Like so many other markets and industries, the COVID pandemic has affected the supply of components used to make semiconductor processing equipment. Semiconductor capital equipment manufacturers are turning to additive manufacturing to help repair the supply chain and enable designers to design for function first. I spoke with Scott Green, Principal Solutions Leader for semiconductors at 3D Systems on this subject. Here are some of the highlights of the interview.
The challenges in semiconductor production
Today’s challenges in semiconductor production (especially as it relates to the effects of COVID) include global production capacity which has been affected by the trade disruptions caused by the Trump administration and the disruptions in working styles all over the planet due to health restrictions.
If the Biden administration policies end up requiring semiconductor foundries to buy new equipment, that will be a supply and demand issue with capital equipment itself. Acquiring fabrication equipment will take time. Even if a semiconductor foundry decided today to add a manufacturing line, there would be a lot of work ahead to get it placed and start pumping new silicon into the supply chain.
So, the major challenge is both a production capacity issue and a supply demand issue. Once COVID restrictions subside, these challenges will be alleviated a bit, but we still have a supply and demand issue where there is an increasing demand for smart devices, smart automobiles, and as new industries tap into the need for semiconductor chips, the capital equipment manufacturing companies are going to have to catch up.
One way that additive manufacturing comes into this is that it generally helps improve the technology developed and brought to market for those systems. Assuming that a foundry is already running equipment at max capacity, the foundry managers can’t really decide they’re going to do something different with that equipment or go to a new, dramatically different process. So, dramatic changes are not really possible with existing capital equipment for processing silicon wafers or lithography.
Since we’re already pushing the boundaries of lithography equipment and what’s possible with physics, one of the technology related barriers for semiconductor production is going to be how do we get the latest generation of fabrication equipment into the hands of those foundries so that they can satisfy the requirements of customers. We need more equipment that’s more capable in the market to satisfy the needs of these manufacturers so that you can have more transistors occupying the same space, but taking lower power.
An example of how additive can improve semiconductor processes
Lithography: There are more than 100,000 components that go into some lithography machines. Every single one of those is built in relatively small quantities, maybe a couple of thousand specialized parts from implementation to production run. What we have is a complex system with a big supply chain of relatively low volume orders from suppliers. So, you have designed compromises pretty much all over the place inside of a lithography machine.
In many cases, additive could enable those systems to work much closer to the theoretical expected working environments, as opposed to making compromises in machine operation because of how you have to manufacture things. The benefits include greater precision, higher production capability, faster cycle times, more wafers produced per machine per week. You’re also going to see a better quality of imaging across the entire wafers. That’s going to mean less waste and higher quality product.
Additive manufacturing allows you to optimize the strength-to-weight ratio. If you have a big armature assembly, additive manufacturing allows you to make use of design flexibility to optimize that component so it only takes up the minimum amount of the space needed, yet has the strength to execute the task. Topology optimization or structural optimization lets designers create relatively massive parts that are light in weight; the use of lattice structures, for example.
But you can’t make such parts using traditional manufacturing processes. They require a different process, such as additive manufacturing.
Another example, there are a number of manifold fluid lines inside a lithography machine. Additive manufacturing is much better at producing parts with conformal or interior cooling structures for better fluid manifold dynamics. A designer no longer must compromise the design to fit the manufacturing technology. For lithography machines, a designer can eliminate the connections of hoses and tubes previously used.
Additive manufacturing allows you to build a fluid manifold or a cooling structure that is prioritizing function over manufacturing capability. You’ll end up with smooth channels or channels that don’t take right angle bends that could cause fluid disturbance.
Another example, a wafer table. You could design any cooling structure inside of a wafer table that you can imagine. And typically, these things are going to be driven from numerical simulation.
Design flexibility is something that you get with additive manufacturing. You are no longer restrained in the design because of the way it will be made.
Why is additive manufacturing in semiconductor applications gaining attention now?
The semiconductor industry has been using additive technology for some time. “Why we haven’t heard about it is because it’s just not as sexy as airplanes and automobiles and missiles and stuff like that.”
Another factor is competition. You have only four huge mega-corporations that make the major chunk of lithography or fabrication machines, “and they ain’t telling anybody anything. They want to make sure that they keep that competitive advantage.”
Semi-con is also highly specialized, so it’s been happening in small pockets. And I think what we’re trying to do is help connect emerging demand with the solutions that we understand for the market.
“We’re also running into physical barriers in the semiconductor industry. Getting down to a 14-nanometer process was a big hurdle in general from what I remember in past history, and it’s going to get even harder. I don’t see us going below a nanometer ever. Some major step change will be needed for the entire system of producing product, and we’re not there yet. But what additive manufacturing is doing, particularly direct metal printing, is very precise and the materials are there. Additive technology is coming down in costs. It’s becoming relatable due to other things like us working with large Hadron colliders, or advanced light source projects, or even automotive and aerospace.”
