C18 Is the Default, Not Always the Answer

Take a 43-residue protein, push it through a 100 angstrom C18 column on a gradient somebody wrote for a different molecule, and the trace that comes back will look entirely respectable. That is the trouble with it. Peaks appear, the software integrates them, and a number with a decimal point drops out at the end. What the run does not do is register everything that went into the injector. Nearly every purity figure in circulation was produced on C18, which is why the one line of a certificate naming the column repays a look.

Begin with the holes, not the coating

Chromatographic silica is riddled with pores, and the overwhelming majority of usable surface lies down inside them rather than on the outer skin of each particle. Retention thus depends on whether a molecule can get in at all. When it cannot, it slides past and meets only the exterior, and three things degrade at once: the grip turns weak, efficiency falls away, and recovery stops being something you can predict.

Convention divides at a single boundary. Pores of 100 angstroms serve small molecules and short sequences, while 300 angstroms is what anything bulkier demands. Put a long peptide on the narrow-pore version and the chromatogram will not flag the error. Excluded species are the largest ones present — aggregated material, chains never cleaved to completion — and those are exactly the population a purity determination exists to expose. The figure returns clean because the worst of the sample never entered the measurement.

Chain length is a dial, and the default is only one setting

That digit following the C counts carbons in the alkyl chain grafted to the silica. More carbons produce a more hydrophobic surface and a firmer hold on whatever passes over it.

Two opposite failures bracket the default. Something short and polar may fail to grip C18 whatsoever and arrive at the solvent front, a region where nothing resolves from anything. Something bulky and hydrophobic may hold on so tightly that only a punishing gradient dislodges it, and whatever never dislodges is recovery quietly lost.

One acid, two jobs that pull in opposite directions

Water and acetonitrile do the carrying; an acidic additive governs how the result looks. Trifluoroacetic acid is the usual one. It ion-pairs with basic residues and tightens peptide peaks, and it stays in routine service even though it damps signal badly in electrospray. Feed the column outlet into a mass spectrometer and formic acid generally takes over instead: peak shape gets worse, ionization gets considerably better. That trade is set out in MALDI and ESI.

Whichever acid is chosen also governs the salt form the isolated material ends up carrying, a wholly separate matter once somebody tries to reconcile a label mass. TFA and acetate counter-ions covers that.

Reading the line nobody reads

A column description occupies one line, and it is the line readers skip. It says whether the method was designed for this compound or borrowed from another. Hardware selection is also among the reasons one vial can yield two different numbers at two laboratories with neither of them being wrong. Why certificates of analysis disagree covers that whole set, while system suitability shows how an analyst demonstrates the column was behaving on the morning it ran.

All material is supplied for laboratory research use only. It is not a drug, not a supplement, and not for use in humans or animals.