Calibration method of a photomask for detecting photomask defects

Abstract

The present invention discloses a method for calibrating a photomask for detecting defects in a photomask, wherein the steps of the method include: S1: selecting a selected area containing a graph in a photomask design document and outputting vector points of said selected area; S2: collecting actual images corresponding to the selected area in a corresponding actual photomask using an AOI detection facility; S3: setting a horizontal tolerance range in the horizontal direction of the selected area and proceeding to step S31; filling the vector points output in step S1 with different horizontal tolerance compensation values ​​within the horizontal tolerance range to render them into a first rendering image; aligning the first rendering image obtained by rendering with different horizontal tolerance compensation values ​​with the actual image, evaluating the influence of different horizontal tolerance compensation values ​​on the degree of agreement between the first rendering image and the actual image using an evaluation method, and using the horizontal tolerance compensation value with the best evaluation result as the tolerance for horizontal correction. Using the said method is advantageous for improving the precision and accuracy of defect detection.

Description

Technology Field

[0001] The present invention relates to a method for calibrating a photomask for detecting defects in a photomask. Background Technology

[0002] Currently, the process of detecting defects in semiconductor masks typically utilizes the DIE to DB (D2DB) mode; that is, defects are identified by comparing the actual image of the final-produced photomask with the mask's design document. However, during the photo-etching process, there is a certain tolerance in the graph size of the manufactured photomask due to differences in process levels. This tolerance can lead to the appearance of false defects during subsequent defect detection, sometimes making it more difficult to determine the authenticity of the defects. The problem to be solved

[0003] The technical problem that the present invention aims to solve is to eliminate defects in the prior art and to provide a method for calibrating a photomask for detecting photomask defects, and using the said method is advantageous for improving the precision and accuracy of defect detection. means of solving the problem

[0004] The technical solution of the present invention, which is intended to solve the above technical problem, provides a method for correcting a photomask for detecting photomask defects, and the steps of the method are,

[0005] S1: A step of selecting a selected area containing a graph in a photomask design document and outputting vector points of the selected area;

[0006] S2: A step of collecting actual images corresponding to selected areas in a corresponding physical photomask using AOI detection equipment;

[0007] S3: A step of setting a horizontal tolerance range in the horizontal direction of a selected area and proceeding with step S31; filling the vector points output in step S1 with different horizontal tolerance compensation values ​​within the horizontal tolerance range to render them into a first rendering image; aligning the first rendering image obtained by rendering with different horizontal tolerance compensation values ​​with the actual image, evaluating the influence of the different horizontal tolerance compensation values ​​on the degree of agreement between the first rendering image and the actual image using an evaluation method, and using the horizontal tolerance compensation value with the best evaluation result as the tolerance for horizontal correction;

[0008] S4: Set a horizontal tolerance range in the vertical direction of the selected area and proceed with step S41; fill the vector points output in step S1 using different vertical tolerance compensation values ​​within the vertical tolerance range to render them into a second rendering image; align the second rendering image obtained by rendering the different vertical tolerance compensation values ​​with the actual image, evaluate the influence of the different vertical tolerance compensation values ​​on the degree of agreement between the second rendering image and the actual image using an evaluation method, and use the vertical tolerance compensation value with the best evaluation result as the tolerance for vertical correction;

[0009] Furthermore, in order to obtain a more accurate horizontal tolerance compensation value, step S3 uses the horizontal tolerance compensation value with the best evaluation result as the horizontal tolerance range for the second evaluation correction and repeats step S31 to obtain the tolerance for the final horizontal correction.

[0010] Furthermore, in order to obtain a more accurate vertical tolerance compensation value, step S4 uses the vertical tolerance compensation value with the best evaluation result as the vertical tolerance range for the second evaluation correction and repeats step S41 to obtain the tolerance for the final vertical correction.

[0011] Furthermore, in step S3, the evaluation method is detailed,

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

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

[0014] Calculating the first differential value sum, i.e., the first error sum, in the vertical edge region of the first rendered image and the actual image;

[0015] The corresponding horizontal tolerance compensation value obtained by setting the minimum of the first differential value sum as the evaluation goal is used as the tolerance for horizontal correction.

