Yield and quality control

Which correctly describes efforts to raise yield by analyzing the large amounts of equipment data, metrology data and test results accumulated in a fab with machine learning and similar methods?

  1. ASimply collecting data automatically brings yield to 100%
  2. BTesting becomes unnecessary
  3. CFinding defect causes and warning signs from relationships among many variables that humans would struggle to see, and using them for improvement and prediction
Show answer and explanation

Answer: C. Finding defect causes and warning signs from relationships among many variables that humans would struggle to see, and using them for improvement and prediction

A semiconductor fab records thousands of process steps and many measurement and test data for each wafer. Linking these and analyzing them with machine learning and similar can reveal defect causes involving many variables, or early signs of abnormalities. Process knowledge remains essential to validate the results.

  • A. Collecting data alone does not raise yield.
  • B. Testing remains necessary.
  • C. Correct. It helps find defect causes and warning signs.
Practice this topic Back to topic

More Yield and quality control questions

  1. Which yield model accounts for defects that occur clustered unevenly on the wafer?
  2. What is the quantity called that, for each type of defect found by inspection, gives the fraction of chips containing that defect that actually fail electrically?
  3. Which correctly states Little's law, relating work in progress (WIP), throughput and cycle time in a factory?
  4. What is the analysis called that statistically identifies, from the manufacturing histories of many lots, the tools, chambers, materials and the like commonly used by low-yield lots?
  5. Which statistical process control technique monitors several mutually correlated characteristics together, rather than one at a time, to detect abnormal combinations?