Why Cleaning Still Matters
Paco Solis, lead investigator at Foresite Inc., speaks with I-Connect007 about why cleaning is more important now, the common pitfalls and misconceptions in cleaning, and strategies to consider to ensure reliable PCB assemblies.
Stephen Las Marias: Paco, tell us more about Foresite and your role in the company.
Paco Solis: Foresite was started in 1992 in Kokomo, Indiana. Our president, Terry Munson, used to be an engineer at Delphi, originally Delco Systems. He spun off his own company as he saw a need for analysis and education in the PCB and PCBA industries. Primarily, we are a consulting and analysis laboratory. We are a consulting firm that looks into process contamination, failure analysis, and ion chromatography studies.
We do a number of things. We assess customers’ boards, assemblies, and related hardware for cleanliness and reliability, and we also educate our customers as far as how clean their process needs to be for “in-field service” and “long-term reliability”. Part of what we do is test PCBs and PCBAs for cleanliness. Other things we do is rescue cleaning, metallurgical studies, and material analysis studies for failures. We help customers through the process of failure analysis to root cause, remove that failure from the process, and teach them how to recover their product lines. In some ways we teach contamination risk assessment as a consulting firm. Once we have our customers back online, we also perform testing and perform laboratory studies for them to monitor their process and prevent the issue from returning.
Las Marias: As lead investigator, what's your main responsibility?
Solis: It can be anything from walking through a PCB manufacturer and looking for opportunities for improvement to cross-contamination and cleanliness control. It could be helping a customer optimize their wash process for PCBs or PCBAs. Or it could be consulting for what grade of wash water they're using. If I’m going to PCBA assembly house, I'm going to go look at the cleanliness of the PCB and the cleanliness of the discrete devices, I.C.s, and ASICs which go into process. This includes the cleanliness of the stenciling, reflow process (both SMT and wave soldering), and rework processes. We are looking at the many factors of bare-board cleanliness, device cleanliness, solder re-flow and end processes. Because, usually when a PCB fails in the field, its typically is not just one factor. It’s multiple factors. We specialize in understanding semiconductor packaging, PCB and PCB assembly processes. And each time we discover a new failure mechanism, we are adding to our knowledge base. This experience, together with the expertise, is what allows us to help our clients get through their issues even quicker, or prevent those pitfalls altogether, especially if they engage early with us.
Las Marias: What are the biggest challenges that your PCBA customers are facing?
Solis: It comes down to cleanliness education and what is important to pay attention to: inner layer cleanliness of PCBs, electrochemical migration issues, either internal on different inner layers of the PCB, or external layers as well. These materials can be left behind by the PCB manufacturer or introduced at any point in the assembly process. One of the biggest challenges is that the industry has been so pinched that profit margins are very narrow. The customers are demanding Class 2 and Class 3 quality, but they’re not willing to pay for it. These narrowing profit margins are forcing some suppliers to ride the quality levels on the edge of the processes. It is very difficult for a PCB manufacturer to still stay profitable and really get a clean product because their customers are not willing to pay for a clean product. When it’s about profit margin, it’s a very, very, tight rope.
Another issue we're finding is that as a customer orders PCBs, they aren’t communicating enough to their PCB manufacturer what they need. It has become a “build to print” industry. They'll give their contract manufacturer the PCB design and say, “Build to this print.” If the customer that is buying the PCB doesn’t know how to specify what his cleanliness needs are, they don’t know how to relay that information to their suppliers. If they did know, they could specify that on their print, and demand that from the supplier. So, there is an education gap between the end customers, intermediate, and all subsequent tier customers.
What we have learned from the PCB industry is “If I don't tell my PCB contract manufacturer how clean to make it, then it is up to the manufacturer to produce what they normally produce” (their standard preferred in-house process package). If the customer doesn't know how clean he needs his PCB to be, then he has done his supplier a disservice. Each tier in the process needs to communicate their cleanliness needs to the next tier. That is where we come in! We help to bridge the gap of knowledge and education. We help our customers by teaching them how clean their processes need to be so in turn, they can work with their contract manufacturer to develop a PCB that is clean enough for his product. It’s not so straightforward. The PCB industry has many different customers with different cleanliness needs based on their individual products. Here is something I say often to my clients: “There’s a difference between the cheapest thing I can buy, and the best I can afford.” You can apply this to every step to the industry. When you drive costs too close to the quality limits, quality will eventually suffer.
Our customers must work with their contract manufacturers at every tier. For example, discussing sensitive architectures or low standoff architectures in their application. We need more of a technology cooperation between customers and contract manufacturers to understand what cleanliness levels they need in order to produce reliable products for the End Consumer. It’s a break of education and communication right now because it is about profit. Usually our clients become true believers in how important this is after they have been bitten by an avoidable issue which cause all parties a great deal of financial pain. I like to call it “baptized by fire”. After this happens to a client they are more sensitized to future products and will usually work to use the lessons learned.
