Slashdot Mirror
Nanoscale Crystals May Be The Future of Silicon
Wire Tap writes: "With all the talk about how silicon is nearing its physical limits in computer systems, and other data processing applications, I found this article quite interesting. "[Brian] Korgel and chemical engineering professor Keith Johnston have found a method to make stable microscopic nanocrystals out of silicon that can emit light. And by toying with the size of the silicon nanostructures, they can change the color of the light that shines through. It can make essentially faster memory and generate less heat and radiate less power.""
Obviously, the semiconductor industry is especially interested in this are. IBM has done some work on "streching" the silicon domains in semiconductor devices to get them to behave differently. By doing this, they can get around some of the size limits that chip/device fabrication is currently running up against. I wish I could find the article I saw on this, but I think it was about a month ago.. What strikes me as really neat about this technology is the possibility for optical computing, rather than electron transport through semiconductors. With chemical and nanoscale design applied together, one could perhaps get different materials to emit different wavelenghts of light, opening up not just optical storage applications, but optical CPUs, memory, etc. I want to say that researchers have found some silicon/germanium crystals that do just this, based on the ratio of silicon to germanium. I think I read it in Science or Applied Physics Letters, but again, I'm not completely sure where I read this, but I know I've seen it somewhere. As a nitpicking aside, there is no such thing as microscopic nanocrystals. Nanocrystals are called such because one or more of their dimensions are on the nanoscale. For nanocrystalline silicon, all of its dimensions are nanoscale, and not microscale.
-When going for broke, go for Ithaca!
Actually, Gallium Arsenide (GaAs) semiconductors AND CPUs have been made - the older Cray Supercomputers at the NSA are nothing but GaAs computer architechure. The biggest problem with these devices was the heat output. To cool these systems, Cray had to put a recirculating ethylene glycol system into the computer, which ran the ethylene glycol OVER the computer chips and circuit boards to cool them. Since ethylene glycol is an electrical insulator there was no worry about short circuit with this system, provided it was kept absolutely dry. Filters were put into the recirculating unit to ensure dryness. It was such a neat visual effect, that Cray put in windows on the side of the casing so you could see the liquid cascade over the chips and boards and fall over the edge like a waterfall.
-When going for broke, go for Ithaca!
When talking about light memories and light computers etc. we must face the fact that one basic light component is missing:
The Light Transistor
We can store light (the light flipflop), we can transport light effictivly, overlay it and all that, but we've no light controlled light amplifier.
Currently we've to take the way around through electricity. Receive the light, transform it to an electrical signal, amplify the signal electrically and retransform the signal to light. This way we loose all the benefits light would have.
Until we find a way to amplify light, directly controlled by light there will be no light computers, and light memories will stay in labratory only.
If one day someone would discover in example some crystal that if shined upon from the side, will change it's up/down transperncy then nothing will stop the light computer, without that all other light components are for funny experiments only.
--
Karma 50, and all I got was this lousy T-Shirt.
Why stop there? Imagine a bucket of "display paint" that you could apply to an entire room (walls and ceiling), and then connect to some kind of output device that would let you configure the output parameters.
I don't think this work is all that new. People have been producing nanocrystals for 10 years or so (check out work by Alivisatos, Bawendi, others). The trick is to get the size distribution narrow, and then to get them into something that's more processible, like a semiconducting polymer, then you can follow simpler fabrication techniques to place them on circuit (and not have perform lithography in a titanium chamber at 450 degrees). It's true that these sorts of silicon nanocrystals would be better suited to biomedical applications, as the goodnanocrystals out there are made of toxic chemicals like CdS and CdSe, hence the line, "Putting cadmium into people doesn't sound so great,". But the article doesn't say anything about size distribution, or whether they've made electroluminescent structures, or just gotten photoluminescence. I just think it should be taken with a grain of salt.
"Now gluttony and exploitation serves eight!" - TV's Frank
Please note, though, that this has nothing to do with making faster Si-based MOSFETs (i.e. smaller transistors). If you're interested in that, look here, here (great story), or here to see just a handful of the ideas people have. With all of these things in development, don't expect anything to overtake Si as the dominant technology for a long, long time (~10 years, maybe even). ;-)