TWI766127B - Method for optimizing focal parameters of lithographic process - Google Patents

Abstract

A method for optimizing focal parameters of lithographic process includes:　Performing a first image capturing step on a first measured pattern with a plurality of focus points; selecting a plurality of the first pattern images corresponding to an intermediate value of the focus from the first pattern images and stacking the selected first pattern images to form a first reference image; removing a plurality of error images from the first pattern images; selecting at least two of the remaining first pattern images corresponding to the intermediate value from the remaining first pattern images and stacking the selected remaining first pattern images to form a first standard image; performing a first similarity analysis on the remaining first pattern images according to the first standard image to obtain a plurality of first image scores; and obtained an optimized focal location according to the first image scores.

Description

Methods of optimizing lithography focus parameters

This specification relates to a semiconductor process, and in particular to a lithography process.

In semiconductor process technology, when a pattern is to be transferred from a photomask to a photoresist, energy is required to activate the photoresist. The source of energy is typically provided by radiation, such as an ultraviolet light source (UV). Focus is to focus the radiant energy of the radiant light on a single point. The accuracy of focusing directly affects the accuracy of the lithography process. In general, different pattern designs require different radiation energy and different focus positions. The conventional technology usually monitors the focus of the lithography process by inspecting parameters such as the critical dimension (CD), end-of-line shortening (LES) and side wall angle (SWA) of the exposed pattern to optimize the process quality. Among them, the shortening of the end of the line refers to the difference between the actual printing position and the predetermined position of the photoresist at the end of the line; the side wall angle refers to the profile of the photoresist after energy exposure.

However, the critical dimension, the shortening of the line end and the side wall angle are all parameters obtained by the detection device to measure the diffraction value of the light emitted from the mask, and then through the simulation. When the background noise is too large, the lithography is out of focus, or the measurement position is wrong, incorrect parameter information will continue to be provided, and the abnormal situation cannot be responded to in real time, resulting in failure of monitoring and better focusing. The operating quality of the film process.

Therefore, there is a need to provide a more advanced method for optimizing the focus parameters of lithography to improve the problems faced by the prior art.

One aspect of the present specification relates to a method for optimizing lithography focusing parameters, including the following steps: firstly, a first image capturing step is performed on a first tested pattern (pattern) with a plurality of focus positions to obtain a plurality of The first pattern image. Afterwards, a plurality of first pattern images corresponding to the intermediate values of the focal positions are selected from the first pattern images, and a first overlay step is performed to form a first reference image. Then, according to the first reference image, a plurality of erroneous images are removed from the first pattern images. Then, at least two of the remaining first pattern images corresponding to the intermediate values of the focal points are selected, and a second overlay step is performed to form a first standard image. Afterwards, a first similarity analysis is performed on the remaining first pattern images according to the first standard image to obtain a plurality of first image scores, thereby determining the best focus position.

According to the above-mentioned embodiments, the present specification provides a method for optimizing lithography focus parameters. It adopts an image analysis method, and firstly selects a plurality of focal positions including a better focal length range to capture the pattern image. Afterwards, a plurality of pattern images are selected from among the plurality of pattern images corresponding to the intermediate values of the focal positions and overlapped to form a plurality of reference images. Perform image comparison analysis between the reference image and the captured pattern image, so as to eliminate erroneous images caused by factors such as excessive background noise, out-of-focus lithography, or wrong measurement positions. After the erroneous images are eliminated, at least two of the remaining pattern images corresponding to the median value of the focus position are selected to perform another overlay to obtain a plurality of standard images. The standard image and the remaining pattern images are compared and analyzed again to calculate the image score representing the quality of each remaining pattern image, and then statistical analysis is performed according to the image score to obtain the best focus position. In some embodiments, the image score refers to the degree of correlation between each remaining pattern image and its corresponding standard image.

With the lithography focusing parameter optimization method provided by the above-mentioned embodiments of the present application, the erroneous images that may cause measurement failure can be eliminated in the lithography process in real time, so as to ensure the monitoring of the process, and obtain better focusing to optimize the lithography Operational quality of the process.

