Sandia and Brown University scientists collaborated on the first vertical-cavity surface-emitting laser(VCSEL) to achieve emission at 380 nanometers in the ultraviolet, while Bell Labs researchers built the longest-wavelength semiconductor laser, which emits at 20-micron wavelength. The Bell Labs laser, called a quantum cascade or QC laser, has a fundamentally different structure from the Sandia VCSEL.

Sandia's 380-nm laser could someday help build solid state "light bulbs" similar to big LEDs, but white. Today, small white LEDs are available. They use 410-nm blue LEDs housed inside a tiny "bulb" coated with white phosphor, just like a fluorescent tube, but the dim "cold" light they emit limits their usefulness. On the other hand, a chip with hundreds of true UV sources could produce the bright "warm" white light of normal-sized light bulbs.

Bell Labs' long-wavelength laser could become the basis of a new type of chemical sensor and analysis system, since the rotational frequency of gas molecules falls into the same frequency region as the QC laser light. That would allow a light beam passing through a gas to identify specific compounds via a selective absorption profile. Since semiconductor lasers can be mass-produced at low cost, the QC laser could form the basis for a new class of chemical analysis products.

Both breakthroughs are the result of incremental improvements to existing laser architectures. The Sandia researchers have found a way to improve the performance of the mirror at the bottom end of the laser cavity. The Bell Labs researchers were building on past successes with the quantum cascade approach, which was invented by Federico Capasso and Alfred Cho, two members of the current team. The quantum cascade effect is able to generate photons in a range of wavelengths, making it a uniquely flexible laser architecture.

Light emission at long wavelengths becomes difficult, since correspondingly thicker semiconductor layers are required to produce the light. The current breakthrough resulted from a new way to guide photons along a single layer. The photons are attracted to the layer by what is known as a surface plasmon -- a layer of photons traveling along the interface between the semiconductor and a metal contact.

Sandia team members said they have not yet achieved an electrically pumped VCSEL but have managed to show that such a device is possible by improving the efficiency of the bottom oxide mirror. "We have finally achieved a laser in the UV range, and though many researchers around the world are also trying to achieve the same goal, we are the first. But I want to emphasize that we have not achieved the holy grail, because our device is optically, not electrically, pumped," said the lead scientist at Sandia, Jung Han. The group hopes to demonstrate an electrically pumped design "within a year," said Han.

The current breakthrough involved shortening the wavelength of the laser beam produced by an optically pumped VCSEL to 380 nm. The previous shortest-wavelength VCSEL was 410 nm. The traditional boundary between blue and ultraviolet is 400 nm, even though 320 nm is the UV range that tans skin. However, the source of excitation for the VCSEL was a big, bulky 355-nm laser shining on the VCSEL microchip.

Sandia has begun work on its next goal: to produce a single-chip solution - the electrically pumped VCSEL that Han referred to as the holy grail.

VCSELs are manufactured by sandwiching thin layers of semiconductors that emit photons when electricity is passed through them. The vertical orientation of the layers is arranged to reflect that light back from the top, as well as up from the bottom, onto the photon-generating microcavity target, thereby prompting it to lase. The reflectivity of the mirrors is directly proportional to the power produced by the resulting laser.

"Our key accomplishment, which has not been discussed elsewhere since we have patents pending, was building an epitaxial bottom mirror -- that was a major engineering accomplishment that has not been duplicated at any other lab that I know about," Han said.

Other VCSEL designers use symmetrical dielectric mirror layers for both top and bottom -- usually an oxide layer, such as titanium oxide. But Sandia used a dielectric hafnium oxide top mirror; and for the bottom layer, instead of an oxide mirror, it used an aluminum-gallium-nitride epitaxial one with almost 100% reflectivity.

"Other labs have had only limited success without an epitaxial bottom mirror, but we mastered the process of building epitaxial mirrors with nearly 100 percent reflectivity and have applied for a patent," said Han.

The problem with building epitaxial mirrors has been stress-induced cracking during annealing. During the fabrication process, chips are raised to thousands of degrees centigrade, and the layers deposited have slightly dissimilar lattice spacing, causing stress between them. During cooling, other labs were getting very low yields because the mirrors were cracking to relieve the stress.

"We have a new process that I can't describe in detail, since we only applied for the patent last month, but I can tell you that it is a recipe you can use with conventional chip fabrication equipment," said Han. "Basically, what we call stress engineering in our patent is a set of special procedures to use during both the growth and cooling phase to prevent epitaxial mirrors from cracking. We are getting 100 percent yields with our stress-management system."

Another advance in the three-year-old project took place more than a year ago, when the researchers added indium to the gallium nitride and aluminum nitride layers, bumping the efficiency of the process from 1 percent to nearly 20%. The introduction of indium lengthened the wavelength of the experimental VCSEL to 380 nm from 360 nm, but the extra efficiency of the device convinced Han to work on shortening the wavelength of an efficient device rather than try to make a shorter-wavelength device more efficient, as did competing labs.

The solid-state light bulb could be enabled by UV VCSELs by building something similar to a fluorescent tube, the inside of which is coated with traditional phosphors but is illuminated by chips covered with hundreds of lasing VCSELS, instead of an electrified glowing plasma as in normal fluorescent tubes. Sandia estimates $100 billion a year in worldwide cost savings, a 120-gigawatt reduction in power generation capacity, and 350 million tons per year less carbon emission to operate such solid-state light bulbs compared with conventional bulbs.

Other applications, however, can utilize the UV laser in its own right. For instance, Sandia's chemical-lab-on-a-chip group plans to integrate the UV laser for detecting radioactive materials and even some pathogens, such as E. coli bacteria. Today, samples must be taken to a 10-cubic-foot UV laser or, for field work, the samples must be tagged with materials that respond to longer-wavelength sources of portable illumination.

"We will work very hard in the next year to create a compact UV laser source, which will benefit groups working on both national security and medical devices to detect chemical and biological pathogens in the field," said Han. Sandia's work was funded by a directed research and development program sponsored by the U.S. Department of Energy's basic energy science program. It was managed on-site by a subsidiary of Lockheed Martin Corp. in Albuquerque, N.M., and Livermore, Calif.

The Bell Labs team included researchers Alessandro Tredicucci, Claire Gmachl, Michael Wanke, Albert Hutchinson, Deborah Sivco and Sung-Nee Chu, in addition to Capasso and Cho. They described their work in a recent issue of the journal Applied Physics Letters.

Additional reporting by Chappell Brown.