Quality control
Reference: reading an HPLC purity chromatogram
9 November 2026 · 5 min read
A certificate of analysis usually leads with one number: purity by reversed phase HPLC, expressed as area percent. The number is useful, but the trace behind it answers more of the questions a laboratory actually has about a batch than the figure printed above it does.
What reversed phase HPLC separates
Reversed phase chromatography sorts molecules by how strongly they hold onto a non-polar stationary phase, commonly silica with C18 chains bonded to its surface. A mobile phase of rising organic strength washes them off in order, the most polar species first. For peptides the gradient typically runs from mostly water towards acetonitrile, with a small amount of an acidic modifier such as trifluoroacetic acid to hold the pH low and keep the peaks sharp.
Peptides suit the technique for two reasons. They are built from residues of very different polarity, so two chains differing by a single residue usually differ measurably in how long they stay on the column. And the peptide bond absorbs ultraviolet light strongly near 214 nm, so a detector set there responds to the backbone rather than to one side chain. A second channel near 280 nm sees only the aromatic residues, tryptophan and tyrosine, which is why two traces from the same injection do not look alike.
Reading the trace in order
The horizontal axis is time since injection, the vertical axis is detector response. Work through it in this sequence.
- Check the conditions. Column, gradient, flow rate, detection wavelength, run length. A chromatogram without them cannot be compared with any other chromatogram.
- Find the injection front at the very start. Anything unretained leaves in the void volume: salts, residual solvent, the counterion. A peak there is usually not a peptide related species.
- Find the main peak and note its retention time. Retention time is the only identity claim the trace makes: the same substance on the same column under the same gradient should come off at the same time from batch to batch.
- Look at the baseline before and after the main peak, not only at the peaks.
- Look at the shape of the main peak.
- Only then read the integration table.
Why area percent is not mass percent
The detector reports absorbance, not mass. Converting one into the other requires a response factor, and every species has its own. Three consequences follow.
Species without a chromophore at the detection wavelength are invisible. Water and inorganic salts contribute nothing. A lyophilised peptide salt can be 99 percent by area and still carry a meaningful fraction of its weight as water and counterion, which is normal for the material rather than a defect.
Related substances do not respond identically to the parent. An impurity that has lost an aromatic residue absorbs less per unit mass, one carrying an extra aromatic group absorbs more. Area percent flatters some and understates others.
Anything that does not elute within the gradient is not counted at all; normalisation runs only over what the integration window includes.
Peptide content by mass is a separate determination: water by Karl Fischer titration, counterion by its own assay, peptide by amino acid analysis or nitrogen determination. What each batch is tested for is set out under quality.
A cluster of peaks next to the main peak
Closely spaced peaks flanking the main peak are the signature of the synthesis, not of dirt. They are chemically related to the target and that is exactly why they elute near it.
As a rough orientation, species more polar than the target tend to elute earlier: oxidation products, deamidated forms, hydrolysis products. Species that are more hydrophobic tend to elute later, and a retained protecting group is the usual reason. A peak sitting almost on top of the main peak, resolved only as a shoulder or a notch, commonly means a single residue deletion or a single altered stereocentre, because one changed position in a long chain barely changes overall polarity.
A lone peak far from the main peak, with a different width or shape, points elsewhere: carryover from a previous injection, a blank contribution, or something leached from a container.
Why the shape matters more than the figure
| What the trace shows | Common cause | Effect on the number |
|---|---|---|
| Shoulder on the main peak | Unresolved co-elution | Area split by an arbitrary drop line |
| Flat topped main peak | Detector saturation or overload | Main peak area understated |
| Long tail | Secondary interactions or overload | Small late peaks buried in the tail |
| Noisy baseline | Low signal, old lamp, mobile phase quality | Smallest reportable peak larger than stated |
| Rising baseline | Normal gradient drift from the modifier | None, if integration accounts for it |
| Step or dip in the baseline | Air bubble or detector fault | Integration unreliable across that region |
A shoulder is the clearest example. Two species under one envelope can be integrated as one peak and reported high, or resolved by a longer gradient and reported lower. Nothing about the material changed, only the method.
A short checklist
- Conditions are stated: column, gradient, wavelength, flow, run time.
- The main peak elutes well inside the gradient and is not cut off at the end of the run.
- The main peak is not flat topped.
- The baseline returns between peaks and the noise sits visibly below the smallest reported peak.
- The integration threshold and limits are stated.
- Retention time is consistent with earlier batches of the same substance.
Terms used here are defined in the glossary.