Copper Foil: Design Rules and Process Limits
Every copper layer on a board begins as a foil, and there are two ways to make that foil. One is deposited onto a rotating drum from solution; the other is rolled from a cast ingot. The two processes produce copper with different grain structure, different surface roughness, and different mechanical behaviour, and those differences propagate into etching, bending, and high frequency loss.
The Two Foil Families
Electrodeposited foil is grown on a polished drum from a copper sulphate bath, so one face is smooth and shiny, the other is rough and matte. Rolled annealed foil is reduced in thickness by successive rolling passes and then heat treated, which leaves an equiaxed grain and a surface that is smooth on both faces.
Neither is universally better. Electrodeposited foil dominates rigid boards because it adheres strongly to laminate and costs less, while rolled annealed foil dominates flexible circuits and very high frequency work because it bends without cracking and loses less energy to the conductor surface.

How Electrodeposited Foil Is Made
Copper is plated onto a titanium drum and peeled off as a continuous web. The grain grows in columns perpendicular to the surface, and the growth face becomes the rough matte side that bonds to the laminate. The drum side stays smooth because it is a negative copy of the polished drum.
Grade follows treatment. Standard foil has a moderate nodular treatment, while a high elongation variant is treated to raise ductility for boards that will be bent or thermally cycled hard. The nodular structure raises the effective surface roughness, which is beneficial for adhesion and unhelpful for loss.
How Rolled Annealed Foil Is Made
An ingot is hot rolled, then cold rolled in successive passes with intermediate anneals, then finish rolled to the required thickness. The final anneal softens the copper and gives it the elongation values that flexible circuits depend on. Because the process is mechanical, the achievable thickness range below about eighteen micrometres is limited compared with the deposited route.
Rolled foil also carries a direction. The grain is elongated along the rolling direction, so the foil is more tolerant of bending across the grain than along it. On a flexible circuit this becomes a design rule: the rolling direction must be specified relative to the bend line.
Surface Treatment And Adhesion To Laminate
Bare copper does not bond reliably to epoxy. Foil is therefore supplied with a treatment that usually consists of a dendritic copper or copper oxide layer followed by a coupling agent that links to the resin. The treatment is applied to the bonding face only, and it is what makes laminate adhesion repeatable at production volumes.
The treatment is sensitive to storage. Treated foil that has absorbed moisture or been exposed to a contaminated atmosphere bonds less well, which is one reason foil has a shelf life and a specified storage condition. A laminate adhesion failure at the press is more often a foil storage issue than a resin problem. Our notes on laminate material properties describe how the bond is tested.
Roughness, Skin Effect, And High Frequency Loss
At high frequency, current concentrates near the conductor surface and the skin depth falls to a few micrometres at gigahertz frequencies. If the surface roughness is of the same order, the current path is lengthened and the loss rises. A rough nodular foil can add measurable loss on a long high speed channel even though the dielectric is the same.
Low profile and very low profile foils reduce this effect by flattening the treatment. They cost more and they adhere less readily, so they are used where the loss budget demands it rather than as a default. The choice interacts strongly with the laminate, and the pair should be selected together. Our notes on high frequency laminates cover the material side of that decision.

Ductility And Flex Applications
Rolled annealed foil extends by fifteen to twenty-five percent before breaking, while an electrodeposited foil of the same thickness may reach only a few percent. That difference decides whether a conductor survives repeated bending. A flexible circuit built on deposited foil will crack at the bend after a modest number of cycles even if the conductor geometry is correct.
Ductility is measured by elongation and by a bend test on the finished construction, not by the foil datasheet alone, because the lamination and plating steps change the properties. Our notes on bendable circuit materials describe how the qualification is run.
Etching Behaviour And Profile
Grain structure changes how copper etches. Rolled foil etches more evenly and produces a straighter sidewall for the same process, because the grain does not channel the etchant. Electrodeposited foil with a columnar grain tends to etch faster along the grain boundaries, producing a rougher edge at fine dimensions.
This matters for the compensation applied at CAM. A foil that undercuts differently requires a different artwork offset, and using one compensation value across two foil types on the same board produces two different finished widths. Our notes on the etching process explain how the offset is established.
Thickness Options And Availability
Standard thickness designations are twelve, eighteen, thirty-five, and seventy micrometres, corresponding to one third, one half, one, and two ounces per square foot. The thinner gauges are needed for fine line work, since etching a thick foil removes too much width before the last copper clears.
Availability differs between the two families. Rolled annealed foil is produced in fewer thicknesses and by fewer suppliers, and lead times can be long. Where a design depends on it, the availability of the exact gauge should be confirmed before the stackup is frozen rather than after the first order.
Specifying Foil On The Stack Drawing
The stack drawing should state the foil type, the thickness per layer, the treatment where it matters, and for flexible work the rolling direction relative to the bend. It should also state the storage condition expected of the laminate supplier if the board is destined for a long shelf life.
Leaving the type unstated means the fabricator will choose the cheaper option, which is reasonable behaviour and wrong for a flex or high frequency design. Stating the type and the reason in one line on the drawing removes the ambiguity and keeps the supply chain aligned with the design intent.
Substitution Risk And What To Lock Down
Foil is one of the few materials a fabricator can change without altering anything visible on the finished board. Substituting a standard electrodeposited foil for a low profile one changes conductor loss, and substituting a deposited foil for rolled annealed copper changes bend life, with no dimensional difference to reveal it. On a design where either matters, the foil type belongs in the controlled specification rather than in a note.
A change notice should be required for any substitution, and the notice should state the effect on loss or on elongation as well as the price. Where the design is not sensitive, allowing substitution is reasonable and can shorten lead times, provided the decision is recorded so a later investigation is not left guessing which build used which foil.
The same logic applies to thickness. A fabricator may propose a half ounce layer where one ounce was drawn, if the current requirement is met, and the change shifts the etch compensation and the impedance of any controlled trace on that layer. Both effects are invisible on the artwork and visible only in the measurement data.
FAQ
Which foil should a rigid board use? Electrodeposited foil is the normal choice. Rolled annealed foil is reserved for flexible circuits and for designs where conductor loss dominates.
Does foil roughness affect impedance? It affects loss rather than the nominal impedance, because the current path is lengthened. The effect grows as frequency rises.
Can the two foil types be mixed in one stack? Yes, but each layer needs its own etch compensation, and the difference should be stated rather than assumed.



