The core technology behind the tunable external-cavity laser is the broadband gain chip. Laser manufacturers have much to consider when choosing a chip. Telecommunications requirements such as dynamic provisioning, channel protection, wavelength conversion and the replacement of fixed-wavelength distributed feedback lasers are driving the adoption of tunable laser sources. External-cavity lasers offer several advantages over other approaches: high power, wide tuning range, high side mode suppression ratio, narrow linewidth and high efficiency. The core technology behind the external-cavity laser is the broadband gain chip, a semiconductor superluminescent diode that acts both as a broadband light source and as a frequency-agile gain medium. The largest market for broadband gain chips is in telecommunications, where external-cavity lasers will act as the sources for metro, long-haul and ultralong-haul networks. The tunable laser telecommunications market is small today, but it is expected to grow rapidly and to reach $869 million in 2005. The chips also have a place in lasers for test and measurement. These lasers display tuning ranges greater than 100 nm and are used primarily for evaluating optical components and systems. In the Littrow external-cavity laser design, the two critical components are the chip and the filter. When current is applied to the chip, it emits a very wide spectrum of light that is collimated onto the filter. The filter is positioned at a precise angle so that only the desired narrow spectrum reflects to the chip. This feedback is then focused back into the broadband gain chip, creating a lasing cavity between the far facet of the diode and the filter. At this point, the chip begins to lase at the desired wavelength. To tune the external cavity, the filter is rotated to reflect a different wavelength. In the Littrow external-cavity laser, the wavelength is determined by the angle of the filter, which controls the wavelength of light that is reflected to the broadband gain chip, creating a lasing cavity. Courtesy of Princeton Lightwave Inc. Many different broadband gain chips are on the market, and it is important for laser manufacturers to select the one that exactly fits their needs. The proper selection of the chip can significantly increase the power and widen the tuning range of an external-cavity laser, but it also can simplify integration and lower manufacturing costs. Furthermore, the reliability and stability of these lasers are directly related to the chip. Selecting a chip Important criteria in selecting a broadband gain chip are gain, tunability, manufacturability, reflectivity and chip design: • Gain and power. Ultimately, the output power of the chip determines that of the laser. The current output requirement of an external-cavity laser for long- and ultralong-haul applications is 20 to 40 mW across the entire tuning range. The metro market requires between 5 and 10 mW. To reach such power levels, high coupling efficiencies, low cavity losses and very high gain are necessary. To achieve a high coupling efficiency into the broadband gain chip, the coupling optics must be designed for the chip’s specific far-field radiation pattern. In addition, it is important that the chip have a large active region so that the alignment tolerances of the coupling optics may be relaxed. Generally, a ridge waveguide structure has a much larger active region than a buried heterostructure. The waveguide design in a broadband gain chip affects the reflectivity at one of the facets, and so the lasing wavelength. A curved or angled waveguide yields an effective reflectivity of 1 x 10—4, relaxing the requirement that the antireflection coatings achieve reflectivity of less than 5 x 10—4 across the tuning range. Both the material structure and the cavity length determine the gain. The choice of cavity length of the broadband gain chip is based on power requirements, constraints on size and the mode spacing of the lasing cavity. To obtain the 20 mW that is necessary for the long-haul market, the chip must have an output power of more than 40 mW to account for losses in the laser. As the power requirements for external-cavity lasers increase, the gain of the chip also must increase; experimental results at Princeton Lightwave Inc. have shown output powers of more than 100 mW in a standard Littrow cavity. An experimental Littrow external-cavity laser at Princeton Lightwave displayed an output of more than 100 mW. The drive current to the broadband gain chip was 450 mA. • Tuning range. The broadband gain chip is the main determinant of the wide tuning range of external-cavity lasers. The current goal of most manufacturers is to have a tuning range that covers 40 nm in either the C- or the L-band. In the future, manufacturers will extend their tuning range to include both bands. The tuning range of the chip is purely a factor of the material structure. To achieve a wide tuning range, the material structure must be designed with that in mind. In addition, the chips should be designed to have a gain curve that is as flat as possible across the entire tuning range. To achieve a 40-nm tuning range for an external-cavity laser, the