer is less than one nominal monomolecular

Page 14 of 31

15 451

layer with the water-methanol mixture and is larger, 2 to 4 monomolecular layers, with the

452

water-acetonitrile mixture. This phenomenon is explained by differences in the structure of

453

water-methanol (exothermic) and water-acetonitrile (endothermic) solutions. The fact that over a wide range of mobile phase compositions the adsorbed layer is

455

enriched in water as compared to the bulk liquid should affect the retention mechanism of

456

ionogenic molecules through permitting higher dissociation degree of such analytes in the

457

adsorbed than in bulk phase

ip t

454

458 459 Acknowledgement

461

This work was supported by the Russian Foundation for Basic Research (Grant No. 13-

462

03-92692) and by the Department of Science and Technology of India (Project No.

463

DST/INT/RFBR/P-147).

us

cr

460

Ac ce

pt

ed

M

an

464

Page 15 of 31

16 464

Figure captions

465 Figure 1. Eremomycin.

467

Figure 2. A model of the surface layer. A dashed line outlines the volume that is

468

inaccessible to large molecules of TtBB but accessible to small solvent molecules. Note that the

469

chiral selector consists of two parts: the chiral ligand proper and the tethering group anchoring

470

the ligand to the surface. Δτ symbolizes the layer of the solvent molecules adsorbed on chiral

471

selectors, an increment to the inaccessible volume. A distance between two neighboring chiral

472

selectors dgap is estimated to be 8 Å since a diameter of an eremomycin moiety is ~ 24 Å and the

473

sum dgap + dL ≈ 32 Å as found from the bonding density of chiral selectors.

cr

ip t

466

Figure 3. Retention volume of the minor perturbation peaks as a function of mobile

475

phase composition for water-methanol and water-acetonitrile mobile phases. The inset shows the

476

xl-VR plot for the water-methanol mobile phase on an enlarged scale.

us

474

Figure 4. Excess adsorption of water from methanol and acetonitrile at 25oC.

478

Figure 5. Composition of the adsorbed phase as a function of bulk phase composition for

479

water-methanol (A) and water-acetonitrile (B) solutions. Temperature is 25°С. Vertical dashed

480

lines mark the adsorption azeotrop composition.

484 485 486

M

ed

483

acetonitrile (B) solutions at 25°С.

Figure 7. Activity coefficients in the adsorbed phase for water-methanol (A) and wateracetonitrile (B) systems at 25°С.

pt

482

Figure 6. Activity coefficients in the bulk liquid phase for water-methanol (A) and water-

Ac ce

481

an

477

Page 16 of 31

17 486

References

487 [1] G. Schay, L.G. Nagy, T. Szekrenyesy, Period. Polytech. Chem. Eng. 4 (1960) 95.

489

[2] G. Schay, L.G. Nagy, J. Coll. Interface Sci. 38 (1972) 302.

490

[3] O.G. Larionov, A.L. Myers, Chem. Eng. Sci. 26 (1971) 1025.

491

[4] N. Le Ha, J. Ungvaral, E. Kováts, Anal. Chem. 54 (1982) 2410.

492

[5] K. László, L.G. Nagy, G. Fóti, G. Schay, Period. Polytech. Chem. Eng. 29 (1985) 73.

493

[6] D.H. Everett, Trans. Faraday Soc. 60 (1964) 1803.

494

[7] D.H. Everett, Trans. Faraday Soc. 61 (1965) 2478.

495

[8] Y.V. Kazakevich, R. LoBrutto, F. Chan, T. Patel, J. Chromatogr. A 913 (2001) 75.

496

[9] F. Chan, L.S. Yeung, R. LoBrutto, Y.V. Kazakevich, J. Chromatogr. A 1082 (2005) 158.

497

[10] S. Buntz, M. Figus, Z. Liu, Y.V. Kazakevich, J. Chromatogr. A 1240 (2012) 104.

498

[11] F. Riedo, E. Kováts, J. Chromatogr. 239 (1982) 1.