[0016] Acquiring the vertical edges of the first rendering image and the actual image is, in detail, to perform a horizontal derivative on the first rendering image or the actual image to set a horizontal derivative threshold, and to mark as a vertical edge an area where the horizontal derivative is greater than the horizontal derivative threshold and its vertical derivative approaches zero;

[0017] Furthermore, correcting the vertical edge area based on vertical edges is done in detail by using the pixel range set by installing adjacent vertical edges as the vertical edge area.

[0018] Furthermore, to obtain a more accurate horizontal tolerance compensation value, the minimum sum of the first differential values ​​is obtained between the horizontal tolerance compensation value and the sum of the first differential values ​​using a fitting method.

[0019] Furthermore, the evaluation method of step S4 is described in detail,

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

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

[0022] Calculating the second differential value sum, i.e., the second error sum, in the horizontal edge region of the second rendered image and the actual image;

[0023] The corresponding vertical tolerance compensation value obtained by setting the minimum of the second differential value sum as the evaluation goal is used as the tolerance for vertical correction.

[0024] Furthermore, acquiring horizontal edges of the second rendered image and the actual image is, in detail, to perform a vertical derivative on the second rendered image or the actual image to set a vertical derivative threshold, and to mark as horizontal edges an area where the vertical derivative is greater than the vertical derivative threshold and its horizontal derivative approaches zero;

[0025] Furthermore, correcting the horizontal edge area based on horizontal edges is done in detail by using the pixel range set by installing adjacent horizontal edges as the horizontal edge area.

[0026] Furthermore, to obtain a more accurate vertical tolerance compensation value, the minimum sum of the second differential value is obtained between the vertical tolerance compensation value and the sum of the second differential value using a fitting method.

[0027] Furthermore, the specific steps for rendering in steps S3 and S4 are,

[0028] A step of reducing and enlarging the size of a graph corresponding to a vector point based on a tolerance compensation value;

[0029] Next, a step of filling the sub-pixel edges of the corrected vector points;

[0030] It includes the step of obtaining a corresponding rendering image by correcting the filled graph through the selected model. Effects of the invention

[0031] The calibration method created by the invention using the above-mentioned technical method performs high-precision detection on the photomask within the allowable tolerance range, thereby preventing the mass occurrence of false defects; furthermore, since the invention is configured so that two-dimensional tolerances have independence by a calibration method divided into two directions, namely horizontal and vertical, errors in the calibration tolerance do not occur due to mutual influence. At the same time, by processing only the information of vertical and horizontal edges and overlooking noise generated by a large number of corners, the calibration tolerance and reliability are greatly improved; next, the invention improves the accuracy of calibration by reducing unnecessary noise generated by alignment through an error summation method that evaluates the quality of the tolerance compensation value. Brief explanation of the drawing

[0032] FIG. 1 illustrates vector points of a selected area indicated in a design document of a photomask; where the intersection points are the locations of each point, and the dotted lines are polygons connecting each point; FIG. 2a illustrates an actual image of a photomask corresponding to a selected area shown in FIG. 1; FIG. 2b illustrates a rendering image formed by rendering the selected region shown in FIG. 1 while passing through a low-pass filter; FIG. 3a is a difference drawing between an actual image and a rendered image with uncorrected tolerances; FIG. 3b is a difference drawing between an actual image and a rendering image with uncorrected tolerances; Figure 4 is a scatter plot of the results of evaluating the error by correcting for each tolerance in the present invention. Specific details for implementing the invention

[0033] Hereinafter, the present invention will be described in more detail based on specific embodiments and combined with drawings so that the content of the present invention can be understood more easily and clearly.

