Electron Beams
Whatever happened to research on electron-beam lithography?
A long-awaited breakthrough in chip-making technology is inching closer towards commercial reality. But progress is slow and remains uncertain. At Thursday's Research@Intel event in Mountain View, California, The Reg and other press folk sat down with Mike Mayberry, Intel's VP of technology and manufacturing, to discuss the …
Same problem EUV is experiencing now, only more so. It's just too bloody slow. The electron beam is also a very touchy thing to get right--too weak and its energy could be disperse in the target or the beam is too unstable for high resolutions; too strong and you'll get the electron version of spalling.
The distinction between EUV and x-rays is arbitrary and set at 10nm, they are both generated by electrons changing energy level, as distinct from from gamma rays which are nuclear in origin.
UV is typically from valence (i.e. outer shell) electrons, whereas X-rays typically originate from inner shell transitions (like when an electron beam removes the inner most electron from a metal atom in a target and the other electrons cascade down to fill the gap).
@ AB3
yes UV less than 200nm is only usually used in a vacuum... for this reason this section of the spectrum is also known as the "vacuum ultraviolet" as all wavelengths in this range are strongly absorbed by oxygen in the air.
Plus I beleive the actual lithographic processes of chip manufacture have been performed in the absence of air for donkeys years...
One thing to watch out for is that Intel doesn't use standard descriptors for the technology nodes. The aim is to introduce EUVL at the 32 nm technology node, which equates to 2013 Year of Introduction in the international technology roadmap for semiconductors. Intel's 32 nm design size equates to the 50 nm technology node. The 32 nm technology node equates to Intel's 16 nm design size (which is also slated for 2013 same as everyone else). Intel have being slated EUV's introduction for 2013 since at least 2006.
@Don Mitchell - Still a couple of companies working on that, Mapper Lithography (parallel beam system) and KLA Tencor (Reflective multibeam system (REBL)).
@ Frank Ly - Yes soft xrays - I'm told the name change was because the public doesn't like things that sound like radiation (ie why the NMR machine in your hospital is called an MRI)
@Charles 9 - Mainly the speed for ebeam - it's not particularly difficult to use (ie to get a good beam) if you don't mind it slow - all the masks for the other techniques are made with ebeam. EUV is a lot faster than ebeam, but still has a way to go to meet the wafer throughput of optical. Being addressed via strong light sources and faster photoresists.
@A B 3 - EUV and ebeam are both vacuum techniques
@ AC - 13.5 nm for EUV lithography
Yes soft xrays - I'm told the name change was because the public doesn't like things that sound like radiation (ie why the NMR machine in your hospital is called an MRI)
Silly really, I would have thought that a public looking at bleeding edge tech would understand, said tech..(proof from point being raised on tech website!).... I guess the marketing guys did a weekday highstreet survey outside poundstretcher and found people who still refuse to have microwaves in thier houses due to the nucular praticles....
In the short term why don’t you use laser interference? Fire one laser, interfere with another one down the same axis and the result is a much smaller spot. Cheap, simple, fast and obvious – i.e. it’s not worth patenting, so nobody is going to persue it!
Look out for the trolls, I just mentioned patents ;-)
@ Dave_H
People do use interference. But unless you want regular arrays of lines, (or regular arrays of dots for 4 beam interference), it's no good (ie you can't produce an arbitrary pattern). Also the aerial image contrast is dreadful.
@soft electrons
Energy distribution in resist is reasonably well understood wrt secondary electrons. Also for the most part, (until you get to doses so low that the shot noise limit is a problem) the overall dose deposited is not variable (although where it is deposited within the film varies a bit). Current EUV and EBL resists seem to show a secondary electron thermalisation length of around 5 nm, which is typically less than the acid diffusion length anyway. So at this stage this is not a problem (give it ten more years though).