KR20040076812A - System and method for detecting and reporting fabrication defects using a multi-variant image analysis - Google Patents

Abstract

PURPOSE: A system and a method for promptly and accurately detecting and recording manufacture defects of a copy by using multi-variable image analysis are provided to accurately identify various detects which are difficult to identify. CONSTITUTION: If an interested area of a copy is captured from an image file, the image file is preliminarily adjusted to correct the image file when necessary(208). The image file is additionally adjusted by using prearranged conversions in numbers corresponding to generation of n adjusted image files(212). If an interested area of a standard is captured from a data file, the data file is preliminarily adjusted to correct the data file when necessary(236). The data file is adjusted by using prearranged conversions to generate n adjusted data files(240). The adjusted image files are arranged with the adjusted data files for extracting residual detect signals including defect information by subtracting each image file from the corresponding data file(248,256).

Description

SYSTEM AND METHOD FOR DETECTING AND RECORDING MANUFACTURING DEFECTS WITH MULTI-PARAMETER IMAGE ANALYSIS {SYSTEM AND METHOD FOR DETECTING AND REPORTING FABRICATION DEFECTS USING A MULTI-VARIANT IMAGE ANALYSIS}

The present invention generally relates to the field of image analysis. In particular, the present invention relates to systems and methods for detecting and recording manufacturing defects using multivariate image analysis.

Accurate and reliable defect inspection is essential to a successful manufacturing process. The manufacturing process seeks to faithfully reproduce the intended design or standard, generally expressed in the form of a computer aided design (CAD) produced by a designer. For example, artwork, reticles, photomasks and product parts, devices and continuous web processes of electronics, semiconductors, displays, microelectronic systems, hard disks and nanotechnology workpieces, It is generally known that variations in standards occur in the manufacture of the same multiple product (eg, making copies of standards). The manufacturing process includes: 1) defects in materials; 2) contamination by particles introduced by various process tools; 3) several harmful effects associated with unwanted environmental pollution; 4) residual film; 5) extra or missing substances; And 6) other defects due to manufacturing process changes are often quite sensitive.

In general, each layer (ie, copy) in a product portion has been formed to similarly represent the specific structure intended according to the particular design of that portion when represented by a standard. Often, product portions are produced through their corresponding masks, reticles, photomasks, or artwork to properly form regions of light that are later directed on the substrate. Depending on the type of resistance (eg, positive or negative) applied over the substrate, the area of light formed by the mask corresponds to the particular structure formed on the substrate or the space between these structures. Corresponding masks, reticles, photomasks, or artwork are patterned with either a 1: 1 representation of the standard, or a derivative of the standard corresponding to the structure formed on the substrate. In other cases, the product portion is produced through direct ablation, machining, milling, etching, or deposition of the material to represent the data structure for which the result is designed.

Conventional inspection systems compare an artwork, reticle, photomask, or product part, i.e., a copy, with either a standard, designed data structure (designer's original intention) or a "golden" image of the part being inspected. It is designed to identify defects. Dissimilarities are recorded as defects.

It has been known that pseudo "false" flaws occur in all automated light inspection systems. The causes of the false defects are various, including the production process of the part to be inspected, the defect observation system, the defect image analysis methodology, the defect classification methodology, and the like. One approach to this problem, disclosed in U. S. Patent No. 4,805, 123 to Specht et al., Is to detect differences between samples to recreate the sample and reduce the effect of false recordings. See US Patent No. 4,532, 650 to Wihl et al. Another approach to this problem, which is disclosed, uses an algorithm that is unique to the defect type (defect at the pattern corner). Another approach to this problem, disclosed in US Pat. No. 6,480,627 to Mathias et al., Is to use an evolution (learning) algorithm implanted in neural networks that affects image features and classification.

Typical conventional inspection systems employ single level analysis of simple images. The quality of inspection can be thought of as being limited to the signal-to-noise ratio that can be obtained by the system. Inspection of artwork, reticles, photomasks and electronics, semiconductors, displays, microelectronic systems, hard disks, and parts of the nanotechnology process using the single-step process described above reliably identifies defects of small size of interest Have limited ability to identify While any single inspection technique is well suited for detecting some form of defect, it is likely to be less suitable for detecting other forms of defect. For any single inspection technique, some of the defects can be easily detected, while other portions of the defects are not well detected or detectable by the inspection system. Different inspection techniques may have different sensitivity to the same group of defects. In addition, any single inspection technique can correctly identify the presence of different types of defects, but cannot distinguish between different types of defects. Therefore, what is required is an improved method of defect detection and recording techniques that detects and discerns defects of interest that are difficult to detect accurately using current inspection techniques, while minimizing the recording of non-defective areas that are recorded as defective. The present invention satisfies these as well as other needs and overcomes the imperfections in current inspection techniques.

In one aspect, the invention relates to a method of inspecting a copy of a defect of interest. The method includes providing an image file containing the region of interest of the copy and converting the image signal with a plurality of conversion functions to obtain a plurality of adjusted image signals. Multiple residual defect signals are extracted using multiple adjusted image signals to determine the presence of the defect of interest. Each residual defect signal of the plurality of residual defect signals corresponds to each of the plurality of adjusted image signals. Rule-based analysis is performed on a number of residual defect signals.

