Chiral drugs:analysis and preparative separation

SummaryTen key points

The following ten points are a condensed version of the whole document, to help judge in advance how relevant it is to the project in hand. The corresponding section is given in brackets after each.

No.How this document is organised and how to read it

The document is organised in the order concept — mechanism — product — method — scale-up. The first two parts explain what chirality is and why drug development places separate requirements on it; Part Three gives the four engineering routes to a single enantiomer and where the chromatographic route sits among them; Part Four develops the recognition mechanism of chiral chromatography, which is the basis for the selection discussed later; Parts Five and Six introduce the technical characteristics and applicable limits of COSMOSIL CHiRAL chiral columns and ChiralONE chiral media respectively; Parts Seven and Eight are directly executable method development and scale-up workflows; Part Nine is a quick reference.

Suggested reading paths: analysts may start at Parts Four and Seven; process development staff at Parts Three and Eight; project management and regulatory staff may read only Parts Two and Nine. Content marked "our inference" indicates that no published statistical source was found and that it serves only as experience-based reference.

Chirality is a purely geometric concept, but it determines the design logic of every separation method that follows. This part sets out the concept and the criteria first, before moving to drugs and chromatography.

1.1Definition of chirality and the criteria

An object that cannot be superimposed on its mirror image by translation and rotation is said to possessChiral(chirality). The left and right hand are the usual analogy: the two hands are mirror images with the fingers in the same correspondence, yet no rotation will superimpose them. The same applies at the molecular level, and a pair of mirror-image, non-superimposable molecules is called a pair ofEnantiomers（enantiomers）。

The criteria can be given in two directions.The positive criterionis to look for a chiral unit in the molecule, most commonly a carbon atom bonded to four different groups, called a chiral centre or stereocentre.The negative criterionis more reliable: if the molecule possesses animproper symmetry element(Sₙ, including a mirror plane σ = S₁ and a centre of symmetry i = S₂), then the molecule is achiral. In practice, looking for a mirror plane first is the more efficient approach — a molecule containing a mirror plane is not chiral.

Fig. 1-1The mirror relationship of a chiral carbon.

The central carbon carries four different groups a/b/c/d, with the dashed bond pointing behind the plane of the page. The two structures are mirror images and cannot be superimposed by rigid rotation.

1.2Four geometric origins of chirality

(S)-lactic acid · central chirality

Four different groups on the same carbon atom — the commonest chiral unit. The physical constants of the two enantiomers are identical, and they differ only in a chiral environment.

(R)-1,1′-bi-2-naphthol · axial chirality

No chiral carbon. The two naphthalene rings cannot rotate freely because of the steric bulk of the ortho substituents, and the chirality comes from the biaryl axis. The configurational stability of such atropisomers falls as temperature rises.

4-carboxy[2.2]paracyclophane · planar chirality

The chirality comes from the position of the substituent relative to the molecular plane. The bridging rings restrict the relative motion of the two benzene rings, so the mirror image cannot be superimposed by rotation.

[6]helicene · helical chirality

Six benzene rings fused ortho form a helix, and the overlap of the two terminal rings gives left- and right-handed configurations. The more rings there are, the higher the barrier to racemisation.

Representative structures for the four geometric origins. Central chirality is the commonest; axial, planar and helical chirality do not depend on a chiral carbon, and the criterion remains "not superimposable on its mirror image".

A chiral centre is not the origin of chirality. Drug molecules include several structural types that are chiral without containing a stereocentre, and their chromatographic behaviour differs appreciably.

1.3Naming and labelling systems: R/S, D/L, (+)/(−)

D-glyceraldehyde · the reference for the D/L label

The D/L system uses glyceraldehyde as its reference and labels according to whether the hydroxyl lies left or right in the Fischer projection; it cannot be derived from, or used to derive, R/S or the direction of optical rotation.

L-alanine · the configuration of protein amino acids

Almost all the amino acids making up proteins are L. The chiral uniformity of living systems is the structural premise for enantiomers behaving differently in vivo.

D/L and R/S belong to two separate systems, and (+)/(−) is the measured direction of optical rotation; none can be derived from another.

The three systems are used in parallel, each with its own meaning, and none can be derived from another.

• R/S (the CIP rules)— describes the absolute configuration. The four substituents are ranked by atomic number (a > b > c > d), the lowest priority d is pointed away from the observer, and if a→b→c runs clockwise the centre is R (rectus), anticlockwise S (sinister). This is the main labelling used in regulatory documents and pharmacopoeias today.
• D/L (the Fischer system)— a relative configuration label referenced to glyceraldehyde, used mainly for amino acids and sugars. Natural protein amino acids are mostly L; natural ribose and glucose are mostly D.
• (+)/(−), or d/l— describes the direction of optical rotation, an experimentally measured physical property with no fixed correspondence to R/S. A compound of a given configuration may rotate in different directions at different wavelengths or in different solvents.

