SHEDS leaves high-power legacy

Initiatives such as the recently completed multimillion dollar super-high efficiency diode sources (SHEDS) program, which was run by Defense Advanced Research Projects Agency (DARPA), have laid the groundwork for building highly efficient 100 kW laser-based defense weapons. This work has also assisted the development of a range of mobile defense applications – including laser interferometric detection and ranging, targeting and imaging – and aided commercial laser systems, which benefit from lower power consumption, a smaller factory-floor footprint and portability.

Before the three-year SHEDS program began in late 2003, the state-of-the-art emitters operating in the wavelength range 800–980 nm were InGaAs/GaAs devices delivering efficiencies of up to 45%. The spatial beam quality of high-power laser diodes is insufficient for cutting metal or for long-distance propagation, so these devices are used to pump another laser cavity that produces a far less divergent beam, usually rare-earth-doped crystals such as Yb:glass or Nd:YAG. This conversion process was typically 25% efficient (in YAG) at converting pump light into a high-spatial-quality output beam. If a 100 kW diode-pumped solid-state laser system was built using this technology, it would dissipate 700 kW of heat and consume nearly 800 kW of electricity. Addressing these cooling and electrical power needs would require two tractor-trailers, which is too great a burden for a tactically usable weapon system.

To make portable laser-based weapons viable, DARPA launched the SHEDS program to drive efficiency improvements in room-temperature laser diode bars operating at 976 nm, which is a key absorption line in ytterbium-doped gain material. A target of 65% power conversion efficiency (PCE) was set for the program s midpoint, and an 80% goal for its completion. Hitting the 80% diode goal and improving solid-state laser efficiency could ultimately cut the weight of the laser weapon s thermal management system from 15,000 kg to around 1,000 kg. Alongside this principal target, the SHEDS program laid down other requirements to ensure that these devices were suitable for laser weapons. These included power density of at least 500 W/cm2 per diode bar, a spectral linewidth of below 2 nm, a spread in peak emission wavelength across a bar of less than 1 nm, and a "wavelength stabilization" target that would prevent drifts in emission wavelength with temperature.

The targets laid out in the SHEDS program have been chased by the laser-diode manufacturers Alfalight, JDSU and nLight, which were supported by teams at the University of Central Florida, Caltech and the National Institute of Standards and Technology (NIST). The chip makers have made progress by introducing fundamental changes to the device s design; the university teams have provided solid-state laser modeling and external wavelength-control gratings; and NIST has provided objective, standardized measurement of device performance. With this approach, all three manufacturers have produced 976 nm emitters that exceeded the 65% milestone, although none have hit the final 80% target.

At Alfalight, our efforts ultimately produced a diode bar with a peak PCE of 73% that delivered 50 W at 10 °C, and a 500 W bar stack built from five 1 cm arrays, which had a maximum, NIST-certified array PCE of 69.3% at 25 °C (see figure 1).

During the course of the SHEDS program we improved the efficiency of our diodes by over 20% by identifying and addressing the various loss mechanisms present in traditional quantum-well diode lasers. These efficiency-limiting losses occur during the transport of electrons and holes from their respective contacts to the active region, and also during the recombination process. The losses include: contact resistance at the metal–semiconductor interface; Joule heating in the bulk semiconductor material, which is caused by the diode s effective series resistance; series ohmic resistance of contacts; heating of the crystal lattice as energetic carriers cross heterojunctions between high and low bandgap material; carrier leakage, caused by electrons that do not combine with holes to form photons in the laser cavity; scattering or absorption of the photons out of the laser; and non-radiative losses, including the current consumed to achieve population inversion.

We tackled these loss mechanisms with a new laser design that featured compositional grading of the interfaces in the quantum well to reduce band alignment losses; a judicious choice of material to provide additional optical and electrical carrier confinement; and doping changes that reduced the contact resistance. Our device s scattering and absorption losses were also reduced through the introduction of tighter manufacturing tolerances for several growth and lithographic processes.

With these changes, all the major loss mechanisms were reduced, aside from the below-threshold mechanisms (see figure 2). However, the increase from this particular mechanism is something of an anomaly, as its actual value was unchanged and its rise is only a comparative one, which resulted from a decline in the contribution from other losses.

