KR100610441B1 - System and method of providing mask defect printability analysis - Google Patents

Abstract

Simulated wafer images and defect reference images of physical masks are used to generate a materiality score for each defect, thus providing customers with meaningful information to accurately evaluate the results of using the mask or repairing the mask. . The defect criticality score is calculated based on a number of factors related to the change in the critical dimension of the neighboring feature of the defect. The common processing window can also be used to provide objective information about defect printability. Other aspects of the mask related to mask quality, such as line edge roughness and contact corner rounding, can also be quantified by using simulated wafer images of physical masks.

Description

SYSTEM AND METHOD OF PROVIDING MASK DEFECT PRINTABILITY ANALYSIS}

An inspection that provides a defect printability analysis for an integrated circuit mask is described.

Mask / Reticle Defects

In order to fabricate an integrated circuit (IC) on a semiconductor substrate, the physical representation of the IC is transferred onto a pattern tool. The pattern tool is then exposed to transfer the pattern onto the semiconductor substrate. Masks are the standard patterning tool used in IC processing. In general, the mask includes a pattern that can be transferred to the entire semiconductor substrate (eg, wafer) in a single exposure. Another standard pattern tool, the reticle, must be repeated step by step to expose the entire substrate surface. For convenience of reference herein, the term "mask" refers to a reticle or mask.

Typical masks are formed from quartz plates with chromium coatings. Generally, a mask is created for each layer of the IC design. Specifically, the portion of the IC layout data file representing the physical layer (such as the polysilicon layer or the metal layer) is etched into the chromium layer. Thus, each mask includes a pattern representing a predetermined circuit layout for the corresponding layer. In high density ICs, the mask may also include optical proximity correction (OPC) features such as serifs, hammerheads, biases, auxiliary bars. These OPC features are sub-resolution features used to compensate for process artifacts and / or proximity effects.

In high density IC designs, those skilled in the IC manufacturing industry recognize the importance of using a mask that provides an accurate representation of the original design layout. Unfortunately, "complete" masks are not commercially viable. In fact, even under optimal manufacturing conditions, some mask defects can occur outside the controlled process.

The defect on the mask is any deviation (ie, irregularity) from the design database that is considered unacceptable by the inspection tool or inspection engineer. 1 illustrates a flowchart 100 of a conventional method for inspecting an integrated circuit. In step 110, an IC is designed. In step 112, mask design data is generated, for example a data file of the layout of the IC. This data is used to fabricate the mask in step 114. At this point, the mask is inspected in step 116 by scanning the mask surface with a high resolution microscope and capturing an image of the mask. Irregularities in the mask are identified in the list by their location. In one embodiment, the mask has a combined grid pattern and the list represents squares in the grid pattern where irregularities are located. These inspections and irregularity identification can be performed by specialized equipment / software provided by companies such as KLA-Tencor or Applied Materials.

To determine whether the mask passes inspection (step 118), an experienced inspection engineer or semi-automatic inspection equipment reexamines the irregularities identified in step 116. Only irregularities that are considered outside the tolerances set by the manufacturer or the user are characterized by defects. If irregularities are found and outside the tolerances, a determination is made at step 128 if the mask can be repaired. If the mask can be repaired, the mask is cleaned and / or repaired in step 130 and the process returns to step 116 of inspecting the mask. If the mask cannot be repaired, a new mask has to be manufactured and the inspection process returns to step 114. If the mask passes the inspection as determined in step 118, the actual wafer is exposed using the mask in step 120.

In order to ensure that the mask produces the desired image on the wafer, the wafer itself is generally inspected at step 122. If irregularities are found and outside the tolerance as determined in the inspection at step 124, a determination is made at step 128 if the mask can be repaired. If the mask can be repaired, the mask is cleaned and / or repaired in step 130 and the process returns to step 116 of inspecting the mask. If the mask cannot be repaired, a new mask has to be manufactured and the inspection process returns to step 114. If irregularities are found on the wafer but determined to be within tolerance, the mask passes the inspection in step 124 and the inspection process ends in step 126.

Unfortunately, the process described above has many significant disadvantages. For example, automated inspection equipment in principle measures tolerances by size. Thus, if the pinhole in the mask has a certain size, the automatic inspection equipment will indicate that pinhole as a defect regardless of its position on the mask. In contrast, experienced inspection engineers may use additional and more subjective methods depending on their level of experience. Specifically, the skilled engineer will be able to determine whether pinholes in critical areas that are much smaller than a given size will have a detrimental effect on functionality or performance, and therefore should be characterized as defects, or not in critical areas that are larger than a given size. It may be determined whether the pinhole will not affect functionality or performance. However, these techniques must be developed over a long period of time at significant cost. Moreover, as with all human activities, even after these technologies have been developed, the quality of retests inevitably changes. Thus, the step of characterizing the irregularity is prone to mistake.

Another disadvantage of the aforementioned process is the triggering of false defect detection. For example, automated inspection equipment can incorrectly report OPC or incomplete OPC features as defective. As mentioned earlier, OPC features are sub-resolution features used to compensate for proximity effects. Therefore, OPC features will generally not constitute nor contribute to defects.

Mask inspection system

To address this drawback, a mask inspection system designed by Numerical Technologies, INC. Provides a mask quality assessment without depending on the actual exposure of the wafer. Such a mask inspection system is described in US Patent Application No. 09 / 130,996 (hereinafter referred to as NTI system), entitled “Visual Inspection and Verification System,” filed on August 7, 1998, and incorporated herein by reference. It is explained.

2 illustrates a process 200 for checking for defects in a mask in accordance with an NTI system. Processing 200 utilizes inspection tool 202 and wafer image generator 209. In one embodiment, the inspection tool 202 includes an image acquirer 203, which is generally a high resolution imaging device, for scanning all or part of the physical mask 201. The defect detection processor 204 compares the mask image provided by the image acquirer 203 with a set of potential defect criteria to determine which areas of the mask contain potential defects. If a potential defect is identified, the defect detection processor 204 signals the defect area image generator 205 to provide a defect area image of the area containing and surrounding the potential defect.

In one embodiment, the inspection tool 202 provides the defect area image data 206 to the wafer image generator 209. In another embodiment, this data is digitized by digitization device 207, stored in storage 208, and subsequently provided to wafer image generator 209. In another embodiment, which analyzes both areas identified as potential defects and areas not identified as potential defects, the scanned image provided by image acquirer 203 is provided directly to wafer image generator 209 or digitized device. It may be provided indirectly via 207 and storage 208.

Wafer image generator 209 includes an input device 210 that receives data directly from inspection tool 202 in a real time feed or offline from storage 208. Image simulator 211 receives information from input device 210 as well as other input data, such as lithography condition 212. Lithographic conditions 212 include wavelengths of illumination, numerical aperture, coherence value, defocus (where the term defocus refers to focal plane positioning), exposure level, lens aberration, substrate state, and Critical dimensions may be included but are not limited to them. Using these inputs, the image simulator 211 can generate a wafer image 213 that simulates the physical mask 201 exposed on the wafer. Image simulator 211 may also generate simulated processing window 214 and job output 215. In one embodiment, image simulator 211 may also take into account photoresist and / or etching processes as indicated by block 216.

Although the process 200 provides important information to the customer through the simulated wafer image 213, for example, the customer may not be able to perform the appropriate action (e.g., repair the mask or manufacture a new mask). The information must still be reviewed to make a decision regarding Thus, the process 200 can be subject to human error. Therefore, there is a need for a mask inspection system and process that provides an objective and accurate measurement of mask defect printability and mask quality.

A system and method for analyzing defect printability are provided. In this analysis, the physical mask and the corresponding zero reference image are examined. In one embodiment, the defect free reference image may be one of the following: a simulated image of the layout of the physical mask, a defect area of the physical mask having the same pattern, or a simulation of the physical mask as processed in manufacturing. Rating image.

This inspection identifies any defects, ie irregularities, of the physical mask in comparison to the reference image. If a defect is identified, not only the corresponding region image from the reference image, but also the defect region image of the defect and the region surrounding the defect from the physical mask are provided to the wafer image generator. Wafer image generators generate image data, i.e. simulations of physical masks and reference images.

In one embodiment, the wafer image generator may receive a plurality of lithographic conditions. These conditions are specific to the lithographic conditions and system parameters by which the physical mask can be exposed by the customer. Such data may include, for example, the wavelength (λ) of the illumination used in the system, the numerical aperture (NA) of the system, the coincidence value (σ) of the system, the type of illumination (eg off-axis or annular), Focus, exposure level, lens aberration, substrate state, critical dimensions of the design (hereinafter referred to as CD). In one embodiment, each parameter may comprise a range of values, thereby allowing the wafer image generator to generate a plurality of simulations based on the range of lithographic conditions possible in different combinations.

Very nonlinear sub-wavelength manufacturing flows can be compensated in the following manner. Specifically, in order to improve the accuracy of the wafer image in sub-wavelength techniques, the wafer image generator may also receive one or more conversion factors. The conversion factor can vary based on features on the mask, such as isolated lines, tightly packed lines, and contacts. Conversion factors may also vary based on any aspect of the manufacturing process, including stepper parameters and photoresist.

In one embodiment, the test pattern provided in the test mask is simulated using a wafer image generator. The test pattern may include variable width isolated lines, variable width tightly packed lines, and contacts of various sizes. Defect analysis, including any change in the threshold value (CD), can be noted for each image pattern on the simulated wafer image. Note the following facts: That is, as used herein, a CD is a measured or calculated size of a particular location, which can be one or two dimensional. From this information, the conversion factor for each feature can be calculated accurately. Moreover, any number of simulations can be provided using various processes (eg lithographic conditions) to obtain conversion factors for these manufacturing processes. Mask shop specific bias can also be included in the lithographic conditions, thus further improving the accuracy of the conversion factors generated by this embodiment.

This method of providing a conversion factor is very cost-effective because it eliminates the time associated with printing the wafer as well as the time to manufacture the wafer. Moreover, because of the simulation environment, this method offers significant flexibility in optimizing system parameters before actual manufacturing.

According to one embodiment, the defect printability analysis generator receives a simulated wafer image and a reference image of the physical mask from the wafer image generator. In one embodiment, two simulated wafer images are aligned in a preprocessing operation. Alignment can be done using a defect free pattern in the mask or using coordinates from the mask. When these patterns or coordinates are aligned, the features provided on this mask (as well as on the wafer image of these masks) are also aligned.

After the alignment, two-dimensional analysis can proceed. In two-dimensional analysis, defects on the simulated wafer image of the physical mask and corresponding areas on the simulated wafer image of the reference mask are identified. Thereafter, any feature (neighbor feature) in proximity to the defect on the simulated wafer image of the physical mask is identified. In one simple implementation, any feature that is within the fix's distance from the defect may be identified as a neighbor feature. In another embodiment, both the size of the defect and the distance of the defect from neighboring features are compared with the measurements in the design rules table. The design rules table can identify the maximum distance from the defect for each defect size (or size range), where if the feature is located within the maximum distance from the defect, the feature is a neighbor feature. Finally, any neighboring features identified are located on the simulated wafer image of the reference mask.

