Excellant summary, Toby. To answer your question, VG was owned by Fison's, but now they are owned by ThermoElectron. Persons interested in digging up more about manufacturers of epitaxial reactors could have a look at Emcore [EMKR], which went public a few weeks ago. All of the other companies are privately held.
I agree with your earlier statement to the effect that MBE is the exception, not the rule, as far as lasers are concerned, with IBM being the most important exception. For example, although Lasertron did go out and get an MBE system when they started making their IBM-style 980nm lasers, they use MOCVD for the rest of their product line.
Two other recent exceptions: Sharp has recently installed a MBE system for phosphide-based lasers and DVD pickups; and Coherent, through its acquistion of Tutcore, also relies on MBE to produce its lasers. Sharp is probably the largest manufacturer of opto semiconductors in the world. Coherent appears to be squaring off toe-to-toe with SLIC: both offering high power Al-free lasers grown by MBE.
So which technique is better? Not surprisingly, as you suggested, it depends on the application. Conventional wisdom would say that you use MOCVD for things needed in bulk at the lowest possible cost, and you use MBE for more demanding devices. But of course, there are exceptions to THAT rule, too. Basically, I think that, provided you have sufficiently experienced and skilled people running the reactors, both can be made to do just about anything.
Mr. Gottleib says "SLIC has made MBE laser production economically feasable by use of the U.S. Air Force DMS technology which I understand monitors temperature and other elements of the wafer production process. SLIC has an exclusive license the the DMS patent and SLIC claims that with this and other state of the art production techniques they can produce a better product at a fraction of the cost."
The problem with this assertion is three-fold.
One: There are other techniques for measuring temperature and other factors available to other companies which "compete" with DMS. They may not be as good, but they are at least serviceable - otherwise, we would not see IBM, Pirelli, Lasertron, and Coherent already SHIPPING material.
Two: The man who invented DMS, and who was touted by SLIC as an important asset, has left the company. Yes, the DMS patent and equipment is still there, but is there anyone who really understands it and can make it work? Many one-of-a-kind systems of this type are fully comprehended only by the person who dreamt them up. Just becasue it was reduced to practice does not mean that it is now a turn-key process that anyone reasonably skilled in the art can operate.
Three: The DMS system is attached to a single-wafer MBE system of the type that was originally developed for research. If SLIC is going to ramp up its production, it is going to need to step-up to a multi-wafer system. The Coherent and Sharp system referenced earlier are capable of doing five wafers at a time. SLIC acknowledges the need for a multiwafer system in their Annual Report. How are they going to implement DMS on the multi-wafer system (it has never been done), and now that the inventor is gone, who is going to do it for them?
Thanks for the nice plug.
Marie Meyer |
