The high-power multimode laser diode is the engine behind all types of fiber lasers, diode-pumped solid-state lasers (DPSSs), and high-power fiber amplifiers. This key component efficiently transforms input electrical current into mid-IR laser light, setting the performance and cost baseline of such laser systems.
Fundamental performance improvements in multimode diode-laser structures and new pumping architectures are poised to make a substantial impact on a wide range of lasers. High power-conversion efficiency (PCE), high reliability, and lower cost drive the development and adoption rate of new laser architectures such as fiber lasers.
Applications of diode pumps
High-power multimode diode pumps have already begun to replace flashlamps, typically in Nd:YAG and Nd:YVO4 (vanadate) lasers, in which the higher efficiency and longer lifetime of the diode outweigh the cost increase. A secondary improvement is a more compact system design because of the smaller power supply required, allowing for the potential of direct air-cooling.
More recently, double-clad optical fiber (DCF), with a single-mode or near-single-mode core and a secondary multimode outer core, has enabled the pumping of fiber amplifiers and fiber lasers with multimode laser diodes, bypassing the 200- to 400-mW (typical) single-mode diode-pump output limit. High-power communications amplifiers based on DCF and with output of +32 dBm are widely deployed using Telcordia-qualified multimode pump diodes.
FIGURE 1. Power-current (left, red) and PCE (left, blue) plots of an Alfalight 50-W, 976-nm laser bar with an advanced high-efficiency structure illustrate the device�s world-record PCE of 73% (10°C submount temperature). Driving diode PCE from 50% of conventional devices to beyond 70% reduces the heat dissipated from a 50-W laser bar down to 20 W (right, green). Conversely, it makes possible a 120-W laser bar with the same 50 W of heat dissipation (right, blue).
Several powerful continuous-wave (CW) and pulsed fiber-laser architectures are being developed and some commercially deployed, with output ranging from 10 W to more than 10 kW (see Laser Focus World, December 2004, p. 81). Fiber lasers are expected to achieve better beam quality and potentially higher pulse repetition rates than their DPSS brethren. Pump diodes with the right properties are a key ingredient in these emerging architectures.
Key specifications
For DPSS applications, 808 nm is the most common pump wavelength, matching the absorption bands of Nd:YAG/YVO4. Erbium- and ytterbium-doped fibers for amplifiers and lasers are most commonly pumped at 915 and 976 nm. For disc-based Yb:YAG DPSS architectures, 940 nm is an emerging requirement. Beyond these applications, a wide range of wavelengths are required, including low-quantum-defect pumping of YAGs, pumping other dopants such as holmium and praseodymium, and pumping using other hosts like ceramics and specialty glasses.
Multimode pump diodes are primarily packaged in free-space and fiber-coupled packages. Single free-space emitters typically emit from 2 to 5 W, depending on wavelength. Free-space laser bars typically produce 30- to 60-W CW, and comprise 19 to 141 individual emitters across a 1-cm length. More than 100 W per bar is commercially available in quasi-CW operation. Fiber-coupled single emitters produce up to 5-W ex-fiber; higher output power, from a few tens to several thousand watts, is achievable by combining the output of several single emitters or bars.
Reliability of these devices is determined by the inherent diode material structure and by the packaging and handling methods used. The requirements range from a few thousand hours-for some pulsed-laser applications-to Telcordia-qualified 75-year median life for high-reliability communications applications.1 Diode lifetime is characterized by infant, random, and wear-out failure. Infant failures can be eliminated through proper burn-in screening. Accelerated lifetime testing of the device and packaging is used to determine random- and wear-out failure rates. The degradation rate, or decrease in output power over time, is typically sufficiently tested for most industrial applications by using temperature acceleration. Extensive Telcordia-like testing creates a matrix of diode degradation over elevated temperature and output power, providing Arrhenius activation energy and power-scaling factors.
Reaching new efficiency highs
One of the most exciting developments in pump laser diodes is the improvement of output power-conversion efficiency (PCE), which is currently about 50% for production devices. A 50% PCE means that for every watt of output, an equal amount of heat is generated. This heat must be carried away to keep the junction temperature at a reliable level. Ultimately this heat limits the amount of output power that the laser can provide.
FIGURE 3. A 20-W fiber-coupled module can be used in both pumping and pulse-amplifying configurations for fiber lasers.
Efficiency improvements directly influence many other key performance parameters. Lower dissipated heat reduces the junction temperature, exponentially increasing the diode lifetime. In addition, it improves tolerance to packaging variations. Reduced electrical input power enables portability and the ability to build multikilowatt-class lasers without an enormous powerplant and cooling facility. Ultimately, high efficiency can reduce cost per watt.
