Photonic integrated circuits could be used to study quantum transport effects that have no classical analogues. In nanoscale devices, for example, metallic atomic-sized contacts and single-molecule junctions, charge and energy transport is governed by quantum mechanics even at room temperature. Such devices therefore behave completely differently to macroscale ones and they are greatly affected by environmental noise and disorder.

The new nanophotonic processor is made up of an array of 88 Mach–Zehnder interferometers, 176 thermo-optic phase shifters and 52 optical modes integrated on a silicon photonics platform. Each phase shifter can be programmed on microsecond timescales, which allows the researchers to perform a large number of experiments in both quantum and classical linear optics.

"To control a processor of this size, we had to develop 240-channel, high-precision analogue electronics and arrays of high dynamic range photodetectors," says lead author of the new study Nicholas Harris, a PhD researcher in Dirk England's group at MIT. "Efficiently coupling light to and from this processor required two more photonic integrated circuits that convert between the small optical fields on the silicon chip and the larger optical fields compatible with standard telecommunications optical fibres."

Simulating static disorder in the processor

In their experiments, the researchers took advantage of the fact that, for light particles (photons), space and time are fundamentally linked by the speed of light, c. They were thus able to simulate how a photon evolves in time by considering the layers in their programmable nanophotonic processor as discrete steps in time. They introduced local disorder into the device by heating the silicon in specific sections of the platform. This alters its structure in these regions and light travels at a different speed through them.

"In one experiment, we varied the phase landscape in time, which allowed us to simulate how environmental noise affects the system. In another one, we held the phase landscape constant in time, but varied it randomly in the direction transverse to that in which light propagated. This allowed us to simulate static disorder."

The researchers say they were able to demonstrate a range of quantum transport regimes. The first was ballistic transport, in which light particles diffuse rapidly from their initial position. This is a signature of coherent quantum transport. The second is Anderson localization of the particles.

Anderson localization, and the "quantum Goldilocks regime"

Thanks to its wave-like properties, light can produce complex interference patterns (that look like water ripples on a lake) when it interacts with materials. Some materials interact so strongly with light that they modify its flow. One example of such a material is a photonic crystal, a periodic structure that acts as a “cage” for light.

For disordered structures, however, random light scattering and interference can produce a new effect called localization in which a light wave becomes "stuck" in closed paths inside the material, bouncing back and forth in complex looping paths called "modes". Photons cannot easily escape but instead travel around in circles inside the medium.

This effect (which works for all types of waves, be they acoustic or electromagnetic) was predicted in 1958 by Philip Anderson, who received the 1977 Nobel Prize in Physics for his discovery. If it could be made to occur inside a device that works by absorbing photons (a solar cell, for example), then the cell could be more efficient at converting light into electricity. In light-emitting diodes (LEDs) the opposite effect would be seen: photons would build up in the device and lasing would occur in certain regions as the photons started interacting with each other – something that would reduce the threshold for light emission.

The third type of quantum transport that the MIT/Elenion researchers observed was environmental-assisted transport in which particles that are initially stuck become unstuck when noise is added to the system. "We observed that there is an optimal strength of noise and too much or too little results in relatively reduced transport of a photon from its initial position to a target position. This is known as the 'quantum Goldilocks regime'," explains Harris.

Boson sampling experiments

As well as allowing the researchers to better understand these fundamental transport phenomena, the new optical processor might help them to perform "boson sampling" experiments. Boson sampling is an algorithm recently developed by Scott Aaronson and Alex Arkhipov, also at MIT, that could only run on a quantum device and its successful implementation would be the first definitive defeat for a classical computer.

A boson sampling machine would use single photons of light and optical circuits to carry out a fixed task. Unlike other quantum algorithms (such as those required for a universal quantum computer, for example), boson sampling could be implemented more easily, but until now it was difficult to generate the dozens of single photons required. The new nanophotonic processor might come into its own here since it could provide the large linear optical networks needed to accurately carry out specific mathematical operations, say Englund and colleagues.

The research is detailed in Nature Photonics doi:10.1038/nphoton.2017.95.