The New, Intensely Competitive Low-k Market
Laura Peters, Senior Editor -- Semiconductor International, 5/1/2001
 |
For example, IBM successfully integrated SiLK dielectric from Dow Chemical about a year ago and will be using it in its 0.13 µm devices. This move essentially forced more competition in the low-k market: After IBM's announcement, many companies that had largely been considering CVD solutions reconsidered spin-on approaches. Since then, IBM partners Infineon and UMC have licensed the technology. In addition, Dow Chemical believes about two-thirds of its customers will license SiLK dielectric processes, while up to a third are looking to integrate the SiLK resin. Sources report that Intel may be one of those customers, although the company has made no formal announcement yet and, in fact, plans to use copper with FSG interlevel dielectric for its 0.13 µm technology node.
The issues are different in the CVD world, but some licensing may still take place. For instance, both Applied Materials and Dow Corning hold patents on the process used to form OSG (organosilicate glass) films using trimethylsilane precursor in a PECVD reactor. Who owns the intellectual property on the process — the material supplier, the tool supplier, both? Or, because Applied expanded on the process integration work done by Dow Corning, does Applied own the improved process, despite the groundwork laid by Dow Corning?
According to recent announcements, Applied Materials has the most customers to date in the low-k CVD market for its Black Diamond CVD film, including TSMC, AMD and Motorola. However, TSMC has announced it also will be using Novellus' CORAL film for the SiC etch stop layer. Since it would bring much higher cost of ownership to support both systems, and because there are technical advantages to integrating the SiOC and SiC films on one platform (or at least between tools from one vendor), we expect TSMC to choose one or the other for its production lines. This implies indecision despite the announcements.
Regardless of all the current action, one can still argue that the low-k market is quite immature, and that many competing materials and CVD suppliers still have opportunities to penetrate the low-k dielectric market (see "Industry Divides on Low-k Dielectric Choices," and "Designing Porous Low-k Dielectrics"). However, time is of the essence. Like copper and 300 mm wafers, the adoption of low-k dielectrics at some point will increase exponentially. Companies that miss the device manufacturers' deadlines will have a hard time penetrating an already split market.
A final issue to consider is the changing nature of semiconductor technology purchases. A low-k decision is a decision to settle on a particular film and its chemical, mechanical and electrical properties — together with its ability to maintain a given k value through etching, cleaning, annealing, CMP and packaging steps. Perhaps the industry will have to get used to making more materials-based solutions, rather than tool-based solutions, as in the past. It will be the ability to successfully and cost-effectively move new materials into the fab environment that, in the future, will set apart leading companies from their competitors.
