Detecting groups of defects in semiconductor feature space

• US 5,991,699 A
• Filed: 04/02/1997
• Issued: 11/23/1999
• Est. Priority Date: 05/04/1995
• Status: Expired due to Term

• Current Assignee: KLA Instruments, Corporation (KLA Corporation)
• Original Assignee: KLA Instruments, Corporation (KLA Corporation)
• Inventors: Kulkarni, Ashok V., Rockwell, Paul
• Primary Examiner(s): WACHSMAN, HAL D
• Application Number:
  US08/886,805
• Time in Patent Office: 965 Days
• Field of Search: 364/468.01-468.03, 364/468.16, 364/468.17, 364/468.28, 364/488-490, 702/34, 702/35, 702/81-84, 702/179, 702/182, 382/147, 382/149, 382/151, 250/559.45, 324/158.1, 324/765, 324/764, 438/16, 438/17, 438/10, 438/12, 438/18, 706/904, 706/912, 706/906, 706/914, 706/921
• US Class Current: 702/83
• CPC Class Codes: H01L 22/20 Sequence of activities consisting of a plurality of measurements, corrections, marking or sorting steps

Abstract

Techniques for improving manufacturing process control based on inspection of manufactured items at intermediate process steps, based on clustering and binning of defect data. Additionally, the using the defect data produced by inspection machines to improve manufacturing process control specifically relating to semiconductor manufacturing process control. Examples described here relate specifically to semiconductor wafers, but may be generalized to any manufacturing process.

Claims

1. A computer program stored on a computer-readable medium for analyzing data associated with defects on a substrate detected during production of semiconductor wafers, wherein for a region of interest on said substrate said data includes a quantitative number and density of defects and other characteristics of each of said defects, said computer program comprising:

instructions for calculating a minimum number of defects in said region to be included in a cluster of defects using said quantitative number of defects in said region of interest;

instructions for calculating a first maximum threshold of a first of said characteristics of said defects included in said defect data in said region to be included in a cluster of defects using said defect density in said region of interest; and

instructions for generating a cluster of defects having at least said calculated minimum number of defects and each included defect having said first of said characteristics that does not exceed said first maximum threshold.

2. The computer program of claim 1 wherein a maximum number of defects in said cluster of defects is calculated as being log(number of defects in region of interest).

3. The computer program of claim 2 wherein said first maximum threshold of said first of said characteristics of said defects included in said defect data of defects in said region potentially includable in said cluster of defects is calculated as being the average of a Poisson distribution of said first characteristic for all of said defects in said region of interest.

4. The computer program of claim 2 wherein said first maximum threshold of said first of said characteristics of said defects included in said defect data of defects in said region potentially includable in said cluster of defects is calculated to have a highest value that maximizes the exclusion of defects in said cluster of defects wherein excludable defects taken together have a Poisson distribution.

5. The computer program of claim 4 wherein said first maximum threshold of said first characteristic of said defects in said region is calculated using the following equation:

##EQU7## wherein r is said first maximum threshold, p is a user selected probability that all defects to be included in said cluster have the same causation, n is said defect density from said defect data for the defects in said region of interest, and N is said minimum number of defects to be included in said cluster.

6. The computer program of claim 1 wherein:

said characteristics of each of said defects includes location coordinates on said substrate and said cluster is to be determined spatially using said location coordinates; and

said first maximum threshold is the maximum distance between any defect in said cluster and the next closest defect in said cluster.

7. The computer program of claim 6 for performing spatial clustering of selected ones of said defects:

wherein said defects and data associated with said defects in said region of interest earlier identified as being in a cluster have been removed from the data to be processed for spatial clustering with the remaining data organized in a forest of trees formatted defect map;

said instructions for generating a cluster of defects includes;

a. instructions for subdividing said region of interest into a grid of rows and columns of square bins, with each bin having a diagonal dimension equal to one half said maximum distance;

b. instructions for searching each bin in each row in said defect map, row by row, for bins that contain at least one defect;

c. instructions for row by row grouping together adjacent defect containing bins as individual groupings of defects with each such grouping of defects being a separate potential spatial cluster of defects;

d. instructions for searching for a first row having at least one defect containing bin and noting a column of said grid in which such bins are located;

e. instructions for searching in a second row that is a next adjacent row to said first row for defect containing bins that are in the same column, or a first, second or third column in either direction from said column in said first row in which a defect containing bin is located;

f. instructions for searching defect containing bins identified in e. for defect containing bins in the same column or next adjacent column to a corresponding defect containing bin in said first row;

