BGA soldering quality issue

[roverider]

Hello,
I am trying to sort out quality issues in soldering 0,5mm grid BGA.  Im making inspection by Dye and pry test and I am seeing in some samples  suspicious golden tinted surfaces mainly on the  pryed off balls from pads of the PCBs, but sometimes on the pads on the pcbs. And also ridges and non uniform color on the surface of the pryed balls. Most time without the flown dye.  I think the golden tint is due to thin oxide layer on the solder balls (which could cause issues of reliability of the solder joint) but it could be also residues of gold (the pads are electrochemically gold plated) and also residues of the solder flux. Solder paste is Pb free paste Indium 10.1 or Indium8.9, both in N2 atmosphere.

[attach=2]
normal color solder joint
[attach=1]
godenn tint-suspicious oxidation
[attach=3]
golden tint and ridges - mainly on right bottom ball

I also made an experiment- i desodered the BGA and then I overheated it to induce oxidation and it seems that the golden tint is heat related.

[attach=4]
-freshly desoldered normally looking but with some faint golden tint area
[attach=5]
first overheating-golden tint with some blue- purple parts
[attach=6]
Second overheating-darker golden color.


Could you please tell me you opinion if the solder joint is oxidated or not and if the ridrges and non uniform color indicates poor quality or not and why?

Thanks.

[helius]

A solder bond is like a mechanical weld: the metal atoms from the solder and the pad dissolve into one another and become a single metal object. If you pry that weld apart, there will be traces of both the solder and the pad on both sides. If you are using ENIG plated boards, the pads have a gold flash. Indium-based solders do not eat that gold away, they dissolve only a fraction of it to create the weld. That means that at the base of the weld zone, there is still gold in a relatively pure state and yellow color. When pried apart, some of that yellow gold-alloy will be torn away and stay on the solder ball, depending on the precise depth at which the solder joint breaks. The ridges are the remains of the plastic deformation during the prying process, as stress builds up in each metal joint before it weakens and breaks. Both of these signs are expected and normal. The dye test is to identify cold joints or voids and you do not seem to have either of those issues.

[roverider]

Thank you,
I understand that there is residue of gold, but, when i left solder on hot soldering tip  for some time, it starts to have  golden  and then even to blue/purple  tint and there is definately not any gold, it is just thin layer of oxides reflecting/ absorbing parts of visible light depending on the oxide layer thickness.
I am confused , if the golden tint on the balls is just parts of the ENIG or or oxide layer like on the soldering tip or both.  And more  confusing for me is that the tint only is visible only in some joints  and also the tint is getting more visible and darker after overheating the balls on air (as showed my crude experiment).

[helius]

Surface oxidation can indeed create those colors. As can chromates, nitrides, sulfides, and other surface reactions. But the way you described your project, I don't see how these could form in an inert atmosphere. Do you see similar coloration on the balls of new components prior to reflow?

[roverider]

No, because I test only final product, because all the assembly is made in our supplier in China.   
There is some background.
By the DnP test, Xray and CT scan I discovered issues with voids (up to 80% pad area) and cracks which caused malfunction of our product. After some adjustments of the soldering process at supplier it gets better, reliability of our product gets better. We got some pre production samples  with good quality soldering and also good reliability, but large percentage of  production pieces fails, so we expect non consistency of quality with the soldering, mainly BGA- sometimes it is cracks sometimes voids and now the golden tint occured.  The tint is only on some  pads  and on somel samples. The inconsistency seems suspicious me.

I saw the golden to purple tint on some components such as pads on ceramic capacitors - some pieces from same batch have classic shiny silver  some got golden tint some even purple/blue.   But as i said it was on final assembly after reflow.

[asmi]

Thank you,
I understand that there is residue of gold, but, when i left solder on hot soldering tip  for some time, it starts to have  golden  and then even to blue/purple  tint and there is definately not any gold, it is just thin layer of oxides reflecting/ absorbing parts of visible light depending on the oxide layer thickness.
I am confused , if the golden tint on the balls is just parts of the ENIG or or oxide layer like on the soldering tip or both.  And more  confusing for me is that the tint only is visible only in some joints  and also the tint is getting more visible and darker after overheating the balls on air (as showed my crude experiment).

That "overheating" looks to me like burned leftovers of the flux. This is what I saw a lot of when I was tuning my soldering profile and repeatedly reflowing the same board over and over.

[SMTech]

I would focus more on what you can see from X-Ray. Could be an idea to get a sample of the BGAs in question (if you didn't supply them) maybe China has sourced a crappy batch or has been storing them badly.

If you are seeing significant voiding and cracking in X-rays I'd be looking at the whole process, printing, paste choice, paste handling, storage, profile but equally there comes a point when you are spending too much time and effort fixing a false economy. If your contractor can't diagnose and fix the issue on their own, you should take your business elsewhere.

[Matt-Brown]

Prying  bga off a board should either damage the bga or the pads on the board. The solder joint should not fail between board paste and ball.
I cannot see the whole board from the pictures, but depending what is around the bga for thermal mass (large smd parts), I am betting they need to increase soak time in the reflow