Ask the Experts
February 13, 2018
BGA Joint Voids - Accept or Reject?
Per IPC-610 acceptability standards, would you accept or reject this BGA solder joint? Is this level of voids something we should try hard to reduce?
I have heard that some level of voids within a BGA solder joint can actually improve reliability. Is this true?
Expert Panel Responses
Too many voids at the interface. Single voids in thebulk of the solder play little role in time to failure. Interfacial voiding cancause significant reductions.
Dr. Craig D. Hillman
CEO & Managing Partner
Dr. Hillman's specialties include best practices in Design for Reliability, strategies for transitioning to Pb-free, supplier qualification, passive component technology and printed board failure mechanisms.
Reject the joint, thelevel of voids and where they are would look to be too high, I would x-ray acomponent that has not been reflowed to see if the balls are already failed.
Global Product Champion
Richard Boyle is a Global Product Champion at Henkel Electronics. He has over 25 years experience in the electronics assembly industry and is responsible for the global technical service of all of Henkel's solder materials.
Thislooks like excessive voiding. However, without a tool (like X-Ray voiding %check) that will give you a quantitative result, this is just an impression.
Definitely something that you want to minimize. I do not know aboutincreased reliability but this is obvious sign that the outgassing of thesolder in the reflow process is not completed - there is a lot still trapped inthe material. It is not uncommon to see voiding especially with lead-freesolder that has more outgassing. Check your oven profile - usually a longersoak and a lower reflow temperature provides better results from the voidselimination point of view.
Engineering and Operations Management
Georgian Simion is an independent consultant with 20+ years in electronics manufacturing engineering and operations.
Contact me at email@example.com.
Let'sfirst address the question of IPC-610 acceptability. The acceptabilitycriterion for voiding in IPC-610 is currently <25% void area on transmissionx-ray. This image is a cross section, which is not directly comparable to anx-ray, because only the voids at the plane of section are visible.
When viewinga transmission x-ray image, all voids, regardless of where they lie in thejoint, are visible. This image contains 16.7% void area at the plane ofsection, as determined using a "threshold" image transformation in AdobePhotoshop. That's high. Based on this, I'd expect an x-ray image of the jointto show at least 25% voiding. If (and I emphasize if) that is so, the jointwould be a reject per IPC-610 or J-STD-001.
Now, let's look at the practical side. The voids in this joint are notlocalized at the component or PWB interface, but are evenly distributedthroughout the joint. Current research seems to indicate that in thissituation, the voiding is unlikely to have a negative effect on reliability.
Itis, however, a process indicator. Were this joint a product of my manufacturingprocess, I would be investigating the root cause(s) and optimizing the processto reduce the occurrence of the voids. For some additional reading on thematter, look at thisarticle published by IPC, and consider reading the referenced paper.
Fritz's career in electronics manufacturing has included diverse engineering roles including PWB fabrication, thick film print & fire, SMT and wave/selective solder process engineering, and electronics materials development and marketing. Fritz's educational background is in mechanical engineering with an emphasis on materials science. Design of Experiments (DoE) techniques have been an area of independent study. Fritz has published over a dozen papers at various industry conferences.
If this was mine I would reject as itdoesn't appear to have reflowed correctly nor does it appear to have pasteprinted on the pad at all. So my guess looking at the picture is
Hope it helps, I would be reflowing on apopulated PCB at 245C, if you have a bare PCB then typically set up the profileto 255C to allow for component losses and density, so this should give you asafety window. Lastly use T4 powder for dense prints = less voids.
- Not hot enough
- No additional Paste printed on the padto start with to aid wetting and reliable solder joint
Technical Sales Manager
BLT Circuit Services Ltd
Greg York has over thirty two years of service in Electronics industry. York has installed over 600 Lead Free Lines in Europe with Solder and flux systems as well as Technical Support on SMT lines and trouble shooting.
Here's my thoughts:
- That's the rattiest looking BGA joint I've ever seen, and I'veseen X-sections of hundreds of BGAs as-received as well asas-assembled, of all types, from most suppliers, assemblers, and re-ballers.
