Projection electron beam lithography apparatus and method employing an estimator
View Patent ↗A process and method for projection beam lithography which utilizes an estimator, such as a Kalman filter to control electron beam placement. The Kalman filter receives predictive information from a model and measurement information from a projection electron beam lithography tool and compensates for factors which cause beam placement error such as wafer heating and beam drift. The process and method may also utilize an adaptive Kalman filter to control electron beam placement. The adaptive Kalman filter receives predictive information from a number of models and measurement information from a projection electron beam lithography tool and compensates for factors which cause beam placement error such as heating and beam drift. The Kalman filter may be implemented such that real-time process control may be achieved.
1. A projection electron lithography system, comprising:
a lithography tool for emitting a beam of electrons and producing measurement information; and
a processor including,
a plurality of different predictive models for producing predictive information, and
an adaptive estimator that iteratively selects a best predictive model from said plurality of different predictive models and controls placement of said beam of electrons based on said predictive information from said best predictive model and said measurement information from said lithography tool, said adaptive estimator employing a tunable strength parameter to determine an optimal adaptation weighting criterion.
2. The system of claim 1 , wherein said adaptive estimator compensates for beating and beam drift effects.
3. The system of claim 1 , wherein said adaptive estimator employs least-squares based linear matrix algebra.
4. The system of claim 1 , wherein said system is a SCALPEL system.
5. The system of claim 1 , wherein said adaptive estimator is an adaptive Kalman filter.
6. The system of claim 1 , wherein said adaptive estimator is an adaptive Kalman filter and each of said plurality of different predictive models is partitioned into wafer scale components and die scale components, said adaptive Kalman filter only employed for wafer scale components.
7. The system of claim 1 , wherein said plurality of different predictive models differ due to a single parameter that varies in each of said plurality of different predictive models.
8. The system of claim 1 , wherein said plurality of different predictive models includes three or more models.
9. The system of claim 1 wherein said plurality of different predictive models are only directed to producing said predictive information for corrections associated with a die scale.
10. A computer implemented process for controlling projection electron lithography, comprising:
emitting a beam of electrons;
producing measurement information on said emitting step;
producing predictive information related to the projection electron lithography process based on a plurality of different predictive models;
iteratively selecting one of said plurality of different predictive models until a best predictive model from said plurality of different predictive models emerges; and
controlling placement of the beam of electrons based on selected predictive information from said best predictive model and said measurement information, wherein said controlling includes determining an optimal adaptation weighting criterion employing a tunable strength parameter.
11. The process of claim 10 , wherein said controlling step employs an adaptive Kalman filter.
12. The process of claim 10 , wherein said controlling step compensates for heating and beam drift effects.
13. The process of claim 10 , wherein said process is a SCALPEL process.
14. The process of claim 10 , wherein said controlling step is implemented as an adaptive Kalman filter and each of said plurality of different predictive models is partitioned into wafer scale components and die scale components, said adaptive Kalman filter only employed for wafer scale components.
15. The process of claim 10 , wherein said plurality of different predictive models differ due to a single parameter that varies in each of said plurality of different predictive models.