1.4A map of stereoisomeric relationships

Putting the several stereoisomeric relationships on one diagram helps in deciding whether a given impurity should be controlled by a chiral method or an ordinary one.

Fig. 1-2Stereoisomeric relationships and the corresponding separation methods.

This diagram gives the first branch point in method selection: a chiral method is needed only if the target impurity and the main component are enantiomers.

1.5Where the property differences between enantiomers appear

In anachiral environmenta pair of enantiomers have completely identical scalar physical properties: molecular weight, melting point, boiling point, density, refractive index, solubility in an achiral solvent, and retention time on an ordinary C18 column. This is the source of every difficulty in chiral separation — they cannot be separated by "polarity difference" in the usual sense.

Differences appear only once achiral environmentis introduced, and include: interaction with plane-polarised light (optical rotation, circular dichroism); salt or bond formation with a chiral reagent (giving diastereomers, whose properties then differ); and binding to a chiral receptor or chiral stationary phase (forming transient complexes of unequal energy). Chiral chromatography exploits the third of these.

1.6Quantifying chiral purity: e.e., e.r., d.e.

Enantiomeric excess (e.e.) is the most widely used measure today:

e.e. (%) = |R − S| / (R + S) × 100

where R and S are the amounts of the two enantiomers (in chromatography peak areas are usually substituted, on the premise that the response factors of the two enantiomers are the same — for UV detection this premise generally holds, since enantiomers have identical UV absorption)

Enantiomers are chemically all but indistinguishable, yet may behave entirely differently in a living body. This part explains where that contrast comes from and gives the corresponding regulatory requirements.

2.1The chiral uniformity of living systems

The basic units of living systems are themselves chiral and occur in nature in a single configuration: proteins are built from L-α-amino acids (glycine excepted), the sugar rings of nucleic acids are D-ribose and D-deoxyribose, and most natural polysaccharides are built from D-glucose. As a result the enzyme active site formed by protein folding, the receptor binding cavity and the transporter channel are allchiral microenvironments。

A drug molecule entering such an environment is like one hand grasping another, fixed hand. A left hand and a right hand inserted into the same glove fit differently — the most direct explanation of pharmacological differences between enantiomers.

2.2The three-point interaction model

The three-point interaction model proposed by Easson and Stedman in 1933 is the basic framework for understanding chiral recognition, and applies equally to biological receptors and to chromatographic stationary phases.

The model requires at leastthree simultaneous interactionsbetween the chiral recognition site and the molecule being recognised, at least one of which is stereochemically dependent. With only two points of interaction both enantiomers can fit equally well and cannot be distinguished.

Fig. 2-1The three-point interaction model.

The complexes formed by the two enantiomers with the same chiral surface differ in free energy (ΔΔG), which appears chromatographically as a difference in retention factor k, that is the selectivity factor α = k₂ / k₁.

2.3Four cases of enantiomer pharmacology

Grouping the pharmacological relationship between enantiomers into four classes maps directly onto development decisions.

2.4Stereoselective ADME and chiral inversion in vivo

Differences between enantiomers are not confined to target binding. Absorption, distribution, metabolism and excretion are all mediated by chiral proteins, so the pharmacokinetic parameters of the two enantiomers may differ:

• Metabolism— CYP isoforms are selective for configuration, so the clearance and metabolite profile of the two enantiomers may differ; the S and R forms of an anticoagulant being metabolised by different CYP pathways is the classic example.
• Plasma protein binding— albumin and α₁-acid glycoprotein are themselves chiral recognition entities, so the free fraction can differ with configuration.
• Transport— efflux transporters such as P-gp have configurational preferences.
• Chiral inversion in vivo— the R form of some 2-arylpropionic acids can be converted enzymatically and unidirectionally to the S form in vivo. Even when such a compound is given as a single enantiomer, exposure in vivo is to a mixture of both, which must be addressed in clinical pharmacology.

2.5The thalidomide case and what it means technically

(S)-ibuprofen · the 2-arylpropionic acids

This class undergoes unidirectional R→S conversion in vivo via an acyl-CoA intermediate. Pharmacokinetic studies must measure the plasma concentration of each enantiomer separately and cannot report only the total.

(S)-thalidomide

The two enantiomers behave differently pharmacologically and interconvert under physiological conditions. The technical conclusion drawn is separate identification and evaluation, not simply administering a single enantiomer.

Esomeprazole · chirality at sulfur

The stereocentre is the sulfoxide sulfur atom. Moving from a racemic drug to a marketed single enantiomer is the typical form of a chiral switch.

Three typical cases: unidirectional conversion in vivo; enantiomers that behave differently and can interconvert; and a racemate relaunched as a single enantiomer.

