Ghavam Shahidi, an IBM research manager, believes the chip industry "should work as a team" on common challenges and avoid "ridiculous secrecy."
Ghavam Shahidi, an IBM research manager, believes the chip industry "should work as a team" on common challenges and avoid "ridiculous secrecy." He told a panel on transistor scaling challenges organized by Applied Materials that "too many companies withhold information in order to achieve a three- to six-month advantage over their rivals. We need to get over that."
To be sure, in 2006 the chip industry needs to make a lot of progress in the search for a high-k gate dielectric and the metal gate electrodes. While most companies have concentrated on hafnium oxide as the high-k material, the search continues for the metal electrodes. Moreover, some propose two metals, which are deposited for the N- and PMOS transistors, while others see merit in the fully silicided, or FUSI, approach.
"We are losing momentum by hiding information" about the right combination of high-k and metal gate electrodes, Shahidi said.
The industry saves money through cooperative research. But there's profit to be made in the market for the fastest processors. If a company achieves a 10 percent performance edge (or reduces active-mode power consumption) by figuring out a high-k/metal gate solution three to six months ahead of its competition, why share that? What is "precompetitive" anyway?
Probably the best that chip makers can hope to achieve is an agreement not to publish disinformation. If a company believes in a technology's commercial value, and chooses not to disclose it, that's fair.
The reality is that consortia looking for research partners have an incentive to publish. IBM publishes aggressively, honing its reputation as an innovator in emerging fields, then invites companies to join its development effort. Companies benefit from sharing the research burden, from presenting world-class papers at industry conferences, from supporting university research and from funding the major research consortia.
Shahidi is correct: Given the industry challenges in the front-end transistor and at the interconnect stack, sharing information helps to sustain momentum. Still, the financial rewards of being a leader are a powerful incentive to maintain big R&D budgets. Balancing competition and cooperation is more important now, when the industry's common challenges are so great.
-David Lammers (firstname.lastname@example.org), editor at-large for EE Times