Mask calibration method for mask defect detection

Abstract

The invention discloses a mask calibration method for mask defect detection, and the method comprises the steps: S1, selecting a selected region with a pattern in a mask design file, and outputting a vector point of the selected region; s2, acquiring an actual image corresponding to the selected area on the corresponding real object mask by adopting AOI (Automatic Optic Inspection) equipment; s3, a horizontal tolerance range is set in the horizontal direction of the selected area, and S31, different horizontal tolerance compensation values in the horizontal tolerance range are adopted to fill and render the vector points output in the step S1 into a first rendered image; first rendering images obtained through rendering of the different horizontal tolerance compensation values are registered with the actual image, the influence of the different horizontal tolerance compensation values on the integrating degree of the first rendering images and the actual image is evaluated through an evaluation method, and the horizontal tolerance compensation value with the best evaluation result serves as the tolerance calibrated in the horizontal direction. The method is beneficial to improving the precision and accuracy of defect detection.

Description

Technical Field

[0001] The invention relates to a mask calibration method for mask defect detection. Background Art

[0002] At present, in the process of defect detection of semiconductor masks, the DIE to DB (D2DB) mode is generally used for detection, that is, the image of the actual mask produced in the end is compared with the mask design file to identify defects. However, in the process of mask lithography, due to different process levels, the graphic size of the manufactured mask will have certain tolerances, and these tolerances will lead to the appearance of pseudo defects in the subsequent defect detection, and sometimes it is more difficult to determine whether it is a real defect. Summary of the invention

[0003] The technical problem to be solved by the present invention is to overcome the defects of the prior art and provide a mask calibration method for mask defect detection, which is conducive to improving the precision and accuracy of defect detection.

[0004] In order to solve the above technical problems, the technical solution of the present invention is: a mask calibration method for mask defect detection, the method steps comprising:

[0005] S1: Select a selected area with a pattern in the mask design file, and output vector points of the selected area;

[0006] S2: Using AOI inspection equipment to collect the actual image corresponding to the selected area on the corresponding physical mask;

[0007] S3: setting a horizontal tolerance range in the horizontal direction of the selected area, and performing step S31: using different horizontal tolerance compensation values ​​within the horizontal tolerance range to fill and render the vector points outputted in step S1 into first rendered images; registering the first rendered images rendered with different horizontal tolerance compensation values ​​with the actual image respectively, using an evaluation method to evaluate the influence of different horizontal tolerance compensation values ​​on the fit between the first rendered image and the actual image, and using the horizontal tolerance compensation value with the best evaluation result as the tolerance for horizontal calibration;

[0008] S4: setting a vertical tolerance range in the vertical direction of the selected area, and performing step S41: using different vertical tolerance compensation values ​​within the vertical tolerance range to fill and render the vector points output from step S1 into a second rendered image; registering the second rendered images rendered with different vertical tolerance compensation values ​​with the actual image, and using an evaluation method to evaluate the effects of different vertical tolerance compensation values ​​on the fit between the second rendered image and the actual image, and using the vertical tolerance compensation value with the best evaluation result as the tolerance for vertical calibration.

[0009] Furthermore, in order to obtain a more accurate horizontal tolerance compensation value, in step S3, the horizontal tolerance compensation value with the best evaluation result is used as the horizontal tolerance range of the secondary evaluation calibration, and step S31 is repeated to obtain the final calibrated tolerance in the horizontal direction.

[0010] Furthermore, in order to obtain a more accurate vertical tolerance compensation value, in step S4, the vertical tolerance compensation value with the best evaluation result is used as the vertical tolerance range for secondary evaluation and calibration, and step S41 is repeated to obtain the final calibrated tolerance in the vertical direction.

[0011] Further, in step S3, the evaluation method is specifically:

[0012] Obtaining vertical edges of the first rendered image and the actual image;

[0013] Demarcate the vertical edge area based on the vertical edge;

[0014] Obtaining a first difference value sum between the first rendered image and the actual image in the vertical edge region, that is, a first error sum;

[0015] The first difference value and the minimum are used as the evaluation target to obtain the corresponding horizontal tolerance compensation value as the tolerance of horizontal direction calibration.

