We recently reworked a BGA component and the corners all bridged. Can these components be removed and reused if we remove the warp? Jim Hall and Phil Zarrow, the Assembly Brothers, reflect on this question and share their own experiences. Board Talk
Board Talk is presented by Phil Zarrow and Jim Hall of ITM Consulting.
Process Troubleshooting, Failure Analysis, Process Audits, Process Set-up CEM Selection/Qualification, SMT Training/Seminars, Legal Disputes
With over 35 years experience in PCB assembly, Phil is one of the leading experts in SMT process failure analysis.
He has vast experience in SMT equipment, materials and processes.
A Lean Six-Sigma Master Blackbelt, Jim has a wealth of knowledge in soldering, thermal technology, equipment and process basics.
He is a pioneer in the science of reflow.
Welcome to Board Talk. This is Phil Zarrow and Jim Hall, the Assembly Brothers, who by day, go as mild-mannered consultants for ITM Consulting; truth, justice and the American way. But we are here at Board Talk up in Mount Rialto in the Board Talk cave to talk to you today about problems to limit situations and process technique. And what do we have today?
Phil, there are common themes that show up and I'm going to read three independent questions that were sent on the same topic.
This comes from B.S.; hope that's no joke because it's certainly a legitimate question. We have been stumped by occasional opens at the corner balls of some BGA components after reflow. We have experimented with many minor adjustments, yet this occasional problem continues. Do you have any suggestions?
The next is from A.S., maybe a brother or a sister. We are having a problem with BGA components bridging. The problem is specific to one corner ball location on one BGA, but it's common to a significant number of circuit assemblies. What could be causing this consistent BGA bridging problem?
From D.W., We recently reworked a BGA component and the corners all bridged. This was a new BGA component that had never been installed on a circuit board before. Can these components be removed and reused if we remove the warp? The plot gets thicker. Would you consider the component to be scrap? What causes the corners of the BGA component to warp and how do we prevent it in the future?
Well, ladies and gentleman, this is a news flash: BGAs warp during reflow.
Yes, they do.
Virtually all do, certainly plastic ones, I mean, may be not ceramic; particularly the ones that have a complex internal structure and they are not thermally balanced. And as my brother, Phil, would want to say: it's a fact; don't fight it, live with it.
Whether you are -- and of course when you think about a BGA warping, the corners are going to have the largest displacement causing open and potentially bridges because of them either warping up and lifting above the circuit board or warping down and pushing into the solder paste therefore causing bridges.
The best technique to try to minimize warp is to slow down your heating and cooling processes. You ramp up during your preheat section and you cool down during your cooling section. Now of course, in cool down, you don't want to go too slow because you don't want to create a coarse-grained structure. So it's again, as in many things with electronic manufacturing, it's compromised, it's trade-off.
Now I am going to open up another moisture barrier bag for a can of worms. Moister sensitive devices, plastic and BGAs are big culprits. What level MSD is this component and are you adhering to the JEDEC standards on this, in terms of your exposure time, bake-out time. If you have a sloppy program, now is the time to get it together.
One experiment you could do with the particular components you know are bad is bake them immediately before reflow. And see if that reduces the problem.
The other things of course are getting a good solder paste that tolerates head-in pillow and other type defects that is specifically formulated to deal with these issues.
Make sure your printing, in as much as the aperture design, your stencils, if you are seeing open some people overprint the corner balls of their patterns, to get a little extra paste to minimize the chance of having it open
If you have seen bridges you might want to try minimizing the aperture; I haven't honestly had any experience with that. But BGA warping is a fact of life and you need to deal with it, and heating and cooling at the minimum acceptable rate is the most common strategy to deal with it.
But most of all fundamentally, before you start any experiments, make sure you have got an MSD program in place. If you don't, you are not alone and that's no excuse.
You have to have the MSD program going. Get with it. You know what is it doing, you are paying for it. And beyond that as far as paying for things -
Don't solder like my brother.
Don't pay my brother to solder for you either.
With new ASIC getting more complex (either bigger or more dense) the warpage can happen either sad face or smiling face plus sometimes the PCB design have certain level of localize warpage then the SMT Reflow result can be either bridging or HIP, non-wet, or open solder.
Shadow Moir Study the warpage of BGA/ASIC & PCB is recommended then we can find solution or mitigation e.g. offset with stencil design (to increase or to reduce solder paste volume in specific area underneath that BGA). We have experiences to develop mfg solutions to many ASIC Suppliers before they release the ASIC to market. This approach is become normal for current ASIC now.
Pongtip P, Celestica Thailand Ltd.
Another cause is known as non-wet opens. If solder paste reflows and wets the BGA ball, and the warpage occurs when the ball has been wetted by the paste, and solidifies before the warpage has relaxed NWO will occur.
There has been tremendous interest in using lower temperature alloys that reduce component warpage to create much higher yields.
