Methods and systems for product failure prediction based on X-ray image re-examination

1. A method comprising, by an X-ray inspection system:

accessing a first set of X-ray images, captured using the X-ray inspection system, of one or more first samples that are labeled as being non-conforming based on a failure of each of the first samples in a field application;

adjusting a classification algorithm based on the first set of X-ray images, wherein the classification algorithm classifies samples into conforming or non-conforming categories based on an analysis of corresponding X-ray images;

analyzing a second set of X-ray images of a plurality of second samples using the adjusted classification algorithm in the X-ray inspection system, wherein the second samples are previously inspected samples that have been classified as conforming by the classification algorithm during a previous analysis using the X-ray inspection system before the classification algorithm is adjusted; and

identifying one or more of the second samples from the second set of X-ray images, wherein each identified second sample is classified as non-conforming by the adjusted classification algorithm.

2. The method of claim 1 , further comprising, prior to accessing the first set of X-ray images:

inspecting the one or more first samples with the X-ray inspection system, wherein the X-ray inspection system indicates the one or more first samples are conforming;

identifying the one or more first samples as being non-conforming during a subsequent failure analysis process after inspecting the one or more first samples; and

labeling the first set of X-ray images of the one or more first samples as being non-conforming responsive to the identifying the one or more first samples as being non-conforming during the failure analysis process.

3. The method of claim 1 , further comprising:

determining, based on the first set of X-ray images, one or more indicative features correlated to one or more failure modes associated with the one or more first samples that are labeled as being non-conforming.

4. The method of claim 3 , wherein the classification algorithm is a machine-learning model, and wherein adjusting the classification algorithm based on the first set of X-ray images comprises:

re-training the machine-learning model based on the first set of X-ray images of the one or more first samples that are labeled as being non-conforming or based on the one or more indicative features correlated to the one or more failure modes of the one or more first samples.

5. The method of claim 3 , wherein the classification algorithm is a rule-based classification algorithm, and wherein adjusting the classification algorithm based on the first set of X-ray images comprises:

adjusting one or more existing rules of the rule-based classification algorithm based on the one or more indicative features; or

adding one or more new rules to the rule-based classification algorithm, wherein the one or more new rules are generated based on the one or more indicative features.

6. The method of claim 3 , wherein the first set of X-ray images of the one or more first samples that are labeled as being non-conforming are captured by the X-ray inspection system during an initial X-ray inspection process during or after a manufacturing process of the one or more first samples and before the one or more first samples are labeled as being non-conforming.

7. The method of claim 6 further comprising:

capturing a third set of X-ray images of the one or more first samples during a failure analysis process after the one or more first samples have been labeled as being non-conforming; and

determining the one or more failure modes based on the third set of X-ray images of the one or more first samples.

8. The method of claim 7 , further comprising:

comparing the third set of X-ray images of the one or more first samples to the first set of X-ray images of the one or more first samples; and

determining the one or more failure modes based on the comparison between the third set of X-ray images and the first set of X-ray images.

9. The method of claim 3 , further comprising:

determining the one or more failure modes based on the first set of X-ray images and information received from another inspection system used in a failure analysis process of the one or more first samples that are labeled as being non-conforming.

10. The method of claim 1 , further comprising:

determining a similarity metric for each of the plurality of second samples with respect to the one or more first samples that are labeled as being non-conforming; and

comparing the similarity metrics of the plurality of second samples to a pre-determined threshold, wherein the one or more second samples are identified based on a determination that the similarity metrics associated with the one or more second samples are less than the pre-determined threshold.

11. The method of claim 10 , wherein the similarity metric associated with each second sample is determined based on corresponding distances from that second sample to the one or more first samples that are labeled as being non-conforming, wherein the distances are within a N-dimensional feature space as defined by N number of features associated with the plurality of second samples or within a M-dimensional process parameter space as defined by M number of process parameters associated with a manufacturing process of the plurality of second samples.

12. The method of claim 10 , further comprising:

predicting a probability of failure for each of the identified second samples or a future failure time based on an associated similarity metric and information associated with the one or more first samples that are labeled as being non-conforming; and

predicting a possible failure mode for each of the identified second samples based on the associated similarity metric associated with that second sample and one or more failure modes associated with the one or more first samples that are labeled as being non-conforming.

13. The method of claim 1 , further comprising:

capturing a third set of X-ray images for one or more third samples;

analyzing the third set of X-ray images using the adjusted classification algorithm; and

classifying the one or more third samples into conforming or non-conforming categories using the adjusted classification algorithm.

