Characteristic variation evaluation method of semiconductor device
First Claim
Patent Images
1. A characteristic variation evaluation method of semiconductor devices, comprising the steps of:
- (a) extracting predetermined parameters concerning a characteristic vale of semiconductor devices and measuring said characteristic value to obtain a measured characteristic value;
(b) eliminating outliers from data measured at said step (a);
(c) selecting a typical semiconductor device having data that is typical among said data with outliers eliminated therefrom, from semiconductor devices used for the measurement of said step (a), and extracting said predetermined parameters of said typical semiconductor device as typical parameters;
(d) selecting specific parameters to be used for the generation of a worst-case parameter, from said predetermined parameters;
(e) generating a worst-case parameter from said specific parameters on the basis of said data with outliers eliminated therefrom; and
(f) verifying a worst-case parameter on the basis of said characteristic value calculated from said typical parameters excepting said specific parameters and said worst-case parameter generated at said step (e), said step (d) including the steps of;
(d-1) extracting equations setting plural combinations of said predetermined parameters, (d-2) inserting parameter values extracted from the semiconductor devices used for the measurement of step (a) into the equations obtained in step (d-1) to obtain a calculated characteristic value;
(d-3) obtaining a correlation coefficient between the measured characteristic value and the calculated characteristic value for each parameter combination in the equations obtained in step (d-1);
(d-4) selecting said specific parameters based on said correlation coefficient.
1 Assignment
0 Petitions
Accused Products
Abstract
Specific parameters used for calculation of a worst-case parameter can be readily selected. Parameters not suitable for characteristic-value expression are eliminated from principal parameters (step ST31), and model equations are extracted (step ST32). Then, parameter values extracted from the same chip are inserted into the model equations to calculate the characteristic value (step ST32). A correlation coefficient between the calculated and measured characteristic values are obtained (step ST33) to select specific parameters (step ST34).
35 Citations
7 Claims
-
1. A characteristic variation evaluation method of semiconductor devices, comprising the steps of:
-
(a) extracting predetermined parameters concerning a characteristic vale of semiconductor devices and measuring said characteristic value to obtain a measured characteristic value;
(b) eliminating outliers from data measured at said step (a);
(c) selecting a typical semiconductor device having data that is typical among said data with outliers eliminated therefrom, from semiconductor devices used for the measurement of said step (a), and extracting said predetermined parameters of said typical semiconductor device as typical parameters;
(d) selecting specific parameters to be used for the generation of a worst-case parameter, from said predetermined parameters;
(e) generating a worst-case parameter from said specific parameters on the basis of said data with outliers eliminated therefrom; and
(f) verifying a worst-case parameter on the basis of said characteristic value calculated from said typical parameters excepting said specific parameters and said worst-case parameter generated at said step (e), said step (d) including the steps of;
(d-1) extracting equations setting plural combinations of said predetermined parameters, (d-2) inserting parameter values extracted from the semiconductor devices used for the measurement of step (a) into the equations obtained in step (d-1) to obtain a calculated characteristic value;
(d-3) obtaining a correlation coefficient between the measured characteristic value and the calculated characteristic value for each parameter combination in the equations obtained in step (d-1);
(d-4) selecting said specific parameters based on said correlation coefficient.
-
-
2. A characteristic variation evaluation method of semiconductor devices, comprising the steps of:
-
(a) extracting predetermined parameters concerning a characteristic value of semiconductor devices and measuring said characteristic value;
(b) eliminating outliers from data measured at said step (a);
(c) selecting a typical semiconductor device having data that is typical among said data with outliers eliminated therefrom, from semiconductor devices used for the measurement of said step (a), and extracting said predetermined parameters of said typical semiconductor device as typical parameters;
(d) selecting specific parameters to be used for the generation of a worst-case parameter, from said predetermined parameters;
(e) generating a worst-case parameter from said specific parameters on the basis of said data with outliers eliminated therefrom; and
(f) verifying a worst-case parameter on the basis of said characteristic value calculated from said typical parameters excepting said specific parameters and said worst-case parameter generated at said step (e), said step (d) including the steps of;
(d-1) obtaining a coefficient concerning said specific parameters; and
(d-2) selecting said specific parameters on the basis of said coefficient wherein said step (d-1) includes the steps of;
determining an equation that expresses said characteristic value by some of said predetermined parameters; and
obtaining a correlation coefficient between said characteristic value measured at said step (a) and a characteristic value calculated from said equation, as said coefficient.
