Solder Paste Printing From the Stencil’s Perspective
Jeff Schake of ASM Assembly Systems discusses the complications surrounding printing and solder paste that he sees from his perspective as a stencil expert.
Barry Matties: Let’s start with some of your background.
Jeff Schake: The division I work for is what was formerly DEK Printing Machines. I work on the printing end, and I’ve been with them for 21 years. My role has been the same since I was hired. I’ve worked in a research and development role. I report to ASM’s Product Line Innovation Center in the Weymouth, U.K., factory where our printers are designed, engineered, and manufactured. I specialize in conducting R&D process projects focused on printing and expanding our knowledge of the behaviors of the printing process. I report that back into the factory to improve our products and knowledge.
I also participate in standards groups and publish the research in proceedings at IPC APEX EXPO, SMTAI, and other conferences. I’m based in the U.S., and for the past few years, I’ve been stationed at the Rochester Institute of Technology CEMA Lab to conduct my projects. Professor Martin Anselm of RIT is the director of this world-class electronics manufacturing laboratory, and ASM has kindly contributed state-of-the-art printing and placement machines into the SMT assembly line there. In that facility, we have the full capability of assembly, inspection, various reliability testing tools, and failure analysis equipment. I enjoy spending time in there doing my projects and interacting with students.
Matties: One of your specialties is also statistical analysis.
Schake: Yes. Anytime you run an experiment, you have to analyze the data.
Matties: In this issue, we are focusing on finer pitch dimensions. We’ve done some research, and everything keeps coming back to printing and getting the stencils and solder paste on the board right. What trends do you see in the printing process?
Schake: The fundamental challenges haven’t changed from where we were 10–20 years ago, but what’s different now is the measurement scale associated with these challenges. Instead of working in a scale of millimeters, we’re down to microns now. We’re trying to assemble metric 0201 passives instead of 0402s. As you would expect, the stencil design is significantly impacted by this miniaturization trend. Aperture sizes are smaller, and stencil foils are thinner. The tough part with stencils and printing comes when miniaturization mixes in with larger legacy components designed on the same board.
In this scenario, designing the stencil with intent to print the smallest pads optimally, which is the typical strategy, may fall short of printing enough solder paste volume on the largest pads because the stencil foil is so thin. To boost the print volume printed on those larger pads, a thicker stencil foil may be required, but in doing so, this may compromise the ability to print controlled paste volume on the smallest pads. We refer to this as the “heterogeneous assembly challenge” within our in-house discussions, but I’ve also heard this referred to as “broadband assembly,” too. Call it what you want, but the main trend to highlight here is the range of component sizes continuing to expand, which is making it harder to design and implement uniform thickness stencils that will work in these scenarios.
Of course, there are tricks that you can play to enhance the print performance of the stencil, like adding nano-coatings to them. Nano-coatings represent a class of proprietary chemistries engineered to repel flux. These materials are applied to the bottom side of the stencil and were originally intended to inhibit paste and flux residue from bleeding out the bottom of the aperture holes and contaminating the stencil foil. But most of the technical papers you’ll find that address the benefits of nano-coatings will feature their claimed influence to improve solder paste transfer and print volume repeatability, which is helpful for challenging heterogenous applications.
We also see an increase in the use of dual-thickness or stepped stencils as a means to achieve printing a wider range of print volumes. A relatively new technology to fabricate stepped stencils utilizes the laser cutting machine that, in addition to cutting aperture holes, can be used to laser weld thinner or thicker foil pieces into the stencil foil. It’s so convenient, and with the programmability aspect of using the laser tool to do this, it lends itself nicely towards improved step resolution not achievable by traditional wet chemistry etch stepping techniques.
As we strive to push the limits of printing to finer dimensions, we see research published on details of the stencil, you wouldn’t have paid much attention to a decade ago. Take, for instance, the surface roughness of laser cut aperture walls and how this impacts solder paste transfer. Not only is the cutting equipment being improved to produce more uniform holes, but new foil material options exist today we didn’t have available before that have been identified to contain unique properties that allow them to be cut better. A higher class of stainless steel stencil foil material, which we generically call “fine grain,” has emerged claiming to offer cleaner cut and smoother aperture holes. Yes, this is a very subtle detail, but taken alone, this may not by itself produce a significant leap in printing capability improvement. However, all these details working collectively together—like fine-grain foils, nano-coatings, higher-quality laser cutting tools, advancements in solder paste formulations in use today, along with the print performance improvements achievable by modern-day printing machines—can achieve impressive printing capability gains to accommodate heterogeneous assembly challenges.
