Can nanolithography ever be a manufacturing technology?

Historically the move to finer dimensions has been accompanied by an enormous reduction in cost per function. Thus despite the cost of new exposure tools, optical lithography is continuing to evolve; transistors with 0.1 μm gates have already been fabricated optically using phase shifting masks and even without such enhancements features of 0.15 μm have been crisply resolved in conventional resist. Thus any sub-optical technology must have the capability of delivering controlled features well below 100 nm to justify the enormous cost of such a technology. There are some special techniques such as self assembly, that can form periodic patterns at nanometer-scale dimensions that we might be able to scale up to useful throughput, but obviously the applications are restricted. Realizing a ‘full service nanolithographic capability’ that can offer arbitrary patterns with well-controlled 50 nm features placed to within a small fractioin of 50 nm and, at the same time, cover 5 cm2 s−1 with less than a few defects per square meter will be an extraordinary undertaking that will have to generate enormous revenue to justify the investment. One possible approach is to use large dense arrays of proximal probes. Even if we have the fashioning capability there are other challenges to making useful product, such as that set by the resistivity of the interconnections.