[0034] As illustrated in FIGS. 1 to 4, in a method for calibrating a photomask for detecting photomask defects, the steps of the method are:

[0035] S1: A step of selecting a selected region containing a graph in a photomask design document through EDA software and outputting vector points of said selected region, as illustrated in FIG. 1;

[0036] S2: As illustrated in FIG. 2a, a corresponding physical photomask is aligned using an AOI detection facility, and then, a real image corresponding to a selected area in the corresponding physical photomask is collected using a detection camera of the AOI detection facility; wherein the purpose of the alignment is to configure the system so as to map the platform coordinates of the AOI detection facility to the coordinates of the photomask design document to collect a real image corresponding to the selected area indicated in step S1;

[0037] S3: Set a horizontal tolerance range in the horizontal direction (i.e., X-direction) of the selected area and proceed to step S31; fill the vector points output in step S1 using different horizontal tolerance compensation values ​​within the horizontal tolerance range and render them into a first rendering image; align the first rendering image obtained by rendering the different horizontal tolerance compensation values ​​with the actual image, evaluate the influence of the different horizontal tolerance compensation values ​​on the degree of agreement between the first rendering image and the actual image using an evaluation method, and use the horizontal tolerance compensation value with the best evaluation result as the tolerance for horizontal correction; wherein, in detail, using different horizontal tolerance compensation values ​​within the horizontal tolerance range involves uniformly selecting a plurality of appropriate numeric points within the horizontal tolerance range and using these numeric points as different horizontal tolerance compensation values;

[0038] S4: Set a horizontal tolerance range in the vertical direction (i.e., Y direction) of the selected area and proceed with step S41; fill the vector points output in step S1 using different vertical tolerance compensation values ​​within the vertical tolerance range to render them into a second rendering image; align the second rendering image obtained by rendering the different vertical tolerance compensation values ​​with the actual image, evaluate the influence of the different vertical tolerance compensation values ​​on the degree of agreement between the second rendering image and the actual image using an evaluation method, and use the vertical tolerance compensation value with the best evaluation result as the tolerance for vertical correction; wherein, using different vertical tolerance compensation values ​​within the vertical tolerance range is, in detail, a step of uniformly selecting a plurality of appropriate numeric points within the vertical tolerance range and using these numeric points as different vertical tolerance compensation values;

[0039] In this embodiment, the process of steps S3 to S4 can be improved by using a multi-thread method.

[0040] In this embodiment, the matching method of steps S3 and S4 can improve matching accuracy by reaching the level of a subpixel by matching using phase.

[0041] Specifically, in order to obtain a more accurate horizontal tolerance compensation value, step S3 uses the horizontal tolerance compensation value with the best evaluation result as the horizontal tolerance range for the second evaluation correction, and repeats step S31 to obtain the tolerance for the final horizontal correction.

[0042] Specifically, in order to obtain a more accurate vertical tolerance compensation value, step S4 uses the vertical tolerance compensation value with the best evaluation result as the vertical tolerance range for the second evaluation correction, and repeats step S41 to obtain the tolerance for the final vertical correction.

[0043] In detail, the evaluation method of step S3 is in detail,

[0044] Acquiring the vertical edges of the first rendered image and the actual image;

[0045] Correcting the vertical edge area based on the vertical edge;

[0046] Calculating the first differential value sum, i.e., the first error sum, in the vertical edge region of the first rendered image and the actual image;

[0047] The corresponding horizontal tolerance compensation value obtained by setting the minimum of the first differential value sum as the evaluation goal is used as the tolerance for horizontal correction.

[0048] Acquiring the vertical edges of the first rendering image and the actual image is, in detail, to perform a horizontal derivative on the first rendering image or the actual image to set a horizontal derivative threshold, and to mark as a vertical edge an area where the horizontal derivative is greater than the horizontal derivative threshold and its vertical derivative approaches 0 (set based on light uniformity, and typically only needs to be less than 5);

[0049] Specifically, correcting the vertical edge area based on the vertical edge is to use the area within the pixel range set by installing adjacent vertical edges as the vertical edge area; in this embodiment, the set pixel range may be a range of three pixels, and of course, may be a different quantity.

[0050] Specifically, to obtain a more accurate horizontal tolerance compensation value, the minimum sum of the first differential values ​​is obtained between the horizontal tolerance compensation value and the sum of the first differential values ​​using a fitting method.