[0016] The vertical edges of the first rendered image and the actual image are obtained by performing horizontal differentiation on the first rendered image or the actual image, setting a horizontal differentiation threshold, and marking a region where the horizontal differentiation is greater than the horizontal differentiation threshold and the vertical differentiation is close to 0 as a vertical edge;

[0017] Further, the vertical edge region is calibrated based on the vertical edge and specifically includes: setting a set pixel range adjacent to the vertical edge as the vertical edge region.

[0018] Furthermore, in order to obtain a more accurate horizontal tolerance compensation value, a fitting method is used between the horizontal tolerance compensation value and the first difference value to obtain the minimum first difference value.

[0019] Further, in step S4, the evaluation method is specifically:

[0020] Obtaining horizontal edges of the second rendered image and the actual image;

[0021] calibrating the horizontal edge area based on the horizontal edge;

[0022] Obtaining a second difference value sum between the second rendered image and the actual image in the horizontal edge region, that is, a second error sum;

[0023] The second difference value and the minimum are used as the corresponding vertical tolerance compensation value obtained by evaluating the target as the tolerance for vertical direction calibration.

[0024] Further, obtaining the horizontal edge of the second rendered image and the actual image specifically includes: obtaining the horizontal edge of the second rendered image and the actual image specifically includes: performing vertical differentiation on the second rendered image or the actual image, and setting a vertical differentiation threshold, marking a region where the vertical differentiation is greater than the vertical differentiation threshold and the horizontal differentiation is close to 0 as a horizontal edge;

[0025] Further, the horizontal edge region is calibrated based on the horizontal edge and specifically includes: setting a set pixel range adjacent to the horizontal edge as the horizontal edge region.

[0026] Furthermore, in order to obtain a more accurate vertical tolerance compensation value, a fitting method is used between the vertical tolerance compensation value and the second differential value and to obtain the minimum second differential value and.

[0027] Furthermore, the specific steps of rendering in step S3 and step S4 include:

[0028] Scale the graphics corresponding to the vector points according to the tolerance compensation value;

[0029] Then fill the sub-pixel edge of the corrected vector point;

[0030] The filled graphics are modified using the selected model to obtain the corresponding rendered image.

[0031] After adopting the above technical solution, the calibration method proposed by the present invention can perform high-precision detection of the mask within the tolerance range without causing a large number of pseudo defects; furthermore, the method of calibrating the horizontal and vertical directions separately in the present invention makes the tolerances of the two dimensions independent and will not affect each other and cause the calibration tolerance to be incorrect. At the same time, only the information of the vertical and horizontal edges is processed, and the noise caused by a large number of corners can be ignored, which greatly improves the tolerance and credibility of the calibration; secondly, the present invention evaluates the quality of the tolerance compensation value by means of error sum, reduces unnecessary noise caused by registration, and improves the accuracy of calibration. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 It is a vector point representation of a selected area in the mask design file; the intersection points are the positions of each point, and the dotted lines are polygons formed by connecting each point;

[0033] Figure 2(a) shows Figure 1 The actual image on the mask corresponding to the selected area;

[0034] Figure 2(b) shows Figure 1 A rendered image obtained by low-pass filtering the selected area;

[0035] Figure 3(a) is a difference diagram between the actual image and the rendered image without tolerance correction;

[0036] Figure 3(b) is a difference diagram between the actual image and the tolerance-corrected rendered image;

[0037] Figure 4 It is a scatter plot of the evaluation error results obtained for different tolerance corrections in the present invention. DETAILED DESCRIPTION

[0038] In order to make the contents of the present invention more clearly understood, the present invention is further described in detail below based on specific embodiments in conjunction with the accompanying drawings.

[0039] like Figures 1 to 4 As shown, a mask calibration method for mask defect detection includes the following steps:

[0040] S1: Select a selected area with graphics in the mask design file through EDA software, and output the vector points of the selected area, such as Figure 1 As shown;

[0041] S2: After the AOI inspection equipment is used to align the corresponding physical mask, the inspection camera of the AOI inspection equipment is used to capture the actual image corresponding to the selected area on the corresponding physical mask, as shown in FIG2(a); wherein the purpose of the alignment is to map the platform coordinates of the AOI inspection equipment and the coordinates of the mask design file, so as to capture the actual image corresponding to the selected area marked in step S1;

[0042] S3: setting a horizontal tolerance range in the horizontal direction (i.e., X direction) of the selected area, and performing step S31: using different horizontal tolerance compensation values ​​within the horizontal tolerance range to fill and render the vector points outputted in step S1 into a first rendered image; registering the first rendered images rendered with different horizontal tolerance compensation values ​​with the actual image respectively, using an evaluation method to evaluate the influence of different horizontal tolerance compensation values ​​on the fit between the first rendered image and the actual image, and using the horizontal tolerance compensation value with the best evaluation result as the tolerance for horizontal calibration; wherein, using different horizontal tolerance compensation values ​​within the horizontal tolerance range specifically comprises: uniformly selecting a number of appropriate numerical points within the horizontal tolerance range, and using these numerical points as different horizontal tolerance compensation values;

[0043] S4: setting a vertical tolerance range in the vertical direction (i.e., Y direction) of the selected area, and performing step S41: using different vertical tolerance compensation values ​​within the vertical tolerance range to fill and render the vector points outputted in step S1 into a second rendered image; registering the second rendered images rendered with different vertical tolerance compensation values ​​with the actual image respectively, and using an evaluation method to evaluate the influence of different vertical tolerance compensation values ​​on the fit between the second rendered image and the actual image, and using the vertical tolerance compensation value with the best evaluation result as the tolerance for vertical calibration; wherein, using different vertical tolerance compensation values ​​within the vertical tolerance range is specifically: uniformly selecting a number of suitable numerical points within the vertical tolerance range, and using these numerical points as different vertical tolerance compensation values.

[0044] In this embodiment, during steps S3 to S4, a multi-threading approach may be used to increase the calculation speed.

[0045] In this embodiment, the registration method in step S3 and step S4 can use phase registration to achieve sub-pixel level to improve the registration accuracy.

[0046] Specifically, in order to obtain a more accurate horizontal tolerance compensation value, in step S3, the horizontal tolerance compensation value with the best evaluation result is used as the horizontal tolerance range for secondary evaluation and calibration, and step S31 is repeated to obtain the final calibrated tolerance in the horizontal direction.

[0047] Specifically, in order to obtain a more accurate vertical tolerance compensation value, in step S4, the vertical tolerance compensation value with the best evaluation result is used as the vertical tolerance range for secondary evaluation and calibration, and step S41 is repeated to obtain the final calibrated tolerance in the vertical direction.

[0048] Specifically, in step S3, the evaluation method is as follows:

[0049] Obtaining vertical edges of the first rendered image and the actual image;

[0050] Demarcate the vertical edge area based on the vertical edge;

[0051] Obtaining a first difference value sum between the first rendered image and the actual image in the vertical edge region, that is, a first error sum;

[0052] The first difference value and the minimum are used as the evaluation target to obtain the corresponding horizontal tolerance compensation value as the tolerance of horizontal direction calibration.

[0053] The vertical edges of the first rendered image and the actual image are obtained as follows: performing horizontal differentiation on the first rendered image or the actual image, and setting a horizontal differentiation threshold, marking a region where the horizontal differentiation is greater than the horizontal differentiation threshold and the vertical differentiation is close to 0 (set according to the uniformity of light, generally less than 5) as a vertical edge;

[0054] Specifically, the vertical edge region is calibrated based on the vertical edge as follows: a set pixel range adjacent to the vertical edge is set as the vertical edge region; in this embodiment, the set pixel range may be a 3-pixel range, and may be other numbers.

