A four-way LED substrate battle is shaping up among sapphire, silicon carbide, silicon, and gallium nitride materials.

The prize is much cheaper LED chips and lighting products.

Soraa’s recent unstealthing added gallium nitride (GaN) to the active contender list with its GaN-on-GaN device. On the silicon (Si) front, Osram Opto CTO Ulrich Steegmueller presented the company's Si research at a recent lighting event. Bridgelux, also researching silicon, raised $25 million in funding from China.

There are some dark horses, including glass, germanium, and aluminum nitride (AlN).

Nitride Solutions of Wichita Kansas closed an oversubscribed $2.5 million round A this month for AlN substrates, with an ultraviolet market emphasis.

High brightness LEDs (other than red) are derivatives of gallium nitride. As with semiconductors, generally, you want to build on a stable substrate with an accurate crystal lattice, but gallium nitride is unstable and defect-ridden. The stable alternatives have a lattice spacing mismatch with the GaN and the result is shattered wafers and efficiency loss. A variety of coping techniques have evolved, such as buffer layers. (For a readable backgrounder on the topic, see here.)

The status quo is sapphire. But the cost of making LEDs out of synthetic gemstones has drawn in the other contenders.

  • Sapphire: the defender. Users are almost everyone but Cree; Lattice mismatch: 13 percent; Advantage: Stable, mismatch well researched; Disadvantage: Too expensive for what you get; Bottom Line: Will never meet DOE's 20X cost reduction goal.
  • Silicon Carbide: the Macintosh of substrates. Users are mainly Cree; Lattice mismatch: 3.5 percent; Advantage: Very stable. Low mismatch aided by cancelling thermal mismatch, highest thermal conductivity; Disadvantage: Priciest, almost proprietary; you may have to buy it from your toughest competitor; Bottom Line: Relatively less costly and difficult if you’re Cree, costlier and harder if you’re anyone else. Cree was a silicon carbide company before becoming an LED company.
  • Silicon: Users -- Research maturing in Osram Opto, Bridgelux, China; Lattice mismatch:17 percent (plus a 56 percent additive thermal mismatch); Advantages:  80 percent substrate cost reduction potential, ubiquitous, big wafers; Disadvantages: Lattice and thermal expansion mismatches from hell.
  • Gallium Nitride: Users: Soraa; Lattice mismatch: 0 percent; Advantage: Lattice and thermal match from heaven, homogeneous material allows higher drive level by tuning the GaN for reduced droop; Disadvantage: Unstable, defect-ridden; Bottom line: Analogous to Cree and silicon carbide, relatively less costly and less difficult if you’re Shuji Nakamura, Steve DenBaars, and Jim Speck, much worse if you’re anybody else.


Soraa is performing a bit of a head fake by not selling LEDs, shipping only a finished MR16 halogen reflector where their compact die is able to win with superior beam concentration. However, that is essentially how all LEDs compete with high pressure sodium lamps to this day. 

While we shouldn’t get ahead of ourselves, new substrates could turn some non-adapting companies into stranded whales. Cree may have a particular hazard here, with a unique, almost emotional, commitment to silicon carbide. On the principle that any new substrate will only be adopted if it has a cost or performance advantage, the battle of the substrates will, in the end, only accelerate and enhance the inevitable transition to solid-state lighting.