Analytical methods
How to read a COA HPLC chromatogram
A certificate of analysis usually prints one purity percentage and a chromatogram. This companion to our HPLC primer shows how to read the picture itself: what every feature is, how the percentage is built, and how much it can move before the chemistry changes.
The reversed-phase HPLC basics page explains the method. This page assumes you are holding a chromatogram from a certificate of analysis and want to read it. Everything below the figures is arithmetic on a constructed chromatogram, not data from any product of ours; our COA fields currently read "Not yet provided", and nothing here is a statement about a batch.
The anatomy of a chromatogram
Figure 1 shows the full trace on a scale where the main peak fits. Figure 2 shows the same trace with the vertical scale zoomed so the small peaks and the baseline are visible, and Figure 3 zooms in on the shoulder. Both are drawn from the peak table below; the peaks are Gaussians with the heights and widths listed, the baseline is a gentle ramp, and there is a solvent-front disturbance near the start.
Figure 1. Constructed chromatogram, full scale. Nothing here is real data.
Figure 2. The same constructed trace, vertical axis zoomed. The dashed line marks where the integration window ends in scenario 3 below.
Figure 3 zooms in on 12 to 15 minutes, where the shoulder (peak B) sits on the leading edge of the main peak.
Figure 3. The same constructed trace between 12 and 15 minutes.
What each feature means
| Feature | What it is | What to check |
|---|---|---|
| Solvent front (near 1.9 min here) | Unretained material and the injection disturbance at the start of the run | It is normally excluded from integration. Check that the integration starts after it, and that nothing important elutes inside it. |
| Baseline ramp | Baselines in gradient runs can drift; the constructed baseline here rises from 3.0 to 10.5 mAU | A drifting baseline is normal, but the integration baseline must follow it. A baseline drawn flat under the ramp changes every area. |
| Main peak (13.55 min) | The target peptide, identified by its retention time and, ideally, a mass spectrum taken across it | Is it a single symmetrical peak, and does the COA show the mass of the species under it? See checking the mass spectrometry. |
| Shoulder (B, 13.12 min) | A species that elutes almost with the main peak, only partly resolved | Whether the software separated it. Closely related sequences such as deletions often elute right next to the target (synthesis impurities). |
| Early and late peaks (A, C, D) | Related species that are more polar (earlier) or more hydrophobic (later) than the target | Whether the run extends long enough to see late eluters, and whether the integration window includes them. |
| Detection wavelength | Peptide analyses commonly monitor near 214 to 220 nm, where the peptide bond absorbs | That it is stated. A different wavelength changes the relative size of peaks. |
The peak table and where the percentage comes from
The software integrates each peak, gives it an area, and reports each area as a share of the total. The shares for the constructed trace:
| Peak | Retention time (min) | Height (mAU) | Area | Area % | What it stands for in this teaching example |
|---|---|---|---|---|---|
| A | 11.55 | 5 | 1.88 | 0.56% | Earlier-eluting, more polar related species |
| B | 13.12 | 22 | 7.17 | 2.15% | Leading shoulder on the main peak |
| Main | 13.55 | 800 | 320.85 | 96.12% | Target peptide |
| C | 17.80 | 4 | 2.01 | 0.60% | Late-eluting, more hydrophobic species |
| D | 24.00 | 3 | 1.88 | 0.56% | Very late peak, after the integration window in scenario 3 |
| Total | 333.78 | 100% |
The main peak is 320.85 of 333.78 total area units, so the reported purity is 96.12%. Nothing in that calculation says what the other 3.88% is, whether it absorbs like the peptide, or what else is in the vial.
How much the number moves with integration choices
The same data, integrated four different ways:
| Scenario | What changes | Reported main-peak area % | Change from honest integration |
|---|---|---|---|
| 1. All five peaks integrated separately | Nothing; the reference case | 96.12% | 0 |
| 2. Shoulder B not resolved from the main peak | B's area is counted in the main peak | 98.27% | +2.15 points |
| 3. Integration window ends at 20 min | Peak D (24 min) is outside the window and ignored | 96.67% | +0.54 points |
| 4. Reporting threshold of 1% area | Peaks A, C and D are below 1% and are dropped | 97.81% | +1.69 points |
None of these is dishonest in itself; they are method decisions. That is why a COA should state the integration window, the reporting threshold and how shoulders are handled, and why the chromatogram, not just the percentage, is the document worth reading.
