Time for a Systems Roadmap?

No one can deny the tremendous service that the International Technology Roadmap for Semiconductors (ITRS) has provided — and continues to provide — to the semiconductor industry. By showing what the process challenges are going to be, such as when lithography will have to function at a certain level or where we will need a specific dielectric constant, the roadmap has been an invaluable planning resource for equipment companies.

However, now that shrinks are reaching critical levels and process windows are becoming smaller, it is becoming apparent that, were the roadmap to go down one additional level, it could become an even more valuable planning tool.

A problem with today's roadmap is that, in many cases, it does not provide information that addresses particular problems pertaining to sensors and subsystems. As an example, it would be useful for many companies to know that, at a certain point, it is going to become necessary to have the means to enable the detection of contamination in process gases at the ppt level — or that to achieve fab repeatability the capability of obtaining stability of rf power measurements of a certain minimum percentage will become desirable.

Currently, the ITRS is focused on the processes that move our industry, essentially leaving it up to the system OEMs to figure out how a process will be delivered. For example, if the roadmap included systems as a result of a consensus between the large fabs, it might, in the metrology area, help sensor developers determine which problems are best solved at a subsystem level. Providing such information to these other industry sectors would make it much easier for companies to focus on these major issues.

Such a reference source would be crucial in helping companies plan by providing the guidance arising from an industrywide consensus, instead of having them go through a process of educated guessing about which problems the industry will require them to address and solve.

Right now, an entire OEM sector has to deduce from the sectors currently addressed by the roadmap which areas may require their attention and R&D efforts. For instance, the roadmap presently may state that, for a certain year, a dielectric constant of 2.7 will be needed, and then possibly indicate that candidate technologies are CVD and spin-on. What it does not show are the challenges that must be overcome for CVD and spin-on to make either or both succeed. There may be gaps that might be solved by the development of the appropriate instrumentation.

Most system suppliers' planning is driven by the roadmap. They design their platforms to address those challenges. The fact that the roadmap lacks a section addressing the instrumentation and subsystem sectors makes the R&D situation for these companies a free-for-all. OEMs design what they think they need, more than often successfully, but there is no one guiding document to help keep them on track. The presence of such a resource would doubtless lead to far more efficient allocation of R&D resources.

The ITRS and the industry as a whole should consider arriving at a consensus that will produce this resource. A determination should be made about which manufacturing challenges will require integrating more metrology into the process. From a chip company perspective and from an International SEMATECH member perspective, we should find out which hot processes need better instrumentation to breach those yellow and red walls.

The information exists. In some cases, the conclusion may be that further instrumentation of a process is not needed — if so, so much the better. Right now, there is probably some low-hanging fruit in the way of easily identified big challenges that might be integrated into the roadmap.

The consensus to carry this out will have to be forged by chip companies, because equipment companies do not carry out much pre-competitive research. The chip companies, however, can sit together and agree on what their needs and most probable benefits might be.

What do you think?