US firms grow UV LEDs on AlN

17 Jun 2002

Researchers in the US have for the first time grown ultraviolet LEDs on single-crystal aluminium nitride.

From Compound Semiconductor magazine.

Researchers at Palo Alto Research Center (PARC), a subsidiary of Xerox, and Crystal IS, a substrate supplier, have demonstrated what they claim is the first UV LED grown on a single-crystal aluminium nitride (AlN) substrate.

The 100 µm device has a top p-contact, a lateral n-contact and incorporates a GaN/AlGaN multi-quantum-well active region. Light is extracted through the bottom of the transparent AlN substrate.

The emission wavelength is close to 360 nm, making the device suitable as a source for phosphor excitation, water purification, or as a compact analytical device for the biotechnology and pharmaceutical markets.

AlN has a high thermal conductivity of approximately 320 W/m.K, which is almost double that of GaN and 10 times that of sapphire, the most common substrate used for the growth of GaN-based LEDs. This provides better heat dissipation during high-power operation.

Other advantages of AlN are that the difference in thermal expansion between AlN and GaN from 1000 °C to room temperature is negligible, and the lattice mismatch is also much smaller than that between GaN or AlGaN and sapphire. This becomes increasingly important for devices that have AlGaN active layers with high Al concentrations, which is required for deep-UV LEDs with emission wavelengths below 300 nm.

Crystal IS supplies AlN single-crystal substrates from 0.5 inch diameter boules. Operating from a leased space at Rensselaer Polytechnic Institute, the company has plans for 2 inch wafers sometime in 2002. At present, substrates are cut and treated using chemical-mechanical polishing and have dislocation densities of less than 10³/cm², at least six orders of magnitude less than typical GaN layers on sapphire.

According to PARC’s Michael Kneissl, the growth technique used reduces the RMS roughness of the surface from 0.83 to 0.38 nm prior to deposition of a graded AlGaN:Si buffer.

“There are a number of issues relating to the device fabrication, as well as growth, that have to be optimized to make a high-output device,” explained Kneissl. “At the moment the drive voltage is quite high because p-type doping at this high aluminum content is hard to achieve. The contact resistance is also not ideal due to the transition layer and the etching down to make a contact to this layer.”

Author
Richard Dixon is news editor of Compound Semiconductor magazine.