Figure 1. Xerox's Palo Alto Research Center has yet to significantly extend the lifetime of its blue diode laser from its original success of 1 hour. However, several new research directions could change that in the near future. Photo © Tom Tracy.
Blue diode lasers could easily double or triple the storage capacity of optical data storage devices, bolstering a flagging growth rate in one of the world's largest optoelectronic industries (see this issue's Industry Focus). With a little more power than that required by the data storage industry, blue diode lasers could also double the resolution of laser printers and scanners, opening new markets for commercial-quality scanners with medical-quality performance. Other potential applications include laser projection displays and military telecommunication devices.
But, you may ask, if blue diodes are so important why hasn't industry succeeded in building them? Unfortunately, blue diode lasers are still plagued by inadequate substrates, crystal layer dislocations, and defects that increase over time with GaN's requisite high drive currents. The ongoing difficulties mean that despite widespread interest, a commercial line of blue and semi-blue diode lasers has yet to emerge.
Scientists around the world are shaking their heads and wondering if blue-diode-laser development paths are heading in the right direction, or if commercial success will once again depend on something completely new.
Active development of blue diode lasers has been ongoing for more than a decade, starting with II-VI compounds such as ZnCdSSe or ZnMgSSe. That changed with the work of one man.
Several years ago, Shuji Nakamura of Nichia Chemical Industries Inc. (Tokushima, Japan) turned his interests to III-V compounds, including GaN. Reportedly, Nakamura simply wanted to go into a new area of research. Nakamura's initial announcement of a pulsed nitride laser in 1995 was followed by similar announcements by Fujitsu (Tokyo, Japan), CREE Research (Durham, NC), Sony (Tokyo, Japan), Xerox PARC (Palo Alto, CA), NEC (Tokyo, Japan), Matsushita (Secaucus, NJ), Toshiba, Univ. of California at Santa Barbara (Santa Barbara, CA), Hewlett-Packard (Palo Alto, CA), SDL, Inc. (San Jose, CA), Pioneer (Tokyo, Japan), and Samsung (Korea); (Figure 1). Today, Nakamura is years ahead of his competitors. This summer, following several missed deadlines and optimistic communications, Nichia released the first commercial "samples" of a violet diode laser with a peak emission near 400 nm.
Most commercial companies have completely abandoned II-VI compounds. However, exceptions include 3M (St. Paul, MN) and several institutions in Germany. Although some of these groups are having success in developing blue LEDS and green diode lasers with short lifetimes (~400 h), continuous-wave (CW) blue diode lasers appear to be the sole province of the III-V group of semiconductors.
Group III-Nitride compounds, commonly referred to as 'nitrides' in semiconductor circles, exhibit bandgap energies of 1.95 to 6.2 eV that include optical emission from 650 nm down to UV wavelengths at 200 nm1. Group III-Ns, however, have a large lattice mismatch with standard semiconductor substrates.
Common substrates include conductive hexagonal silicon carbide (SiC) and insulating sapphire. However, high dislocation densities (109 to 1010 cm-2) between the substrate and GaN layers result in defects that can lead to either inoperative devices or short lifetimes of the diode lasers, despite the fact that group III LEDs made of the same material do not seem to suffer greatly from such dislocations. Marek Osinski of the Univ. of New Mexico's Center for High Technology Materials (Albuquerque, NM), said part of the reason is that the high drive voltages can lead to P-contact metal migration to the p-n junction, resulting in device degradation and failure.
Cree Research Inc. and Fujitsu have both used SiC substrates to create GaN blue diode laser devices. Robert Pomeroy of Fujitsu's Global Public Relations department said the company has not had any significant developments since its announcement in 1997. While breakthroughs have not been forthcoming, interest continues in the use of SiC as a substrate material. SiC is compatible with existing semiconductor manufacturing processes used to make red and IR lasers.
Using metal organic vapor phase epitaxial (MOVPE) growth methods, Fujitsu was able to create a MQW device with a threshold current of 800 mA at 414 nm. The diode laser achieved a maximum output of 20 mW with a five-hour lifetime at 5 mW. At that time, adequate SiC substrates were considered a limitation to development (the substrates were only 0.4 to 1 in. in size).
