Polysaccharide-Based Chiral Chromatography Media in HPLC and SFC

Although the first enantiomeric separations were carried out during the early past century, it was not until 1973 that the first generally useful chiral stationary phase (CSP), cellulose triacetate, was developed. This phase was used primarily for preparative separations until the work of Professor Pirkle in the United States and Professor Okamoto in Japan made analytical CSPs readily available. Although the Pirkle phases provided good separations, the polysaccharide-based CSPs developed by Okamoto had very wide applicability.

It was subsequently found that the majority of enantiomeric separations could be achieved using just four different derivatives—two based on cellulose and two based on amylose. These polysaccharide phases were prepared by first derivatizing native amylose or cellulose and then coating these chiral polymers on a silica-gel base. This preparation method led to phase instability to certain organic solvents in which the polymers are partially or fully soluble.

Many methods have been developed to allow the immobilization or chemical bonding of the polysaccharide-based polymers to the silica support (1). Methods include cross-linking the polymer by use of diisocyanates (2), unsaturated groups that are subsequently polymerized (3, 4), silyl derivatives (5), direct bonding of the polysaccharide to silica by enzymatic or chemical routes before derivatization (6), and cross-linking of the chiral polymer by photochemical or free-radical reactions (7). All of these methods produce somewhat different products with different chromatographic properties.

Immobilization of the chiral polymer to the silica support has several benefits. The CSP is not destroyed on contact with a solvent that swells or dissolves (even partially) the polymer. The sensitivity of the first-generation CSPs to even small concentrations of such solvents often required the use of dedicated equipment to eliminate all chances of accidental introduction of a "forbidden" solvent.

Immobilization of the chiral polymer thus brings a new generation of polysaccharide-based CSPs into line with the stability of other high-performance liquid chromatography (HPLC) phases (such as C₁₈ and CN), which means a significant extension of the column lifetime and the elimination of sudden column failure because of an accident. Others benefits of the immobilization process are that otherwise difficult separations can be achieved by using solvents that previously could not be used with the coated phases and that new selectivity can be developed by using previously unusable chiral selectors.

Immobilized CSP

The 3,5-dimethylphenylcarbamates of amylose and cellulose are the most useful polysaccharide-based CSPs. They are commercialized as "Chiralpak IA" and "Chiralpak IB," respectively (Daicel Chemical Industries, Osaka, Japan) (see Figure 1). Other polysaccharide-based chiral selectors include the 3,5-dichlorophenylcarbamate of cellulose, which was first introduced by Okamoto as a coated CSP. Although it had limited utility because of its solubility in hexane-based mobile phases, it demonstrated sufficiently interesting selectivity when used with pure alcohol mobile phases to be of interest as an immobilized phase. Daicel introduced a CSP of similar structure, immobilized by a proprietary technology, as "Chiralpak IC" (see Figure 1).

Figure 1: Structures of the immobilized chiral stationary phases.

Selectivity and separations

Immobilizing otherwise unstable chiral selectors has great advantages, including the ability to separate enantiomers that cannot be resolved on existing CSPs. The unique structure of Chiralpak IC often allows separations of different selectivity from other polysaccharide-based CSPs. A combination of Chiralpak IC with Chiralpak IA and IB is replacing the older generation coated phases. The three immobilized columns are complementary, and a separation can sometimes be found on only one column. Figure 2 shows the example of prilocaine where the enantiomers are poorly separated with Chiralpak IA and IB but are well resolved with Chiralpak IC (8).

Figure 2: Separation of the enantiomers of prilocaine. Green is Chiralpak IA, blue is Chiralpak IB, and red is Chiralpak IC. Mobile phase is hexane–2-propanol 8:2; flow rate = 1 mL/min; column dimensions = 250 × 4.6 mm.

Immobilization of the chiral polymer increases solvent stability, which allows modification of the selectivity of chiral separations by the modulation of the solvent composition and its components. Figure 3 shows the exploitation of the mobile-phase constituents. Neither of the conventional mobile phases (hexane–2-propanol or hexane–ethanol) gave good resolution of the enantiomers of indoprofen using Chiralpak IC (8). Dichloromethane-based mobile phase improved separation in a significantly shorter time.

Figure 3: Separation of indoprofen enantiomers. Column was Chiralpak IC, 250 × 4.6 mm. All mobile phases contained 0.1% trifluoroacetic acid. Flow rate = 1 mL/min. Green is 30% ethanol in hexane; blue is 35% 2-propanol in hexane; and red is 2% methanol in dichloromethane.