Thalidomide was marketed in the late 1950s as a sedative and antiemetic, and was later withdrawn after being linked to limb malformations in newborns. Subsequent research showed that its two enantiomers behave differently biologically, with the (S) configuration associated with teratogenicity.

It must be said at the same time that the stereocentre of thalidomide undergoesrapid racemisationat physiological pH, so that even when a single enantiomer is administered a mixture of both forms rapidly in vivo. The technical conclusion from the case is therefore not "giving a single enantiomer avoids the risk" but thatenantiomers must be identified and evaluated separately— including their configurational stability. This understanding drove the establishment of the regulatory framework that followed.

2.6The regulatory framework and control requirements for chiral impurities

2.7The product logic of the chiral switch

A marketed racemic drug may be redeveloped as a new product if the single enantiomer can be shown to offer an advantage in efficacy or safety; this strategy is called a chiral switch. It has three technical prerequisites: the two enantiomers do behave differently pharmacologically; the single enantiomer can be obtained reliably at industrial scale; and the single enantiomer does not racemise rapidly in vivo.

For separation technology, a chiral switch means demand expands from "analytical-scale e.e. determination" to "kilogram- to tonne-scale preparation of a single enantiomer" — which is the subject of Part Eight.

Chromatographic resolution is not the only route, but it has a defined place in terms of response speed and range of application at the development stage. This part compares the four routes on the same axes.

3.1The chiral pool route

Starting from a single-configuration compound already available in nature, the target molecule is built by reactions that do not destroy the stereocentre. Common chiral pool starting materials include L-amino acids, D-sugars, terpenes, hydroxy acids (L-tartaric and L-malic acid, for example) and alkaloids.

The advantages of this route are high starting chiral purity, controllable cost and no resolution step; the constraints are that the skeleton of the target molecule must match structurally what is available as a chiral starting material, and that every step of the synthesis must avoid racemising the stereocentre.

3.2Asymmetric synthesis and asymmetric catalysis

The configuration of the new stereocentre is controlled directly in the bond-forming step. Three classes can be distinguished by where the chiral information comes from:

• A chiral auxiliary— a chiral group is attached temporarily to the substrate and removed after the reaction. Stereocontrol is reliable, but two extra steps of attachment and removal are needed and atom economy is lower.
• A chiral reagent— a chiral reducing or oxidising agent used stoichiometrically.
• Asymmetric catalysis— a chiral catalyst transfers chiral information at substoichiometric loading. Metal complex catalysis (asymmetric hydrogenation, epoxidation, dihydroxylation), small-molecule organocatalysis and biocatalysis (enzymes, engineered enzymes) all belong here. The Nobel Prizes in Chemistry of 2001 and 2021 were awarded to asymmetric catalytic hydrogenation/oxidation and to asymmetric organocatalysis respectively, reflecting the standing of this direction in industrial synthesis.

In a mature process, asymmetric catalysis is optimal in atom economy and throughput, but catalyst screening and process development take longer and depend strongly on substrate structure.

3.3Crystallisation resolution and kinetic resolution

3.4Chromatographic resolution

The racemate is separated directly on a chiral stationary phase (CSP), or converted first by chiral derivatization into diastereomers and separated on an ordinary column. Its characteristics are:

• The route is independent of the synthesis— the existing synthetic route is unchanged, and the step can be inserted at any intermediate or at the final product.
• Short development time— for most compounds the column and mobile phase can be screened within days to give a usable method.
• Both enantiomers are obtained— the minor enantiomer can be kept as an impurity reference standard, or racemised and recycled.
• Wide range of application— there is no requirement for the substrate to contain a salt-forming group or a reactive functional group.

The cost lies in solvent consumption and stationary phase cost at the preparative stage. Part Eight gives a quantitative treatment of both.

3.5Matching the route to the development stage

Fig. 3-1Decision relationships for obtaining a single enantiomer.

Development stage and quantity required are the main branching variables; whichever route is chosen, a chiral analytical method is prerequisite work.

Chiral column selection, mobile phase design and temperature setting can all be derived from the recognition mechanism. This part is the basis for the two product chapters that follow.

4.1The molecular basis of chiral recognition

The types of interaction between a chiral stationary phase and the molecule being separated are no different from those in ordinary chromatography; the difference is that they take place in achiral environment of fixed spatial orientationand are therefore sensitive to configuration. The common types are:

The net result of chiral recognition is a free energy difference ΔΔG between the transient complexes the two enantiomers form with the stationary phase. This difference is usually of the order of 0.2–2 kJ/mol, corresponding to α = 1.1–2.2. Robustness studies on a chiral method are therefore more important than on an ordinary one.

4.2Seven classes of chiral stationary phase

The molecular structures on this page are rendered from SMILES by RDKit, with molecular formula and ring count verified by assertion one by one; the structures show the skeleton and key functional groups, and the original source governs the stereochemistry.

