Exploding Solder Balls

Operation Solder Voids

Exploding Solder Balls

Figure 1: Typical BGA pad with extending "dogbone" style via.

Could such a thing occur?

People are soldering tens of thousands of BGA components a day. Most of this activity occurs in the flow of manufacturing and some during rework. Over ninety-nine percent of BGA placement is performed flawlessly. On the other hand, industry's dynamic forces constantly push the envelope. Occasionally things go wrong.

Circuit board designers are continuously looking for more real-estate.

Figure 2: BGA via-in-pad arrangement.

That can be as difficult as looking for apartments in Manhattan, but that doesn't stop them from searching.

Used to be that BGA pads were always routed out to a via on the same side of the board as the pad. Now more and more people are trying to reclaim that space by using the BGA pad itself as a via. See Figures 1 and 2.

That's fine of course when it works and it works most of the time. Yet there are occasions when the via in pad solution causes disruptions in BGA placement. There are a number of different issues that are addressed on various technical forums but today we'll only address one; filled vias.

Figure 3: Vias viewed with oblique x-ray.

Densely spaced BGA locations present numerous challenges for placement. The thermal and electrical characteristics must be carefully calculated and balanced. At a minimum, when the board is exposed to solder reflow temperatures the site must behave, remain stable, not move around or blow up.

In a recent case a customer was experimenting with a variety of different fill materials. T sub g variance and voiding in the fill material caused failures in BGA placement. The T sub g variance at one large BGA pattern caused the board to warp at this location. That problem was addressed and solved and the voiding issue was left. The BGA via voiding looked like this, see Figure 3.

Figure 4: BGA balls at the top of filled vias-in-pads. Note light areas in via column indicating voids.

This picture is taken post board assembly and prior to BGA rework. As you can see, there is a extensive voiding in the pictured vias. What's the problem with voiding you ask? Theoretically, if voids were stable during the thermal transients of BGA rework, there would be no problem, but this is a case where common sense dictates that there will be dynamic forces at work once these voids are exposed to rework temperatures. That was indeed the case. Figure 4 provides a picture of voids in the vias post-reflow. The impact of via voids on the BGA balls is not evident yet.

The via is like a long tower with a minaret-like ball placed on its top. Figure 5 shows what can happen when some of those via voids rise up into the BGA sphere. BGA voiding is a condition that is battled by rework personnel without the presence of

Figure 5: BGA ball void aggravated by gases escaping from via in pad.

integral via voiding. So adding the via void factor to the rework equation can only aggravate the condition, sometimes with disastrous results. Figure 6 is a picture of a void that was so volatile that it caused the BGA ball to "blow out". It blew out so forcefully that it deposited a loose piece of solder under the BGA.

Now that's-a spicy meatball!