Graphene Transistor Speed Record

[CraigS]

Now here's one worthwhile noting … (even though not strictly Astronomy .. but for processing those AP images ...):

"Graphene transistor beats speed records"

Researchers in the US have developed a new way of making transistors from graphene – a sheet of carbon just one atom thick. The technique overcomes a major obstacle facing those who want graphene to replace silicon as the material of choice in future electronic devices. It has also been used to make the highest-speed graphene transistors ever.
…
The finished devices have the highest transconductance, (the inverse of resistance), value ever reported for such devices, of 1.27 mSµm–1. The transconductance of a transistor determines how well it performs.

Microwave measurements on the transistors also show they have a record-breaking intrinsic cut-off frequency in the range of 100–300 GHz, which is about twice as fast as the very best silicon MOSFETs of a similar size.

Finally, the mobility of the devices (which determines how fast electrons move through them) is about 20,000 cm2/Vs – a value that is around two orders of magnitude better than that of similarly sized commercial silicon transistors.
…
The team now plans to fabricate transistors with smaller gate lengths to push the cut-off frequency even higher – perhaps towards 1 THz. "We also hope to scale up the approach to fabricate big arrays of high-speed graphene transistors on large-area substrates, including flexible substrates," revealed Duan.

They are going on to fabricate transistors that push the cutoff frequency towards 1 THz !

Watch this space ... this one is going to be big … I mean ...really big !

Cheers

[renormalised (Carl)]

A few more years and if they bring this out commercially, it will revolutionise the computer industry, for sure!!!.

[bojan]

Don't hold your breath just yet...
The beta of those transistors are still very low.. lower than 1 actually. Those guys just demonstrated that Graphene could be used ....
But before those transistors become really useful. they need to have gain at least 10.
And, they must be cheaper to manufacture...
So, before that happens... we still have silicon and GaAs (for 300GHz and up).

What will REALLY revolutionise the computer industry one day will be parallel processing (in other words: software. We still do not have proper compilers that can produce reliable code for more that 4 cores).. otherwise with graphene processors we will only have 1000x bigger and equally unreliable and slow windows OS or equivalent, just what we have today.

[CraigS]

Originally Posted by bojan

Don't hold your breath just yet...
The beta of those transistors are still very low.. lower than 1 actually. Those guys just demonstrated that Graphene could be used ....
But before those transistors become really useful. they need to have gain at least 10.

The transconductance of a MOSFET decides its gain. In currently available technologies, as you reduce the size of the MOSFET, the transconductance is reduced because dopant impurity starts to impact on electron mobility.

So, seeing as native graphene devices have two orders of magnitude higher mobility than Si or GaAs devices, I'd say there's pretty reasonable potential for tradeoffs when it comes to gain.
Anyway, for digital switching applications all you need is on/off. I'm not so sure analogue applications are where the playing field is (hence the gain may not be the primary factor). Speed is the primary performance criteria for digital switching apps, and reduced heating (due to resistance).

And, they must be cheaper to manufacture...
So, before that happens... we still have silicon and GaAs (for 300GHz and up).

Agreed. But the graphene transistors are now in the ballpark and orders of magnitude better in switching performance up to 300GHz. (That limit will improve .. it always has .. Moore's Law).

What will REALLY revolutionise the computer industry one day will be parallel processing (in other words: software. We still do not have proper compilers that can produce reliable code for more that 4 cores)..

This software is required no matter what technology substrate you use.
The real revolution will be quantum computing. That's further out though .. and graphene substrate will be available in shorter timeframes. Not sure what they'll run quantum computing on .. it seems to be intrinsically linked with photonics developments.

otherwise with graphene processors we will only have 100x bigger and equally unreliable and slow windows OS or equivalent, just what we have today.

I'm not sure I get the 100x bigger ??

Cheers
PS: Good to hear from you on this one, Bojan.

[bojan]

By 100x bigger I meant 100x larger in amount of memory and speed requirements, both HD and RAM (yes, 1000x more features.. But at the end of the day, PC will again need couple of minutes to wake up, and the operator will still have to wait for computer, and not the other way around. And this will be only because the higher speed will allow larger OS.. End result is still the user frustration)

What I am trying to say, it is not the HW that is limiting the performance of today's computers - transputers (http://en.wikipedia.org/wiki/Transputer) were available on the market 20y ago, but noone knew how to use them properly (lack of compilers).. so they sort of disappeared, to be replaced by easier to use and faster processors.. faster but still just Von Neumann concept.