I've found thermal warpage to be very rarely heating rate dependent. Warpage is mainly driven by CTE mismatch between the package materials, which is not a rate dependent phenomenon. Heating rate can affect temperature uniformity, so sometimes rate is indirectly correlated with warpage differences. With some packages extended exposure at higher temperatures can affect warpage, but this gets more into something breaking down inside the package. I have on my shorter lists of studies to perform a study on dynamic thermal warpage with dependency on heating rate and heating method to better understand the correlations.
Neil Hubble, Akrometrix
I agree with all the advices addressed in this article: thermal profile, msd controls, bga design, etc. But if you still have bridges in the corners... put 4 temporary small plates on the corners with a thickness of half diameter of the ball, and remove them after soldering. That will limit the deformation and will reduce/avoid shorts.
Chip Scale components which are simply a die with solder balls attached typically do not have heat spreaders, nor do ceramic BGAs (CCGA or CBGA). The heat spreader causes the top of the BGA package to expand at a faster rate than the bottom of the BGA, pulling the corners down. Moisture in the BGA may contribute to this, but actually the steam that occurs as a result of the moisture causes the component to want to spread in the middle, like a football lying on its side, and in that case the corners of the BGA curl up, not down. Steps you can take to fix these issues: Control the MSD aspects of the part, Pre-bake replacement BGAs, skip using solder paste for BGAs during rework; it is not needed and only contributes to bridging, higher labor cost for rework, extraneous solder fines, etc. Just use the solder volume that is already present in the solder balls that come with the package.
Richard Stadem. General Dynamics Mission Systems
If you are dealing with BGAs, you shall care about MSD. So, let's focus on warpage. The difference among different components in the BGA (silicon die, EMC, substrate, etc.) cause different thermal reaction expanding differently. So, depending on the component, you'll probably find head in pillow or opens in the corner and bridge in the center, or vice-versa. An efficient solution is work on the overprinting where you have HIP or opens, and reduction where you have bridge. This should be a different experiment for each product and component to reach the optmized result. Good luck!
TIAGO DA LUZ ORIVES, Teikon S.A.
BGA warping is real life. I have measured about 100 different BGA-type (but also some QFP's, QFN, LGA and DCB's). The corners are going up (typical, but about 20% of BGA have an alternative behavior). The main source of the warpage is the inner structure of the BGA and there a customer has no influence. Moisture influence exists (example of one BGA-Type about 40 Aum more coplanarity in comparison to a baked BGA).
Overprinting is a solution for edge-balls, but this possibility is limited (bridging still mentioned)! Also a change of the profile is limited! (Please consider, there are also some other components on the board and the recommendations of the solder paste suppliers should be respected). Important recommendation: Please measure the warpage during reflow before you do some changes!
Experiences of the last months. LGA are more critical, there is no ball foor a buffer! Important last remark: The most extremely measurement result I ever had was 400 Aum coplanarity change of one BGA over the full relow cycle!
Heinz Wohlrabe, TU Dresden, Germany
A major reason for warping is coefficient of thermal expansion (CTE) mismatch. Using low-CTE advanced thermal materials, it is possible to tailor CTE, reducing or eliminating this problem.
Carl Zweben, Zweben Consulting, USA
Matching the nozzle size to the BGA is important to minimize the warping of the part during rework. Using a too small nozzle requires all the heat to pass thru the part and into the solder joint. This can cause large temperature differentials and result in BGA warping. A properly sized nozzle - one that is approx 2mm all around larger than the part will allow heat to flow directly onto the PWB and reduce the intensity of the profile. A less intense profile (temp + flow rate) will help to prevent the temp differentials that can cause the BGA to bow down and create corner shorts.
Jerry Wiatrowski, General Dynamics C4 Systems, USA
Does the shape (square vs rectangle), size (large such as 1.25" sq. vs small (say 0.50" x 0.75") or material (such as all plastic or a die on a small PCB) affect the amount of corner or edge warping. At our company, we have the most trouble with memory devices which are plastic and are about 0.5" x 0.75". We rarely have problems with BGAs such as CPUs which are about 1.25" square and have a bonded die flip-chipped onto a small PCB.
We also think that RoHS solder makes for trouble in attaching BGAs. We know of a couple of companies that routinely remove RoHS solder balls and replace them with eutectic lead-tin solder. The tin plus a pinch of this and that metals don't melt uniformly and require a hotter reflow. Since not all component solders are equal for a given BOM, the assembler has to use best judgment to pick the reflow temperature and dwell and ramp-up and ramp-down times.
Bob Anslow, Dynatem, Inc., USA
Main cause behind warping of BGA components is excessive moisture contents in the package. Another reason can be some mismatch between the various components that make up a BGA Package (i.e. the substrate, molding compound and silicon chip).
Later one is the package manufacturing fault and these type of packages require special care while designing the reflow profile.
Whereas if the reason is excessive moisture, special care should be taken while baking.
Maninder Singh, Deltron (A Div. of CDIL), Chandigarh, India