Nova Crystal's Little Secret
When Nova Crystals Inc. announced a 1310 nanometer vertical-cavity surface-emitting laser (VCSEL) a few months ago, it may have been hiding its light under a bushel.
The laser was the most powerful of its type ever produced (see Nova Crystals Demos High-Power VCSEL), but it might also have been something even more significant – the first commercial product to be made using a new technology for fusing together different types of semiconductor material.
If this is the case - and Nova Crystals won’t confirm or deny it – then it might represent a big breakthrough. It addresses a big problem facing developers of optical integrated circuits - that different types of optical device are best implemented in different materials. This makes it very tough to make a complete system comprising different types of device on a single chip.
Right now, some startups are tackling this problem using indium phosphide, one of the few semiconductor materials that can be used to make both active and passive optical devices. However, indium phosphide is a difficult material to work with.
Other vendors are aiming to make devices in different materials and then butt them together to create a complete system. But having to manufacture each piece separately and then precisely align them bumps up costs enormously.
A third solution to the problem was invented by Nova Crystal’s founder and chairman, Yu-Hwa Lo, some time ago. In this case, the different materials are bonded together before the manufacturing process, avoiding the business of having to assemble things afterwards.
This wafer bonding process works well for some mixtures of materials but has proved unworkable for others. In particular, it’s a nonstarter for integrating lasers, made out of gallium arsenide, and their associated electronic circuitry, made out of silicon. Bonding these two materials together requires high temperatures - 700 degrees C – and as the assembly cools, thermal mismatches and residual stress can cause cracking and debonding.
This is where Nova Crystals’ high power VCSEL comes in. It’s unlikely to have been made using the wafer bonding process, and it’s known that the startup has another bonding process up its sleeve - a process that doesn't require high temperatures to work.
It’s called "compliant universal (CU) substrate" and it was invented by another couple of Nova Crystals’ founders, Lester F. Eastman, a professor at Cornell University, and Felix Ejeckam, Eastman's doctoral student.
Ready for some details?
All semiconductors are made of atoms positioned in a regular crystal lattice. To bond one type of semiconductor to another, the atoms in the crystal must line up. If they don't (and there aren't that many winning combinations), then the structure becomes stressed. It develops minute cracks at the atomic level, which kill light emission, or it could even crack and fall apart, especially after heat treatment.
Nova's CU substrate is a kind of glue. Put simply, it's a layer that can stretch to accommodate the mismatch between the crystal lattices of the two materials to be bonded together. But it's a lot better than glue. It's chemically the same as the substrate, so it can conduct electricity.
Nova's 1310nm VCSEL probably makes use of this idea, although the startup won't confirm it. Making long-wavelength VCSELs is a particularly knotty problem that’s beset researchers and manufacturers for the past 20 years. The key is how to grow both the mirrors and the active (light-emitting) region in the same process.
A research paper cited on the startup's Website points to a method of making a 1310nm VCSEL. Published in the journal Applied Physics Letters in 1998, it describes a gallium arsenide-based VCSEL on a CU substrate. 1310nm emission is obtained from a compound of indium and gallium arsenide. Normally, this material cannot be grown on a gallium arsenide substrate because the lattices are too different, resulting in high stresses. However, use of the CU substrate relieves this stress to make the impossible possible.
All the signs point to Nova Crystals having translated theory into practice with its 1310nm VCSEL, samples of which are already being shipped. But if it has achieved this breakthrough, why is it keeping quiet about it? When Light Reading finds out, we’ll let you know. -- Pauline Rigby, senior editor, Light Reading http://www.lightreading.com