And now the CAD industry is focusing on design for additive manufacturing. Earlier CAD programs were a limitation for the whole industry in general.
In the semiconductor industry, you’re only as good as your weakest link inside of a complex system. And when you take something that has a ton of error tolerance, and you start removing all the different thermal gradients, all the fluid turbulences, you start to get a system that functions more efficiently. Again, we’re referencing lithography here. Such changes are going to help us squeak out productivity in the current framework of paradigm production for maybe another 15 or 30 years. And in a relatively short period of time, you’ll start seeing those relatively major effects in consumer products in just a few years.
Inside a wafer table
Inside lithography, for instance, you have a wafer that’s on a plate for lack of a better way of putting it. The “plates” function is to make sure that it keeps that wafer at a stable temperature within a couple of milli-kelvin. Eventually, it’s going to get to an equilibrium where it’s no longer exchanging heat.
But keeping a wafer at a stable temperature has been a limitation because of the wait time involved for the wafer to stabilize, and current cooling and conditioning methods don’t provide uniform thermal environment with precise control. That time is lost productivity. If you can get to a stable temperature faster, you can pump out more wafers per week, improving cycle time.
While this is a functional production improvement, there’s also a quality improvement that’s inherent to having a thermally stable wafer. On a microscopic scale, when temperature is fluctuating, the wafer is actually moving in space. You can’t see with your eyeball, but it’s actually moving as it reaches towards a stable temperature. When you keep something at a very controlled temperature with low thermal gradients, it’s probably going to stay flat. When you project light onto it, you’re projecting it onto a flat surface. And you know that you will get the best possible image projection, which means you get better results.
Traditionally, the conditioning plates and cooling tables have been brazed. Multiple parts are brazed together to create a single component. The advantage that additive manufacturing provides here is that a design can be inspired by artificial intelligence to create cooling channels that serve the function. Additive enables us to design for function first, which it’s really cool and freeing for mechanical engineers.
With additive manufacturing, I can try it, I can print it, and test it on a bench top setup. If it’s not as good as I thought it would be, I can make a change and try it again. That’s a really small meantime between iterations with design engineering. Whereas if I was to go through the traditional supply chain, that part might’ve taken me a month or two to get because it had to go through an ordering system, somebody had to machine it, somebody had to assemble it, somebody probably tested it, did QA. Then they put it in the mail and ship it and then you can finally test it. But with additive manufacturing, you cut out all those steps. You’re able to, for this specific part, iterate quickly on new design concepts, which will allow customers and designers to get to the ideal functional benefits much faster.
If you were to machine that, not only will it take a relatively long machining operation, but you need to assemble the part, and then you’re limited by your ability to iterate because of time, overhead, reprogramming and so on.
Cooling channels can actually surround or shroud around a light source. For instance, you can have an embedded spiral channel instead of a shroud with assembled cooling tubes, if you need to create a cooling jacket. The sky’s the limit when it comes to manufacturing with additive, but there’s not infinite advantage. Of course, you’ve got limitations. They’re just different than other tools and it opens up new realms of possibility.
How much of the semiconductor industry is turning towards additive manufacturing to help create the equipment that they need?
You could actually cast a wide net of categories to describe how much of the semiconductor industry is using additive technology. I would say generally you could break it up into a couple of pieces:
- Major capital equipment manufacturers and their suppliers, and those suppliers are either supplying individual components or subsystems. (A subsystem provider, for instance, might just make lasers, lots of different types of lasers.)
- High-end industrial electronics, which are a little bit more down the value chain. It’s not directly related to it, but examples include cooling plates, thermocouples, heat exchangers in industrial electronics, or other thermal management solution.
Right now, we don’t see a ton of opportunity at the foundries themselves because unless machines are changing hands and new machines are going in or out, you’re not going to retrofit something—if it ain’t broke, don’t fix it. But we’re seeing more opportunity for new products being shipped, new manufacturing lines being created. And so we’re seeing a large, really exciting future ahead of us to work together with the semiconductor industry and so many cool applications.
Do you see the supply chain easing up since the pandemic, or is that still yet to come?
As working limitations change, the foundries are going to be able to run at a faster clip and they’re going to be able to produce more products. But the technical advancements that come into the newest machines, newer processes that are being shipped for the biggest manufacturers, that’s really where we’re going to see the unlocking of speed and potential.
We’ll see consumer electronics that consume less power, are more intelligent, and really interesting, more powerful automotive products.