What the percentage cannot see
- Different response. Kuril (2025) states that peptide impurities from insertion, truncation, deamidation, isomerisation and oxidation may not have the same response as the main peptide, and that applying a relative response factor (RRF) of 1 by default can overestimate or underestimate impurities. The table below shows what that does to the constructed example.
- Co-elution and isomers. Kodidasu et al. (2025) describe two glucagon impurities (des-Gly4 and des-Thr5 glucagon) that coeluted on RP-HPLC and needed UPLC-HRMS to quantify, with detection limits as low as 0.01 to 0.02%. Petersson et al. (2023) developed a 2D-LC-MS approach to main-peak purity because compounds with the same mass-to-charge ratio cannot be told apart by mass spectrometry and must be separated chromatographically.
- Non-absorbing material. Water, TFA and other counter-ions are mostly invisible at 214 nm. Bizanek et al. (1996) describe HPLC with chemiluminescent nitrogen detection as a way to measure peptide content on-line, free of interference from non-nitrogen-containing UV chromophores, which shows the gap between a UV percentage and a content measurement.
| Impurity RRF (relative to the main peak) | Impurity share as reported (RRF = 1) | Impurity share after correction |
|---|---|---|
| 0.5 | 3.88% | 7.46% |
| 1 | 3.88% | 3.88% |
| 2 | 3.88% | 1.98% |
An RRF of 0.5 means the impurity absorbs half as strongly as the peptide, so equal areas hide twice the mass; an RRF of 2 means the area overstates it. The RRF of an unknown impurity is not knowable from the chromatogram alone.
From a chromatogram to the content of the vial
In the reference-material papers, purity is assigned by a mass balance rather than read from the chromatogram: the related impurities are identified and quantified by LC-HRMS, and water, counter-ions and inorganic content are measured separately. The published results show how far the powder is from a bare peptide:
| Material | Result | How it was obtained | Source |
|---|---|---|---|
| Synthetic glucagon | 896.36 ± 0.68 mg/g | Karl Fischer water, ion chromatography for TFA, ICP-MS for inorganic ions, Orbitrap MS for peptide impurities | Wang et al. (2020) |
| Synthetic oxytocin | 796.5 mg/g, uncertainty 6.5 mg/g (k = 2) | 21 related impurities (0.05 to 15.65 mg/g) by LC-HRMS/MS, plus Karl Fischer, ion chromatography, ICP-MS and headspace GC-MS | Wang et al. (2022) |
| Angiotensin II candidate reference material | 691 ± 9 mg/g (k = 2) | 19F-qNMR for the TFA counter-ion (nearly 25% by mass), combined with qNMR and LC-MS/MS amino acid analysis | Melanson et al. (2018) |
| Synthetic human C-peptide study material | Related peptide impurities totalled 83.3 mg/g (± 3.0, k = 2) | More than 65 impurities identified and quantified by LC-HRMS | Li et al. (2018) |
| Angiotensin I candidate reference material | Major impurities estimated at 10.4 mg/g | LC-HRMS/MS impurity screening; failing to correct would cause about a 1% error in amino acid analysis | Stoppacher et al. (2013) |
Wang et al. (2022) call their oxytocin assignment, the first mass-balance purity assignment for a cross-linked peptide, traceable to the SI. McCarthy et al. (2023) review the reference standards that support the quality of synthetic peptide therapeutics. Reading across the table: the peptide fraction of a well-characterised powder is a fraction of a gram per gram, from about 0.69 to 0.90 in these examples, which is why a purity percentage and a content percentage are separate numbers; see purity vs content.
A checklist for any COA chromatogram
- Axes labelled, with units, and a stated detection wavelength.
- The lot or batch number on the trace matches the vial, and the run date is shown.
- Column, mobile phases and gradient are stated, and the run is long enough to see late eluters.
- A peak table with retention time, area and area percent for every integrated peak, and the integration window and threshold given.
- The main peak is symmetrical or its shape is explained; any shoulder is named in the table.
- A blank (solvent) injection, so system peaks are not mistaken for impurities.
