While we often hear about the limitations of pushing the boundaries of semiconductor manufacturing process technologies to meet the needs of higher and higher levels of computing performance, optical circuits are evolving as one potential way of addressing the challenge.

Announcements from two research groups on this subject caught my eye, one led by the Technical University of Munich whose work could pave the way for quantum sensors and transistors, and another from Stanford University whose work on photon diodes could influence the development of neuromorphic computing using light-based components.

While these are both still very much in the research phase, I think it’s worth highlighting these pieces of research to see where we could potentially address the needs of high-performance computing in many applications, including artificial intelligence (AI).

Quantum light sources leading to quantum sensors and transistors

A group of physicists from Germany, the U.S. and Japan lead by the Technical University of Munich (TUM) have succeeded in creating quantum light sources that pave the way for optical circuits.

By placing light sources precisely in atomically thin material layers with an accuracy of just a few nanometers, this could enable a multitude of applications in quantum technologies, from quantum sensors and transistors in smartphones through to new encryption technologies for data transmission.

Instead of having circuits in chips rely on electrons as the information carriers, the photons, transmitting information at the speed of light, could take on this task in optical circuits. The light sources are then connected with quantum fiber optic cables and detectors would form the basis of the building blocks for such new chips.

“This constitutes a first key step towards optical quantum computers,” according to Julian Klein, lead author of the study at TUM. “Because for future applications the light sources must be coupled with photon circuits, waveguides for example, in order to make light-based quantum calculations possible.” He added, “It is possible to integrate our quantum light sources very elegantly into photon circuits.”

The critical point here is the exact and precisely controllable placement of the light sources. It is possible to create quantum light sources in conventional three-dimensional materials such as diamond or silicon, but they cannot be precisely placed in these materials.

As detailed in a paper in Nature Communications, the physicists used a layer of the semiconductor molybdenum disulfide (MoS2) as the starting material, just three atoms thick. This was then irradiated with a helium ion beam focused on a surface area of less than one nanometer. In order to generate optically active defects, the desired quantum light sources, molybdenum or sulfur atoms are precisely hammered out of the layer. The imperfections are traps for so-called excitons, electron-hole pairs, which then emit the desired photons.

Quantum Light Source

Defects in thin molybdenum sulfide layers, generated by bombardment with helium ions, can serve as nano-light sources for quantum technologies. (Image: Christoph Hohmann / MCQST)

A key piece of equipment in this process was the new helium ion microscope at the Walter Schottky Institute’s Center for Nanotechnology and Nanomaterials, which can be used to irradiate such material with an unparalleled lateral resolution.

Together with theorists at TUM, the Max Planck Society, and the University of Bremen, the team developed a theoretical model describing the energy states observed at the imperfections. In the future, they also want to create more complex light source patterns, in lateral two-dimensional lattice structures for example, in order to also research multi-exciton phenomena or exotic material properties.

Alongside scientists from the Technical University of Munich, researchers from the Max Planck Institute for Quantum Optics (Garching), the University of Bremen, The State University of New York (Buffalo, USA) and the National Institute for Materials Science (Tsukuba, Japan) were also involved.

Photon diodes enabling next generation computing

Meanwhile researchers at Stanford University have developed a nanoscale photon diode that could bring us closer to faster, more energy-efficient computers and communications that replace electricity with light. Highlighting the work in a Nature Communications, the team said achieving compact, efficient photonic diodes is paramount to enabling next-generation computing, communication and even energy conversion technologies.

“Diodes are ubiquitous in modern electronics, from LEDs (light emitting diodes) to solar cells (essentially LEDs run in reverse) to integrated circuits for computing and communications,” said Jennifer Dionne, associate professor of materials science and engineering and senior author of the paper. Dionne and co-author Mark Lawrence, a postdoctoral scholar in materials science and engineering at Stanford, have designed the new photon diode and checked their design with computer simulations and calculations.

“One grand vision is to have an all-optical computer where electricity is replaced completely by light and photons drive all information processing,” Lawrence said. “The increased speed and bandwidth of light would enable faster solutions to some of the hardest scientific, mathematical and economic problems.”

The main challenges of a light-based diode are two-fold — one is making the light flow in just one (forward) direction overcoming what’s known as time-reversal symmetry; second, light is much more difficult to manipulate than electricity because it doesn’t have charge. Other researchers have previously tackled these challenges by running light through a polarizer — which makes the light waves oscillate in a uniform direction – and then through a crystalline material within a magnetic field, which rotates the polarization of light. Finally, another polarizer matched to that polarization ushers the light out with near-perfect transmission. If light is run through the device in the opposite direction, no light gets out.

Lawrence described the one-way action of this three-part setup, known as a Faraday isolator, as similar to taking a moving sidewalk between two doors, where the sidewalk plays the role of the magnetic field. Even if you tried to go backward through the last door, the sidewalk would usually prevent you from reaching the first door.

Using light beams instead of magnetic field to create rotation

In order to produce a strong enough rotation of the light polarization, these kinds of diodes must be relatively large — much too large to fit into consumer computers or smartphones. As an alternative, Dionne and Lawrence came up with a way of creating rotation in crystal using another light beam instead of a magnetic field. This beam is polarized so that its electrical field takes on a spiral motion which, in turn, generates rotating acoustic vibrations in the crystal that give it magnetic-like spinning abilities and enable more light to get out. To make the structure both small and efficient, the lab relied on expertise in manipulating and amplifying light with tiny nano-antennas and nanostructured materials called metasurfaces.

The researchers designed arrays of ultra-thin silicon disks that work in pairs to trap the light and enhance its spiralling motion until it finds its way out. This results in high transmission in the forward direction. When illuminated in the backwards direction, the acoustic vibrations spin in the opposite direction and help cancel out any light trying to exit. Theoretically, there is no limit to how small this system could be. For their simulations, they imagined structures as thin as 250 nanometers.

Influencing neuromorphic computing

The researchers are particularly interested in how their ideas might influence the development of brain-like computers, or neuromorphic computers. This goal will also require additional advances in other light-based components, such as nanoscale light sources and switches.

“Our nanophotonic devices may allow us to mimic how neurons compute — giving computing the same high interconnectivity and energy efficiency of the brain, but with much faster computing speeds,” Dionne said. Lawrence added, “We haven’t found the limits of classical or quantum optical computing and optical information processing. Someday we could have an all-optical chip that does everything electronics do and more.”