Pattern recognition method and device and pattern feature size measurement method

Abstract

The invention provides a pattern recognition method and device and a pattern feature size measurement method. The pattern recognition method comprises the following steps: providing a distinguishing feature relationship of CDSEM recognition information of a convex pattern and a concave pattern in a reference structure; cDSEM identification information of the semiconductor structure including a morphology distribution diagram and a characteristic curve diagram is obtained; identifying and cutting the identification signal edge in the semiconductor structure characteristic curve graph, and judging whether the identification signal edge corresponds to the convex pattern or the concave pattern according to the distinguishing characteristic relation of the reference structure. According to the method, the identification characteristic relation of the identification signal edge is calibrated and identified, so that the identification degree of the convex pattern and the concave pattern is improved, the reverse measurement risk is reduced, and the CDSEM pattern identification precision and measurement accuracy are improved; meanwhile, the distinguishing feature relation is obtained through the lowest point height of the reference signal edge, and the pattern recognition efficiency is improved; in addition, the size relation between the slope absolute value and the area integral value of the reference signal edge is used as a preset detection feature, the fitting failure risk is further reduced, and the pattern recognition success rate is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of semiconductor manufacturing equipment, and in particular relates to a pattern recognition method and device and a pattern feature size measurement method. Background Art

[0002] In the semiconductor manufacturing industry, CDSEM (Critical Dimension Scanning Electron Microscope) is a key measuring device and is widely used for accurate measurement of CD (Critical Dimension) of line (line width) and space (line spacing) after the photolithography process. In the photolithography process, the line pattern is the part blocked by the photolithography, forming an unetched raised structure, and the width of its side walls is called line CD; while the space pattern is the part that is not blocked by the photolithography and is thus developed and etched, forming an etched recessed area, and the width of its internal side walls is space CD; the accurate measurement of line CD and space CD from the identification information obtained by CDSEM is crucial to the evaluation conclusion of whether the semiconductor graphics are qualified.

[0003] However, in the prior art, when the measured line CD and space CD are identified from the identification information of CDSEM, the relative displacement of the actual measured pattern and the reference image used for fitting may differ within the range of FOV (Field of view), resulting in inaccurate position alignment of the two images during fitting comparison. At the same time, the difference between the image quality parameters such as the structural contour and image contrast of the actual measured pattern and the image quality of the reference image may also affect the image recognition, making the boundary distinction between the line pattern and the space pattern blurred. These differences lead to low accuracy in distinguishing and identifying the line pattern and the space pattern, and it is easy to cause reverse vector measurement, such as Figure 1 When measuring a space pattern as shown in Figure 2 When measuring the line pattern as shown, the space pattern is measured as the line pattern, which confuses the measurement position identification of the line CD and the space CD, thereby affecting the accuracy of the evaluation conclusion on whether the semiconductor pattern is qualified.

[0004] Moreover, when the image quality and other parameters of the actual measurement pattern differ greatly from those of the reference image, the existing line pattern and space pattern recognition technologies are more likely to cause reverse measurement errors of line CD and space CD or fail to perform measurement due to large differences.

[0005] Therefore, there is an urgent need for a structure or method that can improve the recognition and distinction accuracy and measurement accuracy between line patterns and space patterns.

[0006] It should be noted that the above introduction to the technical background is only for the convenience of providing a clear and complete description of the technical solutions of the present application and for the convenience of understanding by those skilled in the art. It cannot be considered that the above technical solutions are well known to those skilled in the art simply because these solutions are described in the background technology part of the present application. Summary of the invention

[0007] In view of the above shortcomings of the prior art, an object of the present invention is to provide a pattern recognition method and device and a pattern feature size measurement method to solve the problems of poor recognition and distinction accuracy and poor measurement accuracy between line pattern and space pattern in the prior art.

[0008] To achieve the above object, the present invention provides a graphic recognition method, the graphic recognition method comprising:

[0009] Providing a distinguishing feature relationship of CDSEM identification information of a convex figure and a concave figure in a target figure of a reference structure, wherein the distinguishing feature relationship is used to distinguish the convex figure from the concave figure;

[0010] The CDSEM identification information of the target pattern corresponding to the semiconductor structure is obtained by measuring, wherein the CDSEM identification information includes a morphology distribution diagram and a characteristic curve diagram corresponding to the morphology distribution diagram, wherein the characteristic curve diagram includes a signal segment composed of m signal peaks, where m is an integer greater than or equal to 1;

[0011] Each straight line in the characteristic curve graph that passes through the peak point of the signal peak and is perpendicular to the horizontal axis of the characteristic curve graph is a first cutting line, a straight line that is spaced from the first cutting line by a first preset distance in the first direction along the horizontal axis of the characteristic curve graph and is perpendicular to the horizontal axis of the characteristic curve graph is a second cutting line corresponding to the first cutting line, a straight line that is spaced from the first cutting line by a second preset distance in the second direction along the horizontal axis of the characteristic curve graph and is perpendicular to the horizontal axis of the characteristic curve graph is a third cutting line corresponding to the first cutting line, and the first direction and the second direction are two opposite directions along the horizontal axis of the characteristic curve graph; identifying the first cutting line and the second cutting line in the characteristic curve graph of the semiconductor structure, and cutting the signal segment between the corresponding first cutting line and the second cutting line in the characteristic curve graph of the semiconductor structure is an identification signal edge; identifying the first cutting line and the third cutting line in the characteristic curve graph of the semiconductor structure, and cutting the signal segment between the corresponding first cutting line and the third cutting line in the characteristic curve graph of the semiconductor structure is an identification signal edge; obtaining n identification signal edges from the characteristic curve graph of the semiconductor structure, where n is an integer greater than or equal to 2;

[0012] According to the characteristic curve diagram measurement of the corresponding target pattern in the semiconductor structure, obtaining an identification detection feature corresponding to the target pattern in the semiconductor structure and a preset detection feature;

[0013] According to the identification detection characteristics of the corresponding target pattern in the semiconductor structure, an identification feature relationship corresponding to the distinguishing feature relationship obtained between the target pattern in the semiconductor structure and the reference structure is obtained, and it is determined whether the position corresponding to each identification signal edge in the characteristic curve diagram of the semiconductor structure is a convex pattern or a concave pattern.

[0014] Optionally, the method for obtaining the distinguishing characteristic relationship of the CDSEM identification information of the convex figure and the concave figure in the target figure of the reference structure includes:

[0015] Providing a reference structure, wherein the reference structure includes a target pattern, and obtaining position distribution information of a convex pattern and a concave pattern of the target pattern in the reference structure;

[0016] measuring and obtaining CDSEM identification information of a target pattern of the reference structure;

[0017] The signal segment between the first and second cutting lines corresponding to the characteristic curve diagram of the CDSEM identification information of the reference structure is cut as a reference signal edge, and the signal segment between the first and third cutting lines corresponding to the characteristic curve diagram of the CDSEM identification information of the reference structure is cut as a reference signal edge, and a reference signal edges are obtained from the characteristic curve diagram of the reference structure, where a is an integer greater than or equal to 2; the reference signal edges corresponding to the convex figure and the concave figure are respectively calibrated according to the position distribution information;

[0018] The distinguishing characteristic relationship between each reference signal edge corresponding to the convex figure and the concave figure is obtained according to the preset detection characteristics.

[0019] Optionally, the preset detection feature is the height of the lowest point of the reference signal edge; and the identification detection feature is the height of the lowest point of the identification signal edge.

[0020] Optionally, the distinguishing characteristic relationship is the size relationship between the height difference of the lowest points of two reference signal edges and a preset height difference threshold; the identification characteristic relationship is the size relationship between the height difference of the lowest points of two identification signal edges to be identified and a preset height difference threshold.

[0021] Optionally, the preset detection feature is the absolute value of the slope of the reference signal edge; and the identification detection feature is the absolute value of the slope of the identification signal edge.

[0022] Optionally, the distinguishing characteristic relationship is the size relationship between the difference in absolute values ​​of the slopes of two reference signal edges and a preset absolute value difference threshold; the identification characteristic relationship is the size relationship between the difference in absolute values ​​of the slopes of two identification signal edges to be identified and a preset absolute value difference threshold.

[0023] Optionally, the preset detection feature is the area integral value of the closed contour formed by the reference signal edge, the first dividing line corresponding to the reference signal edge, and the integral base of the lowest point corresponding to the reference signal edge parallel to the horizontal axis of the characteristic curve graph; the identification detection feature is the area integral value of the closed contour formed by the identification signal edge, the first dividing line corresponding to the identification signal edge, and the integral base of the lowest point corresponding to the identification signal edge parallel to the horizontal axis of the characteristic curve graph.

