So, you need to find a MEMS foundry? See what these folks have to say first. Each person has a different focus and each one has advice that will benefit you in your search. The participants include:
- Carolyn White, an associate with AMFitzgerald, offering design and fabrication consulting services to MEMS and sensors companies specializing in early-stage development, design and prototyping of MEMS, as well as design for foundry processing.
- Paul Pickering, vice president of business development at Micralyne, a medium-scale MEMS manufacturing company. Micralyne provides “wafer fab and back-end processing, packaging, testing, et cetera," offering a turnkey solution. The company works with customers in later-stage prototyping and developing a manufacturing process with a goal to also provide production manufacturing.
- Jessica Gomez, founder and CEO of Rogue Valley Microdevices (RVM), a MEMS foundry with a focus on the development process through pilot production. RVM works with both large and small companies.
- Brian Stephenson, COO of Tronics, offering custom manufacturing for MEMS foundry services and also have a high-performance gyro on the market. It is a later-stage company for those that are ready to industrialize and move to volume manufacturing, but also has projects in earlier stages of joint development.
Carolyn Mathas: At what point in the product development phase should you consider working with a foundry?
Jessica Gomez: It really depends on the company and the resources available to them at the time they begin the product development phase. Many companies start off at a university or a shared fab space, and they work through first runs on their own; and then they come to a foundry when they’re ready to work with someone on more rapid prototyping, or materials that they don’t have access to but would like to use. A foundry provides greater consistency and expedites the development phase, and the company receives more consistent data that is difficult to obtain out of a shared facility.
Carolyn White: Ideally, when someone is interacting with a foundry they are coming in with a working prototype. We see that as an ideal situation, as it reduces the risk for the company while the foundry itself is more interested in pursuing the opportunity. It’s also important to remember that you need to have a clear business plan and the funds to take the project to final production.
Paul Pickering: Early proof of concept and trial of devices and structures is often accomplished in a shared facility or university. Micralyne takes on projects that have a fairly well understood proof of concept. We then work with the customer towards something that is truly manufacturable.
Brian Stephenson: From our perspective, we work with companies that are serious about taking something into production, have a well-defined development plan, and have the resources and a realistic expectation of what it costs to actually get something made—not only the cost of material but also in time. We’ve had circumstances where people had done a lot of development at a university and then expected to transfer that easily into a production environment, and are surprised to learn that many things must be changed to make it manufacturable. It’s important that potential customers understand the right time to start the process—certainly the earlier the better, but too early can also be a problem. These are things that we determine in our initial discussions.
Carolyn Mathas: Can you successfully work with more than one foundry to fabricate your device and, if so, what are the challenges you may encounter?
Carolyn White: Yes, they can work with more than one foundry, but they would need to understand that it represents twice the work. All of the development you do at one foundry doesn’t necessarily transfer to another. They are two unique facilities with unique capabilities. They would need to consider how important a second source really is. Depending on where they are in the development, it might be better to just focus all of the energy on producing one successful production line rather than splitting your energy between two.
Paul Pickering: I’ve worked with programs where they dual-source because they have expectations of very high volumes or some other critical reason to have alternate sourcing. It can be a tremendous headache and the cost is 2x or more, as bridging the gaps between different fabs is terribly difficult. You just can’t find two fabs with enough similarities to allow much facilitation of dual sourcing, so it is very challenging. In most cases we would discourage it at an early stage.
Brian Stephenson: We have been involved in cases like that, but it is highly difficult in time and cost, and it is extremely difficult to do in MEMS.
Jessica Gomez: if a company is looking to create a primary and a secondary source with two different foundries, it could probably be very challenging, costly and in the end, they may end up with two very different processes. There are many cases, however, where a company has a particular design with challenges of its own, and/or the process is a little unique. In those cases, there may not be one foundry that can support the entire process.
Sometimes what we’ll see is steps 1-3 are performed at one foundry, 4-6 at No. 2 foundry, and maybe the back end is done by the company itself or yet another source. It’s not ideal and it does require a lot of communication between the different teams. There may be cases where you can’t get away from it.
Carolyn White: Regarding what Jessica said, there could be an advantage to having one facility where you do your initial development and then that’s transferred to production at another facility. But that’s not something that would happen early on in the process. That’s something where your ongoing work is to keep your product relevant and up to date.
Jessica Gomez: Yes, that’s actually a cost-effective way to do that, and it can also help create a more manufacturable process early on. Later, you’ll be able to get the economy of scale when going to a larger foundry, so that is a good strategy and can work.
Carolyn Mathas: So when you initially begin working with a company, and it looks like there will be a handoff down the road, do you go ahead and suggest that and also suggest the next possible foundry?