- A mass spectrum, ideally taken across the main peak, so identity and purity are tied together; see HPLC vs LC-MS.
- Peptide content and counter-ion content reported separately from HPLC purity.
Related
- How to read a peptide certificate of analysis
- Reversed-phase HPLC of peptides: the basics
- How to read an LC-MS trace: a worked example
- Peptide purity vs peptide content
- Counter-ion exchange from TFA to acetate
References
- Kumar Kuril A (2025). The Critical Need for Implementing RRF in the Accurate Assessment of Impurities in Peptide Therapeutics. Anal Chem 97:12480-12485. PubMed 40499007 · DOI
- Kodidasu A, Acharyya K, Ramu VG (2025). Absolute Quantitation of Coeluting Impurities in Peptide Drugs Using High Resolution Mass Spectrometry: Glucagon a Case Study in Pharmaceutical Development. J Am Soc Mass Spectrom 36:2072-2078. PubMed 40857144 · DOI
- Petersson P, Buckenmaier S, Euerby MR, et al. (2023). A strategy for assessing peak purity of pharmaceutical peptides in reversed-phase chromatography methods using two-dimensional liquid chromatography coupled to mass spectrometry. Part I: Selection of columns and mobile phases. J Chromatogr A 1693:463874. PubMed 36841023 · DOI
- Bizanek R, Manes JD, Fujinari EM (1996). Chemiluminescent nitrogen detection as a new technique for purity assessment of synthetic peptides separated by reversed-phase HPLC. Pept Res 9:40-4. PubMed 8727482
- Wang X, Zhang F, Li H, et al. (2020). Purity determination of synthetic glucagon using a mass balance approach. Sci Rep 10:4423. PubMed 32157163 · DOI
- Wang S, Wu P, Li M, et al. (2022). Mass balance method for SI-traceable purity assignment of synthetic oxytocin. J Pharm Biomed Anal 207:114401. PubMed 34656934 · DOI
- Melanson JE, Thibeault MP, Stocks BB, et al. (2018). Purity assignment for peptide certified reference materials by combining qNMR and LC-MS/MS amino acid analysis results: application to angiotensin II. Anal Bioanal Chem 410:6719-6731. PubMed 30143839 · DOI
- Li M, Josephs RD, Daireaux A, et al. (2018). Identification and accurate quantification of structurally related peptide impurities in synthetic human C-peptide by liquid chromatography-high resolution mass spectrometry. Anal Bioanal Chem 410:5059-5070. PubMed 29862433 · DOI
- Stoppacher N, Josephs RD, Daireaux A, et al. (2013). Impurity identification and determination for the peptide hormone angiotensin I by liquid chromatography-high-resolution tandem mass spectrometry and the metrological impact on value assignments by amino acid analysis. Anal Bioanal Chem 405:8039-51. PubMed 23708692 · DOI
- McCarthy D, Han Y, Carrick K, et al. (2023). Reference Standards to Support Quality of Synthetic Peptide Therapeutics. Pharm Res 40:1317-1328. PubMed 36949371 · DOI
Frequently asked questions
What should I look at first on a COA chromatogram?
The axes and labels: retention time in minutes, absorbance in mAU, the detection wavelength, and that the lot number on the trace matches the vial. Then find the solvent front, the main peak and every other peak, and check each against the peak table.
Why can a chromatogram look cleaner than the purity number suggests?
Integration choices change the number. In our constructed example the same data give main-peak purities from 96.1% to 98.3% depending on shoulder handling, window length and reporting threshold.
Does area percent equal weight percent?
No. Area percent assumes every species responds equally at the detection wavelength. Kuril (2025) notes that peptide impurities may not have equal responses to the main peptide, and that assuming a relative response factor of 1 can over- or under-estimate them.
Can HPLC show co-eluting impurities?
Not by itself. Kodidasu et al. (2025) describe low-abundance co-eluting impurities in glucagon that RP-HPLC failed to separate, and Petersson et al. (2023) note that isomers with the same m/z must be separated chromatographically because mass spectrometry cannot tell them apart.
Is HPLC purity the same as peptide content?
No. Published mass-balance work on reference peptides assigns peptide mass fractions of roughly 69 to 90% of the powder once water, counter-ions and impurities are accounted for, which is what the table on this page lists.