A segment replacement correction method based on mask weakness pattern classification

Abstract

The application discloses a kind of segment replacement correction methods based on mask weak point pattern classification, first according to test pattern, do specific classification, the weak point of test pattern is divided into bridge, shrinkage, line end narrow and contact hole is covered by silicon gate layer;Then for each type of weak point in test pattern, the weak point image is handled by fast Fourier transform technology, eliminate dark spot and shrink range, obtain the general correction segment of each type and with weak point pattern and simulation pattern together into correction segment library;Finally, actual pattern data is matched with library, according to the classification method of test pattern, the weak point type of actual pattern data is judged, and then the correction is completed by replacing the correction segment library file corresponding to the weak point type.The method saves general correction segment and simulation pattern to form classification weak point correction segment library, which is more efficient than one-to-one comparison of original pattern and correction result, and does not need to establish a large number of graphic library data;It is simple and easy to operate, and shortens the publishing cycle.

Description

Technical Field

[0001] This invention relates to the field of optical proximity correction for microelectronic layout data, and more particularly to a fragment replacement correction method based on mask weakness pattern classification. Background Technology

[0002] As circuit feature sizes continue to shrink, photolithographic weaknesses have become a serious factor affecting manufacturing yield. A weakness is an area within the mask layout pattern; during manufacturing, due to factors such as light diffraction, etching proximity effects, and coverage control, circuit failures are more likely to occur. Therefore, before transferring the design pattern onto the silicon wafer, weakness checks and corrections must be performed in the post-OPC stage. When addressing different weaknesses or defects in OPC, the original OPC script is typically optimized based on the characteristics of the weakness pattern. The optimized script also needs to carefully consider the environment surrounding the weakness, correcting it without affecting the correction of other normal patterns.

[0003] Addressing weaknesses using OPC scripts significantly increases script complexity and reduces readability, making subsequent script optimization more difficult and requiring more verification data. Script optimization necessitates continuous iteration, resulting in substantial computational costs and compromising both workload and efficiency. While yield is guaranteed, it increases the time required to resolve issues, impacting wafer yield development progress and mask release cycles. Furthermore, script optimization encounters bottlenecks in complex environments, making script optimization difficult, as scripts typically directly affect global variables, not just local ones. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies by proposing a fragment replacement and correction method based on mask weakness graphic classification.

[0005] The objective of this invention is achieved through the following technical solution: a fragment replacement and correction method based on mask weakness graph classification, the method comprising the following steps:

[0006] S1: Based on the test pattern, the weaknesses of the test pattern are classified into bridging, shrinkage, line end narrowing, and contact holes being covered by silicon gate layer.

[0007] S2: Denoise the weak point image by using Fast Fourier Transform (FFT) for each type of weakness in the test image, eliminate dark spots and narrow the range to obtain a general correction fragment for each type as an OPC correction image. Store the OPC correction image, weak point image and simulation image into the correction fragment library.

[0008] S3: Perform library matching on the actual graphic data, determine the weakness type of the actual graphic data according to the classification method of the test graphics, and then replace it with the corresponding correction fragment library file to complete the correction.

[0009] Furthermore, the classification method based on the test pattern is as follows: perform photolithography rule checks on the test pattern, measure each position of the simulated contour, if the gap between lines is less than 0.04um, it is judged as bridging; if the line width is less than 0.05um, it is judged as shrinkage; if the gap at the line endpoint is less than 0.07um, it is judged as line end narrowing; if the ratio of the CT target pattern to the poly target pattern being covered by the CT target pattern is greater than 10%, it is judged as the contact hole being covered by the silicon gate.

[0010] Furthermore, the simulated contour is obtained by simulating the test graphic.

[0011] Furthermore, the specific steps for storing the OPC correction graphic into the correction fragment library are as follows: save the graphic as a graphic within the size range of the ambit, centered on the weakness.

[0012] Furthermore, the specific calculation method for the ambit is as follows:

[0013] ambit = 96 * pixelsize, pixelsize = λ / 10NA, where λ is the wavelength of the illumination light, NA is the numerical aperture of the imaging system, and pixelsize is the pixel size of the model.