4.3The structural origin of polysaccharide stationary phases

β-D-glucopyranose unit · underivatised

The backbone unit of a polysaccharide selector. Underivatised cellulose and amylose have limited chiral recognition ability, and the 2-, 3- and 6-hydroxyls must be converted to aryl carbamates.

Cellulose tris(3,5-dimethylphenylcarbamate) unit

The selector of CHiRAL B. Cellulose is β-1,4 linked with a ribbon-like chain; the N—H and C=O of the carbamate provide hydrogen bonding sites, and the methyl groups tune the steric environment of the groove.

Amylose tris(3,5-dimethylphenylcarbamate) unit

The selector of CHiRAL A. The substituents are the same as above; the difference is that α-1,4 linkage makes the chain helical, so the chiral groove has a different geometry and the selectivity is complementary.

Cellulose tris(3,5-dichlorophenylcarbamate) unit

The selector of CHiRAL C. Chlorine substitution raises the polarity and dipole interaction of the carbamate carbonyl, giving recognition of basic compounds and nitrogen heterocycles different from the other two.

The repeating unit of each selector is shown as the methyl glycoside: in the real material the 4-hydroxyl forms the glycosidic bond of the polymer chain, and the free 4-OH in the figure marks where the chain is cut. The three differ only in linkage (β-1,4 / α-1,4) and ring substituent (methyl / chloro), and that is where their complementary selectivity comes from.

Cellulose and amylose are themselves homopolymers of D-glucose, differing in the configuration of the glycosidic bond: cellulose is β-1,4 linked with a relatively extended, ribbon-like chain, while amylose is α-1,4 linked and the chain tends to coil into a helix. Derivatising the three hydroxyls on the sugar unit as phenylcarbamates produces two effects:

• a large number of uniformly oriented N–H / C=O hydrogen bonding sitesandaromatic rings；
• the derivatised polysaccharide chains form a regular higher-order structure during immobilisation, and between and within chains they createchiral grooves。

Once the molecule being separated enters the groove, the spatial relationship of its substituents to the hydrogen bonding sites and aromatic rings on the groove wall depends on configuration — this is the main source of chiral recognition on polysaccharide stationary phases. It is also why the position and electronic character of the ring substituents (methyl, chloro, mixed substitution) change the shape and electron density of the groove and give different selectivity profiles.

Fig. 4-1The structural hierarchy of a polysaccharide chiral stationary phase.

From silica substrate through the bonded layer and polysaccharide helix to the chiral groove, selectivity arises mainly in the outermost layer.

4.4Differences between coated and immobilised types

There are two ways of loading a polysaccharide derivative onto silica, and this difference determines the range of usable solvents — one of the technical attributes with the greatest influence on chiral column selection.

• Coated type— the selector is attached to the silica surface by physical adsorption, with no chemical bond. Preparation is relatively simple and the natural higher-order structure of the selector is better preserved; but the selector can be dissolved by strong solvents and washed off the support.
• Immobilised type (that is, bonded)— a chemical bond is formed between the selector and the silica. The selector cannot be washed off by solvent, and the range of usable solvents widens markedly.

Fig. 4-2The difference between coated and immobilised stationary phases under strong solvent conditions.

The COSMOSIL CHiRAL series uses an immobilised (chemically bonded) selector; see the measured data in section 5.2.

4.5Four elution modes

Fig. 4-3Where the four elution modes sit.

The same chiral column does not give the same selectivity in different modes, and screening one round in each mode during method development is standard practice.

4.6How solvent affects selectivity

In ordinary reversed phase chromatography, changing the proportion of organic phase mainly affects retention (k), with limited change in selectivity (α). Chiral chromatography is different:the solvent takes part in the chiral recognition process itself. Alcohol modifiers compete for the hydrogen bonding sites on the carbamate, and different alcohols (methanol, ethanol, isopropanol) differ in steric bulk and hydrogen bonding ability, changing the solvation state inside the groove and hence α — sometimes even reversing the elution order of the two enantiomers.

In chiral method development, therefore,changing the type of modifier takes priority over adjusting its proportion. It is not unusual for a separation with α = 1.05 in hexane/isopropanol to rise to α = 1.3 in hexane/ethanol.

4.7Temperature effects and reversal of elution order

The relationship between retention and temperature is described by the van 't Hoff equation:

ln k = −ΔH° / (RT) + ΔS° / R + ln φ

Taking the difference for a pair of enantiomers gives ln α = −ΔΔH° / (RT) + ΔΔS° / R. This means:

• When the separation isenthalpy dominated(ΔΔH° is the main contribution), lowering the column temperature raises α. This is the common situation on most polysaccharide stationary phases, which is why chiral methods often run at 15–25 °C.
• When ΔΔH° and ΔΔS° have the same sign and offset one another, there is anisoelution temperature T_iso = ΔΔH° / ΔΔS°. At that temperature α = 1 and the two peaks coincide; on crossing it,the elution order reverses。

4.8Reading the resolution equation in chiral separation

The basic relationship for resolution is:

Rs = (√N / 4) · [(α − 1) / α] · [k / (1 + k)]

The three factors differ greatly in how far they can be manipulated in a chiral separation:

The conclusion is clear: the main effort in chiral method development should go into raising α — that is, changing column and changing solvent — rather than lengthening the column or lowering the flow rate.

