Simulation model creating method, mask data creating method and semiconductor device manufacturing method

Claim 2

A simulation model creating method comprising: computing, for measurement results of a pattern line width of a resist pattern formed by transferring a pattern of a mask onto a wafer with an exposure amount and a focus value varied, a first permissible fluctuation range of the pattern line width from a distribution of the exposure amount and a distribution of the focus value; computing a function which is a polynomial including the exposure amount and focus value as parameters and fits the measurement results of the pattern line width, and computing difference values between measurement values of the measurement results and corresponding approximation values from the function as first differences; comparing the first difference with the first permissible fluctuation range for each measurement value, and, if the first difference exceeds the first permissible fluctuation range, deleting the measurement value and then, until the first difference is less than the first permissible fluctuation range, repeating computation of a function which fits post-deletion measurement results, comparison of the first difference and the first permissible fluctuation range for a new measurement value, and deletion of the new measurement value if the first difference exceeds the first permissible fluctuation range; computing a second permissible fluctuation range for the pattern line width of the resist pattern from an average value and a distribution of a pattern line width of the mask; executing lithography simulation for the mask using a simulation model created based on the measurement results, and computing predicted line widths of a pattern to be formed on a wafer; computing, as second differences, difference values between the measurement values of the measurement results and the predicted line widths; and comparing the second difference with the second permissible fluctuation range for each measurement value, and, if the second difference exceeds the permissible fluctuation range, deleting the measurement value and creating a simulation model based on the post-deletion measurement results.

Claim 3

The simulation model creating method according to claim 1 , wherein the resist pattern is formed by transferring patterns of a plurality of masks, each having a different line width.

Claim 4

The simulation model creating method according to claim 1 , wherein a variance of a distribution of the pattern line width is computed using the variance of the distribution of the exposure amount and the variance of the distribution of the focus value, and the first permissible fluctuation range is computed based on the variance of the distribution of the pattern line width.

Claim 5

The simulation model creating method according to claim 1 , wherein the first permissible fluctuation range is three times a standard deviation of the measurement results.

Claim 6

The simulation model creating method according to claim 1 , wherein the function is a polynomial function with the exposure amount and focus value, and fitting is performed for a case when the exposure amount is constant and for a case when the focus value is constant to calculate the first differences.

Claim 7

The simulation model creating method according to claim 5 , wherein when the exposure amount is denoted “E”, the focus value is denoted “D”, and a, b, c, a′, b′, c′ denote real number coefficients determined in fitting, the polynomial function is expressed as (a·D 2 +b·D+c)·E+(a′·D 2 +b′·D+c′).

Claim 8

The simulation model creating method according to claim 5 , wherein when the exposure amount is denoted “E”, the focus value is denoted “D”, and a, b, c, a′, b′, c′ denote real number coefficients determined in fitting, the polynomial function is expressed as (a·D 2 +b·D+c)·log E+(a′·D 2 +b′·D+c′).

Claim 9

The simulation model creating method according to claim 1 , wherein Mahalanobis distances are computed as the first differences.

Claim 10

The simulation model creating method according to claim 8 , wherein when the exposure amount is denoted “E”, a standard exposure amount is denoted “E 0 ”, the focus value is denoted “D”, and “a 0 ” to “a 8 ” denote real number coefficients determined in fitting, the function is expressed as a 0 +a 1 (1−E 0 /E)+a 2 (1−E 0 /E) 2 +D{a 3 +a 4 (1−E 0 /E)+a 5 (1−E 0 /E) 2 }+D 2 {a 6 +a 7 (1−E 0 /E)+a 8 (1−E 0 /E) 2 }, and, when the measurement value is denoted “w”, an approximation value of the function is denoted “wf”, and a standard deviation of w−wf over all the measurement values is denoted “σ”, the Mahalanobis distance for the measurement value is expressed as |w−wf|/σ.

Claim 11

The simulation model creating method according to claim 1 , wherein Mahalanobis distances are computed as the second differences.

