A major bottleneck to modern computing power is latency -- the time it takes for a CPU to communicate with other devices like a graphics chip, or memory. One possible solution is to switch from using an electrical bus to an optical one. In well-made glass, light can travel at nearly two-thirds its speed in the vacuum an astonishingly fast transfer speed. The key, though, is that you need something to emit a high intensity burst of light (a laser pulse).
Researchers at the University of California, Berkeley have completed a work that inches the computer industry closer to this dream. They've developed a new process that will allow laser-emitting nanopillars to be directly grown onto silicon at milder conditions than in past work.
Silicon itself is remarkably bad at emitting light, so researchers must use alternate semiconductors. One of the most promising candidates is so-called "III-V" semiconductors, which consist of one transition metal element, and one semiconducting element from the fifth group of the periodic table, like gallium.
But growing these elements onto silicon using traditional process technologies has been pretty much impossible, due to the mismatch complexities and high temperature constraints.
The study's lead author, Roger Chen [profile], a UC Berkeley graduate student in electrical engineering and computer sciences, explains [press release], "Growing III-V semiconductor films on silicon is like forcing two incongruent puzzle pieces together. It can be done, but the material gets damaged in the process."
Adds Connie Chang-Hasnain [profile], UC Berkeley professor of electrical engineering and computer sciences who served as the project's principle investigator; "Today's massive silicon electronics infrastructure is extremely difficult to change for both economic and technological reasons, so compatibility with silicon fabrication is critical. One problem is that growth of III-V semiconductors has traditionally involved high temperatures – 700 degrees Celsius or more – that would destroy the electronics. Meanwhile, other integration approaches have not been scalable."
The researchers developed a process to deposit nanopillars of indium gallium arsenide, a III-V semiconductor, at only 400 degrees Celsius. The metal-organic chemical vapor deposition method used is already in commercial use in the solar cell and LED industry.
The resulting nanopillar structure is hexagonal and amplifies an infrared (950 nm) laser signal. Spiral helically up the pillars, the laser is emitted at their ends. This laser-cavity mechanism is complex theoretically, but the bottom line is the researchers have created an on-chip laser that's grown using more affordable process.
States Professor Chang-Hasnain, "This is the first bottom-up integration of III-V nanolasers onto silicon chips using a growth process compatible with the CMOS (complementary metal oxide semiconductor) technology now used to make integrated circuits. This research has the potential to catalyze an optoelectronics revolution in computing, communications, displays and optical signal processing. In the future, we expect to improve the characteristics of these lasers and ultimately control them electronically for a powerful marriage between photonic and electronic devices."
The study on the work was published [abstract] in the prestigious peer-reviewed journal Nature Photonics. The work was funded by a grant from The Defense Advanced Research Projects Agency (DARPA) and a Department of Defense National Security Science and Engineering Faculty Fellowship.