The first blue LED was demonstrated in 1962, shortly after the invention of the first diode laser. This first blue LED was made of silicon carbide (SiC), which like silicon has an indirect bandgap and is, therefore, not useful as a laser material. With the optical-storage application firmly in mind, researchers immediately were intrigued by the possibility of a blue-emitting diode laser, and the search began for a suitable semiconductor. The evolution of blue LEDs has continued to guide this search.
By 1982, the Ministry of International Trade and Industry in Japan was funding a national program for development of continuous-wave (CW) room-temperature diode lasers emitting at blue, green, and yellow wavelengths. But the potential applications for visible LEDs—color displays, outdoor lighting, automobile lights, and traffic signals, which together promise yet another optoelectronic-device revolution in the mass market—has been motivation enough for companies worldwide to fund LED development.
Bright-red LEDs made from compounds of gallium arsenide (GaAs) and arsenic phosphide (AsP) have been available for years. More recently, development of the quaternary alloy indium aluminum gallium phosphide (InAlGaP) has resulted in bright-yellow LEDs. But blue and green LEDs of comparable brightness were missing until recently.
Doubling down
The vision of commercial diode lasers and LEDs with green and blue wavelengths was decades ahead of reality. The possibility of a different opportunity in the near term led to other means of obtaining blue and green output from diode lasers—frequency-doubling and upconversion. Blue and green light at milliwatt CW levels is used in color printing and display and in several applications related to biology and medicine.
The primary laser source for these applications was air-cooled argon-ion lasers, which have their shortcomings, including reliability and cost, which is typically between $3000 and $5000. At this price it seemed possible that the added complexity and cost of a frequency-converted diode-based system might be justified if it could provide longer life along with other benefits.
Frequency-doubling of diodes created a stir in the early 1990s when a number of researchers demonstrated CW outputs of a few milliwatts at around 425 nm. Gallium arsenide-type diode lasers were the pump source for doubling using various methods. A serious constraint on these systems, however, was the strict requirement for the pump source, which typically had to be a relatively low-cost and high-power device that could remain single-mode and spectrally narrow over a range of environmental temperatures.
Frequency upconversion was a later technique that promised higher conversion efficiencies and broader choice of both pump and converted wavelengths. Many materials have energy levels that can serve as intermediate states for multiple absorption of single-frequency photons, allowing infrared photons to stair-step electrons up to levels that correspond to higher energy transitions.
Because fibers can be doped with rare earth elements suitable for upconversion, the interaction length of the pump beam and the conversion medium can be considerable, resulting in higher efficiency and output power. In 1995, SDL Optics (Saanichton, BC, Canada) produced 3 mW of 480 nm light converted from an 80-mW diode laser at 1180 nm by pumping a thulium-doped fiber.
It is perhaps ironic that this technology suffers from lifetime problems, which result from degradation of the conversion medium. The fragility of the doped fibers also was a difficulty. Moreover, these devices attempt to compete in applications where the advantages of gas lasers over most other lasers—in terms of spectral and spatial beam quality—are critical. Efforts continue to provide frequency-converted output in the visible, but mass-market applications such as optical storage still require direct output from diode lasers.
Green-eyed monsters
The III-V nitride family of semiconductors was an interesting possibility from the outset because of their large bandgaps and attractive material properties. The III-V nitrides are robust, as evidenced by their use in high-temperature electronics. But difficult and fundamental problems in producing optoelectronic structures left this research as a small and somewhat isolated effort. Attention turned to II-VI materials, particularly zinc selenide (ZnSe) and its various alloys with cadmium (Cd), sulfur, and magnesium (Mg).
The challenges presented by the II-VI materials were not unlike the challenges of the III-V nitrides. The lattice-matching constraint becomes harder to meet as the laser alloys diverge from the few commonly available substrate materials. Lattice mismatches cause defects that result in the generation of heat, rather than light, from electron-hole recombination, and lead to device failure. In addition, it is extremely difficult to dope ZnSe and related alloys to be p-type.
Advanced device designs developed during the 1980s, and quantum-well (QW) structures in particular, were indispensable to progress in all areas of laser-diode development. The ultrathin structures needed for QW devices were in turn made possible by the advanced deposition techniques of molecular-beam epitaxy (MBE) and metal-organic chemical-vapor deposition (MOCVD). These ultrathin layers also experience less strain from a slightly mismatched substrate and were used to relax the constraint on lattice matching, a critical factor in reducing defects in the crystals.