499

[12] H.L. Wang, U. Duda, C.J. Radke, J. Colloid. Interf. Sci. 66 (1978) 152.

500

[13] F. Gritti, G. Guiochon, Anal. Chem. 77 (2005) 4257.

501

[14] J. Mallette, M. Wang, J.F. Parcher, Anal. Chem. 82 (2010) 3329.

502

[15] B. Buszewski, Sz. Bocian, A. Felinger, J. Chromatogr. A 1191 (2008) 72.

503

[16] A. Cavazzini, G. Nadalini, V. Malanchin, V. Costa, F. Dondi, F. Gasparrini, Anal. Chem. 79

cr

us

an

M

(2007) 3802.

ed

504

ip t

488

[17] A. Cavazzini, G. Nadalini, V. Costa, F. Dondi, J. Chromatogr. A 1143 (2007) 134.

506

[18] L. Asnin, K. Horváth, G. Guiochon, J. Chromatogr. A 1217 (2010) 1320.

507

[19] S.M. Staroverov, M.A. Kuznetsov, P.N. Nesterenko, G.G. Vasiarov, G.S. Katrukha, G.B. Fedorova, J. Chromatogr. A 1108 (2006) 263.

513

Ac ce

508

pt

505

514

2010, pp. 203-222.

509 510

[20]. K. Petrusevska, M.A. Kuznetsov, K. Gedicke, V. Meshko, S.M. Staroverov, A. Seidel– Morgenstern, J. Sep. Sci. 29 (2006) 1447.

511

[21] A. Berthod, H.X. Qiu, S.M. Staroverov, M.A. Kuznestov, D.W. Armstrong, Chiral

512

Recognition with Macrocyclic Glycopeptides: Mechanisms and Applications, in: A. Berthod, (ed.), Chiral Recognition in Separation Methods, Springer, Berlin, Heidelberg,

515

[22] E.N. Reshetova, L.D. Asnin, Russ. J. Phys. Chem. A 83 (2009) 547.

516

[23] E.A. Guggenheim, N.K. Adam Proc. Roy. Soc. Lond. A 139 (1933) 218.

517

[24] D.H. Everett, Pure Appl. Chem. 58 (1986) 967.

518

[25] C.S. Koch, F. Köstner, G.H. Findenegg, J. Chromatogr. 406 (1987) 257.

519

[26] A.V. Vernov, A.A. Lopatkin, Russ. J. Phys. Chem. 55 (1981) 240.

520

[27] F. Gritti, G. Guiochon, J. Chromatogr. A 1155 (2007) 85. Page 17 of 31

18 521

[28] M. Jaronec, J. Goworek, A. Dabrowski, Coll. Polymer Sci. 263 (1985) 771.

522

[29] A.J. Gordon, R.A. Ford, The Chemist's Companion: A Handbook of Practical Data,

523

Techniques, and References, Wiley, New York, 1972. [30] C.A. Rimmer, C.R. Simmons, J.G. Dorcey, J. Chromatogr. A 965 (2002) 219.

525

[31] W.H. Pirkle, C.J. Welsh, J. Liq. Chromatogr. 14 (1991) 1.

526

[32] J.H. Knox, R. Kaliszan, J. Chromatogr. 349 (1985) 211.

527

[33] Y.V. Kazakevich, H.M. McNair, J. Chromatogr. Sci. 31 (1993) 317.

528

[34] J. Lindholm, P. Forssén, T. Fornstedt, Anal. Chem. 76 (2004) 4856.

529

[35] G.C. Benson, O. Kiyohara, J. Solution Chem. 9 (1980) 791.

530

[36] Y.P. Handa, G.C. Benson, J. Solution Chem. 10 (1981) 291.

531

[37] I. Poplewska, W. Piatkowski, D. Antos, J. Chromatogr A, 1103 (2006) 284.

532

[38] Ch.E. Webster, R.S. Drago, M.C. Zerner, J. Am. Chem. Soc. 120 (1998) 5509.

533

[39] S. Bocian, P. Vajda, A. Felinger, B. Buszewski, Anal. Chem. 81 (2009) 6334.

