We are able to lift a SOIC off a PCBA using tweezers and no heat. The pads are bare copper, the PCB was manufactured using tin lead HASL finish. The Assembly Brothers, Phil Zarrow and Jim Hall, 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
Phil Zarrow
With over 50 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
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 with Jim Hall and Phil Zarrow. What's our question of the day?
Jim
The question today comes from P.T. We recently needed to repair a defective PCBA. This required the removal of an SOIC. We were able to simply lift the SOIC off the PCBA using tweezers and light force and no heat. The pads on the PCB are bare copper, but the PCB was originally manufactured using tin lead HASL finish. And the SOIC was soldered with tin-lead solder. What do you think happened?
Well P.T., you've got a bad circuit board my friend. HASL (hot air solder level) coating has been one of the most robust finishes for many years, to the point where even with the problems with lack of flatness on the pads, people still hang onto it because it solders so well.
And the reason it solders so well is because during the process you actually solder the HASL finish to the copper therefore forming the copper intermetallic on the board.
So when you solder the HASL you're really not soldering (according to Howard Manko) because the intermetallic has already been formed. You are simply reflowing the solder. In fact it was the presence of HASL finishes, or the predominance of HASL finish, that coined the term reflow. Because people said you're not really soldering, you're just reflowing solder.
But the point being, if you're got a board where after reflow (or whatever technique you use to attach your SOIC) all of the tin lead lifts off of the pad exposing bare copper. You had a horrible glitch in your HASL process. Because what is HASL? You take a bare board, clean the copper, coat it with flux and dump it in a solder pot and then blow out the extra solder.
Phil
This is a process that's been around since the mid-'70's. Can we say "mature process?" It's pretty hard to screw up. But apparently your board fabricator probably screwed it up. At least that seems the most likely thing.
Jim
The only thing I can think of is a fingerprint. Before the HASL process, the first thing the fab shop will do is scrupulously clean the copper, as they do before they put on any surface finish. In fact, usually they micro-etch the surface to remove a thin layer of copper so that you get all the contaminants off. You have absolutely pristine copper to put on your finish, in this case HASL.
I'm guessing that somehow some contamination got on those pads for this SOIC. And of course it raises the question: you ought to check the rest of the boards.
Phil
Yeah and the rest of the boards of the production too. And maybe some of the stuff shipped?
Jim
One of the most common causes for poor soldering is contamination coming from fingerprints. Somebody puts a thumbprint on a specific board between the cleaning process and the fluxing process.
The flux doesn't have the capacity to clean off the oils. It goes through the process. For some reason the HASL coating stuck a little so that you didn't notice that it was bare copper. Or maybe it was bare copper and nobody picked it up. But the point is it didn't solder right because it was contaminated under the HASL.
Phil
I hope this wasn't a mission critical application. But regardless of whatever surface finish is or is not on your board, whatever you do -
Jim
Don't solder like my brother.
Phil
And don't solder like my brother.
Comments
This problem was quite common in the late 90s. Electronic manufacturers were pushed to drive cost of boards down, so PCB with HASL finish was the preferred option over the ENIG (Electroless Nickel Immersion Gold). The latter was a nice flat pad and ideal for SMD assembly, whereas HASL always had a bump on the trailing edge of a pad as it passed by the air knife. PCB manufacturers came under more and more pressure to try and achieve flat pads on the HASL boards. This was an extremely fine line between a flat pad with minimal solder on and blowing all the solder off as the panel exited the solder leaving only the Cu6Sn5 intermetallic layer.
The HASL board with only the intermetallic layer left on the pad was extremely flat, however it was impossible to make a reliable solder joint on it, either with reflow solder or wave solder processes. The result is solder joints that are not actually bonded to the PCB pad but only to the very small amounts of tin lead dotted around the flat intermetallic layer. This means any assembly subjected to vibration or shock e.g. dropping vehicle remote controls etc. will dislodge components.
If you look closely at the solder pads of a HASL PCB when they are very flat, you will notice a slight pinkish tinge on them. This is a clear indication that there is insufficient tin lead left on the pad for a reliable solder joint. This problem virtually disappeared when OSP (Organic Surface Preservative) processes gained popularity and it was a cost-effective alternative to HASL and cheaper than ENIG.
