Virtual quality control interpolation and process feedback in the production of memory devices
To provide more test data during the manufacture of non-volatile memories and other integrated circuits, machine learning is used to generate virtual test values. Virtual test results are interpolated for one set of tests for devices on which the test is not performed based on correlations with other sets of tests. In one example, machine learning determines a correlation study between bad block values determined at die sort and photo-limited yield (PLY) values determined inline during processing. The correlation can be applied to interpolate virtual inline PLY data for all of the memory dies, allowing for more rapid feedback on the processing parameters for manufacturing the memory dies and making the manufacturing process more efficient and accurate. In another set of embodiments, the machine learning is used to extrapolate limited metrology (e.g., critical dimension) test data to all of the memory die through interpolated virtual metrology data values.
1. A method, comprising:
manufacturing a first plurality of an integrated circuit according to a first set of processing parameter values;
selecting a subset of the first plurality of the integrated circuit during manufacturing as first test samples;
performing one or more first tests on the first test samples;
performing one or more second tests on the first plurality of the integrated circuit subsequent to completing the manufacturing of the first plurality of the integrated circuit;
performing a machine learning process to determine a correlation between results of the one or more first tests and results of the one or more second tests;
subsequent to performing the machine learning process, manufacturing a second plurality of the integrated circuit according to the first set of processing parameter values;
performing the one or more second tests on the second plurality of the integrated circuit; and
interpolating results of the one or more first tests for members of the second plurality of the integrated circuit from the correlation and results of the one or more second tests on the second plurality of the integrated circuit.
2. The method of claim 1 , further comprising:
updating the first set of processing parameter values based on the interpolated results.
3. The method of claim 1 , further comprising:
determining one or more criteria for selecting test samples of the integrated circuit based on results of the one or more second tests on the first plurality of the integrated circuit.
4. The method of claim 1 , further comprising:
performing the one or more first tests on a subset of the second plurality of the integrated circuit, wherein the interpolated results are determined for all of the second plurality of the integrated circuit that are not members of the subset of the second plurality of the integrated circuit on which the one or more first tests are performed.
5. The method of claim 1 , wherein the machine learning process includes a generalized linear model.
6. The method of claim 1 , wherein the machine learning process includes a gradient boosting machine.
7. The method of claim 1 , wherein the machine learning process includes a deep neural network.
8. The method of claim 1 , wherein the integrated circuit is a non-volatile memory circuit.
9. The method of claim 8 , wherein the one or more second tests includes testing for bad memory blocks of the non-volatile memory circuit.
10. The method of claim 8 , wherein the one or more second tests includes determining an early failure rate.
11. The method of claim 1 , wherein the one or more first tests include a photo inspection test.
12. The method of claim 11 , wherein the photo inspection test is performed using a scanning electron microscope.
13. The method of claim 11 , wherein the photo inspection test is performed to determine a result of an etch of a sacrificial layer.
14. The method of claim 1 , wherein one or more of the second tests are performed as part of a die sort process.
15. A method, comprising:
manufacturing a first plurality of an integrated circuit;
selecting a subset of the first plurality of the integrated circuit as first test samples;
performing one or more first tests on the first test samples to determine a corresponding one or more metrology data values for each of the first test samples;
performing one or more second tests on the first plurality of the integrated circuit;
performing a machine learning process to determine a correlation between the metrology data values and results of the one or more second tests;
subsequent to performing the machine learning process, manufacturing a second plurality of the integrated circuit;
performing the one or more second tests on the second plurality of the integrated circuit; and
interpolating metrology data values for members of the second plurality of the integrated circuit from the correlation and results of the one or more second tests on the second plurality of the integrated circuit.
16. The method of claim 15 , wherein the integrated circuit is a non-volatile memory circuit.
17. The method of claim 16 , wherein the one or more second tests includes testing for bad memory blocks of the non-volatile memory circuit.
18. The method of claim 16 , wherein the one or more second tests includes determining an early failure rate.
19. The method of claim 15 , further comprising:
determining one or more criteria selecting test samples of the integrated circuit based on results of the one or more second tests on the first plurality of the integrated circuit.
20. The method of claim 15 , wherein the first plurality of the integrated circuit are manufactured using a first set of processing parameters and the method further comprises:
updating the first set of processing parameter values based on the metrology data values.