In another aspect, the present invention is directed to a defect detection and classification system for inspecting items for defects of multiple defect types. The system includes a defect detection device configured to perform a plurality of analyzes of a region of interest and to provide at least one signal indicative of the shape of each defect. Each of the plurality of analyzes operates to enhance at least one of the plurality of defect types. The rule-based logic system is configured to communicate with the defect detection apparatus, be operable to receive and process multiple signals, and classify the defects in the region of interest. Each of the plurality of signals is associated with one or more of the plurality of signals having characteristics similar to one of the plurality of signals compared to the others of the plurality of signals.

1 shows an image of a region of interest of a standard;

2 shows an area of interest of a copy corresponding to the standard of FIG. 1;

3 is a flow chart illustrating a multivariate defect data extraction method of the present invention for extracting data relating to a defect in a region of interest of a copy of FIG.

FIG. 4A is an illustration of an adjusted image of the region of interest of FIG. 2 adjusted to enhance defects occurring within a defined structure present within the region of interest;

4B is an illustration of a residual image resulting from extracting a defect in the adjusted image of FIG. 4A;

FIG. 5A shows an adjusted image of the region of interest of FIG. 2 adjusted to reinforce defects occurring in the absence of a defined structure present within the region of interest;

FIG. 5B is an illustration of a residual image resulting from extracting a defect in the adjusted image of FIG. 5A; FIG.

FIG. 6A is an illustration of an adjusted image of the region of interest of FIG. 2 adjusted to reinforce defects occurring at the edges of the defined structure present within the region of interest;

FIG. 6B is an illustration of a residual image resulting from extracting a defect in the adjusted image of FIG. 6A; FIG.

Figure 7 shows a composite image of the residual images of Figures 4B, 5B, and 6B;

8 is a high level schematic diagram of a composite image of a copy composed of a plurality of adjacent regions of interest each affected by the defect data extraction method of FIG.

9 is a flowchart showing a multivariate image analysis and defect recording method of the present invention; And

10 is a schematic high level diagram of an inspection system of the present invention.

Referring now to the drawings, FIGS. 1 and 2 show an image 100 of a region of interest 108 of a standard 112 and a copy 116 of a standard, respectively. As discussed in the background section, standard 112 directly generates copies 116, or intermediate structures (not shown) that are later used to make copies, such as photomasks, reticles, engravings or patterns. It may be a data file or a signal (e.g. a CAD file) used to create an intermediate structure such as. In another form, the standard 112 may be a “golden picture”, intermediate structure, or other item corresponding to the copy 116 of a product portion (not shown). In another form, standard 112 may be any data file or signal that contains information that can be compared or reproduced, i.e., copy 116, of the standard. In this regard, it should be noted that the term "copy" as used herein is not limited to a strict original / copy relationship. The only relationship intended in the Examples section and claims appended hereto between the terms "standard" and "copy" is that the copy is a comparison subject with a standard to determine the differences that may exist between the copy and the standard. . This difference is not necessarily the case, but is often described as a defect. As such, the term "defect" is used in this embodiment section and in the appended claims for convenience. However, the term "defect" should not be construed as limiting the invention to defect detection, and generally should be interpreted as difference detection.

As will be discussed in detail below, the present invention relates to an inspection system (FIG. 10) and a method for detecting and recording defects in a copy 116 image relative to a standard 112 image 100 and apparently a defect in the copy itself. will be. The method generally includes two general steps, which include the first step of extracting information about a defect in the copy 116 image 104 (see method 200 in FIG. 3 and accompanying description). A second step of analyzing and recording such defects (see method 500 of FIG. 9 and accompanying description). In one embodiment, the first step is to take each picture 100, 104 using n predetermined transfer functions, or transformations, to adjust various features within the region of interest 108 to increase the shape of the corresponding defect. It is obtained by adjusting. Transformations commonly used to adjust the image include, but are not limited to, morphological, geometric, convolutional, and thresholding operations. Those skilled in the art will appreciate that a number of variable transforms may be applied in different orders to determine defects of interest to the end user. The method of performing such multiple adjustments may be called "multivariate adjustment".

By using this series of predetermined transfer functions to adjust the region of interest 108 of the image 100, 104 of the copy 116 and the corresponding standard 112, the signal-to-signal that can be obtained with conventional single-level image analysis A defect signal having an S / N ratio higher than the noise (S / N) ratio can be obtained. The predetermined transform function is a parameter that reduces the uncertainty of the results recorded in the region of interest and increases the number of degrees of freedom of inspection in the region of interest 108. The result of each scheduled transformation analysis becomes a subcomponent of the second stage, which generally relates to a larger multivariate analysis that analyzes and records the observed defects.

The resulting quality of the multivariate test of the present invention can be considered superior to any single level analysis due to the resulting increase in the complex S / N ratio of the multivariate adjustment.

The reliability of determining the presence of one or more defects of interest can be increased through the identification of subcomponents of the multivariate analysis. The increased S / N ratio of the composite analysis reduces the likelihood of inaccurate recording of defects. With respect to the system of the present invention, one embodiment of a system 600 for implementing the method of the present invention is shown in FIG. 10 and discussed in the accompanying text.

1 and 2, each image 100, 104 is shown as being an array of 10x10 units, for example, an array of 10 pixels 118x10 pixels. With particular reference to FIG. 1, standard 112 generally includes a defined structure 120, a absence 124 of the defined structure, and a CAD generally including an edge 128 between the defined structure and the absence of the defined structure. It can be a file. As mentioned above, the standard 112 can be used to create a photomask for lithography processes used to fabricate any number of items, for example, the connection layer of a microprocessor die. In this case, the photomask would be a copy 116 of FIG. If the copy 116 is a photomask, the defined structure 120 will be the object in the CAD file representing the opaque portion of the photomask used to pattern the photoresist to the location of the wire. Correspondingly, the absence member 124 of the defined structure 120 will be a transparent portion of the photomask used to pattern the photosensitive resin film into a region where no wire is present. In this example, the copy 116 may also be a finished metal layer fabricated using the treated photosensitive resin film layer or photomask.

