Reading an HPLC Chromatogram

Start with the run and the axes

Check the sample ID, run or injection identifier, acquisition date and report revision. Confirm that the trace and peak table refer to the same injection and detector channel. If the report includes several channels, identify which one supports the reported result.

The horizontal axis usually gives time after injection. Retention time locates a peak within that run. The vertical axis gives detector response in the stated units; UV traces may use absorbance units or milli-absorbance units. Detector settings and separation conditions affect what you can compare between runs. [A7]

A tightly cropped image can omit early signals, late signals or the baseline context. Ask for the full relevant run and its axis scales. The HPLC explanation covers separation and detector response in more detail.

Read the educational trace

The example contains three separated peaks. A, B and C are labels for signals, not chemical identities. The shaded region shows the area above the zero baseline for B. All times and areas are invented.

The table totals 100 area units. B therefore accounts for 92% of the included area. That arithmetic does not turn B into an identified compound, establish 92% by mass or assign an identity to A or C. The labels deliberately leave those questions open.

Find B in the plot, follow it down to 5.0 minutes, then compare its entry with the table. Repeat for the two smaller signals. This checks correspondence between the display and reported entries before you assess a chemical claim.

Inspect the integration beneath the curve

An integrator measures area relative to an assigned baseline between peak boundaries. It can divide overlapping signals using different integration rules. Reporting thresholds can also exclude small integrated peaks. Agilent's data-analysis documentation illustrates these choices. [A6]

On a real report, look for baseline and start/end markers. If the trace shows a shoulder or two signals without a clear baseline between them, ask how the analyst split or combined their areas. A shoulder warrants explanation; its appearance alone does not name an impurity.

A screenshot without integration markers gives you less information than the trace together with the processed peak table and method. Ask which peaks entered the percentage calculation and whether the analyst excluded solvent, blank-related or other signals. Do not infer those choices from the height of the largest peak.

A clean-looking peak still needs supporting evidence

A large peak can contain unresolved components. ICH Q2(R2) treats discrimination from interfering substances as part of analytical specificity or selectivity, and discusses evidence for critical chromatographic separations. [A8] A peak's appearance therefore cannot settle identity or absence of interference.

A missing peak also needs context. The detection limit describes the lowest amount a procedure can detect; it does not establish that an undetected substance is absent. [A8] Ask about the relevant detection or reporting limit if the result depends on a small component.

Use common impurities for the classes of material an investigation might consider. Assigning one of those names to a signal requires analytical evidence. The distinction between identity, purity and assay remains relevant even when the report includes a chromatogram.

Write a conclusion the record supports

A useful reading note identifies the sample and run, the detector channel, the assigned main peak, the integration basis and any unresolved question. For the teaching example, the conclusion is: “Signal B has 92% of the three included areas; this example supplies no chemical identification or lot evidence.”

For an actual result, compare the stated calculation with the agreed specification through acceptance criteria and out-of-specification results. Do not invent a universal passing percentage from the plot. If you lack the trace, method or batch connection, record that gap when using the quality documentation route to request the missing record.