AT THE moment, the transistors, diodes and capacitors that do the work in a computer’s microprocessor are carved out from chemical layers that have been deposited onto a silicon substrate. An alternative approach is to make the components separately out of nanowires (rods a thousandth of the thickness of a human hair in diameter), and then connect them up.
Such nanowires can be made in large quantities from various semiconducting materials and metals, using inexpensive chemical processes. Not only will they perform many of the roles of components in a traditional circuit—they can also do other useful things, such as act as light-emitting diodes. But because they are so small they are difficult to work with. Past attempts to build nanowires into successful circuits have depended on being able to place them precisely onto the silicon substrate of a chip and then connect them up in a process similar to the way that electrical equipment was assembled in the days of valves, wires and soldering irons, but on a much smaller scale. That is fine for the laboratory, but hardly suitable for the cost-conscious chipmaking business.
Now a group of researchers at Harvard University, led by Mariano Zimmler and Federico Capasso, have found a way around the problem.