g. instructions for combining defect containing bins identified in instructions f. with a grouping of defects associated with a corresponding defect containing bin in said first row;

h. instructions for searching in a third row that is a next adjacent row to said second row for defect containing bins that are in the same column, or a first, second or third column in either direction from said column in said first row in which a defect containing bin is located;

i. instructions for searching defect containing bins identified in instructions h. for defect containing bins in the same column or next adjacent column to a corresponding defect containing bin in said second row;

j. instructions for combining defect containing bins identified in instructions i. with a grouping of defects associated with a corresponding defect containing bin in said second row;

k. instructions for searching in a fourth row that is a next adjacent row to said third row for defect containing bins that are in a the same column, or a first, second or third column in either direction from said column in said first row in which a defect containing bin is located;

l. instructions for searching defect containing bins identified in instructions k. for defect containing bins in the same column or next adjacent column to a corresponding defect containing bin in said third row;

m. instructions for combining defect containing bins identified in instructions l. with a grouping of defects associated with a corresponding defect containing bin in said third row;

n. instructions for returning to instructions d. to continue to search additional subsequent rows and perform instructions e. through n. on each such row until all rows have been processed;

o. instructions for identifying each grouping of defects generated by instructions a. through n. that have said minimum number of defects to identify said groupings of defects as a spatial cluster; and

p. instructions for preparing a list of said spatial clusters identified by instructions o.

8. The computer program of claim 7 further including the following instructions between instructions c. and d. as follows:

q. instructions for identifying in a first row each of said bins of said groupings of defects that are immediately adjacent an empty bin with said empty bin being immediately adjacent another defect containing bin having less than said minimum number of defects;

r. instructions for measuring in said first row a distance between each of said defects closest to the empty bin of instructions q. in said another defect containing bin and defects closest to the other side of the empty bin of instructions q. in said bin of said grouping of defects adjacent said empty bin of instructions q.;

s. instructions for including in said first row each of said defects in said another bin that are within said maximum distance of a defect in said grouping of defects into said grouping of defects as being a part of a same potential duster of defects;

t. instructions for identifying in said first row each of said bins of said groupings of defects that are immediately adjacent an empty bin with said empty bin being immediately adjacent another bin included within another grouping of defects;

u. instructions for measuring in said first row a distance between defects in each bin closest to and on opposite sides of the empty bin identified by instructions t.; and

v. instructions for creating in said first row an enlarged grouping of defects by merging each of said groupings of defects containing said bins immediately adjacent said empty bin if the measured distance between any two defects on opposite sides of said empty bin are within said maximum distance with said enlarged grouping of defects being a part of the same potential cluster of defects.

9. The computer program of claim 1 wherein:

said characteristics of each of said defects includes defect features and said cluster is to be determined in feature space using at least one of said defect features; and

said first maximum threshold is the maximum feature characteristic of said at least one of said defect features of any defect in said cluster.

10. The computer program of claim 1:

further includes;

instructions for calculating a second maximum threshold of a second of said characteristics of said defects included in said defect data of defects in said region to be included in a cluster of defects using said defect density in said region of interest; and

said instructions for generating a cluster of defects having each included defect with said first and second of said characteristics with neither of said first and second of said characteristics exceeding said first and second maximum thresholds.

11. A computer program stored on a computer-readable medium for analyzing data associated with defects on a plurality of substrates simultaneously processed in a stack with each substrate spaced-apart from the next adjacent substrates during production of semiconductor wafers, wherein for a three dimensional region of interest defined relative to said stack that includes at least a portion of at least two substrates of said plurality of substrates, said data includes a quantitative number and density of defects, an identifier for each of said plurality of substrates relative to the stack position of the corresponding substrate, and other characteristics of each of said defects in said region, said computer program comprising:

instructions for calculating a minimum number of defects in said region to be included in a cluster of defects using said quantitative number of defects in said region of interest with said region including at least a portion of at least two substrates of said plurality of substrates;

instructions for calculating a first maximum threshold of a first of said characteristics of said defects included in said defect data of defects in said region to be included in a cluster of defects using said defect density in said region of interest with said region including at least a portion of at least two substrates of said plurality of substrates; and

instructions for generating a cluster of defects having at least said calculated minimum number of defects and each included defect having said first of said characteristics that does not exceed said first maximum threshold with said region including at least a portion of at least two substrates of said plurality of substrates.

