ENIG is two layers with no overlap in their jobs
Get the division of labour straight and everything else follows from it.
Electroless nickel-phosphorus (the lower, load-bearing layer)
- Separates copper from solder, blocking copper diffusion into the solder. This is the real reason ENIG stores well.
- It is the solderable surface. Solder wets the nickel and forms intermetallic compounds with it, so joint strength is set at the nickel/solder interface.
- Deposits by chemical reduction — no current, no need for a plated connection — so isolated pads plate too, and thickness barely varies with pad geometry. That is where "flat pads" comes from.
Immersion gold (the upper, custodial layer)
- Keeps the nickel from oxidising during storage. An oxidised nickel surface is not solderable; the gold's entire job is to hold that surface intact until reflow.
- Dissolves into the solder the moment the joint forms, so it contributes nothing to joint strength.
- Deposits by displacement: one nickel atom must dissolve for one gold atom to land.
That last line is the key to every ENIG behaviour worth knowing.
Why the gold must stay thin: gold thickness is paid for in nickel
Electroplated gold uses applied current to reduce gold ions onto a surface; plate longer, get more thickness, and the substrate is not consumed. Immersion gold works nothing like that. For gold ions to deposit, they must take electrons from the nickel surface — which means the nickel must corrode first. Whether the reaction continues depends on whether bare nickel is still exposed.
Three consequences follow:
- The gold layer is self-limiting. Once gold covers the nickel completely, the reaction largely stops. Forcing more gold on means forcing the reaction to keep attacking nickel that is already covered.
- Thicker gold means more nickel consumed. An over-thick gold layer is not better protection; it is deeper damage to the layer that carries the joint.
- Gold thickness cannot compensate for insufficient nickel. The two are not interchangeable quantities. Nickel owns strength, gold owns shelf life. If the nickel is thin, no amount of gold fixes it.
Industry standards specify the two windows separately. IPC-4552 (ENIG performance specification — an industry standard, not a Sunking internal value) sets the direction: the nickel must be thick enough to form an effective diffusion barrier, while the gold is given a window that is bounded on both sides — a minimum to guarantee complete coverage and oxidation-free storage, and a maximum precisely to limit how deeply the displacement reaction eats into the nickel. Use the standard revision cited on your drawing for actual figures. Sunking's corresponding inspection window is confirmed per drawing and per order; the surface finish page likewise states that the final scheme is confirmed against the drawing, material and downstream process.
> One sentence is enough to carry this section: gold thickness is not a "more is better" metric, it is a two-sided tolerance.
Black pad: a failure chain that runs entirely through the nickel
Black pad — also called black nickel — usually presents like this. The bare board looks perfectly normal, with even gold colour. After reflow the joints look acceptable too. Then mild mechanical stress or a downstream test snaps the joint brittlely, and pulling the component off reveals a grey-black, grainy pad surface.
The essential point: that black is not tarnished gold. It is the morphology of corroded nickel that was sitting under the gold. The chain runs:
- The electroless nickel layer is a nickel-phosphorus alloy. Phosphorus enriches at the nickel grain boundaries during deposition.
- The immersion gold reaction corrodes the nickel surface. If that corrosion goes too far, or preferentially attacks high-phosphorus grain-boundary regions, the nickel develops localised intergranular corrosion and an abnormally phosphorus-rich surface layer.
- A high-phosphorus, corroded nickel surface cannot form healthy intermetallics with solder. The solder appears to wet, but the interface is brittle.
- Under load the crack propagates along that interface, the joint separates, and the exposed black nickel is what you see.
Two things about this chain deserve emphasis.
First, incoming visual inspection cannot catch black pad. The surface is gold, it is flat, and both visual checks and routine AOI pass it. It only declares itself after soldering and after loading — by which point components are already populated. That is what makes black pad so expensive in quality terms, and it is why bare-board finish is one of the branches you rule out first when investigating the "joint looked fine but broke on contact" pattern discussed in PCBA solder joint failure causes.
Second, the control point is not in the gold. Since the cause is over-corroded nickel, control has to sit on the deposition quality of the nickel itself and on how tightly the displacement reaction is held. Adding gold thickness only deepens the corrosion. That is exactly why a gold-thickness *maximum* is a genuine quality requirement rather than a cost-cutting excuse.
Where "flat pads" physically comes from: why fine-pitch and BGA choose ENIG
The surface finish page recommends ENIG for fine-pitch devices and BGAs on the basis of flatness — and that flatness comes from the deposition mechanism, not from one metal being inherently better.
Compare hot air solder levelling (HASL). Molten solder covers the pad and hot air blows the excess off. Surface tension leaves domes of varying height, and the smaller and denser the pads, the larger the relative effect. BGA pad arrays are especially exposed to this: uneven pad heights mean uneven collapse at reflow.
ENIG deposits chemically. Nickel-phosphorus grows uniformly from solution, following the pad geometry, without surface tension pulling it into mounds. Pad-to-pad coplanarity is therefore good, and printed paste deposits sit consistently. That connects directly to the SMT end: SMT stencil service aperture design assumes the pads are flat. If the incoming pads are not coplanar, the most carefully calculated area ratio does not save the print.