At this point, defect analysis on the simulated wafer image can be done. Defect analysis involves determining mean CD deviation (ACD), relative CD deviation (RCD), and maximum CD deviation (MCD). To calculate the ACD, the CD of an intact feature in the simulated wafer image of the physical mask is subtracted from the CD of the corresponding feature in the simulated wafer image of the reference mask. This difference is then divided by the CD of the corresponding feature in the simulated wafer image of the reference mask. This calculation results in a CD deviation of the simulated wafer physical image from the simulated wafer reference image (ie, the calibration factor used for later calculations). For higher accuracy, one or more defect areas can be analyzed to provide an ACD for the defect areas. In one embodiment, the ACD is calculated for each exposure.

To calculate the relative CD deviation (RCD), the CD of the identified neighboring feature in the simulated wafer image of the reference mask is subtracted from the CD of the corresponding feature in the simulated wafer image of the physical mask. Note the following: That is, features can be one-dimensional, such as lines, spaces, or two-dimensional, such as contact holes, piles, posts, serifs, or other structures based on regions. This difference is then divided by the CD of the corresponding feature in the simulated wafer image of the reference mask. In one embodiment, an RCD can be calculated for each neighbor feature and each exposure. The maximum CD deviation (MCD) among the RCDs can be determined for each exposure level.

According to one embodiment, the defect printability analysis generator may also receive information from the critical area identification generator. The critical region identification generator provides information to the defect left suitability analysis generator that identifies the region of each mask, which is a designated critical region, such as a gate, that requires a high degree of precision to ensure proper performance in the final IC device. to be. This information is called the tolerance for CD deviation (TCD). Defects in the critical region generally have a lower TCD than defects in the noncritical region.

According to one feature, the defect severity score (DSS) determines the mean CD deviation (ACD), maximum CD deviations (MCDs), tolerance for CD deviations (TCD), and the maximum number of exposures used. It can be calculated using the variable N indicative. One typical equation for calculating this defect criticality score is:

In one embodiment, the defect printability analysis generator outputs a DSS with a scale from 1 to 10 in the impact report. These impact reports can be used to reduce human error in defect printability analysis. For example, a given DSS score may indicate that the printed feature (as simulated by the inspection system) has significant performance issues, but that repair of the physical mask is possible. On the other hand, a DSS score higher than the above may indicate that remaking of physical masks is recommended as well as performance issues. Thus, by providing a numerical result with a combined meaning for each number, the technician will be able to efficiently move forward to the next action, either repairing the physical mask or remaking the physical mask, without mistakes.

In another feature, defect printability can also be objectively assessed using various processing windows. An exemplary processing window is provided by a graph of defocus vs. exposure deviation or focal depth vs. exposure latitude. The curves on these graphs indicate areas that contain defects as well as areas of defects. The largest rectangle fit within these curves is called the exposure defocus window, where the common processing window is the intersection of multiple exposure defocus windows. Focuses and exposures that fall within the common processing window produce resist features, such as CD, within tolerances, while focuses and exposures outside the processing window produce resist features outside the tolerances. Thus, analyzing the processing window associated with the feature may provide an objective means of determining the printability of the feature based on the proximity of the defect. In one embodiment, the defect printability analysis generator may determine a common processing window for the features provided in the physical and reference masks, and provide such information in the impact report.

Impact reporting can be advantageously used to analyze repairs that can be performed on the physical mask. Specifically, when using impact reporting (or portions thereof), the bitmap editor may point to possible corrections to the physical mask to remove or greatly minimize the effects of one or more defects. The bitmap editor can then output a simulated mask (repaired mask) that includes these corrections.

The repaired mask can then be inspected by an inspection tool and used by the wafer image generator to generate new simulated wafer images and new impact reports indicating the success of possible corrections provided to the repaired mask. If the correction is acceptable, the bitmap editor can provide the correction information directly to the mask repair tool for repair of the physical mask. If the customer desires further optimization or analysis of other parameters, the above process may be repeated until the bitmap editor indicates that the correction is deemed to be within acceptable range or that certain results cannot be obtained by repairing the physical mask. have.

In one embodiment, the bitmap editor can also point to an optimized mask writing strategy, for example, to identify which tools can be used for certain defects. In addition, the bitmap editor can receive input indicating customer time or customer restrictions, thereby allowing the bitmap editor to optimize the repair process based on these customer parameters. In another embodiment of the present invention, a bitmap editor can be used to provide information to a wafer repair tool. Specifically, the bitmap editor may include a program that compares the effectiveness of mask repairs to wafer repairs.

Defect printability analysis can be performed on individual defects or on multiple defects. In one embodiment, the inspection tool and wafer image generator automatically provide output for all defects found in the physical mask. Thus, the resulting impact report can include a defect criticality score for all defects.

Alternatively, if desired, the impact report may only include a defect criticality score above a certain value. This custom impact report can be provided to the bitmap editor and subsequently to the mask repair tool. Therefore, the inspection system may include complete and automated defect detection and correction processing, thereby greatly reducing the time required to analyze and repair (if appropriate) the mask.

Defect printability analysis also eliminates the need to evaluate OPC features separately from other features. If the OPC feature is to be printed (due to defects) (as determined by the simulated wafer image), defect analysis may point to this error as the CD change is determined. Thus, by removing any complex design rules regarding OPC features, the inspection system ensures a fast, reliable, and accurate way of identifying defects that adversely affect OPC features.

1 illustrates a conventional mask inspection process;

2 illustrates a known mask inspection process and system developed by Numerical Technologies, Inc.

3 illustrates a method of analyzing a defect by using a plurality of masks;

4A and 4B illustrate analyzing defects based on their location relative to various features in the mask,

5 illustrates a mask inspection process and system,

6 illustrates one method of generating the correct conversion factor;

7 illustrates another method of generating the correct conversion factor;

8A-8C illustrate various features of a computer-implemented program coupled with a defect printability analysis generator,

9A and 9B illustrate portions of a physical mask and a reference mask, respectively,

10A (1-3) illustrate a simulated wafer image of the defect free area of the physical mask in FIG. 9A, for three exposures,

10B (1-3) illustrate a simulated wafer image of the defect free area of the reference mask in FIG. 9B, for three exposures,

11A (1-3) illustrate a simulated wafer image of the defect area of the physical mask in FIG. 9A, for three exposures,

11B (1-3) illustrate a simulated wafer image of the defect area of the reference mask in FIG. 9B, for three exposures,

12A illustrates a mask that includes a feature and a defect proximate the feature,

12B illustrates a graph of feature size versus defocus, for the feature of FIG. 12A,

12C illustrates a graph of exposure deviation versus defocus, common processing window graph, for the features of FIG. 12A;

12D illustrates a graph of exposure latitude vs. depth of focus for the features of FIG. 12A;

13A illustrates a mask that includes a feature and a defect proximate the feature, where the defect is greater than the defect of FIG. 12A,

13B illustrates a graph of feature size versus defocus, for the features of FIG. 13A,

FIG. 13C illustrates a graph of exposure deviation versus defocus, for the features of FIG. 13A;

FIG. 13D illustrates a graph of exposure latitude vs. depth of focus for the features of FIG. 13A;

14A illustrates a mask that includes a feature having defects integrally formed therein;

14B illustrates a graph of feature size versus defocus, for the features of FIG. 14A,

14C illustrates a graph of exposure deviation versus defocus, for the features of FIG. 14A, FIG.

14D illustrates a graph of exposure latitude versus focal depth, for the features of FIG. 14A;

15A illustrates a mask that includes a feature having a defect formed integrally therein, where the defect is larger than the defect of FIG. 14A,

15B illustrates a graph of feature size versus defocus, for the features of FIG. 15A,

15C illustrates a graph of exposure deviation versus defocus, for the features of FIG. 15A, FIG.

FIG. 15D illustrates a graph of exposure latitude versus focal depth, for the features of FIG. 15A, FIG.

16A illustrates a mask comprising a contact (or via or post),

16B illustrates a graph of feature size versus defocus, for the contacts of FIG. 16A, FIG.

FIG. 16C illustrates a graph of exposure deviation versus defocus, for the contacts of FIG. 16A

16D illustrates a graph of exposure latitude versus focal depth for the contacts of FIG. 16A, FIG.

17A illustrates a mask with contacts (or vias or posts) with significant critical dimension (CD) variations,

17B illustrates a graph of feature size versus defocus, for the contacts of FIG. 17A,

17C illustrates a graph of exposure deviation versus defocus, for the contacts of FIG. 17A,

17D illustrates a graph of exposure latitude versus focal depth, for the contacts of FIG. 17A,

18 illustrates a mask repair process and system,

19A illustrates a simplified layout showing lines with line edge roughness that may not exhibit critical dimensional variation;

19B illustrates a simplified layout in which line edge roughness is determined, and

20A and 20B illustrate a simplified layout in which corner rounding and / or symmetry have been determined.

preface

Depending on the inspection system / process, all irregularities, ie potential defects, are characterized by actual defects. In one embodiment, a materiality score is provided for each defect, thereby providing the customer with meaningful information to accurately assess the results of using or repairing the mask. The defect criticality score is calculated based on a number of factors related to the change in the critical dimension of the feature near the defect. In another embodiment, the processing window can be used to provide objective information related to mask defect printability. Certain aspects of the mask related to mask quality, such as line edge roughness and contact corner rounding, can also be quantified.

IC layout: identification of critical areas

3 illustrates a feature that facilitates identifying a critical region of an IC. Specifically, the simplified process 300 includes using at least one other mask to identify a defect in the mask and determine whether the defect is located in a critical region. For example, mask 301 represents a layer of polysilicon region 310 in the IC. Two defects 304 and 305 are identified on the polysilicon region 310. Note that the two defects are the same size. Mask 302 represents the diffusion region 311 of another layer in the IC.

Process 300 includes determining the location and size of defects related to features in various masks, such as masks 301 and 302. For example, defects 304 and 305 may be considered insignificant by conventional inspection apparatus when viewed alone with respect to polysilicon region 310 on mask 301, and conventional inspection apparatus may Generally, defects are determined by size. In contrast, the process 300 takes into account the location of the defects 304 and 305 relative to the diffusion region 311 provided on the mask 302 in addition to the size. Specifically, process 300 uses information from various masks to identify critical regions of the IC. The composite IC layout 303 identifies the overlap of the polysilicon region 310 and the diffusion region 311 as the critical region 306. As a key feature of the final IC, the critical region 306, or gate, requires a high degree of precision to ensure proper performance of the transistors in the final IC device. Thus, by analyzing multiple masks and features therein, defect 305 can be characterized as insignificant because it is small and in a non-critical region (eg, interconnect), while defect 305 Although it is small, it can be characterized as important because it is in the critical region (eg, gate) of the IC.

As will be detailed below, defects in the critical region generally have a higher defect criticality score than defects in the noncritical region.

CD variation: identification of defects and neighboring features

4A illustrates a simplified mask 400 showing various layers of polysilicon features in an IC. Mask 400 includes three defects 401, 402, 403 that can affect neighboring polysilicon features 404 and 405. In this example, assume that the defects 401, 402, 403 are the same size.

In general, defects have a greater impact in complex areas than in less complex areas. Thus, assuming that distance X is less than distance Y, defects located in the area defined by the feature away by distance X will have more printability effects than defects located in the area defined by the feature away by distance Y. Can be. However, these general rules have profound limitations.                 

