To Clean or No Clean?
The addition of a cleaning process to your manufacturing line will amount to an added 5–6 cents per board produced. Mike Konrad of Aqueous Technologies believes that is a small price to pay for the added value, testing, and reduced liability and insurance that a cleaning solution can provide; however, he is the first to admit cleaning isn’t for everyone. I sat down with Mike to determine which potential customers could benefit most from cleaning and to discuss the recent change in tolerance, ECM, and the need for updated standards.
Barry Matties: Mike, why don't you begin by talking a little bit about Aqueous and what it does?
Mike Konrad: Aqueous is a manufacturer of cleaning equipment. A lot of people call our machines "de-fluxing" machines. We like to refer to them as cleaning machines because we remove more than flux. We have fully automated cleaning systems designed for post re-flow circuit assemblies. And we have ionic contamination testers, ROSE testers (resistivity of solvent extract). They are designed to test to see how clean boards are after they've been cleaned. It is an IPC and military requirement. Then we also have cleaning systems for stencils and misprinted assemblies. We clean up some mistakes and we clean up the tools that are used in the assembly process.
Matties: What are the most common problems that you see in the industry around cleaning?
Konrad: Pretty much the entire industry used to clean their assemblies after soldering or reflow. Up until the mid- to late-90s. Then, due to environmental regulations, specifically the Montreal Protocol, the solvents that were used to clean circuit assemblies were eliminated and no good drop-in-replacement solvent really took its place. That void fueled a new technology called "no clean" flux and a whole new species of flux and solder pastes evolved that were designed not to be cleaned and they were labeled, quite inappropriately, "no clean." More appropriately, they should be named "low residue.”
Shortly after the introduction of no cleans, many companies, given the choice of adapting a new solder paste or flux vs. going to a new cleaning process, chose the former and stopped cleaning altogether. The problem that we have today is the results of yesterday were actually quite successful. They left a little bit of residue on the board and everything worked well. What we're seeing today, due to a number of reasons that have been widely reported on, like miniaturization and the switch from leaded base alloys to lead-free alloys, combined with a higher reliance on quality standards, is an attitude to not just fix the symptom, but rather determine the cause and fix the problem.
ISO, TQM, Six-Sigma, and all of those quality acronyms really are forcing people to look at what the causes of problems are—and more and more assemblers are realizing the problem today is the amount of residue on the assembly may be no more than yesterday, but today it's just less tolerable. There's not a light that turns on when things become intolerable. It happens slowly and they see their quality yields drop. Manufacturers see more products being returned. In some cases they see more civil litigation because products are failing in the field and then they find out that if they spent 5 cents cleaning the product they would have saved millions of dollars in litigation. Those are the issues we're seeing today.
A lot of people are coming to us knowing they have a problem, but not knowing why they have a problem. They compare the results of five or 10 years ago and demand an answer to “Why isn't that still working today?”
Matties: Isn't the obvious answer or reason tighter tolerances, finer features?
Konrad: That's basically it, tighter tolerances and more harsh manufacturing environments, like with higher re-flow temperatures for example. So really the amount of residue tolerable to an assembly today is much less than it was just a few years ago. What manufacturers frequently fail to realize is when they stopped cleaning flux, they stopped cleaning everything. That's why we don't really like calling it a "de-fluxing" process because that definition is really too narrow. It's really a cleaning process. Even though the flux companies are accurate when they say, "We leave so little residue, it really is practically harmless." It may be true, and at face value let's assume it is true.
When boards were being cleaned as part of the normal assembly process, pre-1990s, not only were we removing flux, which back then was harmful if left on, we were removing etch residues, developer chemicals, silicone mold releases, fingerprints, HASL fluids, Dorito potato chip residues, and flux. While the flux may be less harmful, we were not cleaning anything else. The stowaways, board fab, component fab, and assembly residues, are allowed to be present on the board. When we have the stowaways, we then put flux on it which at best case is just a little bit of residue. The totality of all of those residues is creating a problem whereby the footprint between cathodes and anodes on an assembly is much smaller and the gap between components is much smaller than ever before. That produces kind of a perfect storm for electro-chemical migration.
