Introduction to Modern Liquid Chromatography, Third Edition part 73 pot

If the CMPA is very strongly retained, it may saturate the stationary phase, leading to a situation similar to that of a dynamically coated CSP-column note the similar situation for ion-

(a) Retention of enantiomer (R)-X

(b) Retention of enantiomer (S )-X

[(R)-X ]m + [(R)-CS ]m ([(R )-X ] -[(R )-CS ])m

[(R )-X ]s + [(R )-CS ]s ([(R )-X ] -[(R )-CS ])s

d,CS K d,(R )-X-(R )-CS

Mobile

phase

Stationary

phase

K a,(R)-X

j

[(S)-X ]m + [(R )-CS ]m ([(S)-X] -[(R)-CS])m

[(S)-X ]s + [(R )-CS ]s ([(S )-X ] -[(R )-CS ])s

Mobile

phase

Stationary

phase

K a,(S)-X

j

Figure14.8 Equilibria for the retention of R- and S- enantiomers (chiral–mobile-phase addi-tive, CMPA) mode Subscripts m and s refer to corresponding species in mobile and stationary phases; K a and K drepresent association and distribution constants, respectively

sta-tionary phases with distribution constants K d,(R)-X-(R)-CS and K d,(S)-X-(R)-CS

in the stationary phase (via processes i and j in Fig 14.8) The same equilibria have

to be considered for the S-enantiomer (S)-X The observed solute retention factor k

is then a weighted average of values of k for free and complexed X If the CMPA

is very strongly retained, it may saturate the stationary phase, leading to a situation similar to that of a dynamically coated CSP-column (note the similar situation for ion-pairing in Section 7.4.1 and Fig 7.12)

The consequences of Figure 14.8 can be summarized as follows: Without

the addition of the selector (R)-CS, no separation of the enantiomers (R)-X and

mobile phase, and if its interaction with the analyte is significant and

enantios-elective, then the retention k of the two enantiomers should differ—hopefully

leading to their chromatographic separation In addition to the association of the two solute enantiomers with the CMPA (with different retention of free and complexed solute), stereoselectivity may originate from different retention of the diastereomeric associates, as well as nonequal adsorbate formation via pathway

i or j Stereoselectivity contributions from these individual processes may either

enhance or attenuate each other As the mobile-phase concentration of the CMPA

is increased, there should be an increase in the separation of the two enantiomers However, at sufficiently high concentrations of the CMPA, a decrease in separation

is possible The reason is that a high enough selector concentration can complex

all of each enantiomers in the mobile phase, whereas differences in retention for the two enantiomers is favored when one is complexed to a greater extent than the other

From the description of Figure 14.8, it should be clear that the choice of CMPA and its concentration in the mobile phase are primary determinants of enantioselectivity and a successful separation of the two enantiomers However, other conditions can also play a role; for example, mobile-phase pH, ionic strength, different B-solvents (the organic solvent in RPC), and temperature can be important

and γ -cyclodextrins and their derivatives [18], quinine and quinidine [19, 20],

(+)- and (−)-10-camphorsulfonic acid [20], N-benzyloxycarbonyl-protected di- and tripeptides [19], chelating agents such as amino acids in combination with metal ions (adopting a chiral ligand-exchange chromatography approach) [21–23], and others The CMPA approach appears attractive because of its practical simplicity and relatively inexpensive columns However, this approach suffers from a number of drawbacks, several of which are similar to problems encountered when an ion-pair reagent is added to the mobile phase (Section 7.4.3):

• close control of the temperature necessary because of its possible effects

on the various equilibria of Figure 14.8, with a consequently less robust separation

• system peaks that result from differences in composition of the sample solvent and mobile phase (Section 7.4.3.1)

• incompatibility of some detectors with the presence of the CMPA in the mobile phase (e.g., UV-absorbing additives with UV detection; ion suppres-sion with MS)

• different response factors for the two enantiomers because they can exist partly as the diastereomer-complexes in the mobile phase leaving the column and flowing through the detection cell; see the similar discussion for the indirect method (Section 14.3)

• expense and limited availability of CMPA reagents for all chiral compounds Due to its numerous inherent drawbacks, today the CMPA mode has limited practical value for the HPLC separations of enantiomers However, it should be noted that the CMPA mode is firmly established as the method of choice for

enantiomer separations by capillary electrophoresis.