[0014] Further, S2 specifically involves: performing an FFT transform on the test image to obtain a frequency domain image, performing noise reduction processing in the frequency domain to eliminate 20% of the dark spots and shrink their range, and obtaining a universal correction segment through an inverse Fourier transform.

[0015] Furthermore, the simulation pattern stored in the correction fragment library is based on the OPC correction pattern to perform photolithography simulation to simulate the exposure pattern of the fragment layout pattern on the wafer during the photolithography process, and the simulation process is used to simulate the physical photolithography process.

[0016] The beneficial effects of this invention are:

[0017] The classification weakness correction fragment library refers to the process weaknesses discovered during the original OPC correction method and OPC verification. These process weaknesses can be improved by optimizing the weak points using Fourier transform. The type information of these process weaknesses, general correction fragments, and simulation graphics are saved to form the classification weakness correction fragment library. This is more efficient than comparing the original graphics and correction results one by one, and it does not require the establishment of a large amount of graphics library data.

[0018] This method is simple and easy to implement, does not require the operator to have in-depth professional knowledge, and also saves time in optimizing OPC scripts and shortens the publication cycle. Attached Figure Description

[0019] Figure 1 A flowchart illustrating the steps of a fragment replacement and correction method based on mask weakness graphic classification provided in an embodiment of the present invention;

[0020] Figure 2 A flowchart of weakness graph classification and matching provided for embodiments of the present invention;

[0021] Figure 3 A schematic diagram of the photolithographic weak point region is provided for an embodiment of the present invention;

[0022] Figure 4 This is a schematic diagram of the classification of weaknesses provided in an embodiment of the present invention;

[0023] Figure 5 A schematic diagram of weakness correction provided in an embodiment of the present invention;

[0024] Figure 6 A simulation effect diagram of the weakness correction graphic fragment replacement provided in an embodiment of the present invention;

[0025] Figure 7 This is an example diagram of a weakness correction graphic replacement provided in an embodiment of the present invention. Detailed Implementation

[0026] The specific embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0027] like Figure 1 As shown, this invention provides a fragment replacement and correction method based on mask weakness graph classification. The basic implementation steps of this method can be summarized as follows:

[0028] First, classify the test images according to their specific categories. The classification process is as follows. Figure 2 As shown, Fourier transforms are performed on various types of weaknesses to obtain corrected segments, and a classified weakness correction segment library is established. The library file contains weakness graphics, OPC correction graphics, and simulation graphics.

[0029] The FFT (Fast Fourier Transform) technique is used to denoise the weak image, eliminating 20% ​​of the dark spots and shrinking their range, resulting in a general correction fragment that is then added to the correction fragment library as an OPC correction graphic.

[0030] The actual graphic data is matched against the database; the matching process is similar to... Figure 2 The classification method is the same, except that matching is for random target graphics to be processed, while classification is for test graphics. First, it determines which type of weakness it belongs to, and then replaces it with the corresponding correction fragment library file to complete the correction.

[0031] The Fast Fourier Transform (FFT) is a method for converting a signal from the time domain to the frequency domain. It decomposes a signal into a combination of sine and cosine functions, each combination corresponding to a specific frequency. This allows us to analyze the frequency characteristics of the signal in the frequency domain. In image processing, FFT is widely used for image enhancement, filtering, and denoising.

[0032] The Fourier transform is the inverse operation of the Fourier transform; it can restore a frequency domain signal back to a time domain signal. Mathematically, the inverse Fourier transform is defined as:

[0033]

[0034] Where t represents time, and X(ω) is the signal's spectrum. According to this formula, we can see that the essence of the inverse Fourier transform is to integrate the frequency domain signal and multiply it by an exponential function, thus converting it into a time domain signal. In practical applications, the discrete inverse Fourier transform (DIFT) is typically used to calculate the inverse Fourier transform of discrete signals. Its formula is as follows:

[0035]

[0036] Where n is the length of the signal, and x[k] is the result of the discrete Fourier transform (dft) of the signal in the frequency domain. The inverse Fourier transform is an important mathematical tool with wide applications in signal processing, communication, image processing, and other fields.

