3D Printing Getting Cheaper...

"When laser printers cost more than $5,000, nobody knew they needed desktop publishing." That's an insightful quote from A. Michael Berman, chief technology officer for the Art Center College of Design in Pasadena, which appeared yesterday in a New York Times article about the falling costs of 3D printing.

Beam It Down From the Web, Scotty describes a printer from a company called Desktop Factory that they plan to sell for $4,999. It uses a technology I haven't seen before: it deposits a thin layer of powder on a drum, melts the powder in spots, then rolls it onto a growing object. Bill Gross, the chairman of IdeaLab (the incubator of Desktop Factory), says that the printer can be built for $300 in materials.

The New York Times article also mentions the Hod Lipson's Fab@Home project, which has developed a 3D printer that can be build from a $3,000 kit. And 3D Systems plans to sell a printer for $9,900 this year, but the company's chief executive thinks they can get it down to under $2000 in three-five years.

To date, most 3D printers print only hunks of stuff - no sensors, electronics, motors, or batteries. But Prof. Lipson already has preliminary designs for sensors, batteries, and actuators.

We often talk about how nanofactories will make it possible for people to build large quantities of high-performance products at near-zero cost. 3D printers provide early hints of fabrication-on-demand. But it's worth keeping in mind that nanofactory products will be orders of magnitude more functional and powerful than 3D printer products.

Chris Phoenix

Tags: nanotechnology nanotech nano science technology ethics weblog blog

Comments

[jim moore]

If you want to watch a video of a utility fog like system that Carnegie Mellon and Intel are working on, follow the link to claytronics

[Tom Craver]

"It took IdeaLab a year to prove that the basic approach would work and a second year to develop the technology to get the layers to stick to each other properly. (The model is gently squished, as in a sandwich press, after each layer is applied.) And it has taken two more years to write the required software and to create a working design for the first production model."

A useful data point for discussions of how long it'll take to develop MNT from the point it becomes feasible.

More money and manpower could accelerate this - but then the first MNT unit will likely be significantly more complex, and of course it'll be at the nanoscale.

[jim moore]

I was thinking that there should be an evolutionary pathway from todays 3-D printers to Nano-block Fabbers. *
It should be possible to build a 3-D printer today that uses some kind of planer assembly technique to make an object. Rather than the blocks being ~1 cubic micron make them 1 cubic millimeter. At first you could have simple structural blocks (different colors/ different materials) but soon after functionality (sensors, batteries, computation, actuators, etc.) can be added to each block. For assembly i was thinking something simple like a fractal stream of blocks merging together and mechanically locking in place.

The big question is what is the "killer app" for 3-D printing?
Tele operated robots! At first you have systems like Leggo mind works. Next step it to use 3-D printers to make the structural parts for the robot and buy standardized motors and other electronics . Then you proceed down the path of shrinking the functionality of the motors, electronics, etc. to the size of one of your blocks. At that point your 3-D printer could output a fully operational robot.

And what would people do with their robots?
Robot wars of course.

*I don't think that allowing widespread use of nano-factories that use easily available raw materials is a good idea. Reusable nanoblock fabbers allow for nearly all the same benefits, while limiting many of the ecological problems, security issues, and economic disruptions.

[Chris Phoenix, CRN]

Tom: Yes, that's a useful reality check. Note that half the time was in writing software and designing a production version. If someone built a nanofactory without the software, and then started writing the software, that could indeed add several years to the "flood of products" start. I wouldn't expect anyone to be that short-sighted.

But as nanofactories get easier to develop, it will be easier to design a program that ignores the software aspect. After all, if it's only going to cost you $10M, then you only need to make $10M of profit (modulo risk and net present value) and you don't need very good software to do that.

Jim: Mechanical locking of blocks is far from trivial. Zyvex has developed several locking schemes (for MEMS, no less) that are quite clever, even self-aligning. But they use small tabs with even smaller contact points, so they must be far from the strength of the bulk material. And the exact design depends on the material properties and the fab characteristics. So it'd take some R&D. That's not to say it couldn't be done.

Lessee now... a 200-nm block has space for about 1100 carbon atoms in a row, or 1.4 billion atoms total. An RP system today might have 100-micron voxels, so a comparable block would be 11 cm on a side--too big to build lots of them with RP and combine them into products. If you made the block 1 cm, it would have only 1 million voxels, 100 cubed, which is small enough that it'd be hard to implement intricate functionality in a single block. Could even be hard to implement good snap-fit machinery. And of course an RP voxel isn't nearly as good as an atom: they break apart more easily, the surface is less wear-resistant...

Scaling laws and atomic precision are important benefits of the nanoscale, which you just can't get with an RP machine today. So it'd be a good demo, but I'm not sure it could be more than that. Of course I'd love someone to prove me wrong.

