New Research into Bending Light Could Lead to Light-Based Computing

cageymaru

Fully [H]
Joined
Apr 10, 2003
Messages
22,060
Engineers at Duke University have demonstrated a device that can bend light around tight corners with virtually no losses due to backscattering. The engineers have created a device that uses photonic crystals and the concept of topological insulators to allow light to pass on the surface while blocking it from traveling through the interior. This research could lead to photonic systems where electrons are replaced with light. For this to be true, photons must be able to traverse microscopic spaces such as in a microchip. Photonic systems will be faster than conventional microchips and have higher bandwidth.

Duke engineers had taken the original idea and devised a system that was 100 times smaller at just 35 micrometers long and 5.5 micrometers wide. The original concept used carefully controlled geometry to create the effect, but new research has done away with that requirement. The new research that was released on November 12th has yielded a "device that will work no matter its dimensions or geometry of the photons' path and photon transport is 'topologically protected,'" according to Mikhail Shalaev. "This means that even if there are minor defects in the photonic crystalline structure, the waveguide still works very well. It is not so sensitive to fabrication errors." "The researchers point out that their device also has a large operating bandwidth, is compatible with modern semiconductor fabrication technologies, and works at wavelengths currently used in telecommunications."

The smaller the device the better, but of course we're trying to minimize losses as well," said Wiktor Walasik, a postdoctoral associate in electrical and computer engineering at Duke. "There are a lot of people working to make an all-optical computing system possible. We're not there yet, but I think that's the direction we're going." Now the researchers are trying to make a waveguide that can be turned on or off at will--another important feature for all-optical photon-based technologies to ever become a reality.
 
I get kinda gloomy reading about stuff like this. I attended a network symposium in my university days, and one of the speakers there started his talk by describing 'gravity wells' - how much energy it takes to lift a one-ton spacecraft off of the surface of the Earth, off the surface of the Moon, Mars, Venus, etc., and then how much energy it takes to move that same ton from Earth's orbit to the point in space where it's captured by the next gravity well. Then he described that same power curve as processing power, and used it to describe how newer, better technologies are cut off from the market based on how much the market has already invested in another technology. The new technologies start off as little sounding rockets, but never build enough headway to grow into something bigger. This applies to things like operating systems (he talked about the BeOS, OS/2, and the NeXT), processors, networking systems, or even the fuel we use for our cars. He talked about optical computing and thorium reactors and even milk.

(Note - this was the first time I had heard of thorium reactors, and the fact that we don't have them still bugs me).

Change still happens, but he showed how it happens when a new technology is able to supersede the old technology by coming in from another market where they fill a niche need (think ARM).

We can build simple versions of optical processors now, but only governments can afford them and these simpler chips don't have a purpose, therefore they don't get the iterative growth cycle they need.
 
Last edited:
I get kinda gloomy reading about stuff like this. I attended a network symposium in my university days, and one of the speakers there started his talk by describing 'gravity wells' - how much energy it takes to lift a one-ton spacecraft off of the surface of the Earth, off the surface of the Moon, Mars, Venus, etc., and then how much energy it takes to move that same ton from Earth's orbit to the point in space where it's captured by the next gravity well. Then he described that same power curve as processing power, and used it to describe how newer, better technologies are cut off from the market based on how much the market has already invested in another technology. The new technologies start off as little sounding rockets, but never build enough headway to grow into something bigger. This applies to things like operating systems (he talked about the BeOS, OS/2, and the NeXT), processors, networking systems, or even the fuel we use for our cars. He talked about optical computing and thorium reactors and even milk.

(Note - this was the first time I had heard of thorium reactors, and the fact that we don't have them still bugs me).

Change still happens, but he showed how it happens when a new technology is able to supersede the old technology by coming in from another market where they fill a niche need (think ARM).

We can build simple versions of optical processors now, but only governments can afford them and these simpler chips don't have a purpose, therefore they don't get the iterative growth cycle they need.

Keep in mind Optical chips have a major downside: You have to carry a device on die that can generate the light, which is a major source of power draw. That makes Optical chips a no-go in all form factors up to and including laptops, limiting how much market penetration they can get.
 
At least we know there is a good deal of funding and research going into "post silicon" computing. Eventually something will pan out.
 
Yeah, the thing is, silicon can't get much further. Haven't you guys noticed that already? I mean, we used to get double the speed each generation. Now we are lucky if we get 25% faster GPUs and CPUs each gen. Sometimes it can be as little as 15% improvement. There's going to be a desperate need for a major technological breakthrough.
 
So can you bend all wave lengths ? can you bend heat ( radiated IR ) for instance ??
 
Do the big semi companies dabble in anything like this at all? AMD, Intel, TSMC, etc?
 
I think were about to destroy each other soon in catastrophic global war before we ever see this tech at home.
 
Back
Top