Why do HPLC peaks tail? Causes and fixes

First, measure it: the USP tailing factor

"Tailing" is a slow return of the trailing edge of a peak to baseline. Quantify it with the USP tailing factor, measured at 5 % of peak height: T = W_0.05 / 2f, where W_0.05 is the full width at 5 % height and f is the distance from the leading edge to the peak apex at that height. A perfectly symmetric (Gaussian) peak gives T = 1.0. Most methods accept T ≤ 2.0; above that, area precision and resolution degrade quickly. (The closely related asymmetry factor As is measured at 10 % height — same idea, different reference point.)

Cause 1 — Active silanols (the most common)

On a silica-based reversed-phase column, residual acidic silanol groups (Si–OH) act as weak cation exchangers. A protonated basic analyte (amine, pKa > 8: think β-blockers, tricyclic antidepressants, many drugs) partitions by hydrophobicity and sticks transiently to silanols. The two retention mechanisms desorb at different rates → a long, dragging tail.

Fingerprint: basic compounds tail badly; neutral/acidic ones on the same run are sharp. Worse on older, type-A silica.

Fix: switch to a low-silanol-activity, high-purity type-B or charged-surface / hybrid column; add a competing base or a small amount of ion-pairing agent; or move the pH so the analyte is less protonated. A column's silanol activity is exactly what the Hydrophobic Subtraction Model (HSM) "C" term captures — see our column-equivalence note.

Cause 2 — Column overload

Inject too much mass (or too much volume in a strong solvent) and the stationary phase saturates locally. The isotherm becomes non-linear and you get a characteristic shark-fin shape — a sharp front and a sloping tail (or a fronting peak for some isotherm types).

Fingerprint: tailing scales with injected amount; dilute the sample and the peak sharpens.

Fix: reduce on-column mass (dilute, inject less), or dissolve the sample in a weaker solvent than the mobile phase so it focuses at the head of the column.

Cause 3 — Extra-column dead volume

Unswept volume anywhere between the injector and the detector — over-long or wide tubing, a poorly cut fitting, a worn connection — lets analyte bands smear. This is a system problem, not a chemistry one.

Fingerprint: early-eluting peaks tail most (they have the least column volume to "average out" the dispersion); the effect is roughly constant in absolute time across the run.

Fix: minimize tubing length and ID, use properly seated zero-dead-volume fittings, and keep the detector flow cell appropriate to the flow rate.

Cause 4 — The wrong mobile-phase pH

For an ionizable analyte, retention and peak shape are most reproducible when the mobile-phase pH sits at least 2 units away from the analyte's pKa — so it is fully ionized or fully neutral, not a 50/50 mixture that two slightly different equilibria can broaden. Worse, "water with a pH dialed in" does not hold that pH: only a real buffer near its pKa controls it.

Fingerprint: shape and retention drift between runs or columns; the analyte's pKa is within ~1.5 units of the working pH.

Fix: use a buffer whose pKa is within ±1 of the target pH and at adequate concentration; move the pH well away from the analyte pKa.

A quick decision checklist

• Only basic compounds tail? → silanols. Change column chemistry or pH.
• Tailing grows with injection amount? → overload. Dilute / inject less.
• Early peaks worst? → dead volume. Fix plumbing.
• Shape/retention drift run-to-run, pKa near pH? → buffer/pH. Move ≥2 units from pKa, use a real buffer.

Tailing is rarely mysterious once you read its fingerprint. Measure T, match the pattern, apply the one fix that targets it — and re-measure.