BGA Straight crack

Sounds like the Infamous "Black Pad" Phenomenon to me. A search through the Archives should bring up lots of stuff.

DG

reply »

You are correct. The gold dissolves in the solder. The gold is meant to protect the nickel from oxidizing, so that you can solder to the nickel.

Obviously that didn�t work as planned. Your background information is very sparse. Two things come to mind: 1 Corroded nickel. 2 Black pad. [Search the fine SMTnet Archives for background, articles, and gnashing of teeth.] [An additional article on black pad that is not referenced in the Archives is: Ken Crouse's article in a recent PCFab Magazine http://www.pcfab.com/db_area/archive/2002/0202/cullen.html ]

Continuing with your �surely due to the abnormal stress� theory, what is the abnormal stress on these boards? What do think caused the solder to separate like this?

Here�s what you need to tell us: * What is the breadth and extent of this problem? * Are the defective pads solderable? * Are the pads on a bare board solderable? * What are the color and texture of the pads on the defect board? ... [pix would be nice] * What kind of solder mask is on the board and when, relative to applying the gold plating, was it applied? * What is the % level of P was in the EN tank? * What kind of gold is on the board? * What are the actual thickness of the nickel and the gold, before soldering?

reply »

This could be an electroless nickel / immersion gold specification. While the symmetry of the numbers is pleasing to the eye, but as Joyce Marchand says, "It doesn't blow-up my skirts." Problems with your specification are: * Lower end of the gold is too thin to adequately protect the nickel from corrosion. 0.03um [1 thou] is definitely not thick enough to start with and that leaves no room for fabricator process variation. [Guess the direction that normal variation heads.] * Higher end of the gold is too thick and could contribute to solder connection failure. * Open-ended nickel specification seems unwise. I want to say there is a problem that could be created by overly thick nickel, but I can�t support that thought.

Contrast your limits with the IPC-4552 ENIG specification: * Nickel thickness of 3 - 6 microns [120 - 240 microinches]. * Gold thickness of 0.075 - 0.125 microns [3 - 5 microinches].

ENIG is a bitter / tart [it has little to commend it to a "sweet" status] solderability protection. We are no longer specify ENIG on new boards and are replacing it on older boards. Consider that this may be an opportunity to assessing alternate finishes, rather than improving the specification of a risky material like ENIG.

reply »