Las Marias: You said it well. There has to be communication between the designers, assemblers…
Solis: There must be! I was involved with one client where an uneducated decision from a manufacturer to their supplier resulted in an $18 million issue/recall. It was because of a lack of communication of what was needed versus picking a device which was a few cents cheaper. The application and materials in the application required a slightly more expensive device to withstand the end use environment reliably. The issue also could have been avoided if the customer understood their introduction factors and had spec’d lower contaminating materials when they affixed the PCBA into the final assembly.
Las Marias: In our earlier discussion with some solder manufacturers, they say the PCB assembly industry has some sort of misconception when it comes to no-clean. The customers of these contract manufacturers are expecting very sleek looking boards or PCBAs because they thought it’s no-clean. Can you speak to us more about that?
Solis: When the term no-clean came out, it was kind of a misnomer. There is more than one way you can interpret “no-clean.” If you commit to having a no-clean assembly process there are issues which must be understood and concepts you commit to. If you have a no-clean process, you're not cleaning the bare PCB, not cleaning your parts, not cleaning your ASICs, and not cleaning your contributions as well. So, you have to either know the contributors of all your previous tiers or you blindly accept the cleanliness levels of your suppliers. In short, if you decide not to clean your assembly, you have already committed to accepting all previous tier cleanliness issues.
Here is another consideration. The term “no-clean” has been given to the industry as a process without clear expectations and metrics to the final assembly. Much of this comes from flux suppliers who are trying to supply the industry with different solutions for solder reflow needs. The complication is they don’t often have knowledge of how clean the customer’s assembly and end customer’s needs are. The flux suppliers will put their perceived best chemistries forward based on the input from their flux designers and their assembly customers. So, who is supplying the cleanliness needs for the end use product? Sensitive architectures require cleaner processing levels. This may require a flux product with lower levels of activators to attain a reliable end product. This applies for SMT reflow processes the same as it applies to wave and selective solder processes. The knowledge on how to process these fluxes to achieve these higher cleanliness levels also needs to be considered. Again, we have a gap in communication and education. For us at Foresite, no-clean is not a process. No-clean is a decision! It is a decision not to clean my assembly based on empirical data which suggests my product is clean enough to operate reliably in the selected end use environment. So we call no-clean a decision, not a process. It really comes down to an issue of knowledge, on what sensitive architectures and circuits require.
It's not about how cheap I can do this. It's how well I can do it based on my budget. It isn't as simple as “no-clean”. The bigger question is why am I not cleaning? If my application needs are telling me cleanliness is not a first order concern then I can make an educated decision. It means I've decided I'm not going to clean my boards because I know my board supplier is clean. I know my part supplier is clean, and my ASIC supplier is clean, and I know that all my assembly processes are clean enough for my end use requirements. I make this decision on data. Not just a process but one with the proper data.
Las Marias: Right now, in mission critical applications, you just have to clean your board even though it's no-clean.
Solis: But to what level do we clean the boards? That's another question. How clean is clean? How clean is clean for a mission critical board? The level of cleanliness I need for that environment is going to be different for manned space flight, as it is for military missile operations, as it is for public transportation, as it is for communication devices. The level of cleanliness and reliability is different because the environments are different in which they operate in. You need to start at the end service environment. The more reliable it needs to be, the cleaner it needs to be, and we're not even talking about conformal coating considerations yet.
It's a more complex question, but it's been reduced down to “no-clean.” It is not that simple. The new engineering generation hears the word "no-clean" and they think, "Oh. Okay. I don't have to clean." "Why don't I need to clean?" is the question that needs to be answered. Why do I not need to clean? Why is it no-clean?
Las Marias: It depends on the manufacturer and what they want to do.
Solis: The manufacturer and the end service environment it needs to operate in.
Las Marias: So Foresite also does failure analysis. How does that process work?
Solis: It depends on the failures we’re addressing. If we're looking at a metallurgical issue, then it could be as simple as we’re looking at reflow profile from the SMT assembly, checking to see how much energy and time was used in the process. Did they have a proper ramp rate? Did they have a right activation energy and temperatures for the type of flux they are using? Did they reflow at the right temperature? Or did they over-temperature during reflow? What metal system are they using? What are the reliability needs of the assembly and did they properly process and reflow that solder for that part? Is it the most appropriate alloy and grain structure for the application? Does it have an inner metallic, a continuous and contiguous inner metallic? Is it a lead-free system? Does it have a large amount of copper precipitate or copper solubility into the metallurgy? It depends on the failure. We help them get rid of the factors in the process. That's just one. Metallurgy’s just one disipline.