12. The computer program of claim 11 wherein:

said characteristics of each of said defects includes three dimensional location coordinates to determine position on a substrate and as to which of said plurality of substrates in said stack of substrates, and said cluster is to be determined spatially using said three dimensional location coordinates; and

said first maximum threshold is the maximum distance between any defect in said cluster and the next closest defect in said cluster on the same substrate or on another of said plurality of substrates.

13. The computer program of claim 11 wherein:

said characteristics of each of said defects includes defect features and said cluster is to be determined in feature space using at least one of said defect features on the same substrate or on another of said plurality of substrates; and

said first maximum threshold is the maximum feature characteristic of said at least one of said defect features of any defect in said cluster on the same substrate or on another of said plurality of substrates.

14. The computer program of claim 11:

further includes;

instructions for calculating a second maximum threshold of a second of said characteristics of said defects included in said defect data in said region to be included in a cluster of defects using said defect density in said region of interest with said region including at least a portion of at least two substrates of said plurality of substrates; and

said instructions for generating a cluster of defects having each included defect with said first and second of said characteristics with neither of said first and second of said characteristics exceeding said first and second maximum thresholds with said region including at least a portion of at least two substrates of said plurality of substrates.

15. A computer program stored on a computer-readable medium for analyzing data associated with defects on a plurality of substrates simultaneously processed as a stack of said plurality of substrates each spaced-apart from the next adjacent ones of said plurality of substrates when in said stack during production of semiconductor wafers, said data includes a quantitative number and density of defects, an identifier for each of said plurality of substrates relative to the stack position of the corresponding substrate, location information of each defect on the corresponding substrate, and other characteristics of each of said defects in a region, said computer program comprising:

instructions for generating a defect map having a selected two dimensional sized and shaped region that corresponds to adjacent regions on each of said plurality of substrates when in said stack, with all of the defects from at least two of said plurality of substrates from the corresponding region on said substrates mapped onto said defect map with the same location in said selected two dimensional region as in said corresponding regions of said at least two substrates;

instructions for calculating a minimum number of defects in said defect map to be included in a cluster of defects using said quantitative number of defects in said defect map;

instructions for calculating a maximum threshold of one of said characteristics of said defects included in said defect data of defects in said defect map to be included in a cluster of defects using said defect density in said defect map; and

instructions for generating a cluster of defects having at least said minimum number of defects and each included defect having said one characteristic that does not exceed said maximum threshold from said defect map.

16. A computer program stored on a computer-readable medium for analyzing data associated with defects in a plurality of reticle fields on a substrate produced using the same reticle for each of said plurality of reticle fields on said substrate during production of semiconductor wafers, said data includes a quantitative number and density of defects in each of said reticle fields on said substrate, location information of each defect within each of said plurality of reticle fields on said substrate, and other characteristics of each of said defects on said substrate, said computer program comprising:

instructions for generating a reticle field defect map having a two dimensional size and shape that corresponds to a size of a reticle of interest that was used a plurality of times in processing of said substrate, with all of the defects from each of selected ones of said plurality of reticle fields on said substrate mapped onto said reticle field defect map with the same location in said reticle field defect map as in said corresponding reticle field of said substrate;

instructions for calculating a minimum number of defects in said reticle field defect map to be included in a cluster of defects using said quantitative number of defects in said reticle field defect map;

instructions for calculating a maximum threshold of one of said characteristics of said defects included in said defect data of defects in said reticle field defect map to be included in a cluster of defects using said defect density in said reticle field defect map; and

instructions for generating a cluster of defects having at least said calculated minimum number of defects and each included defect having said one of said characteristics that does not exceed said maximum threshold from said reticle field defect map.

17. A computer program stored on a computer-readable medium for analyzing data associated with defects in a plurality of reticle fields on each of a multiplicity of substrates produced using the same reticle for each of said plurality of reticle fields on each of said multiplicity of substrates during the production of semiconductor wafers, said data includes a quantitative number and density of defects in each of said reticle fields on each of said substrates, location information of each defect within each of said plurality of reticle fields on each of said substrates, identifying information as to which of said substrates each defect is located, and other characteristics of each of said defects on each of said substrates, said computer program comprising:

instructions for generating a reticle field defect map having a two dimensional size and shape that corresponds to a size of a reticle of interest that was used a plurality of times in the processing of said substrates, with all of the defects from each of selected ones of said plurality of reticle fields on each of said substrates mapped onto said reticle field defect map with the same location in said reticle field defect map as in said corresponding reticle field of said corresponding substrate;

instructions for calculating a minimum number of defects in said reticle field defect map to be included in a cluster of defects using said quantitative number of defects in said reticle field defect map;

instructions for calculating a maximum threshold of one of said characteristics of said defects included in said defect data in said reticle field defect map to be included in a cluster of defects using said defect density in said reticle field defect map; and

instructions for generating a cluster of defects having at least said minimum number of defects and each included defect having said one of said characteristics that does not exceed said maximum threshold from said reticle field defect map.

