Photonic Breakthrough: A New Light-Emitting Silicon Eliminates Heat in PCB Design

With data centers increasingly overburdened, one of the most dogging issues electrical engineers face is heat. In electronic circuitry, data is typically transmitted through streams of electrons transiting through conductors. More data means more traveling electrons and more heat. 

But what if we could transmit data without the damaging effects of heat on electronic circuitry? New research out of the Eindhoven University of Technology suggests that silicon can emit photons to transmit data—all without the woes of heat, high energy consumption, and slow on-chip and chip-to-chip communication.

 

Researchers from Eindhoven University of Technology operate an optical setup to measure the light that is emitted. Image used courtesy of the Eindhoven University of Technology

 

Past Frustrations—And Possibility—With Light-Emitting Semiconductors

Photonics, commonly used in fiber optics, carries information not by electrons, but by photons of light. Researchers previously assumed that silicon couldn't emit light, so they turned to wide band-gap semiconductors that could, like gallium arsenide and indium phosphide. 

For instance, a consortium of European researchers experimented with gallium arsenide quantum photonic waveguide circuits while other EE researchers from the University of California Santa Barbara studied high-power indium phosphide photonic integrated circuits. The fundamental problem here is that neither of gallium arsenide nor indium phosphide plays well with silicon, and both are expensive in their own right.

Researchers from the Eindhoven University of Technology recognized that while silicon is still the material of choice for fabricating the vast majority of ICs today, it could be more useful to designers if it too could emit photons and thereby enhance data communication while eliminating heat problems.

 

The Breakthrough: Hexagonal Silicon-Germanium Finally Emits Light

This solution of light-emitting silicon eluded researchers for several years, however. The problems they encountered came back to the fact that silicon is an indirect bandgap semiconductor, which prevents it from emitting light. They then turned their sights to combining silicon with germanium into a hexagonal structure, hoping the end result would be a direct bandgap that could emit and transmit light.

In 2015, the Eindhoven University of Technology researchers published a paper demonstrating the use of a hexagonal shell made of gallium phosphide as a template for hexagonal silicon. While they successfully produced silicon in a hexagonal shell, the shell proved unable to transmit or produce light. 

Fast forward to this week, and many of the same researchers, led by Erik Bakkers, have managed to build an improved hexagonal silicon-germanium shell. The resulting silicon-germanium nanowires, when excited by an external laser, were, in fact, able to transmit light.

 

Light-conducting nanowires fashioned of the silicon-germanium shell. Image used courtesy of the Eindhoven University of Technology

 

According to Bakkers, the next step is creating the actual laser to excite the nanowires, which he says is now only a matter of time.

“By now we have realized optical properties, which are almost comparable to indium phosphide and gallium arsenide, and the materials' quality is steeply improving," Bakkers says. "If things run smoothly, we can create a silicon-based laser in 2020. This would enable a tight integration of optical functionality in the dominant electronics platform, which would break open prospects for on-chip optical communication and affordable chemical sensors based on spectroscopy.” 

 

This Metal Organic Vapor Phase Epitaxy (MOVPE) was used to grow silicon-germanium nanowires. Image used courtesy of the Eindhoven University of Technology

 

The team is now exploring ways to incorporate new nanowires into the dominant electronics platform—by which they mean, of course, silicon.

 

A Future of Designing Without Heat? 

Heat will likely always be a consideration for designers as our AAC contributor Amos Kingatua acknowledges in his articles on the major causes of high temperatures on PCBs and PCB thermal management techniques.

But can you imagine a world in which heat wasn't an issue with silicon data chips? What would this mean for the circuits you design? What possibilities would it open up? Share your thoughts in the comments below.