Enlightenment in the computer chip

For decades, computer builders have dreamed of teaching their favorite semiconductor, silicon, to convert light and electricity directly into one another. Now the goal seems within reach.

Dwarfs Causeway: crystal columns made of germanium and silicon, about half a micrometer thick.  The picture was taken under an electron microscope and is therefore without color in the original.

AEvery year the physical journal awards Physics World, the membership magazine of the British Physical Society, the “Breakthrough of the Year” in the field of physics. Hardly noticed by the public, this title was awarded in December to a research project that could solve a permanent problem in the electronics industry: silicon, the otherwise so flexible semiconductor construction material, does not get along with light. To be more precise, it is very difficult to convert light into electricity – and vice versa.

Then why do photovoltaic systems made of silicon work? “This is only possible because the silicon layers are very thick,” explains Silvana Botti from the University of Jena. The amount of material can make up for the inefficiency to some extent. This no longer works with the tiny components of microelectronics. This is where silicon fails, and Botti wants to change that. Within the cooperation that achieved the award-winning breakthrough, she is responsible for the theoretical calculation of electronic material properties. And in fact, a solution is now emerging in the form of a new form of silicon-containing semiconductor crystals. They consist of hexagonal crystal columns with a diameter of up to one micrometer (millionths of a meter), which are reminiscent of microscopic basalt columns.

In order to understand why these “nanowires”, as they are called in technical jargon, should lead to the goal, one has to look at the current situation in the electronics world. There, optoelectronic components – i.e. photo, light and laser diodes – require other semiconductor materials, above all gallium arsenide. Unfortunately, it’s about as incompatible with silicon as a metric screw with an inch thread. For this reason, optoelectronic components cannot be integrated directly into silicon chips, and we are confronted with this technological brake block every day when dealing with computers – without even suspecting it.

Light is faster than electricity

“After the processors became faster and faster in the 1990s, they have been stuck with their clock frequency in the range of a maximum of three to four gigahertz since the early 2000s,” says Erik Bakkers from the University of Eindhoven, the head of the cooperation. This problem is not as famous as the impending end of Moore’s Law, which aptly describes the progress in miniaturization of electronic components for half a century. But as a brake on progress, it is now almost even more depressing.

The nanowires made of germanium and silicon are grown in this furnace.
The nanowires made of germanium and silicon are grown in this furnace.: Bild: Nando Harmsen, TU / e

The problem has two causes, says Bakkers: on the one hand, the losses due to the resistance of the microscopic lines. “In addition, the electrical resistance delays the flow of current.” Both problems could be solved if the longer cables in the processors in particular could be switched from electricity to light. But that would only work with optoelectronic components, which let the flowing bits transfer from their comparatively lame electron shuttles to a rapid transit network of light and back again. That would be almost a micro version of the technology that has long been established in the global fiber optic network. But it requires optoelectronics that can be integrated into silicon chips. The actual transistors would still work with electrons, because today’s high technology with its sometimes only a few nanometers (billionths of a meter) tiny structures has an advantage here.