18. A computer program stored on a computer-readable medium for analyzing data associated with defects in a plurality of die fields on a substrate, said data includes a quantitative number and density of defects in each of said die fields on said substrate during production of semiconductor wafers, location information of each defect within each of said plurality of die fields on said substrate, and other characteristics of each of said defects on said substrate, said computer program comprising:

instructions for generating a die field defect map having a two dimensional size and shape that corresponds to a size of a die of interest to be produced when processing of said substrate is complete, with all of the defects from each of selected ones of said plurality of die fields on said substrate mapped onto said die field defect map with the same location in said die field defect map as in said corresponding die field on said substrate;

instructions for calculating a minimum number of defects in said die field defect map to be included in a cluster of defects using said quantitative number of defects in said die field defect map;

instructions for calculating a maximum threshold of one of said characteristics of said defects included in said defect data in said die field defect map to be included in a cluster of defects using said defect density in said die field defect map; and

instructions for generating a cluster of defects having at least said minimum number of defects and each included defect having said one of said characteristics that does not exceed said maximum threshold from said die field defect map.

19. A computer program stored on a computer-readable medium for analyzing data associated with defects in a plurality of die fields on each of a multiplicity of substrates simultaneously processed as a stack of said multiplicity of substrates each spaced-apart from the next adjacent ones of said multiplicity of substrates when in said stack, said data includes a quantitative number and density of defects in each of said plurality of die fields on each of said multiplicity of substrates during production of semiconductor wafers, location information of each defect within each of said plurality of die fields on each of said multiplicity of substrates, identifying information as to which of said multiplicity of substrates each defect is located, and other characteristics of each of said defects on each of said multiplicity of substrates, said computer program comprising:

instructions for generating a die field defect map having a two dimensional size and shape that corresponds to a size of a die of interest to be produced when processing of said substrates is complete, with all of the defects from each of said plurality of die fields on each of said multiplicity of substrates mapped onto said die field defect map with the same location in said die field defect map as in said corresponding die field of said corresponding substrate;

instructions for calculating a minimum number of defects in said die field defect map to be included in a cluster of defects using said quantitative number of defects in said die field defect map;

instructions for calculating a maximum threshold of one of said characteristics of said defects included in said defect data in said die field defect map to be included in a cluster of defects using said defect density in said die field defect map; and

instructions for generating a cluster of defects having at least said minimum number of defects and each included defect having said one of said characteristics that does not exceed said maximum threshold from said die field defect map.

20. A computer program stored on a computer-readable medium for analyzing data associated with defects located on a region of a substrate during production of semiconductor wafers, said data includes a quantitative number and density of defects in said region of said substrate, location information of each defect in said region, and other characteristics of each of said defects in said region, said computer program comprising:

a. instructions for generating at least one cluster of defects from said data associated with defects located in said region by one of spatial clustering of defects, feature space clustering of defects, reticle field clustering of defects in several reticle fields in said region, and die field clustering of defects in several die fields in said region;

b. instructions for deleting individual defects included within each of said at least one cluster of defects generated in instructions a. from said data associated with defects located in said region to create a reduced data set and list of defects associated therewith; and

c. instructions for repeating instructions a. if there is data remaining in said reduced data set from instructions b. until there is no longer any data remaining in said reduced data set from instructions b. or until no additional clusters of defects are generated in instructions a. by any of the clustering techniques on the last reduced data set from instructions b.;

wherein no defect is a member of more than one cluster, each cluster of defects is likely to have resulted from a single causation, and any defect data remaining in said reduced data set after the completion of instructions c. represents randomly occurring defects.

