Pattern measurement system, pattern measurement method, and program for measuring edge roughness at the edge of a pattern based on a random noise component

A pattern measurement method that uses a signal obtained by charged particle beam scan or light irradiation with respect to a sample to measure a pattern formed on the sample, the method comprising:

• generating, by a computer that reads a program from a storage device that stores the program for processing the information of the pattern and processes the information of the pattern, based on the signal obtained from the pattern, a plurality of individual line profiles, a sum line profile of the plurality of line profiles, and an original image of the pattern, wherein each of the individual line profiles is created from a line scan of the original image;
• dividing, by the computer, each of the plurality of individual line profiles into two regions;
• executing, by the computer, for each of the plurality of individual line profiles each divided into the two regions and the sum line profile, pattern matching between each of the individual line profiles and the sum line profile;
• executing, by the computer, position correction with respect to each of the plurality of individual line profiles on the basis of the result of the pattern matching;
• generating, by the computer, a correction image of the pattern from the plurality of line profiles that have been position corrected;
• calculating, by the computer, power spectrum density at the edge of the pattern in the correction image and averaging, by the computer, the power spectrum density to calculate a random noise component;
• and
• calculating, by the computer, edge roughness by taking the difference between power spectrum density at the edge of the pattern in the original image and the random noise component;
• generating, by the computer, original images of the pattern having a plurality of types of summation numbers of frames;
• calculating, by the computer, a random noise amount with respect to the original image having each of the summation numbers of frames;
• and
• deciding, by the computer, an optimum summation number of frames based on a relationship between the summation number of frames and the random noise amount, wherein the optimum summation number of frames equals a predetermined ratio of the random noise amount to the summation number of frames.

A non-transitory computer-readable medium storing a program for causing a computer to execute a pattern measurement process that uses a signal obtained by charged particle beam scan or light irradiation with respect to a sample to measure a pattern formed on the sample, the program for causing the computer to execute:

• a process by which the computer that processes the information of the pattern generates, based on the signal obtained from the pattern, a plurality of individual line profiles, a sum line profile of the plurality of line profiles, and an original image of the pattern, wherein each of the individual line profiles is created from a line scan of the original image;
• a process by which the computer divides each of the plurality of individual line profiles into two regions;
• a process by which the computer executes, for each of the plurality of individual line profiles each divided into the two regions and the sum line profile, pattern matching between each of the individual line profiles and the sum line profile;
• a process by which the computer executes position correction with respect to each of the plurality of individual line profiles on the basis of the result of the pattern matching;
• a process by which the computer generates a correction image of the pattern from the plurality of line profiles that have been position corrected;
• a process by which the computer calculates power spectrum density at the edge of the pattern in the correction image and averaging, by the computer, the power spectrum density to calculate a random noise component;
• a process by which the computer calculates edge roughness by taking the difference between power spectrum density at the edge of the pattern in the original image and the random noise component;
• a process for generating original images of the pattern having a plurality of types of summation numbers of frames;
• a process for calculating a random noise amount with respect to the original image having each of the summation numbers of frames;
• and
• a process for deciding an optimum summation number of frames based on a relationship between the summation number of frames and the random noise amount, wherein the optimum summation number of frames equals a predetermined ratio of the random noise amount to the summation number of frames.