Reading an HPLC purity chromatogram

A certificate that reports purity as a single percentage is reporting the output of one chromatographic method, run once, on one sample drawn from one lot. The number is real. What it is a number about is narrower than it looks, and most of that narrowness lives in the method conditions printed underneath it — or not printed at all.

This note is about reading the chromatogram and the conditions that produced it. It does not re-argue the difference between chromatographic purity and peptide content; a separate note in this library covers that. It says nothing about what any material may be used for.

What area percent computes

A UV detector records absorbance against time. Software builds a baseline, integrates each peak it recognises against that baseline, sums the integrated areas, and divides the main peak's area by the total. That quotient, times one hundred, is the purity figure.

Two consequences follow immediately. The figure is relative: it compares the main peak only to the other peaks the method produced and the software chose to integrate. And it is a ratio of areas, not of masses. Nothing in the calculation converts detector response into quantity of substance.

The detection wavelength is usually in the low-UV region because the peptide bond itself absorbs there. That choice is deliberate — backbone absorbance gives a broadly comparable response across peptide-related species regardless of side chain. It is still a choice, and it has consequences.

What the integration silently excludes

Several categories of material can be present in the container and absent from the chromatogram. None of them appears as a deficit.

• Species that never elute — irreversibly retained on the stationary phase, precipitated in the mobile phase, or removed by a sample filter during preparation.
• Species that elute unretained in the void volume, where they may sit inside an excluded solvent-front window.
• Species that co-elute with the main peak and are integrated as part of it.
• Peaks below the integration threshold or below a stated disregard limit.
• Species with little or no absorbance at the detection wavelength.
• Anything insoluble in the sample diluent, which never enters the injector at all.

The percentage still sums to one hundred, because it is defined to. A method that resolves four impurities and a method that resolves eleven both produce a figure that adds up. Only one of them saw eleven.

The wavelength is part of the result

Absorbance is not a general-purpose detector. It responds to chromophores, and different chromophores absorb in different regions. Backbone amide absorbance sits in the far UV. Aromatic side chains — tryptophan, tyrosine, phenylalanine — absorb further out, near 254 to 280 nm.

Move the detection wavelength and the relative peak areas move with it. At an aromatic wavelength, a peptide-related impurity lacking any aromatic residue may be close to invisible, while species rich in those residues are amplified. At a far-UV wavelength, non-peptide process residues that happen to absorb there — scavengers, some counterions, residual reagents — are counted alongside peptide-related peaks with no distinction made between them.

Both figures are correct. They answer different questions. A purity result without a stated wavelength is therefore incomplete in a way that no amount of decimal places repairs. Diode-array detection, evaporative or charged-aerosol detection, and mass spectrometric detection each change what is visible again; a certificate that names only "HPLC" has not told you which of these produced the number.

Gradient and column conditions are the result, not a footnote

The separation decides how many peaks exist to be counted. A shallow gradient through the region where the main component elutes pulls closely related species apart and reports them as impurities. A steeper gradient across the same region merges them into the main peak and reports a higher purity. Neither run is wrong. They are different measurements.

Conditions that change the number include:

• Stationary phase chemistry, particle and pore size, and column dimensions.
• Mobile phase modifier and ion-pairing agent — selectivity under trifluoroacetic acid and under formic acid is not the same selectivity.
• Gradient slope, any isocratic holds, flow rate, and column temperature.
• Injection mass and diluent composition; overload distorts peak shape and can bury a shoulder inside the main peak.
• Detection wavelength, bandwidth, reference wavelength, and data acquisition rate.
• Integration settings: threshold, expected peak width, baseline construction, disregard limit, and whether the solvent front is excluded.
• Total run time, including any final hold. Impurities that would elute after the run stops are not counted, because the run stopped.

None of this is unusual or improper. It is what a chromatographic method is. The point is that the conditions are not context for the result; they are constitutive of it.

Why two labs' figures are not directly comparable

Give the same lot to two laboratories using different columns, different modifiers and different gradients, and you should expect two different purity figures. Both can be defensible. Neither validates or contradicts the other, because they are not measuring the same partition of the sample.

System suitability does not close this gap. USP ⟨621⟩ and Ph. Eur. 2.2.46 describe parameters such as resolution, tailing factor, plate count and injection repeatability, and set out how far chromatographic conditions may be adjusted before a procedure is no longer the stated procedure. A run that meets suitability criteria has demonstrated that the separation performed as the method requires. Two methods can both meet their own criteria and still resolve different sets of impurities.

Comparability therefore depends on the method matching, not on both numbers being high. When you need to compare — lot against lot, supplier against supplier, incoming result against certificate — ask for the following before you compare anything:

1. The chromatogram itself, not only the derived percentage.
2. Detection wavelength or wavelengths, and detector type.
3. Column identity, dimensions, and particle size.
4. Mobile phase composition, gradient table, flow rate, column temperature, and total run time.
5. Injection volume and sample concentration in the stated diluent.
6. Integration parameters, including any disregard limit and any excluded region.
7. System suitability results recorded for that specific run or sequence.
8. Whether individual impurity peaks were identified or merely counted.

That last item is worth separating. ICH Q3A(R2) gives the working vocabulary for organic impurities in a drug substance — reporting, identification and qualification are distinct activities at distinct levels — and USP ⟨1503⟩ describes quality attributes and associated test approaches for synthetic peptide drug substances, including peptide-related impurities and their quantification. A chromatogram in which every peak is integrated and none is identified is a peak count, not an impurity profile.

Naming those documents here describes the framework analysts work within. It is not a statement that any particular material was tested under them, and a certificate that cites a chapter has not thereby demonstrated anything about the lot in front of you.

Read outward from the number. The percentage summarises a chromatogram; the chromatogram summarises a method; the method decides what could have been seen at all. A figure quoted without its conditions is a figure you cannot check, cannot reproduce, and cannot compare to anything.