[0037] To reduce noise, this embodiment needs to filter out the high-frequency components, which can typically be achieved using a low-pass filter. In the frequency domain, the function of a low-pass filter is to truncate the high-frequency signal, retaining only the low-frequency components. The frequency response of a low-pass filter can be expressed as:

[0038]

[0039] Where ωc is the cutoff frequency, indicating that signals above this frequency are truncated. By performing a Fourier transform on the original signal, filtering out the high-frequency components, and finally performing an inverse Fourier transform, the denoised signal can be obtained.

[0040] The method presented in this paper involves performing an FFT transform on the target image to obtain a frequency domain image, and then performing denoising processing in the frequency domain to eliminate 20% of the dark spots and shrink their area. Figure 4As shown in the box, the effect of the replacement image in the frequency domain is obtained. After inverse Fourier transform, the graphic segment to be replaced is obtained. The contour image of the replacement graphic after simulation does not have the weakness problem. After the transformation process is completed, a general correction segment for this type of weakness is obtained.

[0041] Example:

[0042] Step 1: Establish a classification weakness correction fragment library, and save the original target graphics of various types and their corresponding general optimized graphics fragments into the classification weakness graphics library;

[0043] Methods for establishing a classification weakness correction fragment library, such as Figure 2 The workflow is as follows: Data is processed using the original OPC correction method. After OPC verification, process weaknesses are identified and categorized. These weaknesses can be improved or eliminated by obtaining general correction fragments through Fourier transform processing followed by OPC processing. The original target graphic and general graphic fragments at the process weakness location are then saved to form a classified weakness correction fragment library. The graphics are saved as a range of sizes centered on the weakness, such as... Figure 3 As shown. The size of the ambit is determined using the following empirical formula: ambit = 96 * pixelsize, pixelsize = λ / 10NA, where λ is the wavelength of the illumination light, NA is the numerical aperture of the imaging system, and pixelsize is the model pixel size. If NA = 1.35 and λ = 193nm, then ambit is 1.372 microns.

[0044] In this step, the first step is to generate test patterns for the required types of process weaknesses. After identifying each weakness, they are classified. A photolithography rule check is then performed, and measurements are taken at various locations along the simulated contour. Figure 4 As shown in (a), a gap of less than 0.04 μm between lines is considered a bridging, such as... Figure 4 As shown in (b), a linewidth less than 0.05µm is considered shrinkage. Figure 4 As shown in (c), if the value at the line endpoint is less than 0.07µm, it is considered line narrowing. Figure 4 As shown in (d), if the ratio of the ct target pattern to the poly target pattern is greater than a certain limit, such as 10%, then the contact hole is covered by the silicon gate.

[0045] Step 2: Correct the weaknesses using Fourier transform denoising. Different types are corrected using their common graphic segments, and the corrected graphics are saved to the library. The common correction segment is shown below. Figure 5 The image shown is the result of using the target image. Figure 5 The frequency domain image is obtained by performing an FFT transform on (a). Figure 5 (b) Enlarge the image and mark the dark spots. Figure 5 In (c), denoising is performed in the frequency domain to eliminate 20% of the dark spots and shrink their range, resulting in the frequency domain effect of the replacement image. Figure 5 (f) Reducing the size of the image Figure 5 The image segment to be replaced is obtained by inverse Fourier transform of (e). Figure 5 (d)

[0046] Figure 6 The replacement graphic shown Figure 6 (c) Contour image after simulation Figure 6 In (d), no weaknesses were found, and the transformation was completed. Figure 6 The original graphic without any processing Figure 6 (a) Simulated contour image Figure 6 A bridging problem occurs in (b), which is obvious in contrast.

[0047] Step 3: Compare the original target image that needs to be corrected with the original target images of various types in the classification weakness correction fragment library to determine their types;

[0048] In this step, the type determination is not very complex because it only involves determining four categories. Figure 4 The weakness types shown include bridging, line narrowing, line end error, and contact holes being completely covered by the silicon gate layer. The matching process is similar to... Figure 2 They are the same, except that matching is for the random target image to be processed, while classification is for the test image.

