Method for rejecting tuning disturbances to improve lamp failure prediction quality in thermal processes
AssigneeApplied Materials, Inc.
InventorsSubrahmanyam Venkata Rama KOMMISETTI, Haw Jyue LUO, Jimmy ISKANDAR, Hsincheng LAI, Parris HAWKINS
Life of this patent
Estimating the term…
Abstract
Embodiments disclosed herein include methods for reducing or eliminating the impact of tuning disturbances during prediction of lamp failure. In one embodiment, the method comprises monitoring data of a lamp module for a process chamber using one or more physical sensors disposed at different locations within the lamp module, creating virtual sensors based on monitoring data of the lamp module, and providing a prediction model for the lamp module using the virtual sensors as inputs.
Claims18
A method for lamp failure prediction, comprising:
- creating virtual sensors by monitoring and collecting data comprising current and resistance related to an upper lamp and a lower lamp of a lamp module for a process chamber, wherein the virtual sensors are represented as equations selected from one or more of the following:
- TIIR
- =
- log
-
-
- (
- abs
-
- (
- TII
- k
-
-
- 1
- *
- TIR
- )
- +
- 1
- )
- ,
-
- TOIR
- =
- log
-
- (
- abs
-
- (
- TOI
- k
-
-
- 2
- *
- TOR
- )
- +
- 1
- )
- ,
-
- BIIR
- =
- log
-
-
- (
- abs
-
-
- (
- BII
- k
-
-
- 3
- *
- BIR
- )
- +
- 1
- )
-
-
- BOIR
- =
- log
-
-
- (
- abs
-
- (
- BOI
- k
-
-
- 4
- *
- BOR
- )
- +
- 1
- )
- ,
-
- and
- TIR
- =
- log
-
- (
- abs
-
- (
- TII
- k
-
-
- 1
- *
- TIR
- +
- TOI
- k
-
-
- 2
- *
- TOR
- +
- BII
- k
-
-
- 3
- *
- BIR
- +
- BOI
- k
-
-
- 4
- *
- BOR
- )
- +
- 1
- )
- ,
- wherein TII is “Top Inner Current,” BII is “Bottom Inner Current,” TOI is “Top Outer Current,” BOI is “Bottom Outer Current,” TIR is “Top Inner Resistance,” BIR is “Bottom Inner Resistance,” TOR is “Top Outer Resistance,” BOR is “Bottom Outer Resistance,” and the expression “log” in equations is natural logarithm, and where factors k1, k2, k3, k4 are constants extracted from the data;
- and
- providing a lamp failure prediction for the lamp module using the virtual sensors as inputs.
A method, comprising:
- monitoring data of a lamp module for a process chamber using one or more physical sensors disposed at different locations within the lamp module;
- creating virtual sensors based on monitoring data of the lamp module;
- and
- providing a prediction model for the lamp module using the virtual sensors as inputs.
A method, comprising:
- identifying components of a principal factor within a first scatter plot presented in a two-axis coordinate system showing a trend line of a first variable and a second variable plotted against each other, wherein the trend line has one or more discrete value changes, and the first and second variables are associated with one or more physical sensors disposed at different locations within a lamp module for a process chamber;
- identifying a main direction of a cloud of plotted points within the first scatter plot having the largest variance;
- rotating the scatter plot until no discrete value changes in the trend line can be seen when viewing from one axis of the coordinate system to obtain at least one or more rotated vector equations, wherein the one or more rotated vector equations are a function of the first and second variables;
- and
- establishing a predictive maintenance (PdM) model for the lamp module using a specific rotated vector equation containing information of pure nature lamp decay.
Dependencies are inferred from the claim text.