This specification discloses a method for optimizing lithography focus parameters. In order to make the above-mentioned embodiments and other objects, features and advantages of the present specification more clearly understood, several preferred embodiments are hereinafter described in detail with the accompanying drawings. However, it must be noted that these specific implementation cases and methods are not intended to limit the present invention. The present invention may still be practiced using the various features, elements, methods, and parameters of other embodiments. The preferred embodiments are provided only to illustrate the technical features of the present invention, and not to limit the scope of the present invention. Those with ordinary knowledge in the technical field will be able to make equivalent modifications and changes based on the description of the following specification without departing from the spirit and scope of the present invention. In different embodiments and drawings, the same elements will be represented by the same element symbols.

Please refer to FIG. 1. FIG. 1 is a flowchart of a method for optimizing lithography focusing parameters according to an embodiment of the present specification. Wherein, the method for optimizing lithography focusing parameters includes the following steps: firstly, performing an image capturing step on at least one tested pattern at a plurality of focus positions to acquire a plurality of pattern images (step 101 ).

In some embodiments of this specification, the image capturing step may select a plurality of partial patterns (tested patterns) at different positions for image capturing according to the circuit layout to be transferred in the lithography process. In addition, in some embodiments, light with different light energies (eg, ultraviolet light) can also be used to illuminate the same pattern under test to capture different pattern images. The selection of the plurality of focal positions may be based on the critical dimension of the circuit layout, the complexity of the pattern or other process factors, and the selection of the plurality of focal positions that may be included in the preferred focal length range to capture the pattern image.

For example, please refer to FIG. 2 . FIG. 2 shows the result after image capturing of at least one tested pattern according to an embodiment of the present specification. In this implementation, 3 different light energies, eg, energy 1 (eg, 23.5 microjoules (mJ)), energy 2 (eg, 32.2 microjoules), and energy 3 (eg, 28 microjoules), were used in 11 different The focal position (the focal length is arranged from small to large, respectively focal length -5 microns (micrometer, µm), -4 microns, -3 microns, -2 microns, -1 microns, 0 microns, 1 microns, 2 microns, 3 microns , 4 microns and 5 microns), respectively for 5 different tested patterns, such as the first tested pattern 201A, the second tested pattern 201B, the third tested pattern 201C, the fourth tested pattern 201D and the fifth tested pattern For the pattern 201E, image capture is performed, and a total of 165 (3×11×5) pattern images are obtained.

As shown in FIG. 2, each of the upper, middle, and lower solid boxes respectively includes applying different light energies (eg, energy 1, energy 2, and energy 3) to the first tested pattern 201A, The second tested pattern 201B, the third tested pattern 201C, the fourth tested pattern 201D and the fifth tested pattern 201E are 55 pattern images obtained by image capturing. In addition, according to the tested patterns (the first tested pattern 201A, the second tested pattern 201B, the third tested pattern 201C, the fourth tested pattern 201D, and the fifth tested pattern 201E), these 165 pattern images can be They are divided into three groups, namely, a first pattern image 202A, a second pattern image 202B, a third pattern image 202C, a fourth pattern image 202D and a fifth pattern image 202D.

The first pattern images 202A respectively include 11 pattern images (in the first row in the upper solid box) captured by energy 1, 11 patterns (in the first row in the middle solid box) using energy 2 The captured pattern images and 11 (in the first row in the solid box below) captured pattern images with energy 3. The second pattern images 202B respectively include 11 pattern images (in the second row in the upper solid box) captured with energy 1, and 11 patterns (in the second row in the middle solid box) captured with energy 2 pattern images and 11 (in the second row in the solid box below) pattern images captured with energy 3. The third pattern images 202C respectively include 11 pattern images (in the third row in the upper solid box) captured with energy 1 and 11 patterns (in the third row in the middle solid box) captured with energy 2 pattern images and 11 (in the third row in the solid box below) pattern images captured with energy 3. The fourth pattern images 202D respectively include 11 pattern images (in the fourth row in the upper solid box) captured with energy 1, and 11 patterns (in the fourth row in the middle solid box) captured with energy 2 pattern images and 11 (in the fourth row in the solid box below) pattern images captured with energy 3. The fifth pattern images 202E respectively include 11 pattern images (in the fifth row in the upper solid box) captured with energy 1, and 11 patterns (in the fifth row in the middle solid box) captured with energy 2 pattern images and 11 (in the fifth row in the solid box below) pattern images captured with energy 3.