In FIG. 1, the pixels 118 corresponding to the defined structure 120 and the absence of the defined structure 124, respectively, are shown to be in high contrast with each other for the sake of convenience. Depending on the shape and / or standard picture 100, this high contrast may be unique. For example, a CAD file may contain only objects, for example lines and shapes, such that the information corresponding to that location at any given location in standard 112 is binary. That is, the object is present or nonexistent there, or either. Those skilled in the art will recognize that, depending on the subject matter of the standard 112 or the characteristics of the image 100 of the standard, the contrast need not be so sharp.

On the other hand, FIG. 2 shows a copy 116 of the standard 112 in the same 10 × 10 unit area 108 of interest shown in FIG. In the example described above with respect to the standard 112, the copy 116 may be a photomask, a layer of resist treated, or a metal layer fabricated using the standard and / or photomask. Correspondingly, the image 104 of the copy 116 may be obtained using an image capture device, such as the detector 604 of the inspection system 600 shown in FIG. For example, the defined structure 120 is generally brighter in FIG. 2 than in FIG. 1. Similarly, absence 124 of defined structure is generally darker in FIG. 2 than in FIG. 1. This difference may be due to properties such as reflectance or absorptance above all of the various performance characteristics of the inspection system 600 (FIG. 10) and / or corresponding features of the item such as the photomask on which the copy 116 is implemented. This difference can be corrected in conjunction with FIG. 3 as will be described later.

Other significant differences include the presence of defects 131-138 in the region of interest 108 of image 104 of FIG. 2. Defects 131-138 may be generated by inspection system 600 (FIG. 10) or during the manufacture of copy 116; 2) contamination by particles introduced by various process tools; 3) several harmful effects associated with unwanted environmental pollution; 4) residual film; 5) extra or missing substances; And 6) other defects due to manufacturing process changes.

When standard 112 and copy 116 are characterized as having one or more defined structures 120, one or more defined structures absent 124, or one or more defined structures edges 128, Defects can generally be grouped into the following six forms: 1) Lack of components of defined structure (eg defect 131); 2) extra components of the defined structure (eg, defect 132); 3) lack of components in the absence of a defined structure (eg, defect 133); 4) extra components in the absence of a defined structure (eg, defect 135); 5) missing components (eg, defects 134 and 138) at the edge of the defined structure; And 6) extra components at the edges of the defined structure (eg defects 136 and 137). It should be noted that the characterization of the standard 112 and the copy and the corresponding defect types described above never exhaust the scope of characterization and the defect types detected and recorded using the present invention. These are only used to simplify the description of the present invention. Those skilled in the art will appreciate that the scope of the present invention extends to the form of defects and imaging characteristics that can be measured by comparing an image of a copy with a standard (or image of a standard).

Referring now to Figures 3 and 1 and 2, Figure 3 is a multivariate defect data extraction of the present invention that can be used to obtain data to be used in the defect analysis and recording method 500, which will be described below with respect to Figure 9 The method 200 is described. In general, the defect data extraction method 200 may proceed as follows. Once the region of interest (eg, region of interest 108 of FIG. 2) of the copy (eg, copy 116 of FIG. 2) has been captured in the image file 204, the image file is subject to further processing in accordance with the present invention. In order to modify the image file, it may be preliminarily adjusted in step 208 as needed. In this regard, it should be noted that the image file 204 may contain only a single area 108 of interest to be processed at this point. Alternatively, the picture file 204 may be a picture 104 of the full copy 116 or a full picture file that includes a portion of the copy that is smaller than the copy but larger than the area of interest. In this case, it may be desirable to extract only one region of interest at a time. The pre-adjusted step 208 may be used to correct the distortion introduced by the optical, mechanical, and / or other system used to capture the image 104 of the copy 116, for example. ) And / or applying one or more photometric corrections to the image file 204 to correct deformations such as illuminance, reflectance, and electronic light element changes.

In steps 212 (1), 212 (2), ... 212 (n), the picture file 204 is composed of n adjusted picture files [216 (1), 216 (2), ... 216 (n) May be further adjusted using the corresponding number n of predetermined transformations to produce]. Each predetermined transform can be selected to increase detection of the corresponding type of defect, or in some cases, a group of defects, to optimize the detection of that type. As will be appreciated by those skilled in the art, the number of predetermined transformations and the transformations themselves can be selected according to any number of criteria. For example, if three types of defects are appropriate for a certain inspection and each of these forms is efficiently detected by the corresponding transform, then the image file 204 may contain three adjusted image files 216 (1), Can be adjusted in steps 212 (1), 212 (2), 212 (n) (n = 3) using three transforms to produce 216 (2), 216 (n) (n = 3)]. have. In an alternative inspection where the certainty of the presence of a defect type is increased by using multiple different transformations, the image file 204 may be transformed by such multiple transformations. For example, if the copy 116 contains two types of defects and uses three transforms for one of the forms, but only uses one transform for the other, then the image file 204 may be four. Will be converted by three conversions to create three adjusted picture files. In another alternative test, two or more types of defects can be detected using only one modification. For example, if four types of defects are the subject of inspection, two types can be detected with one transform and the other two types can be detected with one transform, respectively, so that the converted and adjusted image file The total number of is four. Those skilled in the art will understand that these examples are never exhausted or limited to the broad scope of the present invention, and that it is not practical to enumerate all possible alternatives due to the wide variety of defect forms and transformations available for detecting such defect forms. Could be.

