Stretching Optical and Stretching Thin
Aaron Hand, Managing Editor -- Semiconductor International, 2/1/2004
I've said it plenty of times before, and so has just about everyone else — optical lithography is still alive. It's a remarkable feat, so I guess we just have to keep shaking our heads in amazement and lauding those that have made it possible.
The incredulity expressed by many over the years is accompanied by a certain invincibility from the lithographers. Can optical lithography make it to the 65 nm node? Of course! Can those wily little photons make it to the 45 nm node? Bring it on! While you're at it, bring on 32! And 22! That's an awful lot of exclamation points, I know, but that's the kind of bravado lithographers are bound to display — and deservedly so.
The industry's engineers should be proud. Heck, photolithography wasn't even supposed to make it into the submicron era, but look how far it has sailed past all the obstacles. Perhaps it's that feeling of invincibility that makes the planners think the industry can develop any number of technologies — all at the same time. Each lithography tool supplier is out there developing the whole gamut of next-generation techniques, including 157 nm, e-beam and EUV lithography, while simultaneously devoting energy to the very solutions that may obviate the need for much of the other work. Nobody wants to be caught empty handed if one of these techniques takes off and they haven't been preparing for it.
The 2003 edition of the International Technology Roadmap for Semiconductors (ITRS) was released in December and, thankfully, it carries two fewer lithography solutions going forward. Proximity X-ray and ion projection lithography have both been pared from the list. But wait, there is still the same number of potential solutions because two more have been added — immersion and imprint lithography. Not that immersion and imprint don't deserve to be considered, because they do. Immersion lithography at least has the possibility of being much more economically viable for the industry as a whole, given that it does not require much of the exposure tool or surrounding infrastructure to be reinvented.
After all, it's not just the toolmakers whose resources get stretched thin with all the indecision. The message that we keep hearing about the viability of any potential technology is that it's got to have the infrastructure behind it to support it. But how are the masks, optics, resists and metrology supposed to continue to be ready and available for an industry that can't make up its mind? How far can we continue to stretch the ever-shrinking resources to maintain ready supplies of fused silica, CaF2 and zero-expansion materials when nobody can commit to which optical material they'll actually need in abundance? How are the maskmakers supposed to improve yield and defectivity while stretching the learning curve over 30 different types of phase-shift masks for advanced lithography?
The toolmakers are hoping for a decision from the chipmakers this year on whether the 45 nm node will be tackled by 193 nm immersion or 157 nm dry lithography. This may be asking a bit much, since the chipmakers won't have had any time working with actual step-and-scan immersion systems. But the toolmakers need some decisions made if they have any hope of getting enough resources behind the true solutions.
The 2003 ITRS Lithography chapter, in talking about the "significant resources" needed to commercialize solutions for the 32 nm node and beyond, notes that development costs must be recovered. But how will they be recovered if the products of all this development get thrown out the window?
As you may have noticed, the exclamation points have turned into question marks. The lithography sector has a history of question marks. The industry needs to stop putting so many resources behind that which will never come to fruition. Of course the exploration has to be done initially, but perhaps we need to be a little quicker about pulling the plug when it's time.