The parameters in this part are taken from Nacalai Tesque published product literature (the COSMOSIL CHiRAL Series product page and product handbook). Microwants is the exclusive agent for this series in mainland China.

5.1Range composition and technical parameters

The COSMOSIL CHiRAL series uses silica as the substrate withchemically bonded (immobilised)polysaccharide derivatives as the chiral selector, in three selectors and two particle sizes.

5.2Solvent tolerance of an immobilised selector

A comparison published by the manufacturer: using trans-stilbene oxide as the probe on a 2.6 mm I.D. × 250 mm column with hexane/isopropanol = 90/10 at 1.0 mL/min, 30 °C and UV 270 nm, tetrahydrofuran (THF) was injected repeatedly.

• A coated selector: THF strips the selector from the stationary phase and the plate count falls below one fifth of its original value.
• An immobilised selector (COSMOSIL CHiRAL): withstands repeated injection of THF.

This difference sets the working boundary of the column directly: a sample solution containing THF or dichloromethane can be injected as it is, with no solvent exchange first. In impurity preparation and direct analysis of reaction mixtures, this saves a sample preparation step.

5.3The usable solvent range and the selectivity space

* Methanol and acetonitrile are immiscible with hexane and should not be combined into the same mobile phase.
Order of eluting strength (normal phase): alcohols, THF ≫ chloroform > MTBE ≫ alkanes.

5.4Efficiency and backpressure of the 3 µm particle size

Flavanone

One of the samples used in the 3 µm peak shape comparison. C2 is the stereocentre of the molecule, detected at 254 nm in hexane/isopropanol 90/10.

Ketoprofen

One of the samples used in the 3 µm peak shape comparison. It contains a carboxyl group, so about 0.1% trifluoroacetic acid must be added to the mobile phase.

COSMOSIL CHiRAL offers 3 µm analytical columns. Relative to a conventional 5 µm, the gains and costs are as follows:

• Efficiency— peak width is reduced; the published literature shows the improvement in peak shape of 3 µm over 5 µm with comparison chromatograms of flavanone and ketoprofen on a 4.6 mm I.D. × 250 mm column. On the R_s ∝ √N relationship, reducing particle size from 5 µm to 3 µm raises N by about 1.67 times in theory and R_s by about 1.29 times — and in a borderline separation where α is close to 1.05–1.10, that increment often decides whether the method is usable.
• Backpressure— on a 4.6 mm I.D. × 250 mm column at 1.0 mL/min and 30 °C in a hexane/isopropanol system: backpressure is 5.3 MPa at 5% (v/v) isopropanol, and does not exceed 12 MPa when isopropanol is raised to 50% (v/v). This range places no pressure constraint on an ordinary HPLC system, so the mobile phase composition can be adjusted over a wide range.

5.5Screening hit rates of the three selectors

trans-Stilbene oxide

The sample used in the solvent tolerance measurement and in the parallel SFC conditions. The epoxide ring is rigid and the polarity moderate, and it separates completely on all three selectors, making it a good benchmark for efficiency and selectivity.

2-(2,4-Dichlorophenoxy)propionic acid

An acidic sample. About 0.1% trifluoroacetic acid is added to the normal phase mobile phase to suppress dissociation; it separates completely on CHiRAL A and C, and only partially on B.

1-(1-Naphthyl)ethanol

A neutral sample with a large conjugated aromatic ring. π–π interaction with the aryl groups of the selector is one of the main contributions to retention.

Nicardipine

A basic sample with a dihydropyridine core. About 0.1% diethylamine is added to the normal phase mobile phase to suppress tailing; it separates completely on CHiRAL C and only partially on A and B.

The probes used in the hit rate trial and the four-sample comparison, covering neutral, acidic and basic classes.

The screening trial published by Nacalai Tesque: 28 racemic samples, with a resolution R_s ≥ 1.5 taken as "complete separation (a hit)".

Fig. 5-1Single-column hit rates of the three selectors and their combined coverage.

The best single column reaches 71.4%; screening on all three covers 27 of the 28 samples.

This has two practical implications. First,a single column is not enough— even the selector with the highest hit rate leaves about 28.6% of samples short of R_s ≥ 1.5, so chiral method development should start from multi-column screening rather than "picking one good column". Second,the failures of the three selectors do not overlap— precisely because of the orthogonality along the backbone and substituent dimensions, combined coverage is markedly higher than that of any single column.