[0024] Optionally, the distinguishing characteristic relationship is the size relationship between the difference in area integral values ​​of two reference signal edges and a preset integral difference threshold; the identification characteristic relationship is the size relationship between the difference in area integral values ​​of two identification signal edges to be identified and a preset integral difference threshold.

[0025] The present invention also provides a graphic recognition device, wherein any of the above-mentioned graphic recognition methods is performed using the graphic recognition device, and the graphic recognition device comprises: a recognition calculation unit, a CDSEM recognition unit, a dividing line recognition unit, a signal segment cutting unit and a recognition judgment unit;

[0026] The CDSEM recognition unit is used to measure and obtain CDSEM recognition information of a target pattern corresponding to the semiconductor structure, wherein the CDSEM recognition information includes a morphology distribution map and a characteristic curve map corresponding to the morphology distribution map; the segmentation line recognition unit is electrically connected to the CDSEM recognition unit, and is used to identify the positions of the first segmentation line, the second segmentation line, and the third segmentation line on the characteristic curve map in the CDSEM recognition information obtained by the CDSEM recognition unit;

[0027] The signal segment cutting unit is electrically connected to the segmentation line identification unit, and is used to cut the signal segment in the characteristic curve graph using the positions of the first segmentation line, the second segmentation line and the third segmentation line obtained by the segmentation line identification unit to obtain n identification signal edges;

[0028] The recognition judgment unit is electrically connected to the recognition calculation unit and the signal segment cutting unit respectively; the recognition calculation unit calculates the respective recognition detection features of two recognition signal edges to be identified selected from the n recognition signal edges, and compares and obtains the recognition feature relationship between the two recognition signal edges; the recognition judgment unit determines whether the positions in the morphology distribution map of the CDSEM recognition information corresponding to the two recognition signal edges to be identified selected from the n recognition signal edges obtained from the signal segment cutting unit are convex figures or concave figures through the recognition feature relationship between the two recognition signal edges obtained by the recognition calculation unit.

[0029] The present invention also provides a method for measuring the size of a graphic feature, the method comprising:

[0030] Providing a semiconductor structure including a target pattern, wherein the target pattern includes a protruding pattern and a recessed pattern;

[0031] Using any one of the above-mentioned pattern recognition methods or / and the above-mentioned pattern recognition device to recognize the positions of the protruding patterns and the concave patterns in the target pattern of the semiconductor structure;

[0032] The position of the convex pattern identified in the target pattern of the semiconductor structure is measured to obtain the characteristic size of the convex pattern, and the position of the concave pattern identified in the target pattern of the semiconductor structure is measured to obtain the characteristic size of the concave pattern.

[0033] As described above, the pattern recognition method and device and the pattern feature size measurement method of the present invention have the following beneficial effects:

[0034] The present invention obtains the corresponding CDSEM identification information of the corresponding figure according to the corresponding position of the calibrated corresponding figure in the CDSEM identification information of the target figure of the reference structure, thereby obtaining the distinguishing characteristic relationship between the reference signal edges corresponding to the convex figure and the concave figure. By using the distinguishing characteristic relationship obtained in the reference structure as a calibration reference, the identification characteristic relationship obtained by the CDSEM identification information of the semiconductor structure to be measured later can be used to distinguish the specific positions of the convex figure and the concave figure, thereby achieving high-accuracy resolution of the convex figure and the concave figure of the semiconductor structure and high-precision measurement of the characteristic size;

[0035] The present invention uses the height relationship of the lowest points of two reference signal edges as a distinguishing characteristic relationship, which makes identification easy and convenient, improves the recognition efficiency of convex and concave figures, and further improves the figure recognition accuracy of CDSEM;

[0036] The present invention divides the characteristic curve graph in the CDSEM identification information of the target pattern of the semiconductor structure obtained by measurement into identification signal edges, and identifies whether the identification signal edge obtained by the division corresponds to a convex pattern or a concave pattern by cooperating with the distinguishing characteristic relationship of the CDSEM identification information of the reference structure calibrated in advance. By identifying and distinguishing the characteristics of a single signal edge, the recognition of the convex pattern and the concave pattern under the influence of various difference factors between the reference structure and the actual process and imaging process is improved, the risk of reverse vector measurement between the convex pattern and the concave pattern is reduced, and the pattern recognition accuracy and key dimension measurement accuracy of the CDSEM are improved;

[0037] The present invention sets the magnitude relationship of the absolute values ​​of the slopes of the two reference signal edges as a distinguishing characteristic relationship, which can be used when the lowest points of the two reference signal edges are close in height and cannot be distinguished. As a replaceable distinguishing characteristic relationship, the risk of failure in fitting the convex figure and the concave figure through the distinguishing characteristic relationship of the reference structure from the identification characteristic relationship of the CDSEM identification information of the semiconductor structure is reduced, and the pattern recognition of the CDSEM is improved;

[0038] The present invention sets the size relationship of the area integral values ​​of two reference signal edges as a distinguishing characteristic relationship, which can be used when the lowest point heights or slopes of the two reference signal edges are similar and cannot be distinguished. As a replaceable distinguishing characteristic relationship, it further reduces the risk of failure in fitting the protruding figure and the concave figure through the distinguishing characteristic relationship of the reference structure from the identification characteristic relationship of the CDSEM identification information of the semiconductor structure, and improves the success rate of CDSEM graphic recognition. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 It is a schematic diagram showing the reverse direction measurement that occurs when measuring a space pattern in the prior art.

[0040] Figure 2 It is a schematic diagram showing the reverse direction measurement that occurs when measuring a line pattern in the prior art.

[0041] Figure 3 Shown is a step diagram of the graphic recognition method in Example 1 of the present invention.

[0042] Figure 4 The figure shows a characteristic curve diagram of CDSEM recognition information of a reference structure target figure obtained in an example of step A1 of the figure recognition method of embodiment 1 of the present invention.

[0043] Figure 5 It shows a characteristic curve diagram of CDSEM recognition information of the semiconductor structure target pattern obtained in step A2 of the pattern recognition method of embodiment 1 of the present invention.

[0044] Figure 6 It shows a characteristic curve diagram of CDSEM identification information of the identification signal edge in the semiconductor structure target pattern obtained in step A3 of the pattern recognition method of embodiment 1 of the present invention.

[0045] Figure 7 Shown is a schematic diagram of a measured space pattern obtained by the pattern recognition method according to Example 1 of the present invention.

[0046] Figure 8 Shown is a schematic diagram of a measured line pattern obtained by the pattern recognition method according to Example 1 of the present invention.

[0047] Fig. 9 Shown is a schematic diagram of the distinguishing characteristic relationship between concave graphics and convex graphics in a characteristic curve diagram of CDSEM recognition information of a reference structure target graphic in an example of a graphic recognition method in embodiment 1 of the present invention.

[0048] Fig.10 Shown is a schematic diagram of the distinguishing characteristic relationship between concave graphics and convex graphics in a characteristic curve diagram of CDSEM recognition information of a reference structure target graphic in an example of a graphic recognition method in embodiment 2 of the present invention.

[0049] Fig.11 Shown is a schematic diagram of the recognition feature relationship of the recognition signal edge in the feature curve diagram of the CDSEM recognition information of the semiconductor structure target pattern in an example of the pattern recognition method of embodiment 2 of the present invention.

[0050] Fig.12Shown is a schematic diagram of the distinguishing characteristic relationship between concave graphics and convex graphics in a characteristic curve diagram of CDSEM recognition information of a reference structure target graphic in an example of a graphic recognition method in embodiment 3 of the present invention.

[0051] Fig.13 Shown is a schematic diagram of the recognition feature relationship of the recognition signal edge in the feature curve diagram of the CDSEM recognition information of the semiconductor structure target pattern in an example of the pattern recognition method of embodiment 3 of the present invention.

[0052] Fig.14 Shown is a schematic diagram of the structure of the graphic recognition device in Example 4 of the present invention.

[0053] Fig.15 Shown is a structural schematic diagram of a graphic recognition device in an example of embodiment 4 of the present invention.

[0054] Fig.16 It shows a step diagram of a method for measuring a graphic feature size in Embodiment 5 of the present invention.

[0055] Description of Reference Numerals

[0056] 01 / 02, first reference signal edge; 03 / 04, second reference signal edge;

[0057] 11. Raised figure; 12. Concave figure; 13. Measurement area; 14. CDSEM identification information; 15. Characteristic curve graph; 16. Horizontal axis;

[0058] 21. First dividing line; 22. Second dividing line; 23. Third dividing line; 24. Reference signal edge; 25. Signal peak; 26. Signal segment; 271. First direction; 272. Second direction; 28. Identify signal edge; 291. First preset distance; 292. Second preset distance;

[0059] 31, lowest point; 311, first lowest point; 312, second lowest point; 313, third lowest point; 314, fourth lowest point; 315, fifth lowest point; 316, sixth lowest point; 317, seventh lowest point; 318, eighth lowest point; 341 / 342, first slope point; 351 / 352, second slope point; 361 / 362, third slope point; 371 / 372, fourth slope point; 36, first base; 37, second base; 38, third base; 39, fourth base;

[0060] 41 / 42, first identification signal edge; 43 / 44, second identification signal edge;. DETAILED DESCRIPTION

[0061] The following describes the embodiments of the present invention through specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention.