However, it is worth noting that the pattern to be tested, the energy used, the focal position and the number of pattern images are not limited by this. Anyone with ordinary knowledge in the technical field can select the desired pattern according to actual needs. , energy and focus, and increase or decrease the number of elements described above.

Afterwards, a plurality of pattern images corresponding to the median value of the focus position are respectively selected from each group of pattern images, and a first overlay step is performed to form a plurality of reference images (step 102 ). For example, please refer to FIG. 3, which shows 12 third pattern images (eg, 55 third pattern images 202C in A schematic diagram of a method for forming a reference image 203 by performing the first stacking step 204 on the selected 12 third pattern images.

In some embodiments of the present specification, the first overlay step 204 is to overlay the selected 12 third pattern images to form a third reference image 203 with overlapping images. At the same time, the first pattern image 202A, the second pattern image 202B, the fourth pattern image 202D and the fifth pattern image 202D are also used in the same way to perform the above-mentioned first overlapping image step 204, thereby forming a first overlapping image with overlapping images. A reference image, a second reference image, a fourth reference image and a fifth reference image (not shown).

In some embodiments of the present specification, before the first overlay step 204 is performed, a noise filtering step may optionally be performed on the selected 12 third pattern images. For example, in this embodiment, Gaussian Blur technology is used to preprocess the selected 12 third pattern images to reduce image noise.

Next, according to the first reference image, the second reference image (not shown), the third reference image 203, the fourth reference image and the fifth reference image (not shown), from the first pattern image 202A, the second pattern image 202B , a plurality of erroneous images are removed from the third pattern image 202C, the fourth pattern image 202D and the fifth pattern image 202E (step 103). In some implementations of this specification, the step of removing a plurality of false images from the first pattern image 202A, the second pattern image 202B, the third pattern image 202C, the fourth pattern image 202D and the fifth pattern image 202E may be It includes calculating the IOU (intersection-over-union, IOU) index of each first pattern image 202A and the first reference image; calculating the IOU index of each second pattern image 202B and the second reference image; calculating each third pattern IOU index of the image 202C and the third reference image 203; calculate the IOU index of each fourth pattern image 202D and the fourth reference image and calculate the IOU index of each fifth pattern image 202E and the fifth reference image. After that, delete one or more (may be zero) first pattern images 202A, second pattern images 202B, third pattern images 202C, fourth pattern images 202D and fifth pattern images with values less than the predetermined IOU index value 202E.

Please refer to FIG. 4. FIG. 4 is a simplified schematic diagram of a method for calculating the IOU index of a pattern image and a reference image according to an embodiment of the present specification. Taking the third pattern image 202C and the third reference image 203 as examples, the method for calculating the IOU index of each of the third pattern image 202C and the third reference image 203 includes comparing each of the third pattern images 202C and the third reference image 203 The image intersection of the two is divided by the image union of the two. The image intersection of the third pattern image 202C and the third reference image 203 refers to the image area A1 where the third pattern image 202C and the third reference image 203 overlap each other. The image combination part of the third pattern image 202C and the third reference image 203 refers to the image area A2 of the third pattern image 202C that does not overlap with the third reference image 203, plus the third reference image 203 does not overlap with the third reference image 203. The overlapping image area A3 of the pattern image 202C is added to the overlapping image area A1 of the two. The IOU index can be expressed by formula (1): IOU=A1/(A1+A2+A3) ....(1) In this embodiment, referring to the calculated result, delete the IOU index value smaller than the self-defined lower limit value 11 wrong third pattern images 202C (as shown by the solid box in FIG. 3 ), leaving 22 remaining third pattern images 202C.

