I was searching other inventor's patents for info about the problem POET is having moving forward with the 100nm. I looked at this one:

http://www.google.ca/patents/US5093280

This might be a clue as to why POET is seeking a non-gold contact:

"Almost every semiconductor electronic substrate has the requirement for ohmic contacts to provide electrical connection to the various electronic components thereof formed in the semiconductor substrate. The pursuit of ohmic contacts has been the subject of much study. In general, for gallium arsenide the best contacts have been fabricated by an alloy techniques where a doping species, such as zinc, is coevaporated with a more noble metal, such as gold, onto the contact area. The deposition is then heat treated for a brief period to allow the zinc to heavily dope the interface layer between the gallium arsenide and the noble metal. The drawback to this process is that the heat treatment temperature must be relatively low to limit disastrous alloying- of the noble metal with the gallium arsenide. Indeed gold forms a liquid phase with gallium arsenide at relative low temperatures of about 350° C. and is known to give rise to a non-uniform interface with the gallium arsenide"

I know that POET uses a high temperature annealing step which Taylor said is one of the things that sets POET lasers apart from competitors (He said something like "nobody would dare subject a laser such a high temperature anneal"). Maybe at such a small feature size as 100nm, the effect of using gold is more pronounced.

This does nothing to diminish my confidence in Taylor's team. They have knocked out numerous problems without any market awareness or concern. I still remember how casually (jokingly almost) Dr. Taylor spoke of the "bowing" of the wafers during the annealing step and how irritating it was to have BAE's wafers bowing when they were annealed at UConn, and vice versa. Taylor's team will overcome.

Importantly for those who *are* concerned, there are other non-gold ohmic contacts that have already been patented which proves that it can be done.

http://www.google.ca/patents/US5480829

Note that this patent explains why lift-off is also something undesirable, just as our NR today suggested:

"The complementary nature of the devices of course implies that they include both N-type and P-type devices on the same substrate. The typical ohmic metallization of the prior art is different for N-type and P-type devices. For N-type, gold germanium nickel (AuGeNi) has been used. For P-type, gold zinc nickel (AuZnNi) or gold zinc gold (AuZnAu) has been used. Both ohmic contacts rely on Au metallization for low resistance, resulting in several disadvantages. The Au-based ohmics are not etchable. Consequently, they must be defined on the device by a lift-off process. As is well understood in the art, lift-off is associated with relatively poor yields and poor pattern definition, and thus limits device geometries to relatively large dimensions. The large dimensions decrease the maximum number of devices which can fit in a given space."

And again with the need for non-gold ohmic contacts:

The Au-based ohmics of the prior art also exhibit the disadvantage of being incompatible with modern Al VLSI metallization used for multi-layer integrated circuit interconnections. The incompatibility stems from the fact that the Au-based ohmics fail to provide a stable ohmic contact through the temperature range required for the Al multi-layer technology. Specifically, the typical Al based interconnect processes occur at or above about 500° C. while Au-based ohmics fail to provide stable ohmic contacts above about 470° C. Consequently, the Au-based ohmics of the prior art limit the commercial application of the complementary devices. Furthermore, over time Au tends to react with the Al, creating a problem known as purple plaque which eventually causes failure of the device."