On June 10, 2000, hundreds of thousands of eager pedestrians traversed the newly opened London Millennium Footbridge. However, as more and more pedestrians filtered across, the bridge started to sway. Panicked, they shifted their weight in the opposite direction to counteract the oscillations. Yet, the pedestrians’ synchronous efforts only compounded the swaying, and the amplitude of the bridge’s oscillations increased. The problem became so severe that the Millennium Bridge had to be closed later in the day, only to be reopened two years later, fitted with a $6-million damping system designed to counteract this effect.
Extensive analysis led engineers to conclude that resonance, the phenomenon in which an external force with a particular frequency amplifies existing oscillations, was behind the dramatic movements of the Millennium Bridge. But resonance can be useful, especially when it comes to laser energetics. At Yale, a team of physicists from Professor Peter Rakich’s lab developed a new silicon-based laser that uses resonant sound waves to amplify light, representing a milestone in the field of silicon photonics.
What is silicon photonics?
The field of photonics emerged with the invention of the laser in 1960. In traditional lasers, light of a specific frequency is emitted by pumping atoms into an excited state, leading to “laser gain.” When the increase in energy reaches a certain threshold, lasing occurs. The next big developments in information technology were optical fibers, which transmit light signals through a glass core. Light can encode more information in its wave properties than electrons over the same period of time, with little loss over thousands of miles.
However, optical devices used to process optical signals are often bulky, spanning entire table tops. Today, silicon-based photonics offers the promise of miniaturizing these systems onto single chips. Rakich believes that silicon photonic systems can revolutionize optics similar to how transistors revolutionized the digital age by eliminating the need for rooms full of vacuum tubes. More importantly, silicon is a naturally abundant element with a massive preexisting processing infrastructure serving the $400 billion semiconductor industry. “If you’re building things out of silicon, as a friend of mine used to say, ‘You’re riding the silicon freight train,’” Rakich said. However, it has proved difficult to produce a laser based purely on silicon—due to certain intrinsic material properties, silicon is not predisposed to provide laser gain. Silicon optical waveguides, which are structures which guide light waves in the medium, suffer from significant dissipation of energy that could otherwise have contributed to laser gain.
Stimulated Brillouin scattering in silicon
The new silicon system uses resonantly-driven sound waves to provide the requisite laser gain. It parallels the Millennium Bridge phenomenon, with silicon as the bridge and light waves as the pedestrians. Just as pedestrians walking at a particular frequency caused the bridge to oscillate, input light, also known as “pump light,” generates elastic waves, commonly known as sound, when oscillating at the natural frequency of silicon. Here, the researchers saw an opportunity to exploit a phenomenon called Brillouin scattering, in which the resulting sound waves in turn interact with the light, moving some light energy into a lower-frequency output channel. The process is self-amplifying—the stronger the sound waves, the more sound-light interaction, and the more energy pouring into the output channel—to the point where the sufficient gain for lasing is achieved.
But Brillouin scattering had never been observed in silicon. It requires a “hybrid” waveguide that can confine both light and sound waves to the same region in silicon, enabling their Brillouin interaction. This is a difficult problem due to the structure of silicon chips, which have a semiconducting silicon layer atop a layer of silica–the oxidized form of silicon. Light waves naturally confine themselves to silicon, but sound waves spontaneously dissipate their energy down into the silica layer.
To address this problem, the Rakich group devised a “suspended” waveguide that encompasses a ridge along the center of the silicon layer to conduct light waves, as well as slots on a micrometer scale punched uniformly along the sides of the ridge. These form “hard boundaries” off which sound waves can bounce, forcing them to stay within the top silicon layer. The researchers then wrapped the linear waveguide back on itself, creating a “racetrack” for the waves to circulate and amplify, leading to enough laser gain to offset the energy loss and produce Brillouin lasing in silicon for the first time.
Unique Brillouin behavior
The first of its kind, the new system does not behave like a conventional Brillouin laser. Traditional Brillouin lasers absorb light noise from the input pump wave, so that the emitted light becomes spectrally purified. In this system, however, the emitted light was spectrally identical to the pump light. Instead, output sound became spectrally purified. For this new regime of Brillouin lasing, the researchers had to construct a new theoretical framework to understand it. “It was all very different, very new physics,” said Nils Otterstrom, lead author of the paper.
This system embodies original design concepts that are important for Brillouin lasing in general. For one, it exhibits forward Brillouin scattering, in which the output wave is emitted in the same direction as the pump wave, due to the hybrid nature of the waveguide. Conventionally, if one were to pump light into a Brillouin-active system, the output wave would “rebound” back out of the system, interfering with the input. Decoding such light output from the input requires devices called circulators, which have yet to be developed in silicon photonic circuits. Forward scattering sidesteps this problem—the new design concept makes integration of Brillouin lasers into chip-scale systems practical for the first time.
Secondly, properties of conventional Brillouin lasers are dictated by the material from which the waveguide is made. That is, for every new laser frequency required, researchers would have to make lasing occur in a new material. As evidenced from this study, this is hardly a trivial problem. In this system, a range of frequencies can be produced by simply varying the width separating the air slots on the waveguide. “You could say, ‘I want this oscillation to happen at these specific frequencies,’ and we can crank out a device that does that. So it’s a very versatile design,” Rakich said. The wide range of laser frequencies potentially accessible would enable even more fine tuning of silicon-based laser systems.
Thirdly, the system does not suffer from cascading, a phenomenon observed in conventional Brillouin lasers. Otterstrom compares cascading to transferring water from one bucket to another. A laser system involves transferring water, or energy in this analogy, from a pump bucket into the output bucket. Cascading occurs when the output bucket has a hole leaking into another bucket, generating a secondary Brillouin laser. “We have been able to plug the hole in the first bucket so we can get an optimal amount of power into it,” Otterstrom said. More power means a purer output laser.
The silicon Brillouin laser has important applications due to its pure sound output and unique lasing properties. Rakich and Otterstrom call sound waves the “unsung heroes of modern technology.” This is because the highly precise clocks that are integral in global positioning systems (GPS) and computers use sound waves to keep time. Spectrally clean sound output from a silicon chip would go one step further in miniaturizing these systems. Moreover, lasers are currently used in high-precision gyroscopes on aircraft, where rotation results in detectable phase changes in the light wave. “Maybe someday we will get to a state of refinement where devices on a chip like this could be cheaper, smaller, more efficient than a fiber-based device. But there is more work to be done,” Rakich said. One of the more immediately accessible improvements is to increase the energy transfer efficiency mediated by Brillouin scattering.
For Rakich, the most rewarding part of this result is not the promises the new silicon system offers, but the needs-driven scientific approach it demonstrates. The development of the silicon laser was motivated by thinking hard about the problem and determining the properties it demanded, before choosing the best material for the problem. “To me, what’s more exciting about this result is that we took something that by all conventional wisdom shouldn’t have the [Brillouin] effect, we’ve made a laser out of it, and we can [use this approach] for any material,” Rakich said.
In any case, by uniting photonics with semiconductor electronics, this new Brillouin laser provides a powerful route to customize on-chip light in silicon-based optical circuits. It provides a glimpse into a yet remote but possible future in which computers run on light.