TED日本語 - ジョゼフ・デシモン: 3Dプリンターを100倍高速化する技術

I'm thrilled to be here tonight to share with you something we've been working on for over two years, and it's in the area of additive manufacturing, also known as 3D printing.

You see this object here. It looks fairly simple, but it's quite complex at the same time. It's a set of concentric geodesic structures with linkages between each one. In its context, it is not manufacturable by traditional manufacturing techniques. It has a symmetry such that you can't injection mold it. You can't even manufacture it through milling. This is a job for a 3D printer, but most 3D printers would take between three and 10 hours to fabricate it, and we're going to take the risk tonight to try to fabricate it onstage during this 10-minute talk. Wish us luck.

Now, 3D printing is actually a misnomer. It's actually 2D printing over and over again, and it in fact uses the technologies associated with 2D printing. Think about inkjet printing where you lay down ink on a page to make letters, and then do that over and over again to build up a three-dimensional object. In microelectronics, they use something called lithography to do the same sort of thing, to make the transistors and integrated circuits and build up a structure several times. These are all 2D printing technologies.

Now, I'm a chemist, a material scientist too, and my co-inventors are also material scientists,one a chemist,one a physicist, and we began to be interested in 3D printing. And very often, as you know, new ideas are often simple connections between people with different experiences in different communities, and that's our story.

Now, we were inspired by the "Terminator 2" scene for T-1000, and we thought, why couldn't a 3D printer operate in this fashion, where you have an object arise out of a puddle in essentially real time with essentially no waste to make a great object? Okay, just like the movies. And could we be inspired by Hollywood and come up with ways to actually try to get this to work? And that was our challenge. And our approach would be, if we could do this, then we could fundamentally address the three issues holding back 3D printing from being a manufacturing process.

One, 3D printing takes forever. There are mushrooms that grow faster than 3D printed parts. (Laughter) The layer by layer process leads to defects in mechanical properties, and if we could grow continuously, we could eliminate those defects. And in fact, if we could grow really fast, we could also start using materials that are self-curing, and we could have amazing properties. So if we could pull this off, imitate Hollywood, we could in fact address 3D manufacturing.

Our approach is to use some standard knowledge in polymer chemistry to harness light and oxygen to grow parts continuously. Light and oxygen work in different ways. Light can take a resin and convert it to a solid, can convert a liquid to a solid. Oxygen inhibits that process. So light and oxygen are polar opposites from one another from a chemical point of view, and if we can control spatially the light and oxygen, we could control this process. And we refer to this as CLIP. [ Continuous Liquid Interface Production. ] It has three functional components. One, it has a reservoir that holds the puddle, just like the T-1000. At the bottom of the reservoir is a special window. I'll come back to that. In addition, it has a stage that will lower into the puddle and pull the object out of the liquid. The third component is a digital light projection system underneath the reservoir, illuminating with light in the ultraviolet region.

Now, the key is that this window in the bottom of this reservoir, it's a composite, it's a very special window. It's not only transparent to light but it's permeable to oxygen. It's got characteristics like a contact lens. So we can see how the process works. You can start to see that as you lower a stage in there, in a traditional process, with an oxygen-impermeable window, you make a two-dimensional pattern and you end up gluing that onto the window with a traditional window, and so in order to introduce the next layer, you have to separate it, introduce new resin, reposition it, and do this process over and over again. But with our very special window, what we're able to do is, with oxygen coming through the bottom as light hits it, that oxygen inhibits the reaction, and we form a dead zone. This dead zone is on the order of tens of microns thick, so that's two or three diameters of a red blood cell, right at the window interface that remains a liquid, and we pull this object up, and as we talked about in a Science paper, as we change the oxygen content, we can change the dead zone thickness. And so we have a number of key variables that we control: oxygen content, the light, the light intensity, the dose to cure, the viscosity, the geometry, and we use very sophisticated software to control this process.

The result is pretty staggering. It's 25 to 100 times faster than traditional 3D printers, which is game-changing. In addition, as our ability to deliver liquid to that interface, we can go 1,000 times faster I believe, and that in fact opens up the opportunity for generating a lot of heat, and as a chemical engineer, I get very excited at heat transfer and the idea that we might one day have water-cooled 3D printers, because they're going so fast. In addition, because we're growing things, we eliminate the layers, and the parts are monolithic. You don't see the surface structure. You have molecularly smooth surfaces. And the mechanical properties of most parts made in a 3D printer are notorious for having properties that depend on the orientation with which how you printed it, because of the layer-like structure. But when you grow objects like this, the properties are invariant with the print direction. These look like injection-molded parts, which is very different than traditional 3D manufacturing. In addition, we're able to throw the entire polymer chemistry textbook at this, and we're able to design chemistries that can give rise to the properties you really want in a 3D-printed object.

(Applause)

There it is. That's great. You always take the risk that something like this won't work onstage, right?

But we can have materials with great mechanical properties. For the first time, we can have elastomers that are high elasticity or high dampening. Think about vibration control or great sneakers, for example. We can make materials that have incredible strength, high strength-to-weight ratio, really strong materials, really great elastomers, so throw that in the audience there. So great material properties.

And so the opportunity now, if you actually make a part that has the properties to be a final part, and you do it in game-changing speeds, you can actually transform manufacturing. Right now, in manufacturing, what happens is, the so-called digital thread in digital manufacturing. We go from a CAD drawing, a design, to a prototype to manufacturing. Often, the digital thread is broken right at prototype, because you can't go all the way to manufacturing because most parts don't have the properties to be a final part. We now can connect the digital thread all the way from design to prototyping to manufacturing, and that opportunity really opens up all sorts of things, from better fuel-efficient cars dealing with great lattice properties with high strength-to-weight ratio, new turbine blades, all sorts of wonderful things.

Think about if you need a stent in an emergency situation, instead of the doctor pulling off a stent out of the shelf that was just standard sizes, having a stent that's designed for you, for your own anatomy with your own tributaries, printed in an emergency situation in real time out of the properties such that the stent could go away after 18 months: really-game changing. Or digital dentistry, and making these kinds of structures even while you're in the dentist chair. And look at the structures that my students are making at the University of North Carolina. These are amazing microscale structures.

You know, the world is really good at nano-fabrication. Moore's Law has driven things from 10 microns and below. We're really good at that, but it's actually very hard to make things from 10 microns to 1,000 microns, the mesoscale. And subtractive techniques from the silicon industry can't do that very well. They can't etch wafers that well. But this process is so gentle, we can grow these objects up from the bottom using additive manufacturing and make amazing things in tens of seconds, opening up new sensor technologies, new drug delivery techniques, new lab-on-a-chip applications, really game-changing stuff.

So the opportunity of making a part in real time that has the properties to be a final part really opens up 3D manufacturing, and for us, this is very exciting, because this really is owning the intersection between hardware, software and molecular science, and I can't wait to see what designers and engineers around the world are going to be able to do with this great tool.

Thanks for listening.

(Applause)