In production we treat FPC copper crack after repeated flexing as a fatigue-driven open-circuit risk that must be caught at CAM and controlled through copper selection, thickness, and bend-zone geometry. Most factories will flag any dynamic flex area where the copper is ED type, thicker than 1/2 oz, or where the designed bend radius falls below roughly 6–8× the total stack thickness. CAM engineers normally force a switch to RA copper, thin the copper if the customer allows, enlarge the radius, or add coverlay reinforcement before the job is released to the floor. That is the practical production rule we apply every day.
How cyclic strain turns surface slip into a through-copper crack
What we typically see on the CAM side is that the copper foil itself is the weakest link under repeated flexing. Every bend cycle puts the outer fibers of the copper into alternating tensile and compressive strain. Once that strain exceeds the elastic limit, fine slip bands form on the surface. After a few hundred to a few thousand cycles those slip bands become persistent and act as crack initiation sites. The crack then grows inward along grain boundaries or through the grains until it reaches the opposite side of the foil and the circuit opens.
ED copper is especially sensitive because its columnar grain structure is oriented perpendicular to the foil surface. Under bending those columns separate easily. RA copper has elongated grains parallel to the surface, so the same strain is distributed over a longer path and crack growth is slower. That is why flexible PCB copper fatigue almost always appears first on ED copper jobs that were never intended for dynamic flex.
Material thickness and residual stress from the manufacturing process also feed the problem. Etching leaves a rough surface that concentrates stress. Any residual tensile stress left after plating or annealing adds to the cyclic load. In the panel the copper is constrained by the polyimide and adhesive; when the stack is bent the neutral axis shifts and the copper sees higher strain than a free foil would. All of these factors stack up in real production and turn a design that looks fine on paper into a crack after a few thousand flex cycles.
Why thicker copper fails earlier under the same bend radius
From a fabrication standpoint the relationship is straightforward: thicker copper experiences higher surface strain for any given bend radius. Strain scales roughly with distance from the neutral axis, so a 1 oz foil (35 µm) sees nearly twice the peak strain of a 1/2 oz foil (18 µm) when both are bent to the same radius. Higher strain means faster accumulation of plastic damage and earlier crack initiation.
In practice we see 1 oz copper start to show micro-cracks after 5 000–10 000 dynamic flex cycles at a 10× radius, while the same construction with 1/3 oz or 1/2 oz copper often survives 50 000–100 000 cycles. The difference is even larger when the copper is ED rather than RA. That is why CAM routinely pushes for the thinnest copper the electrical design will accept in any zone marked as dynamic flex.
Thicker copper also makes the coverlay or adhesive less effective at restraining the foil. The copper wants to stretch or compress more, the coverlay cannot fully follow, and local delamination or buckling appears. Once the copper is free to move independently, crack growth accelerates. This is the manufacturing reason we treat copper thickness as a first-order control variable rather than a secondary electrical choice.
What happens on the floor when the fatigue limit is ignored
If the design is released without compensation, the first symptom is usually intermittent opens during final electrical test or reliability flex testing. Boards that pass initial continuity fail after a few hundred cycles of the customer's flex fixture. Yield drops in the 10–30 % range on high-cycle jobs. Scrap is expensive because the polyimide stack is already complete; there is no economical rework once the copper is cracked.
Worse is the field return. A crack that is only partial at shipment continues to grow under the customer's daily flexing and eventually opens the circuit weeks or months later. That triggers failure analysis, often a CAPA, and can delay the next order while the factory and customer argue over whether the copper type or the bend radius was the root cause. In high-volume consumer flex applications this kind of latent failure is the one that damages both schedule and reputation the most.
Process-wise we also see secondary damage. Cracked copper can lift the coverlay, allowing moisture or ionic contamination into the stack. On boards that still pass continuity the insulation resistance starts to drift, creating soft failures that are hard to catch at AOI or flying probe. All of these effects appear because the original strain limit was exceeded and never corrected in CAM or process setup.
How CAM and process engineers actually stop the cracks
The first action is always copper type and thickness. We specify RA copper for any dynamic flex zone and limit thickness to 1/2 oz or thinner unless the current density absolutely requires more. When the customer insists on 1 oz we enlarge the minimum bend radius to 10–12× total thickness and add a second coverlay layer or a stiffener edge to move the high-strain zone away from the copper features.
On the CAM station we also check the orientation of the copper grain relative to the bend axis. RA copper must be rolled so the grain direction is perpendicular to the bend line; if the supplier panel is oriented the wrong way we rotate the array or reject the material. Etch factor is controlled to keep the copper edge smooth—no undercut that creates additional stress raisers. After etching the panels go through a controlled anneal to relieve residual stress before coverlay lamination.
Coverlay design is equally critical. We keep the coverlay opening well away from the bend apex and use a flexible adhesive system that matches the copper elongation. In some high-cycle jobs we add a thin polyimide reinforcement strip on the outer side of the bend to shift the neutral axis and reduce copper strain. All of these changes are documented in the DFM report so the customer sees the exact compensation that was applied.
Process control on the floor includes limiting the number of pre-bake and press cycles that can embrittle the copper, and verifying the final stack thickness so the calculated bend radius remains valid. Final reliability samples are flexed to the customer's specified cycle count before the lot is released. That combination of material choice, geometry adjustment, and process discipline is how we keep flexible PCB copper fatigue from becoming a shipment problem.
When the factory will accept a tighter design
Exceptions exist for static flex or very low cycle counts. If the board is bent only once during assembly and then stays fixed, we can release 1 oz ED copper with a 5× radius provided the customer signs off on the reduced life expectation. The same applies to designs that flex fewer than a few hundred times in the product lifetime. In those cases we still note the limitation in the DFM file and keep a sample set for accelerated testing, but we do not force the full RA-plus-thin-copper package.
The trade-off is clear: tighter geometry or thicker copper buys cost or electrical performance at the expense of fatigue life. Once the cycle count or the field environment moves into true dynamic flex territory, the factory returns to the strict controls. That is the practical boundary we work to every day.