14.4.2 Chiral Stationary-Phase Mode (CSP)

The chiral stationary-phase mode (CSP) mode is generally the most straightforward and convenient means for chromatographic enantiomer separation; it is the method

of choice for both analytical and preparative applications The chiral selector is

preferentially covalently linked or alternatively strongly physically adsorbed (e.g.,

by coating of a polymeric selector) to a chromatographic support (usually porous silica particles) The mobile phase is achiral, that is, devoid of any chiral constituents During migration of the sample through the column, the individual enantiomers

“ideal fit”

Spacer Support

material

A

C B

a

d b

c

Selector (CS)

S-Enantiomer (S )-X

(R )

K i,S

[(R )-CS]s + [(S)-X]m ([(R )-CS ] -[(S )-X ])s

“non-ideal fit”

Spacer Support

material

A

C B

a

d b

c

Selector (CS)

R-Enantiomer (R )-X

(R )

K i,S

[(R )-CS]s + [(R)-X ]m ([(R )-CS ] -[(R)-X ])s

Figure14.9 Three-point interaction model and associated interactions between R- or S-

enan-tiomers and the CSP

are formed in the stationary phase If resulting values of k for the two enantiomers are sufficiently different, then their separation is possible Consequently successful separation requires a CSP that interacts more strongly with one enantiomer than the other The values of k are determined by the strengths of the resulting complexes, or

the following Section 14.4.3

CSPs and their corresponding enantioselective columns offer a number of striking advantages over both the indirect and additive approaches Solutes that lack appropriate functional groups for derivatization can still be separated by the CSP approach Minor enantiomeric impurities in the selector are only of concern

Other than this, none of the complications described above for the indirect method with an impure selector are present Problems from the presence of the selector

in the mobile phase leaving the column are also avoided with the CSP procedure Thus the selector is absent, which avoids any interference with detection The two enantiomers also possess an equal detector response—so enantiomeric composition

can be directly derived from the area ratio for the monitored R- and S-enantiomer

peaks—without any corrections However, with chiroptical detectors (Section 4.10)

a different response can be intentionally obtained for each enantiomers—with certain advantages (e.g., allowing assignment of the individual enantiomers as [+] and [−]) Disadvantages of the CSP approach include relatively expensive columns in most cases (and often with shorter lifetimes than RPC columns), moderate column

efficiencies (Section 14.5), and sometimes poor chemoselectivity for structural ana-logues (i.e., selectivity for enantiomers is often greater than for structural anaana-logues

that might also be present in the sample—leading to an overlap of the separated (R) and (S) enantiomer peaks by these analogues) It is important to note that enantiomer

separations using CSPs have expanded the preparative isolation of enantiomers from laboratory scale (mg to g) up to production scale (kg to ton) The attractiveness

of CSP separation lies in the short development times, ease of product recovery by the evaporation of volatile mobile phases, ready access to both enantiomers in high chemical and optical yield, and straightforward scalability (Section 15.1.2.1)

14.4.3 Principles of Chiral Recognition

Early attempts to rationalize chiral recognition at the molecular level have led to the

formulation of geometric models, such as the three-point attachment model [26].

The latter model is still frequently utilized to visualize and explain the requirements for enantioselectivity when designing CSPs In its original form this model states that at least three configuration-dependent attractive contact-points between a chiral receptor and a chiral substrate are required for chiral recognition However, a fourth essential requirement is often neglected, namely the fact that the receptor is accessible only from one side and can therefore only be approached in one direction The latter, simplistic model has been under debate since it was first formulated, and

today it is known that not all three interactions need to be attractive (three-point rule by Pirkle [27]) Its adaptation for chiral recognition by CSPs is graphically

