A second peak is not automatically an impurity — it can be the same molecule in a different form, an artefact of the injection, or something that was in the solvent rather than the sample. Working out which costs a few minutes of looking and saves concluding something wrong about a lot. The commonest mistake is treating every extra peak as contamination; the second commonest is treating a single peak as proof of purity.
What a second peak can be
| Cause | Where it appears | Tell |
|---|---|---|
| Solvent front / injection artefact | Very early, at the void volume | Present in a blank injection too |
| Truncation or deletion sequence | Near the main peak | Lower mass by one or more residues |
| Oxidation product | Slightly earlier than the main peak | +16 Da per oxygen; more polar |
| Deamidation product | Close to the main peak | +1 Da; often a shoulder rather than a peak |
| Disulphide isomer | Can be anywhere | Identical mass, different structure |
| Dimer or aggregate | Usually later | Roughly double the mass |
| Carryover from a previous run | Anywhere | Shrinks on a repeat injection |
Only some of those are impurities in the sense of foreign material. A disulphide isomer is the same atoms in a different arrangement; an oxidation product was your peptide until recently. Both count against purity by area, and they mean different things about the batch.
How do I rule out the easy explanations first?
Inject a blank. A peak present in a solvent-only run is not in your sample, and this single test eliminates solvent-front artefacts, carryover and anything leaching from the vial or the column. It costs one run and it is the first thing to do.
Then re-inject the sample. A peak that shrinks on the second injection is carryover from whatever ran before. A peak that stays proportional is real. These two runs between them account for a large share of mystery peaks and neither requires any interpretation.
Does retention time tell me what it is?
It narrows the field. Reversed-phase separates by hydrophobicity, so anything more polar than the target elutes earlier and anything less polar elutes later. Oxidation adds oxygen and increases polarity, so oxidation products run early — often as a shoulder on the leading edge of the main peak rather than a resolved peak.
Truncations are less predictable, because removing a residue changes hydrophobicity in whichever direction that residue pointed. Losing a hydrophobic residue makes the fragment more polar; losing a charged one does the opposite. Reading the trace gives a hypothesis, not an identification.
What actually identifies a peak
Mass. Retention time is a property of the method; mass is a property of the molecule, and the difference between the two peaks' masses usually names the problem outright.
A difference of 16 is an oxidation. A difference of 1 is a deamidation. A difference matching one residue's mass is a deletion, and the residue identifies where in the synthesis it happened. A difference of roughly double is a dimer. Zero difference, with two peaks, means an isomer — same composition, different structure.
That last case is the one a mass check cannot resolve, and it is why a molecule with more than one way to close needs a method that looks at connectivity rather than composition.
Is a shoulder worse than a resolved peak?
Harder to deal with, and often more consequential. A resolved peak can be integrated separately and reported honestly. A shoulder sits inside the main peak's integration window, and where the baseline is drawn decides how much of it is counted as product.
That is a judgement call made by whoever integrated the trace, and it is one of the reasons two laboratories can report different purity for the same sample. A certificate reporting a clean 99% from a trace with a visible shoulder has made a defensible choice that another analyst might not have made.
Does a single peak prove the sample is pure?
No, and this is the symmetric error. A single peak means nothing else was resolved under those conditions, which is a statement about the method as much as the sample. A co-eluting impurity is invisible; so is anything that did not come off the column within the run time.
Running a longer gradient, or a second method with different selectivity, is how that gets tested. A multi-receptor compound where a truncation might retain activity at one target is exactly the case where one clean trace is weaker evidence than it looks.
What to do with the answer
The response depends on what the peak turned out to be, and over-reacting is as costly as ignoring it.
An artefact or carryover. Nothing to do beyond noting it. The lot is unaffected.
An oxidation or deamidation product. This is degradation, and the question becomes when it happened. Present at release, it is a manufacturing matter. Appearing later, it is a storage matter and the date on the certificate is the reference point.
A truncation. Set at synthesis and will not change. The only question is whether the proportion is within specification and whether the impurity plausibly interferes.
A dimer or aggregate. Often reversible and often a handling consequence rather than a batch property. Worth re-running a freshly prepared sample before concluding anything about the material.
Should an extra peak be raised with the supplier?
If it appears on their certificate's own trace, yes — ask what it is, because they may already know. A manufacturer who has characterised their impurity profile can usually name the main one.
If it appeared on your own re-analysis of material that passed at release, the more useful question is about storage and handling on your side before it becomes a supplier conversation. A peptide that was clean at release and shows an oxidation product a year later is behaving normally, not defectively.
Does comparing two lots help?
Considerably, and it is underused. Running two lots of the same compound on the same method, same day, same column, produces a direct comparison with every methodological variable held constant — which is the one condition under which two traces can be compared confidently.
A peak present in both is a process characteristic. A peak present in one is a batch event. That distinction is cheap to establish this way and almost impossible to establish from two certificates issued months apart by different laboratories.
Building a habit around it
The useful discipline is to look at the trace before looking at the number, every time. A purity figure read in isolation carries no information about peak shape, resolution or run length, and all three change what the figure means.
Three things are worth noting whenever a trace is examined: whether the main peak is symmetrical, whether anything sits on its shoulder, and whether the gradient ran long enough for late material to emerge. Those take seconds and they are the difference between reading a chromatogram and glancing at one.
Should traces be kept?
Yes, and it costs nothing. A stored chromatogram can be re-examined when a question arises later; a stored number cannot. Where a discrepancy surfaces months after the fact, having the original trace is frequently the difference between resolving it and speculating about it.
The same applies to keeping the certificate itself filed alongside the experimental record rather than in an inbox. The link between document and data is what makes either one useful.
What if the supplier will not provide the trace?
It is a reasonable request and a reluctance to meet it is informative. A laboratory that ran the analysis has the chromatogram; producing it is a file transfer rather than work.
Where only a number is available, treat the figure as a claim rather than a measurement and weigh it accordingly. A result without a method cannot be compared to anyone else's, and a result without a trace cannot be examined at all.
Does a second peak always reduce the purity figure?
Only if it was integrated. A peak dismissed as solvent front, or falling outside the integration window, does not enter the calculation at all — which is one of the ways two analysts produce different numbers from the same trace.
That is why reading the figure alongside the trace matters more than reading either alone. Purity by area normalisation is the main peak divided by everything counted, and what was counted is a decision rather than a fact.
What is the single most useful habit here?
Running a blank injection before concluding anything. It is one run, it costs nothing beyond the time, and it eliminates the three commonest explanations for a mystery peak at once — solvent front, carryover and anything leaching from the vial or column.
Everything after that is interpretation. The blank is the only step that produces a definite answer about whether a peak belongs to the sample at all, and skipping it is how a batch gets wrongly suspected.
All products referenced here are supplied for laboratory and research use only. They are not drugs, foods, supplements or cosmetics, and are not for human or veterinary use.
