PVD Coating Sourcing Guide: Colours, Substrate, Thickness and Testing

Alan XiaoFounder, Sourcing Pioneer

· Updated October 9, 2026 · 26 min read

A gold-coloured PVD finish contains no gold. PVD is a way of putting a coating onto a surface in a vacuum, and the colour comes from the compound you deposit rather than from a dye or from the metal you named — which is why a "gold PVD" watch case is not gold plated and does not carry a gold content. This guide covers what PVD actually is, why the colour is set by chemistry rather than choice, why the layer underneath decides whether the coating survives, which materials the process rules out, how thin it really is, how to test adhesion rather than trust it, and how a PVD job gets quoted in China.

1. What PVD Coating Is, and What It Is Not

PVD stands for physical vapour deposition. A solid coating material is vaporised inside a high vacuum chamber, travels through the chamber as a vapour, and condenses onto the part as a thin film. That is the whole idea, and it is worth stating plainly because most of the practical consequences follow from it. Coating houses describe it the same way: a solid source is vaporised into atoms or molecules and carried as vapour through a vacuum and a low-pressure gas or plasma to the surface, where it condenses into the coating.

The phrase line of sight matters more than it sounds. Because the material travels in straight lines from the source to the part, the part has to be presented to the vapour, which is why parts are fixtured and rotated on multiple axes inside the chamber, or carried past the source on a conveyor. A surface that faces away from the source throughout the cycle gets less coating, or none.

Four things PVD is not

• It is not a colour. PVD is a family of processes. The colour comes from the compound you choose to deposit, and different compounds are different chemistries rather than shades of one product.
• It is not gold plating. A gold-coloured PVD finish can be produced without gold. It is a hard, thin, coloured film, which is a different product from a plated layer of actual precious metal.
• It is not a thick protective jacket. Films are measured in microns, and often in single digits of microns. Where a part needs real corrosion protection, that protection comes from another layer.
• It is not a wet process. No plating bath, no rinse tanks. Coating houses describe it as greatly reducing the toxic substances that have to be managed and disposed of compared with conventional coating methods that use fluid precursors and chemical reactions.

2. The Two Processes You Will Be Quoted

Two methods account for most PVD work, and a coating house will usually describe both without naming which one your job will run on. Ask, because the method affects what the coating can do and how the part has to be presented.

One more point about the process that has commercial consequences: because the coating is deposited atom by atom, the surface has to be clean and, for the standard processes, electrically conductive. Contamination that would be invisible — fingerprints, polishing compound, a stray oil film — sits between the substrate and the coating and becomes a weak point. This is why an experienced coating house spends real effort on cleaning and preheating the parts, and why the same coating can behave differently at two suppliers who run identical machines.

3. Colour Comes From the Compound, Not a Dye

This is the part that most buyers get wrong first, and it is the answer to the most common question in this category. When a supplier quotes a gold PVD finish, they are not quoting a plating thickness of gold. They are quoting a compound film whose colour happens to be gold, and the colour is a property of that compound.

A coating house data sheet for titanium nitride states the appearance as dark gold, and notes that the coating is hard, inert and lowers friction. That single line explains a lot of confusion in the trade. Titanium nitride is a wear coating used on cutting tools, punches and moulds, and its colour is a side effect of its chemistry. When the same compound is used on a watch bezel or a piece of jewellery, you get the same colour and the same hardness, but it is not a precious-metal layer and it does not contain gold.

The practical consequence for a product range

If you change the colour of a product, you may be changing the compound, and if you change the compound you may be changing the process settings, the number of targets loaded and the run time. A range that offers gold, silver and black is not one coating run in three colours; it can be three setups.

This is worth knowing before you promise a colour range to a retail buyer. Ask the coating house to quote each colour as its own line, with its own minimum quantity, rather than quoting a blended figure. You will get a more honest number and you will find out early which colours are expensive to add.

4. Why the Substrate Decides Whether It Works

Ask any coating house what matters most for adhesion and you will get the same answer: the surface the coating goes onto. One data sheet puts it directly, stating excellent adhesion as long as the substrates were properly cleaned and the process was properly executed. That is a conditional promise, and the condition is yours to manage.

The standard processes need a conductive surface

Conventional PVD deposits onto electrically conductive materials. That is a real limit, and it means the surface matters twice over: it has to be clean enough to bond to, and it has to be the right kind of material to receive the coating.

