Method and system for estimating a state of an uninitialized advanced process controller by using segregated controller data

1. A method, comprising:

implementing a segregated data structure including a plurality of items in an APC controller, each item of said segregated data structure representing a manufacturing context of a manufacturing environment controlled by said APC controller;

determining a threshold criterion for at least one manufacturing context, said threshold criterion indicating, when fulfilled, an insufficiency of data available for said at least one manufacturing context;

comparing, by said APC controller, a current manufacturing context of a substrate to be processed in said manufacturing environment with said threshold criterion; and

when said current manufacturing context corresponds to said at least one manufacturing context and said threshold criterion is fulfilled, estimating a controller state for said current manufacturing context on the basis of one or more of the plurality of manufacturing contexts other than said at least one manufacturing context.

2. The method of claim 1 , wherein estimating said controller state comprises determining at least one difference of said one or more of the manufacturing contexts with respect to said at least one manufacturing context and estimating a contribution of said at least one difference to said at least one manufacturing context.

3. The method of claim 2 , wherein said at least one difference represents at least one of a difference in a process recipe, a different process tool of said manufacturing environment, a different product type, a difference in a minimum critical dimension and a difference in the process history.

4. The method of claim 1 , further comprising determining, by said APC controller, a parameter setting of said manufacturing environment for said substrate on the basis of said current manufacturing context and said estimated controller state and processing said substrate using said determined parameter setting.

5. The method of claim 4 , wherein a plurality of substrates having said current manufacturing context are processed on the basis of said determined parameter setting.

6. The method of claim 4 , wherein said substrate is a product substrate.

7. The method of claim 4 , wherein said substrate represents a pilot substrate and the method further comprises obtaining control data for said current manufacturing context from said processed pilot substrate.

8. The method of claim 1 , further comprising estimating a controller state for said current manufacturing context on the basis of available data for said at least one manufacturing context when said criterion is not fulfilled.

9. The method of claim 2 , wherein estimating a contribution of said at least one difference to said at least one manufacturing context comprises using, in said APC controller, at least one of the following algorithms: a least squares regression; a partial least squares regression; a partial inverse via singular value decomposition; a partial inverse via orthogonal-triangular decomposition; a neuronal network model; a commonality analysis; and any combination thereof.

10. The method of claim 1 , wherein said threshold criterion is determined on the basis of at least one of: a variance error of previously estimated control states of manufacturing contexts that have been initialized on the basis of insufficient control data associated with a corresponding manufacturing context; a number of control state estimates for at least some of the plurality of manufacturing contexts; a frequency of control state updates for at least one of the plurality of manufacturing contexts; and a length of time since the last control state update for said at least one manufacturing context.

11. An APC controller, comprising:

a memory unit configured to receive process-specific data and to store said process-specific data according to a segregated data structure comprising a plurality of items, each of which represents control data of a manufacturing context of a manufacturing environment that is controllable by said APC controller;

a state estimator configured to estimate a control state for a given manufacturing context on the basis of control data of said given manufacturing context;

a comparator configured to determine whether or not a current manufacturing context is a non-initialized manufacturing context; and

a state initializing unit configured to estimate a control state of said current manufacturing context on the basis of control data of one or more manufacturing contexts other than said current manufacturing context when said current manufacturing context is determined to be a non-initialized manufacturing context.

12. The APC controller of claim 11 , wherein said state initializing unit is further configured to determine at least one difference of said one or more manufacturing contexts with respect to said current manufacturing context and to estimate a contribution of said at least one difference to said current manufacturing context.

13. The APC controller of claim 11 , wherein said state initializing unit has implemented at least one of the following algorithms: a least squares regression; a partial least squares regression; a partial inverse via singular value decomposition; a partial inverse via orthogonal-triangular decomposition; a neuronal network model; a commonality analysis; and any combination thereof.

14. The APC controller of claim 11 , wherein said comparator is configured to determine whether or not a current manufacturing context is a non-initialized manufacturing context on the basis of at least one of: a variance error of previously initialized control states of manufacturing contexts; a number of control state estimates for at least some of the plurality of manufacturing contexts; a frequency of control state updates for at least one of the plurality of manufacturing contexts; and a length of time since the last control state update for said at least one manufacturing context.