The demise of sapphire wafers?
"Are we witnessing the demise of sapphire wafers?" I was asked recently. "Not yet!" was my reply.
Resor Associates, Maynard, MA USA
Today gallium nitride (GaN) epitaxial films grown on sapphire wafers provide the best solution for manufacturing light-emitting diodes (LEDs). These chips are used as back lights in all mobile phones, all tablet computers, in most LCD-TVs , and will soon be used in many solid state lighting applications. Yields are improving, costs are dropping rapidly. Many new suppliers have entered the market. The use of LEDs for LCD-TV backlighting has provided a significant jump in unit volume -- an important driver of scale and cost savings. Still people are looking for alternatives. Why?
LED makers want to move to larger wafers so they can lower unit costs. Most LED production is done on sapphire wafers 100mm in diameter and smaller. Semiconductor-grade silicon wafers are readily available at 200mm and 300mm. If LED grade GaN epitaxial films can be grown on large silicon wafers, there could be a clear path to lower unit costs.
Bridgelux Inc. has achieved 135 Lumens per Watt GaN-on-silicon based LED technology
GaN epitaxial films grown on sapphire or silicon have many crystal defects. These defects cause yield loss and decrease the efficiency of converting electric current to light. The ideal material would be single crystal GaN. If large GaN boules can be grown, very efficient LEDs might be built with very few chips lost to crystal defects.
But life is not so simple. LED material systems are very complex. We are operating at the edge of knowledge. It's hard to move fast. Progress is measured in decades, not years. For example, the spacing of atoms in GaN is a poor match to the atom spacing in silicon. Novel buffer layers have been developed to mitigate these problems. Recent results show efficient LEDs can be built. But buffer layers add cost and production is still "2 years" away. In the meantime sapphire suppliers are supplying 150mm and 200mm wafers. It's no longer obvious that silicon wafers are a "must have" technology for LEDs.
Sumitomo and Soitec have joined forces to fabricate high quality GaN engineered substrates. Sumitomo will grow the large boules of GaN. Soitec will use their ion-implant technology to slice off thin GaN films, and then attach these to carrier substrates. The Soitec "slicing" method will conserve expensive GaN material. Still, cost, conversion efficiency and yield remain as serious unknowns.
This competition reminds me of the early days of IC technology. There was a two decade long debate about which technology was best for IC production. Silicon won, not because it was the ideal semiconductor material, but because it moved to high volume and low cost first and kept moving as more suppliers entered the market using silicon wafers. The parallel today in LED manufacturing is sapphire wafers. Sapphire may not be the ideal material, but it is rapidly moving ahead of the competition.
At SEMICON West this year, many of the teams working on these alternative material systems will be presenting their progress. There can always be a breakthrough result. Better check it out.
Solid State Technology, Volume 55, Issue 6, July 2012