For each of the exemplary standards 112 and copies 116 of FIGS. 1 and 2, the copies (or images 104) are: 1) defects occurring within the defined structure 120; 2) defects occurring in the absence 124 of the defined structure; 3) It may be noted that three general types of defects may be included: defects occurring at the edge 128 of the defined structure. If each such defect type is properly detected by converting the image file 204 by the corresponding predetermined conversion, copy in step 212 (1), 212 (2), 212 (n) (n = 3)]. An image file including the image 104 of 116 is adjusted by each conversion to generate three adjusted image files 216 (1), 216 (2), 216 (n) (n = 3)]. Can be generated.

For example, referring to FIGS. 4A and 2, FIG. 4A illustrates a first adjustment adjusted by applying a first transform to the image file 204 that enhances detection of defects occurring within the defined structure 120. The adjusted image 220 is shown. Proper transformation may be a threshold operation that is at a level equivalent to the pixel values of the structure, so that the structure (lighter pixels) and non-structure (darker structures) can be distinguished. Of course, those skilled in the art will recognize that other transformations are also suitable for enhancing the detection of defects occurring within the defined structure 120. The adjusted image 220 is included in the adjusted image file 216 (1). It should be noted that the level of enhancement of the defect is represented in FIG. 5A in the relative contrast of each unit containing the defect, ie the pixel 118 in the example. As a result, the defects 131 and 132 in the defined structure 120 are relatively very dark. In this example, it should be noted that the first transform may also capture at least some of the defects outside the defined structure 120, particularly those defects at the edge 128 of the defined structure. Thus, defects 133, 134, 136, 137, 138 may also be identified by the first transformation, but at a smaller level of enhancement than defects 131, 132 completely contained within the defined structure 120. . Defects 135 (FIG. 2) that are neither within the defined structure 120 nor at the adjacent edge 128 of the defined structure are not identified by the first transform. This example is illustrative and in no way limited in terms of the scope of the invention.

Referring to FIGS. 5A and 2, FIG. 5A is a second adjusted by applying a second transform to the image file 204 that enhances detection of defects occurring within the absence 124 of the defined structure 120. The adjusted image 224 of FIG. Appropriate transformations may be threshold operations that are at the level equivalent to image inversion and non-structured pixel values, thereby distinguishing between structure and non-structure. Of course, those skilled in the art will recognize that other transformations are also suitable for enhancing detection of defects occurring in the absence 124 of the defined structure 120. The adjusted image 224 is included in the adjusted image file 216 (2). Again, the enhancement level of the defect is represented in FIG. 5A by the relative contrast of each unit containing the defect, ie, pixel 118. As a result, defects in the absence 124 of the defined structure 120, in particular defects 135 that are not immediately adjacent edges 128 of the defined structure, are relatively dark. It should be noted that the second transformation may also capture at least some of the defects outside the absence 124 of the defined structure 120, particularly those at the edge 128 of the defined structure. Thus, the defects 132, 133, 134, 136, 137, 138 can also be identified by a second transformation, but at a level of reinforcement that is smaller than the defects completely contained within the absence 124 of the defined structure 120. Is identified. Defects 131 (FIG. 2) that are neither in the absence 124 of the defined structure 120 nor at the adjacent edge 128 of the defined structure are not identified by the second transformation.

Referring to Figures 6A and 2, Figure 6A illustrates a third adjustment adjusted by applying a third transform to the image file 204 that enhances the detection of defects occurring at the edge 128 of the defined structure 120. The adjusted image 228 is shown. The appropriate transformation may be a threshold operation at a level equivalent to the pixel values of the structure on the edge of the structure, so that the structure and the edge of the structure can be distinguished. Of course, those skilled in the art will recognize that other transformations are also suitable for enhancing the detection of defects that occur at the edge 128 of the defined structure 120. The adjusted image 228 is included in the adjusted image file 216 (n) (n = 3). Again, the enhancement level is represented in FIG. 6A by the relative contrast of each unit containing the defect, ie, pixel 118. As a result, the defects 134, 136, 137, 138 at the edge 128 of the defined structure 120 are relatively very dark. It should be noted that the third transform may also capture at least some of the defects located away from the edge 128 of the defined structure 120, in particular defects relatively adjacent to the edge. Thus, defects 131, 132, 133 may be identified by a third transformation, but at a smaller level of enhancement than the defect located at edge 128. Defects 131, 135 (FIG. 2) that are neither at edge 128 nor relatively adjacent to edge 128 are not identified by the third transform.

Referring to Figures 3 and 1 and 2, Figure 3 illustrates that once the region of interest 108 of standard 112 is captured in data file 232, it is necessary to further process the data file in accordance with the present invention. When present, the data file may be preliminarily adjusted in step 236. It should be noted that the term "data file" rather than the term "image file" is used for the standard 112 because the standard may be included in a non-picture form, such as a CAD file or other data file. Similar to the image file 204, the data file 232 may include only a single region of interest 108 to be processed at this point. Alternatively, data file 232 may be an entire data file that includes information about the entire standard 112 or a portion of the standard that is smaller than the standard but larger than the area of interest. In such cases, it may be desirable to extract only information about the area of interest. Preliminary adjusted step 236 includes applying one or more formal manipulations of the data file 232 to compensate for the deformation introduced into the copy 116 by the manufacturing process used to manufacture the copy if necessary. can do. It should be noted that depending on the state of data file 232 and copy 116, preliminary adjustment step 236 may not need to be performed.