[0051] In detail, the evaluation method of step S4 is in detail,

[0052] Acquiring the horizontal edges of the second rendered image and the actual image;

[0053] Correcting the horizontal edge area based on the horizontal edge;

[0054] Calculating the second differential value sum, i.e., the second error sum, in the horizontal edge region of the second rendered image and the actual image;

[0055] The corresponding vertical tolerance compensation value obtained by setting the minimum of the second differential value sum as the evaluation goal is used as the tolerance for vertical correction.

[0056] Furthermore, acquiring horizontal edges of the second rendered image and the actual image is, in detail, to perform a vertical derivative on the second rendered image or the actual image to set a vertical derivative threshold, and to mark as horizontal edges an area where the vertical derivative is greater than the vertical derivative threshold and its horizontal derivative approaches 0 (set based on light uniformity, and typically only needs to be less than 5);

[0057] Specifically, correcting the horizontal edge area based on the horizontal edge involves installing adjacent horizontal edges and using the area within the set pixel range as the horizontal edge area. In this embodiment, the set pixel range may be a range of three pixels, and of course, it may be a different quantity.

[0058] Specifically, to obtain a more accurate vertical tolerance compensation value, the minimum sum of the second differential value is obtained between the vertical tolerance compensation value and the sum of the second differential value using a fitting method.

[0059] Specifically, the rendering in steps S3 and S4 is rendered as a bitmap based on the point set of the photomask design document, and during the rendering process, it is necessary to use corresponding models such as a low-pass filter, a scalar optical simulation model, and a vector optical simulation model to approximate the actual photomask collection drawing, and the rendering result is as illustrated in FIG. 2b. The specific selection method can be selected based on an acceptable error. The specific steps for dividing the tolerance range using significantly large intervals based on the tolerance setting range, performing graph rendering once for each tolerance compensation value, and performing rendering in steps S3 and S4 are as follows:

[0060] A step of reducing and enlarging the size of a graph corresponding to a vector point based on a tolerance compensation value;

[0061] Next, a step of filling the sub-pixel edges of the corrected vector points;

[0062] It includes the step of obtaining a corresponding rendering image by correcting the filled graph through the selected model.

[0063] In this embodiment, the differential method of steps S3 and S4—that is, the method of directly calculating the sum through the differential value—can very effectively neutralize errors caused by inaccurate alignment. If the alignment is inaccurate, the signs of the differential values ​​on the symmetrical edges of the graph are often opposite; when the tolerance value is accurate, the sum of these differential values ​​is approximately zero, and it takes on a significantly large value only when there is a tolerance error. Furthermore, as the deviation of the tolerance increases, the value of this differential sum also increases, which is a reason why the differential method can be used to determine the quality of the horizontal or vertical tolerance compensation value. At the same time, the fact that the differential calculated only the error at the horizontal or vertical edge points and did not calculate the error at the corner points, and specifically overlooked the error at the corner points, also acts as a cause. Since cases where control is unavailable at the corner points occur more frequently due to the process level, more noise information is often provided at the corner points. Similarly, since the differential is calculated by dividing into horizontal and vertical, the two-dimensional tolerances of the horizontal and vertical can be separated in a very novel way, thereby providing higher precision. As shown in FIGS. 3a and 3b, when tolerance compensation is not performed, there is a significant deviation between the rendering drawing and the actual image; however, after the tolerance compensation value is corrected by selecting the minimum error from the photomask design, the consistency between the rendering drawing and the actual image increases, and only a very small deviation remains.

[0064] As described above, specific embodiments have explained in more detail the technical problem, technical solution, and beneficial effects that the present invention aims to solve. The above description is merely a specific embodiment of the present invention and is not intended to limit the present invention. It should be understood that all modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention are all included within the scope of protection of the present invention.

