Is No-Clean the Trend for QFN Components?

Phil
And welcome to Board Talk with Phil Zarrow and Jim Hall, The Assembly Brothers. Today we're coming to you from the ITM Consulting Failure Analysis Cave, high atop Mt. Rialto.

We're here to talk about electronic assembly, materials, equipment, components, practices, procedures among other things. Jim, what's today's question?

Jim
The question. I recently switched businesses from simple single boards to complex double sided boards which are water cleaned.

I had previously been using no-clean processes at all our contract assemblers. That is what I have training and experience on with resolving issues.

Since moving to this new job I have been pushing to reduce cost by moving and testing no-clean processes at our current contract assemblers. I recently had an issue of residue left under a QFN package, that's a form of BTC bottom terminated components, after cleaning.

The resolution was a series of design experiments at the contract assembler to determine the best line speed, water temperature, pressure to run the boards and which saponifier to use.

My question, is there a continuing trend to move to no-clean or is industry in agreement that certain technology will have to remain in a cleaning process?

Phil
I guess there's two answers to your question, the way you pose it anonymous. No, the true answer is there really is no trend per se, obviously people with certain commercial interests might indicate there is, but it's like everything in our industry.

It's application driven. In this case, the decision whether to go to no-clean or clean is depending on the actual residue that you have there and the service environment that the application is going to see.

 Do you agree with that my brother?

Jim
Obviously everybody would love to go to no-clean if they could. It's cheaper, you eliminate the process and so forth. Yes, for high rel applications, long service lives, people have concerns.

And with certain technologies, in this case we're talking bottom terminated components, QFNs and other related packages, there are serious issues.

As we've spoken of before, there are questions even about the safety and long-term reliability of no-cleans with the BTC package. The idea that flux residues and flux chemistry gets trapped under the package, because a low stand-off and high volume of solder paste and flux in the central pad, can you guarantee that you fully deactivate that flux under a BGA package?

Even if you put the whole package through the appropriate temperature cycle during reflow. Yes I do agree. It's application specific.

If you decide you have to clean the bottom terminated components, what you're seeing in the experience with your CEMs is absolutely correct, it is not trivial.

Even with a good water cleaning process and engineered cleaning agents, to get everything right to ensure that you actually clean all of the residue that's supposed to be cleanable or a no-clean residue, simply because, mainly the geometry.

You've got low clearance and very little open space to get fluids to flow under there and flow back out again to carry the potential contaminants away. I think you're on the right path and you're dealing with the reality that exists in the market place and you're going to have to make your own decision of whether no clean is worth it with your specific product.

So I guess I agree with you, Phil.

Phil
This has been Board Talk with Phil and Jim.

And whatever you do ...

Jim
Don't solder like my bother. 

Phil
Don't solder like my brother. 

We work with smaller assemblies with small QFN and smaller SMT packages that make it hard for the traditional cleaning methods. Most applications I choose a NO-Clean paste/flux always. Or a local clean and air blow drying. Why introduce water/solvent and it gets trapped? So we've had very few issues with no-cleans if the assembler is diligent during the hand assembly process.

I love these questions…always anonymous (I think Phil makes them up to stimulate Jim) Let’s visit DMAIC again. And let’s also look at the Industry IPC Assembly Class guidelines. If you are assembling Winky-Blinkies (ClassI), No-Clean is your friend. If you are assembling A device that you only want to work for a year (Class II); No-Clean is your friend. If you are assembling something that peoples’ lives will lost if it fails (Class III), which it sounds like you are, then No-Clean IS NOT YOUR FRIEND. Since many components today are only available in QFN, or BGA formats cleaning processes must be approached outside the conventional process envelope.

The space between component and substrate or board surface is not the same during reflow as during cleaning and thanks to Tg differences between laminate and component body; that gap is virtually impossible to duplicate. The foreign particle is trapped and held firm, but it will also feed dendritic growth over time. If there are tightly coupled high impedance inputs or outputs on the device, they need to be very clean. A measure to prove the impedance is working well after assembly. A very in-depth selection process for the flux and cleaning solvent (b/c water molecules are way, way too big for this job) is critical.

A low surface tension solvent would better remove flux residues from beneath BTCs than an aqueous solution. Water alone has a surface tension of 72 dynes/cm; the surface tension is reduced in water by adding a surfactant but that still registers 35 dynes/cm; a typical solvent has a surface tension less than 20 dynes/cm, making it more likely to flush flux residue from low standoff components such as QFNs. Additionally, most aqueous cleaners do not have reduced surface tension in the rinse cycle (surfactant is added to the wash tank only) so the high surface tension of water means that no rinsing can occur under low standoff parts. The critical steps in a cleaning process are to wet, scrub, rinse and dry.

Modern no clean solder paste has been proven to be reliable without cleaning. The organic acid vapors are much more likely to escape, than a liquid cleaning material is able to penetrate the low standoff assembly, and escape. Using an engineered ratio of rosin and activators, there should be no concern for any residue left in a BTC or QFN assembly using a well designed solder paste.

Consider the liability of decision base on reduce. A one time evaluation does not catch lot to lot variation. Best effort should be to minimize, contain, and properly dispose of "things" that should not be there that others may have to deal with.

In my opinion and experience, utilizing no-clean chemistry for BTCs is becoming the defacto standard for a variety of reasons, one in particular is cleaning performance against the challenge of a low clearance QFN. In our processes, it is our default to use no-clean for any BTCs with clearance concerns. Fortunately, all of our current board designs support this default choice. While we do have assemblies up to 18 layers and very densely populated with QFNs, µBGAs, and other challenging BTCs, we have selected solder pastes that provide excellent results from all perspectives. I could not imagine the struggles that one would face being restricted to an aqueous chemistry AND a low clearance BTC. That combination is definitely a major challenge in the world of SMT.

On a related note, we do have a handful of assemblies that are processed in SMT with no-clean solder pastes, but are ran through a chemical wash to remove the residue, not for product requirements but for process control. The normally high impedance no-clean residue on bottom side QFNs and passives is broken down during the soldering of a very large quantity of thermal hungry TH components on very dense 18 layer boards. Due to the breakdown of the residue, its impedance decreases significantly, thus affecting the integrity of the signals being presented at each matched pair of QFN pins spaced at 0.4mm. Removing the no-clean residue prior to TH processing resolves this issue. We also must use titanium insert based fixtures for selective wave to minimize the heat stress on the SMT parts (as well as protect them from the solder of course). Conclusion: QFNs are your friend and your foe...

No matter if you are go true no-clean, water soluble, or you are cleaning that no-clean, the most important thing to remember is testing under the part to determine if any residues that might be present are still conductive and/or corrosive. Ion Chromatography is the best first step for optimizing any of those processes, HHH or THB testing is the best next step on actual product.

This is the perfect application for a vapor defluxing process and the reason why vapor defluxing has become so popular in the past few years. Vapor can go places that water cannot and the solvents used in vapor defluxing are much stronger than a water based solution. More info on vapordefluxing can be found at vapordegreasing.info