21. A computer program stored on a computer-readable medium for analyzing data associated with defects located on a plurality of substrates simultaneously processed as a stack of said plurality of substrates each spaced-apart from the next adjacent ones of said plurality of substrates when in said stack during production of semiconductor wafers, said data includes a quantitative number and density of defects on each of said plurality of substrates, location information of each defect on a corresponding one of said plurality of substrates, identifying information as to which of said plurality of substrates each defect is located on, and other characteristics of each of said defects on each of said plurality of substrates, said computer program comprising:

a. instructions for generating at least one cluster of defects from said data associated with defects located on a multiplicity of substrates each spaced-apart from the next adjacent ones of said multiplicity of substrates when in said stack by one of spatial clustering of defects on a single substrate, spatial clustering of defects on multiple substrates of said plurality of substrates, feature space clustering of defects on a single substrate, feature space clustering of defects on multiple substrates of said plurality of substrates, wafer stack clustering of defects on multiple substrates of said plurality of substrates, reticle field clustering of defects in several reticle fields on a single substrate, reticle field clustering of defects in several reticle fields on multiple substrates of said plurality of substrates, die field clustering of defects in several die fields on a single substrate, and die field clustering of defects in several die fields on multiple substrates of said plurality of substrates;

b. instructions for deleting individual defects included within each of said at least one cluster of defects generated in instructions a. from said data associated with defects located on a plurality of substrates to create a reduced data set and list of defects associated therewith;

c. instructions for repeating instructions a. if there is data remaining in said reduced data set from instructions b. until there is no longer any data remaining in said reduced data set from instructions b. or until no additional clusters of defects are generated in instructions a. by any of the clustering techniques on the last reduced data set from instructions b.;

wherein no defect is a member of more than one cluster, each cluster of defects is likely to have resulted from a single causation, and any terminal data remaining after the completion of instructions c. represents randomly occurring defects.

22. A method for analyzing data associated with defects on a substrate during production of semiconductor wafers, wherein for a region of interest on said substrate said data includes a quantitative number and density of defects and other characteristics of each of said defects, said method comprising the steps of:

calculating a minimum number of defects in said region to be included in a cluster of defects using said quantitative number of defects in said region of interest;

calculating a first maximum threshold of a first of said characteristics of said defects included in said defect data in said region to be included in a cluster of defects using said defect density in said region of interest; and

generating a cluster of defects having at least said minimum number of defects and each included defect having said first of said characteristics that does not exceed said first maximum threshold.

23. The method of claim 22 wherein a maximum number of defects in said cluster of defects is calculated as being log(number of defects in region of interest).

24. The method of claim 23 wherein said first maximum threshold of said first of said characteristics of said defects included in said defect data of defects in said region potentially includable in said cluster of defects is calculated as being the average of a Poisson distribution of said first characteristic for all of said defects in said region of interest.

25. The method of claim 23 wherein said first maximum threshold of said first of said characteristics of said defects included in said defect data of defects in said region potentially includable in said cluster of defects is calculated to have a highest value that maximizes exclusion of defects in said cluster of defects wherein excludable defects taken together have a Poisson distribution.

26. The method of claim 25 wherein said first maximum threshold of said first characteristic of said defects in said region is calculated using the following equation:

##EQU8## wherein r is said first maximum threshold, p is a user selected probability that all defects to be included in said cluster have the same causation, n is said defect density from said defect data for the defects in said region of interest, and N is said minimum number of defects to be included in said cluster.

27. The method of claim 22 wherein:

said other characteristics of each of said defects includes location coordinates on said substrate and said cluster is to be determined spatially using said location coordinates; and

said first maximum threshold is the maximum distance between any defect in said cluster and the next closest defect in said cluster.

28. The method of claim 27 for performing spatial clustering of selected ones of said defects:

wherein said defects and data associated with said defects in said region of interest earlier identified as being in a cluster have been removed from the data to be processed for spatial clustering with the remaining data organized in a forest of trees formatted defect map;

said generating a cluster of defects includes the steps of;

a. subdividing said region of interest into a grid of rows and columns of square bins, with each bin having a diagonal dimension equal to one half said maximum distance;

b. searching each bin in each row in said defect map, row by row, for bins that contain at least one defect;

c. row by row grouping together adjacent defect containing bins as individual groupings of defects with the defects in each such grouping of bins being a separate potential spatial cluster of defects;

d. searching for a first row having at least one defect containing bin and noting a column of said grid in which such bins are located;

e. searching in a second row that is a next adjacent row to said first row for defect containing bins that are in the same column, or a first, second or third column in either direction from said column in said first row in which a defect containing bin is located;

f. searching defect containing bins identified in step e. for defect containing bins in the same column or next adjacent column to a corresponding defect containing bin in said first row;

g. combining defect containing bins identified in step f. with a grouping of defects associated with a corresponding defect containing bin in said first row;

h. searching in a third row that is a next adjacent row to said second row for defect containing bins that are in a the same column, or a first, second or third column in either direction from said column in said first row in which a defect containing bin is located;

i. searching defect containing bins identified in step h. for defect containing bins in the same column or next adjacent column to a corresponding defect containing bin in said second row;

