Research project: The future beyond silicon

[Oveeri666]

Hi all,

Im doing a research project for HNC at college and my topic is about the next technology once the current silicon technology hits its limits.

So far i have identified various technologies which might be the "next big thing", these are:

Gallium arsenide - this technology has been around since the birth of silicon but had restrictions which made silicone a better material for the time.

DNA processors / computers - massive potential but is only good for a single use (at the moment).

Graphene

Optical

Memristors - a potential component for use in processors which could replace half of the transistors in a processor and each one has the ability to do the work of ten transistors, reducing physical size and thermal output.

Molybdenite

Carbon nanotubes

Part of my assignment is to gather information and opinions from various sources and to this end i am here hoping to get some of your views on these technologies and any other technologies i may not have come across yet.

Your input would be greatly appreciated.

Oveeri666

Ps, Feel free to correct any wrong information i have given as i havent looked into each of the technologies enough to know all the facts in detail.

[cyr]

Ps, Feel free to correct any wrong information i have given as i havent looked into each of the technologies enough to know all the facts in detail.

http://en.wikipedia.org/wiki/Silicone
http://en.wikipedia.org/wiki/Silicon

[fcb]

do you mean SILICON..

[Oveeri666]

yeah fixed that one.

cheers. 

[minime72706]

First of all, it's SILICON, not SILICONE. Silicone refers to the rubber-like polymers that include silicone in their chemical structure.

There's "quantum computers", but I suppose those might fit under "optical".
Organic processors were once a thing, though I'm not sure where they are now in terms of development. I remember them involving threatening bacteria with oblivion unless they calculate things for you
GaAs and other "three-five" semiconductors are in use now, but you won't ever see them used for general circuit design, including processors. The materials are more expensive and they're not even necessarily suitable for general-purpose circuits. GaAs designs would be very power-hungry. You'll basically see this stuff in microwave amplifiers (as discrete transistors!) and in highly-efficient solar panels.
Though silicon is still a major part of it, I would take a look at silicon-carbide. I know very little, but it's popped up a lot lately.

I am or at least I was a very green IC designer for MIT Lincoln Laboratory (laid off), so I at least have exposure to the field. They actually developed the 193nm wavelength photolithography systems that EVERYONE uses today. They tried to develop something newer, but apparently extremely high quality optical glass is hard to come-by, so they weren't able to get anything off the ground. Instead companies these days play tricks with their 193nm lithography systems to make them expose objects 10nm and smaller, which people once thought impossible.

[AG6QR]

Another to add to the list is supercooled superconducting Josephson Junctions.

I remember a lot of research being done on them in the 1980s, when they were predicted to be the future of high speed computing.  I haven't kept up with that field, so I'm not sure if they're still a decade away like they always were, or if they've been largely abandoned, or what.  The fact that I'm not hearing about them these days suggests that not a lot of development research is being done on them now.  Regardless, it's an interesting topic, if for no other reason than to illustrate how difficult it is to predict what the next big thing really will be.  I found a few references; there are plenty more on the web.

http://smithsonianchips.si.edu/augarten/p76.htm
http://ieeexplore.ieee.org/xpl/login.jsp?tp=&arnumber=796539&url=http%3A%2F%2Fieeexplore.ieee.org%2Fxpls%2Fabs_all.jsp%3Farnumber%3D796539
http://www.w2agz.com/Library/Superconductivity/Anacker,%20IBM%20Josephson%20Project%20IBMJ.Res.Dev.24-2-107-112.pdf

Here's an article from 2011, the most recent one I could easily find...

http://www.dwavesys.com/en/pressreleases.html#nature_2011

[penfold]

Its an interesting thing to consider because it seems that as different applications hit their limit in silicon they all seem to be finding their own material that fits the bill, high temperature devices are coming out in SiC and GaN, low wavelength stuff InAs, for microwave stuff there's a fair amount of SiGe mix and some rather interesting materials in the three-nitride family - so diversification is the next big thing

[Stonent]

Always a visionary, Seymour Cray had been exploring the use of gallium arsenide in creating a semiconductor faster than silicon. However, the costs and complexities of this material made it difficult for the company to support both the Cray-3™ and the Cray C90™ development efforts. In 1989, Cray Research spun off the Cray-3 project into a separate company, Cray Computer Corporation, headed by Seymour Cray and based in Colorado Springs, Colorado. Tragically, Seymour Cray died of injuries suffered in an auto accident in September 1996 at the age of 71.

