July 8, 2013, Semicon West, San Francisco—Paul Farrar from the Global 450 Consortium described the efforts towards enabling the semiconductor industry to move to 450 mm wafers. The rationale for the change is to double the number of dice on a wafer without needing a new process node.
The consortium has 5 key members, Intel, Samsung, IBM, Global, and TSMC. All of these big companies are looking to increase productivity and unit shipments through the bigger wafers. The key issues facing the group are defect density, materials, and the development of a new infrastructure and ecosystem. The change needs to demonstrate cost effectiveness for the rest of the industry to follow to the larger wafers.
Among the initiatives, a notch-less wafer is important, since the area loss for a flat is a significant number of potential dice. SEMI and various equipment suppliers, assignees, and SUNY’s College of Nanoscale Science and Engineering (CNSC) are doing the development work, creating standards, and starting to move equipment to CNSC for early trials.
One challenge is to synchronize the process changes for both 300 and 450 mm fabs and processes. The costs for either change, to smaller feature size and to larger wafers, are comparable, so the larger wafer projects may delay a process generation. Some of the early hardware is emerging, but the difficulty is that the two wafer sizes have no commonality. Budgets are limited for all the research to be done.
Due to these cost-dominated developments, the 450 consortium has started working with the US PV consortium with a focus on thin-film photo-voltaic production. The initial work is funded for $300M over 5 years to investigate the requirements for thin-film PV production on 450 wafers. In addition, startups in New York will get tax exemptions for working in some of these areas.
This year, Nikon is planning to finish the first 450mm immersion scanner, with first deliveries slated for next year. Initial results show that the latest scanners have only 80 percent by area of the throughput versus a 300 mm scanner. This needs to be addressed before production begins in '18.
The downturn in '12 dropped sales, leading to a need to reduce capacity, which resulted in a reduction of capital expenditures. This cut some of the available funding for developments in 450. Unit volume is expected to be up this year and next, leading one to consider consortium funding will also increase over the next two years.