In reference to my involvement on the IPC 5-21E committee addressing the IPC 7525 stencil design guideline standard, we recognize the challenge of heterogeneous printing compels people today to use stencil designs containing aperture sizes smaller than our standard recommends. What I’m getting at here is the minimum area ratio stencil design guideline, which is currently prescribed in the standard as a value of 0.66. This figure is outdated and doesn’t reflect the print process capability that is achievable today using modern materials and equipment, so one of the changes that you’ll see in the next Rev C document will be a reduction to the minimum prescribed aperture area ratio.
Nolan Johnson: You’re saying we’re getting down to dimensions where you start to pay attention to things like the metal grain.
Schake: Sure! If you’ve seen all the items listed on the infamous printing process fishbone chart, you know there are so many variables that can impact printing. As the saying goes, “leave no stone unturned.” In fact, there are a handful of conference papers out there documenting printing capability as a function of stencil foil material grain size. The published papers on this topic typically support the use of fine-grain stencils as a means of achieving improved print performance. I believe there’s a correlation between the cosmetic quality of the cut stencil to its capability to print, and fine grain does play into this logic. Apertures cut using a finer grain material probably tend to have less of a wall texture, and the influence of this offers the possibility to benefit printing control. But, as I mentioned before, it’s more likely that it will take a collection of such incremental improvements working in harmony to realize the significant print performance boost we’re all looking for.
Holden: The interesting thing about where you are is that the whole assembly process starts with the solder paste printing. If it doesn’t go exactly right, it can’t get any better going down the assembly line. I’m familiar with DEK printers that we used in Hewlett-Packard. Regarding laser-cutting stencils, do they still do photochemical machining of stencils and stainless steel stencils?
Schake: Chemically etched stencil technology has pretty much dropped away except for the purpose of still using it for creating steps in stencils. Machine-drilled stencils are uncommon except for niche glue and adhesive print applications where thick plastic masks are used. Stainless steel laser-cut stencils account for the majority of stencils currently produced.
Holden: Is there a need for multiple thickness stencils? It’s a dilemma when you’re trying to do mixed parts that are fine and big because you only have a certain pad size and stencil thickness. Is there some flashy way now to get more paste than the thickness of your stencil determines?
Schake: Yes, there is a growing demand for multiple thickness stencils. This is yet another example of how heterogeneous assembly designs are influencing stencil design solutions, which may use step-downs arranged to print smaller deposits on the pads of critical fine pitch components. Step-up stencils are also a possibility and use selectively thicker stencil foil to increase the print volume required on large pads and connector-type parts, for example.
As for getting more paste than the thickness of the stencil permits, the laws of physics still apply, and the thickness of the printed paste brick is ultimately defined by this. That being said, I do recall an academic paper published many years ago out of Germany that did report printing solder paste deposits thicker than the physical stencil foil. The stencil foil used was produced with a matrix of highly co-planar micro-standoffs attached on its bottom side, so the printing that was demonstrated was essentially off contact extrusion printing.
Holden: Earlier today, I watched videos on 3D printing, especially 3D printing of metal and stainless steel. I thought, “Why not selectively print a stainless steel stencil with various thickness of the stencil? If you need more precision afterward, you can refine it with laser cutting or trimming of the stencil.” They put it upside down, so the squeegee side would be flat, but the other side would have various thicknesses in the stencil.
Schake: In my view, the logic is reversed. The flat side of the stencil needs to contact the board you’re printing onto to maintain a sufficient gasket so that you’re not going to bridge. The apertures on the contact side of the stencil need to be sealed against the board to prevent solder paste particles and flux from escaping; ideally, to achieve this, you have two flat surfaces in contact. In the scenario you described, it would be better to present the topographical side of the stencil foil that has different thicknesses levels to the squeegee side. Metal squeegees may not work the best, though, if you have multiple thickness levels or significant step displacements to deal with, but maybe that’s an application better suited to an old-fashioned polyurethane squeegee. It’s an interesting possibility!
