Resource Conservation Via Ultrapure Water Reduction

There is potential for resource consumption (water, chemicals and energy) to increase with the industry moving to a larger wafer size, increased ultrapure water (UPW) requirements and an increase in process steps. In fact, resource conservation is one of the five difficult ESH challenges indicated in the ITRS. Wet cleans, wet etches and their rinses can account for up to 80% of the UPW consumed. Wet cleans also account for a significant portion of the chemicals consumed in wafer fabrication. Researchers at Motorola DigitalDNA Laboratories (Austin, Texas) evaluated several options for UPW reductions and water reclaim, including idle flow reduction and optimized rinses. Their results were presented in July at SEMICON West in San Francisco.

When the wet bench is not processing wafers, UPW is still flowing in the rinse tanks, known as idle flow, which prevents any accumulation of contamination. Since these idle flows can consume a significant amount of UPW, Motorola investigated the feasibility of reducing these idle flows while still providing adequate contamination control. A 30% UPW reduction was achieved in the development laboratory by reducing idle flows. This resulted in saving more than 35M gal/yr of UPW.

To evaluate the feasibility of reducing idle flows further, a bacteria control study was performed in two tanks of a semi-automated wet bench. The results indicate that an idle flow rate of as low as 0.13 gal/min (with no periodic bath dumps) can be used if the point-of-use ozone levels are at least 15 ppb. No bulk liquid phase or biofilm accumulation occurred for up to 80 days. The ozone levels in the bath were shown to recover from a bacterial contamination event as high as 14.4 E6 colony forming units within 2 hr with no bacteria detected within 3 hr of the event.

Another option for water reduction that does not affect the actual cleaning processes is wait time flow reduction (WTFR). WTFR was reduced from high rinse flow (up to 8 gal/min) to a lower flow (1 gal/min). Actual reduction is dependent on the number of wafers processed through the wet benches and not the number of tools. For a fab running 6000 wafers/week, WTFR could result in a reduction of 5.2M gal/yr of water.

The researchers also developed an optimized post-Piranha rinse in a semi-automated test wet bench optimized using wafer surface quality as a performance measure to evaluate resource consumption. A rinse recipe with 60 sec of initial overflow followed by two QDR cycles was shown to yield acceptable wafer surface quality. It resulted in a 76% reduction in rinse cycle time and UPW usage. The overall cycle time for the Piranha-Rinse-Spin Rinse Dryer process was reduced by 26%. On one test wafer bench, overall UPW consumption was reduced by more than 3.2M gal/yr of UPW.

Final particle counts of a contamination comparison between the baseline and reduced RCA cleans. (Source: Motorola)

Other options Motorola considered were a reduction in chemical consumption in wet cleans, water recycling — where clean rinses are collected and reprocessed into UPW — and water reclaim in the UPW process itself using reverse osmosis (RO). Implementation of all these changes could save from 50M to more than 75M gal/yr of UPW. This would reduce energy usage by 2.5-5.5M kWh annually. Further reduction can be achieved with water reclaim and recovery of RO reject depending on factors such as factory sizes, tool types, production rates and product type. Motorola will continue to explore water reduction possibilities in an ongoing effort to increase resource conservation.

For additional information on clean processing, click here.