Wafer bin map based root cause analysis
A template for assigning the most probable root causes for wafer defects. The bin map data for a subject wafer can be compared with bin map data for prior wafers to find wafers with similar issues. A probability can be determined as to whether the same root cause should be applied to the subject wafer, and if so, the wafer can be labeled with that root cause accordingly.
1. A method, comprising:
receiving at least a first wafer bin map associated with a first wafer from a current production run, the first wafer bin map characterized as having a current defect;
identifying a first product flow for making the first wafer;
retrieving a first set of prior wafer bin maps obtained from at least one prior production run that made wafers using the identified first product flow;
analyzing the current defect on the first wafer bin map on the basis of comparison with the first set of prior wafer bin maps; and
assigning a root cause associated with at least one of the first set of prior wafer bin maps to the defect on the first wafer bin map when the first wafer bin map is determined to be similar to the at least one of the prior wafer bin maps.
2. The method of claim 1 , the analyzing step further comprising:
comparing a first pattern on the first wafer bin map with each of a plurality of prior patterns associated with a respective one of the set of prior wafer bin maps; and
assigning the root cause associated with the at least one of the first set of prior wafer bin maps to the current defect on the first wafer bin map when the first pattern on the first wafer bin map is determined to be similar to the at least one of the prior patterns on the at least one of the set of prior wafer bin maps.
3. The method of claim 2 , the comparing step further comprising:
clustering the prior patterns;
generating a statistical distribution for a plurality of calculated distances between the first wafer and a plurality of neighboring wafers based on the clustering of the prior patterns;
determining a first statistical distance dc between the first wafer and a plurality of neighboring wafers that are considered close enough to associate a root cause of the neighboring wafers with the first wafer; and
determining a second statistical distance df between the first wafer and the plurality of neighboring wafers that is considered too far away to associate the root cause of the neighboring wafers with the first wafer.
4. The method of claim 2 , the comparing step further comprising:
calculating zonal bin deltas for the prior patterns;
generating a statistical distribution for a plurality of calculated distances between the first wafer and a plurality of neighboring wafers based on the zonal bin deltas;
determining a first statistical distance dc between the first wafer and a plurality of neighboring wafers that is considered close enough to associate a root cause of the neighboring wafers with the first wafer; and
determining a second statistical distance df between the first wafer and the plurality of neighboring wafers that is considered too far away to associate the root cause of the neighboring wafers with the first wafer.
5. The method of claim 2 , the comparing step further comprising:
convolving the prior patterns;
generating a statistical distribution for a plurality of calculated distances between the first wafer and a plurality of neighboring wafers based on the convolved patterns;
determining a first statistical distance dc between the first wafer and a plurality of neighboring wafers that is considered close enough to associate a root cause of the neighboring wafers with the first wafer; and
determining a second statistical distance df between the first wafer and the plurality of neighboring wafers that is considered too far away to associate the root cause of the neighboring wafers with the first wafer.
6. The method of claim 2 , the analyzing step further comprising:
identifying a first location of the current defect on the first wafer bin map;
retrieving a subset of the first set of prior wafer bin maps each having a prior defect in a second location that corresponds to the first location of the current defect on the first wafer bin map;
comparing the first pattern on the first wafer bin map with each of a plurality of patterns in the subset of prior wafer bin maps; and
assigning the root cause associated with at least one of the subset of prior wafer bin maps to the first wafer bin map when the first pattern on the first wafer bin map is determined to be similar to at least one of the subset of prior patterns.
7. The method of claim 6 , the comparing step further comprising:
generating a cluster of the prior patterns having a similar defect pattern as the first defect pattern;
generating a statistical distribution for a plurality of calculated distances between the first wafer and a plurality of neighboring wafers based on the clustering of the prior patterns;
determining a first statistical distance dc between the first wafer and a plurality of neighboring wafers that are considered close enough to associate a root cause of the neighboring wafers with the first wafer; and
determining a second statistical distance df between the first wafer and the plurality of neighboring wafers that is considered too far away to associate the root cause of the neighboring wafers with the first wafer.
8. The method of claim 6 , the comparing step further comprising:
calculating zonal bin deltas for the prior patterns;
generating a statistical distribution for a plurality of calculated distances between the first wafer and a plurality of neighboring wafers based on the zonal bin deltas;
determining a first statistical distance dc between the first wafer and a plurality of neighboring wafers that is considered close enough to associate a root cause of the neighboring wafers with the first wafer; and
determining a second statistical distance df between the first wafer and the plurality of neighboring wafers that is considered too far away to associate the root cause of the neighboring wafers with the first wafer.
9. The method of claim 6 , the comparing step further comprising:
convolving the prior patterns;
generating a statistical distribution for a plurality of calculated distances between the first wafer and a plurality of neighboring wafers based on the convolved patterns;
determining a first statistical distance dc between the first wafer and a plurality of neighboring wafers that is considered close enough to associate a root cause of the neighboring wafers with the first wafer; and
determining a second statistical distance df between the first wafer and the plurality of neighboring wafers that is considered too far away to associate the root cause of the neighboring wafers with the first wafer.