Matties: In your career spanning over 20 years, you have seen a lot, for sure. What are some of the most common errors you see people have in their printing process?
Schake: If a printing issue is escalated to my attention, then, typically, the checklist of the most common errors has already been probed, and more advanced troubleshooting is required. My troubleshooting proficiency extends beyond the mechanics of the printing machine to consider the various materials that are required to run a robust printing process. I explore and seek to understand the properties of those materials and to verify the design of those materials is suited for producing the best print outcome. The definition of materials includes a host of things like solder paste, stencil, squeegee, tooling, cleaning consumables, and circuit boards. One common aspect of the errors I encounter relates to process design.
Matties: Are you saying to look at it from the total process point of view?
Schake: Yes, we look broadly at what makes a printing process successful and consider anything that can corrupt it. Happy, you referred to a multi-thickness stencil and said you’d put the non-flat side on the circuit board. I said that’s not a good process design choice. We’re talking about bad stencil gasketing there, which, singularly, can really foul the print result. But there can also be a combination of less severe things going on that, if left unchecked, can resonate to produce a similar negative print consequence.
As a metaphorical example, in a card game, you may be dealt a bad hand that requires you to adapt and change your play strategy accordingly. The cards matter. Performance of stencil printing is similarly dependent on the characteristics of the PCB “cards” supplied and influenced by obscure details concerning design and quality aspects such as pad size, solder mask openings, warpage, etc. Again, the cards matter! One of our top challenges is handling and fixturing highly routed thin and flimsy PCB cards designed to be populated with miniature components at extreme density and to position these inside the printing machine in a consistently co-planar fashion to the stencil. As opposed to a playing card, a PCB card has two sides to deal with.
Matties: What do you think about the move toward jet printing?
Schake: My view on it is that there are opportunities to improve assembly capability and performance based on the utilization of jet printing. It complements stencil printing, but won’t replace it. There are good places for it. For example, jet printing is an excellent tool to be used for prototyping and new product introduction. It may also serve well in some low-volume manufacturing and high mix environments. It’s very flexible, customizable, and configurable.
Matties: It addresses the issues you’re talking about with respect to the surface level and different requirements for components.
Schake: In theory, yes, but in practice, the actual process controlled result might be more difficult to achieve. I don’t claim to be a jet printing expert, but I think it’s fair to speculate on what could be challenging. From my casual observation, I’m aware of solder materials that are used in jet printing applications are very formulation sensitive and specific. There’s not a great variety of approved materials for those systems. Managing clogging may be a challenge, particularly when faced with unpredictable and intermittent line stoppages, but these things will be worked out and improved. Where I see those systems fitting in, at least in the scale of a high-speed process, is the utilization of a jet printer working together alongside a stencil printer supplying solder in hard to print locations or locations that require a boost in solder volume. In fact, there’s already jet print head options available inside an SPI tool. I think you’re going to continue to see new versions of this technology and capability emerging.
Matties: The high-mix low-volume players are going to be jumping on the jet printing. That makes a lot of sense for them. One of the things that another expert told us is that a bigger problem is not bridging, but opens and not enough material. Does that ring true to you as well?
Schake: You were saying opens were the concern?
Johnson: The temptation is to use as little solder paste as possible. The tolerances are tight now, where you used to have a couple of mils of tolerance, but now you don’t because that’s your entire pin spacing. The temptation has been, in the field, to use less paste and make a thinner covering of paste. Suddenly, there isn’t enough. The trickier field failure to find is the intermittent open as opposed to the short.
Schake: It’s an interesting statement. I don’t dispute any of that. The trend is to use thinner stencils. I associate your reference to opens occurring with ball array components. The combination of low print deposit volume and warpage can wreak havoc to cause head-in-pillow defects. But for other component types, the bigger problem could be a different defect, like bridging that leads to shorts. Here’s an example. A recent challenge that I’m facing requires assembling 01005 size capacitors, but the spacing between placed parts is one-quarter of the component width. That’s a tall order.