Which is why brass and zinc alloy usually get a base layer first

This is the detail that separates a coated part that lasts from one that fails, and it is the one most often left out of a quotation. PVD films are extremely thin. A film a few microns thick cannot carry the corrosion protection for a part that sees sweat, salt spray or regular washing, because there is simply not enough material there and a scratch goes straight through it.

So on a brass or zinc-alloy watch case, the normal build is a plated layer first and the PVD colour on top. The plated layer does the protective work and gives the coating something sound to sit on; the PVD layer supplies the colour and the scratch resistance. When a quotation mentions only one thickness, ask which layer it refers to. A part specified with a coating thickness but no base layer thickness has an unfinished specification, and the failure will show up as corrosion at the edges rather than as a visible coating defect.

5. Temperature: What PVD Rules Out

Standard PVD is a hot process, and this is the constraint most likely to catch a buyer out because it shows up in the material rather than in the finish. A tool coating house states that it applies titanium nitride at about 800 °F, which is roughly 425 °C, and says plainly that this temperature may adversely affect some materials. Another coating data sheet puts the same coating at 425 to 450 °C. Equipment makers give a broader working range, describing the process as usually running between 50 and 500 °C depending on the coating, the substrate and the process.

So the number is not one number. It depends on the coating you chose and on what the coating house is willing to run. What matters for a buyer is the next sentence, which is the most useful line in any coating data sheet: parts can be coated at lower temperatures, but this requires process modifications that can adversely affect coating properties such as adhesion and cycle time.

Read that carefully. A supplier who says "yes, we can coat plastic" is telling the truth. They are also telling you, if you ask the follow-up, that the process will be changed, and that two of the things the change affects are how well the coating sticks and how long the cycle takes. The second one lands on your price.

6. Thickness: How Thin It Really Is

PVD coatings are thin in a way that surprises people who are used to plating. A published titanium nitride data sheet gives the thickness as 0.0001 to 0.0002 of an inch on the working surfaces, which is 2 to 5 microns. Equipment makers describe films generally as ranging from a few atoms to many microns. Your entire coating, in other words, is thinner than a sheet of household foil by a wide margin.

The reason that is acceptable is hardness. The same data sheet gives titanium nitride a hardness of 2,400 to 2,600 Hv, which it notes is harder than carbide. A film that thin survives normal handling because it is harder than the steel underneath it. The substrate can deform under an impact while the coating, in the ordinary course of wear, resists scratching.

Thickness is not even across the part

Here is the detail that explains most finish complaints, and it comes straight from a coating house's own technical notes. There is variation in coating thickness across a part: edges collect more vapour, so they get more coating, and factors such as process parameters, part geometry and where the part sits inside the vacuum chamber all influence how thick the coating ends up on a particular area.

Two practical results follow. First, a colour that looks slightly deeper or brighter along an edge is not a defect, it is the geometry of the process, and it is worth agreeing with the coating house how much variation you will accept before you reject a batch over it. Second, some faces of a part will always receive less coating than others. If a surface that gets less coating is the one the customer sees, the specification needs to be written that way from the start — which is a fixturing question as much as a coating question.

"Make it thicker" is not a free fix

The obvious thought on reading that the film is only microns thick is to ask for more. It is possible to build up thickness by coating a part more than once, and the same data sheet says so. It also says the limit: as the coating gets thicker, there is a greater chance of spalling, which means patches of coating lifting away from the surface.

So thickness is a range to be specified, not a number to be maximised. Ask for a target with a tolerance, test the result, and treat a supplier who offers unlimited extra thickness without mentioning spalling as someone who has not run the experiment.

7. How to Test Adhesion and Wear, Not Just Thickness

There are two separate things to prove about a PVD coating, and buyers routinely accept evidence for one and assume the other. Thickness tells you how much coating you received. Adhesion tells you whether it is attached. A part can pass a thickness check and fail an adhesion check, and only one of those failures is visible before the product reaches a customer.

Thickness

Film thickness is measured with instruments rather than by eye, and coating houses check it in several ways: a calotest, X-ray fluorescence, a stylus profilometer, or by cutting a cross-section and examining it under a microscope. You do not need to choose the method, but you do need the report to state which one was used and where on the part it was measured — because of the edge variation above, a single number with no location is close to meaningless.

Adhesion

The standard method for rating how well a coating sticks is a tape test, published as ASTM D3359, Standard Test Methods for Rating Adhesion by Tape Test. The idea is simple enough to describe without the document: a grid of cuts is made through the coating down to the substrate, tape is pressed over the grid and pulled away, and how much coating lifts off with the tape is rated against a scale. The method is the same one used across the coatings industry, which means a result is comparable between suppliers.

