How to read a certificate of analysis

What a certificate of analysis is — and what it is not

A certificate of analysis (COA) records what was measured on a specific batch of material and what was found. It is not a specification sheet: a specification states what the material must meet, while the certificate reports the result for the lot in hand. It is not a safety data sheet either — the SDS covers hazards, handling and transport, and does not carry the analytical results.

This distinction matters at audit. A specification with no result attached tells an inspector what you asked for; only a lot-specific certificate tells them what you received.

Product identification — the part that is easiest to get wrong

Check that the catalogue number and the batch number on the certificate match the label on the container. Then check the identity fields against your own record of what you ordered:

• Chemical name and, where applicable, the pharmacopoeial designation — for example “Impurity F” as defined by a pharmacopoeia. The designation is only meaningful if it maps to the definition in the current text.
• CAS number — a useful cross-check, but not a substitute for the name. A CAS number identifies a chemical entity; it does not distinguish salt forms, hydrates or stereoisomers, which frequently carry separate registry numbers or share a parent number.
• Molecular formula and molecular weight — check whether the formula shown is the free base or the salt, and whether the weight is the formula weight or the exact mass. If the material is a hydrate or solvate, the formula should say so.
• Structure — where a structure is printed, check the stereochemistry. A structure drawn without wedges for a compound that has defined stereocentres is a warning, not a simplification.

Purity and content are two different numbers

This is the single most common source of confusion when comparing certificates.

Chromatographic purity (often “purity (HPLC), area %”) is a ratio of peak areas in one chromatographic system. It says how much of what eluted is the main peak. It does not see material that does not elute, does not absorb at the detection wavelength, or is retained on the column. A sample that is 60 % water by mass can still show 99 % area purity if the water is invisible to the detector.

Content or assay is a mass-based figure. It is usually assigned by mass balance: start from 100 %, then subtract the measured volatile content (water by Karl Fischer, residual solvents by GC) and the measured inorganic residue, and correct for chromatographic impurities. The result is expressed on a defined basis.

The basis is part of the number. “99.5 %” means different things depending on whether it is stated:

• as is — as the material is received, including its water and solvent content;
• on the dried basis — corrected to a stated drying loss;
• on the anhydrous basis — corrected for water only;
• on the solvent-free basis — corrected for residual solvents.

Two certificates quoting the same figure on different bases are not comparable. When you are preparing a standard for a quantitative assay, you need the as is value, because that is the material you are actually weighing out.

Reading the test table

A useful certificate lists, for each attribute, the test performed, the specification applied and the result obtained. Read it in that order and ask three questions of every line:

1. Is the method named? “Purity: 99.4 %” without stating the technique, the detection wavelength or the quantitation mode is not verifiable. “Purity (HPLC, area %, 220 nm)” is.
2. Is the specification meaningful for the material? A “conforms to structure” specification for identification is normal; a blank specification on a purity line is not.
3. Do the numbers add up? If water is 0.4 %, residual solvents 0.2 % and inorganic residue 0.1 %, then a content figure of 99.9 % leaves nothing for organic impurities. That arithmetic is worth doing — it catches transcription errors and over-optimistic assignments.

Uncertainty, and what “traceable” should mean

Where content is assigned, an expanded uncertainty U with a coverage factor k is a statement about how well the value is known, not about how pure the material is. A material can be 99.9 % pure with a large uncertainty, and 97.0 % pure with a small one. For a quantitative assay it is the uncertainty that propagates into your result.

The word “traceable” should be followed by an explanation: traceable to what, by what route. A certificate that says the balances are calibrated with weights traceable to a national standard is making a checkable statement. A certificate that says only “traceable” is not.

Storage, retest and release

Three dates and conditions carry real weight:

• Storage conditions — and whether the material as shipped has been kept that way. A stability statement is only valid for material stored as directed.
• Retest date — the date after which the material should be re-examined before further use. It is not an expiry date, and the difference matters: an expired material may not be usable, whereas a material past its retest date may be perfectly usable after re-testing.
• Release date and signatures — who prepared, reviewed and approved the document. A certificate with no named function behind it is harder to defend in an audit than one with a laboratory and a quality function identified.

Questions worth asking before you file the certificate

1. Was this certificate issued for this lot, or is it representative of the product (a “typical” certificate)?
2. On what basis is the content figure stated, and is that the basis your method assumes?
3. Which analytical methods were used, and are the raw data or spectra available?
4. What happens to the assigned value if the material is hygroscopic and has been open to the atmosphere during handling?
5. If a value has been re-assigned since the material was released, is there a revision history?
6. What does the supplier not claim? A supplier who can answer that precisely is usually the one whose other answers are also precise.

A certificate is a document you will have to defend. Reading it critically at receipt costs minutes; discovering at audit that its basis did not match your method costs a great deal more.