Les Watts, Testerion
HASL flux will easily remove fingerprints from either brushed or micro etched copper prior to HASL, these fluxes are based on Hydrobrominc Acid or Hydrochloric acid and have some pretty strong wetting agents incorporated. Most likely what has happened is the Air Knife has been set too powerfull and removed the HASL Solder coating down the the IMC layer and that then becomes unsolderable with anything even a decent soldering iron has a job to build the solder back onto the pad. The Paste will wet to the component body not the pad.
Gregory York, BLT Circuit Services Ltd
It could be any number of things, but if it's made overseas, look into the simple things first. What are they using to micro-etch the copper just prior to flux and HASL? Are they controlling the chemistry and the wait time prior to processing? It could be as simple as the operator went on a break right after micro-etch and let the boards oxidize for 30 minutes before they reflowed them. In my experience with off shore suppliers, they don't do a good job of controlling bath chemistry either. Maybe the PH was off in the nitric tank?
Bradley Fern, Entrust
Might be a plating. If the copper was plated, with copper, as is the general case to create the copper via's, then the plating bath may have been depleted of acid. This depletion will allow copper to be deposited but it does not properly bond with the substrate copper and forms a layer which is mechanically weak. Looks good, but quite defective.
The HASL process then bonds properly to the surface layer of copper but, as you note, the components can easily break free. Examine the component which broke free and see if copper is present on the HASL separation point. If so, then it is clear that the copper plating failed. In any event the boards are inherently defective and must be scrapped.
Jaye Waas, Renkus-Heinz
This is not necessarily a HASL issue. This could also be a soldering issue sometimes called secondary reflow. This frequently occurs when reflow soldered surface mount parts are passed over a wave for soldering through hole components. George Wegner performed investigated this problem back in the 90's. The crack initiates when the top side solder joints come close to the solder melting point. The strength of the solder is zero at the melting point and is close to zero as it approaches the melting point. A CTE mismatch between board and component puts enough strain on the intermetallic layer to initiate a crack. Once formed, the crack will propagate. A search of the Technet archives should find more discussion on this point.
Donald Vischulis, Woodward
I agree; this was some type of contamination and could very well have been a fingerprint. In fact, it most likely was.
As far as the comments about a supply chain issue, well, nearly all board fabricators have a defect slip by them from time to time, and this one was certainly in the "undetectable" category. I would not blame a supply-line manager because a single PWB had a single defect. Anyone with any experience performing non-conforming product disposition knows strange single-incidents happen sometimes. While HASL is usually quite reliable, ENIG is also a very good finish with its own design advantages (especially for very high-sped or lossy circuits), but only if plated by an experienced and diligent fabricator. ENIT is also very resistant to corrosion; I suspect someone meant immersion SILVER is susceptible to sulfur products, but not so ENIG. ENIG does require a longer time in reflow because nickel's rate of dissolution into molten solder is 10X longer than that of copper. Nobody said anything about BGAs, so I am not sure what that comment is about. Bare copper was seen per P.T., so not sure why someone would assume a "thick, brittle IMC layer" was the cause.
The IMC layer is a very self-limiting phenomena, primarily completely formed in the first reflow process. After that, there is very little further IMF growth, not even during subsequent soldering, rework, or baking operations.
So the only question still in my mind after all that is: Whatever led P.T. to attempt to pull the affected SOIC off with a tweezers in the first place?
Richard Stadem, General Dynamics Mission Systems
Every board manufacturer tries their best to ensure that after HASL the pad is as flat as possible for the SMT process.
The problem with this is the hot air knife used on the board as it is lifted out of the solder is set far to strong. The result is a flat pad, however they have actually blown all the solder off of the pad and you are basically left with the Cu6Sn5 intermetalic layer. This layer you can not solder onto and create any reliable bond. If you look at the board under a high magnification microscope you will notice that it has a copper sheen in the HASL finish. They might have provided you with a flat pad but they have also provided a finish that is not possible to create a good solder joint on.
From my experience the solder joint is basically adhered to the pad by flux and a tiny amount of fine solder points. This of course offers no mechanical strength or even electrical connection. I have also found that these joints and up as high impedance joints especially with time.