In 1992, researchers at 3M Co. (St. Paul, MN), using MBE growth on a GaAs substrate and doping techniques developed at the University of Florida, demonstrated the first diode laser in the blue-green. The 3M device was immediately followed by similar ZnSe-based diode lasers at Brown and Purdue Universities. Typical room-temperature CW output powers were a few milliwatts at 525 nm.
Early II-VI lasers had lifetimes measured in seconds. Intensive work over several years by Sony (Tokyo, Japan) has resulted in QW devices using MgZnCdSe that have lifetimes up to 400 hours. The amount of Se in the ZnSe alloys is chosen to provide a lattice match to GaAs, while the magnesium and sulfur fractions are adjusted to achieve the desired bandgap (see figure and Laser Focus World, June 2000, p. 74).
Clearly, these devices are still orders of magnitude away from the million-hour lifetimes common for commercial diodes. The ZnSe materials are very soft, and defects, impurities, and dopants migrate readily through the structure under even moderate current and heating.
Three, five, nitride
The most newsworthy development in both blue LEDs and blue diode lasers is the well-publicized production of GaN based devices by Nichia Chemical Industries (Anan, Japan). The commercial production of blue LEDs in 1993, followed by green LED production, has completed the spectrum of super-bright LEDs across the visible range. The combined market for these highly efficient light sources is projected to soar to billions of dollars, and the potential energy savings from replacing traffic lights alone is many gigawatts annually nationwide.
This success is a tribute to the perseverance of a few researchers, and, among these, Shuji Nakamura (formerly of Nichia) carried the work forward to produce the first blue diode lasers to be offered commercially in 1999. Nakamura used several techniques to overcome difficulties with lattice matching, material quality, and doping. These included
- Two-step MOCVD deposition, which was used to create buffer layers that allowed GaN growth on Al2O3 (sapphire) substrates. A specialized process compensated for flow problems in the deposition created by the required 1000°C substrate temperature. Cree (Research Triangle Park, NC) has fabricated a nitride LED on a conducting SiC substrate.
- P-type doping, which was achieved either irradiating the material with an electron beam or annealing it in nitrogen above 600°C to electrically activate the acceptor dopant.
The robust nature of the nitrides contributes significantly to their success as optoelectronic materials. Their hardness means that defects do not readily propagate through the device structure. The defects themselves appear less likely to create nonradiative transitions (heat sources) in nitrides than in other semiconductors, so a defect density that would produce a failed laser in another material might not harm a GaN-based laser.
The benefits for optical storage may be close at hand. The data density written to the optical storage medium increases inversely as the square of wavelength of the optical stylus. In conjunction with other system improvements, the use of blue diode lasers in next-generation DVD-type devices is projected to lead to another jump in optical storage capacity. With 0.2-µm pits on 0.4-µm tracks read by GaN lasers, these systems have an anticipated double-sided storage capacity of 15 Gbyte, 20 times the capacity of CD-ROM.
Out of the blue
At 400 nm, Nichia's blue diode lasers are really violet and may create lifetime problems for the optics in digital video disks. Moreover, eye sensitivity at this wavelength is only 0.3% of its peak at 555 nm. It is desirable, therefore, to make the wavelength somewhat longer, but this appears to be difficult.
Indium is added to the nitride alloy to increase device efficiency and lower the bandgap (corresponding to 350 nm for pure GaN) to produce visible light. However adding more indium to further lengthen the wavelength may degrade the diode structure. It also appears that piezoelectric effects in the material may play a role in decreasing efficiency as the indium content increases.
Growth on SiC substrates might help resolve these difficulties. SiC is a better heat conductor than sapphire by a factor of ten. Its crystal structure aligns more closely than sapphire to the nitrides, and its lattice constant is a better match, and so should lead to a reduction in device defects. With the market forces of optical storage driving blue diode laser research, it seems safe to predict that the development of low-cost reliable devices will follow the history of red diode lasers.
ACKNOWLEDGMENTS
We would like to thank the following individuals and organizations for providing images used in the Semiconductor Lasers 2000 series timeline: Michael W. Davidson, Florida State University; GE Research and Development Center; Zhores I. Alferov; Nick Holonyak, University of Illinois; Dan Botez, University of Wisconsin-Madison; Lucent Technologies, Bell Labs; Connie Chang-Hasnain, University of California Berkeley; Nichia Chemical Corporation; Jack Jewell, Picolight Corp. -Ed.
Next month the series retraces the development of higher-power diode lasers.