Ask for the test to be run on the actual production stack, on the actual substrate, at the actual thickness — not on a flat test coupon. A coupon proves what the coating can do on an ideal surface, and the interesting failures happen on a real part with a curved surface, a hole, or a thread that did not get the same fixturing attention.

Wear and corrosion test the layer underneath

This is the point most buyers miss. If the part is a brass or zinc-alloy item with a plated base layer and a PVD colour on top, then a corrosion test is largely a test of the plating, not of the PVD. A good salt spray result tells you the plated layer is sound and the stack is closed. It does not tell you the PVD film is thick enough or bonded, because a film a few microns thick is not what is holding the corrosion back.

Wear is the test that belongs to the PVD. Run it in the condition the product will actually meet — a rub test against the surface it will contact, repeated enough times to be meaningful — and compare a coated sample against an uncoated one. A wear result is far more useful when it is comparative, because there is no universal pass mark for a coating on a consumer product. Our first article inspection guide covers how to fold this kind of measurement into the approval stage rather than discovering it at delivery.

8. PVD Against Electroplating: Where Each One Wins

Buyers often frame this as a choice. On a decorative part in a base metal it usually is not one, because the normal build uses both. The useful exercise is to see what each layer is good at.

Two boundaries are worth stating plainly, because leaving them out is how a specification ends up wrong.

First, PVD does not replace the plating on a base-metal part. It is too thin. If you remove the plated layer to save money and keep the PVD colour, you have removed the corrosion protection and kept the appearance, and the failure will show at the edges and in the threads first.

Second, a deeper scratch behaves differently on the two layers. A hard PVD film resists scratching better, but once something does cut through it, there is almost no material there to absorb the damage, and the exposed substrate is right underneath the cut. Plating gives you more material between the surface and the metal, at lower hardness. Neither is universally better; they fail in different ways, and the way a product is used decides which failure matters.

9. How a PVD Job Gets Quoted in China

A PVD quotation is a batch quotation, and understanding that is the difference between a price you can work with and a price that never falls. The cost of running a chamber is largely fixed: pump down, heat, deposit, cool, vent. Once that cycle is running you want as many parts in it as the fixtures will hold.

Four things follow.

• Small parts are cheap to coat and large ones are not, because the constraint is how much fits on a fixture, not how much coating material the parts consume. A price per piece without the chamber load behind it is not a number you can negotiate with.
• Fixtures can be a tooling cost. An unusual shape may need a fixture designed and made before the first run, and that is a one-time cost that belongs in the quotation as its own line.
• Each colour is its own run. Because the colour is a different compound, a three-colour range is three setups with three minimums, not one order split three ways.
• A wrong colour is expensive but often recoverable. Titanium nitride can be stripped chemically so the part can be recoated, though one coating house notes that stripping works on steel alloys but not on every material, because the chemistry attacks some substrates, and that stripping costs roughly 40% of the coating cost. Worth knowing before you approve a colour, and worth knowing that on the wrong substrate it is not an option at all.

One more piece of paper is worth having if you are coating tooling rather than product. There is a published specification for at least one common coating: the aerospace standard AMS 2444 is named by a PVD coating house as the specification for its PVD coating, and by a tool coating supplier as the aerospace specification for titanium nitride. If you are buying coated tools or moulds, writing that number into the purchase documents gives you something to measure against instead of an adjective. Our factory audit guide covers how to check whether a supplier actually owns the process or sends your parts out to someone else, which on coatings is a common arrangement and a common source of delay.

10. A Five-Step Specification Roadmap

1. Write the layer stack from the base metal up. What provides the corrosion protection, what provides the colour, what provides the adhesion, and a thickness for each.
2. Name the compound, not the colour. Titanium nitride for gold, chromium nitride or diamond-like carbon for grey and black, with a target thickness and a tolerance in microns.
3. Check the temperature against the material. Ask for the deposition temperature as a number, and if the part is heat-sensitive, ask what the low-temperature process does to adhesion and to the cycle time.
4. Agree two tests, not one. Thickness with a location, and adhesion by tape test on a production part. Add a wear test in the condition the product will meet, and treat a corrosion result as evidence about the plated layer.
5. Quote by batch and by colour, then keep the sample. Get the chamber load and each colour as its own line, and keep a coated sample sealed as the colour reference for the next order.

Everything in this guide sits inside the Jewelry & Watches sourcing hub, which collects the plating, finishing and marking guides for watches and jewellery in one place.

11. Frequently Asked Questions