The lesser of the 2 evils is to have a bump on the one side of the pad and reflow that solder with the solder paste, rather than a flat pad that does not take solder.
If you have found 1 SOIC that comes off that easy, the entire batch of boards will be suspect.
Les Watts, Testerion
I concur with the original analysis, but I would first eliminate the following. Did you check to see if other components could be removed with the same method? Did you check the component to see if there was bare copper on the component pads. I would suspect a copper laminate to plated copper adhesion issue IE peeling copper as a possible cause.
Ed Uslar, TriTech, USA
Based on the fact that the user clearly states that the pads were "bare Copper" after lifting the SOIC off, I would have to agree that Jim and Phil are definitely correct in their answer. This is clearly a case of the PCB supplier not getting the HASL to Solder to the Copper in this particular area for whatever reason (most likely local contamination). But what might be the true "root cause" is a Supply Chain Management decision to use a some low cost, low experience and low quality PCB shop? Because if this was a good experienced high quality PCB shop, it's rather embarrassing for them, unless this was an extremely rare event!
The other topic related to this that has come up in this blog thread is the subject of "brittleness of IMC". In reality, the IMC formed with the HASL is the rather strong, non-brittle Tin/Copper type versus the truly brittle Tin/Nickel type IMC formed when using ENIG surface finish! I have first-hand experience with this difference on a PCBA with double-sided reflow and BGAs. This PCBA was subject to bending during installation, and when we were building the prototypes made with ENIG surface finish, 60% of the boards were having BGA joints fail due to the bending forces.
When I then recommended a switch to SnPb HASL (and actually vertical in this case) for production, the bending caused failures went to zero. In fact, I did a bending experiment comparing the PCBA finish types on this PCBA. With the ENIG finish the failures would start occurring at around 1mm of PCBA deflection (on a credit card sized PCBA). Testing of the HASL coated PCBA went to 5mm (in both directions on the same PCBA) without any failures (not that I recommend allowing this amount of bend in an actual product)! Of course this is an extreme level of bending, something like a HALT test approach, but it clearly showed the large improvement with HASL vs ENIG. The more brittle nature of Tin/Ni IMC compared to Tin/Cu IMC is a well know and established in the PCB/PCBA industries.
The incident my former co-worker Fritz describes is one of a rather high level of IMC growth in a finish of any type. A 4um / 158uinch IMC layer on a HASL finished PCBA is very high, but as Fritz indicated, this PCBA had been "over-exposed to multiple heat processes."
Typically a HASL PCB will arrive from the supplier with 10 to 20uinch (0.5um) of IMC, and as long as it is covered with at least a thin layer of non-IMC solder then it will be solderable. And IMC itself is in fact solderable, as long as it is not oxidized (which a solder layer above it prevents for quite some time, as the oxygen absorption rate is pretty slow). And of course the IMC is the most important part of the joint, as it is the actual critical "glue layer" that holds the entire joint together! So, IMC is not inherently "bad", it is rather actually "very good."
After a properly temperature controlled Double-sided SMT Reflow processing, and a Selective-wave Solder process (RoHS or non-RoHS), the normal and expected intermetallic growth should result in IMC thickness levels in the 30 to 60uinch (1.50um) range. Each process step will grow a bit more IMC. But this is also why each Reflow process should be optimized to only expose the PCBA to the minimum level of heat input needed to create all the joints and no more! Temp Moles should be utilized and one should definitely not just simply use one common SMT Reflow Profile, especially if it was developed for ENIG surface finish PCBs (which require more heat energy)!
This is one of the key items to realize, and actually a benefit for all the SMT parts being soldered to your PCB, the fact that the actual Solder-joint / IMC has already been created at the PCB shop during the HASL process! As noted, all the SMT process is actually doing is "reflowing the solder". This noticeably reduces the heat energy input needed when "PCBA level processing" a HASL coated PCB vs a ENIG surface finish PCB, since with ENIG you have to create the Tin/Ni Solder-joint / IMC during the PCBA level processing (and thus all the components are exposed to this extra heat).