In some embodiments of this specification, the step of removing a plurality of false images from the first pattern image 202A, the second pattern image 202B, the third pattern image 202C, the fourth pattern image 202D and the fifth pattern image 202E, further It may include evaluating the degree of correlation between each first pattern image 202A and the first reference image; evaluating the degree of correlation between each second pattern image 202B and the second reference image; evaluating each third pattern image 202C and the first The degree of correlation between the three reference images 203; the degree of correlation between each of the fourth pattern images 202D and the fourth reference image is evaluated and the degree of correlation between each of the fifth pattern images 202E and the fifth reference image is evaluated. After that, delete one or more (or zero) first patterns with low correlation degree with the corresponding first reference image, second reference image, third reference image, fourth reference image and fifth reference image Image 202A, second pattern image 202B, third pattern image 202C, fourth pattern image 202D, and fifth pattern image 202E.

來代表；第三參考影像203的影像畫素值，可以由一個25個數字元素構成的第二矩陣

來代表。接著，分別將第一矩陣和第二矩陣轉換為一個25個數字元素所構成的數列。例如，將第一矩陣轉換為第一數列A(x)：[0 6 12.... 132 138 134]；將第二矩陣轉換為第二數列B(y)：[50 52 54....94 96 98]。計算第一數列A(x)和第二數列B(y)的相關係數r_xy ，相關係數r_xy 以算式(2)表示：

其中S_x 和S_y 分別為第一數列A(x)和第二數列B(y)的標準差。在本實施例中，參照計算所得的結果，刪除相關係數數值小於自訂的值的11張錯誤的第三圖案影像202C(如第3圖實線方框所繪示)，餘留下22個剩餘的第三圖案影像202C。Taking the third pattern image 202C and the third reference image 203 as an example, the step of evaluating the degree of correlation between each third pattern image 202C and the third reference image 203 includes calculating each third pattern image 202C and the third reference image The correlation coefficient between the images 203 . First, image pixel values of one of the third pattern images 202C and the third reference image 203 are obtained respectively. For example, in this embodiment, the image pixel value of one of the third pattern images 202C may be a first matrix composed of 25 digital elements

to represent; the image pixel value of the third reference image 203 can be a second matrix composed of 25 digital elements

to represent. Next, the first matrix and the second matrix are respectively converted into a sequence of 25 digital elements. For example, convert the first matrix to the first sequence A(x): [0 6 12.... 132 138 134]; convert the second matrix to the second sequence B(y): [50 52 54... .94 96 98]. Calculate the correlation coefficient r _xy of the first sequence A(x) and the second sequence B(y). The correlation coefficient r _xy is expressed by formula (2):

where S _x and S _y are the standard deviations of the first sequence A(x) and the second sequence B(y), respectively. In this embodiment, referring to the calculated result, 11 wrong third pattern images 202C (as shown by the solid line box in Fig. 3) whose correlation coefficient value is smaller than the preset value are deleted, and 22 images remain The remaining third pattern image 202C.

It is worth noting that the step of removing a plurality of erroneous images from the first pattern image 202A, the second pattern image 202B, the third pattern image 202C, the fourth pattern image 202D and the fifth pattern image 202E can be performed independently by IOU The calculation of indicators or the evaluation of correlations alone can also include both.

Then, at least two of the remaining pattern images corresponding to the median value of the focus are selected, and the second overlay step 205 is performed to form a standard image (step 104 ). Please refer to FIG. 5. FIG. 5 illustrates an implementation method of performing the second overlay step 205 after removing a plurality of erroneous images according to an embodiment of the present specification. In some embodiments of the present specification, the second overlay step 205 includes the following steps: firstly, from the remaining first pattern image 202A, the second pattern image 202A after removing the plurality of wrong images (indicated by a cross) Among the image 202B, the third pattern image 202C, the fourth pattern image 202D and the fifth pattern image 202E, a plurality of pattern images captured with the same light energy are respectively selected. For example, in this embodiment, 11 first pattern images 202A, 11 second pattern images 202B, 7 third pattern images 202C, 10 fourth pattern images 202D, and 9 first pattern images 202D captured by energy 2 are selected. Five pattern images 202E.

