A purity figure on a peptide report looks like a property of the sample. It is closer to a property of an experiment performed on the sample. Change the column, the gradient, the detection wavelength or the way an analyst draws the baseline, and the same vial can return a noticeably different number. None of those results has to be false. Each one answers a slightly different question.
That distinction matters to anyone comparing reports from two suppliers, two laboratories or two production batches. A bare percentage invites comparison. A stated method is what makes comparison possible.
What an HPLC purity number actually is
Most peptide purity figures come from reverse-phase high-performance liquid chromatography. The sample is injected onto a column, usually packed with a C18 material, and pushed through it with a gradient of water and an organic solvent such as acetonitrile, typically with an ion-pairing acid like trifluoroacetic acid in the mix. Components leave the column at different times according to how strongly they stick to it, and a detector records a signal as each one passes.
The result is a chromatogram, a trace with a tall peak for the main compound and smaller peaks for everything else. The purity on the certificate is usually an area percentage: the area of the main peak divided by the total area of all integrated peaks, multiplied by one hundred.
That calculation carries three quiet assumptions. Every impurity that matters appears as its own peak. Every component produces a signal in proportion to its amount. And the software counted the right peaks. Each assumption can fail, and the remaining sections show how.
Wavelength decides what the detector can see
The detector in a typical setup measures ultraviolet absorbance at a chosen wavelength, and that choice changes the picture. Peptide bonds absorb strongly near 214 nanometers, so a trace recorded there sees almost every peptide-related species in the sample. The side chains of aromatic amino acids such as tryptophan and tyrosine absorb at longer wavelengths, in the neighborhood of 280 nanometers, so a trace recorded there sees mainly the molecules that contain them.
Consider a sequence with few aromatic residues. At 214 nanometers the main peak and its impurities all appear. At 280 nanometers the main peak may be small and several impurities may vanish, because they have no signal to record. Reading the 280 nanometer trace, a lab can report a higher purity than the 214 nanometer trace would give for the very same injection.
There is a second wrinkle. Area percent treats every unit of area as equal, but molecules differ in how strongly they absorb light. An impurity that gives a weak signal can hide a larger share of the real mass than its area suggests. A responsible report therefore states the wavelength and treats the percentage as a statement about detected signal, not about mass.
The gradient and the column pull peaks apart, or fail to
Synthetic peptides come with characteristic impurities. Some are chains that lost one amino acid during assembly. Some are truncated, some retain protecting groups, some contain oxidized residues, and some are stereoisomers of the target. Most of them are close relatives of the main product, which means they behave almost the same way on a column and emerge nearly at the same time.
Here the method matters a great deal. A steep gradient changes the solvent mix quickly and compresses everything into a narrow window, so near-identical molecules can overlap. When an impurity co-elutes with the main peak, the integrator sees one peak, and the reported purity is higher than the true value. A shallower gradient spreads the peaks out and can reveal the hidden neighbors, which pulls the number down.
The column chemistry, pore size, temperature, mobile phase acidity and choice of ion-pairing agent all shift how molecules separate. Two validated methods can therefore disagree, and the disagreement is information. It says the sample contains something one method can resolve and the other cannot.
Integration is a human decision in disguise
After the run, software decides where each peak starts and ends. Baselines drift during gradients, small shoulders sit on larger peaks, and blank injections show peaks that come from the solvent and the system and have nothing to do with the sample. Analysts set thresholds that determine which peaks count and which are ignored as noise.
Those settings are choices. Raising the threshold removes small impurity peaks from the total and raises the purity. Lowering it counts more of them and lowers it. Manually adjusting a baseline can move the result as well. Two analysts working from the same raw data file can produce different purity values if neither follows a fixed integration rule.
Well-run laboratories write the integration parameters into the method, record any manual changes and keep the raw file. That documentation is the difference between a number and a result.
Purity answers only one of four questions
Even a perfectly run chromatogram addresses a single question, which is how dominant the main detected component is. Identity, the amount of material present and the concentration in a given volume are different measurements that need different methods. A plain-language explainer on purity, identity and mass from one supplier’s research library separates the four concepts and shows why a result for one does not stand in for another. The takeaway for a reader is to find out which of the four questions a given report answers before drawing a conclusion from it.
How a careful lab checks its own number
Analytical chemists use several habits to avoid fooling themselves. They run an orthogonal method, a second separation that works on a different principle, such as a different column chemistry or a different mobile phase acidity, and compare the purity values. They use a diode-array detector to compare the ultraviolet spectrum across the width of the main peak, since a spectrum that changes from one edge to the other suggests that something is hiding underneath. They couple the separation to a mass spectrometer to identify what each peak is.
They also validate the method itself. Guidance such as the ICH Q2 framework on validating analytical procedures covers specificity, precision, linearity and sensitivity to small changes in conditions, and the general chromatography chapter of the United States Pharmacopeia sets system suitability requirements such as resolution and repeatability. A method that has been tested against these questions earns more trust than one that simply produced a clean-looking trace.
Not every supplier publishes that level of detail, and the gap between a number and the story behind it is where comparison breaks down. The research library at USA Peptide Depot is one example of a supplier explaining how to read certificates of analysis and what a given test can and cannot establish, which gives buyers vocabulary for the right questions.
What a method-complete report should state
A short explainer on what HPLC can and cannot tell you makes the central point plainly: the technique is powerful, and the meaning of its output depends on how it was run. A report that lets a reader judge the result usually includes the items below.
- The detection wavelength used for the purity calculation.
- The column type and the gradient, or a reference to the method used.
- The integration approach, including whether peaks were manually adjusted.
- Evidence that co-elution was checked, by a second method, a spectral comparison or mass spectrometry.
- The lot number, the test date and the laboratory that ran the analysis.
- The raw or annotated chromatogram, not only a summary line.
Any item that is missing is not proof of a problem. It is a gap that limits how far the number can be trusted or compared.
Five questions to ask when two reports disagree
- Were both purity values calculated at the same wavelength?
- Were the columns and gradients comparable, or did one method separate close impurities better?
- Were the integration rules fixed in advance, and were any peaks adjusted by hand?
- Was co-elution checked in either report?
- Do both reports name the same lot and a date close to the date of analysis?
The next time two certificates give different purity values for what looks like the same material, resist the urge to decide that one laboratory is wrong. Ask for both chromatograms and compare the wavelength and the gradient first. More often than not, the answer is sitting in the method section, and once you can see it, the two numbers stop being a contradiction and start being a measurement you can use.