After the data file 232 has been preliminarily adjusted in its entirety, adjustment is made using the predetermined conversion used to adjust the image file 204 in steps 240 (1), 240 (2), ... 240 (n). Then, n adjusted image files 216 (1), 216 (2), ... 216 (n) are generated. The result of this adjustment is n adjusted data files 244 (1), 244 (2),... 244 (n) adjusted for analysis of the region of interest 108. For example, with respect to the exemplary standards 112 and copies 116 of FIGS. 1 and 2, respectively, portions of a file, such as data file 232, or region of interest 108, may be first, first, or third. By each of the first, second, and third conversions described above for the second and third adjusted picture files 216 (1), 216 (2), ... 216 (n). Coordinated data file 244 (1), which acts as a template to enhance extraction of defects within the defined structure 120, and enhances the extraction of defects in the absence 124 of the defined structure. A coordinated data file [244 (2)] serving as a template for providing a template and a coordinated data file [244 (n) () which provides a template for enhancing extraction of defects at the edge 128 of the defined structure 120 ( n = 3)]. Note that additional transformations (not shown) may be applied to both data file 232 and image file 204 to further characterize the observed defects.

After the adjusted image and data files 216 (1), 216 (2), ... 216 (n), 244 (1), 244 (2), ... 244 (n)] are generated, step 248 , Each adjusted picture file 216 (1), 216 (2), ... 216 (n) corresponds to a corresponding respective adjusted data file 244 (1), 244 (2), ... 244. (n)], the corresponding residual defect signal (file) containing the defect information [252 (1), 252 (2), ... 252 (n)], for example, in each step 256 The image file can be extracted by subtracting from each corresponding adjusted data file. For example, with respect to the exemplary standards 112 and copies 116 of FIGS. 1 and 2, the adjusted picture file 216 (1) is aligned with and from the adjusted data file 244 (1). Residual defect signal [252] including residual image 260 (FIG. 4B) in which defects 131, 132 in subtracted and defined structure 120 have been enhanced and other defects 133, 134, 136, 137, 138 have been observed. (1)]. Similarly, the adjusted image file 216 (2) is aligned with and subtracted from the adjusted data file 244 (2) to enhance the defect 135 in the absence 124 of the defined structure 120. Generate a residual defect signal 252 (2) that includes a residual image 264 (FIG. 5B) in which other defects 132, 133, 134, 136, 137, and 138 have been observed. Similarly, the adjusted picture file 216 (n) (n = 3) is aligned with and subtracted from the adjusted data file 244 (n) (n = 3) to define the edge of the structure 120 defined ( Residual defect signal 252 (n) (n) containing residual image 268 (FIG. 6B) in which defects 134, 136, 137, 138 in 128 have been enhanced and other defects 131, 132, 133 were observed. = 3)]. Residual defect signals 252 (1), 252 (2),... 252 (n) may be stored in a common database 272 for further analysis and processing as described below. The database 272 provides relevant information regarding the characteristics of each defect, such as, but not limited to, position, size, density, shape, major axis, longitudinal axis, clusters and shapes, for example, of the defined structure 120. It may include such forms as defects that are fully internal, defects that are fully external to the defined structure 120, or defects at the edge 128 of the defined structure 120. In some probable but unlikely circumstances, it is used to generate n adjusted picture files (216 (1), 216 (2), ... 216 (n) (n = 3)). Note that it may not be necessary to generate n adjusted data files 244 (1), 244 (2), ... 244 (n) (n = 3) by applying the transform. In such an environment, each adjusted picture file 216 (1), 216 (2),... 216 (n) (n = 3) has a residual defect signal [252 (1), 252 (2),... .252 (n)] may be extracted directly from the data file 232 (or a portion thereof) in a circular or pre-adjusted form to produce. Fig. 7 shows an adjusted image file 216 (1), 216 (2), from each of the adjusted data files 244 (1), 244 (2), ... 244 (n) (n = 3). Residual defect signals 252 (1), 252 (2), ... 252 (n) (n = 3) resulting from the extraction of ..216 (n) (n = 3)] and other processing described above. ) Shows a composite residual image 300.

The defect extraction method 200 of FIG. 3 described above is often performed in a small portion (FIG. 2) of the entire image of the entire copy 216 or in a portion of the copy (FIGS. 1 and 2) larger than the region 108 of interest. Will be performed. For example, referring to FIG. 8, the defect extraction method 200 is shown with respect to an area 108 of interest that is 10 pixels by 10 pixels in size, but the corresponding entire image 400 is hundreds, thousands, or more. More than one pixel (or other unit) will be hundreds, thousands or more pixels (units). If one wishes to analyze one or more larger areas of the total image 400 or the entire total image, the individual regions of interest 408 are processed in order in the manner described above with respect to the defect extraction method 200 of FIG. The method may be performed in the same order as the reciprocating scanning method 404 shown in FIG. Of course, multiple regions of interest 408 may be processed in any other order desired.