If we’re working toward a chemistry failure analysis, say they didn’t use enough thermal energy in the reflow and they haven’t rendered their flux activators benign. In this case we help them through thermal dynamics studies to see where they’re missing the thermal energy requirements and what should they have done to the board to properly cross-link the flux residues so that they are benign.
Las Marias: In the SMT process, where is the defect more evident or where does it come from mostly?
Solis: From our experience, it’s process excursions. You have to really look at terms like “touch-up”, “rework”, and what do they mean? To some companies, they’ll add a solder heat alone and call that “touch-up”. Some companies when they add solder or flux they can call it “rework”. Some call this “touch-up” and only call it rework when they take off a device and re-solder in a new one. These rework processes are excursions from the standard engineering of the assembly process. If the company doesn't have the ability to own traceability of the rework and understand the amount of cross-contamination possibilities that are contributors in each step of the rework processes, I don’t advocate reworking without material review activities for that product (MRB).
If I’m adding a cored solder to a rework process, I need to make sure I know what type of flux is in the solder and how to activate the flux completely. If I’m adding liquid flux to that board, I need to make sure I completely activate that liquid flux and I didn’t allow active flux to cross-contaminate the adjacent components. If you haven't educated your rework crew or your production crew on what matters, they could be dry brushing solder balls with a dirty brush and relocating contamination from PCBA to PCBA (just as an example). A little bit of rework and a little bit of touch-up and they think it's value added. The process of taking a brush and brushing the backside of a printed circuit board, which may have flux contamination, and now that same brush goes to brush the next board, and I transfer the contamination from the first board to the second board. Anything that touches the board or goes into the board, you really have to consider what is the possibility for adding cross-contamination?
If I rework it, how am I changing the original design of the material and the chemistry? So, I consider all rework as a process excursion and risk to the product. Unless I understand what rework really is and its contribution to my product. From my perspective, that's why I get called to walk assembly lines. To see where the opportunities are for cross-contamination. Once it comes out of SMT, that’s when it really starts. When operators start touching the boards, they can start transferring contamination. Most engineering communities aren’t aware that rework causes cross-contamination.
Las Marias: What is your advice for the industry on this front? How do they avoid that?
Solis: Back to one of my first statements, when we approach our suppliers we need to get the best quality we can afford and not the cheapest thing we can find. The PCB itself is the second most integrated part of a PCB assembly besides the ICs. After ICs, the PCB is not a commodity like a resistor or capacitor. It is a complex assembly. It needs to be treated like a complex assembly. That is one of my largest recommendations. Buy the best that you can afford. It starts with the PCB. If you don't start with a clean PCB, then you're not going to end with a clean assembly.
Las Marias: Is the cleaning for PCB different then the way that you do it for PCBA?
Solis: That depends on the quality level you want to achieve. It comes down to “How are you washing?” Some PCB manufacturers use tap water in some of their processes and depending on what your architectures are (that may be fine for a toy that gets placed into a fast food meal). Once you get into higher reliability, you have to look at the quality of your water and the positive and negative effects of washing. Do you need to use chemistry to clean up that circuit board? What are the reliability requirements? That will drive how clean you need to be. If you have microvias and through-hole vias and you are trying to use cold DI water, you’re not going to get them clean. If you use hot DI water, you might get them clean. If you use hot DI water with chemistry, you'll lower the surface tension and clean all this architecture that you couldn't with DI water alone. But now that will add cost to your assembly. It's really, what is the end use environment? What is the end quality needed? You’ve got to work backwards.
Las Marias: Paco, do you have any final comments?
Solis: When it comes to the PCB assembly process, the different departments of organizations need to work together. Your quality assurance, your procurement department, your supplier quality department, they all have to work together to understand cleanliness. Every part that goes into that PCB assembly is going to affect its final cleanliness and its reliability in the field. It’s not just the job of one department much less one engineer. Reliability engineering must work with quality assurance, with procurement, with subcons, with suppliers, and all with a good knowledge of all disciplines which go into the assembly. But, they have to be educated on again what the end use requirement are. The breakdown of the process creating a reject could come down to something as simple as a dirty glove. A dirty glove or a chemical that gets substituted because it’s believed to be cheaper and equal (as an example). Replacing a material in the process without really looking at what effect chemically I am doing to my assembly can be a very large expensive mistake. A small change in a process, a different foam, a different sealing agent, a different adhesive tape, a finger cot, a solder, or a flux, the engineering and the education must be there.
Las Marias: It's the skills of the operators as well.
Solis: Very true! You must have a trained and informed workforce at all levels. From my experience most operators do as they have been trained and believe what they are trained to do has added value. If they have not been trained for cross-contamination avoidance they will still operate as if their process is not generating rejects. From Janitors to Managers and all steps in between. All must understand their contributions to cleanliness and reliability and help to decide if cleaning be required? It comes down to education, training, and communication.
Las Marias: Thank you very much, Paco, for that.
Solis: Thank you!