534

[40] A. Wakisaka, H. Abdoul-Carime, Y. Yamamoto, Y. Kiyozumi, J. Chem. Soc. Faraday

cr

us

an

535

ip t

524

Trans. 94 (1998) 369.

[41] T.F. Anderson, J.M. Prausnitz, Ind. Eng. Chem. Proc. 17 (1978) 552.

537

[42] A. Fredenslund, J. Gmehling, P. Rasmussen, Vapor-Liquid Equilibria Using UNIFAC,

541 542

ed

540

[43] R.C. Reid, J.M. Prausnitz, B.E. Poling, The Properties of Gases and Liquids, 4th ed., Mc Graw-Hill, Inc., New York, 1987.

[44] A.V. Kiselev, V.V. Khopina, Trans. Faraday Soc. 65 (1969) 1936.

pt

539

Elsevier, Amsterdam, 1977.

Ac ce

538

M

536

Page 18 of 31

19 542 543 544 Title: Adsorption of aqueous organic mixtures on a chiral stationary phase with bound antibiotic eremomycin

ip t

Authors: Yuliya K. Nikitina, Imran Ali, Leonid D. Asnin Highlights

pt

ed

M

an

us

cr

- Nautilus-E is a new chiral stationary phase with grafted antibiotic eremomycin. - Adsorption of typical hydro-organic mobile phases on this CSP was studied. - A model of the adsorbed layer of a hydro-organic solvent on this CSP is suggested - The adsorbed layer is thin for water-methanol and thick for water-acetonitrile.

Ac ce

545 546 547 548 549 550 551 552 553 554 555 556 557 558

Page 19 of 31

Ac

ce

pt

ed

M

an

us

cr

i

Figure 1

Page 20 of 31

Ac

ce

pt

ed

M

an

us

cr

i

Figure 2

Page 21 of 31

Ac

ce

pt

ed

M

an

us

cr

i

Figure 3

Page 22 of 31

Ac

ce

pt

ed

M

an

us

cr

i

Figure 4

Page 23 of 31

Ac

ce

pt

ed

M

an

us

cr

i

Figure 5a

Page 24 of 31

Ac

ce

pt

ed

M

an

us

cr

i

Figure 5b

Page 25 of 31

Ac

ce

pt

ed

M

an

us

cr

i

Figure 6a

Page 26 of 31

Ac

ce

pt

ed

M

an

us

cr

i

Figure 6b

Page 27 of 31

Ac

ce

pt

ed

M

an

us

cr

i

Figure 7a

Page 28 of 31

Ac

ce

pt

ed

M

an

us

cr

i

Figure 7b

Page 29 of 31

cr

ip t

Table 1

Minor

Minor

Isotopic

disturbance

disturbance

method

H2O-MeOH

H2O-MeCN

an

2.911

25

2.880

30

2.880

40

2.869

2.955

3.045

2.956

M

20

Retention of TtBB

80% MeOH 100% MeOH 80% MeCN 100% MeCN

2.545

2.510

2.061

2.582

5.4

5.9

11.9

5.0

2.954 2.956

d

T, oC

us

Table 1. Summary of the hold-up volume values (ml) found by different methods

ep te

τ*, Å

τ = [2.956 – V0TtBB ]/755, the apparent thickness of the column void volume inaccessible to the molecules of TtBB,

*

755 m2.

Ac c

determined assuming that the whole void volume is 2.956 ml as found by the isotopic method and the total surface area is

Page 30 of 31

Table 2

Table 2. Thickness of the adsorbed layer and adsorption equilibrium constant T, oC

τ, Å

K*

t

τ, Å

H2O-MeOH

H2O-MeCN

1.12

0.45

2.27

25

1.06

0.42

2.20

30

1.15

0.46

2.11

40

1.26

0.49

2.06

9.41

3.94

1.66

ip t

20

The weighted average value evaluated over the whole concentration range

Ac

ce pt

ed

M

an

us

cr

*

K*

t

Page 31 of 31