By ROBERT PEASE
More than a year after embarking on a cooperative research and development agreement, two firms announced a successful technological milestone in the field of vertical-cavity surface-emitting lasers (VCSELs). Cielo Communications (Broomfield, CO) and Sandia National Laboratories (Albu querque, NM) demonstrated the industry's first electrically pumped long-wavelength VCSEL for 1.3-micron applications.
The demonstration confirmed singlemode and continuous-wave operation at 1.294 microns, ideal for telecommunications equipment applications. The VCSEL was also developed to meet the standards for high-speed SONET optical interfaces that specify operation around 1.3 microns. Cielo says the new VCSEL will significantly decrease the cost of delivering high-bandwidth optical interconnects in the telecommunications and Internet infrastructure. Other markets that could benefit from the breakthrough include Fibre Channel, Gigabit Ethernet, 10-Gigabit Ethernet, and fiber-to-the-home (FTTH).
VCSEL technology has been researched for years as an alternative and less-costly light source for use in short-distance applications. Their smaller size, lower power, lower cost, and good reliability enabled their popularity to soar over the years. But most of today's VCSELs run at 850 nm, although the market began in the 980-nm range. But there was always a dream of finding a migration path to 1.3 microns.
"This is a special number, because it's a wavelength in which the glass material in your fiber-optic cable has minimum dispersion," says Bob Mayer, vice president of marketing at Cielo. "That enables the signals to traverse down the path a lot easier, and the edges of your signals don't blur."
Manufacturers first promoted VCSELs at 980 nm because the receiver on the other side could "see" the 980 nm the same as a 1.3-micron. It was believed this would enable easy migration and systems upgrade to the 1.3-micron VCSELs once they were achieved. However, 980-nm VCSELs were never fully adopted in the industry because the cables that were being manufactured, as well as most of those already installed, were never qualified for use at 980 nm. Rather, they were qualified for use an 850 nm or 1.3 microns and the standards bodies pushed to adopt these two wavelengths into the 1-Gbit/sec standards.
"One push was for the 850-nm VCSEL, because that's where cables were qualified and vertical-cavity lasers could be manufactured," says Mayer. "The other push was for the 1.3-micron VCSEL, because that's where the edge emitters were already operating and producing optimal performance. People have been working on 1.3-micron VCSELs for 10 years. It's been a dream of everybody. We're just extremely excited that we finally got it to work."
Cielo and Sandia came together in March 1999 to develop a 1.3-micron VCSEL using gallium arsenide (GaAs). Cielo is a privately held company that researches and develops VCSEL technology and optical packaging. Sandia National Laboratories is a multiprogram laboratory operated by Sandia Corp. a Lockheed Martin Co., for the U.S. Department of Energy.
"Our partnership has quickly demonstrated the impact of our combined technology strengths on the telecommunications industry," says Peter Esherick, manager of the compound semiconductor materials and process department at Sandia National Laboratories. "By obtaining the optical spectrum emission near 1.3 microns, this innovation enables extended distances and data rates to be reached over singlemode fiber at lower costs. In addition, the direct electrical modulation of this device provides instant improvements in speed and reliability over optically pumped approaches. The VCSEL structure enables the ability to create arrays, thus simplifying the overall process of achieving higher bandwidth."
Current 1.3-micron lasers are edge-emitting, either Fabry-Perot (FP) or distributed-feedback (DFB) types. Both must be completely fabricated prior to operational testing. A 1.3-micron VCSEL has significant advantages over edge-emitting lasers in the area of lower manufacturing, packaging, alignment, and testing costs. Other benefits include lower power dissipation and higher reliability. VCSELs can also be grown in arrays to maximize density and bandwidth performance.
Cielo believes the significant cost reduction will result in making increased bandwidth more accessible and cost-effective for telecommunications and the Internet, likely rendering the FP and DFB lasers obsolete for many applications.
Others have attempted to demonstrate long-wavelength VCSEL capability, but have fallen well short of the 1.3-micron objective or have used wafer-bonding techniques, according to Cielo. Wafer bonding requires a second VCSEL as an optical pump for operation. Cielo's VCSEL structure was grown in a single process, similar to that used for 850-nm VCSELs. By growing the structure on GaAs and taking advantage of Cielo's existing GaAs-based mirror structures, the new VCSELs promise simpler manufacturing, improved reliability, and lower overall device cost.
"The goal of our program was to make this a single-laser device-a single, monolithic, electrically pumped 1.3-micron VCSEL grown from the gallium arsenide system and meeting the very stringent SONET specifications," says Mayer. "By growing this VCSEL in one system, we simplify the manufacturing process. It eliminates the wafer bonding approach that cannot support electrical pumping. That makes this announcement very important. We've grown a very simple structure that is performing at the right wavelengths, but is very similar to what the VCSELs look like at 850 nm. Nobody has ever done that before."
While the new VCSELs may challenge FP and DFB lasers in certain applications, they may also replace their lower-wavelength counterparts in others. The use of VCSELs has been limited to relatively short distances because of the 850- and 980-nm operating wavelengths, particularly at very high speeds. The 1.3-micron VCSEL is expected to support 10-Gbit/sec data rates over 10- to 15-km distances.
"That's a big improvement over the 850-nm device that could not be used on singlemode fiber and could only operate at 1 Gbit/sec for about 220 m," says Mayer. "The advantage of going to 1.3 microns is that you're hitting the 'sweet spot' of the performance on the fiber."
Both companies will continue working to bring this long-wave VCSEL technology to market by the second quarter of 2001. Cielo sells integrated modules that have the laser driver and accompanying receiver, so original equipment manufacturers (OEMs) can quickly adapt to their systems. The next step is to prove the reliability of the Cielo/Sandia 13-micron VCSEL structure. And the outlook is promising.
"Hitachi has already proven high reliability with the active region that we've used," says Mayer, "and we're using our own already-proven, highly reliable mirror structure. So while we've got some work to do, we feel pretty good about what the results will be."