16
\$\begingroup\$

We have a board with two BGA chips. Due to the current shortage of many parts, one of the chips is next-to-impossible to find; except that we found a supplier with a good stock of the non-RoHS / leaded version of the chip (which is a discontinued product, but this supplier happens to have a pretty good stock of them).

Problem is: the other BGA chip is only available in lead-free balls.

I know that using leaded process for lead-free BGA chips is not acceptable, because the solder balls of the BGA will not melt and the joint will be unreliable.

My question: is it ok to solder leaded BGA chips with a lead-free (thus, higher temperature) process?

For context:

  • This is for a research project; the boards will be deployed in indoors environments (most likely places with air conditioning, etc.)
  • We do not need to worry about long-term reliability (the boards will be operating for less than a year).
  • Also, the boards are not part of a safety-critical or mission-critical system; sure, we want them to work reliably for the purpose of the quality of the research data; if a small fraction of the boards fail, it is still acceptable.
\$\endgroup\$
13
  • 3
    \$\begingroup\$ Among other things, you need to find out if the materials (like plastics) are of the same type as lead-free since lead-free requires better materials to withstand the higher temperatures. They could just be standard now though (or whenever the ICs were produced) rather than two separate supply chains. \$\endgroup\$
    – DKNguyen
    Commented Aug 11, 2021 at 15:26
  • 2
    \$\begingroup\$ Years ago I had a brief interaction with Prof. Pan at Cal Poly SLO who was doing work on mixing lead and lead-free solder...I can't find his publications list online, but he might be willing to talk to you for a few minutes or refer you to an appropriate paper. \$\endgroup\$
    – The Photon
    Commented Aug 11, 2021 at 15:54
  • 1
    \$\begingroup\$ Here's an old post where I cited a book chapter by Pan (unfortunately the link is dead). Also searching the term "Backward and Forward Compatibility" leads to some app notes at Maxim and TI that might be helpful. \$\endgroup\$
    – The Photon
    Commented Aug 11, 2021 at 15:59
  • 3
    \$\begingroup\$ Ask the assembly house that will be doing the soldering. They are the experts. Some have a strict no cross process policy, others will do it as long as they know. Or, they can place it by hand. I all depends on the assembly house chosen. \$\endgroup\$
    – Aaron
    Commented Aug 11, 2021 at 16:02
  • 4
    \$\begingroup\$ If these are prototypes never to be put on the market, then just go with leaded solder. Particularly if you will take them out of use after a year. If you wish to be very formal, mark them up as non-conforming to RoHS etc directives. \$\endgroup\$
    – Lundin
    Commented Aug 12, 2021 at 9:21

4 Answers 4

13
\$\begingroup\$

I would talk to your assembler and see if you could use two different temperature processes.

Do the Lead free profile first (that goes to a higher temperature) then after those parts are done solder the leaded parts on with a leaded temperature profile. This satisfies the requirements of both. (you don't need to worry about stenciling as BGA's have solder balls)

Another option would be to have the lead free BGA's installed with a IR rework station (which also support temperature profiles) which is more difficult but possible.

At the end of the day it would depend on who is doing the rework and what they are capable of and their experience with BGA parts.

\$\endgroup\$
6
  • 1
    \$\begingroup\$ How would you stencil lead solder with lead-free parts already on the board? A manual process with a mini stencil and a jig to get it in the right place? Sounds expensive and error prone. \$\endgroup\$ Commented Aug 11, 2021 at 16:11
  • \$\begingroup\$ @ScottSeidman -- at first, I understood the suggestion as: put all components, each with the corresponding paste; reflow with lead-free, and the leaded parts will have brittle joints; now reflow at leaded profile, so the brittle joints will melt and "reattach", this time correctly; but on second thought (after reading your comment), I see that my initial interpretation is most likely incorrect (I now see multiple issues with it). But yea, I also see the issues you point out. It would be ideal if we had CNC machines that "print" the solder paste on the board (as an alternative to stencils) \$\endgroup\$
    – Cal-linux
    Commented Aug 11, 2021 at 17:19
  • 2
    \$\begingroup\$ My bad, you can do bgas without paste! You would need flux, though. \$\endgroup\$ Commented Aug 11, 2021 at 18:20
  • 2
    \$\begingroup\$ @ScottSeidman That is what the balls are for, as stenciling would provide different amounts of solder for each pad and potentially make a bad connection (or intermittent with impedance). Flux is probably not as much as an issue, some BGA rework vids just brush on flux and try and keep the thickness even. \$\endgroup\$
    – Voltage Spike
    Commented Aug 11, 2021 at 19:44
  • 1
    \$\begingroup\$ Is this the case regardless of surface finish? For example, with ENIG finish (as opposed to HASL, which provides a thin layer of material equivalent to the solder paste), can BGAs be placed just on top, without applying solder paste, and will the amount of solder in the balls be enough? I've always been curious about this; as an example, KiCAD's library footprints for BGAs do have pads on the PASTE layer, which means that a stencil will provide the openings to apply paste on the BGA pads. \$\endgroup\$
    – Cal-linux
    Commented Aug 11, 2021 at 20:09
8
\$\begingroup\$

