The team at Terecircuits, under the guidance of Wayne Rickard (CEO), Jayna Sheats (CTO) and Joe Fjelstad (Board Adviser) have developed two unique Parallel Placement processes that have the potential to revolutionize certain areas of high volume pick and place and SiP assembly. Trevor Galbraith interviewed the team at SEMICON West in California.
Jayna, if I can start with you. What is the background behind Terecircuits, where did it start and what’s your vision?
Jayna Sheats: It started with my thoughts, toward the end of my time at HP Labs working on organic LEDs where people wanted things to be flexible. There was talk about flexible electronics, but it really wasn’t. With a background in organic chemistry and polymers, I thought polymer transistors were a great idea, but I learned that they just really weren’t practical. Then I learned that you could make silicon very, very thin so that it becomes flexible and has a lot of the properties people wanted from the polymer transistors. And you could make those thin pieces of silicon very small, and you’d still have thousands of transistors on them, which would be enough to do a lot of interesting things in the internet of things.
But the question was how would you handle them? A pick and place machine just didn’t seem to be suited for that. And one thing and another led to this idea of how to use a polymer to do mass transfer of these thin small devices from a source wafer to a substrate, positioning them using light, very precisely. And so, not being a business person, I didn’t know really how to make a business out of that, until a number of years later, I finally met Wayne, who figured out a good business model. That’s where it came from.
Wayne, you have two technologies that we’re going to discuss today around the polymer transfer material. The polymer material can transfer a number of components quickly, and then release them under the use of a high-powered UV laser? Is that essentially what the patents are based around?
Wayne Rickard: Yes, that’s correct. It’s actually quite a flexible technology, it can release thousands of components, or hundreds of thousands of components in a single operation. One of Jayna’s main insights was that you could take the technology that has the resolution of lithography, which goes down to the nanometer scale, and you could apply it to semiconductor assembly. And the challenge there is that pick and place is a technology that’s been around for 30, 40 years. It’s not known for being able to parallelize a lot of operations. An eight head pick and place machine is state-of-the-art, whereas we can transfer thousands and thousands of components in a single operation and we can do it with one micron of accuracy. To do that with a pick and place machine, with that type of placement accuracy, you actually have to slow the machine down because of mechanical settling time and positioning time. This is an amazing approach. It solves a throughput problem, it solves an accuracy problem, and it actually solves, in some cases, a yield problem as well.
The application that I’ve seen you demonstrate is the transfer of LEDs. It’s all the same device. But what volumes and transfer rates can you get?
Wayne Rickard: There’s really no limit. The components themselves are about the size of a red blood cell. You can imagine on a four inch epi wafer or a six inch epi wafer, you can get millions of tiny components on one wafer. We’re only limited by the mask that has apertures for where you want to release a chip. It’s
possible, in one operation, to not just transfer the components, but do a pixel realignment so that in the case of a micro LED display, you’ve got a red, a green and a blue pixel that have to be created. And that’s three different epi wafers and then three different mass transfers and all with precise alignment. But the pixel has a different pitch than the fabrication pitch, so by using a mask, we can actually transfer them at the same pixel pitch.
Jayna, this is a method of transferring these LEDs onto a substrate of some sort, but not interconnecting them. Are you literally placing them and then the manufacturer can decide which way he’s going to do interconnect them, whether he’s going to connect with an epoxy or whether he’s going to solder them or some other method?
Jayna Sheats: That is correct. In the case of what kind of interconnect technology could be used, solder paste or solder bumps, or nowadays, the highest resolution one has thermo compression bonding. We have the ability within the process to control the amount of force that we bring those components down on, which is an advantage compared to laser ablation. Laser ablation is a competitive process, but there’s really very little control. We believe we’ll be able to control that Z force so that almost no matter what the surface is, as long as it has some adhesive properties. Even ‘Van der Waal’s Attraction’ is a form of adhesion.
Do you know what level of force you can control on the Z height?
Jayna Sheats: That’s still in progress, so I don’t want to put a number on that.
You have another process that you’re using for System in Package (SiP), where you’re essentially using the same type of transfer process to transfer chiplets and materials into a package. And this process is solderless. Can you tell us a little bit about that?
Jayna Sheats: Sure, with many cases of a new technology, you have the low hanging fruit, or the first things you can do that are easiest to introduce, and then there are the more complicated things.
As far as the photochemical process that changes the polymer, that creates the gas and a propulsion of a chip, it doesn’t care whether it’s an LED, a micro LED, or a five by five or 10 by 10 millimeter chip. It could be anything. The worlds that you sell into are very different. But, we’ve discovered recently, there are some customers, for example power electronics used in the automotive business where they’re using Silicon Carbide and Gallium Nitride, and those also have to be thin. There are various reasons why that’s the case; heat, thermal control and other reasons. But their pick and place machines are breaking those chips. Even doing those transfers one at a time has a tremendous economic value.
My idea, which I believe we’ll get to eventually, is to make something called a System in Package, similar to what goes into a smartwatch today, for example. I think this is already a few years old, Apple’s SiP had 30 different components inside this one package. You could be making hundreds, let’s say a hundred or 200 of those components, System in Package all at once. You place one chip in each one, microprocessor in each one of them, and you do that all in one second exposure, I mean, less than a one-second exposure. And then you come along and you place the memory chips and so on.
You’re doing the advantages of the parallel processing also for System in Package, and the chiplet world just only takes that to another level.
I want to come to you, Joe, for a moment, because a lot of this technology is based around the original OCCAM principle that you had.
