Written and maintained by CASRAI Editorial Board
Last updated
Enantiomers — non-superimposable mirror-image forms of the same molecule — are chemically identical in every achiral environment: same melting point, same mass, same UV spectrum, same retention on an ordinary C18 column. They can still behave completely differently in a biological system, which is why chiral separation matters for drug substances, agrochemicals and flavor/fragrance compounds, and why regulators routinely require enantiomeric purity data rather than accepting a racemate’s bulk purity as sufficient. (The clearest historical illustration is thalidomide, whose two enantiomers have markedly different pharmacological effects — a standard reference point in any discussion of why chirality is regulated, not just measured.)
Resolving enantiomers requires a chiral stationary phase (CSP): a column packing built from an inherently chiral selector that interacts differently with each enantiomer of a racemic pair. Unlike an achiral separation, where you can reason from analyte polarity and pKa to a sensible starting column and mobile phase, chiral recognition depends on a three-point (or more) interaction between analyte and selector that is genuinely difficult to predict from structure alone. In practice, method development is a screening exercise — run a candidate against a panel of CSPs and mobile-phase modes and see what resolves — more than it is a design exercise. This guide covers how to build that screening panel, how to read early results, and, just as important, how to recognize the point at which further screening stops being worth the column time.
The Two Workhorse Chiral Stationary Phase Families
A handful of CSP chemistries cover the large majority of routine chiral method development. Two families do most of the early-round work:
| Family | Selector | Typical strengths | Typical modes |
|---|---|---|---|
| Polysaccharide-derived | Cellulose or amylose, derivatized as tris-carbamates or tris-benzoates and either coated onto or chemically immobilized on a silica support | Broadest general-purpose hit rate; resolves a wide structural range including many rigid, aromatic and hydrogen-bonding analytes | Normal-phase and polar-organic mode routinely; immobilized (as opposed to coated) versions tolerate a wider solvent range, including dichloromethane and THF, and some grades extend into reversed-phase |
| Macrocyclic glycopeptide | A macrocyclic antibiotic (e.g. a vancomycin- or teicoplanin-type selector) bonded to silica, presenting multiple chiral cavities and ionizable/H-bonding groups | Complementary selectivity to polysaccharide phases; often the better choice for analytes with ionizable amine or acid groups, amino acids and small zwitterions | Genuinely multimodal — normal-phase, reversed-phase and polar-organic all work on the same column, which makes it useful as a single physical column to re-screen across modes |
Other CSP chemistries exist — Pirkle-type (brush/donor-acceptor) phases, native and derivatized cyclodextrins, protein-based phases (alpha-1-acid glycoprotein, bovine serum albumin) for polar and ionic analytes, and crown-ether or ligand-exchange phases for amino acids — but they’re generally second-round options once a polysaccharide/macrocyclic screen has been exhausted, not the starting point, because the two families above resolve a large enough share of drug-like racemates to justify screening them first. See HPLC column selection and stationary phases for how chiral CSPs fit into the broader universe of HPLC packings.
Three Modes, Three Different Selectivities
The same physical CSP column can behave like a genuinely different separation on a different mobile phase, because the balance of interactions driving chiral recognition (hydrogen bonding, dipole-dipole, pi-pi stacking, inclusion) shifts with the mobile-phase environment. Treat mode as a second screening variable, not an afterthought once a column is picked.