[0062] For example, when describing the embodiments of the present invention in detail, for the sake of convenience, the schematic diagrams showing the device structure will not be partially enlarged according to the general scale, and the schematic diagrams are only examples, which should not limit the scope of protection of the present invention. In addition, in actual production, the three-dimensional dimensions of length, width and depth should be included.

[0063] For convenience of description, spatially relative terms such as "under", "below", "below", "below", "above", "on", etc. may be used herein to describe the relationship of one element or feature shown in the drawings to other elements or features. It will be understood that these spatially relative terms are intended to encompass other orientations of the device in use or operation in addition to the orientation depicted in the drawings.

[0064] In the context of the present application, a structure in which a first feature is described as being "above" a second feature may include embodiments in which the first and second features are in direct contact, and may also include embodiments in which additional features are formed between the first and second features, such that the first and second features may not be in direct contact.

[0065] It should be noted that the illustrations provided in this embodiment are only used to illustrate the basic concept of the present invention in a schematic manner, and therefore the illustrations only show components related to the present invention rather than being drawn according to the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component may be changed arbitrarily, and the component layout may also be more complicated.

[0066] In the prior art, in order to improve the accuracy of distinguishing and identifying line patterns (convex patterns 11) and space patterns (concave patterns 12), the industry has developed a variety of recognition modes, mainly including the following four:

[0067] 1) Directly compare the overall image of the actual measurement pattern with the reference image, and identify by emphasizing features such as edge, contrast, uniqueness, image shadow or image model; however, when the actual measurement pattern changes due to various factors in the process, the image matching degree will decrease, and recognition errors are prone to occur;

[0068] 2) Measurement is performed by moving the stage carrying the semiconductor structure to the preset coordinate positions of the line pattern and space pattern determined in advance in the recipe. However, this method requires extremely high stage movement accuracy and is limited by the recipe setup, such as the distance requirement between the actual measurement pattern and the addressing pattern. This can easily lead to reduced stage movement accuracy, resulting in recognition errors.

[0069] 3) Based on the position of the actual measured pattern within the FOV (Field ofview), CD size (critical dimension size) or margin area and the fit between it and the reference image, the judgment is made. However, this method also has strict requirements on the stage movement accuracy and model offset consistency, and has a low tolerance for CD size changes, which makes it easy to make recognition errors.

[0070] 4) Comprehensively consider the signal peak position, signal peak weight, etc. of the line pattern and space pattern in the actual measurement figure and the fit degree of the signal peak in the reference image (such as tilt degree, height difference, etc.) for identification; however, this method is only applicable to the measurement of one-dimensional figures, and has a low tolerance for changes in the actual measurement figures, so it is also prone to recognition errors.

[0071] Therefore, the measurement and recognition method in the prior art is still prone to misidentification of the line pattern (convex pattern 11) and the space pattern (concave pattern 12), so that reverse measurement is prone to occur, such as Figure 1 When measuring a space pattern as shown in FIG. 1 , the line pattern is used as the measurement area 13 of the space pattern for measurement, or as shown in FIG. Figure 2 As shown, when measuring the line pattern, the space pattern is measured as the measurement area 13 of the line pattern, resulting in confusion in the identification of the measurement positions of the line CD and the space CD.

[0072] Embodiment 1:

[0073] like Figure 3 As shown, this embodiment provides a graphic recognition method, and the graphic recognition method includes:

[0074] Step A1: providing a distinguishing characteristic relationship of CDSEM identification information 14 of a convex pattern 11 (line pattern) and a concave pattern 12 (space pattern) in a target pattern of a reference structure, wherein the distinguishing characteristic relationship is used to distinguish the convex pattern 11 from the concave pattern 12;

[0075] Step A2: Measure and obtain CDSEM identification information 14 of the target pattern corresponding to the semiconductor structure, wherein the CDSEM identification information 14 includes a morphology distribution map and a characteristic curve corresponding to the morphology distribution map. Fig.15 , characteristic curve Fig.15 A signal segment 26 including m signal peaks 25, where m is an integer greater than or equal to 1;

[0076] Step A3: Characteristic Curve Fig.15 Each of the peak points of the signal peak 25 and perpendicular to the characteristic curve Fig.15 The straight line of the horizontal axis 16 is the first dividing line 21, along the characteristic curve Fig.15 The first direction 271 of the horizontal axis 16 is spaced a first preset distance 291 from the first cutting line 21 and is perpendicular to the characteristic curve Fig.15 The straight line on the horizontal axis 16 is the second cutting line 22 corresponding to the first cutting line 21, along the characteristic curve Fig.15 The second direction 272 of the horizontal axis 16 is spaced a second preset distance 292 from the first cutting line 21 and is perpendicular to the characteristic curve Fig.15 The straight line of the horizontal axis 16 is the third cutting line 23 corresponding to the first cutting line 21, and the first direction 271 and the second direction 272 are along the characteristic curve Fig.15 Two opposite directions of the horizontal axis 16; Identify the characteristic curve of the semiconductor structure Fig.15 The first cutting line 21 and the second cutting line 22 in the semiconductor structure are cut into the characteristic curve Fig.15 The signal segment 26 between the first cutting line 21 and the second cutting line 22 is an identification signal edge 28; the characteristic curve of the semiconductor structure is identified. Fig.15 The first cutting line 21 and the third cutting line 23 in the semiconductor structure are cut into the characteristic curve Fig.15 The signal segment 26 between the first cutting line 21 and the third cutting line 23 is an identification signal edge 28; from the characteristic curve of the semiconductor structure Fig.15 Obtain n identification signal edges 28, where n is an integer greater than or equal to 2;

[0077] Step A4: According to the characteristic curve of the corresponding target pattern in the semiconductor structure Fig.15 Measuring to obtain an identification detection feature corresponding to a target pattern in the semiconductor structure and a preset detection feature;

[0078] Step A5: According to the identification detection features of the target pattern corresponding to the semiconductor structure, obtain the identification feature relationship corresponding to the distinguishing feature relationship obtained by the target pattern in the semiconductor structure and the reference structure, and determine the characteristic curve of the semiconductor structure. Fig.15 The position corresponding to each identification signal edge 28 in the image is a convex pattern 11 or a concave pattern 12.

[0079] The graphic recognition method of the present invention will be described in detail below in conjunction with the accompanying drawings. It should be noted that the above sequence does not strictly represent the sequence of the graphic recognition method protected by the present invention, and those skilled in the art may change it according to the actual steps of the graphic recognition method.

[0080] First, step A1 is performed to provide a distinguishing feature relationship of CDSEM identification information 14 of the convex pattern 11 and the concave pattern 12 in the target pattern of the reference structure, wherein the distinguishing feature relationship is used to distinguish the convex pattern 11 from the concave pattern 12 .

[0081] Specifically, the “reference structure” refers to a structure used for calibration, and a target pattern is prepared on the reference structure using the same process as the semiconductor structure to be measured.

[0082] In one embodiment, a method for obtaining the distinguishing characteristic relationship of the CDSEM identification information 14 of the convex pattern 11 and the concave pattern 12 in the target pattern of the reference structure includes:

[0083] Providing a reference structure, wherein the reference structure includes a target pattern, and obtaining position distribution information of a convex pattern 11 and a concave pattern 12 of the target pattern in the reference structure;

[0084] Measure and obtain CDSEM identification information 14 of the target pattern of the reference structure;

[0085] like Figure 4 As shown, the characteristic curve of the CDSEM identification information 14 of the reference structure is cut Fig.15 The signal segment 26 between the first cutting line 21 and the second cutting line 22 is a reference signal edge 24, which cuts the characteristic curve of the CDSEM identification information 14 of the reference structure. Fig.15 The signal segment 26 between the first dividing line 21 and the third dividing line 23 is a reference signal edge 24. Fig.15 A reference signal edges 24 are obtained, where a is an integer greater than or equal to 2; and then the reference signal edges 24 corresponding to the convex pattern 11 and the concave pattern 12 are calibrated respectively according to the position distribution information;

[0086] The distinguishing characteristic relationship between the respective reference signal edges 24 corresponding to the convex pattern 11 and the concave pattern 12 is obtained according to the preset detection characteristics.

