Silicon Nanorods bend light in new direction

Swirling nanorods: an axicon lens made of silicon

Ultrathin coatings that arbitrarily manipulate the phase and polarization of electromagnetic waves have been created by researchers in the US. The coatings are made from silicon nanorods using a technique that is compatible with industrial processes such as photolithography. The researchers say that the coatings could be used in new types of optical components that are much less bulky than traditional lenses. The technique could even be used to bend light in ways not possible with conventional lenses.

Fermat's principle – the rule that light travels along the path of least time – says that electromagnetic waves travel along the path on which they accumulate the least phase. In a medium of higher refractive index, the wavelength shortens and so a wave accumulates more phase across the same distance. A wave therefore bends towards the normal to reduce the distance travelled in the medium and the phase accumulated.

Manipulative metasurface

In a conventional optical component such as a lens, phase accumulates continuously as the wave propagates and this determines the nature of the wave that emerges from the lens. However, if the phase of a wave could be changed discontinuously at a surface (called a metasurface), then the wave could, in principle, be manipulated in ways not possible with conventional optics.

While this is straightforward in theory, the challenge facing physicists is how to create such a phase discontinuity using real materials. In 2011 researchers at Harvard University led by Federico Capasso and Zeno Gaburro covered a surface with V-shaped gold antennas so that the surface could be used to introduce any desired phase shift to optical waves passing through it. While this allows the arbitrary redirection of visible light, there are two major problems with this approach. First, the metallic nature of the surface means that most of the visible light is lost as it travels through the surface. Second, thin layers of metal are very difficult to work with and incompatible with the complementary metal-oxide semiconductor (CMOS) process used to make modern electronic devices.

In the new research, Mark Brongersma and colleagues at Stanford University in California use lossless silicon optical antennas. When illuminated by a particular frequency of light (which can be selected by varying its diameter), the antenna will resonate strongly. This causes the light wave to pick up a phase shift that depends on the relative orientations of its polarization axes to the antenna. By appropriately tailoring the orientations and distances between the antennas, the surface can impart any desired phase shift to the light. This allowed the researchers to reproduce the functions of a bulk lens with a single layer of nanorods just 100 nm thick.

Axicons and Bessel beams

The team was able to create various types of "lenses" using this technique. These include traditional focusing lenses and an axicon. The latter is a specialized type of conical lens that transforms an ordinary laser beam into a Bessel beam – a ring-shaped beam used in optical tweezers and eye surgery.

Optics expert John Pendry of Imperial College London is impressed. "If anyone in the electronics or photonics game wanted to use a material, it would have to be silicon," he explains. "You can lay down silicon extremely flat and shape it very precisely. Metals are nowhere near silicon in terms of the precision and the control you can exert over them; so, if you can translate a technology like metasurfaces into a silicon environment, you're on to a real winner because you can hook on to this bandwagon that's been rolling for half a century now."

I think that Intel or other companies based on CMOS technology can implement such a metasurface now

Erez Hasman, Technion-Israel

In the experiment, the metasurfaces were fabricated by electron-beam lithography, but team member Erez Hasman, now at the Technion-Israel Institute of Technology in Haifa, says that commercial companies could produce large quantities using industrial processes such as photolithography. "I think that Intel or other companies based on CMOS technology can implement such a metasurface now," he says.

"The theoretical concept is not surprising at this point, but the fact that they built it and it works is interesting," agrees Andrea Alù, an expert on metasurfaces at the University of Texas at Austin. He looks forward to the development of optical components that are not possible with normal lenses. Hasman suggests that one of the first such uses might be to interface waveguides with free space. "In general, the modes of a laser resonator or a waveguide are very complex and different from the modes of free space," he says. Coupling the two together to allow signals to pass between them, he explains, is very difficult using a lens or a prism but should be no problem using the 2D metasurface.

The research is published in Science