If the paste deposit is too large or shifted and then you start pressing the component down into the printed paste, the solder squeezes out. And then there’s the risk to bridge with the neighboring deposit. Now, you have to minimize the amount of paste printed while still ensuring enough is printed to form a joint. How do you do that? One of the things being looked at is printing through a 75- or 80-micron thick stencil, so it’s less than the standard four mils. You can do that for the 01005s, but you have a lot of other parts that you must consider the challenges of being able to print enough paste volume for. I see the future of where people are going and trying to push the boundaries of density, but by using a thinner stencil on some components, like the warp sensitive ball arrays, this can instigate head-in-pillow opens, as you alluded.
Johnson: What are the challenges you see from the stencil perspective for the solder paste manufacturer?
Schake: As I’ve described, one of the things recently that I’ve been doing is printing small deposits. It’s not necessarily a problem with printing that I have; it’s more the problem I’ve been having with reflowing because I’m getting a bit of graping. I didn’t want it to turn into a solder paste evaluation, but it has turned into something like that because these small deposits are difficult to reflow.
Johnson: Practically speaking, if we need to have less solder paste and a thinner coat, that means the solder paste and soldering process itself needs to be more effective.
Schake: Yes. Every solder ball counts. We’re getting to the point where we’re counting solder balls. You can quantify pass-fail based on the number of solder balls you can see. That’s a scary proposition.
Johnson: That’s precise.
Schake: Yes (laughs). Unfortunately, in my work environment at RIT, we don’t have a nitrogen-inerted reflow oven, and I tend to think this capability would improve my fine feature print soldering results. One of the things that I don’t completely have a good grasp of is a survey of the industry to know who is using nitrogen reflow and how common is it. Dare I say we are heading in a direction where the print requirement will compel people to be incorporating nitrogen in future assembly processes?
Johnson: For the sake of any of our readers who aren’t familiar with why nitrogen is important to the process, could you give a quick explanation?
Schake: Purging reflow ovens with nitrogen effectively creates an atmosphere in the oven where heating occurs without oxygen. The purity of this heating environment reduces exposure to oxidation. For very small printed deposits, like the ones I’ve been having trouble reflowing, these have a large ratio of solder ball surface area exposed to the oven environment compared to their bulk solder volume. In this case, the flux chemistry is working hard to preserve the solderability on all of the exposed solder ball surfaces. And without nitrogen, the chemical activity of the flux has to work harder to keep all the solder balls pure. In a worst-case scenario, flux activity can be prematurely spent, and the solder spheres don’t all coalesce together, leading to a reflow defect known as “graping.”
Johnson: The implication here is that with these very small, thin requirements, we’re starting to get to a point where nitrogen is now a requirement, not a “nice to have.”
Schake: These solder paste deposits can be so small that the surface area of the solder spheres is at a dangerously high level, where you don’t have enough flux to be able to fight off the elements of oxidation.
Johnson: That showed up in an earlier conversation as well. With less solder paste down, there is less flux to do its job, meaning that the flux now has to work harder and be better than ever before to get a good solder joint.
Schake: Correct. But keep in mind also that the chemistries that assist in the flux performance are not necessarily critical ingredients that have an impact on printability. You could have a paste that has a terrible reflow process and has wonderful printing or vice versa.
Johnson: Do you see good progress being made by solder paste manufacturers to help alleviate some of these challenges?
Schake: It’s difficult for me to judge progress because I’ve been so focused on printing and less aware of the reflow side of things. This aspect of paste performance has come as a little bit of a surprise, especially when facing air reflow requirements for these tiny deposits. I was thinking that it wasn’t going to be a problem, and I’ve found that it is, even using solder pastes recommended for fine pitch printability and air reflow capability. There’s still soldering issues that I was hoping wouldn’t exist, even on highly solderable gold pads that I know are clean.
Johnson: This fits with the industry dynamics right now, from a larger scope. We have automotive, autonomous vehicles, medical devices, and IoT. These markets project devices that will be very small, use very small components, and manufactured in huge quantities with a requirement that they are an order of magnitude or two more reliable than devices are now. This all points to the solder joint as a critical factor.
Holden: What’s your background or take on the flex circuit assembly solder pasting? Especially because the flex circuit is still one of the fastest-growing segments of the interconnect market.