[0087] Since the positions of the convex figure 11 and the concave figure 12 in the target figure of the reference structure can be artificially calibrated, the present invention obtains the corresponding CDSEM identification information 14 of the corresponding figure according to the corresponding position of the calibrated corresponding figure in the CDSEM identification information 14 of the target figure of the reference structure, calculates the preset detection features of the reference signal edges 24 corresponding to the convex figure 11 and the concave figure 12, thereby obtaining the distinguishing feature relationship between the reference signal edges 24 corresponding to the convex figure 11 and the concave figure 12. By using the distinguishing feature relationship obtained in the reference structure as a calibration reference, the distinguishing feature relationship obtained by the calibration reference can be used to distinguish the specific positions of the convex figure 11 and the concave figure 12 for the identification feature relationship of the CDSEM identification information 14 of the semiconductor structure to be measured later, thereby achieving high-accuracy resolution of the convex figure 11 and the concave figure 12 of the semiconductor structure and high-precision measurement of the feature size.

[0088] Specifically, from the characteristic curve of the reference structure Fig.15 After obtaining a reference signals 24, each reference signal 24 is numbered and named to obtain a number name corresponding to each reference signal 24, and the reference signal 24 corresponding to the number name of each reference signal 24 is calibrated as a convex figure 11 or a concave figure 12 according to the number name of each reference signal 24 and the position distribution information corresponding to the corresponding reference signal 24.

[0089] In this embodiment, the preset detection feature is the height of the lowest point 31 of the reference signal edge 24 ; and the identification detection feature is the height of the lowest point 31 of the identification signal edge 28 .

[0090] In this embodiment, the distinguishing characteristic relationship is the size relationship between the height difference between the lowest points 31 of the two reference signal edges 24 and the preset height difference threshold; the identification characteristic relationship is the size relationship between the height difference between the lowest points 31 of the two identification signal edges 28 to be identified and the preset height difference threshold.

[0091] The present invention uses the size relationship between the height difference of the lowest points 31 of the two reference signal edges 24 and a preset height difference threshold as a distinguishing feature relationship, which makes identification easy and detection convenient, improves the recognition efficiency of the raised figure 11 and the recessed figure 12, and further improves the graphic recognition accuracy of CDSEM.

[0092] In one embodiment, when using the preset detection features and distinguishing feature relationships in this embodiment, the first preset distance 291 is equal to the second preset distance 292 to ensure that the value of the lowest point 31 is correct; preferably, in this case, the first preset distance 291 and the second preset distance 292 should ensure that all reference signal edges 24 and identification signal edges 28 can include the actual lowest points 31 at both ends of their corresponding signal peaks.

[0093] Specifically, the "preset height difference threshold" means that when the height difference between the lowest points 31 of the two reference signal edges 24 is greater than or greater than or equal to the preset height difference threshold, it is determined that the two reference signal edges 24 can be distinguished as corresponding to the convex figure 11 and the concave figure 12 respectively; otherwise, the two reference signal edges 24 are considered to correspond to the same figure (both are convex figures 11 or both are concave figures 12) and the recognition fails. The specific value of the preset height difference threshold can be set according to the accuracy and size requirements for distinguishing the convex figure 11 and the concave figure 12 and the difference between the heights of the lowest points of the reference signal edges 24 corresponding to the same figures.

[0094] In one embodiment, the preset height difference threshold is equal to 0. When the height difference between the lowest points 31 of the two reference signal edges 24 is greater than 0 (not equal), it is determined that the two reference signal edges 24 can be distinguished as corresponding to the convex figure 11 and the concave figure 12 respectively. Otherwise, it is considered that the two reference signal edges 24 correspond to the same figure (both convex figures 11 or both concave figures 12) and the recognition fails. Specifically, in actual applications, since the signal edges corresponding to the same figure at different positions often have certain differences, setting the preset height difference threshold equal to 0 will increase the possibility of recognition failure. Therefore, the preset height difference threshold is generally set to be greater than 0.

[0095] Then, step A2 is performed to obtain CDSEM identification information 14 of the target pattern corresponding to the semiconductor structure by measurement. The CDSEM identification information 14 includes a morphology distribution map and a characteristic curve corresponding to the morphology distribution map. Fig.15 ,like Figure 5 As shown, the characteristic curve Fig.15 The signal segment 26 includes m signal peaks 25, where m is an integer greater than or equal to 1.

[0096] Next, proceed to step A3, such as Figure 6 As shown, the characteristic curve Fig.15 Each of the peak points of the signal peak 25 and perpendicular to the characteristic curve Fig.15 The straight line of the horizontal axis 16 is the first dividing line 21, along the characteristic curve Fig.15 The first direction 271 of the horizontal axis 16 is spaced a first preset distance 291 from the first cutting line 21 and is perpendicular to the characteristic curve Fig.15The straight line on the horizontal axis 16 is the second cutting line 22 corresponding to the first cutting line 21, along the characteristic curve Fig.15 The second direction 272 of the horizontal axis 16 is spaced a second preset distance 292 from the first cutting line 21 and is perpendicular to the characteristic curve Fig.15 The straight line of the horizontal axis 16 is the third cutting line 23 corresponding to the first cutting line 21, and the first direction 271 and the second direction 272 are along the characteristic curve Fig.15 Two opposite directions of the horizontal axis 16; Identify the characteristic curve of the semiconductor structure Fig.15 The first cutting line 21 and the second cutting line 22 in the semiconductor structure are cut into the characteristic curve Fig.15 The signal segment 26 between the first cutting line 21 and the second cutting line 22 is an identification signal edge 28; the characteristic curve of the semiconductor structure is identified. Fig.15 The first cutting line 21 and the third cutting line 23 in the semiconductor structure are cut into the characteristic curve Fig.15 The signal segment 26 between the first cutting line 21 and the third cutting line 23 is an identification signal edge 28; from the characteristic curve of the semiconductor structure Fig.15 n identification signal edges 28 are obtained, where n is an integer greater than or equal to 2.

[0097] Specifically, the first preset distance 291 and the second preset distance 292 may be equal or unequal, and the recognition system used by the graphic recognition method to recognize the second cutting line 22 and the third cutting line 23 is set and determined in advance according to the requirements of different preset detection features.

[0098] Preferably, the first preset distance 291 and the second preset distance 292 are both smaller than a signal peak 25 along the characteristic curve Fig.15 Half of the length in the direction of the horizontal axis 16.

[0099] Then, step A4 is performed, according to the characteristic curve of the corresponding target pattern in the semiconductor structure Fig.15 The target pattern in the semiconductor structure is measured to obtain an identification detection feature corresponding to a preset detection feature.

[0100] Finally, step A5 is performed to obtain the identification feature relationship corresponding to the distinguishing feature relationship obtained between the target pattern in the semiconductor structure and the reference structure according to the identification detection feature of the target pattern in the semiconductor structure, and determine whether the characteristic curve of the semiconductor structure Fig.15 The position corresponding to each identification signal edge 28 in the image is a convex pattern 11 or a concave pattern 12.

[0101] The pattern recognition method provided by the present invention is to obtain the characteristic curve in the CDSEM recognition information 14 of the target pattern of the semiconductor structure by measuring Fig.15 The segmentation is performed into each identification signal edge 28, and the distinguishing feature relationship of the CDSEM identification information 14 of the reference structure calibrated in advance is compared with the obtained identification feature relationship to identify whether the segmented identification signal edge 28 corresponds to the convex figure 11 or the concave figure 12. By identifying and distinguishing the features of a single signal edge, the recognition of the convex figure 11 and the concave figure 12 under the influence of various difference factors between the reference structure and the actual process and imaging process is improved, such as Figure 7 When measuring the space pattern (depression pattern 12), the space pattern is used as the measurement area 13 for measurement, as shown in FIG. Figure 8 When measuring the line pattern (convex pattern 11), the line pattern is measured as the measurement area 13, which reduces the risk of reverse measurement between the convex pattern 11 and the concave pattern 12, and improves the pattern recognition accuracy and key dimension measurement accuracy of CDSEM.

[0102] Specifically, in step A3, from the characteristic curve of the semiconductor structure Fig.15 After obtaining n identification signal edges 28, each identification signal edge 28 is numbered and named to obtain the numbered name corresponding to each identification signal edge 28; in step A4, according to the numbered name of each identification signal edge 28 and the distinguishing feature relationship obtained from the reference structure, it is determined whether the identification signal edge 28 corresponding to the numbered name of each identification signal edge 28 is a convex figure 11 or a concave figure 12.

