Slashdot Mirror
Optical Microchip Breakthrough In Canada?
_J_ writes: "The Toronto Star has This Article on their Web site about method to "trap light." Since they call it a break-through to making an optical system it implies that light can be stored in a type of memory. I hope that this implies light-using logical gates." While this sounds like one more Holy Grail Found! announcement, the work that professors Ozin and John (mentioned in article) have done makes it sound like they're no slouches in the photonics or nanotech departments.
> Electrons don't travel at the speed of light,
> they have mass and therefore they travel
> slightly slower then the speed of light.
But it isn't the speed of the electrons that counts, it's the speed of the electrical signals, which is much faster. It's similar to a wave in a tank of water, which travels faster than any particular little blob of water travels.
The real advantages of optical technology are that they can be made much smaller because photons are bosons and electrons are fermions. Two identical bosons (such as photons) can be in the same place at the same time, but two identical fermions (such as electrons) cannot - this is the known as the Pauli exclusion principle.
So the Star's article is completely devoid of details - it's a newspaper ! I'll add a few more details so people can get as much information about this topic as they want. First and foremost the latest issue of Nature has an article entitled "Photonics: Opal appeal" specifically about this breakthrough (subscription required). The catch phrase used is a "three-dimensional photonic bandgap material". The team that's accomplished this is a bit more international then indicated so far, consisting of a Spanish team making the opal template, Geoff Ozin's group filling the lattices & then dissolving the template, Henry VanDriel's group performing the laser experiments, and Sajeev John's group providing the theory framework.
For those of you who just want pretty pictures, here are some images of the opals.
Here's the ultimate resource for photonic bandgap materials.
So that should give you more then enough to visit & read. Basically what these materials do is prevent propagation of light of a specific frequency in 3-dimensions. The 'bandgap' of the light can be controlled during the fabrication process allowing these things to block different frequencies. So you could imagine placing one of these materials into an optical fibre & selectively blocking one of the data streams but allowing all others to pass through unimpeded. The current breakthrough is twofold, first these aren't imaginary, they've been made & tested and they aren't decades removed from insertion into optical networks, they're months or years from it, second, this is the first example of a 3D PBG material, previous versions have generally been 2D. One of the neater experiments performed involved putting liquid crystals into the opal holes & then by putting an electric field across the liquid crystals, controlling the transmission through the crystal. A variable transmission photonic bandgap device. Light is fast, electrons are slow, an all optical network would be blazingly fast & these devices bring us a step closer to making that happen.
CJM
Material research has a strong component of wishful thinking and future projections. So many things don't work out because of a few insurmountable details. You need strong sources of motivation to pursue the dire road to success.
In their reasoning and justification of their work, these guys live at least 10 years into the future all the time. The referenced article was probably written by someone who took all their statements at face value. It looks to me as if they still have a long way to go. That's not meant to diminish their merits - these scientist are certainly top notch researchers and their results are truly very impressive. I just don't think they have delivered an imminent disruptive technology.
It's commonly accepted that the existence of a laboratory setup does not guarantee the technological and economical viability of any particular solution in the real world. I would start preparing for an imminent disruptive technology if a successful prototype system did exist. Yet, I don't have the feeling that there are even useful laboratory setups of the presented kinds of photonic devices. It rather looks like promising basic research.
As for the all photonics claim, I think the notion should be scaled down a little to be less prone to misunderstanding. To many people, it sounds like all photons, no electrons. I don't believe there is such a thing within our technological reach. Photons are bosons and interact extremelyweakly. That's not a very good basis for a computing device. Fortunately, photons can be converted into excited states of electrons which are fermions, interact in many ways, and can be used to produce logic gates.
That leaves us with a possible extension of the present use of photonic devices from lines of communication between nodes on a network to nano-lines of communication between old-fashioned electronic gates. And that's certainly not going to happen very soon. So, sorry my friends, no reason to get all excited.
A quick physics lesson. Many people here seem to think it's the speed of the the particle that is important: it isn't it's the ability to change the frequency and amplitude in a given ammount of time. These changes are what carry data.
electron drift: in average high conducting wire and given a good sized (120v) voltage, this speed is roughly 1m/10min. Not exactly something to transfer data with eh?
EM pulse speed in a wire: 2.997 * 10^8 m/s > EMPS > 2.997 * 10^7 m/s. The frequency can be changed quite easily and quickly
Photon speed: Depends on the medum, but 2.997*10^8 >= PS > 2.997 * 10^7 m/s (note this low is an estimate, it might go down to 10^6, but definetly not lower. The frequency of group of photons can change much easier and quicker than that of a EMPS caused by a series of electrons lollygagging in a wire.
This oscillation is what gives them the data transfer speed. This isn't quantum physics, it is taught in the second course of intro physics (not conceptual, but actual) in colleg. Also known as the first E&M course
Well, I'd guess that there will be a lot smaller heat dissipation. Correct me on this, but I assume that since they have no mass (only energy), they have 0 resistance.
:-/
So, the term "cool computer" will probably take a whole new meaning I guess.
However, I am no that excited on the subject. For one thing, I certainly do not posess such knowledge to question their theories.
What worries me, though, is that I kind of expect to be quite a few years before we could get our hands on one of those thingies. Think of the economics.
