Feng scaling down microcavity lasers to achieve energy-efficient high-speed data transmission

The amount of data possessed by humanity has been growing exponentially, driven in part by soaring use of data-generating smart devices and sensors. But how can the physical infrastructure used for storing and transmitting such data keep pace with the dizzying rates of data growth? How can we continue to move these ever-expanding quantities of data in and out of data centers, and to and from the screens of faraway users?

“It’s a big, big network data transfer problem,” says HMNTL’s Milton Feng, dryly.

Feng, who is the Nick Holonyak, Jr., Endowed Chair Emeritus in Electrical & Computer Engineering, has just received a grant from the Army Research Office (ARO) that will enable him to tackle one important piece of that big, big problem. He’s going to look at ways to shrink the lasers used in data centers’ optical data links, with the goals of reducing their energy consumption and increasing their speed.

“Everybody is looking for solutions to reduce energy per bit,” he explains. It’s “required by all the data centers; all the major companies are working on the same problem.”

In the communication data link systems Feng studies, the transmission side is a “microcavity” laser—a laser in which the light is confined to a tiny area. Such lasers already offer important functionality while consuming the lowest possible amount of power. However, there are two respects in which improvements to these lasers could improve capacity.

“One is, reduce the power,” says Feng. “And the second thing is, increase the bandwidth.”

How will he do so? One aim is to reduce the laser threshold—meaning the minimum current required to make the laser lase—by a factor of ten. Feng plans to do so by reducing the size of the microcavity, so a big piece of the puzzle will be to figure out how small the cavity can be. A second aim will be to make the direct laser modulation speed—the speed at which analog signals, such as human speech, are converted to zeros and ones—7.5 times faster.

Today, optical links with an energy efficiency of about 20 picojoules per bit are in wide use. If Feng’s ARO project achieves its aims, microcavity lasers will be able to deliver high-speed data transmission that approaches an efficiency of 10 femtojoules per bit. (A picojoule is 10⁻¹² joule, and a femtojoule is 10⁻¹⁵ joule.

It isn’t an easy problem. Just one of the challenges that Feng must explore is heat. When you apply a current to a microcavity laser—which might be as little as 5 microns across—considerable resistor heat will be generated inside the tiny junction. “You raise the temperature maybe about 20 or 30 degrees higher than room temperature,” he says, “and you need to get rid of that heat. And when you do that, your device is actually operating less efficiently.”

Feng notes that this research area has emerged thanks to the pioneering work of Nick Holonyak, Jr. (co-inventor of the transistor laser and former member of Feng’s Ph.D. committee) and John Dallesasse (a professor in ECE and HMNTL) on oxide for VCSEL and transistor lasers—two types of microcavity lasers the new project will address. He feels his colleagues’ work in this area hasn’t been as widely celebrated as it deserves to be. “The world’s most important laser devices are oxide-aperture VCSEL, and that oxide for laser current and optical confinement was invented by Nick and Dallesasse at the UI,” he says. “And nobody talks about that!”

The ARO project is entitled “Cavity Physics and Tunneling Modulation of Microcavity Laser Toward 10 fJ/bit for Future Optical Networks.” It will run for three years.