The pattern images (11 first pattern images 202A, 11 second pattern images 202B, 7 third pattern images 202C, 10 fourth pattern images 202D and 9 In the fifth pattern image 202E), a plurality of pattern images corresponding to the median value of the focal position (the focal length is 0 μm) are selected (for example, three first pattern images 202A and three second pattern images as indicated by the dotted box). image 202B, 3 third pattern images 202C, 3 fourth pattern images 202D, and 3 fifth pattern images 202E), and then compare the selected 3 first pattern images 202A, 3 second pattern images 202B, The three third pattern images 202C, the three fourth pattern images 202D and the three fifth pattern images 202E are subjected to the second overlay step 205 to form the first standard image 206A, the second standard image 206B, the Three standard images 206C, fourth standard images 206D, and fifth standard images 206E.

Wherein, the implementation method of the second stacking step 205 is substantially similar to the foregoing first stacking step 204, in which the pattern image corresponding to the intermediate value of the focus position is selected. The only difference is that the pattern image selected in the second overlay step 205 must be the pattern image captured with the same light energy.

Subsequently, the remaining first patterns after the error images (indicated by the cross) are removed according to the first standard image 206A, the second standard image 206B, the third standard image 206C, the fourth standard image 206D and the fifth standard image 206E The image 202A, the second pattern image 202B, the third pattern image 202C, the fourth pattern image 202D and the fifth pattern image 202E are subjected to similarity analysis to obtain a plurality of first image scores, second image scores, and third images respectively score, fourth image score, and fifth image score (step 105).

In some embodiments of the present specification, the similarity analysis performed according to the first standard image 206A, the second standard image 206B, the third standard image 206C, the fourth standard image 206D and the fifth standard image 206E includes calculating each Calculate the first correlation coefficient between the remaining first pattern images 202A and the first standard image 20A6, and calculate the second correlation coefficient between each remaining second pattern image 202B and the second standard image 206B , calculate the third correlation coefficient between each remaining third pattern image 202C and the third standard image 206C, calculate the correlation coefficient between each remaining fourth pattern image 202D and the fourth standard image 206D The fourth correlation coefficient and the fifth correlation coefficient between each of the remaining fifth pattern images 202E and the fifth standard image 206E are calculated.

Each of the obtained first correlation coefficient, second correlation coefficient, third correlation coefficient, fourth correlation coefficient and fifth correlation coefficient can be used as each remaining first correlation coefficient after normalization of data. Pattern image 202A, second pattern image 202B, third pattern image 202C, fourth pattern image 202D and fifth pattern image 202E first image score, second image score, third image score, fourth image score and fifth image score Image score.

Next, the best focus position is determined according to the first image score, the second image score, the third image score, the fourth image score and the fifth image score (step 106). In some embodiments of this specification, the step of judging the best focus position includes according to each remaining first pattern image 202A, second pattern image 202B, third pattern image 202C, fourth pattern image 202D and first pattern image 202D. The first image score, the second image score, the third image score, the fourth image score and the fifth image score of the five-pattern image 202E are matched with their corresponding focus positions (focal lengths) to form a first focus-image respectively A score relationship curve 601 , a second focus-image score relationship curve 602 , a third focus-image score relationship curve 603 , a fourth focus-image score relationship curve 604 , and a fifth focus-image score relationship curve 605 .

For example, please refer to FIG. 6 . FIG. 6 is a graph showing a plurality of focus-image score relationships for determining the best focus position according to an embodiment of the present specification. Among them, the horizontal axis of Fig. 6 is the focal position (focal length), and the vertical axis is the image score after normalization. Among them, the focal position is distributed between the focal length of -6 microns to 6 microns, and the image fraction is distributed between -0.1 to 0.1. This shows that each of the remaining first pattern images 202A, second pattern images 202B, third pattern images 202C, fourth pattern images 202D and fifth pattern images 202E are respectively associated with the first standard image 206A, the second standard image 206B , the third standard image 206C, the fourth standard image 206D and the fifth standard image 206E have extremely high similarity. In other words, most of the failed images due to problems such as excessive background noise, out-of-focus lithography or wrong measurement positions have been eliminated.

