Peak Purity: When One Peak Is Actually Two
A chromatogram shows when something left the column. It does not show how many things left at that moment.
One peak on a chromatogram does not prove one compound. Two or more substances can leave the column at the same moment, and the detector reports them as a single peak. This is called coelution. A photodiode array detector can sometimes reveal it by comparing spectra across the peak, but that check has limits of its own, and a pass is not proof.
This article explains how coelution happens, what a peak purity check actually measures, and why a clean result on one method should be read as one piece of evidence rather than a conclusion. It follows on from how to read a chromatogram and peak area and purity.
What a peak actually represents
In liquid chromatography, a sample is pushed through a column. Different compounds move at different speeds, depending on how strongly they interact with the column material. The detector records a signal as each compound emerges. A peak is a plot of that signal against time.
The peak tells you three things: when something emerged, how much signal it produced, and how wide the band was. It does not tell you what the substance is. It does not tell you whether the band held one compound or several.
That distinction matters. Reported purity from chromatography is calculated from peak areas. If a hidden impurity sits inside the main peak, its area is counted as main compound. The number looks better than the sample deserves.
How coelution happens
Separation depends on selectivity, which is how differently two compounds behave in a given column and mobile phase. When two compounds behave almost identically, they emerge together.
This is common with peptides. Many related substances differ from the target by one amino acid, a small modification, or a change in stereochemistry. Examples include a deletion sequence, an oxidised residue, a deamidated residue or an isomer. These share most of their structure with the main compound, so they interact with the column in nearly the same way.
Coelution takes several forms:
- Complete overlap. The two compounds emerge together and the peak looks normal.
- Partial overlap. The peak shows a shoulder, a slight asymmetry or a wider base than expected.
- Hidden under a large peak. A small impurity buried under the apex of a much larger peak adds little visible distortion.
The first and third are the dangerous ones. Neither produces an obvious visual warning.
Why single wavelength detection cannot show it
A single wavelength detector measures absorbance at one chosen wavelength. At each moment it produces one number. If two compounds both absorb at that wavelength, their contributions sum into one trace.
With one number per moment, there is no way to ask whether the signal came from one compound or two. The trace has no information about composition, only about total absorbance. A reader can examine the peak shape, but shape is indirect evidence at best. See UV detection wavelength for how the choice of wavelength shapes what a chromatogram can show.
How a photodiode array peak purity check works
A photodiode array detector, also called PDA or DAD, records a full UV spectrum many times across each peak instead of one wavelength. This adds a second dimension to the data: absorbance across wavelength, at every point in time.
The logic of a peak purity check is simple. A pure compound has one UV spectrum. If the peak is a single compound, the spectrum taken on the upslope, at the apex and on the downslope should be the same shape, regardless of how the intensity rises and falls. If a second compound with a different spectrum is present, the spectral shape changes across the peak, because the mix of the two changes from front to back.
Software quantifies this by comparing spectra. Different vendors use different terms. Common ones are:
- Purity angle. A measure of how different the spectra across the peak are from the apex spectrum, treating each spectrum as a vector. A larger angle means more spectral difference.
- Purity threshold. A limit based on the spectral differences expected from noise and baseline effects alone. In some software it is built from a noise component plus a solvent component.
- Match factor or similarity. An alternative expression of the same comparison, scored so that a higher value means more similar spectra.
In the angle and threshold approach, a purity angle below the threshold means that the spectral variation is within what noise would explain. A purity angle above the threshold means the spectra differ by more than noise would explain, which is a flag for possible coelution. Software also usually draws a purity plot across the peak, which lets an analyst see where along the peak the change occurs.
Reading a result with care
The result is a statement about detectable spectral change. It is not a statement about the number of compounds.
A few practical points follow from how the calculation works.
- Signal size matters. At low absorbance, noise dominates the comparison. At high absorbance, detector response can depart from linearity and distort the spectra. For this reason the peak is best kept within a suitable absorbance range, neither too faint nor saturating.
- Settings matter. Spectral resolution, bandwidth, baseline treatment and the wavelength range used all change the calculated angle and threshold.
- Tails need attention. A peak can pass on a single summary value while the purity plot shows a problem on the tail. Reviewing the plot as well as the number is good practice.
- A flag is not a diagnosis. A purity angle above the threshold can also come from a sloping baseline, a gradient effect or detector artefacts. It tells the analyst where to look.
What a passing result does not show
The core limit is that a spectral check can only detect what changes the spectrum. Several situations defeat it.
- Similar spectra. Related peptides often share the same chromophores, mainly the aromatic residues and the peptide bond. Their UV spectra can be nearly identical. If the impurity looks like the main compound to the detector, the spectrum does not change across the peak.
