The Internationalization of Cooperation— New Challenges
Semiconductor Industry Suppliers Association (SISA)
Dr. Graham pointed out that all the members of the previous panel were from the Semiconductor Industry Suppliers Association (SISA), which is headquartered in Austin, Texas. He praised the organizers of the workshop for the quality of the presenters and the program. He echoed the comment by Ken Flamm that success has many fathers and applied it to the semiconductor industry, underlining just how successful it has been. He then introduced the first speaker, George Scalise, president of the Semiconductor Industry Association, past chair of the Semiconductor Research Corporation, and a former member of the board of SEMATECH.
A U.S. PERSPECTIVE
Semiconductor Industry Association (SIA)
The Unusual Value of Information Technology
Mr. Scalise said he would try to build on what had been said so far, beginning with a comment by Dr. Knorr that the semiconductor industry is the greatest creator of wealth in the United States. He said that roughly 70 percent of semi
conductor product goes to work in information technology, a segment of unusual value. IT, he said, represents about 8 percent of the U.S. economy but nearly 40 percent of the growth in the economy. Information technology is also a deflator, reducing inflation by 0.5 to 1 percent per year and increasing productivity by a point to a point-and-a-half per year.30 Because these are major contributions to the economy, he said, we have the responsibility to “keep this thing going” in order to maintain the growth of the global economy.
A fact of great importance, he said, is that the number of transistors produced each year increases by about 55 percent. This year  the industry will make roughly 40 million transistors for every man, woman, and child on Earth; by 2008 the per-person figure will rise to about 1 billion transistors. This, he said, implies a “major, major transition” that will require significant preparatory steps. One step is to decide whether current industry structures work well enough.
Accomplishments of the Semiconductor Research Corporation
He offered the example of the SRC, the Semiconductor Research Corporation, as a structure that does work well. It was founded in 1982 for two major reasons. The first reason was concern about the insufficient number of engineers coming out of colleges. The second reason was that not enough engineers were being trained in the new solid-state technology. Since then SRC has helped to increase the number of engineering graduates and to enhance technology training.
It has also created an “integrated, virtual semiconductor research laboratory” that funds projects at about 65 universities across the country. One goal has been to find the best principal investigators from any part of society, not just at the major universities. SRC now supports about 275 faculty members and over 800 graduate students, creating a pattern of broad participation that could be imitated in other countries. He advocated extending the SRC model to other associations around the world as a means of both increasing funding and encouraging even broader participation. The current funding level is about $35 million per year without restrictions on the kind of work or the nationality of the students funded. The outcomes of the research are available to all users without restriction.
Another important accomplishment in which SRC has participated, along with the SIA and SEMATECH, is production of the semiconductor roadmap,
which guides the industry today. The domestic roadmap has now been expanded to an international version involving information technology leaders from around the world.
Addressing the Research Gap
He turned to the research gap that had been discussed earlier and acknowledged that the MARCO program designed to address this gap is limited in resources. Two projects were under way to extend this: one for semiconductor design and testing, another for layout. The next steps were to enhance MARCO and add more programs, two of which were then in the planning stage. He conceded that those plans would probably not be enough, considering the proximity of the “brick wall.” He agreed that the effort to get over or around the wall must be international in scope, and he said that SEMATECH could be a key mechanism to promote international research programs on materials structures and devices, circuits systems, and software that would begin to fill a part of the research gap. He recalled that the decision to end government funding to the consortium has helped to move SEMATECH in an international direction. For the SRC as well, he suggested, internationalization is the next logical step.
The Need for a Super-consortium
To pull together a “consortium of consortia,” said Mr. Scalise, would be a more complex exercise than to ask SEMATECH, Selete, MEDEA, the SRC, and others simply to collaborate. One meeting of these groups had already been held, in April 2000, under a structure started by Japan called the International Forum for Semiconductor Technology. Two cooperative first steps were identified: one in environmental safety and health, the other in lithography. At a further meeting in midsummer, however, the group realized it was not sure how to implement the idea. They did agree that a consortium of consortia was the right approach to pursue, because collaborative action is needed to make more efficient use of the R&D dollar. In the United States, for example, public funding over the last decade has declined by 20 percent each for mathematics and physics, while for engineering public funding has declined 30 to 40 percent. The number of electrical engineering graduates has declined by 40 percent. “We’re going to have to create a sea change,” he said, “that will allow us to be far more efficient, far more effective.”
He said that cooperative research was needed to bring about a linked series of necessary steps—more research can drive market competition; more competition can allow the industry to stay on Moore’s Law; the continued improvements implied by Moore’s Law can bring greater functionality and lower cost. “With the issues that are ahead of us in the next five- to seven-year period,” he said, “we don’t have a choice. This has to happen.”
