Question: We're having problems with some BGA devices after SMT assembly.

During testing the BGA devices are failing, however if we press lightly on the top of the problem BGA device during testing, the board passes.

What is likely to be the cause? Is it more likely to be solder joint failure, or something internal within the BGA package?

Expert Panel Responds: 

This sounds like a typical "head-in-pillow" (or "ball-in-socket") defect that is a common issue with BGAs. In this case, the joint is not really metallurgically formed, but the two metal surfaces (the solder and the BGA ball) are mechanically touching each other.

The amount of contact between the ball and the solder can vary with the temperature, or with pressure put onto the part (as you indicated in your question). This is almost certainly not an internal issue with the component itself; it is a soldering defect at the interface between the ball and the solder.

There have been a lot of studies conducted and technical papers written on prevention of head-in-pillow defects. One common cause is the component or board warping just enough that the ball loses contact with the solder paste just before the paste begins to reflow.

Another cause is that the reflow profile drives a Delta T across the component that can create some "tilting" of the component, again just enough to let the ball lose contact with the paste just as the board reaches the liquidus temperature.

This is often a complicated defect resolution, so I would suggest carefully studying the defect location(s) -- are the defects always occurring on the same parts, same locations on the same parts? Is it directional -- if you turn the board 180 degrees heading into the oven, does the amount of locations of the defects change?

I would also suggest to contact your solder paste supplier to see if they have any tips to help with this. If you contact me directly, I can provide a few papers to help get the process started for you.

You are most likely seeing solder balls underneath the device that did not reflow properly and therefore are not making electrical connection to their associated pad(s), although there could be other reasons causing electrical connection failure.

However, in my experience as a contract manufacturer who is exposed to all types of designs this one is the most probable. Applying pressure on the top of the device during operation is causing the improperly reflowed solder balls to forcibly make electrical connection. There are a couple of possible reasons behind the phenomenon that you're seeing but the top three are (and I would review them in this order):

1. The reflow profile for the board assembly is incorrect and there is not enough heat being applied to the BGA package to allow the solder balls to reflow properly (perform proper reflow profiling which is essential especially for large board assemblies);
2. Application of solder paste onto the pads is of insufficient volume to make contact with the solder balls during reflow (check your solder paste foil thickness and/or aperture design to ensure that sufficient paste is being applied);
3. Check the planarity of the board fabrication itself, and make certain that there is no warpage, bow and/or twist (IPC-610 has an equation and specification for this) which will prevent solder balls on (larger) BGA packages from establishing electrical connectivity during reflow (this problem is exacerbated when using high-temperature RoHS processing parameters with thin board fabrications (0.031" or less) built using low Tg materials which essentially causes to the glass to melt and then reform again introducing bow & twist during the reflow process).

There are other factors that can contribute to this issue that are easily overlooked such as contaminated or improperly processed board fabrication plating, not storing or handling plastic-encapsulated high MSL-rated BGA components properly which could cause intermittent package failure after reflow through outgassing, etc. but the three main issues that I mentioned above would be the most appropriate starting point to help resolve your issue.

Most likely cause is a Head on Pillow (HoP) joint failure under the BGA, where the solder paste and the BGA solder ball are not merged, or in contact, following reflow. Downward pressure on the BGA makes an electrical contact on the joint, which releases as the finger pressure is removed and the failure re-establishes.

An x-ray view at an oblique angle should confirm this non-destructively as the ball and the paste will appear as distinct separate entities compared with a single entity that would be a correctly re-flowed ball. An example image is attached - note the variation in the ball shapes - indicating an uneven response to the reflow and the Head on Pillow joints having the separate 'shadow' like structure with respect to the solder ball.

If x-ray inspection shows that all of the BGA joints are consistent then this failure is not due to Head on Pillow joints and it could be caused by warpage of the device cracking and lifting the device away from the board. Alternatively, it could be through a failure within the BGA package itself where an internal joint has failed and pressure applied, again, re-makes the joint.

If I would create a list related to this it would include (note: in random order):

• Oven profile related: head in pillow, improper reflow profile - too hot, too cold, is the profile following the solder paste manufacturer's recommendations.
• Micro cracks in the solder joints
• Other soldering issues - what is the processing time - is there enough time to allow the flux solvents to evaporate, causing an improper wetting?
• Assembly line - is there any vibrations on your reflow oven rails? What about your wave - other than mechanical stress, is there enough heat to reflow the BGA on the top side (if the board is mixed technology and goes through a wave or selective soldering)
• What is the cleaning process? If you perform a wash cycle, how much time do you allow the assembly to cool down after reflow before cleaning? Do you measure the temperature of the assmebly before cleaning?
• Insufficient solder - stencil design,paste type, paste deposition related issues: accuracy, clogged apertures, cleaning
• Component issues: solder ball separating from the body of the component, MSD stored improperly, contamination (finger prints???)
• PCB issues: board flexing, separation in between layers, pad related issue (finish), bad pad design, contamination (finger prints???, Are there via holes around the pads that the solder can wick into. How accurate is the solder resist applied around the pads?
• Last but not least, I wouldn't ever rule out the human factor - there is always a failure that I call MOE (Major Operator Error)

I hope this helps.