Although all of the participants in the SHEDS program produced outstanding results, the room-temperature 80% PCE target remains elusive and invites the question of whether this goal can ever be met. The answer is "yes, but..." because significant development is still required. Until now, progress has been made by attacking the losses inherent to traditional quantum-well devices – any more "belt tightening" will squeeze out only small improvements.

Additional gains will instead have to come from more radical designs such as those using an alternative gain medium, which targets below-threshold losses by introducing fundamental changes to the quantum well. The quantum dot and (110)-plane growth approaches explored in the SHEDS program show promise, but it takes time to optimize the growth conditions of these new structures. Until this is done, it will not be possible to manufacture low-loss devices in this fashion. Encouragingly, some true quantum-dot-confined commercial devices do exist today, but output powers and efficiencies are low and it will require significant effort to develop multi-watt diode lasers with this technology.

The benefits of the SHEDS program have already expanded beyond laboratory results to enable efficiency hikes in commercial laser-diode products. In the real world, device lifetimes are critical and lasers must have proven reliability to win sales, including figures of merit such as characteristic temperatures T0 and T1, mean-time-to-failure and lifetime models. The higher output powers bring a new set of challenges that influence packaging, choice of substrates, attachment technologies, interconnects, exterior packages and cooling methodologies. Volume manufacturing also requires an understanding of the acceptable tolerances for growth, device processing and testing. To address all these production issues and transform a lab device into a commercial product with a proven lifetime can take at least a year.

At Alfalight, the SHEDS program results have already been translated into commercial products. The first stage in achieving this was to incorporate some of the lessons learnt during the project by tightening manufacturing tolerances on existing designs. The second, to realize truly improved PCE levels, required introducing entirely new laser structures and the associated qualifications and life-time test. New commercial high-power laser products are generally delivering a PCE in excess of 50%, and some commercial offerings are even approaching 65% in a single-emitter package format.

The improved efficiency has increased the output powers of both single emitters and bar arrays. It also enables the use of brighter pump sources (watts per emitter or ex-fiber) which are the key to making cost-effective fiber lasers and to boosting the power of compact kilowatt-class diode-pumped solid-state lasers.

Our efforts, and those of other SHEDS program participants, are now focused on extending the efficiency gains achieved at 976 nm to other important pump wavelengths, including 808 nm, 885 nm and 915 nm. The principles for reducing various loss mechanisms at these additional wavelengths are the same, although implementation requires different material systems and quantum-well structures.

We have also made our lasers more suitable for efficient pumping of the narrow absorption lines of rare-earth-doped gain crystals and gain fibers by reducing the emitter s spectral width and its wavelength shift with temperature. This is important for building solid-state lasers, as pump power that falls outside the absorption range of the gain crystal is wasted and actually becomes an additional source of laser system inefficiency.

During the program, the diode bar s output wavelength was controlled with a carefully aligned external grating. This approach can be costly and cumbersome, especially when it is applied to multi-bar stacks and arrays. At Alfalight, with support from the Air Force Research Laboratory, we have now developed an integrated solution that involves adding a semiconductor grating during wafer processing. After both the quantum well and waveguide-index cap are grown, a holographic grating is lithographically exposed across the entire wafer, before it is etched and regrown. The semiconductor grating, which is far outside the laser active region, narrows and stabilizes the shift in output wavelength with temperature and only results in an efficiency loss of around 3–5%.

This monolithically integrated wavelength-stabilization technology has produced passively cooled 976 nm laser bars with 50 W outputs that deliver a record PCE of 58% at 25 °C (see figure 3). The bar has an overall spectral width of less than 1 nm and a wavelength shift with temperature of only 0.07 nm/°C, which is a fivefold reduction over traditional pump diodes. Reduced temperature dependence also eases the demands on the system cooling required to maintain pump wavelength over a system s typical range of operating temperatures and output powers.

Although all these improvements will help towards the eventual realization of a laser-based weapon, there is still work to do. The industry is now much closer to the 80% SHEDS target following the hikes in efficiency delivered by all three diode manufacturers, but further improvements will not be easy as they will require radical changes to the device structure.

Users of high-power laser diodes already benefit from the techniques developed in the SHEDS program through the boost in laser efficiency in commercial products. As these techniques permeate to other types of semiconductor laser this may lead to additional laser products with SHEDS-like efficiencies, including surface emitting designs and quantum-dot devices.