Referring to FIG. 4B, each defect can be analyzed according to a position relative to neighboring features. For example, defect 401 is distanced from feature 405 by distance d1 (A), distance from feature 404 by distance d1 (B), and distance d1 (A) is substantially equal to distance d1 (B). Suppose it is the same as In addition, defect 403 is distanced from feature 405 by distance d3 (A), distance from feature 404 by distance d3 (B), and distance d3 (A) is substantially equal to distance d3 (B). Suppose it is the same. In this example, defect 401 will have a greater printability impact on mask 400 than defect 403. Therefore, general rules apply to defects 401 and 403.

However, the mask also includes a defect located a distance d2 (A) (ie, 0) away from the feature 405 and a distance d2 (B) away from the feature 404. In this case, defect 402 may have more printability impact on feature 405 than defect 401. Moreover, defect 402 will probably have a smaller printability impact on feature 404 than defect 403. Thus, a general rule limited to the spacing of features does not provide an accurate indication of printability impact.

One solution to this problem is to measure the distance (such as d1, d2, d3) from each defect (such as defects 401, 402, 403 respectively) to neighboring features. These distances, combined with the measurement of the size of the defects, may be factored into multiple design rules to provide printability impact. However, this analysis is very complicated to calculate, thus increasing the time required to provide meaningful information to the customer. Moreover, even if the size of the defect and the distance of the defect from the neighboring feature are known, the actual effect of the defect on the neighboring feature cannot be fully predicted by simple inspection of the mask.

Fault Printability Analysis

Therefore, according to one embodiment, a limited number of variables are analyzed. In one embodiment, this limited number of variables includes the critical dimension (CD) of the mask. In particular, any CD change in the feature that occurs due to the proximity of the defect can be determined. To analyze this CD change, the mask image can be simulated as described with reference to FIG. 5.

5 illustrates a process 500 for analyzing defect printability. In process 500, physical mask 501A and reference mask 501B are analyzed by inspection tool 502. In one embodiment, the reference mask 501B may be a physical mask having the same layout as the physical mask 501A but without any defects. In another embodiment, the reference mask 501B may be an image simulated from the layout of the physical mask 501A.

In one embodiment, the inspection tool 502 includes an image acquirer 503 to scan all or a portion of the physical mask 501A and a corresponding portion of the reference mask 501B. Image acquirer 503 may comprise a high precision imaging device, such as a high resolution optical microscope, a scanning electron microscope (SEM), a focal ion beam, an atomic force microscope, or a near-field optical microscope. The image acquirer 503 may also include an interface device for digitizing image information from the image device. In one embodiment, the interface device includes a CCD camera that generates a gray scale bit image representing the image.

The defect detection processor 504 compares the image from the physical mask 501A with the image from the reference mask 501B provided by the image acquirer 503, and identifies any defects in the physical mask 501A. In one embodiment, defect detection processor 504 includes a computer that executes a program of instructions for scanning mask 501. Once a defect is identified, the defect detection processor 504 may not only correspond to the corresponding area from the reference mask 501B, but also the image generator 505 to provide an image of the area surrounding the defect from the physical mask 501B and the defect. Send a signal to Image generator 505 also provides an image of the defect free areas from both masks 501. In one embodiment, image generator 505 may provide an image that includes both defect and defect free areas. To facilitate defect printability analysis described in detail below, the coordinates of these defect and defect free areas can be transmitted along with the generated area image data. Note the following: That is, if the reference mask 501B is provided as a simulated layout, and a complete image of the physical mask 501A is generated, the simulated layout file of the reference mask 501B is as indicated by line 506B. As such, it may be provided directly to the image generator 505.

In one embodiment, the inspection tool 502 sends the region image data from both the physical mask 501A and the reference mask 501B to the wafer image generator 509 as a real-time data feed as indicated by line 506D. to provide. In another embodiment, such data is digitized by digitization device 507, stored in storage 508, and subsequently provided to wafer image generator 509. Storage device 508 can store these digitized information in a format such as Windows BMP on any type of suitable media including computer hard disk drives, CDROMs, servers. In another embodiment of analyzing the physical mask 501A altogether, the scanned image (s) provided by the image acquirer 503 may be image generator 505 or line 506C as indicated by line 506A. May be provided to the digitizing device 507 as indicated by.

Wafer image generator 509 includes an input device 510 and an image simulator 511. Input device 510 is typically a digitized image grabber provided by hardware that reads the type of image data from inspection tool 502 and / or storage 508, namely Matrox ™, Meteor ™, or Pulsar ™. Include known hardware (for real-time data feeds). In one embodiment, image simulator 511 includes a computer-implemented program that runs Windows / DOS at 200 MHz on a suitable platform, such as a personal computer or workstation, having at least 64 MB of memory. The image simulator 511 receives image data from the input device 510 and generates a simulation of the image data for the image data, that is, the physical mask 501A and the reference mask 501B. These simulations are referred to herein as wafer image Phy (for physical mask) 517A and wafer image Ref (for reference mask) 517B.                 

In one embodiment, the simulator 511 further receives a plurality of lithographic conditions 512. These conditions include data specific to the lithographic conditions and system parameters to which the physical mask 501A has been exposed by the customer. Such data may include, for example, the wavelength (λ) of the illumination used in the system, the numerical aperture (NA) of the system, the coincidence value (σ) of the system, the type of illumination (eg off-axis or annular), It may include focus, exposure level, lens aberration, substrate state, and critical value (CD) of the design. In one embodiment, each parameter may comprise a range of values, thereby allowing the image simulator 511 to generate a plurality of simulations based on the range of lithographic conditions possible in different combinations. For example, they can be performed by Monte Carlo simulations with different types of distributions, such as Gaussian distributions. Thus, the wafer image (Phy) 517A and the wafer image (Ref) 517B are the physical mask 501A and the reference mask 501B (or it if the optical lithography exposure is performed under the same conditions as the lithographic conditions 512). May be used to represent a simulated image to be generated.

Conversion factor

For the above (or near) wavelength design, the design rules for the features used in the layout are generally scaled simultaneously by the same factor. If a rule is not scaled as fast as another rule, some modifications can be made to the database, which are generally performed within a relatively short time. In contrast, the manufacturing steps in the sub-wavelength manufacturing flow are highly non-linear. In particular, any mask error can be amplified in the printed pattern on the wafer, resulting in a negative impact on the final device performance.

Therefore, to improve the accuracy of the wafer image 517 in the sub-wavelength technique, the image simulator 511 may also receive the conversion factor 513 according to one embodiment. In one case, the conversion factor is called the mask error enhancement factor (MEEF).

If the conversion factor is "known", then the mask CD can be multiplied with the conversion factor. Currently, "known" conversion factors are generally theoretical estimates. However, these theoretical estimates can be inaccurate for many reasons. First, as recognized by the applicant, the conversion factor may change based on the features of the mask. For example, the conversion factor of an isolated line may be different from the conversion factor of a tightly packed line. Moreover, the conversion factor for a contact may be different than the conversion factor of an isolated or tightly packed line. Second, in addition to design issues on the mask, all aspects of the manufacturing process, including the stepper and photoresist, may affect the conversion factor for a particular feature in the mask, for example. Therefore, theoretical estimates that fail to account for design issues and process parameters are inherently inaccurate.

Alternatively, if the theoretical correction is incorrect, the actual wafer can be manufactured and the device CD can be measured on the wafer using the SEM to determine the conversion factor (s). However, this process generally involves measuring and printing dozens or even hundreds of mask features to measure and calculate the conversion factor (s). Therefore, such treatments are highly expensive and thus commercially impractical.

A cost-effective solution to the aforementioned problem can be provided. 6 illustrates one method 600 for generating the correct conversion factor. In method 600, a test pattern may be provided to the test mask at step 601. The test pattern may include isolated lines with variable widths, tightly packed lines with variable widths, and contacts of various sizes. At this point, a single wafer may be printed at step 602. Defect analysis, including any changes in the CD, may be noted for each test pattern on the wafer in step 603. From this information, the conversion factor for each feature can be accurately calculated at step 604. A limited number of additional wafers can also be printed using various processes to obtain conversion factors for these fabrication processes.

Wafers printed from the test mask may also contain shop-specific information that may affect the conversion factor. Specifically, process variation can vary from one shop to another, and generally fluctuates. This fluctuation results in a slight CD change on the wafer, which is commonly referred to as "bias" in the industry. As described above, by printing one wafer, or a limited number of wafers in a shop and using a test mask, the customer can verify the published bias of the shop or independently determine the bias of the shop.

Test masks with the test patterns described above provide the user with accurate conversion factors as well as shop bias, allowing the customer to accept unacceptable CD changes (generally in the design process, in mask correction operations (described further below). Or potentially by compensating for another shop).

7 illustrates another method 700 for generating the correct conversion factor. In method 700, a test pattern may be provided to a test mask at step 701. Similar to the method 600, this test pattern may also include isolated lines with variable width, tightly packed lines with variable width, and contacts of various sizes. At this point, the wafer image from the test mask can be simulated in step 702 using image simulator 511 (FIG. 5). Defect analysis, including any changes in the CD (described in detail below), can be noted for each test pattern on the simulated wafer image at step 703. From this information, the conversion factor for each feature can be accurately calculated at step 704. Any number of additional wafers may be simulated using various processes (eg, lithographic conditions 512) to obtain conversion factors for these fabrication processes. In addition, note the following: That is, as described with reference to FIG. 6, shop bias may also be included in the lithographic condition 512, thus further improving the accuracy of the conversion factor generated by this embodiment.

The method 700 is highly cost-effective because it eliminates the cost associated with the printed wafer as well as the time required for the manufacture of the wafer. Moreover, because of the simulation environment, the method 700 provides tremendous flexibility in optimizing system parameters prior to actual fabrication.

Image simulation

Image simulator 511 approximates the process of optical lithography by using a simplified version of the Hopkins model as applied to an integrated circuit pattern. In this simplified version, the Hopkins model is considered as a plurality of lowpass filters applied to the input data. Output images from these lowpass filters are added to generate a simulated image (ie, simulated wafer (Phy) image 517A and simulated wafer (Ref) image 517B). Additional information relating to the Hopkins model as used by the image simulator 511 is provided in US Pat. No. 09 / 130,996, and thus will not be described in detail herein.

Calculation of defect criticality score

Defect printability analysis generator 515 receives simulated wafer image 517 from image simulator 511. Generator 515 includes a computer-implemented program that runs Windows / DOS at 200 MHz on a suitable platform, such as a personal computer or workstation, having at least 64 MB of memory. In one embodiment, image simulator 511 and generator 515 run on the same platform.

8A-8C illustrate various features of a computer-implemented program coupled with generator 515. 8A shows that the method 800 for generating a defect criticality score includes a preprocessing step 810, a two-dimensional analysis step 820, a defect analysis step 830, and a critical region indicating step 840.

In preprocessing step 810, the simulated wafer (Phy) image 517A and the simulated wafer (Ref) image 517B are aligned. The alignment is provided by the image generator 505 to the defect pattern (assuming that the simulated image 517 includes both defect and defect regions) or to the input device 510 ), The image simulator 511, sent to the final defect printability analysis 515). When these patterns / areas are aligned, the features provided on the simulated image 517 are also aligned.

