What is Causing Intermittent BGA Failures?



What is Causing Intermittent BGA Failures?
We're having problems with BGA devices after SMT assembly. During testing, the BGA devices are failing, if we press lightly on top of the BGA device during testing, the board passes. The Assembly Brothers, Jim Hall and Phil Zarrow, offer their advice.
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
Phil Zarrow
Phil Zarrow
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.
Jim Hall
Jim Hall
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.

Transcript


Phil
Welcome to Board Talk. This is Phil Zarrow and Jim Hall of ITM fame, but also we're here for Board Talk. I believe today we have a solder defect question, Jim, and BGA related.

Jim
This comes from M.S.  "We're having problems with some BGA devices after SMT assembly. During testing, the BGA devices are failing. However, if we press lightly on top of the problem BGA device during testing, the board passes." 

"What is the likely cause? Is it likely to be a solder joint failure, or something internal within the BGA package?"

Phil
Yeah, I hate those flexible over-molded BGA's. Well, I hate to tell you, but it's your process my boy. And there's a number of different causes.

The first one that comes to my mind is inadequate solder paste height. So the question I have to ask you, "Are you measuring your solder paste height?"

If you are indeed measuring it, are you holding it to intelligent levels? Because that's another mistake we've seen often where people have very wide divergence on the low side. 

So, you definitely want to make sure you've got realistic levels and you're controlling it and you're not just recording it.  You're actually looking at the data, because that's going to tell you a lot right there.  So, that's the first thing that comes to mind.

Jim
One might tend to think about a head in pillow, but I would suspect not since you're getting an open circuit on initial test. Typically with a head in pillow, there's enough contact. Although they're not soldered together, they is usually enough physical contact after the board has cooled down to give you continuity and passing tests. 

That's one of the biggest problems with head in pillow, they seem to pass tests, but then they become intermittent later on. But I do suspect inadequate solder volume coupled with perhaps some amount of warpage of the package.

Look at your heating profile and your cooling profiles and your BGA packages.  Make sure you are fulfilling all of the BGA manufacturer's requirements in your reflow profiles.

Phil
I'm also going to add I hope you're not using a HASL finish on your board, because that can also add topographical variance which could lead to intermittents. 

Jim
Lack of coplanarity on the substrate itself.

Phil
Yeah, I don't care how good they say whatever alloy is for HASL, just when you start getting down to the finer pitch just wasn't meant to be for this. So, good, so we hope that helps and steers you off in the right direction.

On that, this is Phil Zarrow and Jim Hall.

Jim
The Assembly Brothers saying whatever you do, don't solder like my brother.

Phil
And don't solder like my brother.



Comments

If the BGAs have not been stored as per the manufacturer instructions and floor life of the device, moisture can ingress into the device. The subsequent thermal tests (thermal soak and thermal cycling) as part of ESS results in developing water vapour inside the device. In this, process de-lamination of die, lead frame, bond wires, heat sink paddle, the in-built PCB of the BGA with the moulding compound can occur resulting in failure. These failures are further accelerated by mechanical test like vibration and shock subsequently done as part of ESS.

My question is even with the stress being applied and ESS able to captured BGA's failures, we do experience the passing units which PASS ESS and shipped, failed at customer side (the same BGA's). Meaning the ESS gate is not a full proof gate? Any possible suggestion besides than enhancing the ESS profile (fine tune the vibration limits/ramp rate or adding number of ESS cycles)
A. Halim, Honeywell
Given below are the various phases in which a BGA device travels through, in a high reliability environment and the probable causes which can be attributed to the failures in those phases:

1. Electrical testing of BGA

Any observation with respect to intermittent failure, can be concluded as due to lower co-planarity of the device.

2. Post Environmental Stress Screening (ESS)

If the BGAs have not been stored as per the manufacturer instructions and floor life of the device, moisture can ingress into the device. The subsequent thermal tests (thermal soak and thermal cycling) as part of ESS results in developing water vapour inside the device. The water vapour then escapes through these cracks thus relieving the pressure. In this, process de-lamination of die, lead frame, bond wires, heat sink paddle, the in-built PCB of the BGA with the moulding compound can occur resulting in failure. These failures are further accelerated by mechanical test like vibration and shock subsequently done as part of ESS.

