Table of Contents
Last quarter a buyer sent us two competing quotes for the same intermediate and asked which supplier was “wrong on particle size.” One quoted D50: 10 µm. The other quoted 325 mesh, minimum 95% passing. Neither was wrong. Neither was complete, either — and the difference between those two sentences is the kind of thing that stays invisible until, a year into supply, a batch starts feeding badly into your blender and nobody can explain why.
Particle size gets less attention than purity in most procurement conversations, which is understandable: assay and impurity limits decide whether a material is usable at all. But for an intermediate, particle size is not a purity question.
It is a process compatibility question — how the powder flows into your feeder, whether it segregates in the blend, whether your downstream milling or sieving step can absorb it. Those behaviors are governed by the shape of the whole distribution, not by a single number, and they are exactly the properties that a thin specification leaves uncontrolled.
If you are already setting acceptance criteria for chemical attributes, the physical side deserves the same discipline (our guide to specification setting for intermediates covers the overall framework).
We sit on the receiving end of RFQs, so we see how these lines are usually written. Most carry either a bare D50 or a mesh figure. Very few name a method. We probably spend more time clarifying this single line than any other physical property — so here are the five checks we would run before agreeing to anyone’s particle size specification, including ours.
“325 mesh” is the classic trap, and it survives because everyone in the conversation assumes the others share their definition. In practice it can mean at least three different things: a sieve designation whose nominal aperture is 45 µm under the ASTM E11 series, a passing-or-residue claim (95% through? 3% retained? on which sieve?), or simply a grade label the supplier prints on its datasheet.
A buyer who assumes meaning number one and a supplier who means number two will both walk away from the conversation feeling aligned. They are not.
A bare D50 has the same weakness in different clothes. The median tells you where the middle of the distribution sits and nothing about its edges — and the edges are where the process problems live.
Coarse particles beyond the D90 change how a powder flows and whether it survives your sieving step; excess fines below the D10 show up as dust, weight variability in dosing, and material that hangs up in equipment. A specification that controls only the middle of the distribution controls none of the behavior.
What complete looks like: named D-values (D10, D50, D90) each with a target and a tolerance, or a named sieve with a maximum retained percentage. Both forms are fine. A fragment of either is not.
Sieving and laser diffraction do not measure the same thing, and no conversion exists between them. Sieving (covered by USP General Chapter <786> and Ph. Eur. 2.9.31) sorts particles by whether they physically pass an opening.
Laser diffraction (ISO 13320, USP <429>) infers a size distribution from how the powder scatters light, reporting an equivalent-sphere diameter by volume. Needle-shaped and plate-shaped particles — common enough in intermediate crystals — scatter light like a different particle than they behave like in a sieve. The same powder, measured both ways, will produce two different answers, and both will be honest.
USP <429> is explicit that results from different techniques are not comparable, and even instruments built on the same principle can disagree with each other. For a buyer this has one practical consequence: a particle size specification without a named method is not a specification — it is a negotiation deferred.
If your formulation team validated their process against a laser diffraction result, the purchase order has to hold the supplier to laser diffraction, with the optical conditions and dispersion procedure identified. Otherwise every COA becomes an argument about apples and oranges.
Where the method itself has to be settled or transferred, that work belongs upstream — our guide to developing and validating the analytical method covers that side of it. This is also worth writing into the RFQ itself rather than fixing later — the same discipline that makes a precise RFQ work for chemistry applies to physical attributes.
Buyers tend to tighten tolerances because tighter feels safer. With particle size, this backfires in a specific way: every measurement method has a repeatability limit, and a tolerance narrower than what the method can reproduce turns your receiving process into a rejection machine. Batches that are, in substance, fine will bounce across your limit from lot to lot, and you will spend your quality meetings arguing about measurement instead of material.
USP <429> sets repeatability criteria for exactly this reason — instrument alignment, dispersion conditions, and sampling all contribute variation that a specification has to absorb. Before you sign a tolerance, ask the supplier one question: what is the repeatability of this method on this product?
A supplier who can answer that question with data is telling you something about their laboratory that a polished datasheet cannot (how a supplier handles that question is itself diagnostic — we wrote about reading a supplier’s technical answers for exactly this reason).