After the alignment, two-dimensional analysis step 820 may be processed. Referring specifically to FIG. 8B, which further illustrates two-dimensional analysis step 820, defects on the simulated wafer (Phy) image 517A are identified in step 821. Thereafter, the corresponding area of the simulated wafer Ref image 517B is identified in step 822. Note that the coordinates provided by the image generator 505 can be used for the identification steps 821, 822. In step 823, any features that are close to defects in the simulated wafer Phy image 517A (here also referred to as neighboring features) are identified. Finally, at step 824, the corresponding feature (s) on the simulated wafer Ref image 517B may be identified.

The term "proximity" may refer to any feature that causes a change in CD as a result of the proximity of a defect. However, in one simple implementation, any feature within a predetermined distance of the defect can be identified as a neighbor feature. In another embodiment, both the size of the defect (which may be determined at step 821) and the distance of the defect from the neighboring feature (which may be determined at step 823) are compared with the measurements in the design rules table. The design rules table may identify the maximum distance from the defect for each defect size (or size range), where if the feature is located within the maximum distance from the defect, the feature is characterized as a neighbor feature.

After the two-dimensional analysis, defect analysis step 830 is processed. In defect analysis, the defect area is analyzed to calculate the mean CD deviation (ACD) (described in detail below), and the defect area is analyzed to calculate relative CD deviation (RCD) (described in detail below). . Note that the calculation of ACD and RCD can be done in any order. 8C details the defect analysis step 830 in further detail. Specifically, at step 831, CDs for one or more features in the defect area on the simulated wafer image 517 and any in the defect area on the simulated wafer image 517. The CD for the neighboring feature of is measured.

In order to determine the ACD, the CD of the defect free feature on the simulated wafer image (physical mask) 517A firstly corresponds to the CD of the corresponding defect free feature on the simulated wafer image (reference mask) 517B. Is deducted from This difference is then divided by the CD of the same defect free feature on the simulated wafer image (reference) 517B. In order to improve the accuracy of the ACD calculation, multiple features can be analyzed. Specifically, N ACDs can be added and then divided by N, where N is an integer equal to or greater than 1. For example, if two features are analyzed, the ACD can be calculated with the following equation: [(CD (R1)-CD (P1)) / CD (R1) + [(CD (R2)-CD ( P2)) / CD (R2)] / 2), where R represents a simulated wafer image of the reference mask and P represents a simulated wafer image of the physical mask. Note that the ACD can be determined for different intact features or for the same intact features. For example, in one embodiment, a typical gate can be cut every 2 nm (parallel slices of FET channels) across the gate width. Note the following: That is, CD estimation can be performed using standard mask inspection equipment provided by KLA-Tencor, Applied Materials, LaserTech, or any other reticle inspection / metrology tool vendor.

In one embodiment, different exposures may be used to provide a plurality of ACDs for each feature. Note the following: That is, the exposure used can range from values that deviate from the exposure level to be used in the actual manufacturing process, thus providing users with valuable information about the worst outcome. It is further noted that such exposure conditions are generally included in lithographic conditions 512 (FIG. 5) that can be simulated. Therefore, referring again to FIG. 8C, the ACD of each exposure can be calculated at step 832.

In step 833, the relative CD deviation (RCD) is calculated for each neighboring feature (as identified in the detection area of the simulated image) identified in each exposure. For example, for each exposure, a CD of identified neighboring features 904 (R) in the detection area 901 (R) of the simulated wafer image 901 (R) is simulated wafer. It is subtracted from the CD of the same feature (in this case 904 (P)) in the detection area 901 (P) in the image 901 (P). This difference is then divided by the CD of the identified neighboring feature in the simulated wafer image 901 (R) (ie, (CD (P)-CD (R)) / CD (R)). Finally, the maximum RCD (MCD) of the identified neighboring features may be determined for each exposure at step 834.

As described with reference to FIG. 3, features in the critical region, such as gates, require high precision precision to ensure proper performance of the transistors in the final IC device. Thus, by analyzing a plurality of masks and the features therein, a defect can be characterized as insignificant because it is small and in a non-critical region (i.e., interconnect), while it is small but critical region of the IC (e.g., For example, it can be characterized as important because it is at the gate.

Referring to FIG. 5, the defect printability analysis generator 515 also receives information from the critical area identification generator 514. Critical region identification generator 514 may include any standard pattern recognition tool (both hardware and software) for analyzing a physical mask, such as physical mask 501A, used to fabricate an IC. Regardless of the specific tool used, the critical area identification generator 514 provides the defect printability analysis generator 515 with information identifying the area of each mask indicated by the critical area. With this information, the defect printability analysis generator 515 can determine at step 840 whether the defect is within the critical area (FIG. 8A).

Defects in the critical region will generally have a lower tolerance for relative CD changes. In one embodiment, the tolerance for CD change (TCD) may be provided by a lookup table. This lookup table can include values determined by the mask quality control engineer based on experience and various mask specifications. For example, the critical region may have a TCD range between about 3% and 5%, while a noncritical region with very rare features may have a TCD range between about 10% and 15%. In one embodiment, critical area ID generator 514 may include such a lookup table.

Equation 1 provides an example calculation to determine the defect criticality score. Note the following: That is, Equation 1 includes the aforementioned ACD, MCD, and TCD variables, and further includes a variable i indicating a specific exposure and a variable N indicating the total number of exposures analyzed.

9 illustrates a typical portion 900 (P) from a physical mask. Portions 900 (P) include defect regions 901 (P) and defect regions 902 (P). In a non-numerical manner, FIG. 9B illustrates a portion 900 (R) from a reference mask that corresponds to portion 900 (P). Portions 900 (R) include defect regions 901 (R) and defect regions 902 (R).

By using coordinates or defect free patterns in step 810, the corresponding positions of these portions / regions in the simulated wafer image can be aligned. Specifically, for example, simulated wafer images of the defect free regions 902 (P) and 902 (R) can be aligned. In a non-numerical manner, the simulated wafer images of the defect regions 901 (P) and 901 (R) can be aligned. Once regions 901 and 902 are aligned, the features in the simulated wafer image are also aligned. Thus, for example, the features 904 (P) and 905 (P) of the defect region 901 (P) are the features of the defect region 901 (R) in the preprocessing step 810. (904) (R) and (905) (R).

In two-dimensional analysis step 820, defects in the simulated wafer image of mask portion 901 (P) are identified. In this example, an arrow points to the defect 903 in the defect area 901 (P). Features 904 (P) and 905 (P) are then identified as neighboring features that may be affected by defect 903. Finally, any corresponding feature in the simulated wafer image of mask portion 901 (R) may be identified. In this example, features 904 (R) and 905 (R) are identified.

At this point, defect analysis step 830 may be processed. 10 (A (1) to A (3) and B (1) to B (3)) and FIG. 11 (A (1) to A (3) and B (1) to B (3)) are parts ( Illustrates the application of defect criticality score calculation for simulated wafer images of (900) (P) and (900) (R). In order to calculate the average CD deviation for intact features, a number of features are generally measured. For example, FIGS. 10A (1-3) show simulated wafer images from the defect free areas 902 (P) of the physical mask 900 (P) for three exposures. Lines 1001 (P) through 1006 (P) represent cuts made for the two defect-free features of the simulated wafer image at three exposures. Specifically, lines 1001 (P) and 1002 (P) represent cuts made for two features in the first exposure, and lines 1003 (P) and 1004 (P) represent the cuts in the second exposure. Represent the same cuts made for the same feature, and lines 1005 (P) and 1006 (P) represent the same cuts made for the same feature in the third exposure.

In a similar manner, FIGS. 10B (1-3) show simulated wafer images from the defect free areas 902 (R) of the reference mask 900 (R) for the same three exposures. Lines 1001 (R) through 1006 (R) represent cuts made for the two defect-free features of the simulated wafer image at three exposures, where these three cuts are cuts 1001 (P) through 1006 ( Corresponds to P)). Thus, lines 1001 (R) and 1002 (R) represent cuts made for two features in the first exposure, and lines 1003 (R) and 1004 (R) represent the same features in the second exposure. Represent the same cuts made for, and lines 1005 (R) and 1006 (R) represent the same cuts made for the same feature at the third exposure.

Each cut line 1001 (P) to 1006 (P) and 1001 (R) to 1006 (R) provides a combined CD. Therefore, for ease of reference, lines 1001 (P) to 1006 (P) and 1001 (R) to 1006 (R) are subsequently referred to as CDs (1001 (M) to 1006 (M) and 1001 (R) to 1006). (R)).

For the first exposure shown in FIGS. 10A (1) and 10B (1), the average CD deviation can be calculated as follows:

ACD (1) = [(1001 (R)-1001 (P)) / 1001 (R) + (1002 (R)-1002 (P)) / 1002 (R)] / 2

In one embodiment, the actual measurements of the CDs 1001 (R), 1001 (P), 1002 (R), 1002 (P) are 266 nm, 266 nm, 322 nm and 294 nm, respectively. Substituting these values into the equation for ACD (1) yields approximately 0.043 nm.

For the second exposure shown in FIGS. 10A (2) and 10B (2), the average CD deviation can be calculated in a similar manner:

ACD (2) = [(1003 (R)-1003 (P)) / 1003 (R) + (1004 (R)-1004 (P)) / 1004 (R)] / 2

In one embodiment, the actual measurements of the CDs 1003 (R), 1003 (P), 1004 (R), 1004 (P) are 266 nm, 266 nm, 294 nm and 294 nm, respectively. Substituting these values into the equation for ACD (2) yields approximately 0.0 nm.

Finally, for the second exposure shown in FIGS. 10A (3) and 10B (3), the average CD deviation can be calculated in the same way:

ACD (3) = [(1005 (R)-1005 (P)) / 1005 (R) + (1006 (R)-1006 (P)) / 1006 (R)] / 2

In one embodiment, the actual measurements of the CDs 1005 (R), 1005 (P), 1006 (R), 1006 (P) are 252nm, 238nm, 294nm and 294nm, respectively. Substituting these values into the equation for ACD (3) yields approximately 0.028 nm.

In defect analysis, relative CD deviation (RCD) is also calculated for neighboring features in the defect area for each exposure level. 11A (1-3) illustrate simulated wafer images for features 904 (P) and 905 (P) in defect area 901 (P) for three exposures. Lines 1101 (P) through 1106 (P) represent cuts made for the two features of the simulated wafer image at three exposures. Specifically, lines 1101 (P) and 1102 (P) represent cuts made for features 904 (P) and 905 (P) in the first exposure, and lines 1103 (P) and 1104 (P). P)) represents the cuts made for features 904 (P) and 905 (P) in the second exposure, and lines 1105 (P) and 1106 (P) represent features 904 (P in the third exposure). And 905 (P)).

Similarly, Figures 11B (1-3) illustrate simulated wafer images for features 904 (R) and 905 (R) for the same three exposures. Lines 1101 (R) to 1106 (R) represent cuts made for two features of the simulated wafer image at three exposures. Specifically, lines 1101 (R) and 1102 (R) represent cuts made for features 904 (R) and 905 (R) in the first exposure, and lines 1103 (R) and 1104 (R). )) Represents cuts made for features 904 (R) and 905 (R) in the second exposure, and lines 1105 (R) and 1106 (R) represent features 904 (R) in the third exposure. And cuts made for 905 (R)).

