Specimen topography reconstruction
AssigneeADE Corporation
InventorsJaydeep Sinha, Alexander Belyaev
Life of this patent
Estimating the term…
Abstract
This method removes high frequency noise from shape data, significantly improves metrology system (10) performance and provides very compact representation of the shape. This model-based method for wafer shape reconstruction from data measured by a dimensional metrology system (10) is best accomplished using the set of Zernike polynomials (matrix L). The method is based on decomposition of the wafer shape over the complete set of the spatial function. A weighted least squares fit is used to provide the best linear estimates of the decomposition coefficients (Bnk). The method is operable with data that is not taken at regular data points and generates a reduced data field of Zernike coefficients compared to the large size of the original data field.
Claims30
A process for noise reduction from noisy data representing an artifact at sample points in two. dimensional space of a specimen comprising the steps of:
- receiving said noisy data as a vector, each element of which corresponds to one sample point;
- and
- calculating coefficients of a polynomial which converts said noisy data vector to a two dimensional function continuously representing the artifact in the two dimensional space.
A process for the generating a noise correcting. matrix for a measurement apparatus comprising:
- receiving data representative of artifacts in two dimensional space of a specimen obtained by said apparatus, each data point associated with a data position;
- and
- calculating a specimen-independent noise compensating matrix as a function said data position in two dimensional space on said specimen.
An apparatus for noise reduction from noisy data representing an artifact at sample points in two dimensional space of a specimen comprising:
- means for receiving said noisy data as a vector, each element of which corresponds to one sample point;
- and
- means for calculating coefficients of a polynomial which converts said noisy data vector to a two dimensional function continuously representing the artifact in the two dimensional space.
Apparatus for generating a noise correcting matrix for a measurement apparatus comprising:
- means for receiving data representative of artifacts in two dimensional space of a specimen obtained by said apparatus, each data point assocated with a data position;
- and
- means for calculating a specimen-independent noise compensating matrix as a function of data position in two dimensional space on said specimen.
A model-based method of wafer shape reconstruction comprising:
- obtaining a set of noisy data points representing the wafer shape;
- using a complete set of Zernike polynomials as a shape functional space;
- applying a weighted least square fit between said noisy data points and a set of data points calculated from said Zernike polynomials;
- and
- finding decomposition coefficients for said wafer shape.
Dependencies are inferred from the claim text.