3. Post Reflow

This can occur due to:
(i) the loss of coplanarity of (a) BGA (2) the sites of PCB where the BGA is populated during excessive heating during Reflow.
(ii) Mismatch in the constituents of the BGA ball and the solder paste.
(iii) Quick cool down in the last phase of the Reflow profile resulting in stresses and mechanical fracturing due to CTE mismatch

Dr. KUTTIYIL THOMAS OOMMEN THARAKAN, VSSC
Although the issue is likely related to solder integrity, years back I came up against an issue with internal capacitance in the BGA package. If a person touched the package, the capacitance changed and the issue disappeared. Went through a lot of process changes before we realized it was the part, not the process.
Greg Stearns, Emerson
Ball co-planarity has been found to be a significant player in the formation of HIP defects, whether you use paste or tacky flux this can bite you. If you are using a 5 mil stencil and your BGA has 8 mil co-planarity, at some point you are going to get balls that never touch the paste, so the flux never touches the ball, and warping can make it worse. Same goes for tacky flux only. If the flux has been "used" up by the time the part collapses, you may not have HIP, but you will not have the formation of a good inter-metallic.

So you need to check the co-planarity spec of the BGA when you do your stencil design, you may have to Step up the BGA area
Alan Woodford, NEO TECH
BGA placement on the second side reflow may be subject to poorer board flatness. BGA's overlapping on both sides of an assembly can cause stress that can cause joint fracture.
Paul Chyc, BI Inc.
Just adding one more possibility is "BGA Warpage" Especially if it was a lead free BGA and then re-balled to eutectic solder. Some of the solder flow machines used to remove the Lead free balls is so hot that the parts warp during this process and soldering the re-balled parts can be impossible. Die & Pry can prove this out.
Gary R., BAE
I have discovered NOT always a BGA failure passes when it is pressed, is due to a bad process, be careful, BGA is a package with more than 28 years old released, many defects experienced head in pillow, defect passing thru lead free process, ALL of them has been controlled, to my experience one of the reason and we must not underestimate is the BGA package contamination.

I had the luck to face a contamination problem with the same exact symptom, so I strongly recommend double check the lot, and inspect in incoming inspection for the BGA conditions, at production floor will be very difficult to find out, the tool I used in 1998 at that time was 5DX.
Dario Cordova, Leader Technology Solutions
Check the lot code of the BGAs and see if the problem follows that lot code. If this is a flip-chip BGA it sounds like the under-fill underneath the die has delaminated. You can send fresh parts from that lot code out for CSAM which will detect delamination if problem is supplier issue. Also if cause by process, the delamination can be caused buy the soldering process being too hot on the component body, and/or soldering parts that have absorbed moisture.

This is more pronounced concern on larger BGAs. Be sure your parts are baked and you don't exceed package (not solder) temps.
Rick Smith, HP
In addition to the soldering quality of the solder ball for the BGA, there is another issue: delamination of solder ball, why? We received customer complaints about intermittent communication coming from BGA. We performed X-Ray analysis, but no issue related to soldering was observed. We performed a destructive test by micro section of the impacted Pins of the BGA and we found cracks generated by mechanical stress. Finally we found the real root causes and in order to confirm we performed a strain gauge measurement in different step of the process and also at the customer level. We found more than 450µm the limit of BGA delamination.
Drissi Charaf, Lear Corporation, Morroco
Comment to Bill Gavert: The J-STD-001ES does NOT require all gold be removed from ENIG PWB pads. ENIG as defined by IPC 4552 is only about 3 or 4 microinches thick and will not affect the solder joint reliability. The specification does not apply to the PWB plating.
Richard Stadem, General Dynamics AIS, USA
Another possibility is warpage caused by differences in coefficient of thermal expansion (CTE). One solution is to use constraining layer to tailor PCB CTE. In the past, we used copper-Invar-copper. These days, various types of composites are available, e.g. carbon fiber-reinforced polymers and metals.
Carl Zweben, Ph. D., Zweben Consulting, USA
This is very interesting. I am having a similar issue with units failing in the field after a variable time period (so far up to about 9 months). I've x-rayed (basic, and CT) the boards, down to the net in question, and can't find anything definitive from this.

So, what would be the next step in the diagnosis of this issue. At the moment, of a batch of about 225 units, 4 have failed with an intermittent connection under the same BGA device, although on different nets.

I am very keen to know what to do next.
Ed Wood, EW Consult, UK
We do not totally agree with the comment about solder volume being a "common" cause of intermittent BGA's. As Richard has pointed out, solder paste is NOT necessary to create a reliable BGA connection. Aside from reworking BGA's, it is completely common practice to place PoP components with a flux dip and no solder paste on the top component. As long as some paste (or flux) makes it to the pads, the ball should solder.