Jessica Gomez: When we see a project initially, the projections for volume can be really high. We know that these are projections. We try to work with the customer to figure out what their timing is and if they say, “We’re going to ramp up in a matter of six months, and here are our needs," and they are greater than how we can support them, we direct them to another foundry that can meet their needs. If it is further in the future, we may begin working with them and in time start helping them to transfer that process out, either back to their own facilities or to another foundry.
Brian Stephenson: That sometimes works in reverse too. Sometimes we get them at a too-early stage and we refer them to someone who can handle the early stages. Or, sometimes they need to take some additional steps themselves, or they have a special requirement and we refer that out as well.
Carolyn Mathas: Please discuss your relationship as a partnership, and selecting foundries that are a fit on both a technical and a business level.
Paul Pickering: The foundry and the customer need to bond around a set of technical capabilities that match and fit the capability of the fab. One of my biggest criticisms coming from the semiconductor industry to the MEMS industry was the observation that for most MEMS fabs—everything is possible and nothing is possible. There’s this idea that if a customer will pay enough, we’ll figure out a way to do it. That might involve bringing in special capital equipment, and overall, I don’t really support this idea. There are an enormous amount of opportunities out there, and existing foundries can find very good customers that match their capabilities and establish a much better relationship and good development, rather than risk a protracted series of technical failures. For MEMS, it’s important that they be technically well matched.
Brian Stephenson: Managing expectations in these projects can be challenging depending on the experience of the customer and their background. If they are coming from a semiconductor company, their expectations can be wildly different than MEMS realities. Customers need to be prepared and also there needs to be a close and open relationship between the parties. Customers must also understand that what they developed at a university or at another foundry will have to be changed and adapted. Working closely together on the technical side is key. The manufacturing process of the MEMS device itself is so involved with these entire ecosystems, supply chains, the packaging, et cetera, it’s not a stand-alone part, and must be considered to be a part of the whole. Understanding expectations of how long it will take and how much it will cost—this can be eye opening.
Jessica Gomez: It’s very important to form a joint technical team, especially when you’re starting out with a new process and new device that the foundry hasn’t made with the customer before. Typically, initial meetings will include business people and some goals and expectations are set. It is about trust and technical competency all the way around and working through the process development segment to reach the goal. That can be a long process, depending on the complexity of the device. Being open and comfortable with each other really makes a big difference. The last thing you want, when something goes wrong, is an inability to move forward and solve technical challenges.
On the business side, clear expectations are also important, as is looking from a technical standpoint for areas for optimization to meet business goals. Often, the price point that must be achieved much later on is different than what you will see in the initial stages, so it’s good to know in the beginning where those areas are.
Carolyn White: The first thing to remember is that when you get silicon back, there is going to be a constant interaction, and it’s an investment on both the foundry side and the customer side. You want to make sure that not only your near-term goals match but also long-term goals. Things can always change, but keeping both immediate and long-term goals in mind can lead to a very successful relationship.
Carolyn Mathas: What is the approximate time involved in selecting a foundry and getting through initial feasibility and pilot production runs? How do you set expectations regarding both time and money?
Brian Stephenson: There are so many things involved, timing varies. The selection process depends on how quickly they are trying to do something—it may take 2-3 months in most cases, with some that are faster and some that take more time. Once the engagement occurs, typically, unless it’s a very simple device that doesn’t require a lot of complexity in the manufacturing setup, in most cases, it will be 12-18 months minimum to get through the development qualifications and the design iterations needed to adapt the design for a moderately complex device.
How do you set that expectation up front? You just have to be very open and clear on the steps necessary, what you plan to do, and what you expect out of them. We have customers who are very responsive and move things along and others that take more time. We make sure that they understand that, while there will be challenges, we have the technical capability and skill to solve them. It must be a very close partnership to get it all up and running successfully.
Jessica Gomez: This is an interesting question because for MEMS it’s really all over the board. It’s like the construction of a building or a house. It depends on the end result intended. For the customer, it ends up costing more and taking longer than they hope. We try to set realistic expectations up front. If we’re working with a startup company or a company that’s new to the process development piece, we try to set those expectations up with a, “Here’s what we think we can do, and here’s why it’s set up this way.” That’s important.
Getting your first prototype, usually for us, it can be anywhere from six weeks to a couple of months depending on the complexity of the device, and it can take as long as six months until the customer is happy. Sometimes we see the projects as they are refining them, so we do some experiments and they look at the results and say, “I see you can do these things. Can you modify that, because that would actually give us a better result…” and that’s where the technical partnership really starts to bloom in the relationship.
Carolyn White: Being involved in the process in the early part of the development and managing expectations is a huge part of what we do. However, when we’re involved early in the process, we’re sometimes the bearer of bad news. They’re usually very excited about what they’re going to do, have customers lined up, and then they’re provided with the cost and the time associated with it. That can be a deflating experience. But it’s important to get that message across early so that they are as prepared as possible and aware of what the commitment is going to be. There’s a large variability with MEMS. There’s always a so-and-so who is able to get through it in a short period of time, and our customers expect that experience. Unfortunately, with MEMS the experience is very unique to the customer.
