AVS 45th International Symposium
    Manufacturing Science and Technology Group Thursday Sessions
       Session MS-ThP

Paper MS-ThP3
Design of Dynamic Simulation Experiments for Assessing Manufacturing Metrics

Thursday, November 5, 1998, 5:30 pm, Room Hall A

Session: Manufacturing Science and Technology Group Poster Session
Presenter: R.Z. Shi, University of Maryland
Authors: R.Z. Shi, University of Maryland
Z. Han, University of Maryland
K. Moores, University of Maryland
E. Li, University of Maryland
Z. Chen, University of Maryland
G.W. Rubloff, University of Maryland
Correspondent: Click to Email
Evaluation of equipment and processes commonly focuses on raw process time, with equipment overhead contributions to cycle time playing a secondary and/or separate role in decision-making. This work is directed at equipment and process design while treating on an equal footing the contributions to cycle time which arise from raw process time and the cycle time elements associated with establishing process conditions and recovering from them after process completion. Design of experiments (DOE) methodology (using commercial ECHIP@super TM@ software) and dynamic simulation (using a previously designed simulator based on VisSim@super TM@ software) are employed to optimize a rapid thermal CVD polySi process. The RTCVD simulator captures the essential physics and chemistry of mass transport, heat transfer, and chemical kinetics of the RTCVD process as embodied in a specific equipment design. Various parameters for process recipe as well as for equipment design were first selected as possible factors for the response deposition rate and cycle time. A screening design was first carried out to choose those most significant factors, followed by more extensive experiments leading to the generation of response surface models, i.e., for deposition rate and cycle time. The results reveal different regimes of process and equipment design in which cycle time is primarily determined by raw process time, as well as regimes where equipment design is critical in its influence on overhead contributions to the total cycle time, e.g., for establishing reactor pressure initially, or for wafer cooldown. This work demonstrates that the combination of DOE methodology and dynamic simulation provides a powerful tool for examination and optimization of multiple figures of merit; these include more complex but critical metrics like the full cycle time, as well as more specific process measures such as steady-state deposition rates.