Should I glue BGA chips in the corners before soldering them with hot air?

Question

I have often see ball grid array (BGA) chips, mostly those from CPUs or GPUs, being glued around in the corners with some red glue or to the perimeter with a translucent one.

Having to manually solder BGA chips using hot air, should I glue the chips to the board before heating?

In their answers to a quite similar question about soldering small electronic parts (but not specifically BGA chips), some users mention that glue can cause additional problems when it is not applied properly: adhesive glue before soldering

Not having an assistant, such soldering process remains currently challenging for me, as I hold the hot air gun (from a rework station) in one hand and the tweezers in the other one.

Without using glue, I see at least three difficulties:

• positioning and aligning the chips precisely
• maintaining the surface of the chips parallel to thus of the PCB when bringing them to it
• remaining stable during the soldering, without false move nor trembling

So, my questions are:

1. Is the use of glue recommended, given the context?
2. Are there possibly alternative compounds to help keeping the chips into place during the soldering, like a kind of "butter" that would progressively melt when reaching high temperatures?
3. Are there other improvements that I can make to facilitate the soldering of the BGA chips?

Answer 1

To add on to the other excellent answers, and to answer your third question: the red glue you see is likely to be some kind of corner staking or underfilling. After soldering, an adhesive compound is added to mitigate in-the-field failure, particularly when packages are subjected to thermal or physical stresses. It's not intended to improve solderability.

See the ANSYS website for more info.

Answer 2

What you see around the corners is most probably not glue, and certainly not put there to hold the chip in place during automated assembly.

Some SMD components need to be glued down after the soldering, as in case of a PCB with components on both sides, when you flip it upside down to assemble the other side some components might fall when the solder is liquid. This happen in case the surface tension of the molten solder is not enough to hold them in place.

That said, it is possible that it is just a sealing compound that is put there to avoid moisture penetrating below the chip, especially for chips that might get hot, such as a CPU or GPU. Moisture can penetrate below the part, and diffuse inside the chip itself, depending on the package technology, and when the chip heats up the water can become steam, and crack certain parts of the chip. To avoid this, you bake the parts before soldering, you solder, and then shut the sides to avoid any moisture ingress.

How can you solder your BGAs is not a single-answer question. This entirely depends on what pitch we are dealing with, PCB thickness, PCB status (new/used) and equipment you have.

A good recipe is to use a very (very very very) thin amount of really thin (as in runny) flux on the PCB, place the part, hot air & pray. You might need a hot plate if the PCB is particularly thick/big, and if it is small enough you can get away with the hot plate only.

And absolutely, categorically, no glue.

Answer 3

Personally, I'm intimidated by the whole idea of BGA rework with hobby grade equipment, and really wouldn't do it.

But, no, if I were doing it, I would be very hesitant to glue the chip in place. Surface mount soldering relies on letting the surface tension of melted solder being able to pull the chip into alignment. Gluing would prevent that from happening.

The issue is that with hot air, you may have problems getting all the balls molten at the same time. I really wouldn't even consider personally trying it without using a board heater to bring the temp just below the eutectic point, and then using hot air to nudge the region of the board over melting.

Note that facilities that do BGA rework often have xray devices to check the results. Certainly, you would need to be able to tolerate errors, and be able to run a functionality test to verify correct placement.

Answer 4

The "red" glue you are seeing is an SMT red glue and it's a certain type of temperature-set adhesive. Normally most assembly houses will not be using these adhesives, as surface tension will position the components correctly. However that is the "theory" ... In practice and depending on the circumstance, it may be required or used sometimes. I know for sure that hand soldering small BGAs based on the common idea of "surface tension will center it naturally - don't worry" is pie in the sky ! If the hot-air gun blows too strongly, as they often do, even on minimum setting, you will quickly blow the part away, quicker then you can sneeze ! In this case, some sort of manual positioning on your prototype PCB might be the answer ... its not easy to say what type of glue works best (if you really had to go that way), but the temperature sensitive red glue is used because it hardends very quickly over 130 - 150 Celcius, so there's a reason for it being used in some cases ... Having said that, bigger BGA components, say with over 160 BGA balls you hopefully don't need any adhesive at all as the surface tension etc will do the job once the solder is melting ...

Answer 5

What others have said. We use a small under-board preheater (my technician says that is essential) so minimal airflow is needed on top using a hot air pencil. We just bought one of those inexpensive reflow ovens (under USD$500), which with some controller modifications can be quite good for doing small board runs. I saw one in use an an OSHWA meeting a couple years ago and was quite impressed. It has actual profiles and you can add thermocouples etc to do it right but on a small scale. For double sided boards surface tension should hold the upside-down parts on but you can of course use a slightly lower temp solder for the second side. Also you may find a local assembly house that can do small runs for you, and they should even have x-ray inspection for BGAs. There's such a place in Salt Lake City.

Answer 6

• 1. No, absolutely no glue should be used on BGAs prior to reflow. BGA solder balls collapse slightly during reflow, increasing contact with the pad, and any adhesive would interfere with that.
• 2. The solder paste holds the chip in place prior to the melt, then surface tension during. No need for anything else.
• 3. For backside components, limit the BGA size / weight depending on the solder paste contact area. If you can’t avoid this, you may need to do a 2-pass reflow with higher melt solder on back for pass 1, then lower-melt solder on top for pad 2. This adds cost of course.

You can consider using NSMD pads to increase solder-to-pad contact area (and thus, surface tension) during reflow. NSMD pads have proven more mechanically robust than SMD pads for very fine-pitch BGAs.

Finally, make sure your board planarity is well controlled. Also your package planarity needs to be specified and assured by the vendor.

The glue you're seeing isn't for soldering. It's 'corner staking' applied after the reflow process for improved shock-and-vibe mechanical robustness. It's a reduced type of underfill where adhesive is injected between the corners of the soldered BGA and the board, as opposed to complete underfill which is injected underneath the entire IC.

More here: https://www.ansys.com/blog/bga-and-qfn-failure-mitigation-underfilling-edge-bonds-and-corner-staking-physics-of-failure

Answer 7

The main reason for using staking or underfill is to 1) reduce the stress on the BGA solder joints caused by CTE differences between the package and the board, 2) reduce the possibility of the part detaching from the board during a high shock (depth charge near a submarine) or vibration (rocket launch) event and 3) in the case of underfill, provide a better thermal path from the package to the board.

As others have said, you want the part to be able to move a bit during the solder reflow operation, so any staking or underfill is done after soldering and after initial testing shows that the unit works properly.

Whether staking or underfill is needed, and the specific material used, can only be determined after a rigorous structural (for staking) and thermal (for underfill) analysis has been performed.

Staking is preferred to underfill because of the difficulty/impossibility of reworking a BGA that has been underfilled.

Edit 1

Note that on double side SMT boards that go through a solder reflow operation, it is usual practice to use a small bit of adhesive under components to hold them in place, especially those on the underside of the board. However this adhesive is not meant to serve a structural or thermal need.