Tunable EML a key to photonics future
With electro-absorption modulated lasers already the preferred choice for most DWDM networks, tunability now seeks to take the technology to a whole new level.
Lars Andersson and Denis Tishinin
In recent months, perhaps nothing has generated more excitement within the fiber-optic community than the tunable laser. What could be more appealing to a system designer than the ability to easily alter an optical signal's physical route on demand? Even in its infancy, tunable lasers are redefining the way the telecommunications industry configures its networks, and the radical implications of this technology will only grow in the years to come.
The benefits of this technology are enormous. Tunability takes DWDM systems to the next level. While DWDM systems make the most of a network's configuration by carrying multiple channels along a single strand of fiber, the amount of supporting components needed-and the complexity of organizing them-almost belies the efficiency this technology is supposed to bring to telecom networks.
Tunability eases quite a bit of this complexity and clutter by significantly reducing the number of necessary components in DWDM systems, making it an excellent facilitator for optical add/drop multiplexers (ADMs) and optical switching/routing. Tunability enables faster and more flexible network reconfiguration, which is one of the greater challenges to carriers, as their ability to meet customers' demands for effective bandwidth management is crucial to success in this crowded marketplace.
When tunability is introduced into a DWDM system, the optical signal's path is modified throughout the network simply by changing the wavelength. Recent advances in this technology have allowed for the realization of very fast switching times, providing network designers with even greater flexibility.
In addition, multichannel tunable lasers can serve as effective backups to numerous fixed-wavelength lasers, resulting in significant inventory reductions and tremendous cost savings (typically, network engineers keep one backup on hand for each wavelength in service). The ability to utilize one component to back up multiple channels has far-reaching benefits. Aside from obvious monetary savings, valuable board space is utilized with maximum efficiency since the tunable transmitter is no larger than the combination of the fixed laser components it is backing up. Other applications include instrumentation and remote sensing.
As the tunable-laser transmitter becomes more commonplace, the number of product offerings will increase. Tunable technology has been applied to a number of laser transmitters in an effort to meet the various demands of the telecommunications industry-and a plethora of options are available. We'll explore the performance, design, and capabilities of one specific tunable transmitter-a tunable electro-absorption modulated laser (EML) that utilizes 16 channels spread out over the ITU grid at 50-GHz spacing.
The EML is currently the preferred choice for DWDM networks for a lot of good reasons. Its degree of integration and compact size lends itself to lower cost, and it typically has a lower chirp than directly modulated lasers. In addition, EML technology is currently available at OC-48 (2.5-Gbit/sec) and OC-192 (10-Gbit/sec) data rates, and both long-haul and metropolitan system designers can integrate tuned EMLs into their networks (see Table). These factors make an EML the logical choice for our entrance into the world of tunability, and have resulted in the advent of the tunable EML (TEML).
A major part in the effectiveness of a tunable laser is the time it takes to switch channels. In optical networking, speed is of paramount importance because customers demand it. Therefore, networks must make the necessary accommodations to meet this demand. Many current laser modules that use tunability rely on temperature tuning to switch channels. With this method, the temperature of the chip itself is changed to create variations of the distributed-feedback (DFB) laser's refractive index and, therefore, emission wavelength. This process is extremely simple, but it is inherently too slow for fast switching applications, including network system backup, since tuning time is measured in seconds, an eternity in the world of optoelectronics.
An alternative is electronic tuning, which allows for much faster channel switching. Electronic tuning is achieved by injecting current into the laser's tuning (phase and Bragg) sections. Changes in carrier density (caused by the introduction of current) result in variations of the effective refractive index, which yields the desired wavelength tuning. The process of electronic wavelength tuning is fast, with switching time limited only by the speed of the integrated drivers and controllers. Typically, switching times could be expected in the microsecond-to-nanosecond range, depending on the type of driver employed.
The basic principle behind a TEML is the monolithic integration of a tunable laser and electro-absorption modulator (EAM). The tunable laser is used to set desired wavelengths, and the monolithically integrated EAM is used to create low-chirp data modulation. The use of a three-section distributed Bragg reflector (DBR) laser yields ideal end results when compared to other types of lasers.