Next, compare the first focus-image score relationship curve 601, the second focus-image score relationship curve 602, the third focus-image score relationship curve 603, the fourth focus-image score relationship curve 604, and the fifth focus-image score relationship curve 604, respectively. A linear regression is performed on the image scores of the relationship curve 605 to obtain a curve of second order 606 . The quadratic curve 605 is then differentiated, and the obtained coordinates of the vertex 607 of the quadratic curve 606 are the optimal focus positions. In this embodiment, the optimum focus position is 0.45 microns.

According to the above-mentioned embodiments, the present specification provides a method for optimizing lithography focus parameters. It adopts an image analysis method, and firstly selects a plurality of focal positions including a better focal length range to capture the pattern image. Afterwards, a plurality of pattern images are selected from among the plurality of pattern images corresponding to the intermediate values of the focal positions and overlapped to form a plurality of reference images. Perform image comparison analysis between the reference image and the captured pattern image, so as to eliminate erroneous images caused by factors such as excessive background noise, out-of-focus lithography, or wrong measurement positions. After the erroneous images are eliminated, at least two of the remaining pattern images corresponding to the median value of the focus position are selected to perform another overlay to obtain a plurality of standard images. The standard image and the remaining pattern images are compared and analyzed again to calculate the image score representing the quality of each remaining pattern image, and then statistical analysis is performed according to the image score to obtain the best focus position. In some embodiments, the image score refers to the degree of correlation between each remaining pattern image and its corresponding standard image.

With the lithography focusing parameter optimization method provided by the above-mentioned embodiments of the present application, the erroneous images that may cause measurement failure can be eliminated in the lithography process in real time, so as to ensure the monitoring of the process, and obtain better focusing to optimize the lithography Operational quality of the process.

Although this specification has disclosed the above with preferred embodiments, it is not intended to limit the present invention. Anyone with ordinary knowledge in the technical field can make some changes and modifications without departing from the spirit and scope of the present invention. Therefore, the protection scope of the present invention shall be determined by the scope of the appended patent application.

101: Perform an image capturing step on at least one tested pattern with a plurality of focus positions to obtain a plurality of pattern images 102: Select a plurality of pattern images corresponding to the median value of the focus position from the plurality of pattern images, and perform a first Stacking steps to form a plurality of reference images 103: remove a plurality of false images 104 from a plurality of pattern images according to these reference images: and select at least two from among a plurality of remaining pattern images corresponding to the median value of the focus position , perform a second stacking step to form a plurality of standard images 105: perform similarity analysis on the remaining pattern images after removing the plurality of wrong images according to these standard images to obtain a plurality of image scores 106: by these 201A: First tested pattern 201B: Second tested pattern 201C: Third tested pattern 201D: Fourth tested pattern 201E: Fifth tested pattern 202A: First pattern image 202B: Second pattern image 202C: Third pattern image 202D: Fourth pattern image 202D: Fifth pattern image 204, 205: Overlay step 203: Third reference image 206A: First standard image 206B: Second standard image 206C : third standard image 206D: fourth standard image 206E: fifth standard image 601, 602, 603, 604, 605: focus-image score relationship curve 606: quadratic curve 607: quadratic curve vertex

In order to make the above-mentioned embodiments and other purposes, features and advantages of the present invention more obvious and easy to understand, a few preferred embodiments are given and described in detail as follows in conjunction with the accompanying drawings: A flow chart of a method for optimizing lithography focusing parameters according to the embodiment; FIG. 2 shows the result after image capturing of at least one tested pattern according to an embodiment of the present specification; FIG. 3 shows the result From the 55 third pattern images in Fig. 2, 12 third pattern images corresponding to the median value of the focal position are selected to perform the first overlay step to form a reference image. Fig. 4 is based on this specification. Figure 5 shows a simplified schematic diagram of a method for calculating the IOU index of a pattern image and a reference image according to an embodiment of the present specification; Figure 5 shows a second overlay step after removing a plurality of wrong images according to an embodiment of the present specification The implementation method of : and FIG. 6 is a graph showing a plurality of focus-image score relationship curves used to determine the optimal focus position according to an embodiment of the present specification.