Schake: We’re almost approaching flex-like substrates in standard PCBs now. These high-speed printing machines and assembly lines, which are legacy designed for rigid circuit boards, have to accommodate the future of these thinner substrates that are flex like. All equipment manufacturers have to take that into account. How do we prepare ourselves to accommodate this? In many cases, each of these flex-like substrates is attached to its own local support carrier and conveyed through the line in this format. Can we do this somehow without having to palletize everything that goes through the line?
Holden: Because we were making the flex circuits, we could take advantage of the panelized flex, and some of the tooling that went into making and fabricating the flex circuits could then also be used for the assembly. But it was always a fight and always a headache.
Schake: I’m all in on that! It’s rare that I hear about running anything panelized, let alone panelized flex. You did well to take advantage of it. Custom fixturing is indeed a headache.
Matties: What advice would you offer for a designer?
Schake: One of the things that commonly gets overlooked is solder mask. What is the design of your solder mask? What is the spacing of the solder mask from the pad border? How much of a gap do you have between it? How thick is the solder mask relative to the pad? Is it a solder mask defined pad, or a non-solder mask defined pad? Because all of those factors will have an impact on what stencil aperture size you can use. Based on the stencil thickness you have and the pitch, you could have an issue with area ratio, which could possibly challenge the transferability of the solder paste.
The other aspect of solder mask is that it is a topographical feature on the circuit board. Depending on how that’s designed, what the registration of that is, and how close it is to the pad, it’s going to impact the way the stencil gaskets to the circuit board. Solder mask is only one PCB design variable of many. Another one could be legend outlines. The ink marking around a component or around pads. I’m not a fan of legend ink because it tends to raise the level of the stencil in locally affected areas when the stencil is contacting against the board. If that legend ink is close enough to the pad, then the aperture is not going to seat itself properly onto the pad. You’re going to get a bad print.
Matties: That’s probably something people rarely think about.
Schake: Within the IPC 5-21e stencil guideline committee group, we certainly have discussed this, but you’re right; normally, people aren’t thinking along those lines.
Johnson: Because you didn’t have to before, but now we’re down to dimensions where we’re starting to worry about the metal grain in the stencil itself and counting the balls. All of a sudden, the raised dimension of the legend starts to make sense. We’re in the place where that’s no longer noise.
Matties: For the finer pitch conversation, we probably should be talking to some solder mask vendors, and look at the process of applying solder masks. As you’re saying, it’s got to be good coming in before it’s going to be good going out.
Johnson: I hadn’t considered the fact that a solder mask needs to be particularly cleaner.
Schake: I would prefer it. It would help with printing. I know there’s a difference in specifying solder mask technology when we order our test boards. You have LPI, which is “liquid photo imageable.” Then you have LDI, which is “laser direct imaged” solder mask. On some of our test boards that we use for printing, the LPI solder mask, particularly with the fine-pitch printing capability, is not a good combination. LDI solder mask is a much higher quality and a more accurate and controlled result. However, there’s a pretty wide variety of capability out there, even within the scope of LDI solder mask products that we’ve seen.
Matties: Back to what you were saying about the designer, that’s something that they would have to specify if they’re working on a difficult or a fine pitch board. They need to be aware of the impact of the mask as well when they’re making that choice, not driven by how much it costs.
Schake: Right. It can be a deal-breaker if you’re trying to print something that’s challenging. If you have solder mask thickness variation or solder mask misregistration, that can wreak havoc on your printability.
Matties: Especially at the tolerances that you’re talking about.
Holden: A lot of people are adding inkjet solder mask.
Schake: I’m not familiar with that, but it wouldn’t surprise me.
Holden: The inkjet solder mask is catching on pretty quickly.
Matties: It’s been around for years, but now it’s an overnight success.
Holden: The interesting thing is they’re changing all the design rules and using inkjet to outline parts. The solder mask isn’t continuous anymore, which is going to be interesting, because people forget to include it in their impedance calculation the fact that solder mask is a dielectric.
Matties: Jeff, we appreciate your time. Thank you.
Johnson: This has been extremely insightful.
Schake: I’m glad. Take care.