For example, if only one tuning current is used, simpler two-section (gain and DBR) lasers do not allow for quasi-continuous wavelength tuning. The wavelength will jump between the cavity modes, not allowing for coverage of all the channels on the ITU grid. The three-section DBR offers the best combination of ease-of-use and the ability to integrate a modulator section. With more complex lasers (sampled-grating DBR, superstructure-grating DBR, grating-assisted co-directional coupler, etc.) it is more difficult to integrate a modulator.
In the three-section DBR, tuning is achieved by injecting current into the phase and Bragg sections. Again, temperature can be used for tuning, but current is preferred, since it offers fast switching speed and a wider tuning range. The integration of a modulator offers advantages over designs that require polarization-maintaining fiber and an external lithium niobate modulator. The module can be manufactured in a more compact size, fiber management is easier, and costs are more controllable.
Figure 1 shows a typical four-section TEML (three-section DBR laser with an EAM). The gain, phase, and Bragg sections create a laser cavity with a tunable front reflector, and the front-most section is used for modulation. In our electronically tuned model, the three currents (gain, Bragg, and phase) are adjusted simultaneously to achieve desired wavelength, optimum side-mode suppression ratio (SMSR), and output power. We introduced a microcontroller into our module to handle this multidimensional tuning process.
This module requires a more sophisticated control algorithm as compared with a temperature-tuned module, where one current is used to control the power, and temperature controls the wavelength. However, there are still the previously mentioned drawbacks. The Bragg section is used to create a tunable front mirror and the phase section is used to fine-tune the wavelength of the lasing cavity mode. These two sections are transparent to the light emitted by the gain section because the bandgap energy within the quantum well is higher than the photon energy in the laser cavity. Current injected into the DBR section changes carrier density and refractive index, altering the Bragg wavelength and ultimately the lasing wavelength. The injection of current into the phase section is used to stabilize and fine-tune the mode, resulting in quasi-continuous device tuning over the entire output wavelength range (in our example, 16 ITU channels with 50-GHz spacing).
The monolithic integration of the laser with the modulator was created using a selective-area-growth (SAG) pro-cess-a sophisticated, metal-organic chemical vapor deposition (MOCVD) growth on a pre-patterned substrate. The use of SAG allowed us to shift the absorption peak for the modulator and tuning sections relative to the emission peak of the gain region, making them transparent to the emitted light.
In this process, a dielectric mask deposited onto a substrate prior to growth promotes material migration from the patterned area. This enhancement in the multiple quantum wells (MQW) and the barrier growth rate near the edges of a masked pattern is due to material migration and differences in sticking coefficients between the masked and open areas. That increases the growth rate near the masked edges.
For example, the MQW's active gain region (grown in the vicinity of the masked areas) results in a red shift in the transition energies. Therefore, SAG allows the formation of a continuous waveguiding layer with differing transition energies in a single epitaxial growth. This way, the laser and modulator can be grown together with high coupling efficiency and very low loss. Bandgap wavelength offset between the laser and modulator sections can be adjusted via the initial oxide mask dimensions, allowing separate adjusting of the extinction ratio (the ratio of the light emitted in the on and off states) and general performance of the TEML device.
A number of features are needed to supplement a tunable transmitter to ensure proper performance. By optimizing the electronic and mechanical designs, it is possible to integrate into the transmitter a microcontroller, temperature controller, current sources, modulator driver, wavelength reference interface, and monitoring functions.
Power output from the three-section laser configuration is sufficient, making the integration of an amplifier unnecessary and presenting yet another asset. Figure 2 shows a layout of the internal configuration of the TEML transmitter.
An 8-bit reduced instruction-set computer (RISC) microcontroller is used to supervise and control the tunable transmitter. Control of both power and wavelength is achieved through the laser drivers. The bias voltage and modulation amplitude levels for the EAM driver are also handled by the microcontroller. The electronic design and circuit layout of our model has been optimized for low-noise operation.