[0055] Specifically, in order to obtain a more accurate horizontal tolerance compensation value, a fitting method is used between the horizontal tolerance compensation value and the first difference value and to obtain the minimum first difference value and.

[0056] Specifically, in step S4, the evaluation method is as follows:

[0057] Obtaining horizontal edges of the second rendered image and the actual image;

[0058] Calibrate the horizontal edge area based on the horizontal edge;

[0059] Obtaining a second difference value sum between the second rendered image and the actual image in the horizontal edge region, that is, a second error sum;

[0060] The second difference value and the minimum are used as the corresponding vertical tolerance compensation value obtained by evaluating the target as the tolerance for vertical direction calibration.

[0061] Further, obtaining the horizontal edge of the second rendered image and the actual image specifically includes: performing vertical differentiation on the second rendered image or the actual image, and setting a vertical differentiation threshold, marking a region where the vertical differentiation is greater than the vertical differentiation threshold and the horizontal differentiation is close to 0 (set according to the uniformity of light, generally less than 5) as a horizontal edge;

[0062] Specifically, the horizontal edge region is calibrated based on the horizontal edge as follows: a set pixel range adjacent to the horizontal edge is set as the horizontal edge region. In this embodiment, the set pixel range may be a 3 pixel range, and of course may be other numbers.

[0063] Specifically, in order to obtain a more accurate vertical tolerance compensation value, a fitting method is used between the vertical tolerance compensation value and the second differential value and to obtain the minimum second differential value and.

[0064] Specifically, the rendering in step S3 and step S4 is rendered into a bitmap based on the point set of the mask design file. In the rendering process, it is necessary to use corresponding models, such as low-pass filter models, scalar optical simulation models, and vector optical simulation models, to get close to the actual mask drawing. The rendering result is shown in Figure 2(b). The specific selection method can be selected based on the acceptable error. According to the set tolerance range, a larger interval is used to divide the tolerance range, and a graphic rendering is performed for each tolerance compensation value. The specific steps of rendering in step S3 and step S4 include:

[0065] Scale the graphics corresponding to the vector points according to the tolerance compensation value;

[0066] Then fill the sub-pixel edge of the corrected vector point;

[0067] The filled graphics are modified using the selected model to obtain the corresponding rendered image.

[0068] In this embodiment, the difference method in step S3 and step S4, that is, by directly summing the difference values, can well neutralize the error caused by inaccurate registration, because inaccurate registration often leads to the opposite signs of the difference values ​​of the edges on both sides of the symmetry of the figure. If the tolerance value is correct, then the sum of these difference values ​​will be approximately equal to 0. Only when there is a tolerance error will there be a larger value, and the larger the deviation of the tolerance, the larger the value of the difference sum. This is also the reason why this difference method can be used to determine the quality of the tolerance compensation value in the horizontal or vertical direction. At the same time, because the difference only calculates the error at the horizontal edge or the vertical edge, the error at the corner will not be calculated, especially the error at the corner is ignored. Since the process level has more uncontrollable effects on the corner, the corner often provides a lot of noise information. Similarly, because the horizontal and vertical differences are calculated separately, the tolerances of the horizontal and vertical dimensions are cleverly separated, so that higher accuracy can be provided. As shown in Figures 3(a) and 3(b), if tolerance compensation is not performed, there is an obvious deviation between the rendering and the actual image. After the tolerance compensation value calibrated with the minimum error is selected for the mask design, the rendering and the actual image become more consistent, with only a small deviation.

