backtop


Print 16 comment(s) - last by bkslopper.. on Sep 2 at 1:26 PM


Plasmon Laser  (Source: Courtesy of Xiang Zhang Lab, UC Berkeley)
Breakthrough will advance optical computing

Researchers from the University of California, Berkeley have reached a new milestone by creating the world's smallest semiconductor laser. The tiny laser can generate visible light in a space smaller than a single protein molecule.

The new laser breakthrough may one day usher in a new era in computing power by providing CPU makers with the ability to use light rather than electronic circuitry in processors. The key breakthrough was a method that the researchers devised to squeeze the light into a space smaller than its wavelength and keep the light from dissipating as it moved along.

"This work shatters traditional notions of laser limits, and makes a major advance toward applications in the biomedical, communications and computing fields," said Xiang Zhang, professor of mechanical engineering and director of UC Berkeley's Nanoscale Science and Engineering Center, which is funded by the National Science Foundation (NSF), and head of the research team behind this work.

The scientists say that the breakthrough will help usher in innovations like nanolasers that can probe and manipulate DNA molecules, optics based telecommunications and optical computing. Traditionally it is accepted that light can be compressed into a space smaller than half the size of its wavelength. Researchers have been able to compress light down to a couple nanometers by binding it to electrons that oscillate collectively along the surface of metals, otherwise known as plasmons.

Zhang and his team improved on this technique by pairing a cadmium sulfide nanowire with a silver surface separated by an insulating gap only 5nm wide or about the size of a protein molecule. The structure is able to store the light within an area 20 times smaller than the wavelength of the light. The light energy is reportedly stored mostly in the insulating gap between the wire and the silver surface loss is diminished significantly.

"When you are working at such small scales, you do not have much space to play around with," said Rupert Oulton, the research associate in Zhang's lab who first theorized this approach last year and the study's co-lead author. "In our design, the nanowire acts as both a confinement mechanism and an amplifier. It's pulling double duty."

Trapping and sustaining light in the very tight quarters creates conditions where the interaction of light and matter is strongly altered. The researchers say the sign of this altered interaction is a six-fold increase in the emission rate of light in the 5nm gap.

"What is particularly exciting about the plasmonic lasers we demonstrated here is that they are solid state and fully compatible with semiconductor manufacturing, so they can be electrically pumped and fully integrated at chip-scale," said Volker Sorger, a Ph.D. student in Zhang's lab and study co-lead author.



Comments     Threshold


This article is over a month old, voting and posting comments is disabled

I thought this was a dead issue...
By mmcdonalataocdotgov on 9/1/2009 11:20:27 AM , Rating: 4
quote:
The new laser breakthrough may one day usher in a new era in computing power by providing CPU makers with the ability to use light rather than electronic circuitry in processors.


What is the point of using light to transmit on a mobo when the time it takes to turn digital signals into light and convert them back again takes longer than just transmitting the digital signal across the mobo?

The issue with light is that you can mux many channels into a smaller path and thus increase bandwidth across great distances. I don't think this is an issue inside a computer case. Great for fiber networks, though.




RE: I thought this was a dead issue...
By amanojaku on 9/1/2009 11:37:21 AM , Rating: 2
The article doesn't say this will be used on the motherboard. This will be used within the CPU to replace silicon-based transistors. If that works, it's possible RAM and other components will benefit from this, as well.

You bring up a good point, though. Current CPUs are faster than their RAM by 2-4x, if I'm not mistaken. This will just increase the CPU/RAM bottleneck, making . I wonder what strategies will be used to move the signals between components.


RE: I thought this was a dead issue...
By JediJeb on 9/1/2009 5:10:10 PM , Rating: 2
I think this will yield a big advantage in lower heat and higher speeds if it solves the problem with electron migration within the CPU. Pushing more voltage to gain stability at higher speeds is like putting more pressure on a water pipe, sooner or later the electrons leak out and that is what causes CPUs to fail. If this can eliminate that problem then I wonder what the limit on speeds will be.


By niravadesai on 9/2/2009 6:33:02 AM , Rating: 2
Chip to chip communications are now becoming channel limited as the maximum data rate that can be supported right now is 10 Gbps .. The projected need in the next 10 years is for up to 70 Gbps of data rate for chip to chip communications. Optical interconnects are the only technology which can support these data rates. Check out these links on prototype optical interconnect ICs by Intel ..

http://www.eetimes.com/news/latest/showArticle.jht...

http://www.eetimes.com/news/design/showArticle.jht...


RE: I thought this was a dead issue...
By 007silent on 9/2/2009 6:55:58 AM , Rating: 2
quote:
If this can eliminate that problem then I wonder what the limit on speeds will be.

I've studied a couple of articles relating to optronics (digital processing using photons). The research field that is currently making progress on this is called photonics.

A research letter in Nature Photonics features work Intel engineers did on something called Avalanche Photo Diodes, I believe they want to use these to convert light signals to electric current (presumably before entering the socket). They achieved 240 GHz.

The theoretical switching speed maximum of copper wire is cited by some sources as 10 GHz. Comparatively the max switch speed of optic fibre is 10 Tera-hertz and I've heard talk of a debate for 100 THz max. But the theories underlying these predictions are not properly defined. I believe the switching speed is a function of the inertia of the carrier medium. Which explains why electrons with their greater mass propagate slower than photons.

I fully believe we'll reach 3 THz cpus one day with optronics. This is not even factoring in the possibility of multiplexing: if the logic gates can be designed to switch with frequency dependence - meaning allowing or disallowing light of a specific 'colour' independently of another signal passing through the gate at that time. Giving rise to a better flavour of simultaneous multi-threading.


By 007silent on 9/2/2009 7:08:54 AM , Rating: 2
Well what do you know? I knew that is what they were researching APDs for, the above posted link confirms it.


By JediJeb on 9/2/2009 10:51:46 AM , Rating: 2
quote:
This is not even factoring in the possibility of multiplexing: if the logic gates can be designed to switch with frequency dependence - meaning allowing or disallowing light of a specific 'colour' independently of another signal passing through the gate at that time. Giving rise to a better flavour of simultaneous multi-threading.


That is interesting. Depending on how narrow the wavelength of the filtering would be then the number of concurrent data streams could be tremendous. If you can separate at a level of one nanometer then that would give 300 channels just in the visible range of the spectrum. I haven't kept up on my physics ( I am a chemist by trade) so I don't know how practical that level of filtering is on a nanoscale. On macroscale though we have old instruments that easily have resolution of 2nm. It will be interesting to watch the optical computing developments.


By bkslopper on 9/2/2009 1:26:26 PM , Rating: 2
That's simple. Intel will roll out their new QuickER Path Interface. =P


"When an individual makes a copy of a song for himself, I suppose we can say he stole a song." -- Sony BMG attorney Jennifer Pariser











botimage
Copyright 2014 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki