Pump manufacturer Edwards and Fraunhofer implement digital future technologies for smart manufacturing under real-world conditions

Cleanroom - Fraunhofer EMFT

Edwards Pump 1800

The Secure Networked Systems Research Center Munich (SVS, http://www.svs.bayern) has entered into a strategic partnership with the vacuum pump manufacturer Edwards in the field of Smart Manufacturing. The goal of the collaboration is to develop an extended performance characterization for smart/predictive maintenance of subfab devices as a reference for Industry 4.0 applications under real-world conditions. The Fraunhofer Society's research center offers partner companies in the areas of Industry 4.0, connected mobility, and Smart Health/Home an interdisciplinary and application-oriented environment for systematic research.

The research center will provide a cleanroom environment for the agreed cooperation period as part of the project.

Edwards collaborates with the Fraunhofer Institutes AISEC, EMFT, and ESK, which pool their expertise in secure and reliable sensor networks at the Secure Networked Systems Research Center. The collaboration includes research, development, and evaluation of specific sensor and IoT technologies with the aim of generating and securing high-quality data for in-depth analytics. Central to this is the investigation of the correlation between process and pump behavior, which is to be controlled within the semiconductor cleanroom environment of Fraunhofer EMFT. Edwards’ latest EdCentra data platform is expected to play a key role in this.

Research on Industry 4.0 Scenarios

The focus of the collaboration between SVS and Edwards is on implementing predictive or smart maintenance in the specific application of semiconductor manufacturing. To provide efficient and effective service support for high-quality equipment—in this case, vacuum pumps in a semiconductor fabrication environment—predictive maintenance techniques for detecting the condition of devices and their correlation with process conditions are essential.

The 200 mm CMOS (complementary metal-oxide-semiconductor) semiconductor line from Fraunhofer EMFT, combined with the long-standing expertise of the staff in semiconductor manufacturing, offers the optimal environment for monitoring equipment and its performance, as well as for validating analytics and machine learning solutions. The expertise of Fraunhofer ESK and AISEC in secure communication technologies enables the development of secure and flexible solutions for connected factories. Implementation steps within the project include, for example, the development of a sensor setup in the cleanroom environment of Fraunhofer EMFT to collect data from various sources within the infrastructure, as well as preparing a connected sensor node network, including a secure Internet of Things (IoT) infrastructure.

Intelligent and Cognitive Maintenance

Novel machine learning techniques are used to detect anomalies in sensor data. For this purpose, data fusion from multiple sensors for combination and pattern recognition, as well as software algorithms for detecting specific instability states, are required. The predictive maintenance process will be improved by researching new machine learning algorithms based on data collected from sensors above and below the cleanroom floor. This will enable the prediction of future performance and potential anomalies.

An additional important aspect is the use of secure, highly reliable wireless communication between devices, considering factors such as energy consumption, data throughput, data volume, and the uniqueness of the semiconductor manufacturing environment. A secure connection is implemented to exchange data, machine learning models, and device status in real time between secure locations: on-site at the Fraunhofer EMFT CMOS line, remotely in the cloud, or at Edwards' corporate headquarters. This approach not only simplifies response times in service but also facilitates the development and deployment of analytical models. The goal is to provide a new reference architecture that aligns with Industry 4.0 requirements and is immediately tailored to the unique IP concerns of semiconductor manufacturers.

To serve globally operating customers, sensor data from various instances within the factory can be stored in a secure data space that can scale from a single tool upwards. Access to this data and data movement are restricted through the application of uniform access rights and different policies.