[0069] The specific embodiments described above further illustrate the technical problems, technical solutions and beneficial effects solved by the present invention. It should be understood that the above are only specific embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A mask calibration method for mask defect detection, characterized in that: The steps of the method include: S1: Select a selected area with a pattern in the mask design file, and output vector points of the selected area; S2: Using AOI inspection equipment to collect the actual image corresponding to the selected area on the corresponding physical mask; S3: setting a horizontal tolerance range in the horizontal direction of the selected area, and performing step S31: using different horizontal tolerance compensation values ​​within the horizontal tolerance range to fill and render the vector points outputted in step S1 into first rendered images; registering the first rendered images rendered with different horizontal tolerance compensation values ​​with the actual image respectively, using an evaluation method to evaluate the influence of different horizontal tolerance compensation values ​​on the fit between the first rendered image and the actual image, and using the horizontal tolerance compensation value with the best evaluation result as the tolerance for horizontal calibration; S4: setting a vertical tolerance range in the vertical direction of the selected area, and performing step S41: using different vertical tolerance compensation values ​​within the vertical tolerance range to fill and render the vector points output from step S1 into a second rendered image; registering the second rendered images rendered with different vertical tolerance compensation values ​​with the actual image, and using an evaluation method to evaluate the effects of different vertical tolerance compensation values ​​on the fit between the second rendered image and the actual image, and using the vertical tolerance compensation value with the best evaluation result as the tolerance for vertical calibration.

2. The method according to claim 1, characterized in that In step S3, the horizontal tolerance compensation value with the best evaluation result is used as the horizontal tolerance range of the secondary evaluation calibration, and step S31 is repeated to obtain the final calibrated tolerance in the horizontal direction.

3. The method according to claim 1, characterized in that In step S4, the vertical tolerance compensation value with the best evaluation result is used as the vertical tolerance range of the secondary evaluation calibration, and step S41 is repeated to obtain the final calibrated tolerance in the vertical direction.

4. The method according to claim 1, characterized in that: In step S3, the evaluation method is specifically as follows: Obtaining vertical edges of the first rendered image and the actual image; Demarcate the vertical edge area based on the vertical edge; Obtaining a first difference value sum between the first rendered image and the actual image in the vertical edge region, that is, a first error sum; The first difference value and the minimum are used as the evaluation target to obtain the corresponding horizontal tolerance compensation value as the tolerance of horizontal direction calibration.

5. The method according to claim 4, characterized in that The vertical edges of the first rendered image and the actual image are obtained by performing horizontal differentiation on the first rendered image or the actual image, setting a horizontal differentiation threshold, and marking a region where the horizontal differentiation is greater than the horizontal differentiation threshold and the vertical differentiation is close to 0 as a vertical edge; The vertical edge region is calibrated based on the vertical edge and specifically includes: setting a set pixel range adjacent to the vertical edge as the vertical edge region.

6. The method according to claim 4, characterized in that A fitting method is adopted between the horizontal tolerance compensation value and the first difference value and to obtain the minimum first difference value and.

7. The method according to claim 1, characterized in that In step S4, the evaluation method is specifically as follows: Obtaining horizontal edges of the second rendered image and the actual image; calibrating the horizontal edge area based on the horizontal edge; Obtaining a second difference value sum between the second rendered image and the actual image in the horizontal edge region, that is, a second error sum; The second difference value and the minimum are used as the corresponding vertical tolerance compensation value obtained by evaluating the target as the tolerance for vertical direction calibration.

8. The method according to claim 6, characterized in that The step of obtaining the horizontal edge of the second rendered image and the actual image is as follows: vertically differentiating the second rendered image or the actual image, setting a vertical differential threshold, and marking a region where the vertical differential is greater than the vertical differential threshold and the horizontal differential is close to 0 as a horizontal edge; The horizontal edge region is calibrated based on the horizontal edge and specifically includes: setting a set pixel range adjacent to the horizontal edge as the horizontal edge region.

9. The method according to claim 7, characterized in that: A fitting method is used between the vertical tolerance compensation value and the second differential value to obtain the minimum second differential value.

10. The method according to claim 1, characterized in that The specific steps of rendering in step S3 and step S4 include: Scale the graphics corresponding to the vector points according to the tolerance compensation value; Then, the sub-pixel edge of the corrected vector points is filled; The filled graphics are modified by the selected model to obtain the corresponding rendered image.

Images