[0103] In one embodiment, the detailed implementation process of the graphic recognition method of Example 1 is as follows:

[0104] Step a1): Fig. 9 As shown, the characteristic curve from the reference structure Fig.15 4 reference signal edges 24 are obtained, and combined with the positions of the convex pattern 11 and the concave pattern 12 in the target pattern of the reference structure manually calibrated, the Figure 4 The two first reference signal edges 01 and 02 correspond to the raised pattern 11, and the two second reference signal edges 03 and 04 correspond to the recessed pattern 12;

[0105] Step a2): From Fig. 9 The preset detection features (i.e., the heights of the first lowest point 311 and the third lowest point 313) of the first reference signal edge 01 and the second reference signal edge 03 are measured, and the first lowest point 311 of the first reference signal edge 01 is higher than the third lowest point 313 of the second reference signal edge 03, and the height difference is ΔH01;

[0106] Step a3):

[0107] Select the first reference signal edge 02 and the second reference signal edge 04 to repeat step a2), and obtain that the second lowest point 312 of the first reference signal edge 02 is higher than the fourth lowest point 314 of the second reference signal edge 04, and the height difference is ΔH02;

[0108] Select the first reference signal edge 01 and the first reference signal edge 02 to repeat step a2), and obtain that the first lowest point 311 of the first reference signal edge 01 is equal to the second lowest point 312 of the first reference signal edge 02, and the height difference ΔH03=0;

[0109] Select the second reference signal edge 03 and the second reference signal edge 04 to repeat step a2), and obtain that the third lowest point 313 of the second reference signal edge 03 is lower than the fourth lowest point 314 of the second reference signal edge 04, with a height difference ΔH04;

[0110] Step a4): Modeling is performed based on the data obtained in step a2) and step a3), and the distinguishing characteristic relationship is obtained as follows: the lowest point 31 of the reference signal edge 24 corresponding to the convex figure 11 is higher than the lowest point 31 of the reference signal edge 24 corresponding to the concave figure 12, and the height difference is greater than or equal to a preset height difference threshold ΔH;

[0111] Step a5): measuring the characteristic curve of the CDSEM identification information corresponding to the target pattern in the semiconductor structure Fig.15 , and cut it to get Figure 6 The two first identification signal edges 41, 42 and the two second identification signal edges 43, 44 are shown;

[0112] Step a6): From Figure 6 The measured identification detection features of the first identification signal edge 41 and the second identification signal edge 43 (i.e., the heights of the fifth lowest point 315 and the seventh lowest point 317) are such that the identification feature relationship between the first identification signal edge 41 and the second identification signal edge 43 is as follows: the fifth lowest point 315 of the first identification signal edge 41 is higher than the seventh lowest point 317 of the second identification signal edge 43, and the height difference is ΔH11 greater than or equal to ΔH;

[0113] Step a7):

[0114] Select the first identification signal edge 42 and the second identification signal edge 44 to repeat step a6), and obtain the identification feature relationship between the first identification signal edge 42 and the second identification signal edge 44: the sixth lowest point 316 of the first identification signal edge 42 is higher than the eighth lowest point 318 of the second identification signal edge 44, and the height difference is ΔH12 greater than or equal to ΔH;

[0115] Select the first identification signal edge 41 and the first identification signal edge 42 to repeat step a6), and obtain the identification feature relationship between the first identification signal edge 41 and the first identification signal edge 42: the fifth lowest point 315 of the first identification signal edge 41 is equal to the sixth lowest point 316 of the first identification signal edge 42, and the height difference ΔH13=0 is less than ΔH;

[0116] Select the second identification signal edge 43 and the second identification signal edge 44 to repeat step a6), and obtain the identification feature relationship between the second identification signal edge 43 and the second identification signal edge 44: the seventh lowest point 317 of the second identification signal edge 43 is higher than the eighth lowest point 318 of the second identification signal edge 44, and the height difference ΔH14 is less than ΔH;

[0117] Step a8): Determine the recognition result based on the recognition feature relationship obtained in step a6) and step a7) combined with the distinguishing feature relationship obtained in step a4):

[0118] The first identification signal edge 41 corresponds to the raised pattern 11, and the second identification signal edge 43 corresponds to the recessed pattern 12;

[0119] The first identification signal edge 42 corresponds to the raised pattern 11, and the second identification signal edge 44 corresponds to the recessed pattern 12;

[0120] The first identification signal edge 41 and the first identification signal edge 42 correspond to the same pattern and the identification fails;

[0121] The second identification signal edge 43 and the second identification signal edge 44 correspond to the same pattern and the identification fails.

[0122] Specifically, in step a1), the characteristic curve of the reference structure is Fig.15 The reference signal edges 24 obtained may also be other numbers. Those skilled in the art may select a specific number of combinations of the reference signal edges 24 for measuring the preset detection features in steps a2) and a3) as required, which are all within the protection scope of the present invention.

[0123] Specifically, when modeling is performed in step a4) to obtain a distinguishing characteristic relationship, the preset height difference threshold ΔH can be the minimum value x of the height difference between the lowest points of the first reference signal edge 01 or 02 corresponding to the convex figure and the second reference signal edge 03 or 04 corresponding to the concave figure, or it can be the maximum value y of the height difference between the lowest points of the first reference signal edges 01, 02 corresponding to the convex figure and the height difference between the lowest points of the second reference signal edges 03, 04 corresponding to the concave figure. It is also possible to select a suitable parameter between x and y as required so that the difference between the reference signal edges 24 corresponding to different figures and the difference between the reference signal edges 24 corresponding to the same figure will not be confused.

[0124] In one embodiment, the preset height difference threshold ΔH is 5% of the height difference between the lowest point (0% signal point) and the highest point (100% signal point) of the corresponding identification signal edge 28. Specifically, the preset height difference threshold may also be other suitable values.

[0125] Specifically, in step a8), if it is first determined that "the first identification signal edge 41 and the first identification signal edge 42 correspond to the same kind of graphic", and it is not determined whether the first identification signal edges 41 and 42 specifically correspond to the concave graphics 12 or the convex graphics 11, the result of "the first identification signal edge 41 and the first identification signal edge 42 correspond to the same kind of graphic" can be retained first, and according to subsequent judgments such as "the first identification signal edge 41 corresponds to the convex graphic 11, and the second identification signal edge 43 corresponds to the concave graphic 12", the judgment result that the first identification signal edge 42 corresponds to the convex graphic 11 can be obtained based on the previous results; the specific judgment logic can be reasonably set according to the above examples.

[0126] Embodiment 2:

[0127] This embodiment provides a pattern recognition method, and other features of the pattern recognition method are the same as those of Embodiment 1, except that:

[0128] In this embodiment, the preset detection feature is the absolute value of the slope of the reference signal edge 24; and the identification detection feature is the absolute value of the slope of the identification signal edge.

[0129] In this embodiment, the distinguishing characteristic relationship is the size relationship between the difference in the absolute values ​​of the slopes of the two reference signal edges 24 and the preset absolute value difference threshold; the identification characteristic relationship is the size relationship between the difference in the absolute values ​​of the slopes of the two identification signal edges to be identified and the preset absolute value difference threshold.

[0130] The present invention sets the magnitude relationship of the absolute values ​​of the slopes of the two reference signal edges 24 as a distinguishing characteristic relationship, which can be used when the lowest points 31 of the two reference signal edges 24 are similar in height and cannot be distinguished. As a replaceable distinguishing characteristic relationship, it reduces the risk of failure in fitting the protruding figure 11 and the recessed figure 12 through the distinguishing characteristic relationship of the reference structure from the identification characteristic relationship of the CDSEM identification information 14 of the semiconductor structure, thereby improving the success rate of CDSEM graphic recognition.

[0131] Specifically, the “preset absolute value difference threshold” means that when the difference in the absolute values ​​of the slopes of the two reference signal edges 24 is greater than or greater than or equal to the preset absolute value difference threshold, it is determined that the two reference signal edges 24 can be distinguished as corresponding to the convex figure 11 and the concave figure 12, respectively; otherwise, it is considered that the recognition has failed. The specific value of the preset absolute value difference threshold can be set according to the accuracy and size requirements for distinguishing the convex figure 11 and the concave figure 12.

