AVS 53rd International Symposium
    Vacuum Technology Tuesday Sessions
       Session VT-TuP

Paper VT-TuP7
ROR(t) Method in Vacuum Diagnostic

Tuesday, November 14, 2006, 6:00 pm, Room 3rd Floor Lobby

Session: Vacuum Technology Poster Session
Presenter: J. Zhou, Applied Materials Inc.
Authors: J. Zhou, Applied Materials Inc.
H. Gao, Applied Materials Inc.
D. Paul, Applied Materials Inc.
Correspondent: Click to Email
Rate of rise (ROR) in pressure is a key indicator used to diagnose vacuum issues in semiconductor industry. Normally, the final value of measurement will be compared with a preset pass/fail specification. A more useful approach is to evaluate the dynamic moving-average, and the value of this method is presented in this paper. In comparison to the go/no-go data provided by the final vacuum reading, the graph of vacuum over time can provide information on vacuum chamber performance: for example, if it had a leak or an outgassing problem. Thus, this approach minimizes downtime by avoiding unnecessary retests and parts replacement in troubleshooting. In this work, data was collected from more than 100 high vacuum chambers, and three basic types of ROR(t) curves were identified. Also, a controlled-leak experiment was performed to verify the correlation between the ROR(t) curves and chamber status in leak or outgassing. When a controlled leak was added to a leak-tight baseline chamber, the leak not only increased the chamber pressure, but also altered the shape and smoothness of the ROR(t) curve. When the leak was removed, the curve returned to the original shape and smoothness after a certain delay time. The ROR(t) curve signature reflected an evolution process of all gases loaded within the test chamber, which was sensed by the transducer (ion gauge). In a vacuum leak condition, the diffusion process during ROR will experience interference from the normally point-source leak, and this indicator may be especially useful in high volume manufacturing. The possibility of an automatic vacuum diagnostic employing the ROR(t) method will also be discussed.