Beyond FDC—Process Aware Sensors with Smart Data Analysis
ABSTRACT
Semiconductor manufacturing process tools are typically equipped with a broad array of sensors. However, these sensors are usually intended for control system monitoring and not for direct observation of the process. Due to this observational gap, even processes with very similar tool data can produce different results. Process-aware sensors bridge that gap by providing a direct, real-time measurement of process-critical quantities such as deposition or etch rate, film thickness and properties, and dopant concentration. Integrating these key measurements with tool data and process information makes more precise and rapid process control possible, producing less variable processes and products.
Real-time process control on this level is facilitated through the integration of sensor data with online analysis and control capabilities. Traditional control with manually set limits and heuristic models can detect tool faults and, with process-aware sensors, process faults that suggest product risk. Machine Learning can extend this control by providing systems the ability to automatically learn and improve from experience without being explicitly controlled.
This talk will highlight implementation details of a process control solution integrating machine learning and process-aware sensors.
BIOGRAPHY
Lukas Baumgartel was born in Santa Fe, NM. He attended the University of New Mexico (UNM) for undergraduate study and then received a PhD in Physics from the University of Southern California (USC). His dissertation research included work in the USC MEMS group, where Lukas worked on piezoelectric transducers such as high-overtone-bulk-acoustic-resonators (HBAR) and audio range microphones. Following graduate school, he worked for 6+ years at Intel’s Technology Development fab in Portland, OR, doing process engineering, improvement tasks such as yield and performance optimization, as well as manufacturing support. In his current role at Inficon, he works with customer fabs around the R&D to develop sensor and software solutions for complex microelectronics manufacturing problems.