Separation method development

The restrictions on solvents that applied to the early generation of coated polysaccharide CSPs made the development of separation methods using them relatively straightforward. One screened with four mobile phase mixtures—hexane:ethanol, hexane:2-propanol, acetonitrile, and 1:1 ethanol:methanol. Following this, the optimization also was simple because there were either "hits" or not, and the opportunity for modifying the separations achieved were limited. This limitation is an important advantage of these CSPs because the vast majority of separations can be found using these four solvents and only four CSPs—the tris-(3,5-dimethylphenylcarbamate) of cellulose and of amylose, cellulose tris(4-methylbenzoate), and amylose tris(S)-α-methylbenzylcarbamate.

Using immobilized CSPs removes the restrictions of solvents to the extent where any organic solvent with a reasonable viscosity is now accessible. This is a tremendous advantage in that it allows optimization of separations using solvent selectivity effects not available for methods using the coated CSPs. Nonetheless, this also may complicate the development of separation methods simply because of the bewildering array of solvents from which to choose.

To simplify the task, statistical studies have been conducted to assist in solvent selection and reduce the number that should be tried to a reasonable few while retaining maximum resolution and selectivity. Of course, such statistical guidelines must be based upon a wide range of applications to ensure their relevance, and there will always be some exceptions.

Two main sets of data, each based on the separation of a different set of 70 compounds, have been developed. Each set was screened using a wide range of solvents, and the success rates arising from each were determined.

Two measurements of success were used. The first was the number of separations with resolution >1.5 (i.e., baseline separations). The second was the number of separations for which at least a partial separation was attained. Although neither metric is perfect, they can be used to estimate the relative success of separation approaches. Table I shows the results from the first set of data using the three immobilized CSPs available (8). Solvents were the conventional hexane–ethanol and hexane–2-propanol mixtures together with mixtures based on dichloromethane (DCM), methyl t-butyl ether (MTBE) and tetrahydrofuran (THF). These latter solvents were used either pure or with an admixture of hexane and/or alcohols to adjust the solvent strength to allow elution in a reasonable time.

Although no one solvent system showed selectivity for all compounds, it is clear that a suitable combination of them can maximize the success rate (see Table I). Based on these data, one can make recommendations on a set of initial screening solvents that would give a high probability of finding separation conditions (see Table II).

Table I: Success rates using various screening mobile phases.*

In the second study, the separations were not optimized but were simply screened with a range of solvents. In this study, again with 70 solutes, 61.4% were baseline separated, and 94.3% of all chromatograms showed a selectivity greater than 1. Fourteen percent of the baseline separations required use of the extended range solvents (MTBE, DCM, THF etc.).

Table II: Proposed screening solvents*

Gradients

Chiral separations are generally simple, involving the separation of a pair of enantiomers. For this reason, elution gradients are rarely used for enantioselective chromatography. This approach is somewhat different from achiral separations where elution gradients are frequently used.

There are, however, some solvent strength issues, especially for some extended range solvents. Although for many compounds the solvent composition of Table II will suffice, some solutes will either not be retained or will not elute under these basic screening conditions. Thus, before starting the screening, one should establish a suitable solvent composition using several isocratic experiments involving various mixtures of solvents or with a gradient.

Figure 4: Overlay of gradient runs. Column: Chiralpak IC, 250 × 4.6 mm. Mobile phase: A = 30% dichloromethane in hexane; B = 10% methanol in dichloromethane. Both A and B contained 0.1 % diethylamine. Gradient: 0–100% B in 30 min; hold at 100% B 10 min. Flow rate = 1.5 mL/min. UV detection at 250 nm. Blue = Mianserin, Red = Methocarbamol, Green = Sulpiride.

Figure 4 shows an overlay of three gradient runs with three compounds, and Figure 5 shows the final isocratic separations. The gradient was chosen to run from the least-polar solvent mixture (hexane containing dichloro-methane) to the most polar (dichloro-methane containing methanol). A clear correlation exists between the elution time in the gradient and the solvent composition that allows elution in a reasonable time. Of course, this is not only applicable to methylene chloride as a mobile phase but also can be used with the other solvent systems. Such gradients are made possible only by the immobilization of the CSP.

Figure 5: Isocratic separations arising from the separations in Figure 3. Top is mianserin, mobile phase is 20% dichloromethane in hexane. Middle is methocarbamol, mobile phase is 1:1 hexane:dichloromethane containing 2% methanol. Bottom is sulpiride, mobile phase = 10% methanol in dichloromethane. Other conditions are the same as those for Figure 3.

