Microns are not a promise

A thicker coating is the easiest quality claim to make and one of the weakest. What survives daily wear is adhesion, density and edge coverage, and a single micron figure describes none of those on its own.

Published 2026-11-02 · Reviewed by Felix, founder of KUBO Jewelry

Stainless steel jewelry in gold, black and steel PVD finishes on a bench
One process, several colours — the wear life comes from the bond, not the number.
Short answer

Decorative PVD on stainless steel jewelry is typically 0.2–1 µm thick, grown into the surface rather than sitting on it. Thickness is measured by X-ray fluorescence to ISO 3497 on a coupon. Colour, adhesion and edge coverage decide wear life more than the number; above 1.5 µm, stress makes the layer chip rather than wear evenly.

PVD is a line-of-sight vacuum process. Metal is vaporised from a target, reacts with a gas, and condenses on whatever surface it can see. That physics sets a ceiling on how thick a coating can usefully be, and it explains why the micron figure on a supplier's sheet is a weaker signal than most buyers assume.

What a thickness figure describes

A stated thickness is measured on a flat test coupon run through the chamber with the production parts, not on every face of every part. It is a real, repeatable measurement of that coupon.

On a chain link, the outward faces receive the full thickness and the inner faces, where two links overlap, receive a fraction of it, because the vapour could not reach. The base of a recessed logo gets less than the field around it. The inside of a ring band depends entirely on how the parts were racked and rotated. A batch quoted at 0.5 µm is 0.5 µm on the faces the coupon represents, and the buyer's question is which faces those are.

Typical ranges for jewelry

Most decorative PVD on stainless steel jewelry sits between 0.2 and 1 µm, with black coatings at the top of that band. The ranges below are illustrative for lines running jewelry volumes; confirm the figure for the specific part and rack.

FinishTypical thicknessNote
Gold tone (titanium nitride family)0.2–0.5 µmColour is set by gas ratio, not by thickness
Rose gold tone0.2–0.4 µmHeld thin so the pink cast survives
Black (TiCN or DLC type)0.5–1.5 µmDarker and harder; reads as a coating, not a metal
Gunmetal and bronze tones0.3–0.8 µmThe most batch-sensitive family
Industrial wear specification1–3 µmRare on jewelry; too thick for fine edges

The decorative range is narrow, and the colour is not produced by adding material. A gold tone at 0.3 µm and one at 0.8 µm can be the same colour or visibly different, depending on the gas mix and the target.

How the figure is measured

ISO 3497 covers X-ray spectrometric coating thickness measurement, and that method sits behind most supplier figures. Three routes are common in practice.

MethodWhat it givesLimits
XRFThickness and composition, non-destructiveNeeds a flat face; typically ±10–20% on thin layers
Ball crater (calotest)True thickness at one pointDestructive; the test part is scrapped
Cross-sectionThickness and structureSlow laboratory work, used to settle disputes

For production, XRF on a coupon each batch is the normal combination, with a periodic crater test to keep the instrument honest. What matters commercially is the record: a coating line that measures every batch and keeps the numbers can answer a claim months later. One that measures the first batch and then trusts the process cannot.

Why a bigger number is the wrong request

Past roughly 1.5 µm on jewelry geometry, internal stress usually makes the layer chip at edges instead of wearing more slowly. A PVD layer grows in compression, and beyond a certain thickness the stored stress exceeds what the bond to the steel can hold. The failure changes from gradual abrasion to visible flaking.

Three further costs follow. Colour drifts as the layer thickens, so a gold tone darkens and a rose gold loses its pink. Coating time drives the batch charge, so doubling the thickness roughly doubles the cost per piece. And the layer adds dimension: a micron on each side of a diameter is enough to break a press fit that was toleranced to ISO 2768 without a coating allowance.

Where coatings actually fail

Not in the middle of a flat face. The order of frequency is edges and burrs, contact points where a link rubs a link, recesses the vapour could not reach, and any surface coated over contamination. Polishing compound left in a crevice produces a blister rather than a wear mark, and no thickness specification prevents it. Most failures that look like thin coating are cleaning failures.

Salt spray work such as ASTM B117 and the equivalent ISO 9227 cycle can rank coatings for corrosion, but neither predicts abrasive wear on a ring shank or a clasp. For that, the honest tools are a wear trial on sample parts and a thickness record kept per batch.

Writing thickness into the order

Four lines do the work:

  1. A range, not a minimum. 0.3–0.5 µm, measured by XRF on a coupon per batch.
  2. The faces that matter. Outer link faces, the top of a ring band, stated explicitly.
  3. An adhesion criterion. A tape or cross-hatch test on production parts, agreed before the run.
  4. A colour reference. A physical sample, because thickness and colour are separate specifications.

Where a piece carries a piercing post, nickel release is a substrate question governed by EN 1811 testing of the finished item, not something the coating removes.

The FTC's Jewelry Guides, published at 16 CFR Part 23, require coated goods to be described as coated, so name the process on the order rather than the metal it imitates. Buyers weighing finishes for rings and bracelets can follow the question further in our guides to PVD versus electroplating and to why colour fades. For a quotation that states thickness, measurement method and coupon policy in writing, send the drawing through the RFQ page.

Next step

Have a project this applies to?

Request a quotation →