Claim 12

The simulation model creating method according to claim 1 , wherein of functions fitted to the measurement results for each exposure amount after the first difference has become smaller than the first permissible fluctuation range, a function having a maximum value that is a predetermined pattern line width is detected, the exposure amount corresponding to the function is defined as the standard exposure amount, and a focus value at which the function reaches the maximum value is defined as the standard focus value.

Claim 13

The simulation model creating method according to claim 1 , wherein when the pattern line width of the mask is denoted “M”, a mean pattern line width of the mask is denoted “M mean ”, a variance of the distribution is denoted “σ M 2 , and a function expressing the pattern line width of the resist pattern in terms of the exposure amount, the focus value and the pattern line width M of the mask is denoted “f”, the second permissible fluctuation range is computed using a formula expressed as |∂f/∂M|(|M mean |+3σ M ).

Claim 14

A mask data creating method comprising: computing, for measurement results of a pattern line width of a resist pattern formed by transferring a pattern of a mask onto a wafer with an exposure amount and a focus value varied, a first permissible fluctuation range of the pattern line width from a distribution of the exposure amount and a distribution of the focus value; computing a function which is a polynomial including the exposure amount and focus value as parameters and fits the measurement results of the pattern line width, and computing difference values between measurement values of the measurement results and corresponding approximation values from the function as first differences; comparing the first difference with the first permissible fluctuation range for each measurement value, and, if the first difference exceeds the first permissible fluctuation range, deleting the measurement value and then, until the first difference is less than the first permissible fluctuation range, repeating computation of a function which fits post-deletion measurement results, comparison of the first difference and the first permissible fluctuation range for a new measurement value, and deletion of the new measurement value if the first difference exceeds the first permissible fluctuation range; computing a second permissible fluctuation range for the pattern line width of the resist pattern from an average value and a distribution of a pattern line width of the mask; executing lithography simulation for the mask using a first simulation model created based on the measurement results, and computing predicted line widths of a pattern to be formed on a wafer; computing, as second differences, difference values between the measurement values of the measurement results and the predicted line widths; comparing the second difference with the second permissible fluctuation range for each measurement value, and, if the second difference exceeds the permissible fluctuation range, deleting the measurement value and creating a second simulation model based on the post-deletion measurement results; and creating mask data from design data using the second simulation model.

Claim 15

A mask data creating method comprising: computing, for measurement results of a pattern line width of a resist pattern formed by transferring a pattern of a mask onto a wafer with an exposure amount and a focus value varied, a first permissible fluctuation range of the pattern line width from a distribution of the exposure amount and a distribution of the focus value; computing a function which is a polynomial including the exposure amount and focus value as parameters and fits the measurement results of the pattern line width, and computing difference values between measurement values of the measurement results and corresponding approximation values from the function as first differences; comparing the first difference with the first permissible fluctuation range for each measurement value, and, if the first difference exceeds the first permissible fluctuation range, deleting the measurement value and then, until the first difference is less than the first permissible fluctuation range, repeating computation of a function which fits post-deletion measurement results, comparison of the first difference and the first permissible fluctuation range for a new measurement value, and deletion of the new measurement value if the first difference exceeds the first permissible fluctuation range; computing a second permissible fluctuation range for the pattern line width of the resist pattern from an average value and a distribution of a pattern line width of the mask; executing lithography simulation for the mask using a first simulation model created based on the measurement results, and computing predicted line widths of a pattern to be formed on a wafer; computing, as second differences, difference values between the measurement values of the measurement results and the predicted line widths; comparing the second difference with the second permissible fluctuation range for each measurement value, and, if the second difference exceeds the permissible fluctuation range, deleting the measurement value and creating a second simulation model based on the post-deletion measurement results; and creating mask data from design data using the second simulation model.

Claim 16

The mask data creating method according to claim 13 , wherein the resist pattern is formed by transferring patterns of a plurality of masks, each having a different line width.

Claim 17

The mask data creating method according to claim 13 , wherein a variance of a distribution of the pattern line width is computed using a variance of the distribution of the exposure amount and a variance of the distribution of the focus value, and the first permissible fluctuation range is computed based on the variance of the distribution of the pattern line width.

Claim 18

The mask data creating method according to claim 13 , wherein the function is a polynomial function with the exposure amount and focus value, and fitting is performed for a case when the exposure amount is constant and for a case when the focus value is constant to calculate the first differences.

