Seven habits of effective collaborators
08/01/2006
Success will be determined by those companies that can collaborate most effectively.
There has been no shortage of warnings about the impending R&D funding gap in the semiconductor industry, or of calls for greater collaboration, even among competitors, to share costs. But now industry leaders are taking the discussion a step further and offering insights, not just on the value of doing collaboration, but on the importance of doing it well.
That the industry faces an R&D crisis is no secret. The situation has arisen because classical device scaling-the key to the ongoing system price-performance improvements and, therefore, the industry’s success over the past four decades-has essentially ended. Additional scaling results in unacceptable gate leakage, higher voltages, and a rise in power density. The problem, simply, is that we are approaching hard physical limits. A modern gate-oxide layer that is already only a few atoms thick, for example, can no longer scale at historic rates.
Thus, further price-performance progress requires innovation. But innovation is expensive. Consider, for instance, the huge R&D costs associated with moving from dry lithography to immersion to EUV, or from traditional interconnect dielectrics to ultra low-k porous and air-gap materials, or from conventional transistors to strained silicon, high-k gate dielectrics, double gates, and fin-FETs. As IBM CTO Bernie Meyerson remarked recently, “A funny thing happened on the way to the future: We went broke.”
To offset these costs, the trend among independent device manufacturers, foundries, and equipment suppliers alike is toward a globalization of semiconductor R&D, which has been a proven strategy for many companies. IBM, for example, has been involved in partnerships with Sony, Samsung, Toshiba, AMD, Chartered, and Infineon. These and a number of consortia, including Selete, Sematech, and IMEC, are the beginnings of collaborative innovation on a broad scale.
But what’s the best way to design these partnerships? What are the secrets to making them work? In a panel discussion titled “Solutions to the R&D Challenge,” held during The ConFab event (see “The ConFab Special Online Staff Report” on p. 16), representatives from some of IBM’s collaborations shared their views on the following best practices:
1. Choose one location. Conduct R&D work at a single plant. Don’t have engineers from company A in one place and engineers from company B in a different place and have them trade emails once in a while or meet once a quarter. That doesn’t work. If you want to get the work done, send your engineers to one location, give them a good place to work, and make their task a priority.
2. Create an integrated team.Make sure all alliance participants are represented. The lithography R&D department at IBM’s East Fishkill, NY, facility, for example, comprises employees from each of the collaborating companies. Moreover, the R&D managers can be from any of the partners, not just IBM.
3. Align R&D with manufacturing.Because a key strategy for building the next generation of products is to reuse as many existing tools as possible, it’s essential to set up the R&D engineers in the manufacturing facility. Giving them full access to the process recipes, tool parameters, and the workers on the line will facilitate technology transfer and integration.
4. Install processes as developed. Avoid having the manufacturing teams back at your company “reinvent” processes developed by the collaboration partners. Your customers need to seamlessly move products from one partner’s fab to another as demand grows. If they can’t, they will place their business elsewhere.
5. Establish an escalation path. Working out as many intellectual property issues as possible ahead of time is tough, especially during the early attempts at collaboration. So it’s important to agree on a means of passing unforeseen issues up the ranks to the senior executives of the participating companies.
6. Focus on differentiating technologies. To achieve the highest return on collaborative R&D investments, it’s important to concentrate on major cross-industry challenges such as metal gates, high-k materials, low power, and so on that are key to building next-generation transistors and memory devices. Developing better and faster responses to such key issues makes more sense than focusing on less differentiating components such as SRAM cells, memory libraries, and interconnect technologies.
7. Build a business ecosystem. To realize the full benefits of joint investment, it’s necessary to go beyond pure technology development and enhance your business capabilities. One way to achieve this is by building a business ecosystem with your R&D partners to offer joint manufacturing and support to better serve customers.
The industry has passed the point where innovating through collaboration is an option. Now success will be determined by those who can do it most effectively.
 |
Phil LoPiccolo
Editor-in-Chief