When a new semiconductor such as a microprocessor or memory IC is first produced, it often carries a high price, which gradually declines over the lifetime of the part. One way chipmakers cut the cost of the chip is by transitioning production to larger size wafers.
The larger the wafer, the more chips can be produced per wafer and the lower the cost for each individual chip.
“In order for the industry to achieve the cost reductions that are forecast by Moore’s Law, the industry needs to transition to the next generation of wafer size,” says Len Jelinek, senior director and chief analyst for researcher IHS. Moore’s Law states that the number of transistors on a chip doubles every two years. As the number of transistors increases, the cost of the chip declines.
Chip production has been migrating to 300mm wafers. In 2013, 300mm (12-inch) wafers represent about 58% of overall semiconductor capacity and that figure will rise to 70% in 2017, according to researcher IC Insights, based in Scottsdale, Ariz. However, not all semiconductors are produced or will ever be produced on 300mm wafers. About 20% are produced on 200mm (8-inch) or smaller wafers.
For the most part, 300mm wafers are used in high-volume, commodity chips including DRAMs, flash memory and microprocessors. Very recently, some image sensors, power management devices, complex logic and microcomponent ICs with large die sizes have been produced on 300 mm wafers, according to IC Insights.
Larger chipmakers, including Samsung, Toshiba and Micron, all use 300mm wafers for most of their chip production. For instance, Toshiba has 100% of its production on 300mm and is researching plans to transition to 450mm, according to Doug Wong, senior member, technical staff for Toshiba America Electronic Components.
Pure-play foundries such as TSMC and GlobalFoundries, which produce chips for other semiconductor companies, also use 300mm wafers. Foundries and chipmakers such as Samsung “benefit the most from using the largest size wafers available to best amortize the manufacturing costs per die,” according to Trevor Yancey, vice president of technology for researcher IC Insights.
Although chip capacity is transitioning to 300mm wafers, 200mm wafers will not disappear anytime soon. Fabs running 200mm will be profitable for years and will continue to be used to fabricate specialty memories, image sensors, display drivers, microcontrollers, analog products, and MEMS-based devices, Yancey says.
Move to 450mm
The next transition will be to 450mm (18-inch) wafers, but it is not known exactly when that will occur. With the transition to 450 mm wafers, chipmakers will get 2.4 times the number of die per wafer, according to Jelinek.
Last year at least one major chipmaker said it was beginning construction for a new development facility in Oregon that will be capable of manufacturing on 450mm wafers.
“The best estimate for a transition to 450mm in high volume is 2018,” explains Jelinek. “This means there would be some pilot or early production in the 2015/2016 time period.”
The cost to make the transition is high, even for large chipmakers, and only about five or six chipmakers will make the transition.
“There is the cost to develop the tools, for the required R&D, and to develop the manufacturing flows," he says.
Because of the cost, companies may question if it is worth the investment. Yancey says that historically when transition to larger size wafer occurs, chipmakers achieve a 30% cost reduction in chip production. But it may be different with 450mm wafers. Some analysts say cost reduction will be about 20%, while others think the historical 30% cost reduction will hold.
Yancey adds that 450mm wafer capacity will account for only one-tenth of a percent of global IC capacity in December 2017, but will increase each year after that.