ENPA / ENP · Electroless Ni / Pd / Au
ENIG fails at the gold step, because gold deposits by dissolving the nickel underneath it. A 0.1 µm layer of electroless palladium removes the galvanic cell from the pad — and that changes what the machine has to be.
There is no way to make an immersion gold bath stop attacking nickel. It is a displacement reaction: gold goes down because nickel goes into solution. The only fix is to put something between them that is deposited chemically instead.
In ENIG, the immersion gold step is galvanic by design. If the nickel surface is uniform and active, the attack is uniform and the joint is fine. If it is passivated — high phosphorus, oxide, sulphur contamination, a slow entry — the attack concentrates at grain boundaries and the pad fails after assembly. That is black pad, and it is a process-window problem, not an operator error.
Electroless palladium is deposited by chemical reduction. It does not consume the nickel, it covers the nickel completely, and gold then lands on palladium. The consequences are practical: thinner gold is enough, the joint survives more reflows, copper wire bonding becomes possible, and the pad edge stops being a galvanic cell.
An ENEPIG line is not an ENIG line with one more tank. Palladium drag-in decomposes a nickel bath — palladium catalyses nickel reduction, so a few millilitres carried forward can plate the tank walls and the heaters. The layout, the rinse architecture and the exhaust all change to prevent that, and the wetted materials change because palladium baths attack the stainless steel that nickel baths are happy with.
| Station | Chemistry / condition / target |
|---|---|
| 1. Acid cleaner | 45 – 55 °C, surfactant, ultrasonic optional |
| 2. Soft etch | sodium persulfate, 25 – 30 °C, 1 – 2 µm removal |
| 3. Pre-dip / catalyst | palladium-tin or ionic Pd, 25 – 35 °C |
| 4. Accelerator | removes tin shell, exposes active Pd |
| 5. Electroless nickel | hypophosphite, 85 – 90 °C, pH 4.6 – 5.0, 3 – 6 µm, 7 – 9% P |
| 6. Cascade rinse | two stages, conductivity monitored |
| 7. Electroless palladium | 45 – 60 °C, 0.05 – 0.20 µm |
| 8. Cascade rinse | two stages, dedicated to Pd — no shared drain |
| 9. Immersion gold | 80 – 85 °C, pH 4.5 – 5.5, 0.03 – 0.08 µm |
| 10. Rinse and dry | hot DI + air knife, no water marks on pads |
Sequence shown is the standard ENPA configuration. Layer thicknesses follow the customer specification and IPC-4556; the machine is built to hold the window, not to define it.
Drip trays under every lift, a physical gap between the nickel and palladium modules, separate exhaust ducts, and two rinse stages between them with conductivity interlocks. If the rinse conductivity rises, the line stops rather than carrying palladium forward.
This is why an ENEPIG line is longer than an ENIG line for the same output: two extra rinse stages and one extra process tank, not one tank.
Palladium: PVDF and PPS wetted parts, PTFE-coated (Teflon) heaters. A stainless heater surface catalyses palladium plate-out and the bath plates itself out in a week.
Nickel: 316L with PP lining, quartz or Teflon heater at low watt density so there is no local boiling at the element. Gold: PP / PVDF, no stainless in contact. Filtration at 1 µm for nickel, 0.5 to 1 µm for palladium and gold.
At 85 to 90 °C, a 1 °C difference across the tank is roughly a 10% thickness difference corner to centre. Cascaded PID with two sensors per bath, independent over-temperature cut-out, and level interlock on the heaters.
Palladium is more forgiving on temperature and less forgiving on concentration: it is dosed from analysis, not from a timer, because palladium thickness is what decides wire-bond pull strength.
Nickel 4 to 6 MTO with automatic replenishment from a nickel analyser plus pH control; palladium and gold from concentration analysis. The line counts metal turnovers, not days, and raises the bath-change flag before the thickness trend moves.
Every lot is logged with bath age, so a bond failure six months later can be traced to the bath it was plated in.
| Parameter | Range / standard |
|---|---|
| Panel size | up to 620 × 620 mm (larger engineered per project) |
| Nickel thickness | 3 – 6 µm, phosphorus 7 – 9% |
| Palladium thickness | 0.05 – 0.20 µm |
| Gold thickness | 0.03 – 0.08 µm |
| Thickness uniformity | ± 10% within panel |
| Bath temperature control | ± 1 °C (nickel and palladium) |
| Filtration | 1 µm Ni, 0.5 – 1 µm Pd / Au, continuous |
| Wetted materials | PP / PVDF / PPS / PTFE / quartz, selected per bath |
| Heating | PTFE-coated or quartz, low watt density |
| Automation | automatic dosing, MTO tracking, SPC on thickness, MES interface |
| Line configurations | ENPA (Ni/Pd/Au) or ENP (Ni/Pd) when gold is plated elsewhere |
| Reference standards | IPC-4552B (ENIG), IPC-4556 (ENEPIG), customer spec |
ENEPIG capacity is set by the nickel dwell, not by conveyor speed. Three to six micrometres at 85 to 90 °C takes 15 to 25 minutes; palladium needs 8 to 15 and gold 8 to 12. Everything upstream and downstream has to match the slowest module.
racks/hour = nickel tanks × 60 ÷ nickel dwell × transfer efficiency
m²/hour = racks/hour × panel area per rack
Customers usually arrive asking for m²/hour and leave having chosen a nickel tank count. We size the nickel module first, then match palladium and gold to it — adding a second palladium tank buys nothing if the nickel module cannot feed it.
ENEPIG and ENIG lines from this platform run at OSAT-facing PCB plants and at automotive and RF module manufacturers in China, Thailand and India, on boards that are wire bonded after assembly — power modules, sensor packages, and fine-pitch HDI with 0.35 mm pads.
For the electrolytic alternative — thick gold on connector fingers and test pads — see the gantry plating line, which plates nickel and hard gold on the same flight bar.
When the pad has to be wire bonded, when the pitch is below 0.4 mm, or when the board sees more than three reflows. In ENIG the gold step is a displacement reaction that dissolves nickel; if the nickel surface is passivated the attack is local and the joint fails — the classic black pad. Palladium is deposited by chemical reduction, so it does not attack the nickel, and it lets you run a thinner gold layer.
0.10 to 0.20 µm. Below about 0.05 µm the barrier is incomplete and nickel can still diffuse to the gold interface; above 0.25 µm you are buying cost without measurable pull strength. For gold wire, 0.05 to 0.10 µm is usually enough. The final number should come from your assembly house's bond test, not from a table.
Mechanically, yes: insert a palladium module and two rinse stages. In practice the nickel exit section usually has to be rebuilt, because palladium drag-in poisons the nickel bath and the exhaust and drip trays were never laid out for it. We quote both — conversion and new line — and tell you which one holds better thickness uniformity.
Four things, and all four are machine-related: a palladium barrier, nickel at 7 to 9% phosphorus held by bath control, a microetch that is filtered and not contaminated with sulphur, and rinse water that is actually rinsed — cascades with conductivity monitoring, not a single dip.
Nickel runs 4 to 6 MTO with automatic replenishment from a nickel analyser and pH control; palladium and gold are replenished from concentration analysis. The line tracks MTO, not calendar days, and flags a bath change before the thickness trend moves.
Related lines: VCP plating lines · via filling lines · gantry lines for 40:1 HDI
Send us the board: layer count, panel size, hole diameter and depth, copper target and required m²/hour. We will return a cell count, a rectifier schedule and a pilot-run plan.