2012

The reason for the Arthur C. Clarke-ish title is that, according to the International Technology Roadmap for Semiconductors (ITRS), in that year, a monolith of quite a different nature awaits us: the 450 mm silicon wafer.

Much has been written about how this is inevitable for the semiconductor industry, and an important means through which to ensure the continuing efficient and economical production of devices; Intel is hoping for the change to take place by 2014 at the latest. Nevertheless, not everyone is thrilled over the prospect. Among these are the equipment suppliers, who remember the cost and hurdles associated with developing 300 mm equipment. Some of them argue that they still have not gotten their ROI for that tremendous effort in the late 1990s.

One popular view is that the transition from 300 to 450 mm will not be as painful as the one from 200 to 300 mm "because of the lessons learned." The reality is that even if the entire industry closely collaborates in an unprecedented fashion, the move will not be a simple one. As Masahuru Watanabe, formerly of Toshiba, and Scott Kramer of Sematech put it, "Each wafer transition becomes more challenging as complexity increases, business models diverge, and technology dynamics come into play.

"For the next wafer transition, it will be particularly important for the factory architecture to be adaptable across all business models in IC manufacturing: DRAM, high-volume microprocessors, foundry and very high mix logic."¹ Both experts expressed optimism that because the 300 mm transition modeling software has greatly improved both in function and speed, and is capable "of modeling very complex interactions in a short period of time," the transition's cost and disruption will be minimized.² Meanwhile, semiconductor manufacturers are trying to arrive at a consensus on the standards and benchmarks needed to reach the 450 mm goal. The International Sematech Manufacturing Initiative (ISMI) is working on 300 mm Prime, which should result in a series of productivity enhancements expected to assist fabs in evolving to the new wafer size.

It's not that simple.

There is also the matter of cost. Because of the wafer's size (its weight, fragility, etc.), unprecedented levels of automation will be required — most expect 450 mm fabs to not go for <$10B. Much will have to be altered or changed in the process flow; the learning curves will not be as straightforward as hoped, complicating yield. Some, like Chris Mack, a leading expert in lithography and lithographic process modeling, have speculated that processing improvements could make 300 mm the ultimate wafer size, which means 450 mm would be unnecessary.

Another aspect to consider is the supply of the wafers themselves. Not only will the Czochralski puller required for the larger size be slower and costlier, but the diameter-to-thickness ratio for 450 mm wafers requires them to be thicker (~825 μm) to reduce frailness. However, this is expected to increase slip and breakage risks during handling and processing. Added to the complication is that it still needs to be determined what problems an silicon on insulator, silicon germanium or epitaxial wafer of that size will present.

Providers of 450 mm wafers face a situation that did not affect previous transitions: where the silicon will come from. Wafer providers will now have to compete for their basic material with a very strong, rapidly expanding global photovoltaics industry. It seems as if the market demand for solar power is finally taking off. This competition will not be pretty, and is certain to have an upward effect on the cost of silicon, affecting everybody.

Last, but certainly not least, there is the economic and competitive impact that the move — and all the complications and requirements associated with it — will bring to the industry. It is not farfetched to say that it could fundamentally alter the industry from what it is today. The manufacture of equipment and construction of fabs is becoming so expensive that we can expect to see many of today's players either consolidate or disappear altogether. This has happened with each transition to a larger size. The industry's necessity to compete could be dampened, and some energetic aspects of its development slowed down.

And, lest we forget, whispers of 675 mm are beginning to be heard....