Perhaps the lack of competition is partly to blame (essentially only one OS and only one processor manufacturer dominating the market... I don't want to mention names here :-)

[CraigS]

Interestingly, it seems that the transputer was employed in the High Energy Transient Explorer spacecraft observatory, launched in 1996 and then again in 2000, which then led to the discovery of several Gamma Ray Bursts.

Also, as you indicate, the success of technology developments thesedays may be dependent on who is doing, (or funding), the research. It seems that companies like IBM are presently behind graphene transistor technology. I'll keep an eye out for others .. if Intel gets into it, I'd bet it would take off.

There is always a need for faster, more energy efficient hardware processor technologies. By this, I'm talking about the limitations stemming from the physics of the base materials used in the chips themselves. As I mentioned previously, it would seem that Si and GaAs MOSFETs are pushing the basic limitations in the physical parameters (ie: transconductance).

The transputer concept seems to have been about architectural changes in the design of microprocessor (CPU) chips (and also, as you say, matching this to the OS/software). The chips were still designed on the standard base (substrate) materials of the day. Graphene is a new base substrate material, which posesses unique abilities at the transistor level, which would ulitmately permit re-architecture of CPU design, along the same lines as the transputer concept, if anyone ever chooses to have a go at this.

The introduction of Graphene base transistors represents a pure shift in the technology basis of all chips … CPU, logic, etc. This shift is to a large certain extent, independent of CPU and software architectures which would re-evolve to exploit the basic physics hurdles, overcome at the electron etc level.

To me, this is the definition of a pure 'Technology Shift' and is very much distinct from 'Re-Architecture' or redesign.

Cheers & Rgds.

[CraigS]

Bojan (& others):

Here you go … a couple of updates on Graphene Technologies:
Graphene single-transistor amplifier is a first, Oct 21st:

Researchers from the University of California – Riverside and Rice University have made the first single-transistor amplifier from graphene. The device is better than conventional amplifiers thanks to the "ambipolar" nature of graphene, which means that it could find use in wireless and audio applications. It might also be used to design simpler analogue circuits for communications.
...
The team now plans to employ more advanced top-gate transistors, which will allow for higher gain because of the much smaller gate thickness. Balandin and Mohanram's group has already built such transistors with low flicker noise, something that is crucial for graphene transistors in any analogue and communication application. "Now we have to put them to work in amplifiers," added Balandin.

and then (Sept 23rd) ..

Graphene makes 'supercapacitor'

… develop(ed) the first DLC that contains vertically oriented high-surface-area graphene electrodes that aren't porous at all. The device pushes the operating frequency of an electric double layer capacitor to well beyond 5000 Hz, which is a factor of 105 better than commercial DLCs. What's more, it is six times smaller than low-voltage aluminium electrolytic capacitors and can be charged and discharged at high efficiency in times much shorter than 1 ms.

Graphene is also very closely related to the 'BuckeyBalls' (found in space this year .. that's just for the Astronomers out there )

Cheers

[Omaroo (Chris Malikoff)]

http://www.mit.edu/newsoffice/2011/g...tric-1007.html

Looks promising now in direct reference to sensors that like the infra-red...

Graphene, an exotic form of carbon consisting of sheets a single atom thick, exhibits a novel reaction to light, MIT researchers have found: Sparked by light’s energy, the material can produce electric current in unusual ways. The finding could lead to improvements in photodetectors and night-vision systems, and possibly to a new approach to generating electricity from sunlight.

This current-generating effect had been observed before, but researchers had incorrectly assumed it was due to a photovoltaic effect, says Pablo Jarillo-Herrero, an assistant professor of physics at MIT and senior author of a new paper published in the journal Science. The paper’s lead author is postdoc Nathaniel Gabor; co-authors include four MIT students, MIT physics professor Leonid Levitov and two researchers at the National Institute for Materials Science in Tsukuba, Japan.