Electro-chemical migration or ECM takes two popular and one less popular form. The two popular forms are dendritic growth, which is where dendritic growth occurs on a board and it catches fire or it shorts or otherwise fails, and it's quite dramatic. Less dramatic is electrical leakage, AKA parasitic leakage. That problem occurs only when certain conditions are present. When one of the conditions is removed it goes away for a while and then it comes back.
Then there's CAF, conductive anodic filament based failures, which is an ECM issue, but it's between the board layers and that can't be solved by cleaning externally. It can be solved by other things, mainly baking. ECM is presenting itself more frequently than it did before and many assemblers are in denial because they are going by the experience of a few years ago where everything was cool. Now all of a sudden things aren't cool anymore they don't know what changed. That's basically the amount of residue an assembly can tolerate.
Matties: Are you seeing more and more requirements from the OEM’s buyers to require testing on their jobs? It just seems like an easy answer.
Konrad: It's a funny thing. While OEMs will embrace cleaning. Its more common that OEMs come up with a cleaning standard once they use a contract manufacturer. An OEM may not have a cleaning process in-house and the moment they job out that particular assembly to a contract manufacturer they also provide a cleaning standard that the contract manufacturer needs to meet. Contract manufacturers have to clean to a variety of different standards using a variety of different processes using a variety of different chemicals as outlined by their customers, which presents a challenge for the equipment manufacturers. It's more common now to seek equipment that can run a variety of chemicals under a variety of conditions and clean to a variety of standards because of the wide ranging requirements of OEMs.
Matties: Is there a possibility of a universal standard that just applies?
Konrad: We have one, right now. We have a universal standard. It was originally in WS6536 developed out in China Lake, California out in the desert and then it kind of morphed into Mil Standard 2000A and then it made its way into some IPC specifications. That level of cleanliness is determined by a ROSE tester (resistivity of solvent extract tester), and it's expressed in micrograms of sodium equivalent per square inch and the current standard or acceptable industry limit is 10 micrograms of sodium per square inch. We have that already. The problem is that standard was developed in the 1970s.
Matties: It needs to be updated.
Konrad: It needs to be updated for sure, and IPC has been for many, many years trying to update it. I'm not speaking for IPC here, but the problem I can see that IPC would have, or any standards agency would have, is: How can I tell you how clean your board needs to be? If you're NASA or if you're United Airlines or if you're Boeing or Airbus, I'm going to tell you a really low number because I might fly on your airplane one day and I want to make sure it's reliable. Would I tell makers of electronic flea collars the same cleanliness value even though their cost of failure if far different than Boeing’s?
Matties: Is the cost of process dramatically different between the dog collar and a spaceship?
Konrad: No, the cost of cleaning a circuit board is about five to six cents a board.
Matties: So, why wouldn't the higher standard just apply?
Konrad: One would assume that would be the case, however right now we have Class 1, Class 2, Class 3 and Space standards. All different standards. The industry is set up to say, “If it's going in a personal computer it can have this type of solderability standard, and if it's going into an aircraft it has to have that type of solderability standard.” One could ask why we just don't we have one great standard for everything. Well, because consumer goods are made by the millions, and while I say six cents in assembly for cleaning is a really low price, if Apple were to tell Foxconn, "Clean every assembly it's only six cents apiece,” that would cost them millions of dollars annually. The economy of scale sometimes rules out value-added processes.
Matties: Will we stop buying a dog collar or iPhone because it costs six cents more?
Konrad: We wouldn't. There’s a story, probably a little bit of an old wives’ tale, of how United Airlines took the olives out of the Martinis several years ago and they saved several million dollars a year. Now, I would have gladly paid 15 cents extra to have an olive in my martini while I'm flying first class from here to Tokyo.
Matties: They just didn't ask you.
Konrad: Exactly. It's not what the customer is willing to accept, but it's the savings that the manufacturer can pocket. In consumer electronics, or any high-volume electronics, it's not a question of how much will the customer be willing to add on; it's a question of how much can the manufacturer save and invest in other areas of the company.
Matties: Now, when you say it's five cents, that's just the hard costs. There are all the soft costs of real estate, operators and cycle time. So the cost is obviously more than five cents.
Konrad: That's right. The five cents includes the cost of the electricity, the chemical, the water, the DI tanks and things like that. Someone has got to stand in front of a machine, someone has to dedicate 4 by five feet of footprint, someone has to amortize the machine over seven years and depreciate it and all that. That's certainly added on.
