Two grades of ground calcium carbonate can share a d50, a brightness figure and a chemical assay — and behave nothing alike in your mill. The certificate is not lying to you. It is answering a narrower question than the one you are asking.
In this piece
What a CoA is actually for
A certificate of analysis confirms that a specific lot falls inside an agreed specification. That is a compliance document, not a performance prediction. The specification was written to be measurable, reproducible and cheap to test at the plant — three properties with very little connection to how the material disperses in your binder.
This matters because most qualification decisions are made by comparing certificates. If two suppliers both hit d50 2.5 µm and 96% brightness, the cheaper one wins. Six weeks later one of them is thickening in the can.
The numbers that predict behaviour
Top cut, not median
The d50 tells you where the middle of the distribution sits. The top cut tells you whether you will see grit, sedimentation or a blocked filter — although sparse coarse particles show up in sieve residue long before they move a d98. Two grades with an identical d50 and a 12 µm difference at the top cut are different products, and only one will survive a gloss measurement.
The shape of the distribution
Beyond percentiles, the width and modality of the curve drive packing and rheology. A narrow unimodal distribution and a broad bimodal one can share every percentile you are given and behave completely differently at high loading — which is exactly where filler-heavy formulations live.
Oil absorption
Oil absorption is the closest single number to a formulation property. It predicts binder demand, and therefore how much you can load before viscosity, film formation and scrub resistance start to fight each other. Usually on the datasheet, almost never on the certificate.
Surface treatment and moisture
For treated grades, coating level and its evenness decide dispersion behaviour more than particle size does. For moisture-cure systems — silicone and MS polymer sealants above all — residual water is the number that will cause the failure, and the limit you actually need is often an order of magnitude tighter than the standard specification. At that level, specify the method as well: loss at 105 °C will not resolve it, Karl Fischer will.
Specification width is the real risk
A certificate tells you a lot passed. It does not tell you how much room the producer has inside their own limits, and that room is your variability.
Consider two suppliers with a brightness specification of "min 94". One typically delivers 96.5, the other 94.2. Both certificates read as compliant, indefinitely. Only one of them can drift without ever failing — and the drift will show up in your product, not in their paperwork.
This is why lot-to-lot history is worth more than any single certificate. Ask for six to twelve months of results and look at the standard deviation, not the mean.
What to ask for instead
The information generally exists. It sits with the producer's technical service rather than the quality department that issues certificates, and asking for it is normal:
- The full particle size distribution curve, not three percentile points
- Oil absorption and, where relevant, BET surface area
- Coating level and coating agent for treated grades
- Lot-to-lot data over six to twelve months
- Specification limits alongside typical values
- Which of these the producer will contractually hold, as opposed to typically achieve
That last one converts a datasheet into a commitment, and it is the question that most often changes the answer.
A qualification checklist
- Define the property you are buying before you look at any certificate.
- Ask for the distribution curve and oil absorption, not just the CoA.
- Request lot-to-lot history and compare spread, not single values.
- Trial at production shear, not only in a lab disperser.
- Write the numbers that actually matter into the supply agreement, so they are held rather than hoped for.
None of this is exotic. It is the difference between qualifying a material and qualifying a document.
