(New Scientist) A hybrid transistor design that could mean faster and far less power-hungry wireless devices has been unveiled by researchers from IBM. Ghavam Shahidi and colleagues from IBM’s Watson Research Centre in New York claim the transistor could make wireless chips three times faster than current designs while using 80 per cent less power.
Transistors are the switches that control the flow of current through microchips. Improving the efficiency of these components translates into greater overall computing power. And reducing power consumption will extend battery life of mobile computing devices, such as cellphones and handheld computers.
Details of the advance were made at the Bipolar-BiCMOS Circuits conference in Toulouse, France, on Tuesday.
The researchers applied a technique normally used to make conventional CMOS (complimentary metal oxide semiconductor) microprocessor transistors more efficient to bipolar Silicon Germanium (SiGe) transistors. These are used in radio frequency chips, which enable the transmission of data through the air.
Bipolar transistors are used because they are good at amplifying low power radio signals but are typically much less efficient than ordinary microprocessor transistors.
Using a “silicon-on-insulator” layer to reduce the capacitance of the bipolar transistors, the IBM team improved the speed at which they can switch on and off.
The technique is already used to make the latest microprocessors more efficient. IBM says the new transistor design therefore paves the way for placing both SiGe and CMOS transistors on the same silicon-on-insulator wafer, making a faster and more efficient wireless-microprocessor hybrid.
Such a design could be incorporated into advanced wireless devices such as 3G mobile phones and used for wireless connectivity and video image processing. IBM says the new chip design could be implemented within five years.
Yoshio Nishi, a former director of research at Texas Instruments and now at Stanford University, told the New York Times the new design was “a really outstanding achievement”. He does not see any fundamental obstacles to manufacturing the chip.