14. The method of claim 1 , further comprising:

sending feedback information to a manufacturing tool used in a manufacturing process of new samples, wherein the feedback information is generated based on information associated with the one or more first samples that are labeled as being non-conforming; and

causing the manufacturing tool to adjust one or more process parameters based on the feedback information, wherein the manufacturing tool with the adjusted process parameters has a lower probability of producing non-conforming samples.

15. The method of claim 1 , further comprising:

sending feedforward information to a manufacturing tool used in a downstream process of a manufacturing process of new samples, wherein the feedforward information is generated based on information associated with the one or more first samples that are labeled as being non-conforming; and

causing the manufacturing tool to adjust one or more process parameters based on the feedforward information, wherein the manufacturing tool with the adjusted process parameters has a lower probability of producing non-conforming samples.

16. The method of claim 1 , further comprising:

sending feedforward information to an inspection tool used in a downstream process of a manufacturing process of new samples, wherein the feedforward information is generated based on information associated with the one or more first samples that are labeled as being non-conforming; and

causing the inspection tool to adjust one or more parameters based on the feedforward information, wherein the inspection tool with the adjusted process parameters rejects new samples, that are associated with one or more indicative features correlated to one or more failure modes of the one or more first samples, as non-conforming.

17. The method of claim 16 , further comprising:

causing the inspection tool to perform an additional inspection step, wherein the additional inspection step identifies one or more of the indicative features associated with the new samples.

18. The method of claim 1 , further comprising:

generating a recalling recommendation based on the identified one or more second samples, wherein the identified second samples comprise less samples than a sample population used in field applications.

19. The method of claim 1 , further comprising:

capturing a third set of X-ray images for a plurality of third samples that are selected for a reliability test for a sample population;

determining, based on the third set of X-ray images, a correlation metric between the plurality of third samples and the sample population based on a N-dimension feature space as defined by N number of features associated with the sample population; and

determining whether the plurality of third samples are representative to the sample population.

20. One or more computer-readable non-transitory storage media embodying software that is operable when executed to:

access a first set of X-ray images, captured using the X-ray inspection system, of one or more first samples that are labeled as being non-conforming based on a failure of each of the first samples in a field application;

adjust a classification algorithm based on the first set of X-ray images, wherein the classification algorithm classifies samples into conforming or non-conforming categories based on an analysis of corresponding X-ray images;

analyze a second set of X-ray images of a plurality of second samples using the adjusted classification algorithm, wherein the second samples are previously inspected samples that have been classified as conforming by the classification algorithm during a previous analysis using the X-ray inspection system before the classification algorithm is adjusted; and

identify one or more of the second samples from the second set of X-ray images, wherein each identified second sample is classified as non-conforming by the adjusted classification algorithm.

21. A system comprising: one or more processors; and one or more computer-readable non-transitory storage media coupled to one or more of the processors and comprising instructions operable when executed by one or more of the processors to cause the system to:

access a first set of X-ray images, captured using the X-ray inspection system, of one or more first samples that are labeled as being non-conforming based on a failure of each of the first samples in a field application;

adjust a classification algorithm based on the first set of X-ray images, wherein the classification algorithm classifies samples into conforming or non-conforming categories based on an analysis of corresponding X-ray images;

analyze a second set of X-ray images of a plurality of second samples using the adjusted classification algorithm in the X-ray inspection system, wherein the second samples are previously inspected samples that have been classified as conforming by the classification algorithm during a previous analysis using the X-ray inspection system before the classification algorithm is adjusted; and

identify one or more of the second samples from the second set of X-ray images, wherein each identified second sample is classified as non-conforming by the adjusted classification algorithm.

22. A method comprising, by an X-ray inspection system:

accessing a first set of X-ray images, captured using the X-ray inspection system, of one or more first samples that are labeled as being conforming based on a failure of each of the first samples in a field application;

adjusting a classification algorithm based on the first set of X-ray images, wherein the classification algorithm classifies samples into conforming or non-conforming categories based on an analysis of corresponding X-ray images;

analyzing a second set of X-ray images of a plurality of second samples using the adjusted classification algorithm in the X-ray inspection system, wherein the second samples are previously inspected samples that have been classified as non-conforming by the classification algorithm during a previous analysis using the X-ray inspection system before the classification algorithm is adjusted; and

identifying one or more of the second samples from the second set of X-ray images, wherein each identified second sample is classified as conforming by the adjusted classification algorithm.

23. The method of claim 22 , further comprising, prior to accessing the first set of X-ray images:

inspecting the one or more first samples with the X-ray inspection system, wherein the X-ray inspection system indicates the one or more first samples are non-conforming;

identifying the one or more first samples as being conforming during a subsequent analysis process after inspecting the one or more first samples; and

labeling the first set of X-ray images of the one or more first samples as being conforming responsive to the identifying the one or more first samples as being conforming during the subsequent analysis process.

