By Jay Shah, Technical Team, Beta Diamond Products
The sequence our customers run more than any other is 9, 6, 3, then 1 micron diamond. The 6 micron step is the optional one: on softer materials and well ground surfaces many labs skip it and run 9, 3, 1 with no penalty. Some stop at 1 micron, others add a final sub micron oxide polish at 0.05 micron or finer when the application demands it. Each step exists to remove the scratches and subsurface damage left by the one before. Skip too far and you are asking a fine abrasive to erase coarse damage, which it will do slowly, unevenly, or not at all.
First, untangle grits from microns
Before any micron sequence makes sense, you need one piece of translation, because grinding papers are sold in grit numbers and polishing abrasives in microns, and the number one knowledge gap we encounter with customers is the bridge between the two. A 600 grit ANSI silicon carbide paper has particles around 15 to 16 microns. That single fact explains why the diamond sequence starts at 9 micron: it is the next sensible step down from where grinding ends. It gets more confusing because European P graded papers use different numbers for the same sizes, and we keep a full conversion chart in a separate article. For this article, one anchor is enough: 600 grit ANSI is roughly 15 to 16 microns, and diamond picks up from there.
The logic underneath the numbers
A useful rule: each abrasive step should be no more than about half to one third the size of the previous one. From 15 micron paper, 9 micron diamond respects that ratio. Jumping from 9 micron straight to 1 micron does not, and you will spend three times longer at 1 micron chasing 9 micron scratches than the 3 micron step would have taken. The 6 micron step exists for materials hard or damage prone enough that even 9 to 3 is too big a jump.
Remember that scratches are only the visible part. Every abrasive step leaves a deformed layer beneath the surface, typically several times deeper than the scratches themselves. The 3 micron step is not just removing 6 and 9 micron scratches. It is removing the damaged metal underneath them. This is why "it looks shiny" is not the same as "it is ready." A surface can be mirror bright and still show false structure after etching because deformation was polished over rather than removed.
Adjusting for your material
Hard materials tolerate compression. On cemented carbide or hard ceramics, many labs run 9, 3, 1 or even 6 then 1, because deformation depth on very hard materials is shallow.
Soft materials demand the opposite. Aluminum, copper, and their alloys pick up deformation easily and hide it well. Run the full sequence, keep the 6 micron step, use light pressure, and treat the final oxide step as mandatory.
When the final oxide step matters
If your endpoint is routine optical microscopy, a clean 1 micron finish is often enough, and plenty of our customers stop there. If you are doing any of the following, the sub micron step is not negotiable: color or tint etching, image analysis where contrast uniformity matters, microhardness on thin cases, EBSD, or high magnification work on soft alloys. Colloidal silica at 0.02 to 0.06 micron or fine alumina at 0.05 micron removes the last whisper of deformation through combined chemical and mechanical action that diamond alone cannot replicate.
A worked example for steel
Section under flood coolant. Grind 240, 400, 600 grit ANSI SiC, about one minute each, rotating the sample 90 degrees between papers. Then 9 micron diamond on a hard woven cloth, 4 to 5 minutes. Then 3 micron, 3 to 4 minutes. Then 1 micron, 2 to 3 minutes. Add the 6 micron step between 9 and 3 if scratches persist. Finish with colloidal silica on a soft chemical resistant cloth for 1 to 2 minutes if the work demands it. Etch with 2 percent nital. If that does not produce a clean micrograph, the problem is contamination or technique, not the recipe.
Every micron size here ships from our shelf in suspension, paste, or spray. Tell us your material and current sequence and we will point out where it is losing time, free, because that conversation usually ends with a sample request anyway.