Sensors Help Bring Smart Manufacturing to Existing Fabs – Takeaways from SEMI GSMC
As semiconductor device manufacturers investigate digitalization of existing and new plants, they are focusing on new sensing technologies that can increase wafer throughput, reduce downtime, boost yields and drive other production efficiencies by collecting data that provides insights not previously identified.
These sensing technologies came into sharp focus in the second session of the virtual SEMI Global Smart Manufacturing Conference (GSMC), October 19-21. Dr. Chan Pin Chong, EVP at Kulicke & Soffa, opened the session by noting many of the trends behind the three pillars of smart manufacturing – sensing, connecting and predicting – including:
- The installation of 5G networks are accelerating the movement of key data
- Digital twins created by AI modeling are providing tighter tolerances and better process control
- Pandemic-inspired changes with less travel and the need for better visibility up and down the supply chains
Dr. Chong closed with the reminder that it is people who are driving the smarter manufacturing movement and that the industry needs to invest in the next generation of data scientists to continue its progress.
Amazon Web Services (AWS) has established partnerships across the semiconductor industry to help optimize production and drive significant cost savings, Bill Vass, VP of Technology for AWS, said in his opening keynote. Vass highlighted one with NXP on connected cars, and another on how several companies are uploading EDA databases to leverage cloud-based server infrastructure.
Vass shared the progress that AWS has made in transforming industrial plants across many different industries to enable digital twin technologies that are leading to huge savings, with one facility shaving $6 million per month in costs. Some companies are combining inexpensive cameras with computer vision AI technology as a cost-effective approach to retrofitting a plant with analog sensors to monitor pressure gauges and other devices, leading to machine monitoring and cost reductions.
Tool monitoring sensors such as residual gas analyzers or RF detectors are commonly used by semiconductor device manufacturers. The problem is that the processes monitored by the tool sensors are not controlled by the data coming off of these traditional sensors, said Lukas Baumgartel, applications development engineer at Inficon. One solution is to replace the traditional tool-based sensors with process-aware sensors to minimize downtime, enable predictive re-targeting, reduce inline metrology, and improve chamber matching. One such device – the Quartz Crystal Microbalance process-aware sensor offered by Inficon – enables manufacturers to measure deposition rates directly on a wafer. Baumgartel followed his presentation with a case study on how the sensor is used in 300mm fabs.
To more accurately measure the photoresist volume dispensed on the wafer over the process time and optimize photoresist distribution, manufacturers can deploy fluid monitoring sensors, said Dr. Martin Wirz, sales director at Sensirion. Measuring and then controlling the exact volume dispensed can enable greater uniformity, reducing the cost of expensive photoresists.
Panel Discussion Provides Practice Approaches
The GSMC panel discussion, composed of those who presented earlier in the session, focused on how to address the increasing digital needs of older manufacturing lines that use less advanced technologies. Vass said one approach is to use on-tool memory storage for data collection and download it later offline to a cloud server which solves the bandwidth and data rate issues. The discussion led to some very practical approaches to using sensing technology and leveraging it over a wide range of manufacturing sites.
The panel agreed that digital acceleration is key for the future of the chip manufacturing industry and that incorporating new sensors can uncover insights not previously understood in order to improve predictive maintenance and make chipmaking operations more predictable. These types of gains are critical drivers for the industry as it works to maximize device output and confronts the greatest chip shortage in the semiconductor industry’s history.
Access the Conference Proceedings
Video recordings of the presentations and panel discussions at the 2021 Global Smart Manufacturing Conference are still available. Registering for the event will provide access to the platform and allow anyone to view the videos.
About the Author
Dr. Laura Matz is the Chief Science and Technology Officer for Merck KGaA, Darmstadt, Germany, a leading science and technology company, that operates across healthcare, life science and electronics. She is responsible for the corporate innovation teams including the digital office and new digital business models. She possesses 20 years of experience in semiconductor manufacturing and semiconductor materials businesses. Dr. Matz is a strong advocate for science and engineering in young talent. She collaborates with ASU, AICHE (American Institute of Chemical Engineers) ILI board (Institute for Learning and Innovation) and SEMI Smart Manufacturing and AZ Committees.