24. The method of claim 22 , further comprising:

determining, based on the first set of X-ray images, one or more indicative features associated with the one or more first samples that are labeled as being conforming.

25. The method of claim 24 , wherein the classification algorithm is a machine-learning model, and wherein adjusting the classification algorithm based on the first set of X-ray images comprises:

re-training the machine-learning model based on the first set of X-ray images of the one or more first samples that are labeled as being conforming or based on the one or more indicative features associated with the one or more first samples.

26. The method of claim 24 , wherein the classification algorithm is a rule-based classification algorithm, and wherein adjusting the classification algorithm based on the first set of X-ray images comprises:

adjusting one or more existing rules of the rule-based classification algorithm based on the one or more indicative features; or

adding one or more new rules to the rule-based classification algorithm, wherein the one or more new rules are generated based on the one or more indicative features.

27. The method of claim 24 , wherein the first set of X-ray images of the one or more first samples that are labeled as being conforming are captured by the X-ray inspection system during an initial X-ray inspection process during or after a manufacturing process of the one or more first samples and before the one or more first samples are labeled as being conforming.

28. The method of claim 27 further comprising:

capturing a third set of X-ray images of the one or more first samples during a subsequent analysis process after the one or more first samples have been labeled as being conforming; and

comparing the third set of X-ray images of the one or more first samples to the first set of X-ray images of the one or more first samples, wherein the one or more indicative features are determined based on the comparison between the third set of X-ray images and the first set of X-ray images.

29. The method of claim 24 , wherein the one or more indicative features are determined based on the first set of X-ray images and information received from another inspection system used in the subsequent analysis process that labels the one or more first samples as being conforming.

30. The method of claim 22 , further comprising:

determining a similarity metric for each of the plurality of second samples with respect to the one or more first samples that are labeled as being conforming; and

comparing the similarity metrics of the plurality of second samples to a pre-determined threshold, wherein the one or more second samples are identified based on a determination that the similarity metrics associated with the one or more second samples are less than the pre-determined threshold.

31. The method of claim 30 , wherein the similarity metric associated with each second sample is determined based on corresponding distances from that second sample to the one or more first samples that are labeled as being conforming, wherein the distances are within a N-dimensional feature space as defined by N number of features associated with the plurality of second samples or within a M-dimensional process parameter space as defined by M number of process parameters associated with a manufacturing process of the plurality of second samples.

32. The method of claim 31 , further comprising:

determining a probability of failure for each of the identified second samples being lower than a pre-determined threshold probability, wherein the probability is determined based on an associated similarity metric and information associated with the one or more first samples that are labeled as being conforming.

33. The method of claim 22 , further comprising:

capturing a third set of X-ray images for one or more third samples;

analyzing the third set of X-ray images using the adjusted classification algorithm; and

classifying the one or more third samples into conforming or non-conforming categories using the adjusted classification algorithm.

34. One or more computer-readable non-transitory storage media embodying software that is operable when executed to:

access a first set of X-ray images, captured using the X-ray inspection system, of one or more first samples that are labeled as being conforming based on a failure of each of the first samples in a field application;

adjust a classification algorithm based on the first set of X-ray images, wherein the classification algorithm classifies samples into conforming or non-conforming categories based on an analysis of corresponding X-ray images;

analyze a second set of X-ray images of a plurality of second samples using the adjusted classification algorithm in the X-ray inspection system, wherein the second samples are previously inspected samples that have been classified as non-conforming by the classification algorithm during a previous analysis using the X-ray inspection system before the classification algorithm is adjusted; and

identify one or more of the second samples from the second set of X-ray images, wherein each identified second sample is classified as conforming by the adjusted classification algorithm.

35. A system comprising: one or more processors; and one or more computer-readable non-transitory storage media coupled to one or more of the processors and comprising instructions operable when executed by one or more of the processors to cause the system to:

access a first set of X-ray images, captured using the X-ray inspection system, of one or more first samples that are labeled as being conforming based on a failure of each of the first samples in a field application;

adjust a classification algorithm based on the first set of X-ray images, wherein the classification algorithm classifies samples into conforming or non-conforming categories based on an analysis of corresponding X-ray images;

analyze a second set of X-ray images of a plurality of second samples using the adjusted classification algorithm in the X-ray inspection system, wherein the second samples are previously inspected samples that have been classified as non-conforming by the classification algorithm during a previous analysis using the X-ray inspection system before the classification algorithm is adjusted; and

identify one or more of the second samples from the second set of X-ray images, wherein each identified second sample is classified as conforming by the adjusted classification algorithm.