Laser diffraction results depend on choices the testing laboratory makes: wet or dry dispersion, dispersion pressure, sonication, the optical model, and the reporting basis. Change the dispersion and the D90 moves — same powder, same instrument, same day.
When a buyer’s lab and a supplier’s lab report different D90 values on a split sample, the cause is almost never fraud and almost always conditions. It is the most common particle size dispute we see, and it is entirely preventable.
The fix costs one sentence in the specification: name the dispersion route and the critical settings, and require the COA to state them. Then a COA can be audited line by line against the agreed method, the way you would audit any other claim on the document (our walkthrough of reading a COA and the verification checks for suspicious certificates both extend to the physical test lines).
| What the quote says | What it doesn’t tell you | What to ask back |
|---|---|---|
| “D50: 10 µm” | No coarse or fines control, no method, no conditions | D10/D90 limits, method name, dispersion route |
| “325 mesh, min. 95% passing” | Which standard, which basis, what happens below the sieve | Named sieve series, max retained %, and the D-values if available |
| “PSD: typical 5–20 µm” | “Typical” is a description, not a commitment | Replace “typical” with limits tied to acceptance |
Particle size is one of the physical attributes most sensitive to process changes. Crystallization conditions, milling setup, even the batch size of the size-reduction step will shift the distribution — and your supplier’s first production-scale campaign will not match a lab-scale sample exactly.
A specification frozen against a lab lot sets both sides up to fail. The same sensitivity applies across the physical family: polymorph form, bulk density, and flow all move together (we treat the crystallographic side separately in our piece on polymorphism in intermediates).
Two mechanisms keep this honest. First, retain a reference lot: an early production batch, sealed and stored, that both parties can measure whenever a question arises — it converts every future dispute into a comparison instead of a debate.
Second, agree the re-negotiation path in advance: if a legitimate process improvement shifts the distribution, the specification gets revised deliberately, through change control, rather than discovered on a COA. Buyers who set this up during the trial order rarely revisit the question — a well-designed trial order is the natural place to collect the multi-batch data that makes the tolerance real.
Once the material is in routine supply, the same discipline keeps working: distribution trends belong in your batch review, next to assay and impurity trends, because drift in particle size is one of the earliest visible signals of a process change upstream (how we think about that is in our article on batch-to-batch consistency).
Compressed into a checklist you can hold against any quote or COA:
One or two of these is a clarification email. Three or more, in our experience, predicts how the whole technical relationship will go — because a supplier who writes a thin specification will answer thin questions the same way.
If the powder is milled, sieved, blended, or fed into equipment downstream, yes — those operations respond directly to the distribution. If the material dissolves completely in your first processing step, a looser specification may genuinely be fine. Decide from the process, not from habit.
Yes, and for some materials it is sensible — a sieve residue catches the coarse tail cheaply while laser diffraction characterizes the full distribution. What you must define is which result governs acceptance when they disagree, and they will occasionally disagree.
Possibly both. Different dispersion conditions, different instrument models, or a mass-basis versus volume-basis report will produce different numbers from the same powder. The question to ask is not “which lab is right” but “which method governs” — agreed before shipment, not after.
Only if the specification also controls the coarse and fine ends and names the method. A D50 inside the target with an uncontrolled D90 can still be a batch that behaves badly in your process — which is the whole reason the five checks above exist.
Particle size questions are easier to settle before the purchase order than after. If you want a second opinion on a quote line, a COA, or a draft specification, send it to our technical team — clarifying “325 mesh” is a regular Tuesday for us.
Two suppliers can quote the same intermediate with completely different particle size lines, and both may be telling the truth. Here is how to read a particle size specification before you agree to one, so the number on the COA means what you think it means.
A quality agreement divides quality responsibilities between you and your supplier before anything goes wrong. This guide shows intermediate buyers which clauses deserve the first read, what is worth negotiating, and what belongs somewhere else entirely.
A sample proves the chemistry; a trial order proves everything around it — documentation, packaging, logistics, and how a supplier behaves when something slips. This guide walks buyers through both, in the order that costs least and reveals most.
Manufacturer and supplier of pharmaceutical APIs and intermediates. Contact us for technical support and sourcing solutions.