Each line 1101 (P) through 1106 (P) and 1101 (R) through 1106 (R) provides a combined CD. Therefore, for ease of reference, lines 1101 (P) to 1106 (P) and 1101 (R) to 1106 (R) are then referred to as CDs 1101 (P) to 1106 (P) and 1101 (R) to 1106. (R)).

For the first exposure shown in FIGS. 11A (1) and 11B (1), the relative CD deviation (RCD) can be calculated for the feature 904 as follows:

RCD (1 (904)) = (1101 (P)-1101 (R)) / 1101 (R)

In one embodiment, the actual measurements of CDs 1101 (R) and 1101 (P) are 266 nm and 364 nm, respectively. Substituting these equations for RCD 1 (904) yields approximately 0.368 nm.                 

In a similar manner, for the first exposure shown in FIGS. 11A (1) and 11B (1), the relative, maximum CD deviation (RCD) change can be calculated for the feature 905 as follows:

RCD (1 (905)) = (1102 (P)-1102 (R)) / 1102 (R)

In one embodiment, the actual measurements of CDs 1102 (R) and 1102 (P) are 252 nm and 322 nm, respectively. Substituting these equations for RCD (1 (905)), the calculated value is about 0.278 nm.

The RCDs for features 904 and 905 for the second and third exposures can be calculated in a similar manner as shown below.

RCD (2 (904)) = (1103 (P)-1103 (R)) / 1103 (R)

RCD (2 (905)) = (1104 (P)-1104 (R)) / 1104 (R)

RCD (3 (904)) = (1105 (P)-1105 (R)) / 1105 (R)

RCD (3 (905)) = (1106 (P)-1106 (R)) / 1106 (R)

In one embodiment, actual measurements of CDs 1103 (R), 1103 (P), 1104 (R), 1104 (P), 1105 (R), 1105 (P), 1106 (R), 1106 (P) Are 238 nm, 350 nm, 252 nm, 294 nm, 224 nm and 280 nm, respectively. Substituting these values into the formula for RCD (2 (904)), RCD (2 (905)), RCD (3 (904)), RCD (3 (905)), the calculated values are about 0.471nm, 0.167nm, 0.353 respectively. nm, 0.250 nm.

In order to determine the maximum CD deviation (MCD) for each exposure, the largest RCD value is selected. Thus, the maximum CD deviation (MCD (1)) for the first exposure is 0.368 nm (0.368> 0.278), MCD (2) 0.471 nm (0.471> 0.167) and MCD (3) 0.353 nm (0.353> 0.250) to be.

The defect criticality score (DSS) can be calculated using Equation 1. In the example given, N = 3 because three exposures were analyzed.

Substituting these values into Equation 1 results in:

Thus, based on three exposures,

DSS = (3/3) [(MCD (1)-(ACD (1) / 3)) / TCD + (MCD (2)-(ACD (2) / 3)) / TCD + (MCD (3)- (ACD (3) / 3)) / TCD]

Substituting the values calculated above into the MCD and ACD for the three exposures, we get:

DSS = [(0.368- (0.043 / 3)) / 0.1 + (0.473- (0/3)) / 0.1 + (0.353- (0.028 / 3)) / 0.1

DSS = 3.54 + 4.73 + 3.44

Therefore, defect 903 (see Figure 9A) has a DSS of about 11.71.

Defect printability analysis generator 515 (FIG. 5) may output a defect criticality score (DSS) (in one embodiment, scale from 1 to 10) in impact report 516. This impact report 516 may be used to reduce human error in defect printability analysis. For example, a DSS score of 5 would probably indicate that the printed feature will have significant performance issues. However, it will also point out that repair of the physical mask is possible. On the other hand, a DSS score of 7 and above will indicate that the manufacture of the physical mask is recommended as well as performance issues. For example, in one embodiment, a DSS less than 3 means that the CD change due to a defect is within a certain CD tolerance, while a DSS between 3 and 6 indicates that the CD change due to a defect is less than a particular CD tolerance. Although larger, it means that the CD change does not cause serious defects on the wafer (such as open or bridged), while DSS greater than 6 indicates that the CD change due to the defect causes serious defects on the wafer. It means to bring about. Thus, by providing a numerical result with a combined meaning for each number, the engineer can proceed efficiently and without error, for example, to repair the physical mask or to rebuild the physical mask.

Processing window

Defect printability can also be evaluated by using various processing windows. The processing window can be derived from any graph known by those skilled in the art. In general terms, the processing window of a feature is the amount of variation in processing that can be tolerated while maintaining the critical dimension (CD) of the feature within a certain range of the target CD.

One known process variation is the focusing setting of the projection tool, ie the stepper. The focus can vary significantly the resist profile (CD, sidewall angle, resist thickness), and thus can be critical in providing an acceptable lithography process.

Because of the effects of focus and exposure, these variables generally change simultaneously in the focus-exposure matrix. The processing window can be derived from such a matrix. Focus and exposure values located within the processing window produce resist features within tolerance, for example CD, while focus and exposure values located outside the processing window produce resist features outside tolerances. Thus, as will be described in detail below, the processing window can provide an objective means for determining the materiality and printability of the defect.

For example, FIG. 12A illustrates a mask having features 1204 and defects 1203. As described above, the defect 1203 will affect the width of the peer 1204. Specifically, the width of the feature 1204 in the cut line 1201 will be larger than the width in the Kerr line 1202.

12B shows a graph of feature size (in nanometers) vs. defocus (in nanometers). In this figure, the thick horizontal line indicates that the target CD is 200 nm, while the other horizontal line indicates the +/- 10% error of this target CD. Curves 1211 and 1212 expose (or simulate exposure) masks that contain defects 1223 and are printed at cut lines 1201 and 1202 at various defocus levels (in this case, -500 nm to 500 nm). Generated by analyzing the CD of the feature. Curves 1211 and 1212 represent CD analysis at cut lines 1201 and 1202, respectively.

Logically, each feature size on curve 1212 has a corresponding larger feature size on curve 1211. For example, at -300 nm defocus, the feature size at cut line 1202 (see curve 1212) is about 150 nm, while the feature at cut line 1201 (see curve 1211). The size is about 170 nm. Note the following: That is, the acceptable defocus window for both curves, ie the horizontal CD +/- 10%, is between about -208 nm and 208 nm.                 

12C illustrates a graph of percent exposure deviation versus defocus (in nanometers). In this figure, curve 1221 represents the upper and lower limits of exposure deviation for cut line 1201 for various defocus levels, and curve 1222 is the exposure deviation for cut line 1202 for various defocus levels. The upper and lower limits are shown. The largest possible rectangle that fits within the overlap of these two regions defines a common processing window 1223. In this embodiment, the common processing window 1223 indicates that the defocus can vary between about -150 nm and 150 nm, while the exposure deviation varies between about -10% and 10% (tolerance for line CD). Can be maintained within).

12D is a graph of exposure latitude (%) versus depth of focus (DOF) (in nanometers), where exposure latitude refers to the amount of exposure variation, and DOF refers to the amount of focus variation. In this figure, curve 1231 represents the upper and lower limits of exposure latitude for cut line 1201 for various DOF, and curve 1232 represents the upper and lower limits of exposure latitude for cut line 1202 for various DOF. Indicates. Note that curve 1231 and curve 1232 share the same lower limit. The largest possible rectangle that fits below the common lower limit defines a common processing window 1233. In this embodiment, the common processing window 1233 indicates that the DOF can vary between about 0 nm and 300 nm, while the exposure latitude varies between about 0% and 19% (again, tolerates line CD). Can be maintained within).

Note that the information provided by the processing window 1233 can be derived by the processing window 1223. Specifically, the DOF range is the same as the entire range of defocus, and the exposure latitude range is the same as the entire range of exposure deviation.

13A illustrates a mask having features 1304 and defects 1303. Although feature 1304 is the same size as feature 1204, defect 1303 is much larger than defect 1203. Thus, the width of the feature 1304 printed at the cut line 1301 will be wider than the width of the feature 1304 printed at the cut line 1302. Moreover, as will be described below, the defect 1303 will greatly reduce the processing window compared to the defect 1203.

13B shows a graph of feature size (in nanometers) vs. defocus (in nanometers). Again, the thick horizontal line indicates that the target CD is 200 nm, while the other horizontal line indicates a +/- 10% error of this target CD. Curves 1311 and 1312 expose (or simulate) exposure of the mask containing defects 1323 and are printed at cut lines 1301 and 1302 at various defocus levels (in this case, -500 nm to 500 nm). Generated by analyzing the CD of the feature. Curves 1311 and 1312 represent CD analysis at cut lines 1301 and 1302, respectively.

As mentioned previously, each feature size on curve 1312 has a corresponding larger feature size on curve 1311. For example, at -300 nm defocus, the feature size at cut line 1302 (see curve 1312) is about 150 nm, while the feature at cut line 1301 (see curve 1311). The size is about 185 nm. Note the following: That is, the acceptable defocus window for both curves, ie between horizontal CD +/- 10%, is between about -100 nm and -100 nm as well as between about 100 nm and 208 nm.                 

13C illustrates a graph of percent exposure deviation versus defocus (in nanometers). In this figure, curve 1321 represents the upper and lower limits of the exposure variation for CD corresponding to cut line 1301 for various defocus levels, and curve 1322 shows cut line 1302 for various defocus levels. The upper and lower limits of the exposure deviation for the corresponding CD are shown. The largest possible rectangle that fits within the overlap of these two regions defines a common processing window 1323. In this embodiment, the common processing window 1323 indicates that the defocus can vary between about -100 nm and 100 nm, while the exposure deviation varies between about 2% and 15% (line CD within tolerance You can do that).

13D is a graph of exposure latitude (%) vs. DOF (in nanometers). In this figure, curve 1331 represents the upper and lower limits of exposure latitude for CD corresponding to cut line 1301 for various DOF, and curve 1332 corresponds to CD corresponding to cut line 1302 for various DOF. The upper and lower limits of exposure latitude to. Note that curve 1331 and curve 1332 share substantially the same upper and lower limits. The largest possible rectangle that fits below the common lower limit defines a common processing window 1333. In this embodiment, the common processing window 1333 indicates that the DOF can vary between about 0 nm and 200 nm, while the exposure latitude varies between about 0% and 12% (again, tolerance for line CD). Can be maintained within).

Note that the processing window 1223/1233 is significantly larger than the processing window 1323/1333. As can be seen in this example, larger defect sizes reduce the processing window. Therefore, various processing windows can be compared to determine defect printability. In particular, the processing window for a defect feature may be compared to one or more processing windows of a feature having defects proximate such feature (s). In a typical embodiment, the customer can set an acceptable range of deviation from the processing window for the defect free feature.

The processing described above is equally applicable to defects that form part of a feature. For example, FIG. 14A illustrates a mask having a defect 1403 formed integrally with feature 1404. Defect 1403 may affect the width of feature 1404. Specifically, the width of feature 1404 in cut line 1401 will be larger than the width in cut line 1402.