Referring to Fig. 9 and often to Figs. 1 to 3 and other figures mentioned, Fig. 9 shows the n residual defect signals 252 (1), 252 (2), generated in the combined data extraction method 200 of Fig. 3; 252 (n)] illustrates a rule-based multivariate data analysis and recording method 500 of the present invention for use in analyzing and recording defects included in. As can be seen, the analysis and recording method 500 is exclusively in the form of the defects described above with respect to the standards 112 and copies 116 of FIGS. 1 and 2, ie 1) defined structure 120. Included; 2) exclusively included in the absence 124 of the defined structure; 3) Described separately with respect to defects in the edge 128 of the defined structure. However, one of ordinary skill in the art will appreciate that such forms of defects may be replaced and / or supplemented by other forms of defects in accordance with the forms and features of copy 116 and standard 112 contemplated and adapting such other forms. It will be understood that modifications to the analysis and recording method 500 should be made in order to do so.

As described above in connection with the defect data extraction method 200, the residual defect signals 252 (1), 252 (2),... 252 (n) are not essential, but generally in a common database 272. Stored. The analysis and recording method 500 may begin at step 504 by, for example, starting a program, program module, routine, or other program segment or driving an analysis circuit (not shown) that executes the method. Once this is done, the substantial steps of the analysis and recording method 500 will proceed as follows. In step 508, all residual defect signals 252 (1), 252 (2), ... 252 (n) extracted by the defect data extraction method 200 (FIG. 3) are arranged to align all the defects in position. Proceed. For example, in association with the pixel-by-pixel representation described above in connection with applying the defect data extraction method 200 (FIG. 3) to the standards 112 and copies 116 of FIGS. By position in a 10x10 pixel array, for example, the lower left corner pixel 512 of each residual image 260, 264, 268 (FIGS. 4B, 5B, 6B) is positioned at pixel position (1,1). Can be sorted by identification). Thus, for example, referring to FIG. 7, the defect 131 (FIG. 2) is located at pixel 1,5, and the defect 132 is located at pixel position 2,1. You can say

At step 514, a test is made to determine whether all positions (eg, pixel positions) have been processed in the remaining steps of method 500. If all have been done, method 500 ends at step 515. However, if all positions have not yet been processed, the method proceeds to step 516, where each observed defect is one of n residual defect signals [252 (1), 252 (2), ... 252 (n)]. May be performed to determine whether or not it has been recorded by only one. If only one of the remaining defect signals 252 (1), 252 (2), ... 252 (n) has recorded the defect, then the next defect is observed at step 520. *** If one or more defect signals 252 (1), 252 (2), ... 252 (n) have recorded a defect, the set of residual defect signals analyzed so far, i. A test is performed to determine whether the residual defect signals 252 (1), 252 (2), ... 252 (n) are generic, that is, whether all n residual defect signals containing the defect have been identified. . If the set of residual defect signals is not comprehensive, then the next available residual defect signal is analyzed in step 528 to determine whether the signal identifies a problem that is currently problematic.

On the other hand, if the set of residual defect signals that have identified a problem at present is inclusive, then in step 532 the first structure in which the location of the defect represented in all residual defect signals of the set is of interest [e.g., Figures 1 and A test is performed to determine whether or not exclusively within the defined structure 120 of FIG. If the set of residual defect signals is exclusive within the defined structure 120, the signal set may be processed at step 536 for the first major variant (ie, the presence of an exclusive defect within the defined structure). The identified defect is then recorded at step 540 and the analysis proceeds to the next residual defect signal (e.g., one of the residual defect signals 252 (2), ... 252 (n)) at step 544. . If the set of residual defect signals is not exclusive inside the defined structure 120, a test may be performed to determine whether the set is exclusive outside the defined structure at step 548.

If the set of residual defect signals is exclusive outside the defined structure 120, the signal set may be processed at step 552 for a second variant (i.e. the presence of an exclusive defect inside the defined structure). The defect is recorded at step 556 and the analysis proceeds to the next remaining defect signal at the next step 560. If the set of residual defect signals is not exclusive here or inside the defined structure 120, it can be inferred that it is at the edge 128 of the defined structure 120. Thus, in step 564 the set of residual defect signals is processed for a third variant (i.e., the presence of a defect at the edge of the defined structure 120), the defect is recorded in step 568, and the analysis is performed in step 572. Proceed to the next remaining fault signal. Analysis continues through the steps described above until all residual defect signals in the database 272 have been processed. Subsequent defect analysis and recording by unique multivariate properties are available to the user. At the end, that is, in step 515 of method 500, each variation is included in the relevant report (steps 540, 556, 568). Such a report can be further processed to generate a specific end user classification. For example, extra structures or missing structures may be recorded based on size area, location or some other characteristic of importance to the end user.

10 illustrates an exemplary embodiment of the inspection system 600 of the present invention that may be used to capture an image (not shown) of a copy 116 (FIG. 2). This captured image can be used as the image 104 (FIG. 2) by itself or to extract the image 104 for use in the defect extraction method 200 of FIG. 3. When the data file 232 (FIG. 3) includes the "golden image" of the standard 112 (FIG. 1), the inspection system 600 may affect the analysis occurring in the capture process or others and performed in accordance with the present invention. It may be edited to remove any defects to be made and used to capture the standard picture 100 resulting in a golden picture.

Inspection system 600 supports a detector 604, such as a CCD camera, and a main board (not shown) that includes a copy 116 (FIG. 2) (or standard 112 (FIG. 1)) while capturing an image, and optionally It may include a stage system 608 having, for example, X and Y stages 612 and 616 to move, if stage system 608 includes a movable stage such as X and Y stages 612 and 616. If so, the stage system may further include a position measuring device 620 for determining the position of each of the X and Y stages, and a motor driver 622 for driving each of the X and Y stages. It may also include a system controller 624 that controls the movement of the X and Y stages 612, 616 and controls the capture of the image using the detector 604. In this regard, the system controller 624 may include X and Y. Behavioral agent controlling movement of Y stage 612, 616 Frame capturer 632 for capturing images from base 628 and detector 604.