-
-
3. The characteristic variation evaluation method of semiconductor devices, comprising the steps of:
-
(a) extracting predetermined parameters concerning a characteristic value of semiconductor devices and measuring said characteristic value;
(b) eliminating outliers from data measured at said step (a);
(c) selecting a typical semiconductor device having data that is typical among said data with outliers eliminated therefrom, from semiconductor devices used for the measurement of said step (a), and extracting said predetermined parameters of said typical semiconductor device as typical parameters;
(d) selecting specific parameters to be used for the generation of a worst-case parameter, from said predetermined parameters;
(e) generating a worst-case parameter from said specific parameters on the basis of said data with outliers eliminated therefrom; and
(f) verifying a worst-case parameter on the basis of said characteristic value calculated from said typical parameters excepting said specific parameters and said worst-case parameter generated at said step (e), said step (d) including the steps of;
(d-1) obtaining a coefficient concerning said specific parameters; and
(d-2) selecting said specific parameters on the basis of said coefficient wherein said step (d-1) includes the steps of;
obtaining sensitivity from some of said predetermined parameters; and
obtaining respective inner products of said sensitivity and said some of said predetermined parameters to obtain respective correlation coefficients between said characteristic value measured at said step (a) and said inner products, as said coefficient.
-
-
4. The characteristic variation evaluation method of semiconductor devices, comprising the steps of:
-
(a) extracting predetermined parameters concerning a characteristic value of semiconductor devices and measuring said characteristic value;
(b) eliminating outliers from data measured at said step (a);
(c) selecting a typical semiconductor device having data that is typical among said data with outliers eliminated therefrom, from semiconductor devices used for the measurement of said step (a), and extracting said predetermined parameters of said typical semiconductor device as typical parameters;
(d) selecting specific parameters to be used for the generation of a worst-case parameter, from said predetermined parameters;
(e) generating a worst-case parameter from said specific parameters on the basis of said data with outliers eliminated therefrom; and
(f) verifying a worst-case parameter on the basis of said characteristic value calculated from said typical parameters excepting said specific parameters and said worst-case parameter generated at said step (e), said step (d) including the steps of;
(d-1) obtaining a coefficient concerning said specific parameters; and
(d-2) selecting said specific parameters on the basis of said coefficient wherein said step (d-1) includes the steps of;
obtaining sensitivity from some of said predetermined parameters; and
obtaining respective standard partial regression coefficients of said some of said predetermined parameters from said sensitivity, as said coefficient.
-
-
5. The characteristic variation evaluation method of semiconductor devices, comprising the steps of:
-
(a) extracting predetermined parameters concerning a characteristic value of semiconductor devices and measuring said characteristic value;
(b) eliminating outliers from data measured at said step (a);
(c) selecting a typical semiconductor device having data that is typical among said data with outliers eliminated therefrom, from semiconductor devices used for the measurement of said step (a), and extracting said predetermined parameters of said typical semiconductor device as typical parameters;
(d) selecting specific parameters to be used for the generation of a worst-case parameter, from said predetermined parameters;
(e) generating a worst-case parameter from said specific parameters on the basis of said data with outliers eliminated therefrom; and
(f) verifying a worst-case parameter on the basis of said characteristic value calculated from said typical parameters excepting said specific parameters and said worst-case parameter generated at said step (e), said step (d) including the steps of;
(d-1) obtaining a coefficient concerning said specific parameters; and
(d-2) selecting said specific parameters on the basis of said coefficient wherein said step (d-1) includes the step of;
obtaining respective random coefficients of some of said predetermined parameters, as said coefficient.
-
-
6. A characteristic variation evaluation method of semiconductor devices, comprising the steps of:
-
(a) extracting predetermined parameters concerning a characteristic value of semiconductor devices and at the same time measuring said characteristic value;
(b) eliminating outliers from data measured at said step (a);
(c) selecting a typical semiconductor device having data that is typical among said data with outliers eliminated therefrom, from semiconductor devices used for the measurement of said step (a), and extracting said predetermined parameters of said typical semiconductor device as typical parameters;
(d) selecting specific parameters to be used for the generation of a worst-case parameter from said predetermined parameters;
(e) generating a worst-case parameter from said specific parameters on the basis of said data with outliers eliminated therefrom; and
(f) verifying a worst-case parameter on the basis of said characteristic value calculated from said typical parameters excepting said specific parameters and said worst-case parameter generated at said step (e), said step (e) including the steps of;
calculating sensitivity, on the basis of said data with outliers eliminated therefrom, from a relation as follows;
-
-
7. A characteristic variation evaluation method of semiconductor devices, comprising the steps of:
-
(a) extracting predetermined parameters concerning a characteristic value of semiconductor devices and at the same time measuring said characteristic value;
(b) eliminating outliers from data measured at said step (a);
(c) selecting a typical semiconductor device having data that is typical among said data with outliers eliminated therefrom, from semiconductor devices used for the measurement of said step (a), and extracting said predetermined parameters of said typical semiconductor device as typical parameters;
(d) selecting specific parameters to be used for the generation of a worst-case parameter, from said predetermined parameters;
(e) generating a worst-case parameter from said specific parameters on the basis of said data with outliers eliminated therefrom; and
(f) verifying a worst-case parameter on the basis of said characteristic value calculated from said typical parameters excepting said specific parameters and said worst-case parameter generated at said step (e), said step (e) including the steps of;
preparing a table showing a correlation between said sensitivity and said specific parameters; and
calculating a worst-case parameter, using said table, from a relation as follows;
-
Specification