Method and apparatus for automated rule-based sourcing of substrate microfabrication defects

• US 7,987,150 B1
• Filed: 09/04/2007
• Issued: 07/26/2011
• Est. Priority Date: 02/09/2007
• Status: Expired due to Fees

• Current Assignee: SiGlaz
• Original Assignee: SiGlaz
• Inventors: Poreda, John, Dinh, Luan Thien, Truong, Khoa Pham Dang, Luu, Victor V., Trinh, Thieu Nguyen, Nguyen, Phong Van
• Primary Examiner(s): Gaffin; Jeffrey A
• Assistant Examiner(s): CHANG, LI WU
• Application Number:
  US11/849,869
• Time in Patent Office: 1,421 Days
• Field of Search: 706/47
• US Class Current: 706/47
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 5/025 Extracting rules from data
• CPC Class Codes: H01L 22/12 for structural parameters, e.g. thickness, line width, refractive index, temperature, warp, bond strength, defects, optical inspection, electrical measurement of structural dimensions, metallurgic measurement of diffusions electrical measurement of diffusions H01L22/14
• CPC Class Codes: H01L 22/20 Sequence of activities consisting of a plurality of measurements, corrections, marking or sorting steps

Abstract

Automated defect sourcing systems identify root-causes of yield excursions due to contamination, process faults, equipment failure and/or handling. They perform this function in timely manner and provide accurate and timely feedback to address and contain the sources of yield excursion. A signature repository stores known wafer surface manufacturing defect types as set of rules. The signature of a manufacturing defect pattern is associated with the equipment or process that causes the defects, and is used to source the manufacturing defects and to provide process control for changing and/or stopping yield excursion during fabrication. A defect signature rule-based engine matches wafer defects against the signature repository during wafer fabrication. Once the defect signature is detected during fabrication, handling and/or disposing the root-cause of the corresponding defect is facilitated using messages according to an event handling database. Optionally, a real-time process control for wafer fabrication is provided.

Claims

1. A method for generating rules for automatic sourcing of substrate microfabrication defects, comprising:

displaying a wafer map, with a target defect cluster and a grid superimposed on the wafer map;

receiving from a user a set of parameters for defining a set of primitive clusters, using the displayed wafer map, wherein said parameters comprise the maximum allowable distance between grid cells in the same primitive cluster, in which said grid cells contain at least one defect;

selecting a geometric shape to represent said set of primitive clusters;

generating a set of density shape models of said selected geometric shape for the set of primitive clusters defined by the primitive cluster parameters received from the user, the density shape models representing mathematical approximations of the defect distribution and density characteristics of the set of primitive clusters, wherein said density shape models are descriptive of said defect cluster in terms of said geometric shape'"'"'s shape, size, orientation, and location on said wafer map; and

receiving from a user a set of spatial parameters for said density shape model that describe the selected geometric spatial parameters of a defect target cluster associated with said shape'"'"'s shape, size, orientation, and location on said wafer map, wherein said spatial parameters for said geometric shape corresponds to a known physical event that deposits or causes to form defects on a substrate in the microfabrication process.

2. The method of claim 1, wherein the set of parameters for defining a set of primitive clusters includes the total number of defects in the primitive cluster.

3. The method of claim 2, wherein the set of parameters for defining a set of primitive clusters further includes a minimum number of defects in a grid cell of the primitive cluster.

4. The method of claim 2, further comprising:

specifying a set of rules, using spatial attributes of the density shape models, to capture the primitive clusters which the user considers part of the target defect cluster.

5. The method of claim 4, wherein the set of rules express a grouping of a plurality of primitive clusters into a set of one or more combined clusters that captures the target defect cluster that corresponds to a known physical event that deposits or causes to form defects on a substrate in the microfabrication process.

6. The method of claim 5, further comprising:

generating a higher level set of density shape models for the combined clusters; and

allowing the user to generate a higher level set of rules, using spatial attributes of the higher level density shape models of the combined clusters, thereby allowing the user to further combine combined clusters in order to better capture the target defect cluster that corresponds to a known physical event that deposits or causes to form defects on a substrate in the microfabrication process.

7. The method of claim 1, wherein a die stack or a reticle stack is used in lieu of the wafer map.

8. The method of claim 2, wherein a die stack or a reticle stack is used in lieu of the wafer map.

9. The method of claim 3, wherein a die stack or a reticle stack in used in lieu of the wafer map.

10. A method for generating rules for automatic sourcing of substrate microfabrication defects, comprising:

displaying a wafer map and superimposing a grid thereon;

drawing a boundary around a defect cluster, wherein said drawing is performed by a user;

assigning values to one or more parameters such that the values define a set of grid cells which comprise defects and approximate the drawn boundary, wherein said parameters comprise defect density, distance between adjacent defects, and one or more geometric shape;

fitting a geometric density shape model to the set of grid cells; and

generating an if-then rule for recognizing the defect cluster, the rule comprising the assigned values and one or more attributes of the geometric density shape model, wherein said rule further comprises a shape, size, orientation, and location of said geometric density shape model.

11. The method of claim 10, wherein the parameters comprise (a) minimum defects per cell density and (b) allowable distance between cells.

12. The method of claim 11, wherein the parameters further comprise (c) a minimum or maximum number of grid cells per cluster.

13. The method of claim 10, wherein the assigning further comprises searching for values which approximate the cluster within the drawn boundary.

14. The method of claim 13, wherein the searching comprises iteratively increasing the minimum defects per cell, and iteratively decreasing the allowable distance between cells, until the set of grid cells approximates the drawn boundary.

15. The method of claim 10, wherein the geometric density shape model comprises an ellipse, a curved tube, a rectangle, or a polygon.

16. The method of claim 10, wherein the fitting is based on a center of gravity and density distribution of the defects.

17. The method of claim 10, wherein the one or more attributes of the geometric density shape model comprise a centroid, a spine, a spine curvature, a cross section width, a cross section symmetry, or a change of cross section as a function of sweep.

18. The method of claim 10, wherein a die stack or reticle stack is used in lieu of the wafer map.

19. A method for applying rules for automatic sourcing of substrate microfabrication defects, comprising:

superimposing a grid on a set of wafer defects indicated by a results file;

applying an if-then rule to the results file to determine one or more defect clusters in the set of wafer defects, the rule comprising values assigned to one or more parameters;

fitting a geometric density shape model to the defect clusters determined by the if-then rules, using geometric parameters provided by the user, the fitting resulting in one or more density shape model attributes; and

looking up the density shape model attributes in a library of signatures;

comparing said signatures to said geometric density shape model, thereby enabling determination of substrate microfabrication defect source.

20. The method of claim 19, wherein the parameters comprise (a) minimum defects per cell density and (b) allowable distance between cells.

21. The method of claim 20, wherein the parameters comprise (c) a minimum or maximum number of grid cells per cluster.

22. The method of claim 19, wherein the geometric density shape model comprises an ellipse, a curved tube, a rectangle, or a polygon.

23. The method of claim 19, wherein the one or more attributes of the geometric density shape model comprise a centroid, a spine, a spine curvature, a cross section width, a cross section symmetry, or a change of cross section as a function of sweep.

24. The method of claim 19, further comprising:

arbitrating between signatures when a testing step results in more than one signature match.

26. The method of claim 19, wherein the wafer defects represent defects of a stacked set of dies or a stacked set of reticles.

25. The method of 19, wherein the applying step is restricted to defects within a particular wafer region.