So, you will get some IMC growth in the 1st Reflow process, and then you will get some more during the 2nd Reflow process. But if you have controlled your temps correctly, and you are using a HASL or Pb-HASL dedicated Reflow profile, there is no reason you can't achieve fully IPC Class 3 acceptable joints (even on the 2nd side) and with a highly reliable Tin/Cu IMC!
Depending on your thermal masses, pad sizes and paste volume, once in awhile you may see some small areas of "non-wetting/de-wetting" on the outer perimeters of the 2nd Reflow side SMT joint pads due to small areas of IMC finally coming up to the surface in the thinner HASL coating areas. But these are "fully acceptable", even to IPC Class 3, as long as the actual SMT joint meets the IPC requirements for wetting angle and % perimeter, which they seem to always do (my IPC Certified coworkers, our supplier IPC Master trainer, and myself have inspected thousands of such joints and they all were still better then Class 3 requirements).
The topic of these "small areas of 2nd reflow side non-wetted/de-wetted areas" has also been thoroughly and deeply investigated in the past (by some other former co-workers of Fritz and I), as the question was asked: "what if there are also areas of non-wetting/de-wetting under the joint that we can't visually inspect?"
Well, as I said, the investigation was very elaborate and included; hundreds of multiple cross-section slices, 5D-X-ray, and CSAM, thru many joints that exhibited such areas on their pad perimeters, by a large and highly experienced team of experts on PCB/PCBA subjects. The clear conclusion was that it was not a concern. There were no such areas found under the actual solder joints. And when you think about it there are some likely good logical reasons for why this the case, and why the actual solder joint under, and around the perimeter of the Component body/lead is formed before any IMC growth to the surface becomes an issue (body/lead thermal mass, shadowing, latent heat of vaporization of flux, solder paste wicking to the body/lead, etc.)
And once you have created an "IPC acceptable", and actual good solder joint with the HASL coating, even if you have some of these small visible areas of non-wetting/de-wetting on the perimeter of the pads on the 2nd Reflow side, these joints will still also perform acceptably in any subsequent needed rework process. This is because the Solder above the IMC in the actual joint area is now much thicker because of the addition of all the solder from the Solder paste. Rework will be reflowing this solder and any further additional solder paste applied. And in general, and many others have expressed this, reworking of a HASL joint is a much more reliable enterprise then doing so with an ENIG surface finish PCB (and fact some experts even consider ENIG a "non-reworkable" surface finish for their products). There is also the subject of the vastly superior nature of HASL and Pb-free HASL vs ENIG in medium to long term exposure to corrosive industrial G2/G3/GX environments (such as paper mills, tire plants, China, etc) but that is a detailed topic for another day.
The summary is that if you use a high quality experienced PCB shop, control such items as shipping and storage temp, use proper dedicated/optimized HASL/Pb-HASL reflow profiles and understand correct IPC inspection requirements, you shouldn't have problems with excessive IMC growth, and you also shouldn't have problems with areas of no HASL or poor adhesion of HASL due to pre-HASL contamination! As the old saying goes, nothing solders like solder! Although the use of non-RoHS HASL will continue to decline (and drop sharply in mid 2017), according to my most experienced large high quality PCB suppliers, the use of Pb-free HASL is already significant and continues to grow substantially (in particular the Nihon Superior SN100C which is a very good choice).
Steven R. McLaughlin, ABB, Switzerland
As a board manufacturer I was wondering if this board was made in the USA? For a BGA I find it rare to have a HASL Finish, because of the flatness needed.
Richard Kincaid, K & F ELECTRONICS, USA
Gents, I think you may have gotten this one wrong. I suspect that the culprit is a thick, brittle IMC layer. This can happen when the PWBs are baked for long periods after HASL. Heat cycles at the end user's site such as pre-assembly bakes, reflow, wave soldering, post-cleaning bakes, etc. can make it worse.
I once had a bad case of this on a particular PWB part number, so bad that just flexing the board would cause SOICs to fly off! The user can confirm (or reject) my hypothesis by doing the following:
1. If the land surface is dull gray and will not take solder from an iron immediately after removing one of the SOICs, the land surface is IMC.
2. If IMC on the land surface is confirmed, do a vertical cross section through a land, joint and lead. Measure the IMC thickness. It is probably at or above 4 micrometer.