101: Perform an image capturing step on at least one tested pattern with a plurality of focus positions to obtain a plurality of pattern images

102: Select a plurality of pattern images corresponding to the median value of the focus position from the plurality of pattern images, and perform a first stack of steps to form a plurality of reference images

103: Remove a plurality of erroneous images from a plurality of pattern images according to these reference images

104: Select at least two of the remaining pattern images corresponding to the median value of the focus position, and perform a second overlay step to form a plurality of standard images

105: According to these standard image pairs, the remainder after removing a plurality of erroneous images Similarity analysis is performed on pattern images to obtain multiple image scores separately

106: Use these image scores to determine the best focus position

Claims (10)

A method for optimizing lithography focusing parameters, comprising: performing a first image capturing step on a first tested pattern with a plurality of focus positions to obtain a plurality of first pattern images; from the first patterns Selecting a plurality of first pattern images corresponding to an intermediate value of the focus positions in the image, and performing a first overlay step to form a first reference image; according to the first reference image, from the first pattern images removing a plurality of erroneous images from the pattern image; selecting at least two of the remaining first pattern images corresponding to the intermediate values of the focus positions, and performing a second overlay step to form a a first standard image; and performing a first similarity analysis on the remaining first pattern images according to the first standard image to obtain a plurality of first image scores, thereby determining an optimal focus position. The method for optimizing lithography focus parameters as described in claim 1, wherein the first image capturing step includes using a plurality of light energies in combination with the focus positions to capture the first pattern images. The method for optimizing lithography focus parameters as described in item 2 of the claimed scope, wherein after removing a plurality of false images from the first pattern images, the method further comprises: selecting a plurality of erroneous images from the remaining first pattern images a first pattern image captured with a same light energy; selecting a plurality of first pattern images captured with the same light energy from a plurality of remaining first pattern images corresponding to the median value of the focus positions performing the second overlay step; and comparing each of the first pattern images captured with the same light energy with the first standard image to obtain the first image scores . The method for optimizing lithography focus parameters as described in claim 3, wherein the step of obtaining the first image scores includes: capturing one of the first pattern images captured at the same focus position extracting a first pixel value group of the first standard image; and calculating a correlation coefficient between the first pixel value group and the standard pixel value group. The method for optimizing lithography focus parameters as described in item 3 of the scope of the patent application further includes a noise filtering step before performing the second overlay step. The method for optimizing lithography focus parameters as described in claim 5, wherein the noise filtering step includes using a Gaussian Blur algorithm. The method for optimizing lithography focus parameters as described in item 1 of the scope of the patent application further comprises: performing a second image capturing step on a second tested pattern at the focus positions to acquire a plurality of second pattern images ; selecting a plurality of second pattern images corresponding to the intermediate values of the focus positions from the second pattern images, and performing a third overlay step to form a second reference image; according to the second a reference image, removing a plurality of false images from the second pattern images; selecting at least two of the remaining second pattern images corresponding to the median value of the focal points, and performing a fourth The step of stacking images is to form a second standard image; and a second similarity analysis is performed on the remaining second pattern images according to the second standard image, so as to obtain a plurality of second image scores. The method for optimizing lithography focus parameters as described in item 7 of the scope of the patent application, wherein determining the best focus comprises: forming a first focus-image score according to the first image scores and the second image scores, respectively the relationship curve and a second focus-image score relationship curve; perform a linear regression on the first focus-image score relationship curve and the second focus-image score relationship curve to obtain a quadratic curve of second order); and a derivative of the quadratic curve. The method for optimizing lithography focus parameters as described in item 1 of the claimed scope, wherein the step of removing a plurality of erroneous images from the first pattern images includes calculating each of the first pattern images according to the first reference image. An IOU (intersection-over-union, IOU) indicator of a pattern image. The method for optimizing lithography focus parameters as described in item 1 of the claimed scope, wherein the IOU index refers to an image intersection of one of the first pattern images and the first reference image divided by the two A part of an image collection of the author.

Images