Bug-killing chips set for production ramp

One of the many futuristic applications touted for compound semiconductor devices in recent years has been the bug-killing capability of ultraviolet (UV) light emitters. Mercury lamps are already used to rid water and surfaces of bacteria, but concerns over mercury pollution, as well as the potential for semiconductor devices to deliver far more compact, efficient and portable systems, has prompted the development of AlGaN-based LEDs.

Columbia-based start-up firm Sensor Electronic Technology Inc (SETI) and its collaborators in Asif Khan s research team at the University of South Carolina have pioneered the development of specialty deposition techniques, epiwafers, chips and even lamps to this end.

For the first time, that technology is now set to make the transition from development to industrial-scale production. When SETI looked to attract new investment recently, the Korean firm Seoul Semiconductor showed its interest. "We needed to find somebody to do high-volume chip manufacturing, and they wanted to become the number-one player in this emerging market," SETI CEO Remis Gaska told Compound Semiconductor.

Gaska struck an equity and supply deal with Seoul Optodevice Company (SOC), the chip-making subsidiary of the parent firm. While he has given up some of the equity in SETI, Gaska remains the majority shareholder. In return, SETI will supply AlGaN-on-sapphire epiwafers to SOC exclusively, from which the Korean firm will process UV emitters operating at five key wavelengths between 340 and 255 nm.

Although SOC already does epitaxy for the visible-range LEDs that its parent company sells, SETI will retain this part of the operation when it comes to the UV devices. SETI s technology is highly specialized and the epiwafers are grown using the company s proprietary migration-enhanced MOCVD approach.

SOC s executive vice-president Jaejo Kim says that the company currently produces around 60 million chip die per month, and that this is set to ramp to 100 million. Exactly how much of that increase will be attributable to UV LED production is difficult to predict at this early stage, but Kim expects that applications such as air and water purification will demand "several tens of millions" of chips per year. At those volumes he believes that the deep-UV LEDs could be manufactured for as little as $10 per piece, depending on the precise wavelength required.

That kind of price should attract the interest of UV lamp and system developers, who have indicated to Compound Semiconductor that a price lower than $10 for a 0.5 W lamp would ultimately be required for applications such as portable disinfection systems to become economically feasible.

"The price of UV LEDs will [now] go down much faster than if we had decided to do all of the production in-house," admits SETI s Gaska, although he points out that the cost will be highly dependent on the specific nature of the application, and the lamp design.

Now focused on production issues such as yield improvements, a ramp-up in epiwafer manufacture and extending device lifetimes, SETI is also set to move up the value chain to improve lamp designs.

Confident of demand for SETI s epiwafers ramping up before the end of this year, Gaska says that there is still a major need to educate potential customers about the wavelengths that are suitable for the different applications that UV LEDs could serve. This is, at least in part, because of the restrictions inherent to using a mercury lamp, which produces light centered at 254 nm but is used for a wide variety of applications that this particular wavelength may not actually be best suited to.

Devices emitting at 340–365 nm can be used to detect biological species, and for the UV curing of materials such as adhesives. Because the aluminum content of the LED is not so high in this range, the devices are somewhat easier to produce. With less strain in the resulting epiwafers, their production can be scaled up more easily.

Between 280 and 320 nm, biomedical applications such as protein analysis are possible. Gaska says 310 nm could become a critical wavelength, because this is the light to which human skin is most sensitive. It could be useful for treating medical conditions such as psoriasis. Radiation below 280 nm is the critical region for germicidal applications such as water purification, and this is the range where UV LEDs are ultimately expected to find their biggest market.

Gaska and colleagues are now focused on improving the lifetime of 280 nm devices, and hope to set a benchmark of 5000 hours by the end of 2006. They are also looking to demonstrate the feasibility of a germicidal lamp based on their own design. "We are looking at the effectiveness of the 254 nm line – there may be a better wavelength to use," Gaska said.

SETI and SOC both have a number of obstacles to overcome, but a successful partnership between the two could unlock the emerging market for deep-UV LEDs.