Roadmap Gives Targets, Not Solutions

The 2001 International Technology Roadmap for Semiconductors (ITRS) is an amazing work. It provides the industry with technology targets through the year 2016, when devices with 9 nm gate length and a billion transistors will be produced. It highlights the targets for which "no known solutions exist" so that the industry can rally resources around those challenges to meet performance goals and ensure the continuation of Moore's Law. In the document, the roadmap committee especially details the limits of technology or physics that will prevent further progress unless work-around solutions can be made. So what does the roadmap not provide?

It does not give solutions, or limits for that matter. In fact, the 2001 ITRS clearly states, "The ITRS strives to avoid prematurely identifying definite solutions to the future technology challenges," and "a few of the potential solutions are listed, where known, only to inform the readers of current thinking and efforts ... and does not constitute an endorsement by the Roadmap process."

The fact that the roadmap of 1999 was essentially two years behind schedule relative to the 2001 edition proves that companies are in no way limited by the roadmap's targets. It simply provides targets to shoot for, while highlighting the feasibility of that target value at a given time. The feasibility is denoted as "known manufacturable solutions exist" (white), "manufacturable solutions have been identified but require further development" (yellow) and "no known solutions exist" (red). As stipulated in the 2001 roadmap, the red areas are known collectively as the "red brick wall" — where there is consensus in the industry that either the value will never be achieved, or industry experts do not have much confidence that the currently proposed solutions can deliver in the time frame required.

The roadmap also does not exclude potential solutions, even if they do not appear likely at a given time. For instance, it does not say that low-k dielectrics with a net dielectric constant (k_eff) will be 2.1 by the year 2010. It just poses one low-k approach as a potential solution for meeting cost and performance goals at that technology node for some device manufacturer. In another example, the roadmap does not eliminate the possibility that X-ray lithography, for instance, could be used at the 50 nm or future generation. If there is any danger of misinterpreting the roadmap, it's by viewing it as exclusionary or limited. As IMEC fellow Karen Maex recently explained, "The ITRS is great because it gives us guidelines in such a complex environment. On the other hand, we should dare to look at it with more freedom to choose and optimize."

Going forward, there are likely to be many more paths to success than in the past. This is due, in part, to the number of new materials that must be integrated for low-k intermetal dielectrics, high-k gate dielectrics and metal gates, among others. Some IC manufacturers will decide to use different materials than others. Beyond these revolutionary changes in materials, companies will find alternative solutions to the same problem. Device types also continue to grow, as do the number of packaging solutions. The uniqueness of device solutions will keep device manufacturers more competitive, giving them the edge they need in the increasingly global, intensely competitive semiconductor industry.

Finally, the roadmap has been used to identify the areas where precompetitive research is needed most. While no one company has the resources to do all the research to determine the best high-k dielectric/metal gate combinations, for instance, focused groups of researchers have a chance of narrowing down the options. Individual companies take it from there, developing the application-specific solutions through the device manufacturing and packaging processes. Competition will continue to be a foundation for the semiconductor industry, but in today's increasingly expensive and complex manufacturing environment, precompetition is gaining in importance.

What do you think?