14B shows a graph of feature size (in nanometers) vs. defocus (in nanometers). In this figure, the thick horizontal line indicates that the target CD is 200 nm, while the other horizontal line indicates the +/- 10% error of this target CD. Curves 1411 and 1412 expose (or simulate) exposure of the mask containing defects 1403 and are printed at cut lines 1401 and 1402 at various defocus levels (in this case, -500 nm to 500 nm). Generated by analyzing the CD of the feature. Curves 1411 and 1412 represent CD analysis at cut lines 1401 and 1402, respectively. In this example, an energy of 3.9 mJ / cm 2 for development is assumed.

Logically, each feature size on curve 1412 has a corresponding larger feature size on curve 1411. For example, at -300 nm defocus, the feature size at cut line 1402 (see curve 1412) is about 150 nm, while at cut line 1401 (see curve 1411). Feature size is about 165 nm. Note the following: In other words, the acceptable defocus window for both curves, ie between horizontal CD +/- 10%, is between about -208 nm and 208 nm.

14C illustrates a graph of percent exposure deviation versus defocus (in nanometers). In this figure, curve 1421 represents the upper and lower limits of exposure deviation for cut line 1401 for various defocus levels, and curve 1422 shows the exposure deviation for cut line 1402 for various defocus levels. The upper and lower limits are shown. The largest possible rectangle that fits within the overlap of these two regions defines a common processing window 1423. In this embodiment, the common processing window 1423 indicates that the defocus can vary between about -150 nm and 150 nm, while the exposure deviation varies between about -5% and 9% (tolerance for line CD). Can be maintained within).

14D is a graph of exposure latitude (%) versus DOF (in nanometers). In this figure, curve 1431 represents the upper and lower limits of exposure latitude for cut lines 1401 for various DOF, and curve 1432 represents the upper and lower limits of exposure latitude for cut line 1402 for various DOF. Indicates. Note that curve 1431 and curve 1432 share the same lower limit. The largest possible rectangle that fits below the common lower limit defines a common processing window 1433. In this embodiment, the common processing window 1433 indicates that the DOF can vary between about 0 nm and 300 nm, while the exposure latitude varies between about 0% and 14% (again, tolerance for line CD). Can be maintained within).

As mentioned previously, the information provided by the processing window 1433 may be derived by the processing window 1423. Specifically, the DOF range is the same as the entire range of defocus, and the range of exposure latitude is the same as the entire range of exposure deviation.

15A illustrates a mask with feature 1504 and defect 1503. Although feature 1504 is the same size as feature 1404, defect 1503 is significantly larger than defect 1403. Thus, the width of the feature 1504 printed at the cut line 1501 will be wider than the width of the feature 1404 printed at the cut line 1401. Moreover, as will be described in detail below, defect 1503 will greatly reduce the processing window compared to defect 1403.

15B shows a graph of feature size (in nanometers) vs. defocus (in nanometers). Again, the thick horizontal line indicates that the target CD is 200 nm, while the other horizontal line indicates +/- 10% error of this target CD. Curves 1511 and 1512 expose (or simulate) exposure of the mask containing defects 1503 and are printed at cut lines 1501 and 1502 at various defocus levels (in this case, -500 nm to 500 nm). Generated by analyzing the CD of the feature. Curves 1511 and 1512 represent CD analysis at cut lines 1501 and 1502, respectively.

As mentioned previously, each feature size on curve 1512 has a corresponding larger feature size on curve 1511. For example, at -300 nm defocus, the feature size at cut line 1502 (see curve 1512) is about 150 nm, while the feature at cut line 1501 (see curve 1511). The size is about 198 nm. Note the following: That is, an acceptable defocus window for both curves, ie between horizontal CD +/- 10%, is no longer obtainable.

15C illustrates a graph of percent exposure deviation versus defocus (in nanometers). In this figure, curve 1521 represents the upper and lower limits of the exposure deviation for CD corresponding to cut line 1501 for various defocus levels, and curve 1522 represents cut line 1502 for various defocus levels. The upper and lower limits of the exposure deviation for the corresponding CD are shown. In this case, the two regions defined by curve 1521 and curve 1522 do not overlap. Therefore, there is no common processing window. Thus, defect 1503 will effectively prevent feature 1504 from printing within tolerances.

15D is a graph of exposure latitude (%) versus DOF (in nanometers). In this figure, curve 1531 represents the upper limit of exposure latitude for CD corresponding to cut line 1501 for various DOF, and curve 1532 corresponds to CD corresponding to cut line 1502 for various DOF. The upper limit of exposure latitude to. Note that curve 1531 and curve 1532 share no lower limit. Thus, there is no common processing window, thus confirming the information derived from FIG. 15D.

12-15, the use of a processing window to determine defect printability has been applied to the lines. However, this use of the processing window is also applicable to the printability of contacts and vias. 16A-16D are graphs of defect-free contacts 1601 on a mask, graph of feature size versus defocus 1602, graph of exposure deviation versus defocus 1603 (and resulting processing window), graph of exposure latitude versus DOF 1604 (and its resulting processing window), respectively.                 

In contrast, FIGS. 17A-D show the contact 1701 on the mask, the graph of feature size versus defocus 1702, the graph of exposure deviation versus defocus 1703 (and resulting processing window), the exposure latitude versus DOF Each illustrates graph 1704 (and its resulting processing window). Note the following: That is, contact 1701 has a significant profound CD variation, and thus can be considered as a normal defective contact under prior art analysis.

However, compared to the process windows of FIGS. 16C and 16D, the process window of FIGS. 17C and 17D has one process window relatively similar to that of the contact 1601 although the process window exhibits significant CD variation. . Specifically, with reference to FIGS. 17D and 16D, both contacts 1701 and 1601 have a common focal depth between 0 and 600 nm, with substantially similar exposure latitude, ie, 0 for contact 1701. And between 58% and between 0 and 40% for contact 1601. However, although contacts 1701 and contacts 1601 have the same defocus (ie, between -300 nm and 300 nm), these contacts have significantly different percentage exposure variations. Specifically, contact 1701 has an exposure deviation between about 22% and 80%, and contact 1601 has an exposure deviation between about -3% and 37%. As a result, a common processing window may be present for both contacts 1701 and contacts 1601 although small.

Thus, analyzing the processing window associated with a feature can provide an objective means of determining the printability of that feature based on defects. For example, defect materiality can be derived from the amount of overlap between two processing windows (ie, common processing window), while one processing window is extracted from the joining cut line, while another processing window is the reference cut. Is extracted from the line. Specifically, according to this embodiment, the defect printability analysis generator 515 may determine a common processing window for the features provided in the masks 501A and 501B and provide this information to the impact report 516. .

Repair of physical mask

18 illustrates one process that may be used to analyze repairs that may be made for a physical mask. As shown in FIG. 18, when using the impact report 516 (or portions thereof), the bitmap editor 1801 may use a physical mask (eg, to remove or significantly reduce the impact of one or more defects). For example, a possible correction may be indicated to the physical mask 501 A. The bitmap editor 1801 may then output a simulated mask 1802 that includes these corrections. The mask 1802 is a possible, repaired version of the physical mask Note that the bitmap editor 1801 can use the same tools or separate as the wafer image generator 509.

Simulated mask 1802 can be inspected by inspection tool 502 and generates new impact reports and new simulated wafer images (not shown) indicating the success of possible corrections provided to mask 1802. To be used by the wafer image generator 509. If the correction is acceptable, the bitmap editor 1801 can provide the correction information directly to the mask repair tool 1803 for repair of the physical mask. If the customer wants further optimization and analysis of different parameters, the bitmap editor 1801 indicates that the above-mentioned processing is considered to be within acceptable range or that the desired result cannot be obtained by repairing the physical mask. Can be repeated.

In one embodiment, the bitmap editor 1801 may also point to an optimized mask writing strategy. For example, laser tools can be used for opaque defects (eg removal of chromium defects), while focused ion beam tools can be used for transparent defects (eg precipitation of chromium). Note that laser and focused ion beam tools can also be used for precipitation and removal, respectively. In general, a focused ion beam tool provides more precision than a laser tool. However, focused ion beam tools are generally slower than laser tools. Bitmap editor 1801 may receive input (not shown) indicating the customer's time or cost constraints, thereby allowing bitmap editor 1801 to optimize repair processing based on these customer parameters. .

In another embodiment of the present invention, bitmap editor 1801 may be used to provide information to a wafer repair tool (not shown). In detail, the bitmap editor 1801 may include a program for comparing the effect of mask repair with the effect of wafer repair. In one embodiment, the program may convert a non-optical (eg SEM, focused ion beam) image into an optical image for further analysis.                 

Batch processing

Importantly, defect printability analysis can be performed on individual defects or on multiple defects. In one embodiment, the inspection tool 502 and the wafer image generator 509 may automatically output the results for all defects found in the physical mask 501A. Thus, impact report 516 may include a defect criticality score for all defects.

Alternatively, if desired, impact report 516 may include only a defect criticality score above a predetermined value (eg, DSS 5 or greater). This custom impact report can be provided to the bitmap editor 1801 (and subsequently to the mask repair tool 1803). Therefore, a complete and automated defect detection and correction process can be provided, thereby greatly reducing the time required to analyze and repair (if appropriate) the mask.

Consider OPC

Defect printability analysis can also eliminate the need to evaluate OPC features independently from other features. For example, suppose That is, defects located proximate to the scattering rod do not affect the printing of the isolated isolation features. However, this defect can optically interact with the scattering rods, thus resulting in printing at least a portion of the scattering rods. As mentioned previously, OPC features, such as scattering bars, are sub-resolution features and should not be printed.

According to one embodiment, if the OPC feature is printed due to a defect (as determined by the simulated wafer image), defect analysis (step 830) may resolve this error as soon as the CD change is determined (step 831). Can point. Thus, by eliminating any complex design rules with respect to OPC features, this embodiment ensures a fast, reliable, and accurate way of identifying defects that adversely affect OPC features.

Mask quality issue

In addition to CD variations, other printability factors such as line edge roughness must also be considered for mask quality. However, the line edge roughness of the feature on the mask is not currently measured in a meaningful way.

19A illustrates a simplified simulated wafer image 1900 that includes two lines 1901 and 1902. Interestingly, lines with line edge roughness may not necessarily show CD variation. For example, since line 1902 has a substantially symmetrical line edge roughness, line 1902 may not have significant CD variations. However, both lines 1901 and 1902 should be characterized as exhibiting line edge roughness.

Referring to FIG. 19B, the edges of the simulated lines can be analyzed independently, thereby allowing the line edge roughness to be measured accurately. Specifically, when using line 1902 as an example, the centerline 1903 of line 1902 may be determined based on reference mask 501B (FIG. 5). A plurality of theoretical cuts are then made for line 1902 (indicated by line 1904). Each line 1904 includes two "ribs" from the centerline to the opposite edge of the line. For example, lip 1904R extends from centerline 1903 to the right edge of line 1902 and lip 1904L extends from centerline 1903 to the left edge of line 1902. Note that ribs 1904R and 1904L are the same as the CD of line 1902 when added.