10. The method of claim 6 , the comparing step further comprising:
generating a cluster from a plurality of the prior patterns each having a similar defect pattern to the first pattern of the current defect on the first wafer bin map, the similar defect of the cluster is generally located in a second location that corresponds to the first location of the current defect, wherein each of the prior patterns in the cluster has an identified root cause;
wherein the second location in the cluster for the similar defect pattern is defined by a central location and a bounding shape;
comparing the first pattern on the first wafer bin map with the central location and bounding shape for the cluster; and
assigning the identified root cause associated with the cluster to the first wafer bin map when the first pattern on the first wafer bin map is determined to be similar to cluster of prior patterns.
11. The method of claim 1 , wherein the first set of prior wafer bin maps includes wafer bin maps obtained from at least one prior production run that used a different product flow that is similar to the identified first product flow.
12. The method of claim 1 , further comprising:
generating a set of boundary conditions for a plurality of selected wafer parameters; and
tuning the set of boundary conditions in order to identify a first number of neighboring wafers as the subset of prior wafer bin maps.
13. The method of claim 1 , further comprising:
calculating statistics for determining that the current defect is located in one of a plurality of spatial zones on the wafer bin map; and
classifying the current defect on the first wafer by assigning the first wafer to the one of the plurality of spatial zones as determined by the statistical calculations.
14. A non-transitory computer-readable medium having instructions which, when executed by a processor cause the processor to:
receive at least a first wafer bin map associated with a first wafer from a current production run and displaying the first wafer bin map in a first window, the first wafer bin map characterized as having a current defect;
retrieve a first set of prior wafer bin maps obtained from at least one prior production run that used the identified product flow;
analyze the current defect on the first wafer bin map on the basis of comparison with the first set of prior wafer bin maps; and
assign a root cause associated with at least one of the set of prior wafer bin maps to the defect on the first wafer bin map when the first wafer bin map is determined to be similar to the at least one of the prior wafer bin maps.
15. The non-transitory computer-readable medium of claim 14 , further comprising instructions which cause the processor to:
compare a first pattern on the first wafer bin map with each of a plurality of prior patterns associated with a respective one of the set of prior wafer bin maps; and
assign the root cause associated with the at least one of the first set of prior wafer bin maps to the current defect on the first wafer bin map when the first pattern on the first wafer bin map is determined to be similar to the at least one of the prior patterns on the at least one of the set of prior wafer bin maps.
16. The non-transitory computer-readable medium of claim 14 , further comprising instructions which cause the processor to:
cluster the prior patterns;
generate a statistical distribution for a plurality of calculated distances between the first wafer and a plurality of neighboring wafers based on the clustering of the prior patterns;
determine a first statistical distance dc between the first wafer and a plurality of neighboring wafers that are considered close enough to associate a root cause of the neighboring wafers with the first wafer; and
determine a second statistical distance df between the first wafer and the plurality of neighboring wafers that is considered too far away to associate the root cause of the neighboring wafers with the first wafer.
17. The non-transitory computer-readable medium of claim 14 , further comprising instructions which cause the processor to:
calculate zonal bin deltas for the prior patterns;
generate a statistical distribution for a plurality of calculated distances between the first wafer and a plurality of neighboring wafers based on the zonal bin deltas;
determine a first statistical distance dc between the first wafer and a plurality of neighboring wafers that is considered close enough to associate a root cause of the neighboring wafers with the first wafer; and
determine a second statistical distance df between the first wafer and the plurality of neighboring wafers that is considered too far away to associate the root cause of the neighboring wafers with the first wafer.
18. The non-transitory computer-readable medium of claim 14 , further comprising instructions which cause the processor to:
convolve the prior patterns;
generate a statistical distribution for a plurality of calculated distances between the first wafer and a plurality of neighboring wafers based on the convolved patterns;
determine a first statistical distance dc between the first wafer and a plurality of neighboring wafers that is considered close enough to associate a root cause of the neighboring wafers with the first wafer; and
determine a second statistical distance df between the first wafer and the plurality of neighboring wafers that is considered too far away to associate the root cause of the neighboring wafers with the first wafer.
19. A computer-implemented method for a graphical user interface (GUI) for semiconductor processes, the method comprising:
displaying a first window within the GUI containing at least a first wafer bin map associated with a first wafer from a current production run including a display of basic information about the first wafer, the first wafer bin map characterized as having a current defect in a first location;
displaying a second window within the GUI containing a plurality of prior wafer bin maps determined to most closely resemble the first wafer bin map including a display of basic information about the wafers corresponding to the prior wafer bin maps;
assigning a root cause associated with at least one of the set of prior wafer bin maps to the current defect on the first wafer bin map when the first wafer bin map is determined to be similar to the at least one of the prior wafer bin maps; and
providing user controls within the GUI for (i) selecting at least one of the prior wafer bin maps for a visual comparison with the first wafer bin map, (ii) selecting an item of the basic information in order to drill down into the basic information for further analysis, and (iii) overriding and updating the assigned root cause.