Via in Pad for Fine Pitch BGA Pads
As ball pitch falls, the room between the pads disappears before the routing does. At a pitch where no trace can pass between two adjacent pads, the usual escape route is no longer available, and the only way to bring the signal out of a fine pitch array is to place the connection directly in the pad and take it to an inner layer or to the far side of the board. The technique is called via in pad, and it trades routing space for process complexity.
Why the Via Has to Be Filled
A filled via is not a refinement of the process, it is a condition of it. When the pad carries an open hole, the solder paste deposited on the pad flows into the barrel during reflow, the paste that remains is insufficient, and the result is a joint with too little solder, a component that lifts at one end, or a connection that measures correctly and fails later. Filling the barrel and levelling the surface removes that path, which is why the whole family of processes is built around plugging and planarising.
The filling has a second purpose. A pad with a depression in it holds less paste than a flat one, and the placement nozzle can sit at a different height over that pad than over its neighbours, which disturbs the placement. Surface flatness is therefore a specification, not a hope, and the depth of the dish after levelling is normally held within about ten micrometres.
The Three Process Routes
Resin plugging followed by plating is the most widely used route. The via is drilled, the barrel is metallised, epoxy resin is forced into the hole, the resin is cured, the surface is ground back to the pad and copper is plated over the filled area so that the pad becomes a continuous flat surface. The result has good flatness and reliable soldering, and it is what most volume production uses. A vacuum plugging process fills the barrel more completely and leaves fewer voids than a simple squeegee application.
A conductive paste can be used instead of resin. The material is filled into the via and cured, and because it is conductive the pad can be used directly. The flow is shorter and cheaper, but the conductivity and the long term reliability are below those of a plated copper surface, so the route suits consumer products where cost matters and the reliability requirement is moderate.
Plating fill uses copper to close the barrel without resin or paste. The electrical behaviour and the reliability are the best of the three, but the process window is narrow and the plating chemistry and the current parameters have to be controlled closely, which makes it the most expensive option. It is chosen for high speed, radio frequency and other applications where the electrical requirement cannot be compromised. An automatically controlled plating line makes this route practical rather than experimental.
Design Points That Decide the Yield
The aspect ratio of the via has to sit inside the process capability. For resin plugging an aspect ratio of about nought point eight to one is the usual recommendation, because a deeper and narrower hole is difficult to fill completely and leaves a void that later appears as a soldering defect under a ball. The ratio is assessed against the finished board thickness and the drilled diameter together rather than separately.
The BGA pad has to be larger than the via so that an annular ring remains around it. A ring that is too narrow reduces the area available for the joint and weakens the adhesion between the pad and the laminate. An annular ring of at least nought point one millimetres on each side is the usual minimum, and the pad is measured on the finished board rather than in the design file.
The soldermask opening needs the same attention. An opening that is too large exposes the edge of the plugged area, which changes the appearance and can affect the soldering, while an opening that is too small leaves mask on the pad. Whether the pad is defined by the mask or by the copper is a decision that has to be confirmed against the process rather than assumed.
Thermal vias in a power pad are a separate case from signal vias. Their purpose is to conduct heat away rather than to carry a signal, and a partial plug or a small mask window over them can be better for dissipation than a completely filled and plated surface. Treating the two as the same feature and applying one rule to both is a common design mistake.
Where the Process Is Used
Fine pitch ball grid arrays are the main application, and the deciding question is not the pitch alone. A device with few pins and enough routing layers can sometimes be escaped by routing between the pads, and a device with a high pin count on a dense board cannot. The decision is made from the actual routing rather than from a general rule about the package.
High density interconnect boards with stacked or staggered vias also use the process, since the same requirement to keep the surface planar applies wherever a via sits in a pad. Memory interfaces, radio frequency modules and high speed digital boards appear in both groups, because each of them combines a small pitch with a signal integrity requirement that does not tolerate a stub left in the pad.
What the Fabricator Has to Control
The equipment and the records behind the process matter as much as the process itself. Vacuum plugging controls the void content in the barrel, grinding and levelling control the surface, and plating over the filled area restores a surface that can be soldered. Inspection then uses cross sectioning to measure the barrel fill and the copper thickness, optical inspection for the surface, and electrical test for continuity, with the results kept under quality management.
Boards from one to four orders of high density interconnect can be produced in this way, with registration between layers held within about twenty five micrometres, and the process sits alongside ordinary through hole work in the same factory. Our PCB manufacturing group produces the board and SMT assembly places the device onto it, which keeps the filled pad and the ball it carries under one responsibility.
Working With the Fabricator Early
`r`n
The decision belongs in the design review rather than in the fabrication order, because it changes the pad, the mask and the layer strategy together. Once the pad geometry has been set for a filled via and the routing has assumed it, removing the process later means reworking the fanout as well as the fabrication notes. Sending the stack-up and the device data for comment before the layout is finished is the cheapest way to make that decision.
`r`n
FAQ
How much more does via in pad cost? The plugging, the grinding and the second plating add process steps and inspection, so the price is above an equivalent plain board. The difference depends on the number of vias, their diameter and the complexity of the board.
Does a fine pitch array always need it? No. Where the pin count is low and the layer count allows routing between the pads, the process can be avoided. The routing decides.
Does it reduce the reliability of the ball joint? Not when the barrel is filled and the surface is flat. The risk lies in a partly filled via or an over deep depression, which is why the process control and the inspection matter.