As one feature of the invention, the length of the lip on each side of centerline 1903 may be measured independently. In this way, line edge roughness can be accurately determined for each edge of line 1902. In one embodiment, the defect printability analysis generator 515 can calculate the DSS of the LER by using the equations described in detail with reference to FIGS. 8A-8C but modifying these equations to replace the lip length instead of the CD. Since every line inevitably has some LER, the defect printability analysis generator 515 may include a lookup table that indicates the threshold for the LER. If unacceptable line edge roughness (LER) is detected, the defect printability analysis generator 515 may point the LER of the line 1902 to the "defect" described in the impact report 516. Thus, the LER can be repaired in a manner similar to that described above with reference to FIG.

Advantageously, the method using centerline granules can be applied to other features on the mask. For example, even the most complete contact on the mask can be printed on the wafer as a circle or close to a circle because of diffraction. Using high power electron beam (e-beam) lithography to print contacts on the wafer minimizes this diffraction. However, e-beam lithography is significantly more expensive and slower than industry-standard laser raster scans. Unfortunately, using raster scans results in corner rounding of many contacts, although not necessarily all of the contacts on the layout. Such corner rounding can be efficiently detected as described in detail below.

20A shows a line 2002, a lip 2003R (right), including a centerline 2001, ribs (2002TR (top right), 2002BR (bottom right), 2002TL (top left), 2002BL (bottom left); Illustrates a contact 2000 with a line 2003 including 2003L (left). In accordance with one aspect of the present invention, the plurality of theoretical, horizontal cuts made in contact 2000 are unevenly spaced, thus providing more data points for a particular element of features.

In this example, the corner rounding of the contact is particularly interesting. Therefore, the spacing of the cuts is modified to ensure a sufficient number of data points for analysis, especially at the corners of the contacts. Thus, in FIG. 20A, line 2002 has a spacing closer than line 2003. Corner rounding for contact 2000 can be determined by comparing the length of lip 2003L to the length of lip 2002TL for the upper left corner and the length of the lip 2002BL for the lower left corner. In a similar manner, the length of the lip 2003R can be compared with the length of the lip 2002TR for the upper right corner and the length of the lip 2003BR for the lower right corner. Note that there are several known ways of estimating corner rounding effects (eg, missing area, normal distance histogram, etc.).

In some cases, performance issues related to contacts may include whether symmetrical contact shapes are made consistent on the wafer. Advantageously, in addition to the line edge roughness, the symmetry of the contacts can also be determined. For example, the lengths of the lips 2002TL, 2003L, 2002BL may be compared with the lengths of the lips 2002TR, 2003R, 2002BR to determine the horizontal symmetry of the contact 2000 from the centerline 2001. The vertical symmetry of the contact 2000 can be determined by using a vertical cut as shown in FIG. 20B and following a similar process of lip comparison. The overall symmetry of the contact 2000 (ie, "square") is the selected combined horizontal lip (eg, the added length of one of the lip 2002TL and one of the lip 2002TR, ie CD). Can be determined by comparing with a vertical lip.

In one embodiment, the defect printability analysis generator 515 uses the equations described in detail with reference to FIGS. 8A-8C but can calculate the symmetry DSS by modifying these equations to replace the lip length instead of the CD. Since all contacts inevitably have some asymmetry, the defect printability analysis generator 515 may include a lookup table that indicates the threshold for asymmetry. If unacceptable symmetry is detected, defect printability analysis generator 515 may point to that contact / via of line 1902 as a "defect" described in impact report 516. Thus, symmetry can be repaired in a manner similar to that described above with reference to FIG.                 

Note the following: That is, some structures on the layout, such as hammerheads and serifs (outer and inner corners), are provided to facilitate the accurate transfer of the lines on the mask to the wafer. These structures, although not printed independently from the lines, can affect the CD variation of these lines on the wafer. Thus, variations in these structures resulting from the printing of the mask may also adversely affect the printing of lines combined with these structures on the wafer. By examining the CD variation or corner rounding of lines with joined structures, the quality of these structures can also be analyzed effectively.

Another embodiment

Defect printability analysis, defect criticality score, and mask quality assessment are described in various embodiments. Changes and modifications to these examples will be apparent to those skilled in the art. For example, as described above, the defect free reference mask corresponding to the physical mask is inspected. In one embodiment described above, the defect free reference image is a simulated image of the layout of the physical mask. In an alternative embodiment, the defect free reference image is a defect free area of the physical mask having the same pattern. In another embodiment, the defect free reference image is a simulated image of the mask as it is processed in manufacturing. In another embodiment, the defect free reference image is a physical mask image as compensated for the microscope (lens) effect.

In a standard mask fabrication process such as that shown in FIG. 1, the defect print aptitude / mask quality analysis described above may be included in mask inspection step 116. Alternatively, the mask quality analysis described above is equally applicable to wafer repair processing. For example, after the wafer fails inspection as determined in step 124, processing steps for repairing the wafer may be added, rather than proceeding to mask repair steps 128 and 130. In another embodiment, in addition to the defect criticality score, impact report 516 (FIG. 5) includes a phase that includes cross-sectional contour lines, light intensity data, critical dimensions for different defocus, and effects on critical dimensions. ) May contain other job output, such as delivery data.

1. A method of determining the edge roughness of a feature in a mask,

Determining a centerline of the feature based on the representation of the mask;

Measuring a first length of the first lip extending from the centerline to one edge of the feature;

Measuring a second length of a second lip extending from the centerline to that edge of the feature; And

Comparing the first and second lengths to determine edge roughness.

2. The method of clause 1, wherein the representation comprises a layout of a mask.

3. The method of clause 1, wherein the representation comprises a layout of one layer of an integrated circuit.                 

4. The method of clause 1 wherein the feature is a line.

5. The method of clause 1, wherein the feature is a contact.

6. The method of repairing a mask,

Determining an edge roughness of the feature on the mask, and using a lithographic tool to repair the mask if the edge roughness is outside a predetermined value.

7. The method of clause 6, wherein the predetermined value is selected by the user.

8. The method of clause 6, wherein determining the edge roughness comprises determining a centerline of the feature based on the integrity representation of the feature.

9. The method of clause 6, wherein the feature includes at least one of a line and a contact.

10. A method of repairing a wafer,

Determining an edge roughness of the feature on the mask, and using the lithographic tool to repair the wafer if the edge roughness is outside a predetermined value.

11. The method of clause 10, wherein the predetermined value is selected by the user.

12. The method of clause 10, wherein determining the edge roughness comprises determining a centerline of the feature based on the integrity representation of the feature.

13. The method of clause 10, wherein the feature comprises at least one of a line and a contact.                 

14. A method of determining corner rounding of a contact in a lithographic mask, wherein

Determining a centerline of the contact in a first direction;

Providing a plurality of theoretical cuts through the contact in a second direction substantially perpendicular to the first direction, wherein each cut provides a lip extending from the centerline to the edge of the contact; And

Comparing at least two ribs, one lip located near a corner of a contact, and another lip not located near a corner, to determine corner rounding.

15. A method of determining the symmetry of contacts in a lithographic mask,

Determining a first centerline of the contact in a first direction;

Providing a plurality of theoretical cuts through the contact in a second direction substantially perpendicular to the first direction, wherein each cut provides a first critical dimension extending from the first edge of the contact to the second edge of the contact;

Determining a centerline of the contact in a second direction;

Providing a plurality of theoretical cuts through the contact in a first direction substantially perpendicular to the first direction, wherein each cut provides a second critical dimension extending from the third edge of the contact to the fourth edge of the contact; And

Comparing the first and second critical dimensions to determine the symmetry of the contact.                 

16. A method of repairing a mask,

Determining any corner rounding of contacts on the wafer, and using a lithographic tool to repair the wafer if the edge roughness is outside a predetermined value.

17. The method of paragraph 16, wherein the predetermined value is selected by the user.

18. The method of clause 16, wherein determining the edge roughness comprises determining a centerline of the feature based on the integrity representation of the feature.

19. In an integrated circuit,

A plurality of features for performing a function; And

At least one repaired feature, wherein the at least one repaired feature results from an automatic defect criticality score.

20. The integrated circuit of clause 19, wherein the at least one repaired feature comprises a line.

21. The integrated circuit of clause 19, wherein the at least one repaired feature comprises a contact.

22. The integrated circuit of clause 19, wherein the at least one repaired feature comprises an OPC feature.

23. The integrated circuit of clause 19, wherein the at least one repaired feature comprises one of a hammerhead, a serif, and a bias.

24. A mask inspection system,                 

Means for generating a simulated wafer image of a feature on the mask;

Means for determining a centerline for the simulated wafer image based on the defect representation of the feature; And

And a means for measuring the appearance of the simulated wafer image based on the centerline.

25. The system of clause 24, wherein the appearance comprises line edge roughness.

26. The system of paragraph 25, further comprising means for evaluating possible repairs made to the mask based on line edge roughness.

27. The system of paragraph 26, further comprising a mask repair tool that receives a signal from the means for evaluating a possible repair.

28. The system of clause 25, wherein the appearance comprises corner rounding.

29. The system of paragraph 28, further comprising means for evaluating possible repairs made to the mask based on corner rounding.

30. The system of paragraph 29, further comprising a mask repair tool that receives a signal from the means for evaluating a possible repair.

31. An inspection system for analyzing features on a mask, wherein

Means for generating a simulated wafer image of the feature;

Means for determining a centerline for the simulated wafer image of the feature based on the defect representation of the feature; And                 

And a means for determining whether the feature passes a predetermined criterion.

32. The system of clause 31, wherein the predetermined criterion comprises a line edge roughness of the simulated wafer image.

33. The system of clause 32, further comprising means for evaluating possible repairs made to the mask based on line edge roughness.

34. The system of paragraph 33, further comprising a mask repair tool that receives a signal from the means for evaluating a possible repair.

35. The system of clause 31, wherein the predetermined criterion comprises a symmetry of the simulated wafer image.

36. The system of clause 35, wherein symmetry indicates corner rounding of the simulated wafer image.

37. The system of paragraph 36, further comprising means for evaluating possible repairs made to the mask based on corner rounding of the simulated wafer image.

38. The system of paragraph 37, further comprising a mask repair tool that receives a signal from the means for evaluating a possible repair.

39. A method of quantifying the quality of features on a mask, wherein

(a) determining a centerline of the feature;

(b) measuring a first length of the first lip extending from the centerline to one edge of the feature;

(c) measuring a second length of a second lip extending from the centerline to that edge of the feature;

(d) comparing the first and second lengths;

(e) continuing steps (b), (c) and (d) for a plurality of times;

(f) calculating a score for the quality of the feature based on steps (a) through (e).

40. For a physical mask,

At least one feature modified based on the analysis of the centerline of the simulated wafer image of the feature, wherein the centerline is determined by the defect free representation of the feature; And

A physical mask comprising at least one feature that is not modified based on the analysis of the centerline.

41. The mask of paragraph 40, wherein the defect free representation of the feature comprises a reference mask that corresponds to the defect free physical mask.

42. Computer software for determining the edge roughness of features in a mask,

Means for determining a centerline of the feature based on the representation of the mask;

Means for measuring a first length of a first lip extending from the centerline to one edge of the feature;                 

Means for measuring a second length of a second lip extending from the centerline to one edge of the feature; And

Edge roughness determination software comprising means for comparing the first and second lengths to determine edge roughness.