Enrollment and Funding Levels
Dr. Wessner asked two related questions. First, he asked for Mr. Scalise’s impression of how much funding levels for engineering and related disciplines were dropping and to what extent this was responsible for the fall-off in enrollments. Second, he asked if it was feasible to double the size of the SRC’s budget.
Mr. Scalise answered the second question with “absolutely yes.” For engineering and the hard sciences, he said, the SIA is trying hard through a number of programs to get students more involved, including an effort to develop 50,000 K-12 teachers across the country as mentors for young people in math and science. The goal is to foster and reinforce the interest of talented students in those subjects.
Dr. K. C. Das, director of Office of Science and Technology, Department of Technology, Commonwealth of Virginia, observed that the statistics on scientists and engineers are “very frightening,” with little increase for basic research in the universities. He asked at what level of government this issue might most appropriately be addressed.
Funding Engineering and the Hard Sciences
Mr. Scalise said that basic research in the hard sciences and engineering is primarily the responsibility of the federal government and a responsibility that it has backed away from. Funding for the National Institutes of Health has risen by about 47 percent while the funding level for engineering and the hard sciences has declined by about 17 percent. Overall the United States is spending about 3 percent of GDP on basic R&D. In the short term, industry has filled the gap; but this creates an unstable environment, because when economic conditions decline, industry reduces its spending. There has to be a more stable environment, which only the federal government can provide. Encouraging long-term increases in federal funding for basic research is an ongoing major objective of the SIA.
A JAPANESE PERSPECTIVE
Dr. Masuhara said he would try to summarize the Japanese view and then focus mostly on international collaboration. He began with an illustration of the 1999 International Roadmap for Semiconductors, showing “quite a few so-called brick walls which begin around 2005 and 2008.” (See Figure 13.) In the process area the gate CD (critical dimension) control is the major issue in lithography, as
well as gate insulator and gate leak. In the design area a major challenge is to achieve a low-power operation at a very low supply voltage. He said that the international roadmap committee, in discussing the 2000 and 2001 updates, had concluded that the node 130 nm might be accelerated to 2001. He then showed a chart that summarized the major issues discussed by the roadmap technical working group, beginning with the challenges in lithography—not only the tools but also the types of resist, CD control, and gate stack material, equipment, and database. He also summarized the design challenges for developing systems-on-a-chip, including design reuse, system level design, physical synthesis, and testing. Here designers expect the brick wall in 2008.
A Japanese View of U.S. Collaborations
He then gave the Japanese view of government-university-industry collaboration in the United States. In the Japanese view there has been a good balance of support for research by government and industry through the universities. Industry support goes to SEMATECH and the SRC, and university support comes from industry partly through MARCO and the Focus Center Research Project, and partly through SEMATECH, with “very good balance between design and processing.” He said that the overall success of U.S. industry appears to have come from the contributions of five overlapping efforts.
The use of the SIA roadmap to determine the direction of research;
The planning of resource allocation by SIA and SRC;
The allocation of federal funding through Department of Defense, National Science Foundation, and the Defense Advanced Research Projects Agency;
The success of SEMATECH and International SEMATECH in supporting research on process, technology, design, and testing; and
The Focus Center Research Project.
He said that one indicator of the success of U.S. research is the number of papers delivered at the ISSCC and IEEE Field Award events: Nearly twice as many papers originated in the United States as in Japan. As discussed, he said, Japanese industry generated a relatively small number of papers.
The Research Gap
Dr. Masuhara reviewed the research gap discussed by several speakers and the Japanese view of government projects in Europe, Japan, and the rest of Asia. In regard to Europe he emphasized the success of the applications-related technology research center IMEC, created and supported by both national governments and the European Union. This, he said, had led to success in LSI in application areas such as GSM, cell phones, DSP, A/D mixed signal, and SoC. He also mentioned a new regional model created in Scotland called Alba.
The Picture in Japan
Summarizing the picture in Japan, he said it was significant that the semiconductor industry has been generously supported by both STARC and Selete, as well as VDEC. He said that forming partnerships between industry and academia in both process and design would be more important in the future and would require a great deal of funding for development because of changes in technology. A national plan for a government-academia advanced semiconductor research center to study systems, SoC design, and advanced process is needed. STARC has supported industry-academia collaborations in design and testing and Selete has supported advanced process and device R&D.
For international collaboration the Japanese view began with Japan’s participation in the international roadmap. Inter-consortium collaboration is also needed in Japan and on a global scale. Finally, future technology standardization is an effort to achieve global, joint guidance for 300-mm R&D.
He balanced this list by suggesting that participating in consortia does bring tradeoffs that can have negative effects on industry.
First, R&D by individual companies can be more efficient and faster—as long as it commands the best R&D people, good management, and sufficient funds. Recently, he said, these three conditions have become increasingly difficult to meet. In addition, the openness of R&D can be limited by patent policy.