- Something is seriously not right!
- I would "reject the joint", but that's not the whole story. I'dput a lot of effort upstream.
- That looks like a non-melting ball, from its shape (perfect sphere).
- Did it look like that as-received from the packager, before attach tothe circuit board? I suspect it did. If it didn't,if the attach-reflow caused these voids inside the ball, then maybe it' a melting ball, retainingits shape, and I'm baffled.
- I'm guessing there's not much a board assembler can do to, or for, anon-melting ball.
- I note that the top fillet, to the package land, is pretty skimpy. I'd tryto get the BGA supplier to fatten that up, a bit. There'splenty of land real estate available.
- The fillet to the board land is also skimpy, plus it's seriouslymis-registered. There is so little fillet, I cannot determinewhether there are voids in the fillet. That's where attach-reflow conditions can help or hurt.Plus I'd make the PWB lands a bit bigger.
- Bottom line: this swiss-cheese situation is NOT mil-aero quality,by a long shot. It might be OK for a Fisher-Pricetalking mouse toy, but my judgment says NO GOOD for serious harsh-environment reliability.
- Another bottom line: if one grab sample shows something horrible,there's a very good statistical chance that othersamples, for the same population, will be much worse. Justsayin'.
- I'd also say don't count on the concept that voids can be a goodthing. I've seen no data nor analyses, but thehall-way consensus is that goodness might happen, a bit, but in only specific geometries andonly responding to CTE-mismatch shear fatigue, plus voids don't behave nicely to makethat happen .
Lockheed Martin Space Systems
Mike Green is co-chairman of the IPC Terms and Definitions Committee. He has been working with board design and manufacturing for 33 years.
Of all the voids I've looked at, none have shown this type ofcharacteristics. I would ask how this solder ball was made and conduct anevaluation of the solderability of the material to verify the goodness of thepaste and the flux.
Secondly, since thisdoes not look like anything we've ever looked at, it would create somequestions in my mind as to its goodness and reliability. Therefore since thisis unknown, I would reject this condition, from the perspective that you don'thave a homogeneous solder joints, and the joint interface is loaded with planarvoids.
Vice President, Technical Director
At EPTAC Corporation, Mr. Lambert oversees content of course offerings, IPC Certification programs and provides customers with expert consultation in electronics manufacturing, including RoHS/WEEE and lead free issues. Leo is also the IPC General Chairman for the Assembly/Joining Process Committee.
This is a difficult call on a generalization of voids,but I would remind you this is a thin section of the bigger ball and the amountof voids that can be seen on the ball to pad interface would concern me andsuggest that the boards with these voids be temperature cycled to assess thejoint integrity and determine if this amount of voiding has a negative orpositive affect on the balls.
President/Senior Technical Consultant
Mr. Munson, President and Founder of Foresite, has extensive electronics industry experience applying Ion Chromatography analytical techniques to a wide spectrum of manufacturing applications.
What you have heard about voidsincreasing mechanical reliability in BGA solder joints is not necessarilyuntrue.
I am familiar with research conducted byGlen Dody of Motorola in the 90's that found voids in the bulk solder wouldtend to terminate cracks that were propagating.
In the section you have shown I would be concernedabout the amount of voiding at the inter-metallic compound (IMC) interface.While voids in the bulk solder might stop a crack, the IMC is the weakestmechanical location in the solder joint. Voids along this interface will tend to promotepropagation as a separation occurs at the IMC. I would work to reduce voidingin your process.
Process Sciences, Inc.
Stephen Schoppe is President of Process Sciences, Inc., and has 19 years experience providing SMT services to electronics manufacturers. Stephen provides consulting to several Fortune 500 clients on solder and SMT processes, and is a frequent guest speaker at SMT industry events.
ThisBGA definitely has too much voiding and needs to be corrected. Wheneverthis much voiding forms in a single BGA, it is an indication of a poor solderjoint formation. This could be due to either weak flux activity or excessivesolder powder oxidation. Mixing two different solder powders also causes thiskind of phenomenon.
