Physicists in Austria and the US have raised nonlinear frequency conversion efficiency by three orders of magnitude — a factor of roughly 1000 — by pairing a band-structure-engineered multi-quantum-well heterostructure with a patterned titanium dioxide metasurface. The measured result, published in Nature Nanotechnology, exceeds previously reported values for comparable devices at near-infrared wavelengths, the band of direct relevance to telecommunications and quantum information systems.

The collaboration spans TU Graz's Institute of Experimental Physics, the Harvard John A. Paulson School of Engineering and Applied Sciences (SEAS) and the University of Texas at Austin. First author Pernille Undrum Fathi works in Federico Capasso's group at Harvard; TU Graz's Marcus Ossiander and UT Austin's Seth Bank, whose group grew the underlying material, are co-authors.

The device targets second-harmonic generation, the process in which two input photons combine into one photon with double the energy. Nonlinear conversion of this kind underpins frequency mixing, ultrafast signal processing, broadband and pulsed light sources, high-speed light modulation, and the generation of entangled photon pairs for quantum communication. Its practical bottleneck is well known to anyone specifying photonics components: efficient conversion demands high optical powers, because the handful of usable nonlinear crystals have complicated structures with weak nonlinearities.

"Only a handful of such nonlinear crystals exist and these have complicated structures with weak nonlinearities, which means they require high optical powers to make photons interact with each other," says Ossiander. "While new such crystals emerge from time to time, their structure is fixed, which dictates which light wavelengths they work at and how efficiently."

Two engineered layers

The team's approach replaces crystal hunting with material design, stacking two engineered structures.

The first layer is a semiconductor metamaterial grown by Bank's group at UT Austin from nanoscale layers of gallium arsenide and aluminium gallium arsenide. It contains multiple quantum wells that are asymmetrically coupled to each other. The wells confine electrons in one dimension while letting them move freely in the other two, restricting their energy levels to discrete quantum states. Because the wells couple asymmetrically, electrons mainly move in one direction when illuminated. That "one-way street" increases nonlinear electron oscillations, allowing light waves to interact efficiently enough to exceed naturally occurring nonlinearities.

The second layer, added by Ossiander, Fathi and colleagues including Capasso, is a metasurface of titanium dioxide pillars, each several hundred nanometres in size. It generates precisely the polarization needed to deflect incoming light so it scatters along the one-way street, further increasing light intensity inside the structure.

A 0.3° fix

The project nearly stalled at normal incidence. During the experiments, the team found that the designable second-order nonlinear susceptibility tensor element — the required incident polarization — differs from that of conventional materials. Shining light straight onto the sample produced light waves whose polarizations cancelled each other out completely, annihilating the enhancement.

"The new material sent us for an extra lap, however, because, during our experiments, we discovered that the designable second-order nonlinear susceptibility tensor element (the required incident polarization) is different to that of conventional materials," Fathi says.

Tilting the sample by just 0.3° broke the symmetry and solved the problem. That geometry delivered the reported three-order-of-magnitude improvement in effective nonlinear conversion over a non-patterned heterostructure.

The researchers are candid about the learning curve. "Working with new materials of course comes with additional uncertainty, and we spent a lot of time learning about and understanding the mechanisms of these new structures and how the particular nonlinear tensor elements of this material interact with the resonant modes introduced by the metasurface," Fathi says.

"The material we studied is very interesting but also complex," Ossiander adds, noting that the team believes further investigation of its fundamental properties will show how to make better use of the designer nonlinearity.

What to watch next

For R&D managers tracking photonics roadmaps, the near-term significance is twofold: the work points toward smaller, lower-power components for telecom and quantum communication, and the material platform itself is still improving. A separate publication in Optica by Bank's group at UT Austin already addresses how to improve the multi-quantum-well material further.

"Nonlinear optics are ubiquitous in modern science and technology, with examples including the generation of light at new frequencies (where lasers might be unavailable), all-optical signal processing and the generation of entangled photon pairs for quantum communication," Ossiander says. "Enhancing the fundamental processes enabling this technology will allow for much more efficient, compact devices and enable new measurements that would previously have been impossible."