A Catch-22

As transistor dimensions continue to shrink, the channel length (the distance between the source and drain) will shrink as well—from today's 50 nm to the 10-nm lengths expected in the next decade. A shorter channel means faster transistor switching because the charge carriers have a shorter distance to travel. But at the same time, it becomes harder for the gate to maintain control over the channel. Instead, the voltage on the drain begins to lower the energy barrier in the channel, reducing the threshold voltage and freeing carriers to flow even when there is no voltage on the gate. This is what, in essence, is called the short-channel effect; it causes power consumption to rise and ultimately destroys transistor-switching action completely.

To keep the short-channel effect at bay, device designers must sacrifice some transistor performance and endure some increase in power consumption. Reducing the thickness of the depletion region under the gate by increasing the doping in the channel maintains gate control, but it also reduces carrier mobility (a measure of the speed with which carriers move through the semiconductor under the influence of an electric field). When the engineers also decrease the thickness of the silicon-dioxide gate insulation atop the channel to give the gate better control over the channel, the thinner oxide lets more current leak between the gate and the substrate, driving up power consumption.

Up to now, these solutions, which have also required ever more complex doping profiles, have worked well enough. But they are running out of steam. So engineers are coming up with other techniques to keep transistor performance up to par in future transistor generations [see illustration, "The Coming Thing in Transistors"].