To perform the inspection, the main substrate is coupled with the stage system 608 and irradiated with electromagnetic energy (eg, white light). Of course, depending on the nature of the main substrate and / or the nature of the copy 116 (or standard 112) on the substrate and the shape of the detector 604, the main substrate may be a non-white light energy (e.g., spectrally tuned). Light, laser beam, acoustic beam, or charged particle beam). During inspection, the image signal (file) used as the image file 204 (FIG. 3) or data file 232 or to generate the image file 204 or data file 232 is a detector 604, stage system. 608, and generated by the system controller 624 and stored in a suitable storage location (not shown) by the frame catcher 632.

10 and 3, the inspection system 600 further includes a processing system 636 that processes the image and data files 204, 232 in the manner described above with respect to the defect extraction method 200 of FIG. 3. It may include. The processing system 636 includes the adjusted image files 216 (1), 216 (2), ... 216 (n) and the adjusted data files 244 (1), 244 (2), ... 244 ( n)] (digital) signals (not shown) may be digitally stored and processed respectively. Adjusted data files [244 (1), 244 (2), ... 244 (n)] and their respective adjusted picture files [216 (1), 216 (2), ... 216 (n)] Can be stored in an appropriate storage location (not shown), where processing system 636 aligns the corresponding file and extracts the residual defect signals 252 (1), 252 (2), ... 252 (n)]. And store them in the database 272. Residual defects present in any one or more of the residual defect signals 252 (1), 252 (2),... 252 (n) may be considered as raw defects. This residual defect can then be analyzed as described above in connection with the defect analysis and recording method 500 of FIG. 9 to extract the characteristic data on a per-position (eg per pixel) basis. In one embodiment, the processing system 636 captures a defect data buffer 640 that maintains information about the identified defects, image (s), and executes the defect extraction method 200 of FIG. The master controller may be configured to control the entire inspection process from executing the defect analysis and recording method 500 of FIG.

Those skilled in the art will understand that the entire processing of the present invention described above provides for quick and accurate inspection of workpieces and elements (ie copies), for example for the identification of defects having the following characteristics: a) Brightly colored burn defects against any background; b) dark colored burn defects against any background; And c) edge contour defects for any background. The treatment of the present invention is not harmful. Since this process provides the ability to simultaneously or independently optimize the inspection of light colored image defects for any background, dark colored image defects for any background, and edge contour defects for any background, such defects. Can be detected in parallel. The technique is therefore well adapted for identifying several different types of defects that are not normally detectable by existing inspection methods. In addition, the number of incorrect classifications of defects is reduced using the present invention. These and other advantages will be apparent to those skilled in the art after reading this disclosure.

Although the present invention has been described and described with reference to exemplary embodiments, those skilled in the art may appreciate that the foregoing and various other changes, omissions, and additions may be made without departing from the spirit and scope of the invention. Should be understood.

Claims (40)