Paul Pickering: Most MEMS companies work hard early on trying to scope out the project well and set customer expectations correctly. A lot depends on how much experience the customer has with MEMS. I will say to virtually every customer, "As you embark on a long and challenging development, there will always be unforeseen setbacks. It’s a lot like a marriage. There are going to be times when the two of you are not going to want to look each other in the eye, and there are some hard feelings that come up. It takes maturity on both sides to kind of work through that." In companies with large developments, we’ve had huge setbacks because engineers overlooked something. The same thing can happen and it doesn’t matter how experienced the foundry is. We’re dealing in this area of physics, in most cases, that isn’t well modeled and a lot of the effects are not well known. The only way to really understand something is to try it. One other point is that we really try to set up development projects that have good, solid checkpoints for both companies to come together and say, “Before we go to the next level of spending, are we okay with the results so far?” Short loops are the best friends of most MEMS developments.
Carolyn Mathas: How is the introduction of new materials affecting MEMS foundries today?
Jessica Gomez: New materials are really interesting. We sometimes see trends a little earlier than other companies, and we get asked for some really weird things. When the same thing comes up over and over again, it’s time that we need to really look at it. For example, we had very few customers interested in negative resist-types of projects and now, there is increasing interest in it. And the same thing with different process techniques like spray coating, and there’s more interest in polymer-type materials for applications as part of the manufacturing process—things like dry resist and other materials that we don’t have a lot of experience with. Companies will come and say that they would like to try these things and ask us, "What are the barriers to working on this jointly with you to develop a process for these materials?"
Carolyn White: Regarding materials, this question is more related to ensuring manufacturability. New materials can be a very critical part of a MEMS device and can mean the difference between it working and not working. If you are introducing a new material into a foundry, expect a lot of development work and it might affect your ability to transfer. We find that new material tends to be a thing customers develop in-house. Jessica mentioned that you might have a certain part of the process at a foundry, and then it’s transferred back to another foundry that has that expertise. It can definitely be a powerful piece of the puzzle, but you need to be aware that you’re on the leading edge and know what’s associated with that.
Paul Pickering: New materials are an interesting and evolving industry. Some of that work is being done to improve the efficiency of manufacturing. I think the most exciting are some of the Piezo technologies, AlN and PZP, and the set of applications looking at that for low-power and energy harvesting, it’s absolutely fascinating. I wish we had the funds to invest in the capital equipment to do some of that stuff. But we’re a generation probably away from it. I think it’s a very important piece of the evolving MEMS industry.
Brian Stephenson: As far as new materials, we’re looking at polymers, substrates, and it’s kind of a function of the types of projects or markets that we’re working in. For example, right now its biomedical, the medical MEMS space. The new materials are there for a reason—they need to scale down existing traditional types of devices, and the new materials are being adapted to do that. Also for the telecom area or in optics you have specialty coatings and unique depositions. So it depends on the market as to the kinds of materials we’re seeing. If you get outside of the markets that really drove the explosion in MEMS with the accelerometers, gyros, microphones, that were silicon-based and you start to move out to the new markets and applications coming into MEMS—this is where the scaling down is necessary.
Carolyn Mathas: What are the top three things that the customer should have in mind if it is the first time they’re searching for a foundry?
Carolyn White: If you are talking to a foundry, you have a process flow. Is there a technical match? Does the foundry have the capabilities you need? The second thing would be they may be very strong during the development and long term, is there a match there? And the last thing should be cost.
Paul Pickering: For me, it’s the technical capabilities. Does the device match the capability of the foundry? Is there a long-term economic match? We often see customers coming in with a great idea and a reasonably manufacturable structure but in production, it’s a $20 item when they want it to be $2. That happens a lot. It’s trying to apply too sophisticated a technology to a high-volume market. And then, is there technical synergy between the two? At times there are technologists on either side that don’t match up that well and if they are going to be partners, that’s a tough way to start.
Brian Stephenson: I agree with Paul—it’s technical fit and a business plan, as we have to believe that the customer really has a shot at getting there. From the customer side we want to see they are educated or have expectations and an understanding that, regardless of what they’ve done in the past or where they’ve done it, it will probably cost more and take longer than they expect, and that there will also be challenges moving into volume production.
Jessica Gomez: I would say it’s important when coming to a foundry, and if they are not going to be involved with the design part of it, your design should be pretty well vetted. You should have your design materials and have mapped out the real key processes that the foundry needs to have in-house in order to make the project a success. If you can find a match, technically a lot of other things can be worked out. I would also say that making sure that you communicate to the foundry what your pricing goal is in the near term and in the long term is really important for a foundry to be able to evaluate the level of investment that the foundry will be making to work with the company. Finally, I think that having someone on board that is ready and willing to work with the foundry on an ongoing basis is key.