| Mode | Mobile phase | Notes |
|---|---|---|
| Normal-phase (NP) | A non-polar hydrocarbon (hexane or heptane) with a polar alcohol modifier (isopropanol or ethanol, typically 5-40%), often with a small amount of acid/base additive (e.g. trifluoroacetic acid, diethylamine) to sharpen peak shape | The traditional, still widely used mode for polysaccharide CSPs; strong hydrogen-bonding-driven selectivity; poor fit for LC-MS (hexane and non-volatile-additive-free NP methods are workable but require care) and a higher solvent-disposal/HSE burden than the alternatives |
| Reversed-phase (RP) | Aqueous buffer with methanol or acetonitrile | MS-compatible and familiar to any lab already running achiral RP-HPLC; on polysaccharide CSPs the retention/recognition mechanism shifts toward inclusion-type interactions and selectivity is often different from — not just weaker than — the NP result on the same column, so a poor RP screen doesn’t rule out that column in another mode |
| Polar-organic mode (PO) | Acetonitrile and/or methanol (or a mix), with no water and no hexane, usually with a small volatile acid/base additive (e.g. acetic acid/triethylamine) | Fast column equilibration and often the highest efficiency of the three; frequently gives a third, distinct selectivity pattern from the same CSP; MS-compatible when volatile additives are chosen, and avoids both hexane’s HSE burden and RP’s longer re-equilibration |
A practical consequence: don’t discard a CSP after a single mode fails to resolve a pair. The same column screened in the other two modes is, mechanistically, close to three different experiments.
Building a Screening Panel
Because chiral recognition can’t be reliably predicted from structure, method development groups typically run a matrix — a small panel of CSPs against two or three modes each — rather than one column at a time. A representative starting panel:
- 3-4 polysaccharide columns spanning both cellulose- and amylose-derived chemistries (they have different, complementary selectivity even though they’re the same broad family), run first in normal-phase or polar-organic mode.
- 1-2 macrocyclic glycopeptide columns, especially if the analyte has an ionizable or amino-acid-like functional group, run across normal-phase, reversed-phase and polar-organic mode on the same physical column.
- A short list of generic mobile-phase conditions per mode (e.g. two or three alcohol modifiers at a couple of ratios for NP; two organic-modifier choices for PO) rather than an exhaustive optimization at this stage — the goal of round one is to find which column/mode combination shows any separation at all, not to optimize resolution yet.
Labs running high method-development volume automate this with column-selector/mobile-phase-selector valve hardware on the LC system, so a panel of 6-8 columns times 2-3 modes can run unattended overnight as a single sequence rather than requiring manual column swaps. Whether automated or manual, log every combination tried, including the negative results — a “no separation” result on a given column/mode pair is real information that prevents re-testing it later, and it’s useful evidence if a regulatory submission or method-transfer package needs to show the screening rationale behind the final method.
Reading Early Screening Results
A first-round injection on a new column/mode combination usually falls into one of three buckets:
- No separation at all (a single peak, or two peaks that don’t correspond to the two enantiomers) — move on to the next combination in the panel; don’t spend time optimizing a condition that shows zero chiral recognition.
- Partial separation (a shoulder, a fused doublet, or clean peaks with resolution below target) — this is the useful signal. A real, if incomplete, separation means the CSP/mode pair has the right recognition mechanism; from here, small changes (modifier ratio, temperature, flow rate, additive concentration/type) typically improve resolution faster than switching columns again.
- Elution-order ambiguity — when a reference standard of one enantiomer isn’t available, don’t assume elution order transfers between columns or even between modes on the same column; chiral elution order is not generally predictable across CSP chemistries and has to be established for the specific column/mode combination in use, typically by spiking a known-enriched sample or by an independent technique (e.g. polarimetry/ECD on a collected fraction) if that matters for the application.
When to Stop Screening
Exhaustive screening has a real cost — column purchase or rental, solvent, and instrument time — so the practical question is when a hit is good enough to stop on, not just whether one exists. A few criteria, used together rather than any single one in isolation:
- Resolution against the method’s actual purpose. A quantitative, validated release or stability method generally needs baseline or near-baseline separation; pharmacopeial system-suitability practice (general chromatography guidance such as USP <621>) commonly treats a resolution factor (Rs) around 1.5 as the working definition of baseline resolution for a quantitative method, while a qualitative screen (confirming a compound is enantiopure enough to proceed, not measuring an exact ratio) can tolerate less. Match the target to what the method is actually for — don’t chase Rs 2.5 on a screening assay that will never need it, and don’t stop at a partial separation for a method that will be used to report a numeric enantiomeric-excess result.