- No useful absorbance. A compound that does not absorb at the wavelengths monitored produces no spectral signature to detect.
- Low level. A small impurity under a large peak may shift the spectrum by less than the noise.
- Near perfect overlap. If two compounds emerge at exactly the same time, the mix is the same on every part of the peak. Spectra across the peak match each other, even though the peak is not pure. The check compares the peak with itself, so a constant mixture passes.
It is well recognised in chromatographic practice that a diode array purity check can pass while a second, orthogonal separation reveals a coeluting impurity. These are the cases the check was never designed to catch.
Orthogonal methods
The standard answer to the limits of one method is a second method that works differently. Orthogonal means that the separation or detection depends on different properties, so an impurity hidden by one is likely to be exposed by the other.
The common approaches are:
- Change the column chemistry. A stationary phase with a different retention mechanism, for example a phenyl or polar embedded phase rather than another C18, holds compounds by different properties. Compounds that coelute on one phase often separate on another.
- Change the mobile phase. Many peptides carry charges that change with pH, so a shift in pH alters retention differently for compounds with different ionisable groups. The ion pairing or acid additive, for example trifluoroacetic acid versus formic acid, the organic solvent, for example acetonitrile versus methanol, and the column temperature are the other common selectivity levers.
- Add mass spectrometric detection. A mass spectrometer reports the mass of what is eluting. If the mass spectra change across a peak, that can indicate that more than one species is present. See HPLC versus mass spectrometry.
- Two dimensional liquid chromatography. In a heart cutting arrangement, a fraction from the main peak is collected and re-injected on a second column with different selectivity. This is a recognised way to expose impurities hidden under a main peak.
Changing the gradient on the same column sits apart from these. A shallower gradient can spread peaks and resolve a shoulder into two, but it optimises the same separation rather than changing the mechanism, so it is not orthogonal.
Mass spectrometry is not a complete answer either. Isomers have the same mass and can give mass data that is identical to the main compound. Changing charge states, adducts and in source fragmentation can also move a spectrum without a second compound being present, and a high concentration of the main compound can suppress the ionisation of a small coeluting impurity, so it is not seen. This is why a second chromatographic separation remains valuable even when a mass spectrometer is available.
What this means for a certificate
A reported purity value is the result of a specific method, run on a specific sample, with a specific detector. It is conditional on all three.
Two laboratories can report different purity for the same material because their methods separate different sets of impurities. This is discussed in why labs report different purity. A higher number does not mean a cleaner sample. It can mean a method that failed to separate something.
When reading any certificate, useful questions are:
- Which detection was used, a single wavelength or a full spectrum?
- Was a peak purity check performed, and was the result reported?
- Was identity confirmed by a separate technique, such as mass?
- What does the certificate state about the limits of the method?
Purity describes composition. It is not a statement about safety. A sample that is chromatographically pure is not thereby suitable for any use, and a measurement is not a guarantee.
About Novagen
Novagen is a third-party analytical laboratory. It reports what was measured in the sample submitted, and a certificate describes that sample and nothing beyond it. Certificates are published on a verifiable register at novagenlaboratories.com, so a reader can check that a certificate exists and matches what they hold.
The same caution applies to every number on every certificate, including ours. A purity value is a result under stated conditions. Where a question depends on the detail of a method, the right step is to ask. Questions can be sent to admin@novagenlaboratories.com. Current quoted turnaround is 21 days from receipt of samples. For a wider overview of the testing process, see peptide testing.
Common questions
- What is coelution in HPLC?
Coelution is when two or more compounds leave the column at the same or nearly the same time. The detector sees their combined signal as one peak, or as one peak with a slight shoulder. Coelution is a failure of separation, not of detection.
- Can a symmetrical peak still contain an impurity?
Yes. If an impurity has almost the same retention as the main compound, the combined peak can look perfectly symmetrical. Shape is a weak indicator. A clean shape lowers the likelihood of an obvious problem but does not exclude a hidden one.
- What does a passing peak purity result mean?
It means no significant change in the UV spectrum was detected across the peak, within the limits of the instrument and the settings used. It does not prove the peak is a single compound. An impurity with a very similar spectrum, or one present at a very low level, can pass unnoticed.
- How do you confirm a peak is a single compound?
No single measurement confirms it. Analysts combine evidence: a spectral purity check, a second separation using different column chemistry or conditions, and mass spectrometric data across the peak. Agreement between methods that work on different principles is stronger evidence than any one result.
- Why does single wavelength detection miss coelution?
A single wavelength records one number per moment in time. Two compounds that both absorb at that wavelength add together into one trace. With no spectrum to compare, there is nothing to show that the signal came from more than one source.