Copper Foil Roughness: Design Rules and Process Limits

Copper foil roughness is easy to ignore because the foil itself does not look rough. The side that faces the laminate is treated to give the resin something to grip, and the treatment is a surface profile measured in microns. At low frequency the roughness is irrelevant, but the current in a high speed trace travels in a thin layer near the surface, and that layer follows the profile. The result is a longer path and a higher resistance than the design predicted.

What Roughness Means

Roughness is described by the profile of the foil, and it is quoted in microns as the difference between the peaks and the valleys. A standard foil has a profile of several microns, while a very low profile foil is below one micron.

The profile is created deliberately during the manufacture of the foil, by an electrochemical treatment that grows a dendritic structure on the surface. That structure is what bonds the foil to the resin.

The treatment is applied to one side of the foil, and the profile of the other side is much lower. On an inner layer the treated side faces the prepreg, so the outer surface that the current uses is the untreated one but it is still roughened by the lamination.

The lamination presses the foil against the resin, and the resin flows into the valleys. That is why the roughness cannot be removed after the fact, and why the choice of foil has to be made when the stackup is defined.

Skin Effect and Loss

At high frequency the current in a conductor is confined to a thin layer near the surface, and the thickness of that layer falls as the frequency rises. At a few gigahertz the skin depth in copper is a small fraction of a micron.

When the skin depth is smaller than the roughness, the current has to follow the hills and valleys instead of a flat plane. The path is longer and the resistance is higher, and the effect is called the conductor loss or the insertion loss of the material.

The additional loss can be a large fraction of the total on a long line at a high frequency, and it grows with the frequency faster than the loss of a smooth conductor.

The effect appears in the eye diagram as a closed eye at the far end of the link, and it is often mistaken for a problem with the driver or the receiver rather than with the material.

Foil Types and Profiles

Standard foil has the highest profile and the best adhesion, and it is the cheapest. It is used on a general purpose board where the loss does not matter.

Reverse treated foil has the treatment applied to the smooth side, which gives a low profile on the outer surface while keeping a rough interface with the resin. It is a common compromise for a high speed board.

Very low profile and hyper very low profile foils reduce the roughness further, at a higher cost and with a lower peel strength. They are used for a long high speed link where every tenth of a decibel matters.

The choice is therefore a trade between the electrical loss and the mechanical adhesion, and the trade has to be made with the loss budget in hand rather than by habit.

Adhesion and Reliability

Adhesion is the reason the roughness exists. The resin anchors into the treated surface, and a smoother foil has less mechanical grip, so the peel strength falls.

A lower peel strength matters at the edges of the board, at a large copper area, and anywhere that the board sees a thermal cycle. A delamination that starts at a corner is a reliability failure, and it is much more likely on a very smooth foil.

The resin system can compensate to some extent, since a resin with a higher polarity or a different chemistry bonds more strongly to a smooth surface. That is why a smooth foil is usually paired with a specific resin rather than used with any laminate.

The laminate properties guide lists the peel strength as a material property, and it should be checked together with the roughness when the stackup is agreed.

Impedance and Measurement

Roughness also raises the effective dielectric constant of the surface layer, which lowers the impedance of a line slightly. The effect is small at low frequency and grows as the skin depth falls.

Most impedance calculators ignore the effect, so a measured impedance is often a little lower than the calculated one on a roughened foil. The correction is a few per cent and it is usually inside the tolerance of the design.

The measurement of roughness is done with a profilometer on the foil or with a cross section of the laminated board. A cross section shows the actual interface, which is what matters, rather than the profile of the foil before lamination.

The etching process also changes the surface, because the etchant attacks the copper and can increase the roughness of the trace sidewall. The finished trace is rougher than the foil it came from.

Specification for a High Speed Board

The specification should name the foil type rather than the roughness alone, since the suppliers use different designations for similar values. A note that asks for a low profile foil without a type will be interpreted differently by each shop.

The high frequency laminate guide gives the combinations of foil and resin that are used for a low loss board, and it is the right starting point for a new stackup.

The loss should be verified with a coupon, since the material data sheet gives a typical value and the actual board depends on the trace geometry and on the process. A coupon with a known length and a measured insertion loss is worth more than any catalogue figure.

Finally, the high speed design rules should be applied together with the material choice, because a low loss laminate used on a badly routed line will not save the link.

Practical Rules

Start from the loss budget and the length of the line, not from the material. A short link on a dense board rarely justifies a very low profile foil.

Check the peel strength requirement with the mechanical engineer, because the smooth foils are weaker and the board may see a thermal cycle in the product.

Keep the same foil on both sides of a symmetric stackup, since a difference between the two changes the balance and can warp the panel.

Document the foil type, the resin system and the coupon result in the fabrication notes. The three of them together are what makes the material reproducible on the next order.

FAQ

Does copper roughness matter below one gigahertz? It matters less, because the skin depth is larger than the profile. The effect grows quickly as the frequency rises and it becomes significant in the gigahertz range.

Why not use the smoothest foil available? Because adhesion falls with roughness. A very smooth foil needs a matched resin system and it still has a lower peel strength.

Does roughness change the impedance? It raises the effective dielectric constant at the surface and lowers the impedance slightly. The change is small and is usually inside the design tolerance.