Improved designs with novel cooling architecture allow high-power multiple-bar modules to reach new performance levels.
Stephen McComb and Michael Atchley
Compared to other laser types, semiconductor lasers offer the ultimate in compact packaging, lifetime, and reliability. Low-power (milliwatts) semiconductor lasers have long been commodity products that have enabled revolutions in the fields of both data storage and telecommunications. As device output power has been scaled up, semiconductor lasers have begun to penetrate many application areas that were formerly the domain of other lasers, or even nonlaser technology. They also have allowed development of completely new applications.
The parameters that have the most impact on the performance and economics of virtually every semiconductor laser application are output power, brightness, and operating lifetime. The key to market growth has been continued improvement in all three.
The power of a semiconductor laser may be defined two ways. For continuous-wave (CW) lasers, power has its traditional definition of energy flow (measured in watts). Power is important because it determines the maximum throughput or feed rate of a process. The second definition of power applies to lasers operated in a fast-pulsed, quasi-CW mode. Here peak power refers to the maximum power level attained. Peak power is significant because there is a threshold power that must be achieved to perform many processing tasks.
Semiconductor laser devices, particularly bars and stacks, emit light from an extended surface; in contrast, gas lasers usually behave as a true point source. The ability to focus laser output to a small spot has a critical impact on how effectively the output power can be used and is directly related to source size. Brightness is defined as the luminous flux per unit solid angle per unit area of output surface. The goal of semiconductor laser manufacturers is to extract the maximum output power from the smallest emitting surface possible—in other words, to achieve the highest possible brightness.
PHOTO 1 FIGURE 1. A semiconductor laser bar is a monolithic linear array of multiple individual laser emitters. A multibar module is a two-dimensional array of bars. PHOTO END Lifetime impacts the cost-effectiveness of an application in two ways. First, failure of a laser results in the direct replacement cost. In production-line applications, there may also be a significant expense due to the downtime required for replacement and system realignment.
For a given semiconductor laser, lifetime has an inverse, nonlinear dependence on output power; derated operation below maximum specified operating power greatly extends device lifetime. A major goal of semiconductor laser manufacturers is, therefore, to extend lifetime for a given power and brightness level, or, in other words, to increase output power and brightness without decreasing the device lifetime.
Multibar modules *** The first semiconductor lasers were relatively simple devices consisting of a single active junction region or output facet. Today, the brightest high-power single emitters produce more than 1 W of CW power that can be coupled into a 60-µm fiber.
The key to generating really high power has been the development of bars, and more recently, multibar modules or stacks. A bar is a monolithic linear array of multiple individual semiconductor lasers and a stack is a two-dimensional array of multiple bars (see Fig. 1). The most dramatic recent progress has been with these latter products. The industry standard in bars is currently 1-cm-wide devices with 50 W of CW output power and a lifetime exceeding 5000 hours.
The most powerful multibar modules have operated at total powers in excess of 30 kW. The major design goal with these devices is to maximize the output power and brightness without compromising the long lifetime inherent in the bars themselves. The two leading causes of device failure and degradation are burning of the output facets and propagation of so-called dark-line defects in the semiconductor material. With all semiconductor laser types, lifetime is greatly affected by operating temperature because lowering the operating temperature slows the propagation of dark-line defects.
Unfortunately, even though semiconductor lasers are the most efficient devices for converting electricity into laser light, the high powers now in use inevitably result in significant heat generation. This fact dictates that bars incorporate some type of metal heat sink, which is then actively cooled by a thermoelectric cooler or chilled water.
In a multibar module, however, the bars must be positioned as close together as possible to maximize brightness. This precludes the option of mounting each bar on an extended heat sink.
In the case of CW operation, efficient cooling has been the major design issue in multibar modules. Until recently, the approach was to use thin copper heat-sink spacers between the bars and to mount this assembly on a copper heat sink. Cooling water then flows behind and around the bars, as well as through the main heat sink.
