How to interpret an HPLC chromatogram

What a chromatogram actually records

A reversed-phase HPLC trace plots detector response — almost always UV absorbance at 214 or 220 nanometres, where the peptide bond itself absorbs — against elution time. Sample is injected onto a hydrophobic stationary phase and washed off with an increasing gradient of organic solvent. Compounds leave the column in order of hydrophobicity, and each produces a peak.

Purity is calculated from areas, not heights: the target peak area divided by the total integrated area of all peaks, expressed as a percentage. That definition carries two consequences. Anything that does not absorb at the detection wavelength is invisible to the calculation, and anything excluded from integration — for example a solvent front cut off at the start of the run — does not count against the result.

Retention time identifies nothing on its own

Retention time is a property of the method, not of the molecule. Change the gradient slope, the column chemistry, the column dimensions, the temperature or the flow rate and the same peptide elutes somewhere else. A retention time is only informative when compared against a reference standard run under identical conditions on the same instrument.

This is the single most common misreading of a certificate. A stated retention time confirms that the laboratory ran a method; it does not confirm identity. Identity comes from mass spectrometry, which is why a complete release package always pairs the two.

Reading peak shape

A well-behaved peak is narrow and close to symmetrical. Departures from that shape are diagnostic. A shoulder on the leading or trailing edge usually indicates a closely related species — most often a deletion sequence differing by one residue, or a diastereomer formed by racemisation during coupling. A broad, low, poorly resolved peak in a long or hydrophobic sequence commonly reflects on-column aggregation or conformational exchange rather than chemical impurity.

Tailing across all peaks, including any system peaks, points to a column or instrument issue rather than to the sample. Fronting typically indicates column overload — too much material injected — which also compresses apparent resolution and can inflate the reported purity by hiding small neighbours under a distorted main peak.

Small satellite peaks eluting just before the target are frequently more polar oxidation products; a peak sixteen daltons heavier on the corresponding mass spectrum confirms methionine oxidation. Late-eluting minor peaks are usually more hydrophobic adducts, often incompletely scavenged protecting-group fragments from cleavage.

Integration and baseline placement

Every reported purity figure embeds analyst or software decisions about where each peak begins and ends and where the baseline lies. Drawing the baseline across a rising background rather than under it, or setting a rejection threshold that discards small peaks, will shift a result by several tenths of a percent. None of this is misconduct — it is the ordinary discretion built into chromatographic integration — but it does mean purity figures should be read as measurements with a method-dependent uncertainty, not as exact constants.

In practice this makes within-laboratory, within-method comparisons far more meaningful than cross-vendor ones. Two laboratories reporting 98.6 and 99.1 percent for equivalent material are not necessarily disagreeing about the material.

What to check before trusting a trace

Confirm the sample identifier and lot number printed on the chromatogram match the vial and the certificate header. Check the method block states column, mobile phases, gradient, flow rate, wavelength and injection volume. Verify the run time is long enough that the gradient reached its end — a trace that stops early can leave late-eluting hydrophobic impurities uncounted. Look for a stable baseline before and after the peak of interest.

Finally, check the trace is an instrument output rather than a redrawn image. Genuine chromatograms carry axis labels, timestamps, integration marks and a peak table with retention times and area percentages. A smooth, unlabelled curve in a marketing PDF is an illustration, not analytical evidence.