“Diamond transistor” clocks 120GHz

Diamonds are Forever! 

;-)

a single transistor operating at 120 GHz is nothing special.

At a 5ps (5 pico-second) switching speed, Intel's existing Silicon CMOS transistors can operate at 200 GHz if you only wanted to operate a single xtor.

Making an integrated device with a hundred million connected transistors that have to deal with wire-delay and clock skew, and having the entiire IC clock at 120GHz...that would be special.

2001: IBM demonstrate a 210 GHz SiGe heterojunction bipolar transistor.

2002: Scientists at the IBM's Burlington, Vt. labs say they have tweaked previous transistor designs with new SiGe bipolar technology to reach speeds of 350 GHz.

2005: Walid Hafez and Milton Feng at the University of Illinois at Urbana-Champaign developed a transistor that work at a frequency of 604 GHz. The device is made from the semiconductors indium phosphide and indium gallium arsenide.

2006: IBM (Georgia Tech) has unveiled a silicon germanium (SiGe) bipolar transistor that operate at frequencies above 500 GHz at 4.5 Kelvins (-451 degrees Fahrenheit), or at 350 GHz at room temperature.

2008: "Diamond transistor" clocks 120GHz...   :-(

I predict the end of the ceramic space heater...

Diamonds are a geek's best friend!

Great! Now I know what I will get for the wife...
"But it has diamonds, honey!"

Knowing that it's made from super-cheap materials, future chips made like this would be a dime a dozen.

Diamonds are a nerds best friend.

3.3 terahertz transistor was reached in 2001 by AMD / IBM team. Intel made it to 1THz a week before that. A French team hit the 1 THz in 2004, so 120GHz is just slow.

How do they know it will run that fast? They don't make equipment that can count that fast. It's kind of like when they say that a clock is accurate to 1 second every million years. How do they know?

It must have been hard to make

hat, coat taxi...

You gotta subscribe to view the article. :( Personal details etcetc. to subscribe.. Sounds amazing, so the question is: why not send a press release to everyone?

Search around a little and you will find 1THz transisors (here is one: http://www.semiconductor.net/article/CA6514491.html ).

So who cares about 120GHz.

A second every million years or 1*10^-6 second a year?

This is a big deal for many reasons. First, although diamonds are expensive, carbon is inexpensive and plentiful. With the right manufacturing processes, diamond based semiconductors are clean, green, and potentially inexpensive. Next, diamond based semiconductors would have dramatically improved heat tolerance. They won't cook when they get hot. Also, diamond is a fantastic conductor of heat. This will make cooling of chips much easier as hot spots will be reduced, and heat carried away faster and easier. Diamond based semiconductors also have the potential to dramatically reduce power requirements for a given level of compute capability. And diamonds are transparent in the frequencies most used in communications, thus making optical interconnects on chip much easier to integrate. Also, 120 gHz is not the limit for diamond. It's just a step on the path. Over time Diamond could become the new semiconductor material of choice.

Measurement. It's one of those science things, you know, banned by the church and all that? It's all the rage since Galileo.

You can take a measurement over a short period of time and interpolate the graph to a longer time. Oops, one of those science things again, like space travel.

It's very simple: they are comparing the length of a REAL second (the half-life of a Cesium atom) with the length of a second as the clock measures it. In your case, if a real second has a length of 1, the clock would have a length of 1 + 1/(number of seconds in a million years) ... at this rate, the clock would be off by 1 second every million years. I hope that explains your question.

Youre all missing the point, 
This is CARBON, not sillicon. 
Fair enough sillicon transistors with woefull leakages, opperating voltages and currents, pathetic efficiencies and assorted other technical aggonies, have exceeded the speed of this new little toy. 
But youre talking about a transition that is nearly as important as the thermonic/solid state revolution more than 50 years ago this is new science.

These poor bastards in the lab have to learn it all over again the same way Bardeen and Brattain did in 1947, But when carbon semiconductors are fully established, we could be looking at efficiencies far far better than what we currently have especialy in the UHF frequencies and higher brackets and also a huge increase in the level of precision obtainable compared to current technology since at this point carbon nanotechnology will probably start making a stand and slicing zero's off our current process technologies, forget 15 nanometers guys, were talking about 2 or 3.

The title of the article is misleading.

Diamond transistors are not important because they can('t) switch any faster. They can't. Switching speed is determined by other factors.

Because a carbon (crystallized carbon = diamond) nucleus has fewer protons than a silicon nucleus, it has 1 fewer shells, and thus the electrons are less shielded from the nucleus. This means the electrons are more tightly bound to the atom, and are less likely to be kicked off by thermal energy, an unwanted side effect.

That is, a diamond semiconductor would have smaller leakage currents and enable more energy efficient designs.

The only reason hitting 120Ghz is important is because there's no point in having a more energy efficient alternative if the alternative isn't very fast. 120Ghz isn't faster, but it is plenty fast for any application in which you might be interested in saving energy.