j. combining defect containing bins identified in step i. with a grouping of defects associated with a corresponding defect containing bin in said second row;

k. searching in a fourth row that is a next adjacent row to said third row for defect containing bins that are in a the same column, or a first, second or third column in either direction from said column in said first row in which a defect containing bin is located;

l. searching defect containing bins identified in step k. for defect containing bins in the same column or next adjacent column to a corresponding defect containing bin in said third row;

m. combining defect containing bins identified in step l. with a grouping of defects associated with a corresponding defect containing bin in said third row;

n. returning to step d. to continue to search additional subsequent rows and perform steps e. through n. on each such row until all rows have been processed;

o. identifying each grouping of defects generated by step a. through n. that have said minimum number of defects to identify said groupings of defects as a spatial cluster; and

p. preparing a list of said spatial clusters identified in step o.

29. The method of claim 28 further including the following steps between steps c. and d. as follows:

q. identifying in a first row each of said bins of said groupings of defects that are immediately adjacent an empty bin with said empty bin being immediately adjacent another defect containing bin having less than said minimum number of defects;

r. measuring in said first row a distance between each of said defects in said another defect containing bin closest to the empty bin of step q. and defects closest to the other side of empty bin of step q. in said bin of said grouping of defects adjacent said empty bin of step q.;

s. including in said first row each of said defects in said another bin that are within said maximum distance of a defect in said grouping of defects into said grouping of defects as being a part of a same potential cluster of defects;

t. identifying in said first row each of said bins of said groupings of defects that are immediately adjacent an empty bin with said empty bin being immediately adjacent another bin included within another grouping of defects;

u. measuring in said first row a distance between defects in each bin closest to the empty bin and on opposite sides of the empty bin identified by step t.; and

v. creating in said first row an enlarged grouping of defects by merging each of said groupings of defects containing said bins immediately adjacent said empty bin if the measured distance between any two defects on opposite sides of said empty bin are within said maximum distance with said enlarged grouping of defects being a part of the same potential cluster of defects.

30. The method of claim 22 wherein:

said characteristics of each of said defects includes defect features and said cluster is to be determined in feature space using at least one of said defect features; and

said first maximum threshold is the maximum feature characteristic of said at least one of said defect features of any defect in said cluster.

31. The method of claim 22:

further includes the steps of;

calculating a second maximum threshold of a second of said characteristics of said defects included in said defect data of defects in said region to be included in a cluster of defects using said defect density in said region of interest; and

said step of generating a cluster of defects having each included defect with said first and second of said characteristics with neither of said first and second of said characteristics exceeding said first and second maximum thresholds.

32. A method for analyzing data associated with defects on a plurality of substrates simultaneously processed in a stack with each substrate spaced-apart from the next adjacent substrates during processing, wherein for a three dimensional region of interest defined relative to said stack that includes at least a portion of at least two substrates of said plurality of substrates during production of semiconductor wafers, said data includes a quantitative number and density of defects, an identifier for each of said plurality of substrates relative to the stack position of the corresponding substrate, and other characteristics of each of said defects in said region, said method comprising the steps of:

calculating a minimum number of defects in said region to be included in a cluster of defects using said quantitative number of defects in said region of interest with said region including at least a portion of at least two substrates of said plurality of substrates;

calculating a first maximum threshold of a first of said characteristics of said defects included in said defect data of defects in said region to be included in a cluster of defects using said defect density in said region of interest with said region including at least a portion of at least two substrates of said plurality of substrates; and

generating a cluster of defects having at least said minimum number of defects and each included defect having said first of said characteristics that does not exceed said first maximum threshold with said region including at least a portion of at least two substrates of said plurality of substrates.

33. The method of claim 32 wherein:

said other characteristics of each of said defects includes three dimensional location coordinates to determine position on a substrate and as to which of said plurality of substrates in said stack of substrates, and said cluster is to be determined spatially using said three dimensional location coordinates; and

said first maximum threshold is the maximum distance between any defect in said cluster and the next closest defect in said cluster on the same substrate or on another of said plurality of substrates.

34. The method of claim 32 wherein:

said other characteristics of each of said defects includes defect features and said cluster is to be determined in feature space using at least one of said defect features on the same substrate or on another of said plurality of substrates; and

said first maximum threshold is the maximum feature characteristic of said at least one of said defect features of any defect in said cluster on the same substrate or on another of said plurality of substrates.

