Defect detecting apparatus, defect detecting method, information processing apparatus, information processing method, and program therefor

A defect detecting method comprising:

• capturing, with a first magnification, an image of a microstructure formed on each of a plurality of dies on a semiconductor wafer, with respect to every first division region obtained by dividing each die in plurality;
• storing the image of each first division region together with a first identification information for identifying a position of each first division region within each die as a first inspection target image;
• creating an average image as a first model image for the every first identification information, the average image being obtained by averaging, among the first inspection target images, the first inspection target images of respective first division regions having the corresponding first identification information over the dies;
• detecting presence or absence of a defect of the microstructure on each first division region within one of the dies by comparing the first model image with each first inspection target image corresponding to the first identification information of the first model image;
• capturing, with a second magnification higher than the first magnification, an image of the microstructure on the first division region from which the defect is detected and microstructures on respective first division regions, within other dies, corresponding to the first identification information of the first division region from which the defect is detected, with respect to every second division region obtained by dividing each first division region in plurality;
• storing the image of each second division region together with second identification information for identifying a position of each second division region within each die as a second inspection target image;
• creating an average image as a second model image for every second identification information, the second average image being obtained by averaging, among the second inspection target images, the second inspection target images of respective second division regions having the corresponding second identification information over the dies;
• and
• determining the kind of the detected defect by comparing the second model image with each second inspection target image corresponding to the second identification information of the second model image.

An information processing apparatus comprising:

• a storage unit configured to store therein a captured image with a first magnification, of a microstructure formed on each of a plurality of dies on a semiconductor wafer, with respect to every first division region obtained by dividing each die in plurality together with first identification information for identifying a position of each first division region within each die as a first inspection target image;
• a model image creating unit configured to create an average image as a first model image, the average image being obtained by averaging, among the first inspection target images, the first inspection target images of respective first division regions having the corresponding first identification information over the dies;
• a detecting unit configured to detect presence or absence of a defect of the microstructure on each first division region within one of the dies by comparing the first model image with each first inspection target image corresponding to the first identification information of the first model image;
• a control unit configured to control the storage unit to store therein captured images with a second magnification higher than the first magnification, of the microstructure on the first division region from which the defect is detected and microstructures on respective first division regions, within other dies, corresponding to the first identification information of the first division region from which the defect is detected, with respect to every second division region obtained by dividing each first division region in plurality together with second identification information for identifying a position of each second division region within each die as a second inspection target image, and control the model image creating unit to create an average image as second model image for every second identification information, the second average image being obtained by averaging, among the second inspection target images, the second inspection target images of respective second division regions having the corresponding second identification information over the dies;
• and
• a defect classifying unit configured to determine the kind of the detected defect by comparing the second model image with each second inspection target image corresponding to the second identification information of the second model image.

An information processing method comprising:

• storing a captured image with a first magnification, of a microstructure formed on each of a plurality of dies on a semiconductor wafer, with respect to every first division region obtained by dividing each die in plurality together with first identification information for identifying a position of each first division region within each die as a first inspection target image;
• creating an average image as a first model image, the average image being obtained by averaging, among the first inspection target images, the first inspection target images of respective first division regions having the corresponding first identification information over the dies;
• detecting presence or absence of a defect of the microstructure on each first division region within one of the dies by comparing the first model image with each first inspection target image corresponding to the first identification information of the first model image;
• storing captured images with a second magnification higher than the first magnification, of the microstructure on the first division region from which the defect is detected and microstructures on respective first division regions, within other dies, corresponding to the first identification information of the first division region from which the defect is detected, with respect to every second division region obtained by dividing each first division region in plurality together with second identification information for identifying a position of each second division region within each die as a second inspection target image;
• creating an average image as a second model image for every second identification information, the second average image being obtained by averaging, among the second inspection target images, the second inspection target images of respective second division regions having the corresponding second identification information over the dies;
• and
• determining the kind of the detected defect by comparing the second model image with each second inspection target image corresponding to the second identification information of the second model image.

A program for executing, in an information processing apparatus, the processes of:

• storing a captured image with a first magnification, of a microstructure formed on each of a plurality of dies on a semiconductor wafer, with respect to every first division region obtained by dividing each die in plurality together with first identification information for identifying a position of each first division region within each die as a first inspection target image;
• creating an average image as a first model image, the average image being obtained by averaging, among the first inspection target images, the first inspection target images of respective first division regions having the corresponding first identification information over the dies;
• detecting presence or absence of a defect of the microstructure on each first division region within one of the dies by comparing the first model image with each first inspection target image corresponding to the first identification information of the first model image;
• storing captured images with a second magnification higher than the first magnification, of the microstructure on the first division region from which the defect is detected and microstructures on respective first division regions, within other dies, corresponding to the first identification information of the first division region from which the defect is detected, with respect to every second division region obtained by dividing each first division region in plurality together with second identification information for identifying a position of each second division region within each die as a second inspection target image;
• creating an average image as a second model image for every second identification information, the second average image being obtained by averaging, among the second inspection target images, the second inspection target images of respective second division regions having the corresponding second identification information over the dies;
• and
• determining the kind of the detected defect by comparing the second model image with each second inspection target image corresponding to the second identification information of the second model image.