gain chip should have a 3-dB full width half maximum gain spectrum of at least 40 nm — and preferably closer to 70 nm. A wider full width half maximum spectrum indicates flatter gain, which means that a manufacturer must perform less attenuation or fewer adjustments to the operating current of the chip to keep the power constant across the tuning range. Experiments at Princeton Lightwave have demonstrated the wide tuning range of broadband gain chips. A standard chip in a Littrow external cavity displayed a range of more than 170 nm. • Manufacturability. Because there are different designs for external-cavity lasers in the marketplace, the manufacturers of broadband gain chips must be able to modify their products for different customers, yet maintain their capacity for high-volume production. Some customers have unique submounts or facet coatings that are optimized for their external-cavity design. Moreover, the manufacturers must keep to a minimum the variation over time of the gain, peak wavelength and placement of the chip on a submount. As the variation increases, laser manufacturers are forced to make costly design changes. The key is that chip providers have their own material growth reactors, foundry and automated pick-and-place production equipment. • Reflectivity and design. A critical characteristic of a broadband gain chip is that one of the facets has a reflectivity of less than 5 x 10—4 across the tuning range. It is essential that the chip lase only at the feedback wavelength selected by the filter. The low reflectivity on one of the facets suppresses self-lasing, which effectively extends the length of the lasing cavity to include the filter. Without it, the chip will lase because of the feedback from the facets and will overpower the feedback of the filter. This leads to modal instability and to poor performance of the external-cavity laser. Moreover, as the power requirements of the lasers increase, the reflectivity at the facet must be reduced to prevent self-lasing. There are two approaches to reducing reflectivity. The first is to use a standard straight waveguide and a coating with very low reflectivity. The second — and preferred — approach is to use an angled or curved waveguide design. Broadband gain chips with straight waveguides are essentially Fabry-Perot lasers. Not designed specifically for external-cavity lasers, they are a less than optimal solution. The main issue with these chips is that they need a very effective antireflection coating on one of the cleaved facets. Most manufacturers of straight-waveguide gain chips cannot produce these coatings themselves. Either they must outsource the coatings, or the manufacturer of the external-cavity laser must send the chips to a coating vendor. Unfortunately, few vendors can deposit multilayer antireflection coatings with a reflectivity of less than 5 x 10—4 across the tuning range. Because such coatings are state-of-the-art, the process is expensive, and yields are low. Furthermore, it is questionable whether the coating vendors can accommodate the large volumes that are necessary for the telecommunications market. The better solution is to use a gain chip that has been specifically designed for use in an external-cavity laser. By using a properly designed angled or curved waveguide in which the waveguides intersect the facet of the chip at an angle, the effective reflectivity will be less than 1 x 10—4. By intersecting at an angle, the light that is incident to the facet is either reflected at an angle equal to the incident angle or is refracted. A simple, low-cost, one-layer coating with a reflectivity of 2 x 10—2 can be added to the angled or curved side of the chip to further decrease reflectivity. Combining an effective reflectivity of 1 x 10—4 and a 2 x 10—2 coating further reduces it to 2 x 10—6. Currently, this is the only way to get such a low level of reflectivity over the entire tuning range. The main benefit of the angled/ curved waveguide design is that chip manufacturers can apply the coating in-house, enabling them to supply components to the laser manufacturer that are easy to integrate and that reduce total costs. If the power requirements increase beyond a point where the reflectivity must be less than 2 x 10—6, the angle of the waveguide can be increased or the reflectivity of the coating reduced. Future developments As performance requirements increase and tunable external-cavity lasers further penetrate the long-haul, ultralong-haul and metro markets, broadband gain chips will continue to evolve. Each specific market segment will need a chip with a wider and flatter gain spectrum. The ultralong-haul market also will require chips with higher gain because of increased attenuation in the fiber. In the long-haul market, there is a need for a chip with a monolithically integrated electroabsorption modulator to ensure small form factor and low cost compared with a continuous-wave external-cavity laser with an external modulator. Lastly, to penetrate the price-sensitive metro markets, a high-gain, directly modulated chip will be required. Meet the author Todd Swanson is a product line manager at Princeton Lightwave Inc. in Cranbury, N.J.