illustrated in Figure 14.9

Considerable confusion with the three-point interaction model has arisen from the question: ‘‘what does ‘interaction’ mean?’’ [27] As noted above, both attractive and repulsive forces are included in ‘‘interaction,’’ which can either stabilize or destabilize the formation of the analyte-CSP complex Moreover many interactions are actually multipoint in nature, which minimizes the need for additional support-ive interactions For example, whereas hydrogen bonding and end-to-end dipole interactions are regarded as single-point interactions—therefore counting for only

multipoint interactions and may count as at least two interaction points each [27] Similarly molecules that contain chiral centers incorporated into rigid elements such

as a cyclic ring require fewer interactions [28] Two of the four bonds of the asymmetric centers are incorporated into a rigid ring, which enforces a molecular rigidity and makes the two stereoisomers more easily recognizable It is a common perception in the field that a single interaction with a rigid plane or its surface can count for at least two interaction sites

Consider next a chiral stationary phase that has a chiral selector CS, bonded

to the surface of a suitable support via a spacer (Fig 14.9) In this idealized model the analyte interacts solely with the selector, not with the spacer or support The

relative retention of the analyte enantiomers (R and S) are determined by their

strength of interaction with the selector shown in Figure 14.9, as influenced by the three-dimensional orientation and spatial arrangement of complementary sites in the binding partners Ideally there will be a perfect match between the chiral selector

and the respective enantiomer of the analyte (ideal fit) The preferred enantiomer

turns out to be the stronger-bound one [(S)-X in Fig 14.9], which is characterized

strongly retained and elutes as the second enantiomer peak For the other enantiomer

(R)-X, there is a considerable spatial mismatch—at least for one of the interaction sites Hence there is a non-ideal fit between (R)-X and CS, so that the corresponding

earlier Thermodynamic considerations for this selector-solute association and the adsorption process, respectively, are treated in more detail at the end of this chapter (see Section 14.7.)

An ideal fit and effective binding between analyte and selector can be achieved,

• if there is a size and shape complementarity, so that the analyte sterically fits the binding site of the selector—which is often arranged as a preformed

pocket or cleft, as in the case of cyclodextrin selectors (steric fit)

• if the analyte and selector have complementary interaction sites (functional groups) arranged in a favorable geometric and spatial orientation, so that

attractive noncovalent intermolecular interactions can become active (func-tional fit based on complementary interacting groups) These interactions

drive the association between analyte and selector enantiomers and are

basically electrostatic in nature; they comprise:

nega-tively charged groups)

(induction forces), and induced dipole–instantaneous dipole (dispersion forces) interactions

electron-rich and electron-poor aromatic groups)

• if hydrophobic regions of the selector and analyte are spatially matching so

as to enable binding by hydrophobic interactions (hydrophobic fit).

• if the flexibility of the analyte and selector allow an optimized binding in the

stationary phase (dynamic fit) In particular cases, an induced fit through a

conformational change of analyte and selector molecules upon complexation may further enhance the complex stability

Intuitively, chiral recognition might be thought to increase for higher binding affinities between analyte and selector (i.e., large binding constants or low dissocia-tion constants generally increase enantioselectivity) This concept has been known for a long time in pharmacology as Pfeiffer’s rule [29], which states that the stereos-electivity of drugs will increase with their potency The validity of this rule has been debated, and it is now accepted that for various reasons the principle is of limited applicability

Chiral recognition is commonly regarded as a bimolecular process of analyte-CSP interaction, and the effect of the mobile phase is frequently ignored The mobile

phase can in fact play a significant role in enantioselectivity by determining the degree of solvation of interactive sites of the analyte and selector, and whether these sites are then available for intermolecular contact between the analyte and CSP The choice of solvents, buffer salts, and pH are also major determinants

of conformational preferences and of the ionization states of both analytes and CSPs The mobile phase can also suppress detrimental, nonspecific interactions that deteriorate enantioselectivity (Section 14.7.3) Consequently a proper choice of mobile phase represents an important step in the development of both indirect, but especially direct enantiomer separations

