Quality control on diamond powder comes down to three questions: is the size distribution what it claims, is the shape what the process needs, and can the supplier prove that every batch was measured the same way. Everything else is paperwork around those three.
PSD: Which Method And Why It Matters
Particle size distribution is the single most useful measurement, and also the one most often reported without its method. That is a problem, because different methods give different numbers on the same powder.
| Method | Measures | Best for | Limitation |
|---|---|---|---|
| Sieve analysis | Passing / retained by screen | Coarse grit, mesh grades | Impractical below roughly 40 um |
| Laser diffraction | Equivalent sphere distribution | Micron and sub-micron powder | Assumes spherical particles; needs dispersion control |
| Image analysis | Projected particle dimensions | Shape plus size together | Sampling dependent, slower |
| Electrical sensing zone | Volume-based count | Narrow distributions | Requires conductive electrolyte |
| Dynamic light scattering | Brownian motion | Nano range | Very sensitive to agglomeration |
For W-series and finer, laser diffraction is the working standard. For mesh grades, sieving is the definition of the product. A supplier who reports “D50 3.2 um” without saying which instrument produced it has given you half a measurement.
Reading The Numbers
Four points on the curve carry most of the meaning:
- D10 indicates the fine fraction, which affects slurry stability and cut rate.
- D50 is the midpoint and correlates with nominal grade, but hides the tails.
- D90 starts to show the coarse tail that causes scratches.
- D97 or D99 is the number that predicts your defect rate in polishing and wafer work.
A powder with a perfect D50 and a fat D97 will pass incoming inspection and fail in production. This is the most common quality gap in the trade.
Morphology
Shape is read from images, not from a number. Two powders with identical PSD can differ in cutting behaviour because one is blocky and the other is sharp and irregular.
Ask for an image at the size you are buying. Shape at 40 micrometres says little about shape at 2 micrometres, because the milling process treats different size fractions differently. Ask which microscope and magnification were used, since a 4x image and a 10x image show different things.
Ash And Toughness
Ash is the non-diamond residue left after combustion. In our measured data across high-strength grit it ranges from 0.0002% to 0.0093%, which is a forty-fold spread. For bonded tools that spread rarely matters; for plating, slurry and thermal work it does.
Toughness index covers the mechanical side. Measured TI on the same product family ranges from roughly 32% at MBD4 to 84% at SMD30, and it varies with particle size within a grade. A TI number without the test size cannot be compared with anything.
What A Batch Record Should Contain
- Lot or batch number, production date, and the grade and size produced.
- PSD with the method and instrument named, plus the sampling procedure.
- Ash value, and impurity analysis if the application requires it.
- TI value with the test size stated.
- Morphology image reference for the lot.
- Coating weight gain and process, for coated product.
- Deviation notes, if the lot was outside the normal window.
Sampling Decides Whether The Test Means Anything
A certificate is only as good as the sample behind it. Fine powder segregates in a bag: the coarse fraction migrates upward during handling and vibration, so a scoop from the top is not representative of the lot.
Proper sampling uses a sample thief that takes material from several depths, or a full-bag riffling split. Ask how the sample was taken. If the answer is “from the bag”, treat the certificate as indicative rather than definitive.
Two related practices separate a controlled process from a lucky one: keeping a sealed reference sample per lot, and re-testing one lot per quarter against a third-party laboratory. The reference sample is what settles a dispute months later.
When A Lot Is Out Of Specification
The useful question is not whether the lot is rejected, but which parameter failed and how far. A D50 slightly high with a clean D97 is often usable for a coarser step; a D50 on target with a fat D97 is not usable for finishing work no matter how good the rest of the report looks.
- Size out of window, tail clean. Usually re-gradable to an adjacent step.
- Tail out of control. Not acceptable for scratch-sensitive work.
- Ash out of window. Application dependent; often usable in bonded tools.
- Coating weight gain out of window. Return to supplier; it cannot be corrected downstream.
Common Mistakes
- Accepting D50 as a specification. Without D97 you cannot predict scratches.
- Comparing results from different instruments. Two labs using different dispersion settings can report different numbers for the same powder.
- Skipping incoming inspection because the supplier is “known”. Lot variation happens to good suppliers too; the record is what catches it early.
How To Verify
Send a sample of the delivered lot to a third party once, and compare against the supplier certificate. Once the numbers align, you can rely on certificates for routine lots and re-test periodically. Keep the reference sample sealed, because it is the only way to settle a dispute six months later.
Next Step
Related reading: sample testing checklist, ash and magnetic content.