Needed: new crystal growth procedures
Figure 2. Sony used a raised-pressure metal organic chemical vapor deposition (MOCVD) to create its first blue diode laser. Experts say that an even higher pressure deposition system may lead to improvements over today's devices.
Several of the aforementioned companies are looking at sapphire substrates as a base for GaN blue diode lasers. Companies such as Nichia, Sony (Figure 2), and Pioneer have all used sapphire as a starting point for GaN growth. However, while sapphire offers good thermal dissipation (0.35 W/cm-K), GaN on its own is far better (1.3 W/cm-K).
Osinski said blue diode lasers using conventional sapphire substrates exhibited material defects that limited the device lifetime to approximately 300 hours. To move beyond that point, scientists developed photomasking techniques that isolated small areas of the substrate. Epitaxial growth in these small areas were seeded with GaN through epitaxial lateral overgrowth (ELOG) processes that, eventually, led to larger, dislocation-free regions over the masked areas.
Nichia and other groups have taken this process a step further by continuing the ELOG growth beyond the point where the isolated areas eventually coalesce. By continuing the ELOG process, the GaN layers can reach a thickness of 100 to 200 µm. With such a thick layer, the original sapphire substrate can be completely etched away, leaving a GaN conductive substrate with a perfect lattice match for the following bandgap layers and higher thermal dissipation.
Looking at the work of Nichia, David Bour and associates at Xerox PARC hope future experiments using ELOG will improve their device's performance. Curently, however, Bour's group is concentrating on improving the GaN material quality, using metal organic chemical vapor deposition (MOCVD) on sapphire substrates to create multiple quantum well structures.
"We've used a ridge waveguide structure," Bour said. "The best results achieved are with 2- and 3-µm ridges. We use coated mirrors to get the reflectivity upour best CW thresholds are around 8 kA/cm." He said these high threshold current densities make packaging all the more important.
"Nakamura is achieving threshold current densities of about 3 to 4 kA per centimeter, but those are still very high compared to IR or red laser diodes," Bour said. "That's why packaging is more important, as well as fabricating narrow ridges. In red or IR diodes, you're usually dealing with 4- to 6-µm-wide ridges. For nitride ridge waveguide lasers of those dimensions, you wouldn't have a single- mode laser and the heat dissipation is poor. We must use narrow ridges to get the performance you need, to promote lateral heat dissipation and single-mode operation."
Reproducibility is a crucial problem for GaN diode lasers. Bour said, "These devices are much less reproducible than arsenides and phosphides used in the IR or red lasers. That's a consequence of the MOCVD growth and the more difficult processing required for the nitrides. Growth is performed at a high temperature and in ammonia gas, and that tends to create a convection-driven turbulence in the reactor, which makes the process somewhat chaotic, a little different from run to run."
A new direction?
Although Nichia has produced devices using ELOG that exceed 10,000 hours operational lifetimes at 20 deg. C and 5 mW, Osinski said a new approach may actually bring the device home.
Osinski writes, "In spite of these very encouraging data, ELOG substrates may not, after all, offer the ultimate solution for III-N lasers. [The] present offering of diode laser 'samples,' introduced by Nichia in January, illustrates the difficulties that still existNichia only offers single samples at a very high cost (about $2000 per laser) for 'evaluation purposes.' This indicates that while the ELOG approach is capable of producing champion devices, it may be very difficult to achieve sufficiently high yield, necessary for low-cost high-volume manufacturing."3
Instead, Osinski draws attention to several new processes in the GaN field, such as high-pressure growth (HP), sublimation growth (SG), hydride or halide vapor phase epitaxy (VPE), pendeo-epitaxy, and nanohetereoepitaxy.
As stated earlier, GaN crystals can be created by starting with a sapphire substrate, laterally overgrowing thick layers of GaN and then etching away the sapphire. However, a commercial process will have to reduce the time and cost of these procedures.
High-pressure growth, developed by the High-Pressure Research Center (Warsaw, Poland), has delivered the highest quality GaN crystals with dislocation densities of less than 104 cm-2. However, Osinski said the process is expensive and takes approximately 200 hours to grow a 100-mm2 crystal. MOCVD homeoepitaxy on SG crystals are even smaller than those produced by HP methods, although they have resulted in "improved-quality materials." Finally, Osinski characterizes VPE-grown GaN as "relatively easy and inexpensive" but has yet to beat the optical quality of the structures produced on sapphire.
