A certificate of analysis, or COA, is the document that turns a vial of powder into a known quantity. It is also the document most often faked, most often misread, and most often ignored. This is a working guide to reading a real one and recognizing a fake one.
The four things a real COA tells you
Identity. A mass spectrum confirming the molecular weight matches the published sequence within the expected tolerance. Look for the actual spectrum image, not just a number — a real COA shows the trace and the integration.
Purity. A reverse-phase HPLC chromatogram with a clearly labeled main peak, a published gradient, and an integrated purity percentage. The chromatogram should show the full run, not a cropped close-up of the main peak.
Endotoxin. A limulus amebocyte lysate result reported in endotoxin units, with the test method and the detection limit stated. If the COA reports endotoxin as a simple pass or fail with no number, treat it as missing data.
Residual solvents and counter-ion. Most synthesis routes leave traces of acetonitrile, trifluoroacetic acid, or acetate. A real COA acknowledges this and reports residual levels against a specification.
Reading the HPLC trace
Start with the main peak. It should be symmetrical, not skewed, and it should sit cleanly on the baseline. A shoulder on the leading edge usually indicates a deletion product. A shoulder on the trailing edge usually indicates an oxidation product. A flat baseline that suddenly lifts halfway through the run is a sign of late-eluting impurities that the integration may have ignored.
Then look at the headline purity number in context. Ninety-eight percent purity on a clean trace with a symmetrical main peak is meaningfully different from ninety-eight percent purity on a trace with three visible shoulders. The number is the same; the molecule is not.
Reading the mass spectrum
The main mass peak should match the published monoisotopic or average mass within the instrument's tolerance — usually a fraction of a dalton on a modern instrument. A real COA reports both the theoretical and observed mass, not just one. Common adduct peaks — sodium, potassium, ammonium — are normal and should be labeled if present. An unlabeled mass peak near the main signal is a question, not an answer.
What fakes look like
Fake COAs have a few consistent tells. The chromatogram is a low-resolution image, often the same image reused across multiple lots — check the file metadata if you can. The mass spectrum has no axis labels or no integration callouts. The endotoxin section reports a pass with no number. The document has no lot number, or the lot number does not match anything on the vial label. The signing analyst is a generic name with no credentials and no contact information.
The most common fake of all is a real COA from a different lot, photocopied and re-stamped. The defense against this is the lot lookup on the manufacturer's site — if you cannot independently verify the lot number through the manufacturer, the document is decoration.
What to do with the COA you receive
File it with your lot records. Compare the headline numbers across lots over time — purity should be stable, endotoxin should stay well under the published limit, and the main-peak retention time on the HPLC trace should not drift unless the column has been changed. A supplier whose COAs become quietly worse over six months is a supplier worth a conversation.