The researchers, from Hewlett-Packard Laboratories in Palo Alto, and University of California, Berkeley, claim this is the first time that optical modulation has been seen in a NIM at near infrared wavelengths.

An optical modulator is a device that encodes information in a beam of light by changing the intensity of the light to create trains of pulses. The team based their modulator on a "fishnet" NIM, because this structure is known to have both a negative electrical permittivity and a negative magnetic permeability for infrared light a certain wavelengths. As well as being responsible for the negative refraction of light, this “double resonance” also affects how much light is transmitted by the NIM.

To make their modulator, the team sandwiched an 80-nm layer of silicon between two 25-nm layers of silver. The silver layers were perforated with tiny rectangular holes using nanoimprint lithography and electron-beam lithography to form an square array of criss-crossed wires. The wires were separated by 320 nm in both directions; the wire widths were 220 nm in one direction and 110 nm in the perpendicular direction.

"These are basically resonant structures," explains Hewlett-Packard's Shih-Yuan Wang, who led the work. The magnetic resonance is related to the sandwich structure, while the electric resonance is caused by the crosshatching of wires.

The device was then studied by shining two laser beams on it. One beam was a 1700-nm wavelength infrared laser that was shone through the device to see how much light it would transmit. The second beam was a 532-nm wavelength visible laser that was used to control the transmission of the infrared light.

The visible laser was able to control infrared transmission by creating electron—hole pairs in the silicon layer, the presence of which causes a slight shift in the wavelength at which resonances occur. The result is a doubling of the amount of infrared light transmitted by the device when the visible laser is on.

Such a change is already enough to make a useable modulator, says Wang, but he believes that a much higher contrast is possible with further refinement of the structure. What's more, by repeating the experiment using very short laser pulses, the team found that the material could be switched in as little as 58 ps – a figure defined by the time it takes for the electrons and holes to respond to the laser light. This means that the modulator could be switched on and off very fast, at several tens of gigahertz in the current device and Wang believes that 100 GHz could be possible with device optimization. This is much faster than any commercial modulator available today.

Their modulator operates in the near infrared part of the spectrum, making it a good candidate for use in optical communications devices. "The closest competitor is the electro-absorption modulator," says Wang. "But with that you are limited to the bandgap of the semiconductor you are using. With our structure we can design to any wavelength if we change the dimensions of the criss-cross pattern."

There's plenty of work to do, however. The modulator is optically pumped in its existing form; a useable device would require electrical activation.