Copper Roughness on PCB: 6 Effects From Adhesion to Loss

Copper roughness is the microscopic texture of a copper surface, measured in fractions of a micrometre. It exists for a reason: a rough surface gives the resin something to grip, so the copper and the laminate stay together through drilling, thermal cycling and rework. It also has a cost, because the same texture increases loss in a high frequency conductor.

The two effects pull in opposite directions, which is why roughness is a design decision rather than a process detail. A stack-up built for a 10 Gbps link and a stack-up built for a power board want different surfaces, and the difference follows from the foil and the treatment chosen at lamination.

What Copper Roughness Means

Roughness is expressed as a profile measurement, commonly as an arithmetic mean deviation or a ten point height, taken over a defined sampling length. Two surfaces can share the same number and behave differently, because what matters to a conductor is the shape and the period of the profile rather than its average height.

The relevant scale also depends on the question being asked. Adhesion depends on texture over the length of a resin molecule and its mechanical interlock, while loss depends on texture compared with the skin depth at the operating frequency. The same surface can be smooth for one purpose and rough for the other.

Foil Types and Their Surface Structure

Electrodeposited foil is produced by plating copper onto a drum, and its matte side carries a nodular structure that provides mechanical grip. Rolled annealed foil is rolled from a billet, and its surface is smoother and denser, which suits high frequency work and flexible circuits.

The choice interacts with the resin system. A standard FR-4 with a high-flow resin can tolerate a smoother foil because the resin wets deeply into whatever texture exists, while a low-flow or a filled resin may need more texture for the same bond strength. That is why the stack-up and the foil are specified together.

<img src="https://www.gopcba.com/wp-content/uploads/2026/05/工业物联网-1.jpg" alt="Conductor loss measured on a test coupon after copper roughness change” />

Oxide and Bonding Treatments

Before lamination, the copper receives a treatment such as brown oxide or an alternative bonding chemistry, which grows a dendritic structure on the surface. That structure adds roughness deliberately, and it is the main reason a treated inner layer bonds far better than a clean, bright copper surface.

The treatment thickness and morphology have to be controlled, because an oxide that is too heavy becomes brittle and separates under thermal stress. Modern alternatives aim at the same adhesion with a thinner, less brittle layer, which is one reason the surface can be made smoother for electrical reasons without losing bond strength.

Etching and the Profile It Leaves

Etching shapes the conductor profile and also changes the surface of the copper that remains. An aggressive etchant undercuts the trace and leaves a rougher top surface, while a gentler process holds a more rectangular profile. The etching method therefore affects both the geometry and the surface a designer will rely on.

Surface preparation before resist also matters, because a micro-etch that is too deep leaves a texture that survives the whole process. Where impedance and loss are specified, the combination of foil, treatment, plating and etch should be treated as one system rather than adjusted independently.

Roughness and Conductor Loss

At high frequency, current crowds towards the surface of a conductor, so the current path follows the profile of the copper rather than running along a smooth plane. A rough surface lengthens that path and increases both resistance and the phase delay of the signal travelling along the line.

The effect grows with frequency and is one of the reasons measured insertion loss on a real board can exceed the value predicted from a smooth conductor model. Designs that matter at these frequencies should be verified on a high speed test coupon built with the production stack-up and finish.

Skin Depth and Frequency

Skin depth is the depth at which current density falls to about a third of its surface value, and it shrinks as frequency rises. At a few gigahertz it is a small fraction of the copper thickness, so the surface profile is a large share of the conductor the signal actually uses.

This is why the relationship between roughness and loss is not linear. Once the profile height is comparable with the skin depth, further roughening has a disproportionate effect. Below that point, roughness contributes little to loss, which is why low frequency boards are unaffected by surfaces that matter at microwave frequencies.

Measuring Roughness

Roughness is measured with a profilometer on a prepared sample, or inferred from cross sections and from loss measurements on coupons. Profilometry gives a direct profile, but it is a local measurement, so several positions should be taken across a panel to see the variation.

For production, the more practical control is the foil and treatment specification backed by coupon loss data. Where a supplier changes foil or treatment chemistry, the change should be confirmed with the same coupon measurement rather than with a roughness number alone, because the profile shape can change even when the average height does not.

Balancing Adhesion Against Loss

The compromise is normally expressed in the stack-up. High frequency layers use a smoother foil and a bonding treatment that delivers adhesion without deep texture, while digital and power layers use a conventional treatment. A hybrid stack-up lets each layer carry the surface it needs.

Where a design has both requirements on the same layer, the decision falls to measured data. Coupon loss and a peel test together show whether the chosen surface holds reliably and performs electrically, and that pair of numbers settles an argument that roughness values alone cannot.

When Roughness Becomes a Defect

Roughness turns into a defect when it is uneven or when it comes from the wrong source. A nodular deposit from contaminated plating, a pitted surface from over-etching and a patchy oxide all create local variation that affects adhesion and impedance in the same place.

Because these faults are local, they are easy to miss. Spot checks on a coupon can pass while a region of the panel is affected, which is why process control of the plating and treatment chemistry matters more than inspection of the finished surface. The copper weight record, plating data and etch control together describe the surface that was actually built.

FAQ

Is smoother copper always better for signal integrity? For loss, generally yes, but smooth copper offers less mechanical grip, so the bonding treatment and the resin system have to compensate. The right answer is a matched combination rather than the smoothest available foil.

Does copper roughness change the characteristic impedance? It changes the effective resistance and the phase velocity more than the nominal impedance, though a very rough surface does slightly increase the effective path length. Low loss designs should be validated with coupon measurements that include the real surface.

Can roughness be measured on a finished board? A cross section shows the profile at the cut, and coupon loss data describes the electrical effect on production material. Direct profilometry on a finished board is difficult, which is why the surface is best controlled through the foil and treatment specification.