Engineers have been trying to get silicon to emit light since the early days of solid-state displays and optical communications, in hopes of integrating optical and logic circuits on a single silicon chip. Doing so would eliminate the need for a separate device, typically composed of some form of gallium arsenide, to generate light, thus reducing system size, power and cost. Although many companies are still seeking this holy grail by leveraging the bandgap properties of silicon, Applied Plasmonics has taken a radically different route.

The approach uses a technology the company calls plasmon wave propagation. In this scheme, an electron beam is directed across the surface of a chip, accelerated by a high-voltage bias source. On the chip surface, one or more arrays of nanodimension antennas of specific height and spacing are fabricated using standard lithographic and etch technology. When the e-beam passes over the top of the antenna array, it excites the natural surface plasma on the tips of the array to generate photons, thus creating light.

"The nanoantenna arrays allow any specific wavelength or combination of wavelengths of light to be generated," said Henry Davis, director of market development at Applied Plasmonics (Gainesville, Fla.).

Light emission is independent of the bulk material, and no diodes or transistors are used in the display portion. Logic and some high-voltage circuits, however, will be needed to drive the control elements that switch the beam on and off and steer it across the surface.

In addition to paring the size and cost of semiconductor devices, the scheme would allow system designers to create simple inter- and intrachip communications systems, or even inter- or intraboard links, with minimial overhead. Additional applications might include clock broadcast schemes, distributed computing systems, short-haul networking, secure networks, displays and printers.

The electron beam is generated by a lateral field-emission tip, biased at about - 20 kV. "Such tips would typically be fabricated on the chip, with one or more per antenna array," said chief technology officer Jonathan Gorrell. "Since the tips do wear out, multiple tips can be fabricated and then switched, so that a new tip goes into use as its predecessor deteriorates below a predetermined level." The electrons can travel without scattering and are easily manipulated.

The beam is then accelerated over the surface of the silicon, much as a cathode-ray tube accelerates electrons to affect the phosphors (see illustration at left).

Special nanometer-sized control elements, similar to the deflection plates in a CRT, can also be fabricated on the surface of the silicon. By applying the proper bias (typically about 10 volts) and timing on those elements, the electron beam can be deflected or blocked. "Thus, a single beam can be used to excite multiple nanoantenna arrays to create characters, images for displays or just single points of light for chip-to-chip communications," Gorrell said.

The optical devices can operate at high frequencies — switching times are better than 10 - 13, the company reports — and can generate light with a frequency range from far infrared to ultraviolet.

Nanolithographic techniques are the key to the antenna arrays, Davis said. Light with the same spectrum as sunlight, or any other spectrum desired, can be generated, and the efficiency of the light generation is mostly determined by the effectiveness of the electron trapping of the e-beam by the nanoantenna arrays. The company said the plasmon wave propagation scheme is several times more efficient than fluorescent lamps or LEDs. Moreover, the energy from the electron beam can be recovered.

A single electrodeposited metal layer of silver is used to create the nanoantenna array, and the periodicity of the antenna elements controls the light frequency. The antennas are 100 nm high, and, when fabricated using lithographic tools targeted for 90- and 65-nm features, antennas with a pitch of 155 to 250 nm can be created. Each antenna might be 60 to 180 nm long and at least 30 nm thick (or up to 90 percent of the pitch), said Gorrell. A basic linear array of antennas might be several microns long, Gorrell said, but depending on the light output and shape, the antenna array can be configured in many patterns — linear, rectangular, square and so on.

Multiple arrays, each with a different light frequency, can be fabricated on a single substrate. That substrate can be any integrated circuit, but of high interest are the highly integrated submicron CMOS ICs, said Davis. "Many of these ICs have limited high-speed I/O ports, and through the incorporation of nanoantenna arrays, the surface of the chip can be used as an additional I/O resource," he said. "Antenna arrays interspersed with optical receivers could form free-space high-speed optical links with a nearby chip or be used to drive a Raman laser."

Independently verifying the plasmon wave propagation, Paul Holloway, director of the Material Science and Engineering Department at the University of Florida, has re-created the structures in his lab using a standard material-processing flow. Visual confirmation of the technology's ability to generate light and vary the colors was done using a single e-beam and a simple line of nanoantenna structures. Signal timing was straightforward, Holloway said, but some issues remain. Focusing the electron beam and keeping it on target are key issues for stable displays, he said. Other issues that must be dealt with to make the technology commercially viable include the vacuum packaging, creating a compact electron source and small high-voltage supplies.

Voicing some similar concerns, Rob Lineback, the senior market research analyst for IC Insights (Scottsdale, Ariz.), thinks the startup may be facing a chicken-and-egg dilemma. "To get companies to adopt the technology means spending lots of time educating designers and providing proof-of-concept demonstrations," Lineback said. "At the same time, Applied Plasmonics must convince the process-equipment and other tool vendors — as well as suppliers of electron sources, power supplies and hermetic packages — to put together the infrastructure needed to turn the concepts into a manufacturing reality."

Still, Lineback sees "many positive aspects" in the technology. "Unlike compound semiconductors, which degrade over time due to heat or other material changes, the antennas used in the plasmon wave propagation technology have no wear-out mechanism, and the electron beam generates no heat, thus potentially making the technology very reliable," he said. Some of that reliability may depend on the supporting infrastructure (power supplies, electron source and so on), he said.

Regardless, the company plans to license the technology and develop commercial products based on it. A developer's kit will be available in the fourth quarter.

    See related chart