Can visible light ever be manipulated so that it bends the wrong way? If it could, a range of futuristic devices would be tantalizingly close to reality, such as a lens for imaging features smaller than the wavelength of light, or a shield to render objects invisible.

Several scientists have written off such 'negative refraction' in the visible range as practically impossible but a group is now claiming to have achieved it, spurring a debate about what constitutes true refraction.

Light bends in a specific way when it passes from one medium to another — an effect called refraction. Negative refraction describes a situation in which light bends the opposite way. It happens only if the direction in which the peaks and troughs travel along a light wave can be reversed relative to the direction in which the wave itself is travelling.

A material with a negative refractive index would focus light perfectly instead of dispersing it. This led John Pendry of Imperial College London to predict that a 'perfect' lens could be made, which would image features smaller than the wavelength of light. Some asserted that refraction could only ever have a positive value. But the debate was settled in 2003 when negative refraction was demonstrated for microwaves1,2 and later for infrared waves.

A straw in a glass of water seems disjointed because of refraction (left). But in this rough mock-up of what would happen if water had a negative refractive index (right), the effect is startling. The underside of the water's surface can be seen but not the bottom of the glass. For more accurate models, see ref. 4. Credit: E. SCHREMPP/SPL

Researchers achieved the effect with 'metamaterials' that had components of roughly the same size as the light's wavelength. More recently, Pendry used a metamaterial to bend light around an object to create an 'invisibility shield', also for microwaves3.

But achieving similar effects for visible light has seemed well out of reach. Radiation in the microwave and infrared ranges has wavelengths in the order of micrometres or centimetres, so the components of the material used to negatively refract them are also on this scale. But building something equivalent for visible light, with a wavelength of some 500 nanometres, is a huge challenge.

Now Jennifer Dionne and Henri Lezec, working in Harry Atwater's group at the California Institute of Technology in Pasadena, have unveiled a material that they say has a negative refractive index for visible light. Dionne presented the results on 11 January at Nanometa 2007, a conference on nanophotonics and metamaterials held in Seefeld, Austria, and the group has submitted them for publication.

Rather than try to create a material with components as small as the wavelength of visible light, theoreticians recently suggested taking advantage of electromagnetic waves called surface plasmons, created when light hits free electrons oscillating on the surface of a metal, to guide the light in the desired direction. This is what Dionne and Lezec have now done. Their device, called a waveguide, consists of the insulator silicon nitride sandwiched between two sheets of silver.

Light enters the device through a slit in the upper silver sheet. Once inside, the light wave couples with oscillating electrons in the silver to create a surface plasmon wave that travels along the metal's surface. But embedded in the silicon nitride is a gold-coated prism, with a gap between it and the upper silver sheet that is just 50 nanometres wide (see graphic). As the surface plasmon wave crosses this gap, it is refracted. Dionne says that she has detected light with wavelengths of 480–530 nm (blue-green) emerging from the device having undergone negative refraction. The refractive index reached as low as −5 (compared with +1.33, for light travelling from air into water).

figure 1

Figure 1

Passing the surface plasmons through the thin gap above the prism confines their movement, so only one mode of surface plasmon wave can get through. At certain wavelengths of light, the frequency of the surface plasmon wave is close to the frequency of the oscillating electrons within the bulk of the metal. In this case the surface plasmon wave and the oscillating electrons interact in such a way that the direction of travel of the wave's peaks and troughs is reversed, giving negative refraction.

For Dionne, the goal of “peeking round the corner” has been achieved. “It's like alchemy,” she says. “But it works.”

Others in the field are more cautious. Mark Stockman, a theoretician at Georgia State University in Atlanta, is concerned about the system's inefficiency, pointing out that only about 1% of the light gets through. Dionne emphasizes that enough light gets through to be detected directly and says she thinks improvements can be made.

And some are unconvinced that it offers true negative refraction. Allan Boardman, a theoretician from the University of Salford, UK, and Vladimir Shalaev from Purdue University in West Lafayette, Indiana, who are also trying to negatively refract visible light, argue that the experiment simply shows negative refraction of plasmons, rather than of light itself. “It's not negative refraction per se,” says Boardman. “They've got to qualify it a lot more.”

But others such as Nikolay Zheludev of the University of Southampton, UK, say this doesn't really matter, because the end result is the same. “If everything is correct, this is a grand claim,” says Zheludev. “Yes, they had negative refraction,” agrees metamaterials and plasmonics expert Eli Yablonovitch from the University of California, Los Angeles. “I don't see much controversy there.”

Pendry is also convinced, although he says he didn't expect to see the effect demonstrated so soon. “It is very impressive,” he says. “They've done it in a most spectacular way.”

Whether the approach counts as true negative refraction or not, to do anything useful with it will require turning the two-dimensional system into a three-dimensional device. Atwater envisages stacking a dense array of waveguides on end: “We have not done this yet, but at least this work illustrates the inherent possibility of doing so.”