Slashdot Mirror

Ball Semiconductor has an interesting take on creating integrated circuits: Put them on little spheres of silicon!

The beauty of this approach is that you can create different balls with different functionality and then cluster them in 3D shapes. You also can cool by using fluid or air-flow through the spaces between the balls. Of course you have to use something to create the connections between the balls, but it requires so little material that you can go back to using a pinch of lead in the mix.

Result? All the functionality of the computer on your desk (other than long-term storage) could be put into a 3" cube of bb-sized spheres. And each one could be custom made to your specifications by picking which spheres are included in what configuration.

Ball Semiconductor has an interesting take on creating integrated circuits: Put them on little spheres of silicon!

The beauty of this approach is that you can create different balls with different functionality and then cluster them in 3D shapes. You also can cool by using fluid or air-flow through the spaces between the balls. Of course you have to use something to create the connections between the balls, but it requires so little material that you can go back to using a pinch of lead in the mix.

Result? All the functionality of the computer on your desk (other than long-term storage) could be put into a 3" cube of bb-sized spheres. And each one could be custom made to your specifications by picking which spheres are included in what configuration.

> You don't just spray silicon like a coat of paint!
> You need a perfect (~99.9999999999%) pure crystal.

Well, no, you don't. You just need pure crystals in the spots where the transistors are supposed to be. Perhaps it would be easier to create a grid of small pure crystals instead of a single perfect one. In fact, how about growing them chemically and then painting them on? You could even have P and N crystals made separately. I am not sure how to go about connecting them (monolayer deposition followed by CVD to fill the gaps perhaps?), but I think something like this could warrant some consideration.

> When you sandwich together different materials there
> is serious risk of cracking over time as they heat up and cool down.

How about pre-cracking them then? Build in natural fault lines into the circuits and let them break, leaving the pieces connected with stretchable metal wires.

> But how do you think timing will be impacted if one layer is typically hotter than another?

But you will know about this situation at design time. At steady state the temperature gradient is constant, and with proper cooling should remaing that way. In fact, it would be more stable than in a flat circuit because of the ball's greater thermal mass.

> How do you connect both layers together? You would
> need to grow metal posts *through* the new substrate layer.

No, you would grow the metal posts *while* you grow the substrate layer. Masking and etching should work just fine as long as the posts are wide enough to not suffer greatly from misalignment. You can also do inductive coupling.

> How is signal integrity impacted?

Why would it be? Signals travel on a spherical surface as easily as on a flat one. Look at Earth.

> How do you supply enough power?

Radially. Because the radial connections must already be thicker, they would offer a lower voltage drop, and by being able to route power directly to any point on the sphere you would be able to reduce the length of the power lines in the circuit.

> When something goes wrong and an entire batch of wafers
> is destroyed (at millions of dollars in cost)

And that's another problem. Shouldn't they be thinking really hard about how to make the process cheaper? Take a look at Ball Semiconductor, for instance. They claim a 5 day turnaround time! If you could do that with your bad chips, perhaps you wouldn't spend too much time "laser fixing" a particular bad wire. It would cost less to just restart the process.

> Ball Semiconductor

Hah! I knew it! :) /me does the happy dance.

That company also have discovered some even more interesting advantages of doing spherical circuits. It turns out that the balls can be made without a cleanroom, which is obviously an enormous expenditure. They also cite lower environmental cost, due to less silicon going to waste. And making an inductor coil on the surface for communication is way cool!

Ball Semiconductor

...spherical semiconductor circuits. Ball Technologies

Heh heh... good one. And it got modded up as Informative instead of Funny -- I love it when that happens!

Or maybe I just didn't poke around the ballsemi.com site enough to find the pictures of their 3-D wafer fab. :)

Your ideas are good, thinking out of the box, and check this out for thinking out of the box, spherical semiconductor circuits.

Ball Technologies

burnin

Just FYI, there's also a company called Ball Semiconductor, Inc. that has been developing various design, manufacturing and processing technologies revolving around spherical semiconductors.

Ball Semiconductor have at least as interesting a plan to deposit semiconductors on small spherical surfaces. They have some small gates working already.

Apparently Ball Semiconductor is the company making this product. What I had believed to be 3d modular architecture is a stand-alone product. The 3d sphere itself has the whole chip etched on it. Ball Semiconductor states that they can produce chips at a much higher rate .

This could all be propaganda. I am interested in the heat dissipation on these bad mamajamas though, so I emailed them to get some tech specs on how they deal with this.

l8rs

First guess at an URL was www.ballsemi.com.

You can find this yourself by searching "silicon spheres integrated circuits" on Google.

Ball Semiconductor have been working quietly away on a process which uses small, cheap spheres of silicon instead of expensive slices. Maybe that's where the ball bit comes from. I didn't think they were ready for production, but their fab is potentially very cheap to make once perfected. Vik :v)

Combine a little AFM with massively paralleled MEMS robotics, and you have your nanotech assembler.

Do it using balls instead of wafers, and you could (in theory) build an entire microfab into a rod, say 5ft high, 2" diameter. But I digress... ;-)

Personally, I prefer the ball semiconductor approach for microelectronics on curved surfaces. :-)

Or for antenna-type structures, go for EFAB.

...and then there's that microphone reported a while back.