Many operational parameters are constantly monitored through the microcontroller, all of which are available to the user through the digital interface. We designed our digital interface to be compatible with a PC's enhanced parallel port to simplify transmitter evaluations. Control functions were also integrated, allowing the user the ability to monitor performance of the laser and several internal subsystems. Alarms are available to system designers to indicate malfunctions. In the event a given application requires additional stability, a generic interface for mating to all known wavelength lockers is incorporated and applicable with both etalon- and filter-based types.
Thermal management is an issue that must be addressed when designing a tunable transmitter, and several features can be incorporated to improve thermal characteristics. Typically, the most power-consuming part of a module of this type is the thermo-electric cooler (TEC). By operating the TEC from its own power supply, its dissipated heat can be reduced by as much as 50%. To increase the heat transfer from the tunable transmitter to its surroundings, the TEC can be mounted directly to the motherboard, eliminating the need for cooling fins. This feature also results in a lower overall package height, the advantages of which are obvious.
Performance testing of tunable technology coupled to an EML has yielded promising results. The intrinsic nature of this setup lends itself more easily to 10-Gbit/sec transmissions. Typical fiber-modulated output power is 0 dBm after the modulator section, and extensive testing has resulted in extinction ratios of 10 dB or better on all 16 channels. In addition, SMSR testing has been consistently 35 dB or better on all channels under modulation. Figure 3 shows the wavelength tuning spectrum of the TEML and Figure 4 shows the eye-pattern of the TEML operating at 10 Gbits/sec.
Tunability is an excellent, natural progression of DWDM technology for the telecommunications industry as the spacing between channels continues to decrease. The marriage of this technology with EMLs offers a compact, cost-effective approach to tunability with a high degree of built-in integration. The net result of this high degree of integration is that overall system capacity is increased while utilizing a minimum of valuable board space. Additionally, since many connections are eliminated because support systems are an internal part of the module, assembly times are reduced.
Lars Andersson is manager of tunable-laser technology and Denis Tishinin is manager of tunable-laser device design at Multiplex Inc. (South Plainfield, NJ).
All distributed Bragg reflector (DBR) lasers have at least two sections: gain and Bragg. In a DBR laser, the cleaved facet next to the gain section and the grating in the Bragg section form the laser cavity. The Bragg grating works as a wavelength-selective mirror. The wavelength of peak reflectivity (the Bragg wavelength) is determined by the grating period and the effective refractive index of the waveguide.
Lasing will occur on the cavity mode that is in some sense closest to the Bragg wavelength. By injecting current into the Bragg section, the effective refractive index and therefore the Bragg wavelength can be varied. By doing this, a coarse tuning is achieved. When the current is increased, the Bragg wavelength decreases and the lasing mode will simultaneously jump from cavity mode to cavity mode in a stepwise fashion. The positions of the cavity modes are more or less constant during this coarse tuning.
The laser has an infinite number of cavity modes equally spaced in frequency. The optical length of the laser cavity determines the distance between two adjacent cavity modes. Adjusting the optical length of the laser cavity can therefore move the cavity modes. This adjusting is often referred to as fine-tuning, achieved by altering the effective refractive index of the laser cavity.
Two common ways of altering the effective refractive index, are injecting current or changing the temperature of the laser. When the correct wavelength has been achieved, the mode stability has to be considered. There are a large number of possible Bragg-phase current combinations that generate the desired wavelength, and there are several ways to find the "best" one for that given wavelength.
One of the more common ways is simply to choose the combination that gives the highest side-mode suppression ratio (SMSR). However, hysteresis effects in the vicinity of the mode jumps also have to be considered. Other methods aim for selecting a parameter that is more easily accessible but still gives a good SMSR. One control scheme would be to use the phase section to fine-tune the wavelength and the Bragg current to optimize the output power. In this way, if used together with a wavelength reference, the operation point can be continuously monitored while the transmitter is running.