Gold Wire Bonding: Surface Requirements on PCB Pads
Wire bonding places a far more demanding requirement on a pad surface than soldering does, because the bond is formed by pressing a fine gold wire against the pad with heat and ultrasonic energy. Any contamination, porosity or thickness variation at that interface reduces the bond strength, and the failure may not appear until the product has been through a thermal cycle or a mechanical shock. Wire bonding is therefore a finish and cleanliness problem as much as it is an assembly process, and the board specification has to reflect that.
What Wire Bonding Needs from a Pad
The bond forms by diffusion between the gold wire and the metal on the pad, driven by the combination of temperature, pressure and ultrasonic vibration. For that to happen consistently, the surface has to be clean, chemically stable and metallurgically compatible with gold.
A surface that oxidises, that carries organic residue or that is plated with a metal that forms a brittle compound with gold will produce a bond that looks acceptable and fails a pull test. The bonding process itself is very repeatable, so the variability almost always comes from the board. When a bonding line suddenly produces weak bonds on one product and not on another, the difference is almost always in the surface rather than in the machine.
Finish Options for Bonding
ENIG is the most common finish for bond pads, because the nickel provides a diffusion barrier and the gold provides a surface that gold wire bonds to readily. Immersion gold over nickel is used where the wire is aluminium, although that combination requires more care. Aluminium wire on gold forms a brittle intermetallic at elevated temperature, so the interface has to be controlled carefully where that combination is used.
Hard gold is unsuitable for bonding because of the hardening elements in the deposit, and OSP is unsuitable because the film has to be removed and the exposed copper is not stable enough. The choice is therefore narrow and it is usually made at the design stage. Changing the finish after the layout has been released means new artwork and new tooling, so the decision should be made before the design is frozen.
Nickel and Gold Thickness
The nickel layer has to be thick enough to stop copper diffusing to the surface, and the gold layer has to be thick enough to present a consistent bonding surface without being so thick that it changes the mechanical behaviour of the bond interface. Thickness should be specified as a band with a minimum and a maximum, and the band should be confirmed on a coupon rather than on a product pad.
Thickness uniformity across the pad matters more than the average value. A pad that is thin at one edge will bond well at the centre and weakly at that edge, and the pull test will show the weakness even though the nominal thickness is inside specification. Where porosity is suspected, a section through the layer and a simple chemical test will both give useful information. Our plating thickness guide covers the measurement methods. X-ray fluorescence on several points across the pad is the practical way to check uniformity without destroying the board.
Surface Cleanliness and Contamination
The most common cause of a weak bond is contamination, and the sources are mundane: mask residue, flux from a previous assembly step, silicone from a release film, and fingerprints. Each of them forms a barrier that stops the diffusion that makes the bond.
Silicone is the worst of them because it spreads and because it is difficult to remove once it is on the surface. Any silicone based material in the vicinity of a bonding area should be excluded from the process, and the exclusion should be written into the shop rules rather than left to individual judgement. A single silicone wipe used near a bonding area can affect a whole batch of boards, and the effect is not visible.
Plating Quality and Porosity
Porosity in the gold layer exposes the nickel beneath it, and the exposure changes the way the wire deforms at the bond interface. The result is an inconsistent bond that may pass a pull test on one pad and fail on the next one on the same board.
Porosity is reduced by a well maintained bath, good filtration and adequate agitation. Our solderability guide covers the surface tests used on finishes, and a porosity test belongs in the qualification of any bonding surface. The test is cheap compared with the cost of a batch of weak bonds discovered at the end of the assembly line.
Pad Geometry and Mask
The pad has to be large enough for the bond footprint and for the second bond if the wire lands on the board. It also has to be separated from the mask by enough distance that the mask does not encroach during bonding, because a mask edge inside the bond area prevents the wire from deforming as intended. The mask should be kept clear of the bonding area with a margin that accounts for the mask registration tolerance.
Pad spacing is driven by the bonding equipment and by the wire diameter. A design that pushes the pitch below what the machine and the finish can support will produce consistent failures, and the limit should be confirmed with the assembly house before the layout is released. A supplier who says a process is possible is not the same as a supplier who has qualified it, so a sample run is worth requesting.
Bond Strength and Pull Testing
Bond strength is checked by a wire pull test or a shear test on a sample, and the result has to be compared with a limit that accounts for the wire diameter and the bond type. The failure mode is as important as the force, because a bond that fails at the interface is a board problem while one that breaks in the wire is a process setting problem. Recording the failure mode with the force is what turns a routine test into a diagnostic.
A sample plan is essential, because bonding is a statistical process and a single successful bond proves very little. Our quality documentation describes how these results are classified at gopcb.
Storage and Shelf Life
A gold surface does not oxidise, but it does accumulate contamination from the atmosphere and from its packaging. A board that has been stored for a long period may need a plasma clean before bonding, and the requirement should be established from the storage history rather than from the calendar.
Packaging therefore matters for bonding boards more than for soldering boards. A clean, sealed bag with a desiccant keeps the surface in the condition it left the finishing line, and it removes an entire class of bonding failure from the process. Where a board will be bonded after a long storage period, a plasma clean before bonding is a common and effective precaution.
Process Control Points
The controls are the finish specification, the nickel and gold thickness and their uniformity, the bath maintenance, the cleanliness of the handling environment, the storage and the periodic bond test. Each of them is measurable and each of them should be recorded with the lot. The combination of thickness data, bath records and bond test results is what allows a bonding problem to be resolved quickly.
Where a bonding process is qualified, the qualification should include a thermal cycle followed by a pull test, because a bond that is marginal at room temperature may fail after the product has been through its assembly flow. That test is what links the board to the reliability of the finished device. A finish that passes every incoming check and fails after assembly has not really been qualified.
FAQ
Which finish is best for gold wire bonding? ENIG is the usual choice, because the nickel barrier and the gold surface give a stable and bondable interface. Hard gold and OSP are not suitable for this application.
Why does a bond fail on one pad but not the next? Usually porosity or a local thickness variation, or a contamination spot from handling. The pull test result and the failure mode will show which of those is responsible.
Does a bonding board need special packaging? It should be packed clean and sealed with a desiccant, because airborne contamination and silicone vapour degrade a bonding surface even though the gold itself does not oxidise.
