Optical researchers here have demonstrated a 140-GHz serial-to-parallel conversion using nanojoule optical pulses with a novel self-organized quantum-well material. The initial experiments, part of a collaboration between physicists at Sophia University here and materials scientists at the University of Tokyo, used 515-nanometer pulses. Though these are inappropriate for communications, other resonance energies — which range from the ultraviolet to the infrared — could be exploited instead.
"This is the first demonstration of serial-to-parallel conversion of nanojoule-order terabit/second signals at room temperature at a conversion rate of 140 GHz," said Kazuhiro Ema, professor of physics at Sophia. "Previous reports were only demonstrations at low temperature 10 Kelvin, or were using high-intensity submicrojoule signals. Our present results should bring the time-to-space conversion to a practical application level."
Ultimately, time-to-space conversion, another name for the serial-to-parallel variety, "may lead to ultrahigh-speed optical transmitter and receiver boxes with lower cost or higher functionality," said Andrew Weiner, a professor in the school of electrical and computer engineering at Purdue University, who is involved with similar work.
In this technique, Weiner said, "a block of ultrashort optical pulses in a single serial stream are separated in a single operation into a series of pulses separated in space." In other words, "An entire block of pulses is converted simultaneously and in parallel," he said. By contrast, "in conventional high-speed demultiplexer approaches based on ultrafast optical gates or high-speed integrated optic modulators, demultiplexing occurs essentially one bit at a time."
In the future, Weiner suggested, "The unique 'word-at-a-time' nature of this approach may also alleviate some of the stringent timing requirements of traditional mux-demux approaches as they go to very high bit rate."
The Sophia-Tokyo conversion technique works by combining diffractive spreading of an incoming serial pulse with a nonlinear optical process called four-wave-mixing (FWM) that takes place in the multiple quantum-well material. FWM involves the active changing of the optical properties of the material by the light passing through it, and thus offers a way for different beams of light to interact with each other.
The grating begins the conversion by spreading out the signal's spectrum: This transforms it so that it is effectively half temporal, half spatial. Next, FWM is used to nonlinearly combine the spectrally decomposed pulses both with each other and with a reference pulse. The different colors end up creating structures within the multiple quantum well that affect their own wavelengths — a non-linear optical response. Through cross-correlation within the signal itself and with a reference pulse, the transformation from time to space is completed.
This system has advantages over a competing three-wave-mixing technique because the frequency of the output is the same as that of the input: This means that such systems can be cascaded. In addition, the new scheme can be used for conversion in both directions.
A new material makes this process plausible for networking applications. The organic-inorganic crystal layers are made up of a two-dimensional network of corner-sharing octahedra self-organized quantum wells and layers of alkylammonium chains. The dielectric constant of the latter layers is low, so the excitons — or excited quantum-well states — are tightly confined. Researchers say that the room-temperature nonlinearity of the material at the excitation resonance is very high. In addition, the longitudinal and transverse relaxation times of the material, which set the speed at which the material can work, are fast: 7 and 0.5 picoseconds, respectively.
Fully relaxed
In their experiments, the researchers multiplexed single 300-femtosecond pulses to produce 4-bit signals at 1 nJ and reference pulses at 60 nJ. The time between reference pulses was 30 ps, so that each one would arrive after the material had fully relaxed, while the time between signal pulses was 2 ps.
The signal pulse stream is diffracted by a grating, and then focused down to a spot to maximize intensity. The reference pulse stream is also focused down, and acts to temporally select from the spread-out signal stream. The result was recorded on a streak camera. Unfortunately, the camera's speed limited the operation of the system to 140 GHz. Nevertheless, researchers were able to calculate, by manipulating bit separation times, that terabit/second operations can be converted in real-time.
Materials scientists face several problems in trying to make the technique practical in real applications. Most obviously, they have to shift wavelength: Their experiments have so far been conducted using green light. Aware of this deficiency, they are already looking at other materials with resonance energies more suitable for communications.
In addition, "Although our time-to-space conversion has high sensitivity, the conversion efficiency — about 1 percent — is still low," said Ema, the Sophia University professor. "For a real system we need more than 10 percent efficiency. To overcome it we have to improve the design of the material to have a larger nonlinearity."
Tsuyoshi Konishi, an assistant professor in the department of material and life science at Osaka University, pointed out another difficulty. "The main disadvantage of the time-space conversion would be a power budget," Konishi said. "It pulls up multiple information including an ultrashort pulse and expands them on a spatial domain for parallel processing. This procedure causes a reduction of power for each spatial channel."
Nonetheless, he is optimistic. "I believe that the time-space conversion method could contribute to the next generation of photonics as a bridge between the fiber optics time domain and spatial optics spatial domain," Konishi said.