Be very careful with this...

Lead-free BGA's tend to be Sn-Ag-Cu and that copper content will react with the lead in SnPb solder causing very brittle joints.

http://thor.inemi.org/webdownload/newsroom/Presentations/APEX08/Papers/SAC_BGAs_in_SnPb.pdf

--EDIT--

Just to add some clarification to this, it is risky and it does cause cracks, but as the OP stated, it is for a short period of time. At the start of 2019 I ended in a similar situation and as long as you speak to the assembly house that there is a mix-tech process and they the reflow is at appropriate levels then it does work.

The only additional comment is I had Pb and Pb-Free BGA's (due to sourcing) and a test card then went through a RoHS type flow. Post reflow there were inspections performed and there were signs that the Pb balls started to collapse because obviously the temperature and duration was longer and thus tin-lead was more fluid

\$\endgroup\$
1
  • \$\begingroup\$ I wonder if that's what plagued early Playstation consoles (the common repair was to re-work the BGA.) \$\endgroup\$
    – rdtsc
    Commented Aug 11, 2021 at 17:59
2
\$\begingroup\$

It's possible to get BGA's reballed. When my last employer was switching over to RoHS, we had a batch of BGA's from our fab with balls incompatible with the process (can't remember which way round).

We sent them to a company with a laser reballing machine.

\$\endgroup\$
2
\$\begingroup\$

Component Prep is Required if Reliability is Important

A double-solder component prep removes the lead contamination.

If component prep is not done, the joint will fail prematurely, as this causes embrittlement. There is a metallurgical incompatibility.

If component prep is done, the issue is mitigated, and the assembly can be conforming and acceptable for class 2, and with process testing, class 3.

This process can be called double tinning or, for BGA parts it is less common but can be dubbed double reballing. Although you may consider some variations or hybrid approach as suggested in the comments.

  1. reball to dilute lead.
  2. reball for mounting

This is a prototype process, There is a target concentration level that is considered acceptable. As long as your process reasonably expects to hit this concentration that is what matters.

The process for cleaning lead and cleaning gold and the applicable standards are largely the same and may already be familiar to your contract manufacturer/assembler, so check with them.

Note: The lifetime reduction depends on several factors, if the product can tolerate higher failure rates or appropriate testing is done. You may opt to pursue a more basic mitigation, like basic reballing, or no mitigation altogether.

\$\endgroup\$
4
  • \$\begingroup\$ Double-reballing seems a bit excessive. Would it not be good enough to deball, tin with SAC, clean then reball? \$\endgroup\$
    – SiHa
    Commented Aug 12, 2021 at 16:06
  • \$\begingroup\$ @SiHa , potentially, depending on what the assembly house recommends. \$\endgroup\$
    – crasic
    Commented Aug 12, 2021 at 16:08
  • \$\begingroup\$ Note that if you have engineering resources to spend (e.g. this is a continuous occurrence or several 10KU), there are empirical ways to determine how much prep you can get away with \$\endgroup\$
    – crasic
    Commented Aug 12, 2021 at 16:13
  • \$\begingroup\$ Yeah, I suppose the assembly houses leadfree processes are a bit more mature now. When I was doing this, we were having to tell them what to do! The 'black pad' phenomenon was not widely understood, for example. \$\endgroup\$
    – SiHa
    Commented Aug 13, 2021 at 7:46

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.