In production we plate FPC gold fingers with hard gold at 0.3 µm to 1.5 µm depending on the insertion life the customer specifies. Nickel underplate stays fixed at 3–5 µm. CAM engineers pull the thickness from the drawing or from the mating-cycle note; if neither is clear we default to 0.5 µm hard gold and flag the file for confirmation. Soft gold or ENIG is rejected for fingers because the deposit wears through after a few dozen cycles.
Why hard-gold thickness is locked to insertion life rather than electrical resistance
Contact resistance on a fresh finger stays low even at 0.15 µm of hard gold. The real driver is wear. Every insertion scrapes a few nanometers of gold. After the gold is gone the nickel oxidizes and contact resistance climbs. Factories therefore map thickness directly to expected mating cycles:
- 30–50 cycles → 0.3 µm hard gold is normally enough
- 100–300 cycles → 0.5 µm becomes the practical minimum
- 500–1000 cycles → 0.8–1.0 µm is required
- Above 1000 cycles or harsh vibration → 1.25–1.5 µm plus tighter nickel control
These numbers come from internal wear tests run on polyimide flex material with the same surface roughness and plating chemistry used on production panels. The numbers are not theoretical; they are the point at which we start seeing resistance rise above 20–30 mΩ on the test fixtures.
Where the process window forces thickness decisions during panel plating
Hard gold on FPC is electrolytic, selective, and run after the coverlay is already laminated. The polyimide and adhesive outgas slightly in the plating tanks, so bath chemistry drifts faster than on rigid boards. Current density across a flex panel is also less uniform because the material is thin and can float or wrinkle on the plating rack. That combination means the gold thickness variation across a single panel is typically ±15–20 %. If the designer asks for 0.3 µm minimum, the process must target 0.4–0.45 µm average just to keep the low points above 0.3 µm. That is why CAM engineers rarely accept a 0.3 µm call-out without an explicit "minimum after plating" note; otherwise the plating line will push the average higher to protect yield.
Nickel thickness is less variable but still critical. Below 3 µm the nickel becomes discontinuous on the rolled-annealed copper used for most FPCs; above 6–7 µm the flex becomes stiffer and the gold finger can crack when the cable is bent near the connector. Production therefore holds nickel between 3 µm and 5 µm and only opens the window for high-reliability or thick-gold builds.
What happens on the line when thickness is under-specified or uncontrolled
If the gold is too thin, the first failure mode is not open circuit; it is intermittent high resistance after a few dozen insertions. The customer sees it as "connector works on the bench but fails after assembly into the product." In the factory we catch some of these with a 50-cycle wear test on sample fingers, but the test is destructive and only runs on first articles or process audits. Once the job is in volume, thin gold simply ships and the problem appears in the field.
Over-thick gold creates different headaches. Above roughly 1.5 µm the deposit becomes brittle; when the flex is bent or the connector is inserted at a slight angle the gold can micro-crack. Those cracks expose nickel and the same oxidation path begins. Thick gold also lengthens plating time, raises cost, and increases the risk of gold bridging between closely spaced fingers if the dry-film mask is not perfectly sealed. Scrap rates climb because the gold cannot be stripped and re-plated without attacking the nickel or the polyimide.
Registration between the gold finger and the coverlay opening is another related failure. If the coverlay edge sits on top of the gold instead of stopping short, the gold thickness in that overlap zone is reduced by the mechanical stress of lamination. We regularly see 0.1–0.15 µm loss under the coverlay edge; that is enough to push a marginal 0.3 µm finger below the wear limit.
How the factory actually controls and compensates for FPC hard gold thickness
CAM starts by converting the customer's thickness call-out into a process target. For a 0.5 µm minimum we set the plating recipe to 0.6–0.65 µm average and program the rectifier for the panel size and copper area. After plating, XRF measurements are taken at five points on every panel—two ends, center, and both sides of the longest finger set. Any panel with a reading below the minimum is stripped and re-plated or scrapped if the nickel is already compromised.
For high-cycle designs we also add a hard-gold "keep-out" zone in the coverlay data so the gold surface is fully exposed and the thickness is not reduced by lamination pressure. On panels with mixed rigid-flex areas the plating current is adjusted zone-by-zone because the rigid sections steal current and the flex fingers would otherwise come out thin.
Bath control is tighter than for rigid boards. We run cobalt or nickel hardeners at the high end of the supplier range so the deposit Vickers hardness stays above 150 HV; softer gold wears faster even at the same thickness. Free cyanide and gold concentration are checked every four hours because the thin flex material increases drag-out and accelerates bath aging.
When the customer supplies a connector drawing with a stated life (for example 500 cycles per the connector manufacturer), we treat that number as the governing requirement and ignore any lower thickness that may appear in the PCB fabrication notes. The connector OEM's wear data is more reliable than a generic 0.3 µm call-out on the flex drawing.
When thinner gold or process exceptions are accepted
Low-cost consumer products that will see fewer than 20 insertions in their lifetime can run 0.2–0.25 µm hard gold if the customer accepts the risk and signs a process deviation. In those cases we still keep the nickel at 3 µm minimum so the surface remains solderable if rework is ever needed. Prototype lots that will only be mated a handful of times on the bench are sometimes released with ENIG instead of hard gold; the file is clearly marked "prototype only – not for production."
Designs with gold fingers that are only used for test-point probing (never for a permanent connector) can also drop to 0.15–0.2 µm because the wear mechanism is different. Once the application moves to a real edge connector or ZIF socket, the full thickness table applies again.
In short, FPC hard gold thickness is not chosen for appearance or for a generic IPC number. It is chosen so the gold survives the exact number of insertions the product will see. Production sets the plating target high enough to absorb the natural thickness variation of flex plating, measures every panel, and only relaxes the rule when the real mating life is known to be very low.
