How to Read an HPLC Chromatogram

The moment the plot stops being decorative

A purity figure on a certificate of analysis is a summary of a decision someone made while looking at a chromatogram. If you never see the trace, you are accepting both the measurement and the judgment that produced it. Learning to read the plot lets you separate the two, which is the difference between trusting a supplier and trusting a specific document.

The goal here is not to turn a procurement reviewer into a chromatographer. It is to make the graphic legible enough that you can tell a clean, well-resolved separation from a crowded one, and to know which follow-up questions expose a weak result. Everything below assumes reverse-phase HPLC with UV detection, the most common configuration for peptides.

What the two axes are actually telling you

The horizontal axis is retention time, usually in minutes, measured from injection. It reflects how strongly each component interacts with the column under the specific gradient, temperature, and flow rate of the method. Retention time is a property of the method as much as the molecule, so a peak at 9.2 minutes means nothing without the method that produced it.

The vertical axis is detector response, commonly milli-absorbance units for UV. Response scales with concentration and with how strongly a compound absorbs at the chosen wavelength. That second point matters: two impurities present at the same molar amount can produce very different peak heights if one absorbs strongly and the other barely absorbs at all. A quiet baseline is reassuring, but it is not proof that nothing is there.

Baseline: the reference everything is measured against

Peak area is measured relative to the baseline, so the baseline is not background detail — it is the ruler. A flat, stable baseline that returns to roughly the same level before and after the peaks suggests a controlled run. Drift, in which the baseline climbs or sags across the gradient, is common with UV gradients and is usually corrected by the software, but heavy drift can swallow or inflate small peaks depending on where the integration draws its line.

Noise sets the floor for what can be seen at all. Anything smaller than the noise band is invisible, which is why a genuinely honest report often lists a reporting threshold: impurities below a stated level are neither integrated nor claimed. Absence of a peak is not the same as absence of a compound; it can simply mean the compound is below the threshold of the method.

Integration is a decision, not a measurement

Integration is the step where the software — guided by an analyst — decides where each peak begins and ends and draws a baseline underneath it. For a tall, symmetric, well-separated peak, that decision is nearly automatic and hard to get wrong. For two peaks that touch, it is a judgment call, and the two most common approaches give different answers.

When peaks are not fully resolved, an analyst chooses between dropping a vertical line to the baseline at the valley between them or drawing a tangent skim across a shoulder. The choice shifts area from one peak to the other and therefore changes the reported purity. This is legitimate, routine, and method-defined — but it is also exactly why a purity percentage without the underlying trace is a weaker claim than it appears.

• Perpendicular drop splits overlapping peaks at the valley; it tends to over-assign area to a small peak riding next to a large one.
• Tangent skim treats a shoulder as sitting on the tail of a larger peak; it can under-assign that shoulder.
• A visible manual integration mark is not a red flag by itself, but it is a reason to ask how the method defines the rule.

Reading the main peak and the neighbors

The principal peak is normally the tallest and is reported as a percentage of total integrated area — area normalization. A purity of, say, the high nineties by area means the principal peak accounts for that share of everything the detector counted, not that the vial is that fraction peptide by mass. Those are different measurements, and conflating them is a frequent source of confusion.

The neighbors matter more than their size suggests. A cluster of small peaks eluting just before the main peak is a familiar pattern for deletion sequences and related process impurities. A single small peak eluting slightly later can point to an oxidation product. Peak shape is a clue too: a sharp front with a long tail (tailing) or a slow rise with a sharp back (fronting) can indicate column or method stress rather than a real second species.

System suitability: proof the instrument earned the run

Before a result counts, the method should demonstrate that the system was performing correctly on that day. System suitability is a set of pre-defined checks run with a reference or standard, and a chromatogram tied to a compliant method should be backed by suitability that passed. If suitability failed, the sample data is not interpretable regardless of how clean the trace looks.

Common suitability parameters include resolution between a critical pair of peaks, theoretical plates or peak efficiency, tailing factor, and the repeatability of replicate injections. USP General Chapter 621 and ICH Q2(R2) define these terms and expectations. You do not need to memorize the formulas; you need to know that the numbers exist and to ask whether they were met for the batch in question.

• Resolution describes how cleanly the critical pair separates; low resolution undermines the purity split.
• Tailing factor flags peak asymmetry that can distort integration.
• Injection repeatability shows the run was stable, not a lucky single injection.

Artifacts that masquerade as data

Not every feature on a trace is the sample. The injection itself can produce a solvent front or an early disturbance near the void time; a peak eluting there is often the sample diluent, not a component of interest. Air bubbles, pressure spikes, and detector saturation can all create shapes that look like peaks or like missing peaks.

Detector saturation is worth calling out because it silently corrupts purity. If the principal peak is so concentrated that the detector flattens at the top, its true area is underestimated and every impurity looks proportionally larger. A flat-topped or clipped principal peak is a reason to question the concentration and the result, not to read the impurities more strictly.

A schematic is not a batch record

It is worth stating plainly: an illustrative chromatogram — the kind used to teach or to decorate a page — represents nothing about any real lot. Only a batch-specific trace, tied to a lot number, an analytical date, and a named method, carries information about material you might buy. Treating a generic figure as evidence of a specific batch is a category error, and a supplier that offers one in place of the other is answering a different question than the one you asked.

When you evaluate a real result, the trace and the certificate should agree with each other and with the method. Compound name, lot number, wavelength, gradient, and analytical date should be consistent across documents. Disagreements between the summary number and the trace are the single most useful thing an untrained reader can catch.

The short list of questions to ask

You can get most of the value of chromatographic literacy from a handful of questions, even without deep expertise. They move the conversation from adjectives to evidence and tend to reveal quickly whether the documentation is real.

• Is this trace batch-specific, with a lot number and analytical date that match the certificate?
• What wavelength and gradient were used, and does purity mean area percent under those conditions?
• Did system suitability pass for this run, and what were the resolution and tailing values?
• What is the reporting threshold, and how were overlapping peaks integrated?