A wafer fab built with current equipment has to have capacity of 20,000 wafers or so a month to be economical. New minifabs would run only 5,000 wafers per month, with smaller, lower throughput equipment, but the key is reducing tool costs proportionally as well.

Unless a line has capacity to run 15,000 to 20,000 wafers/month, the different throughputs of the different tools are not in balance, and some are not fully utilized, pushing up the cost of ownership. If the line runs at less than this rate, the excess capacity of the highest throughput tools is wasted, and average utilization goes down. Even with a 200mm fab, the cost of underutilizing such expensive assets hurts profits. With 300mm lines it’s a huge problem.

So there’s growing interest in lower volume mini lines to try to reduce the huge capital investment that brings huge losses in the downturns. The industry needs to go beyond just tools with ever-higher throughput, to consider instead tools that bring down the cost of ownership for the entire process, and improve the whole economics of production.

The real question is, can we build a 5,000 wafer/month line with the same cost of ownership as a 20,000 wafer/month line? If the cost of high throughput tools like ion implanters could be reduced to one-quarter their current levels, throughput could be reduced to one-quarter too. Though most of the focus on minifabs so far has been on small footprints and small volumes, the real issue has to be cost of ownership.

Tool makers and users working together on ASET’s planned new project to develop a highly efficient fab aim to have 1/4 price tools on the market by 2005.

— H. Omiya, former director of SELETE, for Nikkei Micro Devices Online