5.6SFC compatibility

1-Phenoxy-2-propanol

A sample used in the parallel SFC conditions. Moderately polar with no ionisable group, its retention under CO₂ + 10% isopropanol is comparable to normal phase HPLC.

1-Phenylethane-1,2-diol

A vicinal diol structure, able to form multiple hydrogen bonds with the carbamate. One of the applications listed on the manufacturer's site.

2,2,2-Trifluoro-1-phenylethanol

The strong electron-withdrawing effect of the trifluoromethyl group changes the hydrogen bond donor ability of the adjacent hydroxyl, making it a common probe for examining the hydrogen bonding contribution; one of the applications listed on the manufacturer's site.

COSMOSIL CHiRAL columns can be used in supercritical fluid chromatography. The published literature gives two sets of parallel HPLC and SFC conditions:

Flow rate under SFC is three times higher while, according to the published literature, the separation performance is comparable to HPLC. For preparative work this means higher throughput per unit time, and since the bulk of the mobile phase is CO₂ the solvent evaporation load in work-up is far lower than for normal phase HPLC.

5.7Conditions for neutral / acidic / basic compounds

Methyl DL-mandelate

A neutral sample. The hydroxyl and ester carbonyl of the α-hydroxy ester form two adjacent hydrogen bonding sites; detected at 220 nm on CHiRAL B.

N-benzyloxycarbonyl-DL-alanine

An acidic sample. The carbamate protecting group itself provides hydrogen bonding sites; detected at 210 nm on CHiRAL B with about 0.1% trifluoroacetic acid added.

Mecoprop

An acidic sample of the phenoxypropionic acid class. Separated on CHiRAL A with hexane/isopropanol/trifluoroacetic acid 95/5/0.1.

Carbinoxamine

A basic sample containing a pyridine nitrogen and a tertiary amine. About 0.1% diethylamine is added to the mobile phase; detected at 225 nm on CHiRAL A.

Tramadol

A basic sample containing a tertiary amine and a quaternary carbon hydroxyl. Separated on CHiRAL C with hexane/isopropanol/diethylamine 95/5/0.1.

Brivaracetam

A pyrrolidinone drug with two stereocentres; one of the applications listed on the manufacturer's site. A product with several stereocentres requires both enantiomeric and diastereomeric impurities to be examined.

Arranged in the order neutral, acidic, basic. Adding about 0.1% acid for an acidic sample and about 0.1% base for a basic one is standard practice in normal phase chiral separation.

The published application conditions can be summarised in one practical rule:add about 0.1% acid for an acidic compound and about 0.1% base for a basic one, to suppress ionic interaction between the solute and residual silanols on the silica and improve tailing and detection.

The published literature also includes application data from users, covering indole amides, oxadiazolopyrazines, β-hydroxy ketones, branched carboxylic acids, phenyl glycols, trifluoromethyl benzyl alcohols, exo/endo adducts of bicyclic enones, tetrahydropyrrolocarbazolones, 3-butyl-7-hydroxyphthalide and brivaracetam, ranging from simple alcohols and acids to complex skeletons with several stereocentres.

5.8Available formats and selection advice

This part is compiled from the published information on the chiral drug separation solution page at microwants.com. Specific batch parameters and available formats are confirmed on enquiry.

6.1Positioning and coverage of the three modes

ChiralONE chiral columns usepolysaccharide derivatives as the chiral selectorand can be run innormal phase, reversed phase and polar organicmodes. Their place in the Microwants product system is alongside self-packing capability, the packing station and the preparative column packing service, forming a continuous chain from method screening to preparative scale-up.

The fourteen ChiralONE models

The table below groups them by fixing method and selector, with parameters from Product Centre · ChiralONE chiral columnson this site. The 5 µm series is for routine analysis and the 10 µm series for preparative separation.

6.2The link from analytical screening to preparative scale-up

Chiral projects commonly break at one point in execution: the analytical method is established on a finished analytical column, and only on reaching the preparative stage is it found thatbulk media with the same selector are unobtainable, or that the lead time and cost of a preparative column are unacceptable — so the method has to be redeveloped on a different stationary phase.

ChiralONE addresses this break by supplying media and packing capability together: the selector and mode settled at the analytical stage can be reproduced directly with the same media in a self-packed preparative column, avoiding method redevelopment.

Fig. 6-1The three-layer link in a chiral project.

The key is that the "stationary phase selection conclusion" produced by the first layer can be reused directly in the second and third, without breaking down because the media are unobtainable.

6.3Economics of self-packed preparative columns

The solution page notes that chiral stationary phases carry a high unit cost, so loading and solvent consumption at the preparative stage weigh heavily on total cost; unit production cost can be reduced by raising the injection concentration, optimising the gradient and recovering solvent, and by self-packing preparative columns on a packing station rather than buying finished columns.