As a way of inspecting copies for defects of interest, a) providing an image signal comprising an area of interest of the copy; b) converting the image signal with a plurality of transform functions to obtain a plurality of adjusted image signals; c) extracting a plurality of residual defect signals using the plurality of adjusted image signals to determine the presence of a defect of interest, wherein each residual defect signal of the plurality of residual defect signals is derived from the plurality of adjusted images. Extracting a plurality of residual defect signals, corresponding to each of the signals; And d) performing rule-based analysis on the plurality of residual defect signals. The process of claim 1, wherein step c) e) providing a data signal comprising said region of interest of a standard; And f) extracting each residual defect signal of the plurality of residual defect signals using a corresponding one of the plurality of first adjusted pictures and the data signal. The process of claim 2, wherein step f) g) converting the data signal with the plurality of transform functions to obtain a plurality of adjusted data signals; And h) extracting each residual defect signal of the plurality of defect signals using the corresponding respective ones of the plurality of adjusted data signals and the plurality of adjusted image signals. 4. The method of claim 3, wherein step h) includes subtracting each of the plurality of adjusted image signals from the corresponding respective one of the plurality of adjusted data signals to obtain a corresponding one of the plurality of residual defect signals. How to. The method of claim 3, wherein the copy comprises at least one defined structure, at least one absence of at least one defined structure, and at least one edge of the at least one defined structure, wherein the defect of interest is at least one. A defect in the defined structure of the defect, a defect in the at least one absence and a defect at the at least one edge. 6. The method of claim 5, wherein the plurality of transformations comprises: a first transformation that enhances a defect of interest in at least one defined structure, a second transformation that enhances a defect of interest in at least one absence; A method comprising a third transformation that enhances the defect of interest at one edge. The method of claim 1, further comprising preconditioning said image signal prior to step b). 8. The method of claim 7, wherein preconditioning the image signal comprises applying a geographic correction to the image signal. 8. The method of claim 7, wherein preconditioning the image signal comprises applying photometric correction to the image signal. 4. The method of claim 3, further comprising preconditioning said data signal prior to step g). 11. The method of claim 10, wherein preconditioning the image signal comprises performing morphology. The method of claim 1, wherein the data signal comprises at least a portion of a CAD file. The method of claim 1, wherein the data signal comprises at least a portion of a golden picture. The system of claim 1, wherein the copy comprises at least one defined structure, at least one absence of at least one defined structure, and at least one edge of the at least one defined structure, wherein the defect of interest is at least. A defect in one defined structure, a defect in at least one absence, and a defect at at least one edge. 15. The method of claim 14, wherein the plurality of transformations comprises: a first transformation that enhances the defect of interest in at least one defined structure, a second transformation that enhances the defect of interest in at least one absence; A method comprising a third transformation that enhances the defect of interest at one edge. The method of claim 1, wherein the rule-based analysis determines the location of each of the defects of interest relative to the at least one structure of interest in the region of interest, and based on their location, at least a portion of the defects of interest. Recording. The method of claim 1, wherein the rule-based analysis comprises determining whether each defect of interest is present in one or more of the plurality of residual defect signals. 18. The method of claim 17, wherein the rule-based analysis comprises recording only one of the defects of interest present in the plurality of residual defect signals. The method of claim 1, wherein the rule-based analysis comprises determining the location of each defect of interest with respect to one of the defined structure, the absence of the defined structure, and the edge of the defined structure. 20. The method of claim 19, wherein the rule-based analysis comprises determining whether a defect of interest is exclusive within a defined structure, exclusive outside a defined structure, or at the edge of the defined structure. As a way of inspecting copies for defects of interest, a) providing an image signal comprising an area of interest of the copy; b) providing a data signal comprising a region of interest of a standard; c) converting the image signal into a plurality of transform functions to obtain a plurality of adjusted image signals; d) transforming the data signal with the plurality of transform functions to obtain a plurality of adjusted data signals; And e) extracting each residual defect signal of the plurality of defect signals using the corresponding respective ones of the plurality of adjusted data signals and the plurality of adjusted image signals. 22. The method of claim 21, further comprising performing a rule based analysis of the plurality of residual defect signals to record at least a portion of the defect of interest. The method of claim 22, wherein the rule-based analysis determines the location of each of the defects of interest relative to the at least one structure of interest in the region of interest, and based on their location, at least a portion of the defects of interest. Recording. 23. The method of claim 22, wherein the rule-based analysis comprises determining whether each defect of interest is present in one or more of the plurality of residual defect signals. 25. The method of claim 24, wherein the rule based analysis comprises recording only one of the defects of interest present in the plurality of residual defect signals. The method of claim 22, wherein the rule-based analysis comprises determining the location of each defect of interest with respect to one of the defined structure, the absence of the defined structure, and the edge of the defined structure. 27. The method of claim 26, wherein the rule-based analysis comprises determining whether a defect of interest is exclusive within the defined structure, exclusive outside the defined structure, or at the edge of the defined structure. A system that checks a copy of a standard for a defect of interest. a) first means for extracting a plurality of residual defect signals in the region of interest using copies and standards; And b) second means for performing rule-based analysis on the plurality of residual defect signals to record at least a portion of the defect of interest. 29. The system of claim 28, wherein the first means comprises third means for converting, by a plurality of transformations, an image file comprising an area of interest of a copy to produce a plurality of adjusted image files. 30. The system of claim 29, wherein the first means comprises fourth means for converting, by the plurality of transformations, a data file comprising a standard region of interest to produce a plurality of adjusted data files. 31. The system of claim 30, wherein the first means comprises fifth means for aligning and subtracting corresponding respective ones of the plurality of adjusted image files and the plurality of data files. 29. The method of claim 28, wherein the second means determines the location of each of the defects of interest relative to the at least one structure of interest in the region of interest, and based on their location, at least a portion of the defects of interest. And a sixth means for recording. 29. The system of claim 28, wherein the second means comprises a seventh means for determining whether each defect of interest is present in one or more of the plurality of residual defect signals. 25. The system of claim 24, wherein the second means comprises eighth means for recording only one of the defects of interest present in one or more of the plurality of residual defect signals. 23. The system of claim 22, wherein the second means comprises a ninth means for determining the location of each defect of interest with respect to one of the defined structure, the absence of the defined structure, and the edge of the defined structure. 27. The system of claim 26, wherein the ninth means further determines whether the defect of interest is exclusive in the defined structure, exclusive in the absence of the defined structure, or at the edge of the defined structure. A defect detection and classification system for inspecting items for defects of multiple defect types, a) a defect detection device configured to perform a plurality of analyzes of a region of interest and to provide at least one signal indicative of the shape of each defect, each of the plurality of analyzes to enhance at least one of the plurality of defect types; A fault detection device in operation; And b) a rule-based logic system in communication with the defect detection device, operatively configured to receive and process the plurality of signals, and classifying a defect in the region of interest, wherein each of the plurality of signals is a plurality of signals. And a rule-based logic system associated with one or more of the plurality of signals having characteristics similar to one of the plurality of signals compared to the other of. 38. The defect detection and classification system of claim 37, wherein the item is selected from the group consisting of artwork, reticles, photomasks and electronics, semiconductors, displays, microelectronic systems, hard disks, and product portions of nanotechnology processes. . 38. The system of claim 37, further comprising a detector operably connected to the defect detection apparatus. A method of constructing a rule-based logic defect detection and classification system that inspects artwork, reticles, photomasks, and electronics, semiconductors, displays, microelectronic systems, hard disks, and product parts of nanotechnology processes. a) measuring multiple processing conditions for the defect inspection equipment using multiple analysis of at least one region of interest and providing a plurality of signals indicative of one or more defects in the at least one region of interest; b) processing the plurality of signals to determine the nature and presence of the one or more defects in the at least one region of interest; c) processing the plurality of signals using a rule based logic system to classify defects in the at least one region of interest.