Transistor Design - lowering necessary voltage possible with alternative materials?

[TurtleCrusher]

I've been reading up on low power IC operation and have noticed that even many ARM cpu's need in the area of 1.5-2.1V just to keep transistors switching properly. From my military training and other readings, essentially a transistor is a two way diode, with three contacts, one each end and a center contact between that facilitates current flow between collector and emitter when the necessary base voltage has been reached.

My question is, since germanium diodes have a .3V requirement (silicon has .7V) to activate the diode, would producing germanium transistors inside of IC's significantly lower voltage requirements (and power) and is it possible to manufacture?

Probably an elementary question, but it's been on my mind.

 

[GammaLaser]

Most logic chips today use a transistor type called MOSFET, the type of transistor you mention is called a Bipolar junction transistor (BJT). The way MOSFETs operate is quite different--they are composed of a drain, source, and a gate electrode which regulates the channel between the drain and source, and is separated from it by a thin gate oxide. The important characteristic which dictates the supply voltage of a MOSFET circuit is the "threshold voltage"--the required voltage at the gate to turn on the channel between the drain and source. And the main thing that is needed to reduce threshold voltage is either to reduce the oxide thickness at the gate or to use a higher dielectric material for the oxide. For the former we have pretty much made it as thin as we could because the oxide thickness is now on the order of several atoms thick for planar MOSFETs, and for the latter new materials (like hafnium oxide) have been developed which are called high-K dielectrics which are being used now instead of silicon dioxide. An indirect issue with decreasing threshold voltage is that you can increase leakage--which can mean you wind up increasing power, as well as manufacturing variation can lead to large percentage differences in the thresholds between transistors on the same chip.

So switching to Germanium in itself doesn't affect threshold but it can affect what gate dielectrics you might be able to use.

 

[A5]

Germanium is also far more expensive to produce. A lot of EEs would have a much easier life if Ge were as common and cheap as Si, but alas

 

[soccerballtux]

Germanium is also far more expensive to produce. A lot of EEs would have a much easier life if Ge were as common and cheap as Si, but alas

we could do it with our tech now its just that silicon has already had so much already invested in it there's not enough financial incentive to ditch it. We can grow 300mm+ ingots of silicon no problem, doing that with germanium would require a lot of investment that we don't really need to make yet.
Hm it would be nice to clock twice as fast though...

 

[dkozloski]

Germanium came first and had the market cornered. The switch to silicon was the big breakthrough that reduced cost and increased reliability because it produced a tougher, more heat resistant product.

 

[soccerballtux]

really? I didn't know that. Hm.