[0132] In one embodiment, the detailed implementation process of the graphic recognition method of Example 2 is as follows:

[0133] Step b1): Fig.10 As shown, the characteristic curve from the reference structure Fig.15 The two reference signal edges 24 are obtained by combining the positions of the convex pattern 11 and the concave pattern 12 in the target pattern of the reference structure manually calibrated, and the Fig.10 The first reference signal edge 01 corresponds to the raised pattern 11, and the second reference signal edge 03 corresponds to the recessed pattern 12;

[0134] Step b2): From Fig.10 The preset detection features of the first reference signal edge 01 and the second reference signal edge 03 (i.e., the absolute values ​​of the slopes of the first reference signal edge 01 and the second reference signal edge 03) are measured, and the absolute value of the slope of the first reference signal edge 01 is smaller than the absolute value of the slope of the second reference signal edge 03, and the difference in the absolute values ​​of the slopes is ΔK01;

[0135] Step b3):

[0136] Modeling is performed based on the data obtained in step b2), and the distinguishing characteristic relationship is obtained as follows: the absolute value of the slope of the reference signal edge 24 corresponding to the convex pattern 11 is smaller than the absolute value of the slope of the reference signal edge 24 corresponding to the concave pattern 12, and the difference in the absolute value of the slope is greater than or equal to a preset absolute value difference threshold ΔK;

[0137] Step b4): measuring the characteristic curve of the CDSEM identification information corresponding to the target pattern in the semiconductor structure Fig.15 , and cut it to get Fig.11 The first identification signal edge 41 and the second identification signal edge 43 are shown;

[0138] Step b5): From Fig.11 The measured identification detection features of the first identification signal edge 41 and the second identification signal edge 43 (i.e., the absolute values ​​of the slopes of the first identification signal edge 41 and the second identification signal edge 43) are such that the identification feature relationship between the first identification signal edge 41 and the second identification signal edge 43 is: the absolute value of the slope of the first identification signal edge 41 is less than the absolute value of the slope of the second identification signal edge 43, and the difference in the absolute values ​​of the slopes is ΔK11 greater than or equal to ΔH;

[0139] Step b6): Determine the recognition result based on the recognition feature relationship obtained in step b5) and the distinguishing feature relationship obtained in step b3):

[0140] The first identification signal edge 41 corresponds to the convex pattern 11 , and the second identification signal edge 43 corresponds to the concave pattern 12 .

[0141] Specifically, in the above steps, only two reference signal edges 24 and two identification signal edges 28 are selected for calibration and identification. In actual application scenarios, multiple reference signal edges 24 may appear. In step b2), multiple reference signal edges 24 are arranged and combined in pairs (any two different reference signal edges 24 are combined) for comparative measurement to obtain preset detection features, so that in step b3), multiple groups of preset detection features are modeled to obtain distinguishing feature relationships; in actual application scenarios, multiple identification signal edges 28 may also appear. In step b5), multiple identification signal edges 28 are arranged and combined in pairs (any two different identification signal edges 28 are combined) for comparative measurement to obtain identification detection features, so that in step b6), identification feature relationships are obtained through multiple groups of identification detection features to determine the graphics corresponding to each identification signal edge 28.

[0142] Specifically, the reference signal edge 24 and the identification signal edge 28 are both signal edges, the highest point of the signal edge is the 100% signal point, and the lowest point of the signal edge is the 0% signal point.

[0143] In one embodiment, the absolute value of the slope of a signal edge can take the highest point of the signal edge as the first slope point, and the position below the highest point (100% signal point) by a preset ratio (the height difference of the signal edge between the 100% signal point and the 0% signal point as the denominator of the ratio, and the height difference of the signal edge between the position determined by the preset ratio and the 100% signal point as the numerator of the ratio) as the second slope point, and the absolute value of the slope of the straight line between the first slope point and the second slope point as the absolute value of the slope of the signal edge, so that two signal edges belonging to the same signal peak can have the same first slope point and intercept a point below the first slope point by the same preset ratio from the same slope point as the second slope point, thereby improving the comparability and distinguishability of the absolute values ​​of the slopes between different signal edges. Specifically, other suitable methods can also be used to determine the first slope point, the second slope point and the preset ratio to meet actual application requirements, all within the scope of protection of the present invention.

[0144] In one embodiment, Fig.10As shown, the signal peak where the first reference signal edge 01 and the second reference signal edge 03 are located together is taken as the 100% signal point position, the first lowest point 311 of the first reference signal edge 01 and the third lowest point 313 of the second reference signal edge 03 are respectively the first slope points 341 and 342 of each other, and the positions of the first reference signal edge 01 and the second reference signal edge 03 that are higher than the 0% signal point by a preset ratio (the signal edge height difference between the 100% signal point position and the 0% signal point position is taken as the denominator of the ratio, and the signal edge height difference between the position determined by the preset ratio and the 0% signal point position is taken as the numerator of the ratio) are respectively taken as the second slope points 351 and 352 of each other, the absolute value of the slope of the straight line between the first slope point 341 and the second slope point 351 is taken as the absolute value of the slope of the first reference signal edge 01, and the absolute value of the slope of the straight line between the first slope point 342 and the second slope point 352 is taken as the absolute value of the slope of the second reference signal edge 03.

[0145] In one embodiment, Fig.10 As shown, the preset ratios corresponding to the second slope points 351 and 352 are 10%-20%.

[0146] In one embodiment, Fig.11 As shown, the signal peak 25 where the first identification signal edge 41 and the second identification signal edge 43 are located together is taken as the 100% signal point position, the first lowest point 311 of the first identification signal edge 41 and the third lowest point 313 of the second identification signal edge 43 are respectively the third slope points 361 and 362 of each other, and the positions of the first identification signal edge 41 and the second identification signal edge 43 that are higher than the 0% signal point by a preset ratio (the signal edge height difference between the 100% signal point position and the 0% signal point position is taken as the denominator of the ratio, and the signal edge height difference between the position determined by the preset ratio and the 0% signal point position is taken as the numerator of the ratio) are respectively taken as the fourth slope points 371 and 372 of each other, the absolute value of the slope of the straight line between the third slope point 361 and the fourth slope point 371 is taken as the absolute value of the slope of the first identification signal edge 41, and the absolute value of the slope of the straight line between the third slope point 362 and the fourth slope point 372 is taken as the absolute value of the slope of the second identification signal edge 43.

[0147] In one embodiment, Fig.11 As shown, the preset ratios corresponding to the fourth slope points 371 and 372 are 10%-20%.

[0148] Embodiment 3:

[0149] This embodiment provides a pattern recognition method, and other features of the pattern recognition method are the same as those of Embodiment 1, except that:

[0150] In this embodiment, the preset detection features are the reference signal edge 24, the first cutting line 21 corresponding to the reference signal edge 24, and the lowest point 31 corresponding to the reference signal edge 24 parallel to the characteristic curve Fig.15 The area integral value of the closed contour formed by the integral bottom edge of the horizontal axis 16; the identification detection feature is the identification signal edge 28, the first cutting line 21 corresponding to the identification signal edge 28, and the lowest point corresponding to the identification signal edge 28 is parallel to the characteristic curve Fig.15 The area integral value of the closed contour formed by the integral base of the horizontal axis 16 .

[0151] In this embodiment, the distinguishing characteristic relationship is the size relationship between the difference between the area integral values ​​of the two reference signal edges 24 and the preset integral difference threshold; the identification characteristic relationship is the size relationship between the difference between the area integral values ​​of the two identification signal edges 28 to be identified and the preset integral difference threshold.

[0152] The present invention sets the size relationship of the area integral values ​​of the two reference signal edges 24 as a distinguishing characteristic relationship, which can be used when the heights or slopes of the lowest points 31 of the two reference signal edges 24 are similar and cannot be distinguished. As a replaceable distinguishing characteristic relationship, it further reduces the risk of failure in fitting the protruding figure 11 and the recessed figure 12 through the distinguishing characteristic relationship of the reference structure in the identification characteristic relationship obtained from the CDSEM identification information 14 of the semiconductor structure, thereby improving the success rate of CDSEM graphic recognition.

[0153] Specifically, the “preset integral difference threshold” means that when the difference between the area integral values ​​of the two reference signal edges 24 is greater than or greater than or equal to the preset integral difference threshold, it is determined that the two reference signal edges 24 can be distinguished as corresponding to the convex figure 11 and the concave figure 12, respectively; otherwise, it is considered that the recognition has failed. The specific value of the preset integral difference threshold can be set according to the accuracy and size requirements for distinguishing the convex figure 11 and the concave figure 12.