Broken down into a calculable cost structure:

6.4The division of roles with COSMOSIL CHiRAL

The two are not alternatives. The more common combination is to complete analytical screening and establish the e.e. method on the three immobilised COSMOSIL CHiRAL selectors, then, once the selector type and elution mode are settled, scale up at the preparative stage with ChiralONE media self-packed in the same mode, while retaining the analytical method for process control and release.

This part gives a directly executable screening and optimisation workflow, together with the key points of e.e. method validation and how to handle common problems.

7.1Designing the screening matrix

The hit rate data in section 5.5 show that the probability of success with a single column under a single condition is limited, so screening should be organised as a matrix. One executable three-tier screening scheme is as follows.

Fig. 7-1The three-tier screening workflow.

Tier 1 covers most samples with a small number of conditions; tiers 2 and 3 address the samples tier 1 missed, expanding along the solvent and mode dimensions respectively.

7.2Mobile phase systems and additives

7.3The order in which to work the optimisation variables

1. Type of modifier— the largest effect on α; try this first. Ethanol and isopropanol often give markedly different selectivity.
2. Column temperature— lowering the temperature raises α in most systems (section 4.7). Examine the range 10–40 °C in 5 °C steps, watching for proximity to the isoelution temperature.
3. Proportion of modifier— mainly affects k. The aim is to keep k between 2 and 10; below k = 1 resolution is noticeably lost, above k = 10 the time cost is not worth it.
4. Type and concentration of additive— affects peak shape and, in some systems, α.
5. Flow rate— affects N. At the analytical stage the flow can be dropped to 0.5 mL/min to buy resolution; at the preparative stage the opposite applies and the flow should be raised as far as the resolution margin allows.
6. Column length and particle size— adjust last. R_s ∝ √N, with diminishing returns; but changing from 5 µm to 3 µm raises N without lengthening the run.

7.4Where reversed phase and polar organic modes fill the gaps

• Sample solubility in an alkane system is too low— a sample solution of sufficient concentration cannot be made up under normal phase conditions and preparative loading is limited, so switch to polar organic mode.
• MS coupling is needed— a normal phase mobile phase is incompatible with an ESI source, so reversed phase or polar organic mode (with ammonium formate/acetate) is the workable choice.
• Bioanalysis— for aqueous matrices such as plasma and tissue homogenate, reversed phase mode allows direct injection.
• α is insufficient in normal phase— switching mode itself provides new selectivity; this is tier 3 of the screen in section 7.1.

7.5Indirect approach: chiral derivatization

Marfey's reagent (FDAA)

A chiral derivatization reagent. It reacts with the amino group of an amino acid to give diastereomers separable on an ordinary C18 column, and the dinitrophenyl group provides UV response at the same time.

N-acetyl-L-cysteine · the chiral thiol used with OPA

Used with o-phthalaldehyde for pre-column derivatization of primary amines, forming isoindole diastereomers. The derivative has limited stability, so the injection timing must be fixed.

Two commonly used pre-column derivatization reagents. Derivatization converts enantiomers into diastereomers, and the separation is carried out on an ordinary C18 column.

A chiral derivatizing reagent (CDR) is reacted with the enantiomers to give a pair ofDiastereomers, whose physicochemical properties differ and which can be separated on an ordinary C18 column (see Fig. 1-2).

The position taken on the solution page is that the direct approach applies more broadly, while the indirect approach retains a cost advantage for amino acid compounds. Chiral purity determination for amino acids can be established as a pre-column derivatization method with the DL amino acid kit.

7.6Key points of method validation for e.e. determination

An e.e. method is essentially aboutquantifying a trace of the minor enantiomer in the presence of a large excess of the major. That positioning makes the focus of validation different from an ordinary assay method.

7.7Common problems and troubleshooting

An analytical method aims at resolution; a preparative method aims at throughput per unit time and unit cost. This part gives the conversion from analytical to industrial column and the engineering constraints involved.

8.1Loading, overload and the touching-band mode

Preparative chromatography runs in thenon-linear region— once the amount injected is large enough the adsorption isotherm is no longer a straight line, and the peak changes from a symmetrical Gaussian to a right triangle. This is the main difference in principle between preparative and analytical work.

Fig. 8-1The relationship between injection amount and peak shape.

The aim of preparative method development is to establish the "touching-band injection amount" and the acceptable overload range above it, not to maintain analytical-grade resolution.

Two kinds of overload must be distinguished:

• Concentration overload— high sample concentration, small injection volume. The peak becomes a right triangle and the retention time moves earlier. This is the form commonly used in preparative work.
• Volume overload— low sample concentration, large injection volume. The peak broadens into a rectangle. It is forced on you when sample solubility in the mobile phase is limited, and throughput is lower than with concentration overload.