[0049] Step 4: Extract a universal correction fragment from the classification weakness correction fragment library that matches the type of the original target image for which optical proximity correction data is required;

[0050] In this step, after step 2 determines the category of the target image, a general correction fragment for that category is extracted from the category weakness correction fragment library. Figure 7 Taking bridging graphics as an example, determine the original graphics. Figure 7 If (a) represents the bridging category, then the general correction fragment for the bridging category is extracted. Figure 7 In (b), the corrected pattern is obtained after replacement. Figure 7 In (c), no bridging problem occurs.

[0051] Step 5: Replace the weak points of the original target graphic that need to be corrected with the extracted general graphic fragments to generate a new target graphic and use the new target graphic for subsequent OPC processing and verification.

[0052] In this step, the weak areas in the original graphic are directly replaced with general correction fragments to obtain a new target graphic. Figure 7After replacing the generic graphic fragment in the example, the resulting new target graphic can be used for subsequent OPC processing and verification.

[0053] After Fourier transform denoising in this embodiment, the dark spots in the transformed image of the layout corresponding to the weak point were reduced by 20%, and the weak points were eliminated after simulation of the layout obtained by inverse transform.

[0054] After treatment, the process window PW in the weak area increased by 31.5%, resulting in higher manufacturability.

[0055] The method of establishing a general fragment correction saves script iteration complexity and time, and the photomask publication cycle is shortened by 5.6%.

[0056] The above embodiments are used to explain and illustrate the present invention, but not to limit the present invention. Any modifications and changes made to the present invention within the spirit and scope of the claims shall fall within the protection scope of the present invention.

Claims

1. A fragment replacement and correction method based on mask weakness graph classification, characterized in that, The method includes the following steps: S1: Based on the test pattern, the weaknesses of the test pattern are classified into bridging, shrinkage, line end narrowing, and contact holes being covered by silicon gate layer. S2: Fast Fourier Transform (FFT) is used to denoise the weakness images of various types of weaknesses in the test images. Specifically, the FFT is performed on the test images to obtain the frequency domain image, and denoising is performed in the frequency domain to eliminate 20% of the dark spots and shrink their range. After inverse Fourier transform, a general correction fragment is obtained. Each type of general correction fragment is obtained as an OPC correction graphic. The OPC correction graphic, weakness graphic, and simulation graphic are stored in the correction fragment library. The specific steps for storing the OPC correction graphic in the correction fragment library are as follows: the graphic is saved as a graphic within the range of the ambit size, centered on the weakness. The ambit is specifically calculated as follows: ambit = 96 * pixelsize, pixelsize = λ / 10NA, where λ is the wavelength of the illumination light, NA is the numerical aperture of the imaging system, and pixelsize is the pixel size of the model; S3: Perform library matching on the actual graphic data, determine the weakness type of the actual graphic data according to the classification method of the test graphics, and then replace it with the corresponding correction fragment library file to complete the correction.

2. The fragment replacement and correction method based on mask weakness graph classification according to claim 1, characterized in that, The classification method based on the test pattern is as follows: perform photolithography rule checks on the test pattern, measure each position of the simulated contour, if the gap between lines is less than 0.04um, it is judged as bridging; if the line width is less than 0.05um, it is judged as shrinkage; if the gap at the line endpoint is less than 0.07um, it is judged as line end narrowing; if the ratio of the contact hole target pattern to the silicon gate layer target pattern is greater than 10%, it is judged as the contact hole being covered by the silicon gate.

3. The fragment replacement and correction method based on mask weakness graph classification according to claim 2, characterized in that, The simulated contour is obtained by simulating the test graphic.

4. The fragment replacement and correction method based on mask weakness graph classification according to claim 1, characterized in that, The simulation pattern stored in the correction fragment library is based on the OPC correction pattern to simulate the exposure pattern of the fragment pattern on the wafer during the photolithography process, and the simulation process is used to simulate the physical photolithography process.

Images