Chromatogram Peak Integration: How Choices Shift a Purity Figure
A purity figure looks like a direct reading from an instrument, but it passes through a step that involves choices: deciding where each peak starts and ends, where the baseline runs beneath it, and how overlapping peaks are divided. That step is chromatogram peak integration, and two careful analysts working from the same raw data can finish it with purity results a few tenths of a percent apart. For labs comparing certificates across many lots, or checking a supplier’s result against their own, understanding integration explains much of the small variation that otherwise looks like a real difference in the material.
Chromatogram peak integration: from trace to number
The detector records a continuous signal. Integration software turns that signal into a table of peaks, each with a retention time, a height and an area. For purity by area percentage, the calculation is then simple: the main peak’s area divided by the total area of all counted peaks, multiplied by 100.
Every term in that calculation depends on integration decisions:
- Detection settings, such as slope sensitivity and expected peak width, decide which small bumps are recognised as peaks at all.
- Baseline placement decides how much of each signal counts as area.
- Peak splitting rules decide how overlapping peaks share their area.
- Exclusions, such as the solvent front, blank peaks and a reporting threshold, decide which peaks enter the total.
Baseline placement under a gradient
On a flat, quiet baseline the question hardly arises. Peptide methods, though, usually run a gradient at low UV wavelengths, where the baseline drifts as the mobile phase changes. The software has to project a baseline beneath each peak, and small differences in that projection change the area.
The effect is proportionally small for a large main peak and proportionally large for a small impurity. Drawing the baseline a little high under a minor peak can shrink its area towards the reporting threshold, and if it falls below that threshold the peak disappears from the total altogether. Our article on gradient elution HPLC explains why that drift happens.
Dividing peaks that overlap
When two peaks are not fully resolved, their area has to be split. The common conventions give different answers:
| Approach | How it divides the area | Typical effect |
|---|---|---|
| Perpendicular drop | A vertical line from the lowest point between peaks down to the baseline | Each peak keeps part of the other’s tail |
| Valley to valley | Baselines drawn between the valley points, so each peak sits on its own raised base | Tends to reduce the areas of both, especially the smaller peak |
| Tangent skim | A small peak on the tail of a large one is cut off along a tangent to the tail | Assigns the tail to the main peak and gives the small peak less area |
| Exponential or Gaussian skim | The tail of the large peak is modelled as a curve beneath the small one | Often more realistic, but depends on the model chosen |
None of these is universally right. Each is a convention, and a good method states which one it uses so the same rule is applied every time.
Shoulders and hidden impurities
A shoulder is a visible bulge on the side of the main peak, usually a closely related impurity that the method has only partly resolved. Splitting it off as a separate peak or folding it into the main one is often left to software defaults, and folding it in always raises the purity figure.
A hypothetical example shows the scale. Suppose the main peak has an area of 980 units and the separate impurities total 20, giving 98.0%. If a shoulder worth 6 units had been cut from the main peak and counted separately, purity would read about 97.4%. The raw data are the same; only the integration differs.
Some impurities produce no shoulder at all because they co-elute with the main peak. No integration choice can reveal those. Our piece on co-elution and peptide purity covers how they are detected.
How much precision a purity figure can carry
Because integration contributes to the uncertainty of the result, differences of a few tenths of a percent between two reports are often within the range that different integration choices alone can produce. Reading 98.4% and 98.7% as meaningfully different is usually a mistake unless both came from the same method, instrument and integration rules. Our article on measurement uncertainty explains how that range is estimated.
It also explains why the chromatogram itself is more informative than the percentage. A number cannot be checked; a trace can be re-examined, and anyone with doubts can see where the baselines were drawn and how shoulders were handled.
Consistent integration across a large order
For a lab evaluating many lots of the same compound, a few habits make chromatographic comparisons fair:
- Keep the chromatogram image, not only the percentage, for each lot in your records.
- If you run your own analysis, fix the integration parameters in a written method and apply them to every lot, avoiding manual adjustments unless they are documented.
- Run a blank and exclude only the peaks that appear there, so system artefacts are not counted as impurities.
- When comparing a new lot with an earlier one, re-integrate both under the same rules instead of comparing the printed numbers.
- Note any shoulder or unusual peak shape in your inventory log against the lot, even if the percentage looks normal.
Integration affects proportions only. It does not change whether a measured mass matches the sequence, and it says nothing about endotoxin or sterility, which require separate tests. Bulk Peptides products are sent for third-party HPLC purity testing, and certificates for some products, with their chromatograms where provided, are available on our certificates of analysis page.
This article discusses analytical methods for research laboratories. Bulk Peptides compounds are intended for in-vitro study only and must not be used in humans or animals.