Supercritical fluid chromatography

Supercritical fluid chromatography (SFC) is important in enantioselective separations, especially for the preparative purification and isolation of individual enantiomers. Mobile phases based on alcohols (methanol, 2-propanol) or acetonitrile have been used. Of these, the alcohol-based solvents have been the most useful. Acetonitrile is a much weaker solvent in SFC than it is in HPLC and is often used with added alcohol to improve its elution power. The major problem with existing solvents in preparative separations is that many compounds have limited solubility in alcohols and are better soluble in solvents of intermediate polarity such as chloroform or THF. The sample in this case may precipitate when it is mixed with the mobile phase, thus blocking the inlet frit of the column. Under certain conditions, some coated polysaccharide stationary phases may be stable to small amounts of, for example, chlorinated solvents in CO₂. But problems arise if the column depressurizes during operation. The immobilized chiral selectors are not affected by these considerations and allow any organic solvent as SFC cosolvent, thus eliminating solubility problems.

Selectivity in SFC follows much the same trends as in HPLC, although as noted, the lower solvent strength of aprotic solvents requires the addition of alcohols for the more-polar solutes, not only to ensure rapid elution but also obtain good peak shapes. Figure 6 illustrates the effect on peak shape and retention of 1,2,3,4-tetrahydro-1-naphthol by the addition of 5% methanol to the THF cosolvent. Compounds containing polar groups appear to tail in aprotic solvents. Addition of methanol (or other alcohol) removes this effect.

Figure 6: Influence of methanol additive in SFC separation of 1,2,3,4-tetrahydro-1-naphthol enantiomers. Column is Chiralpak IA, 250 × 4.6 mm. Mobile phase for (a) is 25% THF in CO2 and for (b) is 25% of 5% methanol in THF in CO2. Flow rate = 4 mL/min, back pressure = 150 bar, and temperature = 35°C.

Method development in SFC

Separation method development in SFC usually is carried out by investigating the selectivity derived from alcohol–CO₂ mixtures. This task may be modified by using suitable additives such as ethanesulphonic acid (9), trifluoroacetic acid, or diethylamine (see Figure 7). Sulfinpyrazone is not eluted from the Chiralpak IC column with CO₂–2-propanol as a mobile phase. Addition of 0.5% ethanesulfonic acid to the cosolvent allows elution.

Figure 7: SFC separation of sulfinpyrazone enantiomers. Column is Chiralpak IC, 250 × 4.6 mm. Mobile phase is CO2 containing 20% of 0.5% ethanesulfonic acid in 2-propanol. Flow rate = 5 mL/min. Back pressure = 100 bar, Temperature = 35°C.

Once the alcohols have been investigated, other solvents may be tried. The most useful to date are THF and methyl acetate, both with an admixture of methanol to adjust the solvent strength. This approach gave at least partial separations for 92% of a 38 sample set. This work continues with the study of additional cosolvents, particularly MTBE and the chlorinated solvents.

Preparative chromatography

An important goal in the industry is to bring products quickly to clinical trial. Enantioselective chromatography is used extensively in this activity because it is usually much faster to develop a chromatographic separation that can be applicable to the quantities required for discovery, toxicology, and early Phase 1 testing than it is to develop a crystallization, an enzymatic resolution or, especially, an enantiospecific chemical route. The most accepted stationary phases for this activity are the polysaccharide-based CSPs because they combine the widest range of selectivity with good loading characteristics (10).

The biggest challenge in the development of a preparative separation is to find suitable solubility. Many compounds show very little solubility in hexane-containing mobile phases and even where polar solvents such as methanol or acetonitrile can be used, the solubility of many products remains low. Thus, a major need for preparative chromatography is access to the solvents in which most samples are well soluble. Several separations involved adding chloroform or methylene chloride to the sample, and sometimes the mobile phase, to enhance sample solubility. Such operations are fraught with peril because the coated polysaccharide CSPs can be damaged very easily by such mobile phases. With immobilized CSPs, this problem disappears, and in a number of separations with HPLC and SFC, chlorocarbon-based mobile phases were used with great success.

Studying the effects of solubility, however, we found other influences of the mobile phase on preparative separations. The effects of the nature of the mobile phase on the separation of α-methyl-α-phenylsuccinimide enantiomers were studied with a view to establishing the highest production rate for HPLC and SMB separations. Screening against a number of possible mobile phases gave the solubility, retention, and selectivity shown in Table III, which shows considerable variations in the values among the solvents.