Matties: So in terms of true cost...
Konrad: That's a few more cents. Cleaning is not an expensive process.
Matties: But the cost of failure is so big. For a dog collar maybe it’s no big deal, your dog scratches a bit more.
Konrad: In fact, in one of our workshop presentations we show the cost of failure. We show the cost of failure of an implantable defibrillator and we show a picture of an ambulance. In our workshop, we show the cost of failure of an electronic flea collar and we show a dog scratching himself. Clearly the cost of failure is quite different.
The challenge the cleaning industry has is most of our customers are coming to us as first-time buyers of cleaners. A customer will approach us and they’ll have one of two engineer types. They’ll either have the old engineer, the old sage that's near retirement that remembers cleaning from the ‘70s and ‘80s. He's been told or she's been told it's time to buy a cleaner. The first thing they do is they look through one of your SMT Magazine’s resource guides and they look up DuPont, because they want to buy Freon TMS or they look up sources for trichloroethylene and they realize it's not sold anymore. They look up Baron Blakeslee and Unique Industries and other manufacturers of vapor degreasers and they find out that either they're not around anymore or they are doing other things now. They have knowledge of cleaning but it's all obsolete.
Alternative to that, we have engineers that approach us that say, "I don't understand why I'm even looking at a cleaner because the jar on this paste says ‘no clean.’” It might as well say to them "don't clean." It's an oxymoron to clean a "no clean" in their mind. Potential customers come to us with either old obsolete information or zero information.
Matties: And their own mindset.
Konrad: We have to educate our customers that cleaning may not be necessary. I wouldn't dare to say everyone needs to clean. I would like to be able to say that and have it be accurate, but it's not true. Not everyone needs to clean. The amount of products requiring cleaning today is greater than it was five years ago and I imagine it will be greater five years from now. It's like a shadow slowly enveloping a land, and if you're in that land and the shadow passes over you, you have to start cleaning. Not everyone is in the shadow. It has to do with the cost of failure. It has to do with the climatic environment that the assembly is going to be placed into. A board in a CISCO server in a server farm at Google or Amazon really doesn't have to be quite as clean as a board on the North Sea oil platforms.
Because the environment is much harsher and the environment plays a huge role in reacting with the residue to create ECM. We look at the climatic environment that the part is going to go in. We look at the cost of failure, airplane crash, defibrillator fails, itchy dog, etc. Then we also, when we get really geeky about it, start looking at the spacing between components. Because of the nature of the way we test for cleanliness, we look at the geographic diversity or differences between spaces of components on boards. That plays a role into how low of a number or high of a number we shoot for in terms of cleanliness. There are a lot of factors in there.
Matties: You just mentioned testing. Talk a little bit about testing and that process. When people use it, how they use it, and that sort of thing.
Konrad: The technology used today and for the last 30+ years for post re-flow circuit assemblies is ROSE testing (resistivity of solvent extract testing). There are four or five brands of testers on the market and they've been around forever. Thousands of them have sold. They express contamination in micrograms of sodium equivalent per square inch. Basically, how much salt would we have to add to the test solution to equal how dirty your boards are? We're not really measuring for salt, rather we're just measuring for how much salt we would have to add to this really sensitive solution to make it as dirty as your board made it dirty.
They're not perfect machines because they don't have the ability to test for every known contaminate. But their kind of like tracer bullets, that the military uses when firing bullets at night. They can't see every single bullet, but they don't need to. They need to see where the majority of them are going and then they can calibrate their aim. Same with these testers. They don't detect everything, but they detect enough stuff common in assembly residues that if they see something you know that there is probably more there. If they don't see anything, then there's really nothing else harmful left on the assembly.
The only thing that one has to know is that the pass-fail limit, which we still use today, of 10 micrograms of sodium per square inch, or 1.56 micrograms per centimeter square, was established many, many years ago before surface mount technology was even around. Although the pass-fail limit is 10, if I were testing boards and I received a 9.9 and my board was going in a medical device or on an airplane or something where failure is unacceptable, I would not go to sleep at night knowing it was a 9.9. because the nature of the test based on averages. So 9.9 could be 20 somewhere else on the board and zero on other places of the board and it's averaging a number over the entire surface area of an assembly.