43. Computer software for determining the corner rounding of contacts in a lithographic mask,

Means for determining a centerline of the contact in a first direction;

Means for providing a plurality of theoretical cuts through the contact in a second direction substantially perpendicular to the first direction, wherein each cut provides a lip extending from the centerline to the edge of the contact; And

Contact corner rounding determination software comprising means for comparing at least two ribs, one lip located near a corner of a contact and another lip not located near a corner to determine corner rounding.

44. Computer software for determining the symmetry of contacts in a lithographic mask,

Means for determining a first centerline of the contact in a first direction;

Means for providing a plurality of theoretical cuts through the contact in a second direction substantially perpendicular to the first direction, wherein each cut provides a first critical dimension extending from the first edge of the contact to the second edge of the contact;

Means for determining a centerline of the contact in a second direction;                 

Means for providing a plurality of theoretical cuts through the contact in a first direction substantially perpendicular to the first direction, wherein each cut provides a second critical dimension extending from the third edge of the contact to the fourth edge of the contact; And

And means for comparing the first and second critical dimensions to determine the symmetry of the contact.

As such, the invention is limited only by the appended claims.

Claims (70)

A method of providing printability analysis for defects on a physical mask, the method comprising: Generating a simulated wafer image of the physical mask; Generating a simulated wafer image of the reference mask corresponding to the intact physical mask; Identifying a first feature adjacent the defect on the simulated wafer image of the physical mask; Identifying a second feature on the simulated wafer image of the reference mask, wherein the second feature corresponds to the first feature; And Calculating a critical dimension deviation that includes the first and second features to provide the printability analysis. The method of claim 1, Wherein said calculating step includes determining a first critical dimension of said first feature and a second critical dimension of said second feature. The method of claim 2, And said calculating step includes calculating relative critical dimension deviations for the first and second features. The method of claim 3, wherein And calculating the relative critical dimension deviation comprises subtracting the second critical dimension from the first critical dimension and dividing the result by the second critical dimension. The method of claim 3, wherein Identifying a plurality of first features proximate the defect on the simulated wafer image of the physical mask; Identifying a plurality of second features on the simulated wafer image of the reference mask, wherein the plurality of second features correspond to the plurality of first features; And Calculating a plurality of relative critical dimension deviations for the plurality of first and second features. The method of claim 5, And providing a maximum critical dimension deviation by determining the largest one of the plurality of relative critical dimension deviations. The method of claim 1, Identifying a third defect free feature on the simulated wafer image of the physical mask; Identifying a fourth feature on the simulated wafer image of the reference mask, wherein the fourth feature corresponds to the third feature; And And calculating a critical dimension deviation that includes the third and fourth features. The method of claim 7, wherein Calculating the third and fourth features includes calculating a first critical dimension of the third feature and a second critical dimension of the fourth feature. The method of claim 8, Wherein said calculating step includes calculating critical dimension deviations for the first and second features. The method of claim 9, And calculating the critical dimension deviation comprises subtracting the first critical dimension from the second critical dimension and dividing a result by the second critical dimension. The method of claim 9, And estimating the critical dimension deviation for N defect features on the simulated wafer image of the physical mask, wherein N is an integer of 2 or greater. The method of claim 11, Wherein the critical dimension deviation for each defect feature is added and the result is divided by N to provide an average critical dimension deviation. A method of providing printability analysis for defects on a physical mask, the method comprising: Generating a simulated wafer image of the physical mask; Generating a simulated wafer image of the reference mask corresponding to the intact physical mask; Calculating an average critical dimension deviation for the defect free areas of the physical mask using the simulated wafer images of the physical mask and the reference mask; Calculating a maximum critical dimension deviation for a defective area of the physical mask using the simulated wafer image of the physical mask and the reference mask; And And using said average critical dimension deviation and said maximum critical dimension deviation to provide said printability analysis. The method of claim 13, And determining a defect criticality score based on the use step. The method of claim 13, And determining the tolerance for the critical dimension change. The method of claim 15, And wherein said using step comprises using said tolerance for a critical dimension change to provide said printability analysis. The method of claim 15, Wherein said using step includes determining a number of analyzed exposures. A method of providing printability analysis for defects on a physical mask, the method comprising: Generating a simulated wafer image of the physical mask; Identifying a first feature affected by the defect on the simulated wafer image; Identifying a second feature on the simulated wafer image that is not affected by the defect, wherein the first and second features have substantially the same critical dimensions in the absence of the defect; Calculating a first critical dimensional deviation for the defect free area of the physical mask using the simulated wafer images of the physical and reference masks; Calculating a second critical dimension deviation with respect to a defective area of the physical mask using the simulated wafer image of the physical and reference masks; And Using the first and second critical dimension deviations to provide the printability analysis. The method of claim 18, And determining the tolerance for the critical dimension change. The method of claim 19, And wherein said using step comprises using said tolerance for a critical dimension change to provide said printability analysis. The method of claim 19, Wherein said using step includes determining a number of analyzed exposures. In the method of manufacturing a physical mask, Designing an integrated circuit; Generating mask design data for one layer of the integrated circuit; Fabricating a physical mask that matches the mask design data; Examining the physical mask based on the simulated wafer image of the physical mask and the simulated wafer image of the reference mask, wherein the reference mask corresponds to a flawless physical mask, The inspection step Calculating an average critical dimension deviation for the defect free areas of the physical mask using the simulated wafer images of the physical and reference masks; Calculating a maximum critical dimension deviation for a defective area of the physical mask using the simulated wafer images of the physical and reference masks; Using the average critical dimension deviation and the maximum critical dimension deviation to provide the printability analysis; And Determining whether the physical mask passes an inspection based on the printability analysis. The method of claim 22, Wherein said inspecting step comprises using tolerances for critical dimensional changes to provide printability analysis. The method of claim 23, And said inspecting step further comprises determining a defect criticality score based on using said tolerance for said average critical dimension deviation, said maximum critical dimension deviation, and a critical dimension change. . The method of claim 22, Wherein said using step includes determining the number of exposures analyzed. The method of claim 25, And wherein said printability analysis comprises a defect criticality score for at least one exposure. The method of claim 25, Wherein said printability analysis comprises a defect criticality score for a plurality of exposures. A method of generating a defect criticality score for a defect on a mask, Providing a two-dimensional analysis of the defect and a first feature on a mask proximate the defect; Providing a first wafer image of the mask; And Providing a defect analysis for a second feature on the wafer image, The second feature corresponds to the simulated first feature, And providing the defect analysis comprises calculating various critical dimension deviations based on the first and second features. The method of claim 28, Identifying a third feature on the defect free mask of the mask; Providing a second wafer image of the reference image; And Further comprising providing a defect analysis for the fourth feature, The third feature represents the first feature, The second wafer image includes the fourth feature corresponding to the third feature being simulated, And providing defect analysis comprises calculating various critical dimension deviations based on the first, second, third, and fourth features. The method of claim 29, And providing defect analysis for the second and fourth features comprises comparing critical dimensions for the second and fourth features. The method of claim 30, The defect analysis providing step further comprises the step of determining the change in the critical dimension by varying the exposure. The method of claim 31, wherein And providing a defect analysis further comprises determining a maximum critical dimension change for each exposure. The method of claim 32, The defect analysis providing step further comprises the step of calculating a relative maximum critical dimension change for each exposure. The method of claim 29, And providing a calibration between the first and second wafer images. In a system for analyzing defects on a physical mask, An inspection tool for generating a mask image from the physical mask and a reference image from a reference mask; A wafer image generator for generating a stepper mask image from the mask image and a stepper reference image from the reference image; And A defect printability analysis generator for comparing the stepper mask image with the stepper reference image, The defect printability analysis generator uses the stepper mask image and the stepper reference image to determine an average critical dimension deviation for the defect free area on the physical mask, The defect printability analysis generator uses the stepper mask image and the stepper reference image to determine a maximum critical dimension deviation for a defective area on the physical mask, Wherein the defect printability analysis generator uses the average critical dimension deviation and the maximum critical dimension deviation to provide printability analysis of the defect. 36. The method of claim 35 wherein Further comprising a critical region identification generator for determining whether the defect is located in a critical region of the physical mask, And the critical region identification generator provides an output to the defect printability analysis generator. 36. The method of claim 35 wherein And a bitmap editor that receives data from the defect printability analysis generator and provides the defect with suggested repairs. The method of claim 37, And a mask repair tool corresponding to the proposed repair. The method of claim 38, The system automatically analyzes the defect and the mask repair tool automatically corresponds to the proposed repair. 36. The method of claim 35 wherein The system automatically analyzes the defect and the defect printability analysis generator automatically provides a materiality score for the defect. A system for generating a defect criticality score for a defect on a physical mask, Means for generating a first image of a defect on the physical mask proximate the defect and a second image of the feature on the reference image; Means for simulating a first wafer image of the first image and a second wafer image of the second image; And Means for generating the defect criticality score based on the first and second wafer images, The generating means Using the first and second wafer images to determine an average critical dimension deviation for the defect free area on the physical mask, Use the first and second wafer images to determine a maximum critical dimension deviation for a defective area on the physical mask, And use the average critical dimension deviation and the maximum critical dimension deviation to provide the defect criticality score. 42. The method of claim 41 wherein Means for identifying whether the defect is within a critical region, and And the identification means provides data to the defect criticality score generating means. 42. The method of claim 41 wherein Wherein said simulating means comprises means corresponding to a plurality of lithographic conditions. delete delete delete delete delete A computer-readable recording medium having recorded thereon computer software for analyzing defects in a mask, The computer software is a computer, Means for generating a simulated wafer image of the first mask; Means for generating a simulated wafer image of a second mask corresponding to the defect free first mask; Means for calculating an average critical dimensional deviation for the defect free areas of the physical mask using the simulated wafer images of the physical and reference masks; Means for calculating a maximum critical dimension deviation for a defective area of the physical mask using the simulated wafer images of the physical and reference masks; And And means for utilizing said average critical dimension deviation and said maximum critical dimension deviation to analyze printability with respect to said defect. The method of claim 49, And said means for use comprises means for generating a defect criticality score for said defect. The method of claim 49, And means for providing repair information for the defect based on the printability. 51. The method of claim 50, And means for providing repair information for the defect based on the defect criticality score. In the method of inspecting a physical mask containing a defect, Generating a simulated wafer image of the physical mask; Generating a simulated wafer image of the reference mask corresponding to the intact physical mask; Calculating an average critical dimension deviation for the defect free areas of the physical mask using the simulated wafer images of the physical and reference masks; Calculating a maximum critical dimension deviation for a defective area of the physical mask using the simulated wafer images of the physical and reference masks; And Using the average critical dimension deviation and the maximum critical dimension deviation to provide printability analysis of the defect. The method of claim 53 wherein And said comparing step comprises generating a defect criticality score. The method of claim 53 wherein And forwarding said information about said defect to a mask repair tool. The method of claim 54, wherein And passing said defect criticality score to a mask repair tool. The method of claim 53 wherein The reference mask is A simulated image of the layout of the physical mask, A defect free area of the physical mask having a pattern substantially the same as the area containing the defect, and And wherein the physical mask comprises one of the simulated images of the physical mask as it is processed at the time of manufacture. The method of claim 53 wherein And wherein said simulated image generation step of said physical mask compensates for image distortion generated during image capture. delete delete delete delete delete delete delete delete delete delete delete delete

Images