WIRELESS technology is often credited with making us more productive. Now it looks like it could also improve the inner workings of our computers. Wireless transmission may become the most efficient method of moving data the length of tiny processors.
Microchips continue to become more powerful as designers find ways to cram greater numbers of transistors into tighter spaces. But in those cramped conditions the electronic signals sent down metal wires 鈥� the interconnects between transistors 鈥� can suffer interference, degrading performance. The answer is to go wireless, claim at the University of Pennsylvania, Philadelphia, and at the University of Texas in Austin.
Right now, most chip designers are considering systems based on so-called nanometallic waveguides to overcome interference. Here, a pulse of light is fired onto the solid waveguide, generating a ripple in the electrons that hug the material鈥檚 surface. These electrons whizz along the surface carrying information with them, without interference from signals in neighbouring waveguides.
Advertisement
But these waveguides cannot carry signals far enough, says Engheta. The electronic ripples degrade as they interact with electrons in the waveguide鈥檚 surface. Losses are such that waveguides are only suitable for sending data a few micrometres 鈥� too short to shuttle information from one side of a chip to another, he says.
Engheta and Al霉 have now shown that communication via air could be more effective at the microscale. 鈥淪ignal loss through the air follows an inverse square law,鈥� says Engheta. 鈥淭here are still losses, but compared with the alternative 鈥� exponential decay through a waveguide 鈥� it is not so bad.鈥�
Their calculations show that silver antennas, each 120 nanometres tall and 20 nanometres wide, could outperform solid links for transmission distances of 17 micrometres or more (Physical Review Letters, in press).
, a researcher in optical nanocircuitry at Stanford University in California, agrees in principle with the proposal. But where there are many on-chip nano-antennas, each broadcasting a distinct signal to a receiver elsewhere on the chip, there鈥檚 a high risk that crosstalk will degrade each signal and impair performance, he says.
鈥淪ilver nano-antennas could outperform metal links inside chips over some transmission distances鈥�
Engheta reckons there are solutions to that problem. Although his calculations involve simple dipole antennas, it would be possible to minimise crosstalk using directional nano-antennas, he says.
