Photonic computing is closer than ever.
Researchers at Duke University have developed a light-emitting device that can be switched on and off up to 90 billion times per second. This 90 GHz is roughly twice the speed of the fastest laser diodes in existence, potentially offering a whole new level of optoelectronic computing. Central to the technology are the infinitesimal crystal beads known as quantum dots.
The computing devices we're used to are based on shuttling electrons around via wires and switches. This has worked out pretty well through the history of computing, but electronics have limits, both in speed and in scale. Optoelectronics swap out electrons for pure light: photons. A computer based on information carried via photon is just by definition optimal, offering the literal fastest thing in the universe. Other advantages over electronic systems: less heat, less power, less noise, less information loss, less wear.
One potential of many is in eliminating or reducing what's known as the memory bottleneck. This is an increasingly damning problem having to do with standard computer architectures in which memory (data, machine instructions) is stored in a different location than the processors themselves. As those processors get faster, any actual performance improvements are hampered by having to, in a sense, send away for the needed data. This limitation is in some part a consequence of electronic speed limits.
In a computer architecture even only partially based on optics, where the system bus—which moves data between processors, memory, and input-output devices—passes along information via photons, we can imagine a pretty big improvement. This would also lay a foundation for quantum encryption schemes in which secure keys are transmitted using entangled photons.
But how does it work?
A quantum dot is indeed a tiny crystal semiconductor, but it's the "tiny" that really enables the desired quantum properties. We can imagine the crystal as a sort of dish or bowl, in this case around 6 nanometers across. Inside that bowl is an exciton, which is a quasiparticle pairing of an electron and an electron hole, which is itself a gap within the atomic structure of the underlying crystal where an electron should be but isn't. It's kind of like the electron is a prisoner in a cell pacing wall to wall. As the cell gets smaller and the walls closer together, the prisoner paces at increasing wavelengths (traversing less distance at the same speed). This is the gist of what's known as the particle-in-a-box model.
The result of this setup is that the particle's energy becomes a function of the geometric size of its cell (or bowl or well). Different diameters, different possible energy levels. These energy levels correspond to different wavelengths, which in turn correspond to different colors. This is why quantum dots are about to become very important in display technology.
The Duke optoelectronic scheme takes advantage of plasmonics, which is the general field of study concerned with the interactions between electromagnetic fields and free electrons in a metal. The researchers' idea is to make a sandwich of two very particular materials: a thin sheet of gold and a collection of silver nanoscale cubes. Trapped in the middle are our quantum dots. Fire a laser at the cubes and the result is that their surface electrons become energized, creating an electromagnetic field that excites the quantum dots, causing them to kick out photons corresponding to their various energy levels. The researchers were able to increase the rate of these spontaneous emission events by a factor of 880 (known as the Purcell factor), compared to the the emission rate of unexcited quantum dots. What's more, the emission events were 2,300 times brighter, with vastly more "fluorescence intensity."
"The most near term application is not for optical computing but for optical communication at short distances; for example, between processor cores," Maiken Mikkelsen, co-author of a new study in Nature Communications describing the work, told IEEE Spectrum. "Short distance communications really need good, cheap, and energy-efficient light sources that can be made directly on chip, which is not possible with lasers."