Reading a Chromatogram: a Field Guide
ON THIS PAGE
- Key findings
- Retention time belongs to the method
- The baseline is a decision, not an observation
- Peak shape, and the three figures that describe it
- Area percent is arithmetic over what the detector saw
- What a related-substance peak can be
- A field key for the trace in front of you
- What the trace cannot show
- Sources
Key findings
- A peak means something left the column then and absorbed at the detector's wavelength. What co-elutes, absorbs elsewhere or never elutes leaves none.
- Retention time belongs to the method, not the molecule. The harmonized chapter gives relative retentions for information, not as requirements.
- USP 621 and Ph. Eur. 2.2.46 are one harmonized text: symmetry factor 0.8 to 1.8 by default, plate count defined only for isocratic data.
- ICH Q3A(R2), adopted 25 October 2006, is where 0.05, 0.10 and 0.15 percent come from. Its preamble excludes peptides by name.
- Area percent assumes every peak answers the detector equally per unit mass. Counterion, salt and water add weight no peak measures.
A chromatogram is one detector's signal plotted against time, and most of what it can say was settled before the trace was drawn: by the column, the wavelength, and where the software put the baseline. A peak is not an identity, and the percentage beside it is not a mass fraction.
Retention time belongs to the method
Change the column, the modifier, the gradient slope or the oven temperature and every number on the time axis moves. That is why pharmacopoeial impurity tables print relative retention, the peak's time over a named reference peak's: the ratio survives an instrument transfer, the raw minutes do not. The harmonized chapter goes further, stating that retention times and relative retentions are not requirements but are given for information, and the Ph. Eur. has retired the term relative retention time. Identity comes from mass spectrometry, against the calculated mass of the sequence. The expected minute proves the method ran, nothing more.
The baseline is a decision, not an observation
Area is measured against a baseline drawn by software under parameters a person chose. Where it starts and stops decides how much of a peak's edges are counted, and whether two touching peaks are split by a vertical drop or by skimming decides how their areas divide. Noise sets the floor, and the harmonized chapter puts a number on it: the limit of quantitation, at a signal-to-noise ratio of 10, has to sit at or below the reporting threshold, the level above which a peak is reported at all. A certificate reporting a 0.03 percent peak without stating a quantitation limit is reporting a number its own trace may not carry.
Peak shape, and the three figures that describe it
USP 621 and Ph. Eur. 2.2.46 carry one harmonized text. The symmetry factor, which that text also calls the asymmetry or tailing factor, is taken at one twentieth of peak height: the full width there over twice the distance from the leading edge to the apex perpendicular. 1.0 is symmetric; above it the peak tails and can bury a small impurity under its back edge, below it the peak fronts, usually overload. The default range, and the figure to hold a certificate against, is 0.8 to 1.8, widened from 0.8 to 1.5. Plate count measures efficiency and falls as a column ages, but the chapter defines it only for isothermal, isocratic or isodense data, so a plate number beside a gradient peptide method came from a system suitability run, not the trace next to it. Resolution is the one to read first. The chapter scales the gap between two retention times by 1.18 and divides by the sum of the two peak widths at half height, the older baseline-width formula having been dropped. Peaks that do not return to baseline are not baseline resolved, and their areas are estimates whatever decimals the report carries.
Area percent is arithmetic over what the detector saw
Area-percent purity is the main peak's area over the summed area of every integrated peak, on an assumption almost never true: that every substance answers the detector equally per unit mass. Peptides absorb in the ultraviolet mainly through the amide backbone, which is why detection sits near 214 nanometers, with weaker aromatic absorbance near 280. Something with no chromophore does not answer at all.
Three exits follow: what co-elutes under the main peak is counted as main peak, what does not absorb never enters the sum, and neither does what never leaves the column. So counterion, water and salt in a lyophilized cake can hold a real share of a vial's weight while a figure of 98 percent by HPLC sits truthfully on the same page.