- Peak shape, not just peak spacing. A pair that meets a resolution number on paper but shows a badly tailing minor enantiomer peak (common near the detection/quantitation limit for a trace-impurity enantiomer assay) may not actually be usable at the concentration the method needs to work at — check peak shape at the relevant concentration, not just at a convenient loading level.
- Robustness and transferability. A condition that only works on one column serial number, or that depends on a mobile-phase ratio sitting right at a knife-edge between resolving and co-eluting, is a fragile method even if it screens well once. Favor a combination with some margin, on a CSP chemistry that’s commercially available from more than one source if the method will need to be transferred or the column re-ordered years later.
- Detection and downstream compatibility. If the method needs LC-MS detection (see LC-MS: how liquid chromatography and mass spectrometry are coupled), a polar-organic or reversed-phase hit with volatile additives is a more practical stopping point than a normal-phase hit that technically resolves the pair but fights the ion source.
- Diminishing returns. If a reasonably built panel — several polysaccharide columns and at least one macrocyclic glycopeptide column, each across two or three modes — produces nothing usable, that’s a real signal to change strategy rather than keep adding polysaccharide variants: move to a second-round CSP chemistry (cyclodextrin, protein-based, Pirkle-type) suited to the analyte’s functional groups, or reconsider the separation technique itself (supercritical fluid chromatography is a common escalation path for compounds that resist HPLC-based chiral screening).
In short: stop screening at the first combination that clears the resolution bar the method actually needs, shows acceptable peak shape at working concentration, and isn’t sitting on a knife-edge — not at the first combination that merely separates the pair, and not only after every column in the lab has been tried.
How This Differs From Size-Based Separation
Chiral screening is sometimes confused with size exclusion chromatography (SEC) because both involve “picking the right column” from a manufacturer’s range, but the two solve unrelated problems. SEC separates by hydrodynamic size using an inert, non-interacting resin — see size exclusion chromatography: column selection, calibration and molecular weight determination — and a well-run SEC method deliberately avoids any specific chemical interaction between analyte and packing. Chiral separation is the opposite: it depends entirely on a specific, three-point interaction between analyte and a chiral selector, and a molecule’s size is close to irrelevant to whether it resolves. A method that needs both size information and enantiomeric purity on the same sample uses two separate, unrelated analyses, not one column doing both jobs.
Frequently Asked Questions
Do I need a reference standard of both enantiomers to develop a chiral method?
Not to find a separating condition, but you do need at least one enantiomerically enriched or single-enantiomer reference to assign elution order and to know which peak is which. Without it you can confirm a separation exists but can’t reliably report which peak corresponds to which enantiomer.
Can polysaccharide and macrocyclic CSPs be used on the same LC system as ordinary achiral columns?
Yes — they’re standard HPLC/UHPLC columns dimensionally and connect to any conventional LC system; no specialized instrument is required, only the column-switching hardware if a lab wants to automate panel screening across multiple columns.
Why does the same column give different selectivity in normal-phase versus polar-organic mode?
Because the interactions driving chiral recognition on a given selector (hydrogen bonding, dipole-dipole, pi-pi stacking) are weighted differently depending on the mobile-phase environment, so the mobile phase is effectively a second, independent screening variable rather than just a tuning knob on a fixed separation.
Is gas chromatography ever used for chiral separations?
Yes, for sufficiently volatile analytes, typically using a cyclodextrin-derivatized capillary phase, but that’s a distinct technique from the HPLC-based polysaccharide/macrocyclic screening covered here and is chosen based on analyte volatility, not as a substitute screening step.
How many column/mode combinations does a typical screen actually need before stopping?
There’s no fixed number — it depends on the analyte — but many labs find a usable condition within a handful of polysaccharide-column/mode combinations, with macrocyclic glycopeptide phases as the next tier if those don’t work; the stopping point is defined by the resolution/robustness criteria above, not by exhausting a fixed column count.
