iLED iLED, October issue : 23
ISSUE N°2 OCT OBER 2011 Packaged LED revenue per application (In M$) (LED Packaging Report, Yole Développement, 2011) mm x4 mm. For electrical testing you have driving currents from 5 mA to 20 A. LED structures can have top or bottom contacts, vertical or mesa structures. Also, if you have a fl ip-chip mounted package the contacts are in different places and con fi gurations. It makes it very complicated for LED testing equipment, because you cannot really design a one-size-fi ts-all system.” Assembling fl ip-chip packages has been a key area of innovation in die-attach equipment, and is one that is being increasingly exploited. “Lumileds was the historical leader, the fi rst to develop fl ip chip bonding and for many years they were the only one,” Virey said. “Now we're seeing more companies coming out with fl ip-chip or fl ip-chip-like die attach. For example, Cree and Everlight have also recently begun using fl ip-chip systems.” Meanwhile, die singulation companies using laser technology could compensate for the impact increased die sizes might have on them by providing laser lift-off tools. This process directs a high-intensity laser through the LED's sapphire substrate, targeting the LED's GaN epilayers. This locally breaks down the GaN into nitrogen gas and gallium metal, creating a shockwave at the interface that separates the epilayers from the sapphire. “There is de fi nitely an increasing number of companies using laser lift-off,” Virey said. “We've identi fi ed at least eight doing laser lift-off with vertical structures. Also, I was surprised, but this process can have a very high yield. It's tricky to develop a stable process, but once it's done the lift-off is close to 99 per cent yield.” Such high yielding processes will become more critical to manufacturers' success as competition increases in the LED market, and pressure builds to fi nd advantages over their rivals. Perhaps the largest area where there is scope for improvement in this regard is MOCVD yields. “The industry still pretty much throws away 50 per cent of the chips that are being made at the epitaxial level,” Virey noted. “The ugly truth is that those die still have to be diced and separated. Right now, they're singulating all the die that they know aren't good. This number is going to decrease.” Though higher yields would clearly be good news for the LED industry overall, testing and die singulation equipment producers might be excused for having mixed emotions about it. That's because the degree of reliance on their tools will decrease. “Basically, there will be many fewer wafer starts needed for a certain number of chips sold at the end,” Virey commented. “It's going to reduce the opportunity for LED testing and die singulation, but there should be signi fi cant improvement in manufacturing yield. Another positive is that the cost of ownership of LED solutions will decrease, which will speed up adoption and therefore generate more demand.” www.yole.fr Eric Virey holds a Ph-D in Optoelectronics from the National Polytechnic Institute of Grenoble. In the last 12 years, he’s held various R&D, engineering, manufacturing and marketing positions with Saint-Gobain. Most recently, he was Market Manager at Saint-Gobain Crystals, in charge of Sapphire and Optoelectronic products. 23 i LED
Publication List
Using a screen reader? Click Here
