Method and apparatus for analyzing correlation for semiconductor chips

A method for analyzing a correlation for semiconductor chips, comprising the steps of:

• measuring a geometrically physical quantity of a device in a semiconductor chip and measuring an electrically physical quantity of said semiconductor chip, for each set consisting of a plurality of semiconductor chips, each of the sets being produced under approximately same conditions;
• classifying data of a combination of said geometrically physical quantity and said electrically physical quantity for each of said sets into a plurality of groups based on numeric order of said geometrically physical quantity;
• determining central values of said geometrically physical quantity and said electrically physical quantity for each of said groups;
• and
• examining a correlation between said geometrically physical quantity and said electrically physical quantity.

A semiconductor chip yield adjusting method, comprising the steps of:

• measuring a physical quantity of semiconductor chips or a state quantity of an apparatus for producing said semiconductor chips as a monitored quantity, said semiconductor chips being produced under approximately same conditions and being divided into sets each of which has a plurality of semiconductor chips;
• determining substantially a good or defective product ratio for each of said sets;
• classifying data of combination of said good or defective product ratio and corresponding said monitored quantity into a plurality of groups based on numeric order of said monitored quantity;
• determining central values of said good or defective product ratio and said monitored quantity for each of said groups;
• determining a regression equation with respect to the combination of the central values of said good or defective ratio and the respective central values of said monitored quantity;
• and
• changing the process conditions based on said regression equation so that said good product ratio may increase.

A semiconductor chip yield adjusting method, comprising the steps of:

• measuring a physical quantity of semiconductor chips or a state quantity of an apparatus for producing said semiconductor chips as a monitored quantity, said semiconductor chips being produced under approximately same conditions and being divided into sets each of which has a plurality of semiconductor chips;
• determining substantially a good or defective product ratio for each of said sets;
• classifying data of combination of said good or defective product ratio and corresponding said monitored quantity into a plurality of groups based on numeric order of said monitored quantity;
• determining central values of said good or defective product ratio and said monitored quantity for each of said groups;
• and
• determining a regression equation with respect to the combination of the central values of said good or defective ratio and the respective central values of said monitored quantity,
• wherein said steps of through is performed with respect to a plurality of the monitored quantities, said method further comprising the steps of:
• selecting said monitored quantities whose correlation coefficient are more than a predetermined value and whose absolute inclination value is within a predetermined range;
• determining a multiple regression equation with respect to combinations of the central values of said good or defective ratio and the respective central values of said selected monitored quantities;
• and
• changing the process conditions based on said multiple regression equation so that said good product ratio may increase.

A storage medium having program stored therein, said program comprising the steps of:

• measuring a physical quantity of semiconductor chips or a state quantity of an apparatus for producing said semiconductor chips as a monitored quantity, said semiconductor chips being produced under approximately same conditions and being divided into sets each of which has a plurality of semiconductor chips;
• determining substantially a good or defective product ratio for each of said sets;
• classifying data of combination of said good or defective product ratio and corresponding said monitored quantity into a plurality of groups based on numeric order of said monitored quantity;
• determining central values of said good or defective product ratio and said monitored quantity for each of said groups;
• and
• determining a regression equation with respect to the combination of the central values of said good or defective ratio and the respective central values of said monitored quantity.

An apparatus for analyzing a correlation for semiconductor chips, comprising a computer having program installed therein, said program comprising the steps of:

• measuring a physical quantity of semiconductor chips or a state quantity of an apparatus for producing said semiconductor chips as a monitored quantity, said semiconductor chips being produced under approximately same conditions and being divided into sets each of which has a plurality of semiconductor chips;
• determining substantially a good or defective product ratio for each of said sets;
• classifying data of combination of said good or defective product ratio and corresponding said monitored quantity into a plurality of groups based on numeric order of said monitored quantity;
• determining central values of said good or defective product ratio and said monitored quantity for each of said groups;
• and
• determining a regression equation with respect to the combination of the central values of said good or defective ratio and the respective central values of said monitored quantity.