35. The method of claim 32:

further includes the step of;

calculating a second maximum threshold of a second of said characteristics of said defects included in said defect data in said region to be included in a cluster of defects using said defect density in said region of interest with said region including at least a portion of at least two substrates of said plurality of substrates; and

said step of generating a cluster of defects having each included defect with said first and second of said characteristics with neither of said first and second of said characteristics exceeding said first and second maximum thresholds with said region including at least a portion of at least two substrates of said plurality of substrates.

36. A method for analyzing data associated with defects on a plurality of substrates simultaneously processed as a stack of said plurality of substrates each spaced-apart from the next adjacent ones of said plurality of substrates when in said stack during production of semiconductor wafers, said data includes a quantitative number and density of defects, an identifier for each of said plurality of substrates relative to the stack position of the corresponding substrate, location information of each defect on the corresponding substrate, and other characteristics of each of said defects in a region, said method comprising the steps of:

generating a defect map having a selected two dimensional sized and shaped region that corresponds to adjacent regions on each of said plurality of substrates when in said stack, with all of the defects from at least two of said plurality of substrates from the corresponding region on said substrates mapped onto said defect map with the same location in said selected two dimensional region as in said corresponding regions of said at least two substrates;

calculating a minimum number of defects in said defect map to be included in a cluster of defects using said quantitative number of defects in said defect map;

calculating a maximum threshold of one of said characteristics of said defects included in said defect data of defects in said defect map to be included in a cluster of defects using said defect density in said defect map; and

generating a cluster of defects having at least said minimum number of defects and each included defect having said one characteristic that does not exceed said maximum threshold from said defect map.

37. A method for analyzing data associated with defects in a plurality of reticle fields on a substrate produced using the same reticle for each of said plurality of reticle fields on said substrate during production of semiconductor wafers, said data includes a quantitative number and density of defects in each of said reticle fields on said substrate, location information of each defect within each of said plurality of reticle fields on said substrate, and other characteristics of each of said defects on said substrate, said method comprising the steps of:

generating a reticle field defect map having a two dimensional size and shape that corresponds to a size of a reticle of interest that was used a plurality of times in processing of said substrate, with all of the defects from each of selected ones of said plurality of reticle fields on said substrate mapped onto said reticle field defect map with the same location in said reticle field defect map as in said corresponding reticle field of said substrate;

calculating a minimum number of defects in said reticle field defect map to be included in a cluster of defects using said quantitative number of defects in said reticle field defect map;

calculating a maximum threshold of one of said characteristics of said defects included in said defect data of defects in said reticle field defect map to be included in a cluster of defects using said defect density in said reticle field defect map; and

generating a cluster of defects having at least said minimum number of defects and each included defect having said one of said characteristics that does not exceed said maximum threshold from said reticle field defect map.

38. A method for analyzing data associated with defects in a plurality of reticle fields on each of a multiplicity of substrates produced using the same reticle for each of said plurality of reticle fields on each of said multiplicity of substrates during production of semiconductor wafers, said data includes a quantitative number and density of defects in each of said reticle fields on each of said substrates, location information of each defect within each of said plurality of reticle fields on each of said substrates, identifying information as to which of said substrates each defect is located, and other characteristics of each of said defects on each of said substrates, said method comprising the steps of:

generating a reticle field defect map having a two dimensional size and shape that corresponds to a size of a reticle of interest that was used a plurality of times in the processing of said substrates, with all of the defects from each of selected ones of said plurality of reticle fields on each of said substrates mapped onto said reticle field defect map with the same location in said reticle field defect map as in said corresponding reticle field of said corresponding substrate;

calculating a minimum number of defects in said reticle field defect map to be included in a cluster of defects using said quantitative number of defects in said reticle field defect map;

calculating a maximum threshold of one of said characteristics of said defects included in said defect data in said reticle field defect map to be included in a cluster of defects using said defect density in said reticle field defect map; and

generating a cluster of defects having at least said minimum number of defects and each included defect having said one of said characteristics that does not exceed said maximum threshold from said reticle field defect map.

39. A method for analyzing data associated with defects in a plurality of die fields on a substrate during production of semiconductor wafers, said data includes a quantitative number and density of defects in each of said die fields on said substrate, location information of each defect within each of said plurality of die fields on said substrate, and other characteristics of each of said defects on said substrate, said method comprising the steps of:

generating a die field defect map having a two dimensional size and shape that corresponds to a size of a die of interest to be produced when processing of said substrate is complete, with all of the defects from each of selected ones of said plurality of die fields on said substrate mapped onto said die field defect map with the same location in said die field defect map as in said corresponding die field on said substrate;

calculating a minimum number of defects in said die field defect map to be included in a cluster of defects using said quantitative number of defects in said die field defect map;

calculating a maximum threshold of one of said characteristics of said defects included in said defect data in said die field defect map to be included in a cluster of defects using said defect density in said die field defect map; and

generating a cluster of defects having at least said minimum number of defects and each included defect having said one of said characteristics that does not exceed said maximum threshold from said die field defect map.