For the reasons above, the mobile phase must be considered as an important factor in enantiomer separations with CSPs, as it defines the interaction environment where chiral recognition takes place Solvents can interfere or promote specific analyte–selector interactions and thus affect enantiorecognition Solvents of high polarity attenuate the strength of electrostatic interactions, whereas hydropho-bic interactions are present only in aqueous or hydro-organic mobile phases Mobile-phase pH is especially significant in the case of ionizable analytes or CSPs, as ionic interactions require ionized entities These ionic interactions can

be weakened by an increase in mobile-phase ionic strength, similar to the case of ion-exchange retention (Section 7.5.2) Many other factors influence the selection of the mobile phase, but these are similar for all liquid chromatographic procedures: reversed-phase, normal-phase, ion-exchange, and so forth For further details on mobile-phase selection, the reader is referred to Section 14.6 for individual CSPs, as well as relevant sections in earlier parts of the book

Compared to separations by RPC, the chromatographic efficiencies of enantioselec-tive columns in following Section 14.6 are often fairly low Plate numbers seldom

to a greater extent than in other forms of HPLC It can be assumed that these enantioselective column efficiencies are dependent on the same factors as for other HPLC columns (Section 2.4.1), so some additional factor must be involved, namely slow adsorption–desorption kinetics at the chiral sites It is commonly accepted that the desorption process can be slow because of the formation of analyte-selector complexes that are stabilized by simultaneous, multiple interactions Slow kinetics can be improved by the use of higher temperatures, but enantioselectivity often decreases with an increase in temperature Consequently the lower efficiency of many CSP columns must be accepted as a fundamental problem, with no obvious means of its correction

AND THEIR CHARACTERISTICS

Success in the use of enantioselective columns depends on the selection of a suitable CSP—one that is able to separate the target enantiomers It is therefore important

to know which chiral stationary phases are available, as well as their principal characteristics and typical operation conditions It can also be helpful to understand how they distinguish between enantiomers and what analytes they are most useful for These issues will be covered in the present section

A broad variety of chiral molecules, of both natural and synthetic origin, have been considered as prospects for useful CSPs A few hundred CSPs are today offered commercially, among which perhaps 20 to 30 CSPs are used most frequently—these columns are capable of separating most enantiomers likely to be presented for analysis Selectors for the latter columns fall into the following CSP classes:

Section 14.6.1)

• macromolecular selectors of synthetic origin (poly(meth)acrylamides, (poly-tartramides; Section 14.6.2)

• macromolecular selectors of natural origin (proteins; Section 14.6.3)

• macrocyclic oligomeric or intermediate-sized selectors (cyclodextrins, macro-cyclic antibiotics, chiral crown ethers; Section 14.6.4– 14.6)

• synthetic, neutral entities of low molecular weight (Pirkle-type phases, brush-type CSPs; Section 14.6.7)

• synthetic, ionic entities of low molecular weight that provide for ion exchange (Section 14.6.8)

• chelating selectors for chiral ligand-exchange chromatography (Section 14.6.9)

The following sections summarize the most important characteristics of the most commonly employed, commercially available CSPs and enantioselective columns These detailed insights into the way different CSPs achieve enantioselecitiv-ity can be helpful in an initial selection of promising CSPs and other conditions for chiral separation, followed by trial-and-error experimentation to achieve an

accept-able separation Specific recommendations for the use of certain, more favored CSPs and conditions are italicized and indented for easy access.

14.6.1 Polysaccharide-Based CSPs

Polysaccharide selectors have a long tradition in enantioselective liquid chromatog-raphy In 1973 Hesse and Hagel introduced microcrystalline cellulose triacetate (MCTA) as a polymeric selector material (without supporting matrix) for enan-tioselective liquid chromatography [30] While MCTA exhibits widely applicable enantiorecognition and favorable loading capacities for preparative separations, it suffers from poor pressure stability, slow separations, and low chromatographic efficiency A solution to the mechanical stability problem of MCTA was proposed

by Okamoto and coworkers in 1984 The cellulose derivatives were coated at about

20 wt% onto the surface of macroporous silica beads (100 or 400 nm pore size) [31] These materials exhibited considerably improved mechanical stability and much better efficiencies, and permitted HPLC enantiomer separations Such coated polysaccharide-based CSPs were state-of-the-art for several decades