Joe Fjelstad: Let’s put it this way, it’s similar. Jayna and I, we’re of a common mind it turns out. Like a lot of things, innovation happens in lots of different areas and people see the same ‘kinds of obvious’, shall I say, when considering the way things ought to be. Jayna was focused. I gave a talk on the OCCAM concept back in 2008, I think it was, in Florida, at an SMTA event. And Jayna approached me afterwards and we just started chatting. She understood exactly what I was talking about, and I thought, “Oh, great. I don’t have to explain it to her.”
But her vision is similar to what she just described. That idea of what could be possible for electronics manufacturers at a bigger level. I think some of the first efforts were up in Waterloo, Canada, where the intent was to see what might be done in terms of building RFID chips. And so she had a good handle on how things get done. Of course, there’s a lot of effort in that and certain things happen
in her own way. But to me, the real beauty of the process, is the idea of being able to have an attachment material that disappears with light. It just goes to gas.
It vaporizes?
Joe Fjelstad: Yes, there’s nothing to clean up. It’s done and everything is right where it needs to be, and it’s virtually untouched, so it’s incredibly pristine. Well, that same kind of technology is attractive to me. In fact, back when Jayna asked me first to be an advisor to the company, I started looking at where are the opportunities in my space where it might work, and one that struck me as an opportunity area would be pick and place of these discrete devices, which are now so small, you can put them in your pepper shaker, put them on your eggs in the morning and wouldn’t know that you ate them. They’re that tiny. And so that ability to be able to place these things, create masses of them, and then deliver them to, and then just drop the components where they’re needed was really, really attractive. And it remains attractive for the things that I’ve been talking about with OCCAM.
It must be really exciting for you to see this idea coming to fruition. There’s been a lot of people that really liked the idea and have liked the idea for years, but we needed to see proof of concept. We need to see it actually working. And here we are coming to the point where we’re actually producing solderless packages.
Joe Fjelstad: I think everything has it’s day. As Victor Hugo said, “nothing is more powerful than an idea whose time has come”, and I think that’s still the case. We start to see things. The one thing that I think is going to be incredibly important for this, is in the world of chiplets. I wrote an article about 10 years ago or more, talking about disintegrating ICs. Break them down into IP blocks, and then package them. And the key for me is standards Going back to my days at Tessera, they were focused on standards. In other words, being able to make flip chip technology available to the common man.
And you do that by adopting a standard footprint and then letting things fall together. And so when I put my little module together, with the help of my colleague, Verne Solberg, it’s basically using Lego bricks. I said, “Here, Verne. I’m going to get you a bunch of Lego bricks. Make a module for me.”
Well, I remember back in the early days of Tessera, you were intending to be a manufacturing outfit. You gave me a tour me around the factory, and I remember you saying “that’s where the manufacturing’s going to be”. But after you got proof of concept, you literally turned into a licensing company.
Joe Fjelstad: Yes, we couldn’t handle the world’s volume. I mean, we were actually developing materials at the time in our own in-house for silicones for the die attach to be able to create the buffer layer between the chip and the board. In other words, overcome the major flaw or failure mode in electronics is always a solder joint. If you take the stress off the solder joint, then you have something that will work, or should work nicely for a long time. But we went to Dow and asked them to make something like this, and they said, “Sure, we can probably do that for you. How many car loads do you want?” And we said, “No, no, no. You don’t understand.” There’s a value to this and that value proposition doesn’t always register with large manufacturers.
I want to turn back to Wayne for a minute because Wayne, you came into the company about four years ago with a strong business background to help bring this process to market. Where are we in terms of realizing the concept, getting into production and getting licensing deals or whatever it is you’re trying to do?
Wayne Rickard: I have a background in angel investing, and one of the concerns I had when I first met Jayna, my first thought was, “Wow, this is huge. Semiconductors had their day in the sun, but trying to get angel investors, or even VCs, to invest in a technology like this today, it’s going to be a long road”. We’ve been fortunate in that we’ve been able to raise $5 million over the course of the last three or four years. We’ve also received some government grants from the National Science Foundation. We ended up with a pretty good source of seed funding, and we thought originally we might have to build a tool to support this process, which can be very capital intensive and it’s not really something you want to take on. But when we started looking at the environment, there were lots of competing approaches, lots of ideas. No big company had committed to building a tool that would support a laser-based transfer.
Today, there’s eight separate companies that we’re aware of that have laser-based tools. They will all be compatible with our material. And what’s exciting is that they’ve put the ‘cart ahead of the horse’ in a sense, because the norm is a tool company will build a tool and ask, “What’s an off the shelf material that I can work with?” But they found that all the off the shelf materials use a lot of energy, they get ablated and they leave residue. And we have a purpose-built material that’s met the crossroads of when the tools and materials are ready. We’re working with, I would say, a majority of those companies now, and we’re qualifying our material. A lot of those tools will work with our process and our material.
And I think for us, it’s been exciting because number one, we don’t have to build a tool. And number two, they’re attacking different verticals. We have the micro LED opportunity that we mentioned earlier. There’s some tool companies that are completely focused on that, and others are focused on semiconductor and pick and place replacement, and we have partners in that space as well. For us, it’s all about now getting qualified and getting some pilot lines up and running. And as you know, in the semiconductor world, nothing gets qualified overnight. It’s going to take several more years of work, but we have the right partners and it’s looking very positive right now.
Can you give us a sense of a timeline at this point?
Wayne Rickard: Well, I can say that we did sign our first joint development agreement with a major semiconductor provider, and that timeline will be about a year. And at the end of that timeline, we’ll be exiting with a tool partner that they’ve also identified, and a process that works. Well, we’ll look forward to that. Thank you all for coming in and telling us about this today.
–TREVOR GALBRAITH