Claims

Claim 1 A method for calibrating a photomask for detecting defects in a photomask comprises the steps of the method: S1: selecting a selected area containing a graph in a photomask design document and outputting vector points of said selected area; S2: collecting actual images corresponding to the selected area in a corresponding actual photomask using an AOI detection facility; S3: setting a horizontal tolerance range in the horizontal direction of the selected area and proceeding to step S31; filling the vector points output in step S1 with different horizontal tolerance compensation values ​​within the horizontal tolerance range to render them into a first rendering image; aligning the first rendering image obtained by rendering with different horizontal tolerance compensation values ​​with the actual image, evaluating the influence of different horizontal tolerance compensation values ​​on the degree of agreement between the first rendering image and the actual image using an evaluation method, and using the horizontal tolerance compensation value with the best evaluation result as the tolerance for horizontal correction; S4: setting a horizontal tolerance range in the vertical direction of the selected area and proceeding to step S41; A method for correcting a photomask for detecting photomask defects, characterized by comprising the step of: filling vector points output in step S1 with different vertical tolerance compensation values ​​within a vertical tolerance range to render them into a second rendering image; aligning the second rendering image obtained by rendering with different vertical tolerance compensation values ​​with the actual image, evaluating the effect of different vertical tolerance compensation values ​​on the degree of agreement between the second rendering image and the actual image using an evaluation method, and using the vertical tolerance compensation value with the best evaluation result as the tolerance for vertical correction. Claim 2 A method according to claim 1, wherein step S3 is characterized by repeatedly performing step S31 using the horizontal tolerance compensation value with the best evaluation result as the horizontal tolerance range for the second evaluation correction to obtain the tolerance for the final horizontal correction. Claim 3 A method according to claim 1, wherein step S4 is characterized by repeatedly performing step S41 using the vertical tolerance compensation value with the best evaluation result as the vertical tolerance range for the second evaluation correction to obtain the tolerance for the final vertical correction. Claim 4 In claim 1, the evaluation method of step S3 is characterized in detail by acquiring a vertical edge between a first rendering image and an actual image; correcting a vertical edge region based on the vertical edge; calculating a first differential value sum, i.e., a first error sum, in the vertical edge region between the first rendering image and the actual image; and using the corresponding horizontal tolerance compensation value obtained by setting the minimum of the first differential value sum as an evaluation target as the tolerance for horizontal correction. Claim 5 A method according to claim 4, wherein obtaining a vertical edge of a first rendering image and an actual image is characterized in detail by performing a horizontal derivative on the first rendering image or the actual image to set a horizontal derivative threshold, and marking an area where the horizontal derivative is greater than the horizontal derivative threshold and the vertical derivative is close to 0 as a vertical edge; and wherein correcting a vertical edge area based on a vertical edge is characterized in detail by using a pixel range within which adjacent vertical edges are installed as a vertical edge area. Claim 6 A method according to claim 4, characterized in that the sum of the first differential values ​​is obtained using a fitting method between the horizontal tolerance compensation value and the sum of the first differential values. Claim 7 In claim 1, the evaluation method of step S4 is characterized in detail by acquiring horizontal edges of a second rendering image and an actual image; correcting a horizontal edge area based on the horizontal edges; calculating a second differential value sum, i.e., a second error sum, in the horizontal edge area of ​​the second rendering image and the actual image; and using the corresponding vertical tolerance compensation value obtained by setting the minimum of the second differential value sum as an evaluation goal as the tolerance for vertical correction. Claim 8 A method according to claim 6, wherein obtaining horizontal edges of a second rendering image and an actual image is characterized in detail by performing a vertical derivative on the second rendering image or the actual image to set a vertical derivative threshold, and marking an area where the vertical derivative is greater than the vertical derivative threshold and the horizontal derivative is close to 0 as a horizontal edge; and wherein correcting a horizontal edge area based on the horizontal edge is characterized in detail by using a pixel range within which adjacent horizontal edges are installed as a horizontal edge area. Claim 9 A method according to claim 7, characterized by obtaining the minimum sum of the second differential values ​​between the vertical tolerance compensation value and the sum of the second differential values ​​using a fitting method. Claim 10 A method according to claim 1, wherein the specific steps for performing rendering in steps S3 and S4 include: a step of reducing and enlarging the size of a graph corresponding to a vector point based on a tolerance compensation value; then, a step of filling the sub-pixel edges of the corrected vector point; and a step of correcting the filled graph through a selected model to obtain a corresponding rendering image.

Images