[0154] In one embodiment, the detailed implementation process of the graphic recognition method of Example 3 is as follows:

[0155] Step c1): Fig.12 As shown, the characteristic curve from the reference structure Fig.15 The two reference signal edges 24 are obtained by combining the positions of the convex pattern 11 and the concave pattern 12 in the target pattern of the reference structure manually calibrated, and the Fig.10 The first reference signal edge 01 corresponds to the raised pattern 11, and the second reference signal edge 03 corresponds to the recessed pattern 12;

[0156] Step c2): From Fig.12The preset detection features of the first reference signal edge 01 and the second reference signal edge 03 (i.e., the area integral values ​​of the first reference signal edge 01 and the second reference signal edge 03) are measured, and it is obtained that the area integral value of the first reference signal edge 01 is greater than the area integral value of the second reference signal edge 03, and the difference in the area integral values ​​is ΔS01;

[0157] Step c3):

[0158] Modeling is performed based on the data obtained in step c2), and the distinguishing characteristic relationship is obtained as follows: the area integral value of the reference signal edge 24 corresponding to the convex pattern 11 is greater than the area integral value of the reference signal edge 24 corresponding to the concave pattern 12, and the difference in the area integral values ​​is greater than or equal to the preset absolute value difference threshold ΔS;

[0159] Step c4): measuring the characteristic curve of the CDSEM identification information corresponding to the target pattern in the semiconductor structure Fig.15 , and cut it to get Fig.13 The first identification signal edge 41 and the second identification signal edge 43 are shown;

[0160] Step c5): From Fig.13 The identification detection features of the first identification signal edge 41 and the second identification signal edge 43 (i.e., the area integral values ​​of the first identification signal edge 41 and the second identification signal edge 43) obtained by measurement are as follows: the area integral value of the first identification signal edge 41 is greater than the area integral value of the second identification signal edge 43, and the difference in the area integral values ​​is ΔS11 greater than or equal to ΔS;

[0161] Step c6): Determine the recognition result based on the recognition feature relationship obtained in step c5) and the distinguishing feature relationship obtained in step c3):

[0162] The first identification signal edge 41 corresponds to the convex pattern 11 , and the second identification signal edge 43 corresponds to the concave pattern 12 .

[0163] Specifically, in the above steps, only two reference signal edges 24 and two identification signal edges 28 are selected for calibration and identification. In actual application scenarios, multiple reference signal edges 24 may appear. In step c2), multiple reference signal edges 24 are arranged and combined in pairs (any two different reference signal edges 24 are combined) for comparative measurement to obtain preset detection features, so that in step c3), multiple groups of preset detection features are modeled to obtain distinguishing feature relationships; in actual application scenarios, multiple identification signal edges 28 may also appear. In step c5), multiple identification signal edges 28 are arranged and combined in pairs (any two different identification signal edges 28 are combined) for comparative measurement to obtain identification detection features, so that in step c6), identification feature relationships are obtained through multiple groups of identification detection features to determine the graphics corresponding to each identification signal edge 28.

[0164] Specifically, the reference signal edge 24 and the identification signal edge 28 are both signal edges, the highest point of the signal edge is the 100% signal point, and the lowest point of the signal edge is the 0% signal point.

[0165] Specifically, Fig.12 As shown, the area integral value of the first reference signal edge 01 is the first reference signal edge 01, the first dividing line 21 corresponding to the first reference signal edge 01, and the first lowest point 311 corresponding to the first reference signal edge 01 are parallel to the characteristic curve Fig.15 The area integral value of the closed contour formed by the first bottom edge 36 of the horizontal axis 16; the area integral value of the second reference signal edge 03 is the second reference signal edge 03, the first dividing line 21 corresponding to the second reference signal edge 03, and the third lowest point 313 corresponding to the second reference signal edge 03 parallel to the characteristic curve Fig.15 The area integral value of the closed contour formed by the second base 37 of the horizontal axis 16 .

[0166] Specifically, Fig.13 As shown, the area integral value of the first identification signal edge 41 is the first identification signal edge 41, the first cutting line 21 corresponding to the first identification signal edge 41, and the fifth lowest point 315 corresponding to the first identification signal edge 41 are parallel to the characteristic curve Fig.15 The area integral value of the closed contour formed by the third bottom edge 38 of the horizontal axis 16; the area integral value of the second identification signal edge 43 is the second identification signal edge 43, the first cutting line 21 corresponding to the second identification signal edge 43, and the seventh lowest point 317 corresponding to the second identification signal edge 43 parallel to the characteristic curve Fig.15 The area integral value of the closed contour formed by the fourth base 39 of the horizontal axis 16 .

[0167] Embodiment 4:

[0168] This embodiment provides a graphic recognition device, and the graphic recognition method described in any one of Embodiments 1-3 is performed using the graphic recognition device, such as Fig.14 As shown, the graphic recognition device includes: a recognition calculation unit, a CDSEM recognition unit, a segment recognition unit, a signal segment cutting unit and a recognition judgment unit;

[0169] The CDSEM identification unit is used to measure and obtain CDSEM identification information 14 of the corresponding target pattern in the semiconductor structure, and the CDSEM identification information 14 includes a morphology distribution map and a characteristic curve corresponding to the morphology distribution map. Fig.15 The dividing line identification unit is electrically connected to the CDSEM identification unit for obtaining the characteristic curve of the CDSEM identification information 14 in the CDSEM identification unit. Fig.15 Identify the positions of the first cutting line 21, the second cutting line 22 and the third cutting line 23;

[0170] The signal segment cutting unit is electrically connected to the segmentation line recognition unit, and is used to use the positions of the first segmentation line 21, the second segmentation line 22 and the third segmentation line 23 obtained by the segmentation line recognition unit to compare the characteristic curve Fig.15 The signal segment 26 in the image is cut to obtain n identification signal edges 28;

[0171] The recognition judgment unit is electrically connected to the recognition calculation unit and the signal segment cutting unit respectively; the recognition calculation unit calculates the respective recognition detection features of two recognition signal edges 28 to be identified selected from the n recognition signal edges 28, and compares and obtains the recognition feature relationship between the two recognition signal edges 28; the recognition judgment unit determines whether the positions of the two recognition signal edges 28 to be identified selected from the n recognition signal edges 28 obtained from the signal segment cutting unit in the morphology distribution map corresponding to the CDSEM recognition information are convex figures or concave figures through the recognition feature relationship between the two recognition signal edges 28 obtained by the recognition calculation unit.

[0172] In one embodiment, the recognition judgment unit controls the recognition calculation unit to select any one of the preset detection features in Embodiments 1-3 to obtain the corresponding distinguishing feature relationship.

[0173] In one embodiment, a method for measuring and obtaining CDSEM identification information 14 of a target pattern of a semiconductor structure includes:

[0174] Using a scanning electron microscope to measure and obtain CDSEM identification information 14 of the semiconductor structure;

[0175] Select the CDSEM identification information 14 of the target pattern in the semiconductor structure, the CDSEM identification information 14 of the target pattern only includes the morphology distribution diagram and characteristic curve corresponding to the target pattern Fig.15 .

[0176] In one embodiment, Fig.15 As shown, the CDSEM recognition unit is also used to measure and obtain CDSEM recognition information 14 of the corresponding target figure in the reference structure; the segmentation line recognition unit is also used to measure the characteristic curve in the CDSEM recognition information 14 of the reference structure obtained by the CDSEM recognition unit. Fig.15 Identify the positions of the first cutting line 21, the second cutting line 22 and the third cutting line 23; the signal segment cutting unit is also used to use the positions of the first cutting line 21, the second cutting line 22 and the third cutting line 23 obtained by the cutting line identification unit to compare the characteristic curve Fig.15 The signal segment 26 in the reference signal edge 24 is cut to obtain a reference signal edges 24; the recognition calculation unit is also connected to the signal segment cutting unit, and is used to analyze the preset detection features between the convex pattern 11 and the concave pattern 12 in the reference signal edge 24, so as to obtain and store the distinguishing feature relationship between the convex pattern 11 and the concave pattern 12; the distinguishing feature relationship stored by the recognition calculation unit is used to compare with the recognition feature relationship of the recognition signal edge 28 obtained from the target structure of the semiconductor structure later, so as to identify whether the corresponding recognition signal edge 28 is a convex pattern 11 or a concave pattern 12.

[0177] Embodiment 5:

[0178] This embodiment provides a method for measuring the size of a graphic feature. Fig.16 As shown, the method for measuring the graphic feature size includes:

[0179] B1: providing a semiconductor structure including a target pattern, wherein the target pattern includes a protruding pattern 11 and a recessed pattern 12;

[0180] B2: using the pattern recognition method described in any one of embodiments 1 to 3 or / and the pattern recognition device described in embodiment 4 to recognize the positions of the protruding pattern 11 and the recessed pattern 12 in the target pattern of the semiconductor structure;

[0181] B3: The position of the convex pattern 11 identified in the target pattern of the semiconductor structure is measured to obtain the characteristic size of the convex pattern 11, and the position of the concave pattern 12 identified in the target pattern of the semiconductor structure is measured to obtain the characteristic size of the concave pattern 12.

[0182] It should be noted that the above sequence does not strictly represent the graphic feature size measurement method protected by the present invention, and those skilled in the art may change it according to the actual steps of the graphic feature size measurement method.