For completeness, here is the division of labour across finishes, taken from this site's surface finish page; for a wider side-by-side view see PCB surface finish comparison:
| Finish | Application stated on this site | Point relevant to this article |
|---|---|---|
| HASL / LF-HASL | Mature and economical; standard boards, cost-sensitive projects | Molten solder coverage; flatness limited by surface tension |
| ENIG | Fine-pitch, BGA, high reliability | Chemically deposited and flat; nickel solders, gold stores |
| OSP | Volume SMT, lower cost | Organic film; constraints on storage and reflow cycles |
| Immersion silver | Telecom, high frequency, some automotive | Good flatness and high-frequency performance |
| Hard gold | Gold fingers, connectors, repeated mating | Wear resistant; ENIG does not fill this role |
That last row needs stating on its own: the ENIG gold layer is not wear resistant. It is thin, and it is designed to dissolve during soldering. Contact areas that mate repeatedly — gold fingers, connector contact surfaces — use electroplated hard gold, and the two are not interchangeable. If one board carries both BGA pads and gold fingers, the drawing must specify the areas separately.
Writing "ENIG" on a drawing is not enough
This is the most actionable section here. When a drawing says only "Surface finish: ENIG," the nickel thickness, the gold thickness, the standard revision used to judge them and any area exceptions are all left to fab defaults. Four additions make it verifiable:
- Cite the standard and its revision. State that thickness is judged to IPC-4552 and name the revision. This is safer than inventing your own numbers and removes the possibility of the two sides measuring against different expectations.
- Treat gold as a two-sided window. State a maximum, not just a minimum. When only a minimum appears, the supply-chain instinct that "a bit more is safer" pushes in exactly the direction of black pad risk.
- Specify by area. Mark soldering areas as ENIG and mating contacts as electroplated hard gold with area callouts, rather than naming one finish in the title block.
- State the verification method and sampling. Which method measures thickness, at which locations, how many boards per lot, and what the accept criteria are. Follow the logic of the manufacturing tolerance page: a capability has to be written as a measurable quantity before incoming inspection can reach a conclusion.
What incoming inspection can and cannot establish
Within the quality framework, surface finish breaks into three verifiable items and a set of clear limits.
What can be established:
- Thickness. Nickel and gold thickness can be measured lot by lot. This is the direct evidence of whether the drawing window was met.
- Appearance and coverage integrity. Skip plating, exposed copper, contamination and colour variation are catchable visually and by AOI.
- Solderability. A wetting test to the applicable standard shows whether the nickel surface still wets properly after storage.
What cannot be established, and must be acknowledged:
- Black pad cannot be judged from incoming appearance. As above, it only manifests after soldering under load. What incoming control can do is manage the causes — nickel quality and a gold layer kept under its maximum — and retain thickness evidence, not "detect black pad."
- Joint strength is not a bare-board test. It belongs to the PCBA stage, in the territory of solder joint failure causes.
- Plating process data does not substitute for finished-board judgement. Electroplating quality inspection and release covers the copper line; that is a separate set of criteria from the finish layer, and neither stands in for the other.
Put differently: the correct incoming objective for this item is not "guarantee no black pad." It is guarantee that every lot carries thickness evidence, and that the evidence falls inside a two-sided window. Black pad is the outcome of process drift, not a defect sampling can filter out.
FAQ
Q: Is thicker gold more reliable?
A: No. Immersion gold dissolves nickel to deposit gold, so a thicker gold layer means deeper nickel corrosion. Gold is a window bounded on both sides, not a one-sided "more is better" metric.
Q: Is black pad caused by bad gold plating?
A: No. The black is the corroded nickel-phosphorus layer beneath the gold. Both the cause and the control point are in the nickel deposition and in how tightly the displacement reaction is held.
Q: Does gold or nickel determine joint strength?
A: Nickel. The gold dissolves into the solder at reflow; strength comes from the intermetallic interface between nickel and solder.
Q: Why is ENIG recommended for fine-pitch and BGA?
A: Because chemical deposition follows the pad geometry and is not shaped by molten-solder surface tension, giving good pad-to-pad coplanarity. See the surface finish page.
Q: Can gold fingers use ENIG?
A: Not advisable. The ENIG gold layer is thin and not wear resistant. Per the division on this site's surface finish page, repeated-mating areas use electroplated hard gold. One board can specify both, by area.
Q: How long can an ENIG board be stored?
A: No shelf-life figure is published on this site, so none is given here. Storage life depends on gold coverage integrity and packaging conditions, and is confirmed against drawing requirements and the order; see manufacturing capabilities for related handling notes.
Q: What happens if the drawing only says ENIG?
A: Thickness windows, standard revision and area exceptions all fall to fab defaults. Add at least the four items above: standard citation, two-sided gold window, area callouts, and verification method.
To confirm a finish scheme and inspection window for a specific build, send the drawing and stackup for DFM and capability review: request a quote.
---