Claim 19

A semiconductor device manufacturing method comprising: computing, for measurement results of a pattern line width of a first resist pattern formed by transferring a pattern of a mask onto a first wafer with an exposure amount and a focus value varied, a first permissible fluctuation range of the pattern line width from a distribution of the exposure amount and a distribution of the focus value; computing a function which is a polynomial including the exposure amount and focus value as parameters and fits the measurement results of the pattern line width, and computing difference values between measurement values of the measurement results and corresponding approximation values from the function as first differences; comparing the first difference with the first permissible fluctuation range for each measurement value, and, if the first difference exceeds the first permissible fluctuation range, deleting the measurement value and then, until the first difference is less than the first permissible fluctuation range, repeating computation of a function which fits post-deletion measurement results, comparison of the first difference and the first permissible fluctuation range for a new measurement value, and deletion of the new measurement value if the first difference exceeds the first permissible fluctuation range; computing a second permissible fluctuation range for the pattern line width of the first resist pattern from an average value and a distribution of a pattern line width of the mask; executing lithography simulation for the mask using a first simulation model created based on the measurement results, and computing predicted line widths of a pattern to be formed on a wafer; computing, as second differences, difference values between the measurement values of the measurement results and the predicted line widths; comparing the second difference with the second permissible fluctuation range for each measurement value, and, if the second difference exceeds the second permissible fluctuation range, deleting the measurement value and creating a second simulation model based on the post-deletion measurement results; creating mask data from design data using the second simulation model; transferring a mask pattern based on the mask data to a photoresist on a second wafer; forming a second resist pattern on the second wafer by development processing; and with the second resist pattern as a mask, processing a film to be processed on the second wafer.

Claim 20

A semiconductor device manufacturing method comprising: computing, for measurement results of a pattern line width of a first resist pattern formed by transferring a pattern of a mask onto a first wafer with an exposure amount and a focus value varied, a first permissible fluctuation range of the pattern line width from a distribution of the exposure amount and a distribution of the focus value; computing a function which is a polynomial including the exposure amount and focus value as parameters and fits the measurement results of the pattern line width, and computing difference values between measurement values of the measurement results and corresponding approximation values from the function as first differences; comparing the first difference with the first permissible fluctuation range for each measurement value, and, if the first difference exceeds the first permissible fluctuation range, deleting the measurement value and then, until the first difference is less than the first permissible fluctuation range, repeating computation of a function which fits post-deletion measurement results, comparison of the first difference and the first permissible fluctuation range for a new measurement value, and deletion of the new measurement value if the first difference exceeds the first permissible fluctuation range; computing a second permissible fluctuation range for the pattern line width of the first resist pattern from an average value and a distribution of a pattern line width of the mask; executing lithography simulation for the mask using a first simulation model created based on the measurement results, and computing predicted line widths of a pattern to be formed on a wafer; computing, as second differences, difference values between the measurement values of the measurement results and the predicted line widths; comparing the second difference with the second permissible fluctuation range for each measurement value, and, if the second difference exceeds the second permissible fluctuation range, deleting the measurement value and creating a second simulation model based on the post-deletion measurement results; creating mask data from design data using the second simulation model; transferring a mask pattern based on the mask data to a photoresist on a second wafer; forming a second resist pattern on the second wafer by development processing; and with the second resist pattern as a mask, processing a film to be processed on the second wafer.

Claim 21

The semiconductor device manufacturing method according to claim 17 , wherein the first resist pattern is formed by transferring patterns of a plurality of masks, each having a different line width.

Claim 22

The semiconductor device manufacturing method according to claim 17 , wherein a variance of a distribution of the pattern line width is computed using a variance of the distribution of the exposure amount and a variance of the distribution of the focus value, and the first permissible fluctuation range is computed based on the variance of the distribution of the pattern line width.

Claim 23

The semiconductor device manufacturing method according to claim 17 , wherein the function is a polynomial function with the exposure amount and focus value, and fitting is performed for a case when the exposure amount is constant and for a case when the focus value is constant to calculate the first differences.