Table III: Solubity, retention, selectivity, and productivity

Loading studies, which involve the injection of a range of sample loads to evaluate the isotherms of adsorption of the solutes, demonstrated major differences among the solvents. The adsorption behavior of the α-methyl-α-phenylsuccinimide enantiomers depended considerably on the nature of the mobile phase. In some cases (chloroform and acetonitrile/2-propanol), the separation collapsed rapidly on loading with the second peak merging with the first at relatively low load. In the case of MTBE, the second component peak moved to longer retention time at moderate load and only began to merge with the first eluting component at high loadings. The high loading capacity in this system led to high productivity for the separations, both in HPLC and SMB mode. The productivity of the separations (in kilograms of enantiomer produced per kilogram of CSP per day) is shown in Table III for each of the mobile phases.

Separation in ethyl acetate also had a high loading capacity, which, combined with the high solubility in this solvent, resulted in the highest production rate for both HPLC and SMB. Thus, solubility is only part of the answer for preparative separations. The effect of the solvent on the loading capacity, presumably through its interactions with the stationary phase, is of great importance in preparative separations. This result has been observed only by virtue of the solvent stability of the immobilized polysaccharide-based CSPs.

Conclusion

The new generation of CSPs, based on immobilized polysaccharides, has proven to be versatile in development of enantioselective separation methods, requiring relatively few stationary and mobile phases for success. The range of selectivity of the CSPs is extended by using solvents hitherto forbidden for the older generation of coated polysaccharide stationary phases not only for HPLC separations but also for SFC methods. Furthermore, the nature of the solvents in preparative separations may have a profound influence on both the solubility of the samples and on the loading capacity of the preparative columns. The extended range of solvents now accessible with the availability of immobilized CSPs extends the potential of preparative enantioselective chromatography to new and larger-scale separations.

Geoffrey B. Cox is vice-president of technology at Chiral Technologies, 800 N. Five Points Rd., West Chester, PA 19380, tel. 610.594.2100, fax 610.594.2325, gcox@chiraltech.com.

References

1. T. Zhang and P. Franco, "Analytical and Preparative Potential of Immobilized Polysaccharide-Derived Chiral Stationary Phases" in Chiral Separation Techniques, G. Subramanian, Ed. (Wiley-VCH, Weinheim, 2007).

2. Y. Okamoto et al., ""Chiral Stationary Phases for HPLC: Cellulose Tris (3,5-dimethyl-phenylcarbamate) and Tris (3,5-dichlorophenylcarbamate) Chemically Bonded to Silica Gel," J. Liq. Chromatogr. 10, 1613 (1987) .

3. L. Oliveros, C. Minguillon, and P. Lopez, "Polysaccharide Derivatives and their use for Preparing Stationary Chiral Phases Suitable for Separating Isomers," French Patent, 2714671 (1994).

4. T. Kubota et al., "Cellulose 3,5-Dimethyl-phenylcarbamate Immobilized onto Silica Gel via Copolymerization with a Vinyl Monomer and Its Chiral Recognition Ability as a Chiral Stationary Phase for HPLC," Chem. Lett. 724 (2001).

5. R. Duval, "Polysaccharides and Oligosaccharides Crosslinked with Bissilane-, Bisthioether-, Bissulphoxyde-, Bissulphone- and Butanediyl-Derivatives and Their Shaping as Substrate Materials," French Patent 2784108 (2000).

6. N. Enomoto et al., "Preparation of Silica Gel-Bonded Amylose through Enzyme-Catalyzed Polymerization and Chiral Recognition Ability of Its Phenylcarbamate Derivative in HPLC," Anal. Chem. 68 2798 (1996).

7. E. Francotte and T. Zhang, "Photochemically Cross-Linked Polysaccharide Derivatives Having No Photopolymerisable Functional Groups," PCT International Patent Application, WO 97/04011 (1997).

8. T. Zhang, D. Nguyen, and P. Franco, "Use of Immobilized Polysaccahride-Based Chiral Stationary Phases in Enantiomeric Separation," presented at Enantiosep '07, Ferrara, April 2007.

9. R.W. Stringham, "Chiral Separation of Amines in Subcritical Fluid Chromatography using Polysaccharide Stationary Phases and Acidic Additives," J Chromatogr. A, 1070, 163 (2005).

10. E Francotte, "Chiral Stationary Phases for Preparative Enantioselective Chromatography" in Preparative Enantioselective Chromatography, G B Cox, Ed. (Blackwell, Oxford, UK, 2005).