Purity: what HPLC measures

With the molecule confirmed (lesson 2), the next question is how clean the sample is. A purity figure never names the molecule (lesson 2), never counts milligrams (lesson 4) and never weighs salt or water (lesson 5). This lesson looks at what it does measure, and at the four places it is made: separation, detection, integration and the reporting threshold.

What happens to the sample inside an HPLC system?

A pump pushes a liquid, the mobile phase, through a narrow steel column packed with tiny coated particles, the stationary phase. A measured volume of dissolved sample is introduced into the flowing liquid just before the column. Each molecule in the sample spends part of its time stuck to the particles and part of its time carried along. Molecules that stick more travel more slowly and leave the column later.

For peptides, the usual arrangement is reversed phase: the particles carry long oily chains (C18, eighteen carbons long, is a typical choice) and the mobile phase starts mostly watery, then becomes steadily richer in an organic solvent such as acetonitrile. This rising mix is called a gradient. A published reference-standard method for the peptide leuprolide, for example, used a C18 column and monitored peptides and impurities at 220 nm (McCarthy et al., Pharmaceutical Research, 2023). More water-loving peptides leave early; more oil-loving ones leave late. A near copy of the target, missing one residue, often behaves slightly differently and leaves slightly earlier or later.

The time a molecule takes to emerge is its retention time. Under fixed conditions it is reproducible, which is why a retention time match against a reference standard counts as one identity check. The HPLC-UV method page lists typical columns and conditions.

What does the UV detector actually see?

At the column exit, the liquid passes through a small cell where ultraviolet light shines across it. Molecules that absorb light at the chosen wavelength reduce the light reaching the sensor, and the instrument records that as signal. Peptide bonds absorb at low UV wavelengths, which is why peptide methods monitor in the region around 210 to 220 nm.

Two consequences follow. First, area percent covers only the peaks that are integrated. Water gives no peak, and counter-ions are not part of the peptide peak table, so they need methods of their own: a published mass balance study of synthetic glucagon measured water by Karl Fischer titration and trifluoroacetic acid by ion chromatography, separately from the peptide impurities (Wang et al., Scientific Reports, 2020). Second, different molecules absorb different amounts of light per milligram. Area percent assumes, in effect, that impurities and the main peptide respond similarly, which is close to true for near copies of the same peptide and less true for unrelated material.

How is the purity percentage calculated, step by step?

The software draws a baseline under the signal and measures the area of each peak between its start and end points. This is integration. The analyst then reports area percent: each peak's area divided by the total area of all integrated peaks, times 100. The main peak's area percent is the purity.

The peak table below is invented for teaching and comes from no real sample.

Purity is 486,000 divided by 496,750, which is 97.84%. Relative retention time is each peak's retention time divided by the main peak's, so peak 3 at 13.4 minutes sits at 1.04, just after the main peak. Impurities close to 1.00 are the hardest to separate and the most likely to be near copies of the target.

What is a reporting threshold, and why does it matter?

Not every bump in the signal is reported. Methods set a reporting threshold, below which peaks are left out of the table. For synthetic peptides, the European Medicines Agency's guideline states that, under the Ph. Eur. general monograph on substances for pharmaceutical use, peptide-related impurities "should be reported above 0.1%, identified above 0.5% and qualified above 1.0%", and that methods should be able to meet the 0.1% reporting threshold (EMA, synthetic peptides guideline, 2025).

Suppose the illustrative run also had five tiny peaks of 400 counts each. Each is about 0.08% of the new total of 498,750, below a 0.1% threshold, so none is listed. Together they are 2,000 counts, about 0.4%: counted, they would take purity from 97.84% to 97.44%. Leaving them out is legitimate if the threshold is stated. A report that gives purity without its threshold leaves you unable to tell how much was set aside.

What can hide inside the main peak?

An impurity that leaves the column at almost the same time as the target can merge into the main peak. The software then counts it as product, and purity looks higher than it is. The EMA guideline asks method developers to take into account "the risk of co-eluting impurities" and says that when co-eluting impurities are observed as one peak, the 1.0% qualification threshold applies to them unless otherwise justified (EMA, synthetic peptides guideline, 2025).

ICH Q2(R2) gives the standard way to show a method separates what it should: "for critical separations in chromatography, specificity can be demonstrated by the resolution of the two components which elute closest to each other" (ICH Q2(R2)). A laboratory with a validated method has that evidence on file. A buyer usually sees only the result, which is one reason identity by mass spectrometry is a useful partner to HPLC: it can show whether the main peak contains more than one mass.

How often does a printed purity figure hold up?

Not always. In a study of 98 synthetic peptides, every one arrived with a supplier certificate stating at least 95.0% purity, but only 43 met that figure when the researchers ran their own analysis. Of the impurities they could identify, 74% involved at least one missing amino acid: exactly the near copies that elute close to the main peak (Verbeke et al., Journal of Pharmaceutical Analysis, 2015). A purity figure is a measurement only when you know which sample, which method and which laboratory produced it.

Why can two laboratories report different purity for the same vial?

Because each of the four steps involves a choice. None is wrong if validated and stated, but purity figures are only truly comparable when they come from the same method.

For any peptide, semaglutide included, a stated method and threshold matter more than an extra decimal place in the purity figure. The chromatogram tool recalculates area percent from a peak table. Next, the same HPLC run is used to count milligrams: Amount: content vs label.

Check your understanding

1. Why can water and counter-ions not lower an HPLC-UV purity figure? Purity is calculated only from the integrated peptide peaks. Water gives no peak and counter-ions are not counted in the peak table, so they need separate tests.
2. A report shows 98.9% purity with no reporting threshold stated. What can you not tell? How much area from small peaks was left out of the total, which could move the figure.
3. What does a relative retention time of 1.02 suggest about an impurity? It elutes just after the main peak, so it is likely a close relative of the target and at higher risk of merging into the main peak.