40. A method for analyzing data associated with defects in a plurality of die fields on each of a multiplicity of substrates simultaneously processed as a stack of said multiplicity of substrates each spaced-apart from the next adjacent ones of said multiplicity of substrates when in said stack during production of semiconductor wafers, said data includes a quantitative number and density of defects in each of said plurality of die fields on each of said multiplicity of substrates, location information of each defect within each of said plurality of die fields on each of said multiplicity of substrates, identifying information as to which of said multiplicity of substrates each defect is located, and other characteristics of each of said defects on each of said multiplicity of substrates, said method comprising the steps of:

generating a die field defect map having a two dimensional size and shape that corresponds to a size of a die of interest to be produced when processing of said substrates is complete, with all of the defects from each of said plurality of die fields on each of said multiplicity substrates mapped onto said die field defect map with the same location in said die field defect map as in said corresponding die field of said corresponding substrate;

calculating a minimum number of defects in said die field defect map to be included in a cluster of defects using said quantitative number of defects in said die field defect map;

calculating a maximum threshold of one of said characteristics of said defects included in said defect data in said die field defect map to be included in a cluster of defects using said defect density in said die field defect map; and

generating a cluster of defects having at least said minimum number of defects and each included defect having said one of said characteristics that does not exceed said maximum threshold from said die field defect map.

41. A method for analyzing data associated with defects located on a region of a substrate during production of semiconductor wafers, said data includes a quantitative number and density of defects in said region of said substrate, location information of each defect in said region, and other characteristics of each of said defects in said region, said method comprising the steps of:

a. generating at least one cluster of defects from said data associated with defects located in said region by one of spatial clustering of defects, feature space clustering of defects, reticle field clustering of defects in several reticle fields in said region, and die field clustering of defects in several die fields in said region;

b. deleting individual defects included within each of said at least one cluster of defects generated in step a. from said data associated with defects located in said region to create a reduced data set and list of defects associated therewith; and

c. repeating step a. if there is data remaining in said reduced data set from step b. until there is no longer any data remaining in said reduced data set from step b. or until no additional clusters of defects are generated in step a. by any of the clustering techniques on the last reduced data set from step b.;

wherein no defect is a member of more than one cluster, each cluster of defects is likely to have resulted from a single causation, and any defect data remaining in said reduced data set after the completion of step c. represents randomly occurring defects.

42. A method for analyzing data associated with defects located on a plurality of substrates simultaneously processed as a stack of said plurality of substrates each spaced-apart from the next adjacent ones of said plurality of substrates when in said stack during production of semiconductor wafers, said data includes a quantitative number and density of defects on each of said plurality of substrates, location information of each defect on a corresponding one of said plurality of substrates, identifying information as to which of said plurality of substrates each defect is located on, and other characteristics of each of said defects on each of said plurality of substrates, said method comprising the steps of:

a. generating at least one cluster of defects from said data associated with defects located on a multiplicity of substrates each spaced-apart from the next adjacent ones of said multiplicity of substrates when in said stack by one of spatial clustering of defects on a single substrate, spatial clustering of defects on multiple substrates of said plurality of substrates, feature space clustering of defects on a single substrate, feature space clustering of defects on multiple substrates of said plurality of substrates, wafer stack clustering of defects on multiple substrates of said plurality of substrates, reticle field clustering of defects in several reticle fields on a single substrate, reticle field clustering of defects in several reticle fields on multiple substrates of said plurality of substrates, die field clustering of defects in several die fields on a single substrate, and die field clustering of defects in several die fields on multiple substrates of said plurality of substrates;

b. deleting individual defects included within each of said at least one cluster of defects generated in step a. from said data associated with defects located on a plurality of substrates to create a reduced data set and list of defects associated therewith;

c. repeating step a. if there is data remaining in said reduced data set from step b. until there is no longer any data remaining in said reduced data set from step b. or until no additional clusters of defects are generated in step a. by any of the clustering techniques on the last reduced data set from step b.;

wherein no defect is a member of more than one cluster, each cluster of defects is likely to have resulted from a single causation, and any terminal data remaining after the completion of step c. represents randomly occurring defects.