The lens bottleneck is going to be much more difficult to solve than the drug screening bottleneck. With drug screening, as I understand it, the bottleneck is actually sample and test vehicle preparation. Once you have a sample of a particular drug, and a sample of the test material (virus, abnormal protein, whatever), actually performing the test is fairly straightforward. And so combinatorial chemistry and similar approaches can be very useful because they expedite sample preparation.
In lenses, in contrast, the problem is the fundamental physics of the lens. You have to very slowly heat a very large piece of glass to a given temperature, hold it there for a given time, and cool it down very slowly. The necessary times, temperatures, and heating rates all depend on heat transfer in the glass, defect diffusion rates, and the glass's thermal expansion/contraction behavior. [Insert lots of messy math that I'm too lazy to look up.] Obvious solutions like bigger furnaces don't work because the thermal shock creates more problems than it solves.
Moreover, this kind of issue affects all optical components to some degree, not just those for semiconductor manufacturing. The existing lens companies have a huge incentive to fix the problem, if it can be fixed. And lots and lots of capital equipment is required before you can even do the experiments to work on it. So I don't think this is where the next hot startup is going to come from.
The more interesting alternative would be a solution that avoids the bottleneck by avoiding the lens. Is it possible to write sub-0.18 micron circuit patterns by some means other than projection lithography, while still achieving the throughputs that will pay for all your other process equipment? I'm skeptical, but lots of companies are trying.
Or, less radically, is there an alternative lens design that will achieve the same results with smaller, more easily manufactured lens elements? SVG claims that there is, and they've got it.
Katherine |