In the following years a large variety of distinct polysaccharide derivatives,

O O

O

O R

O O R

O

O α β

n

Cellulose

O O

O

O R

O O R

O

On

Cellulose tris (4- methylbenzoate)

a

Cellulose tris (3,5- dimethylphenyl-carbamate)

b

CH3

CH3 N H

CH3

CH3

CH3

Amylose tris (3,5- dimethylphenyl-carbamate)

d

Cellulose tris (3,5- dichlorophenyl-carbamate)

c Cl

Cl

Polymer backbone Residue Name Tradename

N H

N H

a Chiracel OJ not available

b Chiracel OD Chiralpak IB

c not available Chiralpak IC

d Chiralpak AD Chiralpak IA

Restricted solvent range

Full solvent range

Figure14.10 Polysaccharide-based CSPs derived from cellulose and amylase, with corre-sponding column trade names

and coworkers and coated onto wide-pore silica (Fig 14.10) The three most ver-satile coated polysaccharide CSPs are commercially available from Daicel Chemical Industries, Ltd and Chiral Technologies:

• Chiralcel® OJ, based on cellulose tris(4-methylbenzoate)

• Chiralcel® OD, based on cellulose tris(3,5-dimethylphenylcarbamate)

• Chiralpak® AD, based on amylose tris(3,5-dimethylphenylcarbamate) Since the expiration of the patents covering these CSPs, generic materials have become available from a number of suppliers (e.g., Eka Chemicals, Regis, Macherey-Nagel, Phenomenex) under different tradenames (e.g Kromasil®, CelluCoat™, and ArmyCoat™, from Eka; Regis Cell from Regis; Lux Cellulose from Phenomenex) However, the overall performance of these generics often differ significantly from the original products in terms of chiral recognition capability, retention behavior, and column efficiency—because distinct supports and coating protocols are used for their fabrication (just as all C18 RPC columns are not equivalent)

Daicel and Chiral Technologies (as well as other suppliers) offer these CSPs

in normal-phase (NP; to be operated with alkane-alcohol eluents or similar elution conditions) and reversed-phase versions (RP; for use with hydro-organic mobile

phases; i.e., usually buffered aqueous-organic mixtures) These are presumed to differ in the type of support onto which the selectors are coated The user manual recommends that columns with extension R be used with RP conditions, while other columns should avoid aqueous mobile phases Whatever mobile phase is used, it is strongly recommended to dedicate the column to a specific mode and avoid switching between RP and NP conditions The extension H indicates a high-performance version based on 5-μm particles Most recently CSPs based on

corresponding RP-versions, Chiralpak AD-3R, Chiralcel OD-3R) to provide higher column efficiency and faster separations for high-throughput chiral analysis Because

of slow adsorption-desorption kinetics (Section 14.5), however, the advantage of smaller particles can be less than for achiral separation

The uniquely broad chiral-recognition capability of polysaccharide-type CSPs allows their use for a very broad range of sample types This feature originates from various molecular and supramolecular structural features peculiar to these semi-synthetic macromolecules The current understanding of chiral recognition principles for polysaccharide CSPs (largely derived from chromatographic experi-ments) is still at a rather immature state, despite their extensive use and importance

A limited number of studies have addressed the basis of polysaccharide molecu-lar recognition: solution NMR of oligomeric surrogates [32–34], solid-state NMR [35], computational studies [34–37], ATR-IR [35], thermodynamics [36, 38, 39], and quantitative structure-property relationship studies [39–42] However, these approaches have failed to provide much direction for the more effective use of polysaccharide CSPs