What a related-substance peak can be
A small peak beside the main one has a short list of usual identities in synthetic peptide work. Deletion and truncation sequences, from incomplete coupling, come off the resin a residue or more short. Oxidation of methionine or tryptophan adds 16 daltons per oxygen and generally shifts the peak earlier. Deamidation of asparagine or glutamine adds roughly 1 dalton, and Dykstra and colleagues, in Analytical Chemistry in 2025, needed ion mobility alongside a shortened gradient to separate those forms from the parent. Chromatography sorts. Spectrometry names.
The numbers beside those peaks come from ICH Q3A(R2), adopted 25 October 2006. In its band for a maximum daily dose at or below 2 grams, 0.05 percent is where a peak is reported, 0.10 percent where it is identified, 0.15 percent where it is qualified. The last two are not flat percentages: each is that percentage or an intake of 1.0 milligram a day, whichever is lower. The pharmacopoeial reporting threshold, generally 0.05 percent, is what the harmonization renamed from disregard limit; peaks at or below it never enter the percentage.
The scope line is the part that gets skipped. Q3A(R2) covers registration applications for new drug substances made by chemical synthesis, and its preamble excludes both substances at the clinical research stage and, by name, peptides. So 0.05 and 0.10 on a peptide certificate are borrowed twice over: across the line between a registered medicine and laboratory research material, and across a substance class it excludes outright. A laboratory writing a certificate reaches for them anyway. They are conventions, not the thresholds that govern the vial.
A field key for the trace in front of you
| What you see | What it can be | What would settle it |
|---|---|---|
| Shoulder on the main peak | An unresolved second component, or column tailing | Resolution, and the symmetry factor against the 0.8 to 1.8 default |
| Small peak just before the main peak | An oxidized form, generally earlier. A deamidated form, which the trace does not place | Mass spectrometry on the collected fraction |
| Peak at the very front of the run | Void volume: solvent, salt, anything unretained, all in the denominator | An unretained marker for the void time |
| Baseline climbing across the run | Gradient drift from the modifier's own absorbance | A blank gradient injection with no sample |
What the trace cannot show
The silences are specific. A chromatogram cannot show whether the method was ever challenged with deliberately degraded material, and ICH Q2(R2), Validation of Analytical Procedures, adopted 1 November 2023 in place of Q2(R1), is where that demonstration of specificity is set out. It describes one injection of one lot on one day.
Sources
Everything below was read 2026-09-10. USP General Chapter 621 and Ph. Eur. 2.2.46 are one harmonized text. Its Stage 4 version, published by the United States Pharmacopeial Convention, official 1 December 2022, gives resolution, plate number, the symmetry factor and its 0.8 to 1.8 default, the signal-to-noise ratio of 10 and the reporting threshold; that body's notice of adoption for the chapter, posted 19 November 2021, records the dropped baseline-width resolution formula and the retired term disregard limit. From the European Directorate for the Quality of Medicines and HealthCare: the 28 September 2021 Pharmacopoeial Discussion Group sign-off, the 1 January 2023 Ph. Eur. 11th Edition implementation date, the earlier 0.8 to 1.5 symmetry range, relative retentions given for information not as requirements, and the retirement of relative retention time. The International Council for Harmonisation publishes both ICH guidelines: Q3A(R2) for the thresholds, the 1.0 milligram alternative, the 25 October 2006 date and the exclusion of peptides from scope, and Q2(R2), Validation of Analytical Procedures, for its 1 November 2023 adoption in place of Q2(R1) and its specificity provisions. The deamidation separation is from Dykstra and colleagues, Analytical Chemistry, 2025, 97, 749 to 757, abstract. The impurity chemistry, the 214 nanometer absorbance and the 16 and 1 dalton shifts are textbook analytical chemistry, not sourced to a document here. No certificate, laboratory or supplier is named here, because none was examined.
Email updates
New articles and news by email, at most once a day or once a week, whichever you pick.
Respond to this analysis
Corrections, contradictions, and anything you have seen yourself. Responses are read and checked. Verified ones become entries in the ledger or corrections to this page. Nothing posts unverified, and no response is published with anything but the handle you type, or the display name of the account you sign in with. Reporting an error, or asking for this article to come down? Use the Report form below.
Report an error or request a takedown
Reports go to the site's administrators. Nothing is removed automatically.