I would drive that number down and I recommend in no case numbers greater than five. I recommend in most cases numbers lower than one. We like to see cleanliness values in the point range 0.1, 0.2, 0.3, etc. We don't like to see whole numbers. Because that mandates a person start determining the cost of failure and climatic environmental conditions. Rather than do that, just clean a little bit better. Drive a number so low that even if you round it up a little bit it doesn't really matter.
Matties: When people think of test and inspection, this probably isn't a test that usually comes to mind. Or is it?
Konrad: If you're a military contractor, or you're doing high-reliability jobs for flight, medical, etc., this is a common test.
Matties: It's required.
Konrad: Yeah. These testers have been sold by all four manufacturers in the thousands so they're very plentiful and they're very common. They're new to people who are new to cleaning. While not everyone is required to test, I would recommend people to test. The cleaners that we build and that other companies build all have testers built into them. They're not, however, the IPC approved ROSE test. Even if they say they are, there's not one cleaner on the market that has an approved ROSE tester built into it because IPC has a very strict standard of how ROSE testers operate. They have to operate with a very specific test solution. They have to have a very specific algorithm that they run. All of that is controlled through standards. The testing equipment built into cleaners doesn’t meet those standards, because that would actually make a rotten cleaner. It makes a good tester, but a rotten cleaner.
Not everyone needs to go to that extra effort and ROSE test their boards unless they're really concerned about cleanliness or if they're under a requirement to test. If you're under a requirement to test, pick your favorite tester and buy it and get on with it. If you're not under a requirement to test and the cost of failure is so high that you really want to cover your ass, it is a good idea to test. Because now you're meeting an accepted published industry standard.
Matties: Well, civil litigation is also driving down to the component level now.
Konrad: A couple years ago, I served as an expert witness in a civil litigation matter. It was a company that had a technology and was fueled by venture capitalists. They had this technology, but they didn't have the ability to make it so they used a contract manufacturer to build their product. Then they installed their product all throughout North America in streets, below the ground. Their product began to show signs of failure. Their product was run by a battery which was guaranteed to run for 10 years. After several months, the batteries were failing and they couldn't figure out why.
The owner of the technology sued the contract manufacturer and said, "You screwed this up and we're going to sue you for $30 million." I helped represent the contract manufacturer and other experts were representing the OEM. It turned out that this particular contract manufacturer was told that they had to clean the part but not to use any cleaning chemical, just water. Water's great when you want to remove water soluble substances, but even though they were soldering with a water soluble flux, they had other stowaway contaminants on the board which didn't really sign up to be in the water soluble army. None of those contaminant species were removed.
Then a part was added during the assembly process manually, hand soldered, and it was a very large part with a lot of mass to it. It used a lot of flux, soldered with “no clean", and did not clean the assembly. Then it gets better. They wanted to pot the boards in silicone so they could put them in the ground. Their thought was they didn't want any of the contaminant species that are in the ground, moisture and anything else, to get onto the board, which is good thinking. The problem was they sealed in all the sins of the manufacturing process. Nothing could get out.
What couldn't get out was moisture, and there was moisture between the layers of the board. They never baked prior to potting. What they ended up having was electrical leakage on the board which drained the battery. They had two units here, two units there, 12 units there, all over North America that had to be dug up on the streets and removed and replaced. All over an avoided cost of six cents per assembly. If you ask them, had they known what would happen, would they have spent six cents more and the answer is of course yes. Now they're cleaning, but only after this giant litigation matter. It does hit home. The decisions of manufacturing do come back to bite you or reward you. Either way.
Again, I'm not saying everyone needs to clean. Frequently we don't realize things even need to be cleaned until things start failing.
Matties: Yeah, you need to be mindful of these variables and influences that you're mentioning.
Konrad: Cleaning adds value and it never has a downside. Although I sound very self-serving when I say this, spend six cents. The worst that can happen is they waste six cents per assembly. Now that could be a huge number to some people. Generally, in the real world for us mere mortals that don't produce 100 million boards a day, it's not a big number, and it's a huge insurance policy. It's self-serving for me to say it, but it's true. That's why we're in this business. The cleaning process does add value and it reduces liability.
Matties: It does. Mike, I think this is really good. We certainly appreciate you sharing your expertise.
Konrad: My pleasure. Thanks for asking good questions.
Matties: Thank you.