>>PGILD permits single step doping. An excimer laser is used to heat those precise
geographies where the dopant is required. This is done in a process chamber
containing the required concentration of the dopant. Thus much wafer processing,
movement, handling is removed. It could substantially lower the costs associated
with making chips relative to the current processes, chemicals etc to achieve the
same result.<<
Pretty good description. The big benefit isn't cost, though. Laser annealing theoretically allows much more precise junctions, therefore tighter CD control. Better CD control means faster chips, and a narrower distribution of speeds around the mean, both of which translate *directly* to increased selling price for the chip maker. Even further in the future, laser annealing supports things like raised junctions that are not even possible with current implant-and-RTP methods.
So why isn't the marketplace eating this stuff up? You'd have to ask UTEK about that one. However, they are definitely not marketing it as aggressively as they might, for reasons that are not quite clear to me.
Re: someone else's comments about PGILD missing the window for Intel's first 300mm fab, and therefore missing all Intel fabs for that generation due to Copy Exactly. I have heard rumors that Intel is deviating from the Copy Exactly protocol in order to increase flexibility. That makes sense: the sub-0.25 micron (and especially sub-0.18 micron) process picture is pretty confused at the moment, and the "best" process depends very strongly on what sorts of devices you are making. IMO, it would be premature for Intel to rule out process changes from fab to fab.
Katherine |