The part where he starts talking is at 6m49s



If you're interested in the circuit technology used in these older supercomputers, that starts around 37 minutes into the video.

At 42 minutes in, he starts talking about GaAs used in the Cray 3.

He said the processors were mounted bare to the board and ground down to make them thinner. The die was 25 mils thick and they ground them down to 12 mils so they could get a higher board density.

[minime72706]

Oh god... I forgot about the GaAs-powered Cray supercomputer --- the thing basically needs to be drowned in Flourinert to keep it running.

[Stonent]

Oh god... I forgot about the GaAs-powered Cray supercomputer --- the thing basically needs to be drowned in Flourinert to keep it running.

Yep the world's most expensive loveseat. I used to have a telnet account on a Cray C90. It wasn't useful for much as it was outdated by the time I had access to it, but fun to say "I have access to a Cray"

There was a video a few year back where someone overclocked their computer to 5GHz or so and bathed it in supercooled Fluoirnert. Apparently Fluroinert isn't really designed to be used at subzero temperatures and turned into supercooled Jello and seized up the pump.

[IonizedGears]

Look into diamond substrates, they've made 81Ghz diamond transistors. They can run perfectly fine for hundreds of degrees or at much higher temps than silicon at least. The only thing holding them back is the ability to make large diamond wafers quickly and cheaply.

[G7PSK]

Look into diamond substrates, they've made 81Ghz diamond transistors. They can run perfectly fine for hundreds of degrees or at much higher temps than silicon at least. The only thing holding them back is the ability to make large diamond wafers quickly and cheaply.

That and they only work at elevated temperatures, work fine on Venus but on Earth you have to put diamond transistors in an oven and the bandgap is 5.5 eV.

[Oveeri666]

Thanks everyone for these insights, I'll be going over them thoroughly over the next week and they should help my project greatly. 

[ejeffrey]

* GaAs is the semiconductor of the future and always has been -- this joke has been around longer than I have been alive.  Same goes for the other III-V and II-VI semiconductors -- they have niche applications in microwave electronics but are too power hungry.  That means that even ignoring the cost if you made a GaAs CPU today it would probably have lower performance than a silicon CPU because you couldn't pack it tight enough.
* DNA -- Mostly seems like a fad caused by grant application buzzword bingo.  It might be useful for something.
* Organics -- organic semiconductors are terrible (low mobility).  They are useful because they are cheap in large volume -- you can basically paint them on.  That is why they are used in OLEDs and possibly organic PV, but they won't be competing with silicon for performance.
* Optical -- Optical processing is fast but low density because optical waveguides have to be much larger than metal contacts.  It is useful for applications like optical switching where you don't need high density logic and you want to avoid the optical->electrical->optical transduction.  You wouldn't want to make an optical CPU.
* memristors -- I don't know much about these.  They sound promising, and also like they might be relatively easy to integrate with silicon tech, which would be a huge advantage.
* Carbon nanotubes -- appear to be inferior to graphene
* Graphene -- Graphene is probably still the most promising novel semiconductor, but that is partly because it is also the newest.  So far it is most promising in the same sort of applications as GaAs -- high frequency analog discrete transistors or very small scale integration microwave devices.  I think that because it doesn't have a bandgap it never turns off.  I am not up-to-date on the graphene literature, but if that isn't solved it would make graphene integrated circuits a non-starter.
* Diamond -- Last I heard, it was still really hard to get the doping at the levels you needed.  Also, as I understand it requires a lot of power due to the high bandgap.  In principle that is OK because it can handle the high temperature and has insanely high thermal conductivity, but you need to make sure the rest of the system can handle the heat load.
* Superconducting logic.  Very fast and near zero on-chip power dissipation.  It requires cryogenic cooling.  More importantly, current feature sizes are huge and they are limited by the need for inductors which are hard to make small.

[con-f-use]

Quantum computing using qdots or supraconductivity.

[madshaman]

You might be interested in this: http://m.spectrum.ieee.org/nanoclast/semiconductors/nanotechnology/ibm-demonstrates-graphene-transistor-twice-as-fast-as-silicon