The DARPA Super-High Efficiency Diode Sources (SHEDS) program, whose ultimate target is 80% PCE, has driven recent performance advances. As a result of these efforts, Alfalight has demonstrated a 50-W diode bar with a record 73% PCE (see Fig. 1).2 These efficiency improvements were achieved by altering the laser-diode quantum-well structure and characteristics of the optical cavity.3
A laser diode consists of a quantum well that defines the wavelength and converts injected current into photons, a waveguide defined by the index of the surrounding material, and a pair of mirrors produced by cleaving the end facets of the laser diode.
The laser diode is not 100% efficient at turning an injected electron-hole pair into a photon because of a number of loss processes (see Fig. 2). Below-threshold losses stem from the fact that a certain amount of current is needed simply to achieve population inversion (that is, to get the laser to threshold), as well as from nonradiative processes. Carrier leakage refers to injected current that does not pump the quantum well, but leaks around it instead. Joule heating is resistive heating from the metal-semiconductor contact resistance and from resistance of the heterostructure itself. Band-alignment loss comes from the bias voltage required to align the bands of the heterostructure, which does not contribute directly to lasing. Scattering and absorption results from photons leaking from the waveguide, being absorbed by free carriers, or being scattered by internal structures.
The leap from 50% to more than 70% efficiency was achieved primarily through evolutionary improvements to an existing laser design. Elements included reducing joule heating through better doping and contact resistance, engineering the heterostructure to reduce the band-alignment voltage deficit (and thus operating voltage), and reducing scattering and absorption through improved crystal quality, dopant choice, and waveguide design.
Future developments
The demonstrated efficiency improvements will result in production-level devices by proving manufacturability and reliability. Additional efficiency improvements may come from revolutionary work using quantum dots in the heterostructure or growing the quantum well on a (110) crystal plane.
Improved cooling techniques and packaging also affect overall laser-system efficiency by reducing the energy required to cool the diodes. The latest techniques include new water-cooled mounts that reduce pressure and flow requirements, cooling fluids that contain nanoparticles, and cooling techniques using phase-change materials to temporarily store heat removed from the diodes.
Brightness-the amount of laser power per emitter area and per solid angle will also improve. Increased brightness can increase system efficiency by reducing the total pump power required or by enhancing the absorption of pump light. This is especially true for fiber lasers, some microchip laser configurations, and many high-power direct-diode applications. Many fiber-laser architectures benefit from having the maximum amount of pump power provided into a 200-, 250-, or 400-µm-core fiber with a numerical aperture (NA) of 0.22.
Improved coupling efficiency into the optical fiber, made possible both through incremental improvements in optical-train design and improved confinement of the optical field in the waveguide, will enhance future brightness. Brightness can also be enhanced by the use of pumps coupled with low-NA fiber (0.15), which also eases output-coupling constraints and allows for the use of high-efficiency tapered fiber bundles to combine the output of single emitters. This approach is useful both for pumping fiber lasers and for coupling pump light into passthrough DCF to build pulse amplifiers (see Fig. 3). Furthermore, as the thermal properties of packaging are improved, diodes will be able to reliably provide more power out of a given facet area, increasing brightness.
System-pumping schemes that rely on narrow absorption bands, such as 976 nm for erbium/ytterbium or 808 and 885 nm for Nd:YAG, will benefit from stabilization of the laser output wavelength. Stabilization enables wide-temperature-range operation of the pump diodes or allows passive cooling, important for field-deployed DPSSs and fiber lasers as well as reduced-cost fixed installations.
The use of wavelength-stabilized multimode diodes makes spectral beam combining feasible and improves manufacturability; such techniques also significantly enhance the combined-source brightness. This is important for fiber-laser pumping and in some direct-diode applications.
Wavelength-stabilization techniques, such as the use of an external volume Bragg grating (VBG), have already been successfully implemented. These methods add cost to the pump, however, and could reduce reliability due to additional packaging components. Future stabilized pump lasers are expected to have wavelength stabilization integrated into the diode through holographically written gratings incorporated into the laser material itself. Side benefits are reduced manufacturing costs from improved wavelength yields.
Diode total cost and cost per watt are expected to gradually decrease, enabling the adoption and growth of more DPSS and fiber-laser architectures. Costs will improve, not only because fewer diodes are required due to increased brightness and power per emitter, but also because of the expected increase in volume production as more diode-pumped architectures are adopted. Simplified cooling and coupling schemes will reduce packaging costs, and more-mature III-V processing methodologies will reduce manufacturing overhead.
The reliability of bar and single emitters will continue to improve, and will rise to meet the demanding requirements of industrial-laser and communications-system manufacturers. Industrial customers should come to expect 10,000 to 30,000 hours of life from bars as well as high-power single emitters. ❏
REFERENCES
1. M. Kanskar et al., Proc. SPIE 4995, 196 (2003).
2. M. Kanskar, et al., LEOS 2004 Proc., WC-4 (November 2004).
3. M. Kanskar et al., Solid-State Diode Laser Tech. Rev. Proc., DIODE-1 (June 2004).
ROB WILLIAMSON is director of technology and business development at Alfalight Inc., 1832 Wright St., Madison, WI 53704; e-mail: [email protected].