Outlook bright for gallium nitride devices

Gallium nitride (GaN) researchers will remember 1997 as the year that 10,000 hours of extrapolated continuous-wave (CW) diode-laser lifetime was demonstrated by Shuji Nakamura of Nichia Chemical Industries (Anan, Japan). It was, therefore, timely that the same fall MRS meeting (Boston, MA; Dec. 1-5, 1997) that honored Nakamura with the MRS medal for his achievements commercializing GaN light-emitting diodes (LEDs)--and soon diode lasers--also included a tutorial session entitled "Industrial Appl

Outlook bright for gallium nitride devices

Gallium nitride (GaN) researchers will remember 1997 as the year that 10,000 hours of extrapolated continuous-wave (CW) diode-laser lifetime was demonstrated by Shuji Nakamura of Nichia Chemical Industries (Anan, Japan). It was, therefore, timely that the same fall MRS meeting (Boston, MA; Dec. 1-5, 1997) that honored Nakamura with the MRS medal for his achievements commercializing GaN light-emitting diodes (LEDs)--and soon diode lasers--also included a tutorial session entitled "Industrial Applications of GaN Diode Lasers." The tutorial comprised three invited talks given by experts in the printing, optical-recording, and display industries who were encouraged to speculate on the future impact of GaN lasers in their respective fields.

Ross Bringans of the Xerox Palo Alto Research Center (Palo Alto, CA), whose grou¥recently demonstrated a GaN laser, presented an overview of the printing industry and its motivation for moving to GaN diode lasers. By the year 2000, he said, the worldwide digital printing market is expected to reach $100 billion. Future printer technologies are driven by the need for speed, color, resolution, power consumption, and multiple functionality--that is, scan, fax, and print functions combined in a single unit. The major motivation driving short-wavelength diode-laser development for printers is reducing the beam spot size while maintaining an adequate depth-of-field (DOF) and optical aperture size.

Bringans used the example of a mid-range 600-dot-per-inch (dpi) laser printer operating at 60 pages per minute (ppm) and observed that current 780-nm output devices in a typical optical system need 6-mm-aperture optics to achieve the 35-µm FWHM spot size required in the scanning direction (see Fig. 1 on p. 16). For true 1200 dpi-- 17-µm spot size--780-nm lasers would require more-expensive and bulky 12-mm-aperture optics and larger polygon scanners, while producing a DOF of only 0.5 mm. A submillimeter DOF requires a more-costly mechanical system capable of maintaining the photoreceptor surface within the laser focus.

Alternatively, a 390-nm-output GaN diode laser could operate at 1200 dpi with a 1-mm DOF using 6-mm optics. A GaN diode laser producing 6 mW of single-mode CW output power is required to discharge the photoreceptor at speeds corresponding to 60 ppm. Higher power will permit increased printing speeds. Rise- and falltimes of about 1 ns are required to allow image enhancement. Finally, a GaN device emitting at a wavelength above 430 nm is preferred to avoid decomposition of commonly used organic photoconductor materials.

Digital video disks

The development of digital-video-disk (DVD) technology was addressed by Akito Iwamoto of the Toshiba Multimedia Laboratory (Tokyo, Japan). DVD-ROM products are now being shipped in volume. They use diode lasers emitting between 635 and 650 nm that can read 0.4-µm pits separated by 0.74-µm tracks to store a total of 4.7 Gbyte of data, readable at a rate of 11 Mbyte/s. Iwamoto also announced that Toshiba would begin shipping 2-Gbyte DVD-RAM drives based on writable phase-change media in January 1998. Toshiba predicts it will shi䂦-Gbyte DVD-ROM and 20-Gbyte DVD-RAM drives by 2005 (see Fig. 2). Iwamoto also reported that the seven-times increase in DVD-ROM storage relative to CD-ROM was achieved via improvements in the diode-laser wavelength, system numerical aperture, and signal processing, with the signal-processing contribution being roughly twice as important as the other two.

Although the recent shift from 780 to 650 nm had a moderate effect on the recording density, the shift to GaN devices scheduled for 15-Gbyte DVD-ROM drives shipping in 2000 is expected to yield a much larger gain. These drives will have 0.2-µm minimum pit lengths on a 0.4-µm track pitch.