Second, government-funded consortia can be bureaucratic. Once a target is determined, it is difficult to change direction and achieve progress quickly.
Third, consortia of private companies are suitable in commonly shared, basic technologies. It is difficult for industries in different sectors, such as devices and equipment, to collaborate on more specific applications. Too often industry managers are reluctant to send their best R&D people to work in a non-competitive area.
Academia-industry consortia often seem to be the best solution because of openness, availability of top R&D people, and clear benefit to industry; but the value of academic accomplishments is traditionally measured by numbers of papers, degrees, and awards, whether or not the work is relevant to industry’s needs. Often, he said, the R&D tends to be too academic. He suggested the example of Dr. Kilby’s Nobel Prize in physics as excellent academic work with practical results.31 Dr. Masuhara further recommended a prize be named after Dr. Kilby for work in the technology area in order to inspire academic people to work in technology.
Five Criteria for Organizing a Successful R&D Consortium
He offered five criteria for organizing a successful R&D consortium.
Business merit: Is the technology applicable to industry? Can the market accept the new technology? Can the concept lead to new business?
Technical merit: What is the technical merit of each company? Can new patents or technologies be generated? Are there pitfalls in application, technology matching, suitability, or reliability?
Participant merit: Does the consortium provide good opportunities for participants and a good career path?
Academic merit: Can the consortium lead to research papers, master’s and doctoral degrees, and faculty success?
Industry manager merit: Are managers willing to send the best R&D people from industry? Dr. Masuhara said that this is a “negative value” in many cases.
He suggested the use of “Masuhara’s success equation” to measure whether a consortium would be successful, giving a weighting to each function, with the highest weighting for “A” and the lowest for “E.” (Dr. Masuhara suggested that the value of “E” is “maybe zero” in most cases.)
Dr. Masuhara concluded by suggesting three points about successful government-industry cooperation. First, semiconductors will continue to be the key technology in the twenty-first century for information, communications, and consumer technologies. Second, the industry faces major challenges in the years immediately ahead if it is to avoid hitting the brick wall. Third, reaching solutions will require the industry’s best efforts in both competitive R&D and interconsortium collaboration.
A TAIWANESE PERSPECTIVE
Genda J. Hu
Taiwan Semiconductor Manufacturing Company
Dr. Hu began with the message that Taiwan has always been a player that believes in international cooperation. Even as a late-comer to the industry, in whose R&D activities it has not played a significant role, Taiwan has tried not only to improve its internal R&D capabilities but also to participate in international activities, especially in the International Technology Roadmap for Semiconductors. Taiwan has been an active participant in ITRS since its inception in 1998, and last December Taiwan hosted the annual conference for updating the year 2000 roadmap.
The Industry Consortium Called ASTRO
He also discussed in more detail the planned Taiwanese consortium called ASTRO, which has been placed on hold due to issues beyond the control of the industry. The attempt to form that organization, he said, is a clear demonstration that Taiwan intends to participate in R&D consortia. Part of the objective of ASTRO, he said, is to facilitate participation in international R&D activities when there is an opportunity. To him the important message is that virtually all the companies in Taiwan are willing to join international-level R&D activities.
Without ASTRO, the most feasible strategy for the time being is for individual companies to join consortia activities. He said that trying to participate internationally through the various Japanese research institutes would be difficult because they are funded primarily by the government. He suggested that interna-
tional R&D activity is best handled through either individual companies or consortia of private industries.
Industry Must Play a More Active Role
The plan for ASTRO was to start out with some government funding, but less than 50 percent, so that decision making would be primarily in the hands of the industry. Individual companies would have to play a more active role to make the consortium a success.
In the semiconductor industry in Taiwan almost every company is engaged in some type of international collaboration. TSMC has a major working relationship with Philips and is an active member of International SEMATECH. TSMC participates in ITRS activities and has numerous programs with U.S. as well as Japanese equipment manufacturers. UMC is involved with SRC and IBM-Infineon, and with Hitachi in Japan. Winbond is working with Toshiba.
At the present time, Dr. Hu concluded, even though Taiwan does not yet have a true consortium similar to SEMATECH in the United States or Selete in Japan, there are still “very high levels of intention within our industry” to collaborate internationally. He said that the industry will watch for any opportunity to do so in the near future.
A EUROPEAN PERSPECTIVE
Semiconductor Equipment and Materials European Association (SEMEA)
Dr. Kamerbeek began by explaining that the name of his association had recently changed. The name was once Semiconductor Materials Association, but in 1997 it was decided that a combined activity was needed for both the materials and equipment industries; the two were combined under a single title. It used existing organizations in Germany, France, and the United Kingdom. The Dutch companies joined to create an umbrella European association.
