Answer to Question 1
It is possible the root cause of your problem is insufficient paste.

Assuming you're running a one-to-one pad to aperture ratio, the "Area Ratio" (AR) for a 0.34mm aperture using a 0.005" thick stencil is 0.67, which is generally accepted as the lowest AR a robust process can handle. This means that all other aspects of the printing process must be in control to ensure a robust process. The addition of a Lead Free paste may further effect the release characteristic shifting your process to an unstable print.

I would suggest a post print inspection of the BGA is warranted to qualify the process. If, after optimizing the printing process, paste transfer instability is the issue, one solution may be the stencil type used. Laser cut stainless steel foils have a better release characteristic then those made using the Chemical Etch process. Nickel foils, either laser cut or E-form, have the best release quality.

Answert to Question 2
We faced a problem with lead-free soldering of a BGA (0.8mm pitch, 0.5mm ball size, BGA PCB pad = 0.34mm) -- it failed open intermittently (product PASS test at factory BUT failed when reach customer). We are using a SAC (Ag=3, C=0.5) lead-free formula.

1) We are following the recommended profile from manufacturer (Pre-heat 60-120s @150-190C; reflow 30-60s @220C). What could have gone wrong?

It sounds like "head in pillow." This is a situation where the solder paste doesn’t fully wet the BGA ball. Typically this is an issue with the component supplier and we have seen a lot of this issue with lead free BGA. The best recommendation is to change the chemistry of the paste to get a higher activity level to improve the wetting eliminating the “head in pillow”.

Answer to Question 1
Could be that there is a solder wetting issue or a coplanarity issue. Those seem to be the main issues.

Answer to Question 2
One method to rework BGA is to print solder paste on top of the Balls of a new BGA package and place on the board after removing the defective package.

Answer to Question 1
The primary root cause for this could be inadequate melting of the bump and the solder particles in the paste, causing intermittent contact and eventual failure in the field. Here are some recommended corrective actions ...

1. The 220 deg C for reflow is a bit lower than normal. I would recommend that the reflow temperature be increased to atleast a 230-240 deg. C range. Maintaining the time above liquidus between 30-60 secs. if possible is adequate.

2. If using a Ramp-Soak-Spike Profile then the preheat temperature should prabably go up to 150 and not 190, at a ramp rate of around 1-2 deg. C per second. After reaching the 150 preheat provide a soak with a slow climb to 190-200 deg C so as not to keep the flux temperature high for extended periods of time. This will burn the flux off depending on the chemistry and result in solder defects during reflow.

3. If it is a Ramp-to-Spike profile then increase the slope to around 1 deg C per second and make sure that the assembly is not soaked for too long a period above the 190 deg. C temperature before reflow.

Answer to Question 2
It will, but rememebr to reflux the bottomside of the BGA, as the flux from the paste is no longer available to protect the surfaces.

Some thoughts on this:

1) Its’ hard to come to solid conclusions on profile without seeing it. In other words, the profile may be “in spec” but at an upper or lower limit that is causing the failure. So it is recommended to provide the profile to the paste vendor or to us for further review.

2) Product can have an impact—if the board is a big one, your thermocouple can be measuring a surface temp but not picking up the colder “core temp” of the product. So the profile can look acceptable but not acceptable in real terms.

3) Thermocoupling is key! It is so hard to put t/c’s on BGA’s. You need to drill up under the BGA to mount the t/c at the joint. We can provide data and instructions on how to do this if desired-just let us know.

4) Coplanarity of the component is important. Sometimes the part can potato chip due to moisture so when it heats, the coplanarity changes. So you can have every aspect of the process set up perfectly and still see the defect.

5) Paste quantity can have an impact for sure. Too little can also be a factor.

Suggest gathering up the profile data, sample board with t/c’s attached, paste data and component sample with packaging and storage method identified and send it to us or any other trusted oven type or consultant like ITM or STI and this can be sorted out quickly.

As for re-work, that’s out of my power band so no meaningful comments on that one but I am sure others will weigh in. Of course, we would want to get it sorted out process-wise first so we can make the problem go away and not have to re-work! (But understand that re-work is needed for the time being)