Furthermore diagnosing the root cause of an intermittent BGA can be extremely expensive, tricky, and time consuming business. There are a myriad of test that can be performed. Such as:
  1. X-Ray Inspection - Non-destructively examine for internal defects,
  2. Solderability Analysis - Determine solderability of bare boards,
  3. Contamination Analysis - Fourier Transform Infrared Spectroscopy (FTIR) and/or Scanning Electron Microscopy/Energy Dispersive X-Ray Spectroscopy (SEM/EDS),
  4. Dye-n-Pry Analysis,
  5. Microsection Analysis (Cross-section),
  6. Differential Scanning Calorimetry (DSC) - Determine Tg of laminate and / or component, Thermomechanical Analysis (TMA) - Determine CTE of laminate and / or component,
  7. Optical and/or Scanning Electron Microscopy (SEM) - Examination of internal structures ...etc. All of which can also be inconclusive.
This can be a tricky defect and in my experience I have never seen the printing, placing, or reflow process be the root cause. I would be more likely to suspect defective components, circuit board plating or some handling step that creates an internal stress.
David Molyneux, TT Electronics, USA
Back in 1972 at JPL there was a paper written: (Investigation of Gold Embrittlement in Connector Solder Joints By Frank Lane) The results (cut & pasted): "Since the maximum average gold content of all solder joints mapped by the electron microprobe was 2.8%, with one joint having a maximum average of only 3.9%, and there were no cracks in any joint after thermal shock (-100 to +130°C), vibration (up to 100 g peak) and mechanical shook (200 peak g), it is concluded that untinned and unwicked JPL connector solder joints do not contain gold in quantities sufficient to cause embrittlement."

This double tinning rule was written More than 40 years ago. I used to give in at the first mention of this issue because it was easier and less time consuming to just do it. layouts/footprints are getting tighter/smaller and double tinning makes no since. I think that the folks writing the Standards should re-visit this issue for current Assembly Techniques.
Jose Tamayo, Jose Tamayo Co., USA
One of the most common problems that we see is related to the use of lead free solder. If you are not using vapor phase or nitrogen reflow. When the lead free solder ball becomes liquid during the reflow process, air will come in contact with the connection between the solder ball and the BGA component. The pad on the BGA actually oxidizes because there is no flux protecting the solder joint. There is only enough flux in the solder paste to protect the bond from the PCB pad to the solder ball. If you remove the BGA, you will note that there are BGA pads that are a flat dull grey. Solder will not wet to them. This is what is causing a bad connection.
Dennis O'Donnell, Precision PCB Services, Inc., USA
We are fabricating boards with Fine Grid BGAs. We have as in the past called out ENIG. However, there is a catch 22.

This board is part of an experimental instrument for the International Space Station. We do not have the traditional NASA requirements. However, our QA wishing to follow the current NASA workmanship standards.

QA is requiring that per IPC J-STD-001ES and IPC J-STD-001E all gold must be removed from the board surface with a double tinning process.

This of course defeats the whole purpose of using ENIG to make the BGA surface flat.

Am I missing something here? I would be very grateful for your insight on this.
Bill Gavert, Fibertek Inc.
We also have had the "Finger Fix" effect. The BGA assembly in question has multiple package types; i.e. molded plastic, metal housing (CPU), and several ceramic based. Since then I have reduced the overall energy of the solder flow, while keeping within the packaging parameters and solder paste manufacturer recommendations. After the change, the PCB assembly has been working flawlessly. Our test engineers have also seen noticeable increases in performance.

Robert Meyn, Senior Systems Technologies
Intermittent opens can be due to five (5) most common causes:

1. Open internal via in main PWB. This is usually caused by initial reflow or rework without pre-baking the PWB to remove moisture. The high moisture content can lead to z-axis expansion of the PWB during reflow or rework (not always manifest as delamination), shearing the internal traces from the via barrel. This is especially true in the area of most BGA designs, with many vias. Pushing down on the BGA can cause a intermittent re-connect.

2. Same as number one above, except for the BGA substrate, which (usually) also has many vias.

3. Head-in-pillow, as was mentioned. However, solder paste is NOT NECESSARY for BGA soldering. Many experienced houses are seeing better yields by simply leaving the paste off of the pads and solder the BGA directly to the pads using the BGA solder balls only. Eliminating the paste gets rid of bridging due to misprints, head-in-pillow defects, and many other defects associated with the paste. It is NOT NEEDED in 95% percent of instances where BGAs are soldered, and is certainly not needed for rework. Just leave the apertures off of the stencil, and use a very small amount of a good tacky flux instead.

4. Pad cratering, leading to intermittent opens between the pads on the PWB and their traces.

5. Many BGAs are made with temperature sensing circuits that switch off the BGA if the temperature increases above a certain limit. If it is operating at around 100 deg. F, and you press your thumb on it, this can sometimes be enough to kick it off.
Richard Stadem, General Dynamics AIS, USA

Submit A Comment


Comments are reviewed prior to posting. You must include your full name to have your comments posted. We will not post your email address.

Your Name


Your Company
Your E-mail


Your Country
Your Comments