Pendeo-epitaxy on SiC substrates may suffer from the same defect qualities as other ELOG processes. "The lateral overgrowth proceeds from side walls of GaN trenches etched down in the conventional epitaxial layers," Osinski said. "Most likely this approach will suffer from...difficulties with obtaining high yield from the wafer that contains alternating regions of high-quality and low-quality material."
Nanoheteroepitaxy and other compliant substrate methods may hold some future promise. These methods use different techniques to create high, narrow ridges on silicon-on-insulator, lithium gallate (LiGaO2), and other materials. These ridges absorb the stress created by lattice mismatch in subsequent layers.
Today, no one can say for sure what the final solution will be to the blue diode laser question. Manufacturers of optical storage devices are struggling to create new optical heads and other approaches that push red diode laser-based systems further up the recording-capacity scale. These and other external factors may give developers of blue diode lasers the time they need to find a new approach to the blue diode laser problem. Certainly, with optical storage devices accounting for $500 million in revenues this year, the pressure is on to find a "next-generation" solution.
However, even if a 2- or 3-mW diode laser appears on tomorrow's horizon, problems remain for the scanner and printing industries. Adding one more challenge to an already tall stack, Xerox's Bour estimates that blue diode lasers would have to reach 10s of mW and operate in multiple beams in order to meet the needs of his industry.
1. G. Foulon, A. Alexandrovski, R. Route and M. Fejer, Stoichiometric and Oxidized LiNbO3, Stanford University, 1998 Annual Report, Center for Nonlinear Optical Materials, Stanford, CA.
2. Kenji Kitamura, Shunji Takekawa, Yasunori Furukawa, Improvement in performance of optical crystals for frequency conversion devices, NIRIM 1998 Annual Report.
3. Marek Osinski, Green, blue, and beyond: current status and future prospects for short-wavelength diode laser development, Center for High Technology Materials, Univ. of New Mexico.
CREE Research Inc.
4600 Silicon Drive · Durham, NC 27703
Phone: (1) 919/313-5300
Fax: (1) 919/313-5452
3000 Hanover St. · Palo Alto, CA 94304
Phone: (1) 650/857-1501
Fax: (1) 650/857-7299
Matsushita Electric Corp. of America
One Panasonic Way · Secaucus, NJ 07094-2917
Phone: (1) 201/348-7000
Fax: (1) 201/348-8378
7-1, Shiba · 5-chome Minato-ku
Tokyo 108-8001, Japan
Fax: 03-3798-1510 / 03-3798-1511 / 03-3798-1512
Nichia Chemical Industries Inc.
491 Oka · Kaminaka-Cho, Anan, 774-8601
Landic Toranomon Bldg. · 7F 3-7-10, Toranomon 105
Phone: 33 438-4731
Fax: 33 438-4730
Samsung Electronics Co. Ltd.
Samsung Main Building 9F,
250, 2 Taepyong-Ro, Chung-Ku, · Seoul 100-742, Korea
80 Rose Orchard Way · San Jose, CA 95134-1365
Phone: (1) 408/943-9411
Fax: (1) 408/943-1070
Sony Semiconductor Company of America
3300 Zanker Rd. · San Jose, CA 95134
Phone: (1) 408/432-1600
Toll-Free Phone: (1) 800/288-SONY
Fax: (1) 408/955-5176
Toshiba America, Inc.
1251 Avenue of the Americas, Ste. 4100
New York, NY 10020
Phone: (1) 212/596-0600
Fax: (1) 212/593-3875
Univ. of California at Santa Barbara
College of Engineering · Engineering I, Room 1038
Univ. of California · Santa Barbara, CA 93106-5130
Phone: (1) 805/893-3207
Univ. of New Mexico
Center for High Technology Materials
1313 Goddard SE · Albuquerque NM 87106, USA
Phone: (1) 505/272-7800
Fax: (1) 505/272-7801
Xerox Palo Alto Research Center (PARC)
3333 Coyote Hill Road · Palo Alto, CA 94304
Phone: (1) 650/812-4000
R. Winn Hardin
R. Winn Hardin is a journalist based out of Jacksonville, FL.