The ideal GaN diode laser for DVD-RAM in 2000 is envisioned to have a CW operating point at 60°C as low as 3 V and 100 mA, while producing a single-mode 30-40-mW beam with an aspect ratio of less than four. DVD-ROM has reduced beam-power requirements of 3-5 mW, but otherwise similar specifications. Ultraviolet absorption in the glass substrate becomes an issue with GaN devices, so a wavelength between 400 and 430 nm is targeted, corresponding to at least 80% transmission.

Display technology

The invited talk on the impact of GaN devices on future display technology was presented by Robert Melcher of the IBM T. J. Watson Research Center (Yorktown, NY). His major message was that the potential impact of laser light sources on display technology is enormous, while the obstacles that the lasers must overcome are equally so. Melcher explained that the major issue is cost. Present-day halide-lam¥technology provides efficient 60-l/W white light for $100-$300 including power supply. Consequently, displays based on gas- and YAG-laser technologies are shipping in low volume only in the extremely high end of the display market.

IBM recently announced a reflective liquid-crystal display/silicon projection display in which the potential advantages of replacing the halide lam¥with diode lasers can be seen (see Fig. 3). Currently, the white light is divided into primary colors using dichroic mirrors. Because liquid-crystal spatial light modulators (SLMs) are used, the beams must be polarized, resulting in additional losses and complexity. After re flecting off the SLMs, the three images are recombined within a prism element and projected onto the viewing screen. The optics required for collimating, splitting, polarizing, and recombining three broadband primary-color beams are elaborate and costly, although the result is an industry-leading 2048 ¥ 2048-pixel projected color image.

If the lam¥could be replaced by three diode lasers, the packaging would immediately benefit from simplified power supply and cooling requirements because diodes operate at higher overall efficiency and require only a few volts--versus kilovolts--for halide lamps. Polarized, monochromatic laser output eliminates the need for polarizing optics, eases the specifications on the other optical elements and SLMs, and makes it easier to filter out ambient light. If the diode lasers were pulsed in rapid succession, a single, simplified optical system could replace the three-beam approach shown in Fig. 3 and a more saturated image spanning a broader color gamut could be built u¥from overlaid red, green, and blue images using a scheme called field sequential color.

At the end of his talk, Melcher sketched out the specifications of a laser-based high-definition television system based on 656-, 532-, and 457-nm diode lasers operating at a 6500 K white-point color balance. To achieve a screen brightness of 500 cd/m2 over a 55-in.-diagonal 16:9 aspect ratio screen, assuming 33% optical efficiency and a 3X screen gain, the lasers must produce 6.6, 1.8, and 1.2 W at red, green, and blue wavelengths, respectively. Existing broad-area red-output diode-laser arrays readily achieve outputs of tens of watts. If the blue and green wavelengths are to be provided by diode lasers, GaN devices must be extended to longer wavelengths and their output power drastically improved.

Nichia has supplied blue-output (about 450 nm) and green-output (about 525-nm) GaN LEDs since 1995, but extending diode-laser wavelengths into the visible will be more difficult due to heightened indium gallium nitride (InGaN) active-layer-material quality requirements for lasing. Furthermore, the best GaN diode lasers today are grown on insulating sapphire substrates and fabricated as a mesa structure with two topside contacts. This approach is unsuitable for power laser and broad-area array designs because it requires lateral conduction within the heteroepitaxial GaN.

At the MRS meeting, recent improvements in selective area metal-organic vapor-phase epitaxy of GaN were reported that suggest the sapphire substrate could eventually be removed after growth. Substrate removal, or the use of conductive silicon carbide (SiC) or GaN substrates, would enable broad-area GaN power lasers with conventional to¥and bottom contacts to be fabricated. Once watt-class blue-emitting GaN diode lasers are available, perhaps by 2000-2001, some high-end projection-display platforms should begin to migrate to red- and blue-output devices, while using diode-pumped, frequency-doubled Nd:YAG technology to supply the green.

As Melcher repeated many times in his talk, however, the entire industry will not adopt diode lasers until their cost is competitive with that of conventional lamps, plus whatever additional system costs can be saved by replacing those lamps. But, a mature, growing $2 billion annual projection-display market awaits the successful developer of GaN-based broad-area power lasers.

Toby Strite

ACKNOWLEDGMENTS

The author gratefully acknowledges proofreading and suggestions from Drs. Bringans, Iwamoto, and Melcher.

SAMUEL (TOBY) STRITE is with Uniphase Laser Enterprise, Switzerland; e-mail: tobystrite@ UniphaseLE.com.

More in Lasers & Sources