This design approach resulted in the "1°C/W" rule of thumb relationship; increasing the output power of a bar by 1 W raises the temperature of that bar by 1°C relative to the heat-sink temperature. Because there is a practical limit (approximately 15°C in a dry atmosphere) on the minimum temperature of the cooling water, this places a limit on the output power. As with early semiconductor lasers, designers were once again confronted with a trade-off between operating lifetime and output power.
In the past year, engineers at Opto Power have broken through this thermal-gradient barrier, using minichannel cooling technology. Individual bars are separated by copper cooling plates approximately 1 mm thick, through which the cooling water flows. The space inside these hollow plates is partitioned into flow channels by thin cooling fins separated by a few hundred microns, maximizing the metal surface area in contact with the cooling water (see Fig. 2).
PHOTO 2FIGURE 2. The use of minichannels enables efficient cooling and high-power operation of multibar modules. PHOTO END Using this approach, the bars can be closely stacked with a 1.8-mm bar-to-bar pitch, yet with a temperature differential of only 0.3°C/W. This design has allowed stacks of 1-cm bars to be operated at 50 W CW per bar with lifetimes still measured in the 5000-10,000-hour range. For example, a 2 x 10 array can produce 1 kW from a total emitting surface of only 4 cm2 (see Fig. 3).
There is also a growing market for quasi-CW devices, particularly for military applications such as target ranging and simulator illumination. These applications typically require a low duty cycle (a few percent at most), which greatly reduces the cooling requirements, simplifying product design and cost. For these uses, up to 16 bars are stacked on top of a single water-cooled heat sink, with no additional cooling elements required between the individual bars. This approach allows the bars to be operated at 100 W peak power per bar, with duty cycles up to 2% and pulse durations of 1 ms or less. This type of 16-bar stack delivers a peak output power of 1.6 kW, yet has a typical operating lifetime of more than 1 billion pulses.
Other design considerations *** These performance levels result from ongoing improvements in device design and fabrication, including bonding methods, cavity length and emitter size optimization, fiber-coupling techniques, and epitaxial growth control.
The size (both length and width) of the active stripes in a bar is another factor that has been carefully optimized. A longer cavity means that the bar has a larger surface area attached to the heat sink. Different emitter widths allow for optimization of cooling from emitter to emitter. Each emitting facet then generates 10 mW/µm2, which is currently the maximum before risking facet damage.
Both bars and stacks produce large, divergent output beams that have a rectangular cross section in the far field. In the case of bars, fiber coupling can circularize this output. Each emitter in the bar can be coupled into separate fibers that are then formed into a circular bundle. The typical brightness currently achievable with this approach is 30 W from a 770-µm-diameter fiber bundle. A variety of novel optical elements have also been developed in recent years to reshape the output of a stack to a focused line or small circular cross section.
Applications *** For many years, the dominant application for high-power semiconductor lasers has been as 808-nm pump sources in Nd:YAG, Nd:YLF, and Nd:YVO4 solid-state lasers, replacing lamp-pumping in most neodymium lasers of less than 100-W output. Recent performance advances have led to explosive growth in applications at several other wavelengths. In the graphics industry, 830-nm devices are now used extensively for applications such as color proofing and digital direct-to-plate printing.
PHOTO 3 FIGURE 3. Multibar modules, such as the Monsoon family of devices, can now deliver several kilowatts of CW laser power. PHOTO END Lasers at 915 nm are finding use as pump sources for fiber lasers, and 940-nm devices are similarly used to pump ytterbium lasers. Of course, the market for 980-nm devices as pump sources for erbium-doped fiber amplifiers also continues to grow rapidly. In addition, semiconductor lasers with output at 1.7 µm are finding use in a number of new medical and military applications, including target illumination and ranging. Lasers at the shorter wavelengths are also being used in a range of materials-processing applications, such as soldering, plastic welding, paint stripping, and marking.
STEPHEN McCOMB is marketing manager and MICHAEL ATCHLEY is product market manager at Opto Power Corp., 3321 E. Global Loop, Tucson, AZ 85706; e-mail: [email protected].