Coated vs immobilized chiral HPLC columns: how to choose

Why do polysaccharide columns dominate enantiomer separation?

Because enantiomers have almost the same physical properties, and a selector that fits one of them better is what pulls them apart.

Two molecules that are mirror images of each other (enantiomers) have almost identical physical properties, so an ordinary reversed-phase column does not separate them. In the body, however, they behave as different compounds: sometimes only one acts, and sometimes one is harmful. For small-molecule APIs, the amount of the unwanted enantiomer can end up in a specification (small-molecule API QC (in Japanese)).

To separate them you use a column packed with a chiral stationary phase (CSP). Silica particles carry a molecule (the selector) that has a better fit with one of the two enantiomers, and the difference in fit shifts the retention times. The mainstream choice in small-molecule drug development is the polysaccharide type, with cellulose or amylose derivatives as the selector. Daicel's CHIRALPAK and CHIRALCEL lines are the best-known examples.

Two axes define the choice: how the selector sits on the silica, which sets the usable solvents, and what the selector's backbone is, which sets the compounds you can separate.

How does coating versus immobilization decide your solvents?

Coated phases dissolve in strong solvents. Immobilized phases do not.

A coated column has the polysaccharide derivative physically laid over the silica surface. Examples are CHIRALPAK AD-H (amylose), CHIRALCEL OD-H (cellulose) and Phenomenex's Lux Amylose-1. An immobilized column has the same kind of selector chemically bonded to the silica. CHIRALPAK IA carries the same selector as AD-H, IB the same selector as OD-H, and IC is cellulose tris(3,5-dichlorophenylcarbamate). Lux i-Amylose-1 is another immobilized example.

Because the polysaccharide dissolves in strong solvents, a coated column is at risk from them. Daicel's AD-H instruction manual (July 2013 edition) names acetone, chloroform, DMF, DMSO, ethyl acetate, methylene chloride and THF as solvents that can destroy the stationary phase if they only remain in the flow path, and it warns that even traces are dangerous. Before you connect the column, it asks you to flush the whole instrument, including the injection port and the sample loop, with a solvent compatible with the column, and to replace the autosampler wash solvent. The shipping solvent is hexane and 2-propanol at 9 to 1.

On an immobilized column the selector does not wash off, so strong solvents can be used both in the mobile phase and to dissolve the sample. Daicel's IB instruction manual says miscible organic solvents can be used, including chloroform, ethyl acetate, THF and MtBE.

In practice, a coated column fits when you can stay with hexane and alcohols and want to narrow down conditions quickly under normal phase. An immobilized column fits when the sample does not dissolve well in hexane, when you want to explore by changing solvents, or when preparative work uses strong solvents. One caveat: for some compounds, comparisons show that the coated column separates better.

Same selector, different attachment, different usable solvents; mix up the types and a single injection can end a column.

What order should you choose in, from solvent to particle size?

Decide the solvent first. Everything else follows from it.

Start with what the sample dissolves in and what you want to use as the mobile phase. If hexane and alcohol are enough, you can begin with the coated AD-H or OD-H, which have a thick track record. If the sample dissolves poorly in hexane, or you may move between reversed phase and polar organic conditions, starting with an immobilized column means less rework.

Next comes the selector. Cellulose and amylose phases differ in selectivity, and a compound that does not separate on one may separate on the other. Selectors with different substituents also exist: CHIRALCEL OJ-H is cellulose tris(4-methylbenzoate), and IC has dichlorophenyl groups that pull electrons. Because immobilized phases can be tried while changing the solvent, they suit a screen that lines up a few columns with different backbones and substituents.

Particle size comes after that. Going from 5 µm to 3 µm raises efficiency, and the same resolution can be reached with a shorter column in less time, but selectivity does not change. A compound that does not suit the selector will not separate however fine the particles are. Settle particle size after a separation is confirmed, to match the analysis time or the preparative quantity.

When you must match an official compendium or an already approved method, check the packing specification too. The United States Pharmacopeia (USP) specifies packing material by L code. In manufacturers' product tables, the coated type of amylose tris(3,5-dimethylphenylcarbamate) is listed as L51 and the immobilized type as L99. The same selector in a different type has a different code, so whether one can replace the other is checked against the method's wording and system suitability.

If preparative work is likely, favor makers that offer larger-diameter columns of the same selector. Nacalai Tesque's COSMOSIL CHiRAL 5A has preparative versions with 10 mm and 20 mm internal diameter, and Regis Technologies' RegisPack also comes in a 3 µm grade and works for both HPLC and SFC. You can move the selector you found in screening to a thicker column without changing it.

You may also be able to source the same selector from another maker. YMC's CHIRAL ART series uses selectors from the same family as IA and IC. But even when the chemical structure of the selector is the same, a different packing or lot does not guarantee the same measured resolution. When you switch, confirm resolution and retention time with your own sample.

Decide the solvent first, try two backbones for the selector, and decide particle size last.

What breaks a chiral column, and what misleads you?

Three things: strong solvents on coated columns, additive residue, and unwatched elution order.

The most common is bringing a strong solvent to a coated column. It need not be in the mobile phase. It can come in as the sample dissolved in DMSO or THF, as autosampler wash solvent left behind, or as the previous user's solvent still in the flow path. This is why Daicel's manual says even traces are dangerous. It shows up as lower resolution or distorted peak shape.

The second is acid or base additive left on the column. Additives such as diethylamine and trifluoroacetic acid are known to stay on polysaccharide columns and change retention and separation (the memory effect). After you change additives, confirm with a standard that retention time and peak shape have returned before you use the column. A safe practice is to assign columns by additive type.

The third is not looking at elution order. There are reports of the elution order of enantiomers reversing when only the additive type was changed. When a trace enantiomer tails behind the main peak and overlaps it, quantitation becomes difficult. A condition where the trace enantiomer comes out first lets you read down to a lower amount (detection limit and quantitation limit (in Japanese)).

Check three things before each injection: what is left in the flow path, which additive this column has seen, and which enantiomer elutes first.

What to check before you select, quote or run a chiral column

Use these when you choose a column, or when you confirm a quotation and specification with a manufacturer.

1. What does the sample dissolve in? Will the injection solvent contain a strong solvent such as THF, dichloromethane or DMSO?
2. What solvent do you want to use in the mobile phase? Might you move to reversed phase or polar organic solvents later?
3. Have you tried both amylose and cellulose phases?
4. Does an official compendium or an existing method specify the packing, for example by L code? If so, does it match down to coated or immobilized?
5. Is it analysis only, or will you go to preparative scale? Is there a larger-diameter column with the same selector?
6. Is anything from the previous solvent left in the injection port, the loop or the autosampler wash solvent?
7. If you use additives, have you recorded which additive was used earlier on this column?
8. Does the trace enantiomer come out before or after the main peak?
9. Have you asked the manufacturer whether the solvent you want to use is guaranteed for that selector?

Makers and products are collected on the chiral columns page of the product database. Why chiral purity ends up in a specification is covered in small-molecule API QC (in Japanese), and what to confirm when you move an analytical method to another site is in analytical method transfer (in Japanese). The selection logic differs from that of the RP-HPLC columns used for ordinary related-substance analysis.