[0183] In summary, the graphic recognition method and device and the graphic feature size measurement method of the present invention can segment the characteristic curve graph in the CDSEM recognition information of the target graphic of the semiconductor structure obtained by measurement into recognition signal edges, and use the distinguishing feature relationship of the CDSEM recognition information of the reference structure to identify whether the segmented recognition signal edge corresponds to a convex graphic or a concave graphic. By calibrating and identifying the features of a single signal edge, the recognition of the convex and concave graphics under the influence of various difference factors between the reference structure and the actual process and imaging process is improved, the risk of reverse vector measurement between the convex and concave graphics is reduced, and the graphic recognition accuracy and key dimension measurement accuracy of CDSEM are improved; at the same time, by setting the preset detection feature of the distinguishing feature relationship obtained by the reference structure as the height of the lowest point of the reference signal edge, the recognition efficiency of the convex and concave graphics is improved, and the graphic recognition accuracy of CDSEM is further improved; in addition, the size relationship of the absolute value of the slope of the reference signal edge and the size relationship of the area integral value are used as preset detection features for further distinguishing the convex and concave graphics, the risk of fitting failure is further reduced, and the success rate of CDSEM graphic recognition is improved.

[0184] Therefore, the present invention effectively overcomes various shortcomings in the prior art and has high industrial utilization value.

[0185] The above embodiments are merely illustrative of the principles and effects of the present invention, and are not intended to limit the present invention. Anyone familiar with the art may modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or alterations made by a person of ordinary skill in the art without departing from the spirit and technical concept disclosed by the present invention shall still be covered by the claims of the present invention.

Claims

1. A pattern recognition method, characterized in that: The graphic recognition method comprises: Providing a distinguishing feature relationship of CDSEM identification information of a convex figure and a concave figure in a target figure of a reference structure, wherein the distinguishing feature relationship is used to distinguish the convex figure from the concave figure; The CDSEM identification information of the target pattern corresponding to the semiconductor structure is obtained by measuring, wherein the CDSEM identification information includes a morphology distribution diagram and a characteristic curve diagram corresponding to the morphology distribution diagram, wherein the characteristic curve diagram includes a signal segment composed of m signal peaks, where m is an integer greater than or equal to 1; Each straight line in the characteristic curve graph that passes through the peak point of the signal peak and is perpendicular to the horizontal axis of the characteristic curve graph is a first cutting line, a straight line that is spaced from the first cutting line by a first preset distance in the first direction along the horizontal axis of the characteristic curve graph and is perpendicular to the horizontal axis of the characteristic curve graph is a second cutting line corresponding to the first cutting line, a straight line that is spaced from the first cutting line by a second preset distance in the second direction along the horizontal axis of the characteristic curve graph and is perpendicular to the horizontal axis of the characteristic curve graph is a third cutting line corresponding to the first cutting line, and the first direction and the second direction are two opposite directions along the horizontal axis of the characteristic curve graph; identifying the first cutting line and the second cutting line in the characteristic curve graph of the semiconductor structure, and cutting the signal segment between the corresponding first cutting line and the second cutting line in the characteristic curve graph of the semiconductor structure is an identification signal edge; identifying the first cutting line and the third cutting line in the characteristic curve graph of the semiconductor structure, and cutting the signal segment between the corresponding first cutting line and the third cutting line in the characteristic curve graph of the semiconductor structure is an identification signal edge; obtaining n identification signal edges from the characteristic curve graph of the semiconductor structure, where n is an integer greater than or equal to 2; According to the characteristic curve diagram measurement of the corresponding target pattern in the semiconductor structure, obtaining an identification detection feature corresponding to the target pattern in the semiconductor structure and a preset detection feature; According to the identification detection characteristics of the corresponding target pattern in the semiconductor structure, an identification feature relationship corresponding to the distinguishing feature relationship obtained between the target pattern in the semiconductor structure and the reference structure is obtained, and it is determined whether the position corresponding to each identification signal edge in the characteristic curve diagram of the semiconductor structure is a convex pattern or a concave pattern.

2. The pattern recognition method according to claim 1, characterized in that: The method for obtaining the distinguishing characteristic relationship of the CDSEM recognition information of the convex figure and the concave figure in the target figure of the reference structure includes: Providing a reference structure, wherein the reference structure includes a target pattern, and obtaining position distribution information of a convex pattern and a concave pattern of the target pattern in the reference structure; measuring and obtaining CDSEM identification information of a target pattern of the reference structure; The signal segment between the first and second cutting lines corresponding to the characteristic curve diagram of the CDSEM identification information of the reference structure is cut as a reference signal edge, and the signal segment between the first and third cutting lines corresponding to the characteristic curve diagram of the CDSEM identification information of the reference structure is cut as a reference signal edge, and a reference signal edges are obtained from the characteristic curve diagram of the reference structure, where a is an integer greater than or equal to 2; the reference signal edges corresponding to the convex figure and the concave figure are respectively calibrated according to the position distribution information; The distinguishing characteristic relationship between each reference signal edge corresponding to the convex figure and the concave figure is obtained according to the preset detection characteristics.

3. The pattern recognition method according to claim 2, characterized in that: The preset detection feature is the height of the lowest point of the reference signal edge; the identification detection feature is the height of the lowest point of the identification signal edge.

4. The pattern recognition method according to claim 3, characterized in that: The distinguishing characteristic relationship is the size relationship between the height difference of the lowest points of the two reference signal edges and the preset height difference threshold; the identifying characteristic relationship is the size relationship between the height difference of the lowest points of the two identification signal edges to be identified and the preset height difference threshold.

5. The pattern recognition method according to claim 2, characterized in that: The preset detection feature is the absolute value of the slope of the reference signal edge; the identification detection feature is the absolute value of the slope of the identification signal edge.

6. The pattern recognition method according to claim 5, characterized in that: The distinguishing characteristic relationship is the size relationship between the difference in absolute values ​​of the slopes of two reference signal edges and a preset absolute value difference threshold; the identifying characteristic relationship is the size relationship between the difference in absolute values ​​of the slopes of two identification signal edges to be identified and a preset absolute value difference threshold.

7. The pattern recognition method according to claim 2, characterized in that: The preset detection feature is the area integral value of the closed contour formed by the reference signal edge, the first dividing line corresponding to the reference signal edge, and the integral base of the lowest point corresponding to the reference signal edge parallel to the horizontal axis of the characteristic curve graph; the identification detection feature is the area integral value of the closed contour formed by the identification signal edge, the first dividing line corresponding to the identification signal edge, and the integral base of the lowest point corresponding to the identification signal edge parallel to the horizontal axis of the characteristic curve graph.

8. The pattern recognition method according to claim 7, characterized in that: The distinguishing characteristic relationship is the size relationship between the difference in area integral values ​​of two reference signal edges and a preset integral difference threshold; the identifying characteristic relationship is the size relationship between the difference in area integral values ​​of two identification signal edges to be identified and a preset integral difference threshold.

9. A pattern recognition device, characterized in that: The graphic recognition method according to any one of claims 1 to 8 is performed by using the graphic recognition device, the graphic recognition device comprising: a recognition calculation unit, a CDSEM recognition unit, a dividing line recognition unit, a signal segment cutting unit and a recognition judgment unit; The CDSEM recognition unit is used to measure and obtain CDSEM recognition information of a target pattern corresponding to the semiconductor structure, wherein the CDSEM recognition information includes a morphology distribution map and a characteristic curve map corresponding to the morphology distribution map; the segmentation line recognition unit is electrically connected to the CDSEM recognition unit, and is used to identify the positions of the first segmentation line, the second segmentation line, and the third segmentation line on the characteristic curve map in the CDSEM recognition information obtained by the CDSEM recognition unit; The signal segment cutting unit is electrically connected to the segmentation line identification unit, and is used to cut the signal segment in the characteristic curve graph using the positions of the first segmentation line, the second segmentation line and the third segmentation line obtained by the segmentation line identification unit to obtain n identification signal edges; The recognition judgment unit is electrically connected to the recognition calculation unit and the signal segment cutting unit respectively; the recognition calculation unit calculates the respective recognition detection features of two recognition signal edges to be identified selected from the n recognition signal edges, and compares and obtains the recognition feature relationship between the two recognition signal edges; the recognition judgment unit determines whether the positions in the morphology distribution map of the CDSEM recognition information corresponding to the two recognition signal edges to be identified selected from the n recognition signal edges obtained from the signal segment cutting unit are convex figures or concave figures through the recognition feature relationship between the two recognition signal edges obtained by the recognition calculation unit.

10. A method for measuring the feature size of a graphic, characterized in that: The method for measuring the graphic feature size comprises: Providing a semiconductor structure including a target pattern, wherein the target pattern includes a protruding pattern and a recessed pattern; Using the pattern recognition method described in any one of claims 1 to 8 or / and the pattern recognition device described in claim 9 to recognize the positions of the raised patterns and the recessed patterns in the target pattern of the semiconductor structure; The position of the convex pattern identified in the target pattern of the semiconductor structure is measured to obtain the characteristic size of the convex pattern, and the position of the concave pattern identified in the target pattern of the semiconductor structure is measured to obtain the characteristic size of the concave pattern.

Images