Suppose that in a month's time Mr. Ozin somes out and says "I've got a processor ready, architecture, ISA, layout, blueprints, the whole lot. Along with exquisite details on the manufacturing process!" (ok, it's not his job to design the processor, but let's say that somehow he got one ready from some processor designer at his uni)
For one thing, all the major corps will jump on it immediately (IBM for example). But the manufacturing process will be a new one and it's gonna be bloody expensive to make them and not the most efficient.
Another reason they're gonna be *really* expensive, even if the manufacturing process is just 5% more expensive than current practices: the corps so far have spent billions on investments in both product development and the respective manufacturing infrastructures. And they will want to milk that cow first on us and THEN, in a few years' time, introduce the optical chips as the high-prestige ones with equivalent prices...
No need to rush. Even a company rushing to beat the competition (take AMD for example, my favourite) will be held back a bit. No manufacturer is gonna make such big jumps.
A bit off-topic, but think about it. I wouldn't expect this technology to become mainstream anytime soon (btw, does the word "military" ring a bell??? I'm sure they'll want it first)
Trian
I'm no longer fed up with MS Windows: I go rid of them
I'm dissapointed with the lack of technical detail
in the article. I'm still trying to figure out
what is so novel about this. There has been an
aweful lot of work done for years now on trapping and guiding light. The big issue is efficiency.
The most promising technology I have seen for
photonic computing is guiding along defects of a photonic band-gap in a photonic crystal. This is
lossless guiding!!! Thats right, no photons can
escape! This research is lead by Joannopoulos at MIT http://ab-initio.mit.edu/photons Pretty
interesting possibilities since a photonic crystal
restricts photons of a given wavelength range from
propagating throught the material. A defect in
the 'crystal' allows the forbidden light to be
guided along the defect without leaking into the
bulk. Light can even be guided around right
angles without loss.
So we have the pipes, now we need the light
equivolent of transisters. But thats coming.
Jeremy
Oooh, I'm impressed. Slashdot already has links to the homepages of the two main subjects of the story of interest. Within which details of what is likely being talked of in the Toronto Star article can be found. I wish I had noticed that before I did all that searching.
Anyways, you'll notice that the publications start back in the early 90's. The 'new' thing they've discovered together might be what is talked about here, and is more clearly described here and here (Sajeev John's page contained links to this stuff...).
It's just a new way of making something that's been researched for the past 10 years, photonic band gap materials.
I haven't seen anything yet to tell us if this is such a better way of making this class of material that it counts as a 'revolution'. We have to find someone who knows a lot more about the current state of the art in creating photonic band gap materials and get this person to analyze this new method and it's results, to tell us if it's a significant advance, or what it's advantages are.
AKA: More peer review please.
I was always under the impression that the big advantage light would have over electricity would be in the size of the circuits.
With current chip technology, people have estimated all sorts of physical limits to how small we can make chips because of interference and such. Two wires (or etched copper or whatever) have to be physically seperated - but you can have two beams of light cross at a point and it wouldn't affect either "wire." In fact, it would seem that you could have two photon channels in completely oposity directions, but sharing the same space, and it would still be alright.
The advantage would come from being able to make insanely small chips, or chips the size we have now with a LOT more stuff on 'em.
--Me
I have a sig, and this statement is false.
I just wanted to address a couple of issues that seem to come up repeatedly (and sometimes incorrectly).
Heat: It's not obvious that optical computers would not have the heating problems the electon-based ones have. Sure, it wouldn't be based on the same mechanism (resistance), but you still have the problem of absorption. The same process by which the sun heats up your car in the afternoon would be a problem here.
Any time you shine light through something, some of it is transmitted, some is scattered and some is absorbed. The last two will cause signal losses and absorption will cause heating.
Heating may not be the biggest hurdle, but it will still be an issue.
Electron vs Photon speed: As a number of people have pointed out, wires to not carry signal at the speed of the electrons. A good (medium level) analogy to understand this is marbles in a plastic tube.
Let's say I have 100ft of plastic tubing full of marbles. We decide that every second, I'm gonna push a marble in my end (1) or I'm not (0). That's a 1bps data rate. Now, the speed at which the data travels is 100ft devided by the time between when I push on a marble and when one falls out the other end. Obviously, that's gonna be pretty fast.
The point is that the bit gets from one end of the tube very quickly even though any given marble will take a long time to get from one end to the other. Similarly, the electrons can carry information faster than they actually move.
(Disclaimer: This analogy is correct only in the sense of this last paragraph. I am not claiming otherwise)
This article was pretty sparse on technical details...all it said that there was some kind of silicon material coating microscopic bubbles in opals. So is the way that they store a piece of information by trapping the little photons in the bubbles, where they bounce around a few hundred trillion times, until they are allowed to go free?
I feel sorry for those poor photons, trapped in their little opal bubble cages.
On the other hand, if they ever built a server out of these...we could /. it. We haven't slashdotted a precious stone yet.
Hopefully I didn't put any [] around my words.
Also, since photons do not posses charge, they can not be interfered with by any kind of static electricity, magnetic fields, etc. Their signal stays truer.
Hopefully I didn't put any [] around my words.