The next two years - Commentary - Column

Electronic News, Feb 25, 2002 by Naoki Ayata

IN LOOKING AT WHERE lithography will go the next two years, each point leads back to a central one - increased collaboration. As the market returns and we all hurry to move forward again, I hope we will take time to examine what the downturn has taught us, especially about what to expect from each other as we move along.

Distinctions will continue to be less clear between where lithography begins and ends. Customers will increasingly want an integrated, automated, optimized solution of exposure platform, retide and RET, track equipment, etching, metrology and automation. This means more intense, expert and integrated work-collaboration-- in R&D, process and product development, delivery, implementation and ramp-up.

So joint development will increase, and its goals will be to focus on a total litho bay solution that will ensure optimal printing and etching results. This will involve both pure play and combined system vendors, materials suppliers, and, of course, device manufacturers. We are also likely to see an increased role from university teams and national or international consortia, and from those who can provide automation and e-diagnostics expertise.

The immediate work of joint efforts will be to continue to push DUV (248nm) ever deeper. Of course the costs are high for extending an existing technology node -- which is another reason why collaboration will be more necessary. But customers, especially DRAM makers, will choose to squeeze out additional volume with current 248-nm scanners rather than making the costly shift to 193-nm and are likely to stick with 248nm, maybe up to 100nm. So 248nm extension will continue along with the introduction of 193nm.

I believe 193nm generation will be ramping up to mainstream in next two to three years provided the economy continues to grow. Technology-wise, the pace of 193nm changeover should achieve customer requirements, as will [CaF.sub.2] material supply and quality.

However, it is difficult to separate the insertion of 193nm from the introduction of 300mm. I expect 300mm investment in mass production will come by 2004, and mostly at the 193nm node. Logic people are most eager for 193nm. But for many it may be that 193nm introduction will follow 300mm introduction, because the customer does not want two new technologies at one time.

It is easy to predict that the roadmap (ITRS) will remain on its current pace in spite of the investment roadmap. The ITRS never takes a downturn because the cycle time for inserting a new technology is much longer than any likely period of market slowdown. So here is another reason why collaboration is important: managing risks and costs. It is also why I would mention the importance of having strong resources for internal collaboration.

But it may be wise to ask, is the roadmap going to show more and more exits, intersections, and major junctions? So far, it has mapped the course for the high-volume production of memory and microprocessor devices. Will these devices continue to be the predominant traffic on the roadmap?

Like others, Canon still maintains the pursuit of" Optics Forever," which, in general, affirms that after 248nm will come 193nm, and then 157nm--then EUV (a natural extension to optical projection lithography). In the next two years we will continue to apply our heavy investment in R&D for the 157nm and EUV generation, even knowing the big roadblocks to be overcome.

But optics forever will be a costly attitude for reasons besides pursuing technology from generation to generation and having to qualify from, say, 248nm to 193nm generation. There is also the reason that optics forever is not just progression of technology steps but also incremental steps within each technology, or, so to say, inching (or nanometering) each node to its own extreme.

We will be facing very difficult mask issues in the next two years in order to make either incremental steps or next-node strides. In both cases, masks need to be optimized with the whole system, which means close collaboration with a mask maker and others with RET/OPC expertise. But this will mean big costs for node extensions, as well as for solving the roadblock for 157nm and EUV masks.

Even more so, we will have the dilemma with " optical extension" that its own high development and infrastructure costs must occur in tandem with the costs of developing those next generation technologies that will take the place of optic systems. So, as in every twoyear period since i-line, vendors will continue to invest heavily in NGL technology.

Naoki Ayata is senior VP and general manager, Semiconductor Equipment Division of Canon U.S.A., Inc.