Raising sample solubility is therefore a direct way of raising preparative throughput. A stationary phase that allows a strong dissolving solvent (sections 5.3 and 6.3) has a real advantage here.

8.2Scale-up calculations from analytical to preparative column

Keeping the same particle size, column length and linear velocity, scale by the ratio of column cross-sectional areas:

8.3Packing technique and bed stability

The performance of a preparative column is largely determined by packing quality. The usual approaches along the scale ladder:

The value of DAC is that the bed is under compression throughout use, voids from settling are compensated automatically by the piston, so efficiency is maintained over long operation and the media can be discharged, cleaned and repacked at the end of column life. For chiral media with their high unit cost, this bears directly on the effective depreciation period of the media.

8.4Continuous chromatography: SMB and MCSGP

The inherent inefficiency of batch chromatography is that for most of the time only a short section of the column is actually separating, while the rest is occupied by washing and equilibration. Continuous chromatography improves on this with columns in series and valve switching.

SMB fits chiral resolution well because chiral resolution is fundamentally abinary separation— only two enantiomers have to be separated, which is exactly the premise SMB requires. In published reports, the specific solvent consumption and stationary phase requirement of SMB in tonne-scale production of a chiral API are usually markedly lower than for batch chromatography (a qualitative conclusion inferred here; the actual figures depend on the system).

8.5Preparative SFC

Preparative chromatography with supercritical CO₂ as the bulk mobile phase suits chiral separation for three reasons:

• Low mobile phase viscosity— it can be run at high linear velocity with controllable backpressure, giving higher throughput per unit time than normal phase HPLC; in the example in section 5.6 the SFC flow rate is three times that of HPLC.
• Simple work-up— CO₂ vaporises on depressurisation, leaving only a small amount of alcohol modifier in the fraction, so the evaporation load is low.
• Lower organic solvent usage— this cuts solvent purchase and waste treatment costs directly, with a clear improvement in PMI.

The constraints are the capital cost and the maintenance requirements of a pressure system, and sample solubility in the CO₂/alcohol system. COSMOSIL CHiRAL columns can be used in SFC and, according to the published literature, their separation performance is comparable to HPLC (section 5.6), so a method developed on HPLC can be transferred fairly quickly to an SFC platform for evaluation.

8.6Solvent recovery and unit production cost

The specific solvent consumption of chiral preparative work (L solvent / kg product) is usually higher than for ordinary reversed phase preparative work, because α is small in a chiral separation and the loading per run is limited. Solvent recovery therefore carries greater weight in the cost structure.

A usable formula for costing:

Unit production cost = media depreciation/kg + net solvent consumption × solvent unit price + energy and labour/kg + yield loss converted to cost

where net solvent consumption = total solvent used × (1 − recovery). Raising recovery from 70% to 90% cuts net solvent consumption to one third, and the weight of this item in total cost falls noticeably.

8.7Column life and regeneration

• Prevention first— filter samples through 0.22 µm before injection; fit a guard column packed with the same medium as the main column; avoid mobile phases outside pH 2–9; avoid incompatible solvents (coated types especially).
• Routine regeneration— flush with a strongly eluting solvent (isopropanol or THF in a normal phase system; dichloromethane or chloroform on an immobilised type) for 20–30 column volumes, then return to the working mobile phase and equilibrate.
• Column head treatment— where a void at the column head causes split peaks, the head of a preparative column can be removed and media added; a DAC column compensates automatically by adjusting the piston compression.
• Deciding on end of life— judge on the resolution and plate count in the system suitability test, set a percentage fall from the initial value as the retirement threshold, and record it in the equipment log.

This part is a quick reference that can be used on its own, separately from the main text.

9.1Quick reference for mode and column selection

9.2Chinese–English glossary

9.3Sources

• COSMOSIL CHiRAL Series product page (technical parameters, solvent tolerance data, hit rate trial, SFC conditions, application conditions, formats) — NACALAI TESQUE, INC.:https://www.nacalai.com/global/cosmosil/index/CHiRAL.html
• COSMOSIL CHiRAL Series product handbook (PDF) — NACALAI TESQUE, INC.:https://www.nacalai.com/global/download/pdf/COSMOSIL_CHiRAL_Series.pdf
• Chiral Drug Separation · Chromatography Solutions (ChiralONE positioning, the three modes, the indirect approach, e.e. determination, preparative work and scale-up) — Microwants:https://www.microwants.com/solution-chiral.html
• ICH Q6A, Q3A(R2), Q3B(R2), Q2(R2) guidelines
• FDA, Policy Statement for the Development of New Stereoisomeric Drugs (1992)
• Easson L.H., Stedman E., Biochem J., 1933 (the original proposal of the three-point interaction model)