Instead, the MIT researchers found that shining light on a sheet of graphene, treated so that it had two regions with different electrical properties, creates a temperature difference that, in turn, generates a current. Graphene heats inconsistently when illuminated by a laser, Jarillo-Herrero and his colleagues found: The material’s electrons, which carry current, are heated by the light, but the lattice of carbon nuclei that forms graphene’s backbone remains cool. It’s this difference in temperature within the material that produces the flow of electricity. This mechanism, dubbed a “hot-carrier” response, “is very unusual,” Jarillo-Herrero says.

Such differential heating has been observed before, but only under very special circumstances: either at ultralow temperatures (measured in thousandths of a degree above absolute zero), or when materials are blasted with intense energy from a high-power laser. This response in graphene, by contrast, occurs across a broad range of temperatures all the way up to room temperature, and with light no more intense than ordinary sunlight.

The reason for this unusual thermal response, Jarillo-Herrero says, is that graphene is, pound for pound, the strongest material known. In most materials, superheated electrons would transfer energy to the lattice around them. In the case of graphene, however, that’s exceedingly hard to do, since the material’s strength means it takes very high energy to vibrate its lattice of carbon nuclei — so very little of the electrons’ heat is transferred to that lattice.

Because this phenomenon is so new, Jarillo-Herrero says it is hard to know what its ultimate applications might be. “Our work is mostly fundamental physics,” he says, but adds that “many people believe that graphene could be used for a whole variety of applications.”

But there are already some suggestions, he says: Graphene “could be a good photodetector” because it produces current in a different way than other materials used to detect light. It also “can detect over a very wide energy range,” Jarillo-Herrero says. For example, it works very well in infrared light, which can be difficult for other detectors to handle. That could make it an important component of devices from night-vision systems to advanced detectors for new astronomical telescopes.

The new work suggests graphene could also find uses in detection of biologically important molecules, such as toxins, disease vectors or food contaminants, many of which give off infrared light when illuminated. And graphene, made of pure and abundant carbon, could be a much cheaper detector material than presently used semiconductors that often include rare, expensive elements.

The research also suggests graphene could be a very effective material for collecting solar energy, Jarillo-Herrero says, because it responds to a broad range of wavelengths; typical photovoltaic materials are limited to specific frequencies, or colors, of light. But more research will be needed, he says, adding, “It is still unclear if it could be used for efficient energy generation. It’s too early to tell.”

“This is the absolute infancy of graphene photodetectors,” Jarillo-Herrero says. “There are many factors that could make it better or faster,” which will now be the subject of further research.

Philip Kim, an associate professor of physics at Columbia University who was not involved in this research, says the work represents “extremely important progress toward optoelectric and energy-harvesting applications” based on graphene. He adds that because of this team’s work, “we now have better understanding of photo-generated hot electrons in graphene, excited by light.”

The research was supported by the Air Force Office of Scientific Research, along with grants from the National Science Foundation and the Packard Foundation.

[Poita (Peter)]

There seems to be an endless amount of graphene news over the last 5 years or so, always promising some major change to the way the world works, but no products yet.
I'm hoping some come to fruition sometime soon, I was so excited by graphene a few years ago, but gave up following it a while back. Hopefully some actual products are much closer to release now.

[Poita (Peter)]

Originally Posted by bojan

By 100x bigger I meant 100x larger in amount of memory and speed requirements, both HD and RAM (yes, 1000x more features.. But at the end of the day, PC will again need couple of minutes to wake up, and the operator will still have to wait for computer, and not the other way around. And this will be only because the higher speed will allow larger OS.. End result is still the user frustration)

What I am trying to say, it is not the HW that is limiting the performance of today's computers - transputers (http://en.wikipedia.org/wiki/Transputer) were available on the market 20y ago, but noone knew how to use them properly (lack of compilers).. so they sort of disappeared, to be replaced by easier to use and faster processors.. faster but still just Von Neumann concept.

Perhaps the lack of competition is partly to blame (essentially only one OS and only one processor manufacturer dominating the market... I don't want to mention names here :-)

I worked on the OS for the Atari Transputer project.. yes, Atari of all people designed a working Transputer all those years ago, and yes, the major challenge was harnessing the power.