What an HPLC purity figure actually measures, and what it does not

Almost every research-peptide certificate in circulation leads with a single number. It is usually somewhere between 98 and 99.9 percent, it is usually attributed to HPLC, and it is usually presented without any explanation of what was integrated, at what wavelength, or against what. This note sets out how that number is produced, what it legitimately supports, and the four things it is routinely assumed to mean but does not.

How the number is produced

In reverse-phase HPLC, a dissolved sample is pushed through a column packed with hydrophobic C18-modified silica while the mobile phase becomes progressively less polar, typically a gradient from aqueous 0.1% trifluoroacetic acid into acetonitrile. Components partition between the stationary and mobile phases according to hydrophobicity and elute in order, each producing a peak at the detector.^4,5

The detector records absorbance against time. Software integrates the area beneath each resolved peak, and purity is reported as:

purity % = (main peak area ÷ total integrated peak area) × 100

That is the whole calculation. Everything interesting is in the qualifiers around it.

What the number does not tell you

1. It is an area percentage, not a mass percentage

Peak area is proportional to absorbance, and absorbance depends on how strongly a species absorbs at the chosen wavelength. Two compounds present at identical molar concentration can produce very different areas. So 99.6% area does not mean 99.6% by weight. A lyophilised peptide vial reporting 99.6% HPLC purity will still typically contain several percent water by mass and a meaningful proportion of trifluoroacetate counter-ion. Neither is an impurity in the chromatographic sense, and neither appears in the calculation at all. This is why we also report Karl Fischer water content on every certificate.

2. It is wavelength-dependent

We detect at 214 nm because the peptide bond itself absorbs strongly between roughly 210 and 220 nm, so every residue contributes to the signal.⁴ Detection at 280 nm instead relies on aromatic residues (tryptophan, tyrosine and phenylalanine), which means a peptide containing none of them is close to invisible. A certificate that omits the detection wavelength is not fully interpretable, and two certificates run at different wavelengths are not directly comparable.

3. It says nothing about identity

Chromatography measures how much of one detected thing there is relative to other detected things. It does not establish what that thing is. A single sharp peak integrating to 99.8% is entirely compatible with a highly pure sample of the wrong compound: a deletion sequence, a diastereomer, or an unrelated peptide of similar hydrophobicity. Identity has to come from an orthogonal technique. Electrospray mass spectrometry supplies it by matching observed molecular weight against the value calculated from the sequence, which is why we treat identity and purity as two separate release gates rather than one combined claim.

4. It only counts what elutes and absorbs

Species that are retained irreversibly on the column, elute in the void volume, or simply do not absorb at 214 nm are excluded from the denominator. Inorganic salts are the obvious case. This is a well-recognised limitation of area-normalisation approaches generally, and it is the reason pharmacopoeial impurity work pairs chromatographic profiling with orthogonal assays rather than relying on a single normalised figure.^2,3,6

How to read a certificate properly

A certificate you can actually act on states, at minimum:

• The batch number, matching the vial label and the outer seal.
• Column chemistry and dimensions, the gradient, and the detection wavelength.
• Retention time for the main peak, so the trace can be checked against the table.
• The integration table, showing every peak, not just the one that flatters the number.
• An orthogonal identity result. Expected mass versus observed mass.
• The testing laboratory and date, named, so the work is attributable.
• Water content, because it is the largest unstated mass term in a lyophilised vial.

If a supplier will only release that document by email, on request, after purchase, the practical effect is that the number in the product listing cannot be checked at the moment it matters. That is the specific gap our open batch registry exists to close: the trace, the integration table and the identity result are published against the batch number before anyone buys anything, and they stay there afterwards.

Our current position

Kelvin's release floor is 99.0% main-peak area by RP-HPLC at 214 nm on C18, with identity confirmed by ESI-MS against the calculated mass. Mean released purity across 2025–26 was 99.4%. Batches that fail either gate are not released, are not repriced, and do not enter the catalogue. Every released batch is published with its full chromatogram. For example, batch KB-26-0342.