Chris

[jim moore]

"Lessee now... a 200-nm block has space for about 1100 carbon atoms in a row, or 1.4 billion atoms total. An RP system today might have 100-micron voxels, so a comparable block would be 11 cm on a side--too big to build lots of them with RP "---- but the perfect size for "hands on" demonstration of the idea of nano-blocks. Think about it Chris, I bet if you designed a few blocks, had someone 3-D print a few copies of each design, they would be a hit with almost every audience. (especially those very important small audiences.)

[jim moore]

Chris,
I also think did not explain myself clearly enough. None of todays 3-D fabrication system uses the planer assembly (or convergent assembly) model for making objects, but I think that it can be done.
The reason I think that it would be a good idea if people started to use the planer assembly technique is: the system can assemble things with simple blocks of material today and 15 years from now the same technique can use highly functional nano blocks.

Why do you think todays RP's have a voxel size of ~100 microns?
nanoventions looks like they can print with an accuracy of 1 micron or better. Now, Nanoventions does not do Rapid Prototyping but my point is that maybe the reason the size of the voxel is not smaller may not have anything to do with the limits of precision. I think the reason the voxel size is not smaller is two fold: smaller voxel - longer time produce an object of the same size; smaller voxel - bigger computer file. (both problems are much easier to handle with planer assembly) If you go with the 1 micron voxel size a 1 millimeter block is analogous to one of your nano-blocks.

One way the assembly process might look is:
An empty tube travels back and forth underneath a series of dispensers, The dispensers push different types of blocks into the tube in a programed order until the tube is filled. The process repeats itself many times and the filled tubes are put in a precise array. After the array of tubes has been assembled, the contents of all of the tubes get uniformly pushed through an extrusion plate. Inside the extrusion plate the streams of blocks are pushed and locked together (if mechanical bonds are too weak or difficult, adhesives could be used) A solid block of material gets extruded, in order for you to get the object you made the "place holder" blocks must be removed.

[Tom Craver]

It's probably not too soon to start mocking up some simple nanoblocks. I'd bet that some of the fabbing companies would be interested in making some samples, for the publicity they'd get from the various talks CRN does.

So how should they be designed? While not perfectly safe, if nanoblocks require a key for disassembly (for recycling) and maybe for assembly, any device that would disassemble objects to steal nanoblocks would have actually incorporate a recycling unit, making it large enough to be obvious. And to self-replicate, the device would have to steal recycling units.

Separate issue: If we assume we're within 15-25 years of having nanofactories, maybe it'd be smart for CRN to begin building a "patent pool" - providing some leverage over future 'design policy' for products based on those patents, via license control.

I.e. should nanoblocks be recyclable? Should fabbers be able to build complete copies of themselves out of nanoblocks, or will some parts require atom precise MM that are only distributed as integral parts of nanofactories? Require DRM, allow DRM, require no DRM? Etc.

This might also provide an answer to how MM can get financed. Create macro version of simple nanoblocks for hobbyists - call them "LOKA blocks" - about the size of LEGO blocks, but designed to lock firmly together. Later, make a fabber that can use a millimeter-scale version of LOKA blocks. Profits would go to a non-profit foundation, financing key MM R&D and ultimately aiming to insure that MM is a blessing for the entire world.

[Chris Phoenix, CRN]

I know that 2D processes (including lithography) can make far smaller features than 100 microns. But I don't know that 3D RP machines typically get a lot smaller. As you say, it would drive up replication time. Current RP seems to focus on handheld-size models, and to find post-processing acceptable. I'm sure smaller layers/voxels would be possible, but I don't know if it's available from a mass-market machine.

On patents, there are a lot of subtleties of patent law: for example, you can't patent something that's not buildable yet. To do anything useful with patents would take a lot of money for a high-quality IP lawyer.

A demo nanoblock sounds great, but it'd just be a non-functional and mostly-opaque model. Most of the really cool stuff in MM is done with moving parts, and you couldn't get that with RP.

That's not to say models are useless. Even a static model of some of the designs that have been coming out of Nanorex would be really cool. But I think a nanoblock with non-working parts too small to see would be more confusing than enlightening.

Chris

[Tom Craver]

What I have in mind is somewhat simpler. The ONLY property to be demonstrated is the secure locking of block to block.

Think more in terms of the simplest possible mechanical design that can allow 3D interlocking of blocks - only slightly more complex than LEGO blocks. Something that can not only be built, but in a really crude implementation could even be cast in a single piece from plastic or metal.

Think in terms of those simple nylon "zip strips" that ratchet closed to lock so firmly that the police use them as temporary handcuffs.

Now imagine taking a handful of LOKA blocks, clicking them together in the shape of a wrench, and then using that to change a tire.

[Chris Phoenix, CRN]

A self-replicating factory-type robot built of RP blocks was done a few years ago. The blocks snapped together. It was slightly too weak - one of the beams sagged too much. But if you just want snapping blocks, I think we have it already!

If I were going to demonstrate mutable matter, I might look for some other way of joining/fusing the material. Plastic-coated metal beads - hit it with RF and it melts in place. Or something like that.

Chris