Analytical Documentation
How to read a certificate of analysis — and spot the ones that mean nothing
HPLC tells you how much of the peak is your target compound. Mass spectrometry tells you whether that target is the right molecule at all. Most buyers of research-use material check one number, on a document that was never lot-specific to begin with.
There is a specific failure mode in laboratory procurement that has nothing to do with chemistry and everything to do with paperwork. A researcher orders a compound. The listing says “99% purity, COA available.” A PDF arrives. It has a logo, a molecular weight, a retention time, and a percentage. Everybody relaxes.
The PDF is often the least interrogated document in the entire workflow. And in the research-use-only supply chain — where there is no FDA oversight of manufacturing, no lot tracking requirement and no adverse event reporting — that document is essentially the only quality control the buyer ever sees. One analysis published in 2026 found that roughly three in ten research-grade products failed basic quality checks: mislabeled, under purity thresholds, or dosed incorrectly.[1]
This guide is about reading the paperwork properly. It is not a protocol, not a buying recommendation for any specific compound, and not medical guidance of any kind. It is about one skill: looking at a certificate of analysis and knowing within ninety seconds whether it is evidence or decoration.
A certificate of analysis is not a claim about a product. It is a report about one specific batch, on one specific date, using named methods. If it can’t tell you which batch, it isn’t a certificate.
The two tests that answer different questions
Nearly every purity dispute in this market comes from conflating identity with purity. They are separate measurements produced by separate instruments.
HPLC — high-performance liquid chromatography — separates the contents of a sample as they pass through a column at different speeds. The output is a chromatogram: a baseline with peaks. Purity is calculated as the area of your main peak divided by the total peak area, expressed as a percentage. For research-grade material, ≥98% is the working standard, and 99%+ is what serious catalogs advertise.[2]
What HPLC cannot do is tell you what the main peak is. If a supplier shipped an entirely different but structurally similar compound, HPLC would happily report it as 99.4% pure. Pure something.
Mass spectrometry answers identity. It ionises the sample and measures mass-to-charge ratio, producing an observed molecular weight you compare against the theoretical molecular weight for the compound you ordered. Agreement within a fraction of a dalton is what confirms you received the molecule on the label. A certificate with HPLC and no MS is half a certificate.
Suppliers that publish both chromatograms and mass spectra per lot are the minority. Our advertising partner is one of them.Sponsored
See per-lot reports →Purity is not one number
Here is the part that surprises people coming from consumer products. The “purity factor” used for reference standards in pharmaceutical analysis is not the HPLC number at all. It is the HPLC number discounted by everything else in the vial. The standard formulation is:
Purity factor = % purity by HPLC × [1 − 0.01(% water + % residual solvents + % residue on ignition + % salt)][3]
A material reporting 99% by HPLC but carrying 6% water and 2% counter-ion salt is not a 99% material by mass. That gap is why gravimetric calculations based on the label figure drift.
Water content — usually measured by Karl Fischer titration or loss on drying — matters disproportionately for lyophilised powders, which are hygroscopic by nature. Residual solvents are the traces of the synthesis and purification process left behind, typically measured by gas chromatography. Residue on ignition captures inorganic ash. None of these appear on a one-line certificate, which is precisely why one-line certificates are popular.
The ninety-second checklist
Open the PDF and look for these fields in this order. Anything missing is a question, not a dealbreaker — but three or more missing means the document is marketing.
| Field | What good looks like | Red flag |
|---|---|---|
| Lot / batch number | Alphanumeric ID that matches the label on the physical vial | No lot number, or one certificate reused across the whole catalog |
| Date of analysis | Specific date, plus retest or expiry date | Undated, or dated years before your order |
| Testing laboratory | Named third-party lab with contact details | “Tested in-house” with no signature or analyst name |
| Method | Named method and conditions — column, gradient, detection wavelength | “HPLC” with no parameters |
| Specification vs. result | Two columns: the acceptable range, and the measured value for this batch[4] | Only a result, with no specification to judge it against |
| Raw chromatogram | Full trace image with baseline, retention times and integration table | A typed number with no trace attached |
| Mass spec data | Theoretical vs. observed molecular weight | Absent entirely |
| Water & solvents | Karl Fischer or LOD figure, residual solvent screen | Absent — you cannot compute mass accurately |
Reading the chromatogram itself
If the trace is attached, two details carry most of the information. First, peak position: impurities eluting before your main peak tend to be more hydrophilic species, while later peaks tend to be more hydrophobic — often related structures from incomplete synthesis.[5] Second, baseline behaviour: a noisy or drifting baseline inflates integration and can quietly manufacture a percentage point or two of apparent purity.
A trace that looks suspiciously clean — one perfect Gaussian peak, no solvent front, no noise — is worth more scepticism than a trace with visible minor peaks that were honestly integrated.
Documentation you can check before you spend anything
Our advertising partner publishes third-party HPLC and mass spectrometry reports tied to individual lot numbers, for research use only. Read the certificates first — that is the entire point of this article.
Open the partner catalog →Sponsored link. Research use only. Not for human or veterinary use, food, drug or cosmetic applications.
Six questions to ask a supplier
- Is the certificate lot-specific, and can I see it before ordering? A vendor who emails a generic PDF is telling you their documentation is not tied to inventory.
- Who performed the analysis? Third-party beats in-house. In-house with a named analyst and instrument beats an unsigned template.
- Do you retain samples from each lot? Retained samples make independent verification possible after the fact. Very few resellers do this.
- What are the storage and shipping conditions? If the answer is vague, the material’s history between QC and your bench is unknown.
- What is the written policy if my own testing disagrees with your certificate? The existence of a policy matters more than its generosity.
- Are you the manufacturer, an importer or a reseller? All three can be legitimate. Only one of them controls the synthesis.
Material sold as “research use only” occupies a distinct regulatory position. Compounds in this category are not FDA-approved for human or veterinary use, and RUO material cannot lawfully be used in human or veterinary compounding — pharmaceutical-grade API is required for that.[6] Regulatory attention to this space has been increasing: the FDA placed a set of peptide drugs on its “unsafe” compounding list in 2023, and an advisory committee revisited those restrictions in 2026, which means the rules you operate under may not be the rules you learned.[7]
Nothing in this article should be read as guidance on human use of any compound. Neural Research does not publish protocols, dosing information or therapeutic claims.
The uncomfortable conclusion
Certificate literacy is a workaround, not a solution. In a supply chain with no mandatory lot tracking and no adverse event reporting, the buyer is performing the quality assurance function that regulation performs elsewhere. That is a real burden and it is unevenly distributed — the labs with instrument access verify independently, and everyone else reads PDFs.
Reading them well is still worth something. A supplier that publishes complete, lot-specific, third-party documentation with raw traces attached has chosen to be checkable. That choice is expensive, which is exactly why it is informative.
References
- Hone Health, survey and quality-failure analysis, June 2026.
- Onyx Biolabs, How to read a COA: HPLC vs. mass spec, January 2026.
- LCGC / Chromatography Online, Certificate of analysis and calculations for small-molecule drugs, June 2024.
- Alliance Chemical, How to read a chemical COA, November 2025.
- PH Labs, How to read an HPLC certificate of analysis, May 2026.
- Older Lundy Law, Research-use-only material and compounding, February 2026.
- ProPublica, An FDA reversal could open the market, April 2026; Johns Hopkins Public Health, How the FDA regulates peptides, August 2026.
Keep reading
A certificate describes the material on the day it was tested. What happens between that day and your freezer is a separate problem.
Cold chain & reagent stability →