Cooperation as a Way of Life in Europe
In discussing support measures for semiconductor technology, he said that internationalization is a common concept for any European. He said that if he drives for more than one hour from his home, he is in Germany, Belgium, or the sea. In other words, anything a company does is likely to involve a form of international cooperation. European countries are relatively small, so that if a company wants to develop new equipment, new materials, or new processes in this industry, it will probably find its partners in other countries.
He said that European cooperative programs can be divided into two categories: national projects, which are very country-specific and too numerous to address, and international projects.
He mentioned first the IST (Information Society Technologies) Programme of the European Union, which has the longest history.32 The European Union has developed the IST into a single integrated research program that builds on the convergence of information processing, communications, and media technologies. The other major international effort is EUREKA—a decentralized effort that is divided into a variety of sub-categories or subjects.33 The largest EUREKA project is MEDEA (Micro-Electronics Development for European Applications).
In all European collaborative projects the rules call for international cooperation, which is defined as a minimum of two partners coming from at least two countries. Usually the areas where this work will take place are described and the nature of the research must be advanced or pre-competitive R&D. The work is either centralized, as it is for IST and other projects under the European Framework 5 Program, or partly decentralized, which is the case for EUREKA projects.
The IST, a Centralized Effort
The European Union IST program is planned and organized by the commission with the support of industry. The representatives of the 15 countries participating in the European Union approve the programs. The commission issues regular calls for projects which are evaluated by commission experts who are employees of the European Union in Brussels, independent consultants, or industry representatives. The program is centralized, and the commission issues contracts from its own funds. Commission employees or hired independent experts and consultants review R&D progress. The typical funding level for each European Union IST program and project is 50 percent of eligible costs.
The interests of the equipment and materials industry in this program are
spelled out in subchapters of the IST program. These include microelectronics R&D, microelectronics semiconductor equipment assessment, and microelectronics SEA300, which is the 300-mm support program. This subchapters program was created because it was clear at the outset of 200-mm development that the costs related to equipment testing, wafer supplies, and other components would be extremely high for the next technology and very difficult for the smaller companies to support.
EUREKA, a Less Centralized Effort
The less centralized EUREKA supports numerous projects on very different topics, from transportation to agriculture; MEDEA is devoted to information technology. Each EUREKA project must be accepted by the ministers’ conference of the EUREKA countries, and each country has its own EUREKA secretariat. An important point is that all projects are initiated and guided by industry.
EUREKA funding differs from European Union funding in that each country decides to participate in a project. Industrial partners in different countries participating in a particular project will enter into contracts with their own national government. This means that partners from different countries must each have a contract with their national government. National governments have different rules for issuing contracts, so the countries are still learning how to accommodate their rules to the various EUREKA projects. The learning process has helped the authorities learn to cooperate and coordinate contracts in order to start projects smoothly. One difficulty, and a difference from the European Union initiatives, is that funding for EUREKA cannot cross borders. This is because national governments want their funding to be used within the country of origin, whether by institutes or companies.
He reviewed the MEDEA program, which ran from 1996 through 2000 with a budget of nearly 500 million euros. It worked as an umbrella program for projects in semiconductor R&D and was run by a MEDEA board consisting of top industry executives. The overall program was controlled by industry.
Other EUREKA programs in information technology are ITEA and PIDEA. ITEA, or Information Technology for European Advancement, supports the development of software technology for European industry and has a planned budget of some 3.2 billion euros for 1999-2006. PIDEA, or Packaging and Interconnection Development for European Applications, supports R&D on high-density interconnection and packaging technologies. PIDEA is a EUREKA cluster program with a total budget of 400 million euros for the period 1998-2003.
A Need to Enhance Core Competences
Dr. Kamerbeek said that the European equipment and materials industry has been interested in MEDEA because it needs help in seven core competences.
The 300-mm development platform;
Innovative manufacturing processes;
Next-generation gas and chemicals;
Product testing; and
The Motivation to Create Partnerships
Europe has a greater need for this kind of support than does a single large country like the United States. Europe is a patchwork of small- and medium-size economies, and each of them is more or less sub-critical in IC activities. However, Europe is not an easy place to form partnerships because of different languages, currencies, and cultures. Countries therefore need incentives to create partnerships, and MEDEA, serving as an information clearinghouse, provides those incentives by helping to arrange funding and partners. Most of the equipment and materials companies participating in MEDEA are small- and medium-size enterprises that can benefit from this kind of support.
As the follow-up program, MEDEA Plus begins in 2001, and the industry is once again supportive. The important research feature for MEDEA Plus, concluded Dr. Kamerbeek, is the enabling technologies charter. Most work in MEDEA Plus for the industry will focus on sub-100-nm research and development to help its participating companies toward the next generation of information technology.
Dr. Graham concluded the panel session and the symposium by thanking all participants, especially those who had traveled long distances to attend. He thanked the organizers again for a rich and educational program featuring excellent material and the participation of the leaders of the international information technology community.