Director New Product Development
Metallic Resources, Inc
David Bao has more than fifteen years of experience in developing new solder paste, wave soldering fluxes and other SMT consumables. He currently serves as the Director of New Product Development at Metallic Resources Inc. He received a Ph.D. in Chemistry at Oklahoma State University.
Forspecific BGA void acceptability recommend to refer IPC 7095 and incorporatesame in Shop floor X-ray machine. This std has specific criteria based on sizeof void and numbers to Pass or Fail. As per picture it looks not meeting IPC7095 void criteria or having very low reliability BGA Balls.
Beyond that, can see the ball has not being melted sufficiently andcollapsed with a BGA placement shift.
Supplier Quality Leader
Subrat has 10 year of extensive experience in PCB assembly process optimizing for quality, process includes screen printing, wave, reflow. He has a copyright in stencil design published in Apex Expo2010 at Las Vegas US.
As Mike Green noted in his point 5, there seemsto be something seriously strange with the initial solder ball. Have youconsidered this might be a "counterfeit part?" And as noted by others, itappears there was a very low amount of solder paste available to make goodjoint fillets (both at the PCB & Component side). Along with the previouslynoted lack of registration and small PCB pad size.
I would also be concerned about such large "notch voids"open to the outer surface of the ball. These would be "stress concentrationpoints" under shear forces and could become crack initiation sites.
I am also concerned about the relatively high level ofvoids along the PCB side interface, especially if this was an ENIG surfacefinish with it's much weaker and brittle Tin/Nickel IMC interface vs aTin/Copper based IMC interface (when using HASL, Pb-HASL, ImSn, ImAg or OSP).
In regards to the question of some level of voids leadingto higher reliability, there is indeed some level of reasonable and potentiallyrelated scientific extrapolated truth to a possible answer. There have beentest results in the past on IGBT Chip to DBC solder joints that clearly showedthat such joints, with a evenly distributed pattern of small voids, hadnoticeably better power cycle life.
The reasonable theory being that the smallvoids allowed an increased level of, shall we say "flexibility", in the jointand absorbed some of the shear strain caused by the CTE mismatch of thematerials during the Power cycle induced temp cycling. Therefore, I believe itwould be very reasonable to postulate that an evenly distributed array of smallvoids in a BGA Solder ball would also result in greater flexibility of theball, and thus result in greater temp cycling reliability (in say an IPC-97012nd level interconnect test regime). Taller balls and solder columns have thesame positive result due to increased flexibility (as does the Z-axis thicknessof any solder joint).
The challenge/problem, of course, is trying to get suchevenly distributed and controlled small size voids in a Solder ball (andwithout getting to many voids at the interface), and preferably with none ofthe voids exposed on the outside surface. Basically a "closed cell foam solderball with a full skin". Something interesting to think about.
Steven McLaughlin, ABB, Switzerland
This solder joint appearsto have a wetting issue. Voiding issecondary to a wetted solder joint. Ifwetting has not occurred then voiding is the least of his problems.
Dean Edwards, APT Electronics, Inc., USA
Bob Willis, bobwillisonline.com, UK
I was just shown the void image which is a very small thumb nail and to me honest there is no way you can make a call on what the black blobs are? In any investigation consider each of the things that can and do create voids. I have provided a list based on my own failure analysis on void reduction on different terminations:
- Solder Paste volume
- Via holes in mounting pads
- Filled vias in mounting pads
- Filled and capped vi in pad
- PCB surface finish, outgassing
- Component plating outgassing
- Normal reaction from soldering process
- Non wetting or surface dewetting
- Stencil design
- Solder paste type
- Temperature profile
- Component footprint
The thing that stands out in the image to me isthe lack of the ball collapsing after reflow. If most or all of the balls underthe BGA have this appearance; and that's a big if; then it indicates the ballsdid not totally melt and we might deduce the balls had these voids on the part"as received".
I would look seriously at the supplier. If the BGA is high on one side and low on the other, andthis image was of a ball on the high side of the BGA, then this mess might havebeen caused during the reflow process from one of the many theories mentionedabove.
Jack Lucas, Ametek Programmable Power, USA