Researchers from Graz University of Technology, Harvard and the University of Texas at Austin have developed an innovative method for coupling light non-linearly. This opens up new possibilities for telecommunications and quantum technology.

Using a newly developed combination of a semiconductor layer and a meta surface, a research team comprising members from Graz University of Technology (TU Graz), Harvard and the University of Texas at Austin (UT Austin) has achieved a breakthrough in converting light into new frequencies. Their method transforms light around 72,000 times more efficiently than with previous materials. The researchers have published their findings in the journal Nature Nanotechnology.

As a medium for transmitting information, light is by no means an unknown quantity; its linear propagation, for example, enables communication in modern fibre-optic networks. However, this is insufficient for complex computational operations, quantum technology and cryptography, or frequency combs for high-precision measurement technology. Here, non-linear polarisation is required, enabling photons to interact with one another and exchange information.

Miniaturisation previously impossible

“Traditionally, extended crystalline structures such as lithium niobate are used to couple light waves non-linearly,” says Marcus Ossiander from the Institute of Experimental Physics at TU Graz. “However, this requires quite large volumes and a great deal of light energy, which has so far stood in the way of miniaturisation and many applications. Our method makes it possible to generate non-linear polarisation with light in smaller structures and with significantly less energy consumption.”

The starting point for the development of the “light converter” was an idea from the research group led by Seth Bank at the University of Texas at Austin. Using molecular beam epitaxy, they grew nanometre-scale semiconductor layers of gallium arsenide and aluminium gallium arsenide containing asymmetrically coupled quantum wells. These wells restrict the movement of electrons spatially to such an extent that quantised, atom-like states arise. Because the quantum wells are deliberately designed to be asymmetrical, the electrons move predominantly in one direction when exposed to light. In this artificially created one-way street, asymmetric or non-linear electron oscillations build up, enabling light waves to interact with one another extremely efficiently.

Solution for appropriate light incidence

However, this development from Texas still had one limitation: in order to harness the full effect of the semiconductor layers, the light had to propagate parallel to them, a configuration that is cumbersome in most applications. In technical implementations, the electrons would therefore not have moved along the artificial one-way street, and much of the non-linear effect would have been lost. So there was a promising material, but no suitable way to access its properties. Marcus Ossiander then came up with the crucial idea: a meta surface consisting of a precise chequerboard pattern of titanium dioxide pillars, each several hundred nanometres in size. This meta surface is placed directly on top of the semiconductor layer and deflects the light in such a way that it scatters along the one-way path. He put this idea into practice in collaboration with Federico Capasso’s research group at Harvard, in particular his colleague Pernille Fathi. As a result, they succeeded in utilising the properties of the semiconductor layer to excite non-linear light polarisation and even amplify it.

During the experimental tests, the researchers encountered a new phenomenon. If light strikes the component – comprising the metamaterial and the semiconductor layer – in a perfectly straight line, the optical fields cancel each other out due to geometric symmetry. However, tilting the sample by a mere 0.3 degrees broke this symmetry and enabled highly efficient non-linear polarisation using light. “We are achieving this conversion of light, which is 72,000 times more efficient, at wavelengths used in telecommunications,” says Marcus Ossiander. “Compared with the technology used over the past 30 to 40 years, this is a huge leap forward. This increased efficiency means not only smaller components in the future, but also lower energy consumption, for example when transmitting information from computer to computer within data centres. This is likely to be just as interesting for optics manufacturers specialising in integrated photonics as it is for companies that use this technology to process vast amounts of data.”