When wire-bond pull strength comes in low, the instinct is to tune the bonder. More often, the ceiling was set weeks earlier — in the plating specification, not at the bonder.
Force, ultrasonic energy, temperature, bond time — these are real process variables, and optimizing them matters. But on builds where pull strength stubbornly underperforms regardless of bonder adjustments, the root cause is usually already committed into the substrate before the first wire goes down.
What the plating specification actually controls
The surface the wire lands on is not a fixed boundary condition. It is a variable, and it is set by the substrate specification weeks before assembly begins. Four parameters do most of the work:
- Gold thickness over the nickel. Below a threshold, the wire is effectively bonding to the nickel barrier, not to gold. Above a plateau point, additional gold thickness stops adding pull strength and adds cost without adding reliability. The useful window between those two limits is narrow.
- Nickel barrier integrity. Phosphorus content and bath control determine whether the nickel–gold interface is sound or predisposed to the interfacial fracture mode commonly called “black pad” — parts that pull clean but weak, with no visible surface defect.
- Surface contamination and oxide state. The condition of the pad between plating and bonding sets real bondability. Elapsed time, storage conditions, and handling between the plating bath and the bonder all compound.
- Finish selection. ENEPIG and ENIG behave differently under gold wire. ENEPIG adds a palladium barrier that changes the gold-wire interaction and expands the usable thickness window. The substrate specification, not the bonder operator, makes that choice.
The pull-strength plateau
The pattern is consistent enough to be predictable: pull strength climbs with gold thickness, then flattens. Below the knee of that curve, small plating-thickness variation produces large swings in pull numbers. The process is operating on a cliff. Above the plateau, the relationship decouples — more gold, same strength.
The practical consequence is that a build tuned to operate below the knee of its finish’s curve is vulnerable to every normal batch-to-batch variation the plating supplier considers acceptable. None of that variation is a plating defect. All of it moves the pull numbers.
Where to look first
When a build is chasing pull strength at the bonder and cannot get there, the most productive first step is an audit of the incoming substrate finish — not another pass at the bonder recipe. Measure the gold thickness. Request the plating bath records. Compare the finish specification against the bond wire and geometry combination actually being run.
If the substrate is below the knee for its finish type, bonder optimization cannot close the gap. The constraint is upstream. Identifying it early — before the program concludes the bonder is the problem — is the difference between a one-lot correction and a multi-lot debugging campaign that leaves the root cause intact.