The glucopyranose chains in cellulose and amylose derivatives have been shown to form helices, with the helical twist being less pronounced for the cel-lulose derivatives as compared to amylose derivatives (left-handed 4/3 helical structure for amylose tris(3,5-dimethylphenylcarbamate) [33] Chiral recognition with polysaccharide-type CSPs may arise from three distinct features: (1) molecular chirality due to the presence of several stereogenic centers of the glucopyranose units, (2) conformational chirality due to the helical twist of the polymer backbone, and (3) supramolecular chirality from the alignment of adjacent polymer chains that form chiral cavities These features, further enhanced by derivatization of the polysaccharide, provide the exceptional and versatile stereodiscriminating abilities

of today’s polysaccharide-type selectors

From this general mechanistic picture it is clear that enantiorecognition for polysaccharide-type CSPs is defined not only by the type but also by the specific func-tionality of the respective cellulose and amylose derivative The effect of the aromatic substituents on chiral recognition for cellulose tris(phenylcarbamate) derivatives has been studied in detail by Okamoto [43] Chromatographic data were obtained for

a set of 18 CSPs coated with different mono- or di-substituted phenylcarbamate derivatives; it was found that inductive and steric effects are crucial for the over-all chiral recognition capacity Ortho derivatives, regardless of the nature of the substituent, performed poorly; para-substituted derivatives carrying methyl-, ethyl-, chloro- and trifluromethyl-groups produced improved enantiorecognition The best results were found with 3,4- and 3,5-dimethylphenyl- and dichlorophenylcarbamate derivatives, which have been subsequently selected as first choice for the preparation

of commercial CSPs For specific analytes, however, other derivatives may produce

still higher enantioselectivities For this reason, it is worthwhile to also include less common CSPs in the screening process—especially for preparative separations (see the extensive coated-polysaccharide-column lists of Daicel and Chiral Technologies)

As noted above, the supramolecular structure of a polysaccharide-type CSP has an enormous influence on its enantioselectivity In the early studies of Hesse and Hagel [30] with microcrystalline MCTA material, it was demonstrated that this polymer showed major changes in chiral recognition performance after dissolution and re-precipitation This observation was interpreted in terms of specific enantios-elective binding sites (microcrystalline domains) In similar fashion Francotte and Zhang [44] found that the supramolecular organization of polysaccharide deriva-tives also has a major impact on the enantioselectivity of coated CSPs Cellulose tris(4-methylbenzoate) coated from different solvents exhibited very different enan-tioselectivity, and—in some cases—inversion of enantiomer elution orders [44] X-ray diffraction experiments support the hypothesis that solvent-induced alterna-tions in the supramolecular structure account for this observation These findings underline the fact that polysaccharide-type selectors respond very sensitively to external stimuli such as solvents, temperature, and additives that can lead to confor-mational changes of the CSP—with altered binding processes, memory effects, and

so forth

The complexity of the chiral recognition processes for polysaccharide-type CSPs renders a rational approach to method development difficult Current strate-gies for chiral method development involve trial-and-error screening of various polysaccharide-type CSPs under multiple mobile-phase conditions—often using fully automated column- and solvent-switching A number of studies have focused on identifying efficient screening routines to maximize the chance for success [45–48]: the most promising CSP in the NP mode is

If serial instead of parallel screening is utilized, columns should be tested in this order [45] Interestingly, in the RP mode, Chiralcel OD-RH appears to be first choice rather than Chiralpak AD-RH From these extended screening studies it is also clear that polysaccharide CSPs have extremely broad applicability For example, Borman et al reported the results of a comprehensive screening campaign for a set of over 100 chemically diverse racemates, employing HPLC, SFC, and CE separation techniques and systems [49] With only three polysaccharide-type CSPs (Chiralpak

AD, Chiralcel OD, and Chiralcel OJ), more than 70% of the racemic analytes could be resolved It was also found that the individual polysaccharide derivatives often show complementary chiral recognition with respect to analyte structure; that

particularly true for the cellulose and amylose tris(3,5-dimethylphenyl) derivatives (Chiralcel OD and Chiralpak AD), which for a given analyte frequently display reversed elution of the enantiomers

Polysaccharide-based CSPs have traditionally been used as normal-phase pack-ings; that is, their preferential mobile phase is hexane or heptane with 2-propanol

or ethanol as strong solvents (B-solvents) [52] Note that 2-propanol tends to show

a different selectivity compared to ethanol for Chiralpak AD; elution can even be reversed by a change of these two solvents [53] Recently an independent solid-state