Indium phosphide benefits high-performance transmission

Hava Volterra

Micha Zimmerman

The rapid increase in optical networking speeds and distances is placing heavy demands on the underlying optoelectronic components. In particular, transmission at very high speeds (40 Gbit/s and above), and over very long distances (2000 km and above), requires special device characteristics. Indium-phosphide-based devices offer certain advantages for such high-performance transmission.

The challenges

As transmission speeds and distances increase, various signal distortions that are not significant at lower speeds and distances become important and can limit transmission performance. Particularly well-known is chromatic distortion, which increases by the square of the bit rate and can severely limit transmission distances at speeds of 2.5 Gbit/s and higher. As the bit rate grows so does the peak power in order to maintain the energy per bit. At rates of 10 Gbit/s and above additional factors become important, such as fiber nonlinear distortions like self-phase modulation (SPM). The higher bit rate reduces the period so that effects like polarization-mode dispersion, negligible at lower bit rates, come into play.

While channel speeds and distances are increasing, fiber capacity also is being increased in a third dimension-channel count. Higher channel counts combined with the higher power demand associated with high-speed transmission lead to much higher aggregate signal power being transmitted through the fiber. This, in turn, leads to additional phenomena such as four-wave mixing and cross-phase modulation that can further degrade transmission performance.

Increased channel counts and the higher power generally associated with higher speed transmission leads to a third constraint, that of component size. As the number of channels and their speeds increases, there is a corresponding need to reduce optical component size and power requirements.

Modulation

Most of today`s high performance systems use some form of NRZ (non-return-to-zero) transmission, which is a simple on-off format in which light on indicates a "1," and light off indicates a "0." The NRZ format is relatively straightforward to implement and works reliably in many applications. The technological challenges of higher speed and longer distance transmission, however, have raised the interest in an alternative transmission format-in return-to-zero (RZ) transmission, a "1" is indicated by a pulse of light and a "0" is indicated by absence of a pulse. This format is named "return-to-zero" because the signal is turned off between each pulse (see Fig. 1).

Such RZ-modulated signals have several advantages over NRZ signals in high-performance transmission. While the bandwidth of an RZ signal is higher, leading to higher chromatic dispersion, RZ signals afford better immunity to fiber nonlinear effects and to PMD. Since the nonlinear effects are much more difficult to compensate for than the linear ones, RZ signals can offer better performance. As a result, RZ has been the modulation format of choice on many higher-performance systems, including new 40-Gbit/s, ultralong-haul, and submarine transmission systems.

Optical time-division multiplexing (OTDM) takes advantage of the higher speed potential of optical components relative to electronic components, in order to increase optical channel speeds. Optical TDM is achieved by multiplexing in the time domain several short-optical-pulse RZ data streams to achieve a bit-interleaved, denser RZ signal. For example, OTDM can be used to achieve 40-Gbit/s transmission from four 10-Gbit/s electronic data streams. A similar configuration could be used for achieving much higher speeds-four 40-Gbit/s data streams could be multiplexed to achieve 160-Mbit/s transmission, for example.

Current components

Most of today`s high-performance transmitters are built from two basic building blocks, lithium niobate (LiNbO3) modulators and high-power, continuous-wave (CW) distributed-feedback (DFB) lasers.

The LiNbO3 modulators act as electronic shutters, turning light on and off using the Mach-Zender effect. This is achieved by splitting the light, then changing the refractive index on one or both of the two modulator arms via an externally applied electrical signal. When the light is recombined, the difference in refractive index causes the light to interfere destructively, effectively turning off the modulator.

While providing good modulation performance, LiNbO3 modulators have several limitations for high-speed transmission. One is that the size of the modulator increases with modulation speed. Thus, while 2.5-Gbit/s LiNbO3 modulators are available in packages under 3-in. long, 40-Gbit/s devices are significantly larger-5 in. and longer (see Fig. 2). In addition, the drive voltage demands of LiNbO3 modulators increase with frequency- placing a larger burden on the drive electronics, in which larger voltage swings are difficult to produce as the frequency rises.

Indium phosphide

Indium phosphide is a III-V material, meaning that it is a binary crystal with one element from the metallic group III of the periodic table, and one from the non-metallic group V. Another notable member of this family is GaAs (gallium arsenide). Indium phosphide has been a focus of development since the early 1980s. The material system of choice for long distance fiberoptic transmission is InGaAsP/InP because it can provide varying bandgaps in the region from 1.2 to 1.6 µm in which silica fiber has minimum dispersion (1.31 µm) and loss (1.55 µm). It is currently used for a variety of devices, including the DFB lasers commonly utilized for DWDM, semiconductor optical amplifiers, and detectors.

A key advantage of InP in relation to LiNbO3 is device size. Because its refractive index is relatively high, bends in devices can be made smaller and sharper. In addition, the bandgap of InP is closer to that of the light being used in optical communications. Thus electro-optical effects in InP are stronger, which means they can be achieved in shorter distances and with lower drive voltages. For the technology described in this article, InP device sizes are well under 1-mm square, which is negligible compared to the surrounding elements- connectors, drive electronics, and hermetic packaging.

Indium phosphide also has the potential for use in creating integrated devices. In addition to active optical elements, such as lasers, modulators, and amplifiers, passive waveguides can be created with InP. By combining these devices, complex optoelectronic integrated circuits can be created.

In addition to its use in creating optoelectronic components, InP is showing great promise as a semiconductor for use in very high-speed electronics, including the drive electronics for 40-Gbit/s and higher speed optics. In this case it may successfully compete with other high-speed materials, including gallium arsenide and silicon/germanium, for hegemony as the material of choice at higher speeds.

This application to high-speed semiconductors opens the possibility of integrated electronic/optoelectronic components made of InP providing very powerful and cost-effective solutions for the most demanding optical networking systems.

New indium phosphide devices

Two new highly versatile InP devices are compelling alternatives for high-speed transmission. The first is an InP-based electro-absorption modulator (EAM). Such EAMs use the electro-absorption effect in InP materials to modulate light. As a reverse bias voltage is applied to InP, it becomes opaque to light. When the voltage is removed, it becomes transparent. This effect is achieved over less than 300 µm (3 × 10-9 m) of material, and with drive voltages of 3 V and less (see Fig. 3).

While InP electro-absorption modulators currently are being used as part of integrated lasers/modulator combinations (known as EMLs, or electro-absorption modulated lasers), these integrated devices do not provide the performance required for the most demanding applications. Such EMLs suffer from reflections from the modulator back into the laser that change the signal characteristics. In addition, other characteristics of the device, including electrical response, suffer from compromises made in monolithic integration of the laser and modulator.

In contrast, stand-alone EAMs can achieve the modulation performance available today on LiNbO3 modulators, at a fraction of the size and drive voltage. In fact, the size of the actual semiconductor die is less than 100th the size of the equivalent LiNbO3 die, and the size of the packaged device is about one-sixth the size of an equivalent device.

The second InP device is a pulse generating laser (PGL). Unlike the laser/modulator combination traditionally used to generate pulses for RZ modulation, PGLs actually pulse at a predetermined frequency, synchronized to a reference RF signal. Since the laser concentrates all its energy in these short pulses, the peak output power of the laser can be very high, while maintaining a reasonable average power. In addition, the pulses can be set to very narrow widths-thus, as pulse width is reduced, peak power increases to maintain the same average power (see Fig. 4).

In addition to their use in NRZ and RZ transmitters, PGLs and EAMs are ideal devices for implementation of OTDM transmitters and receivers.

FIGURE 4. A pulse-generating laser can provide very short, high-power pulses, suitable for both RZ modulation and OTDM.