A spectrum matching method for optical key dimension measurement equipment and equipment used therefor

Abstract

The present invention relates to the technical field of semiconductor chip manufacturing measurement equipment, and proposes a spectral matching method for optical critical dimension measurement equipment and equipment to which it is applied. The method includes: providing two-dimensional and three-dimensional samples for optical critical dimension measurement; collecting the golden machine spectral signal of the sample on a golden machine, and converting the golden machine spectral signal into a spectral image and then performing high-order quantization; collecting the matching machine spectral signal of the sample on a matching machine, and converting the matching machine spectral signal into a spectral image and then performing high-order quantization; and using a preprocessing regression algorithm to adjust the system parameters of the matching machine, so that the matching machine high-order quantization spectral image matches the golden machine high-order quantization spectral image. The present invention uses the high-order quantization spectral image as the matching target, and uses the preprocessing regression algorithm to find the global optimal value of the system parameters, which greatly improves the matching efficiency and matching accuracy of the optical critical dimension measurement equipment.

Description

Technical Field

[0001] The present invention generally relates to the technical field of semiconductor chip manufacturing measurement equipment. Specifically, the present invention relates to a spectrum matching method for optical critical dimension measurement equipment and equipment used therein. Background Art

[0002] In the process of semiconductor chip manufacturing, process technologies such as lithography, development, etching, thin film deposition, ion implantation, metallization, cleaning, and chemical mechanical polishing are widely used. After the processes of lithography, etching, thin film deposition, and chemical mechanical polishing are completed, measurement equipment is needed to accurately measure the line width, groove depth, sidewall inclination, relative position offset, relative position height difference, and other key dimensions of the two-dimensional or three-dimensional structure that are crucial to the yield and device performance. According to the high-precision measurement data, the parameters of the process equipment can be adjusted, so that the key dimensions of the subsequent process improvements are as close to the design value as possible, thereby improving the yield and device performance. Compared with CDSEM (scanning electron microscope) which can only measure line width and TEM (transmission electron microscope) which is a destructive measurement technology, OCD (optical critical dimension) technology based on ellipsometry technology has the advantages of non-destructiveness, fast measurement speed, multiple measurement parameters, and high measurement accuracy.

[0003] In the mass production of semiconductor chips, the matching of multiple measurement equipment is very important. If the matching degree of multiple measurement equipment is poor, it will definitely affect the consistency of semiconductor chip manufacturing, resulting in a decrease in the yield of the production line. Therefore, in order to improve the yield of semiconductor chip mass production, a more accurate matching method suitable for optical critical dimension measurement equipment is urgently needed.

[0004] Currently, commonly used optical key dimension measurement equipment usually uses a set of wafers coated with different thicknesses of SiO2 (thickness usually ranges from 20 angstroms to 10,000 angstroms). By measuring the film thickness, the equipment system parameters are adjusted to make the SiO2 film thickness values ​​measured by different equipment from thin to thick match as much as possible. This technology requires the use of a film thickness measurement model that has been established and optimized in advance. This model requires the floating thickness of SiO2, and sometimes also requires the floating correlation coefficient in the SiO2 dispersion model. Through a professional regression calculation engine, a better fit is achieved. In the process of matching film thickness measurement equipment, the correlation coefficients in all SiO2 dispersion models are usually fixed, and only the thickness of SiO2 is floating. The commonly used regression algorithm is the LM algorithm for solving nonlinear least squares problems. The above method is usually well applied to film thickness measurement equipment. After adjustment, different equipment can achieve better film thickness measurement matching. However, in the field of semiconductor manufacturing, the application scenarios of optical critical dimension measurement technology are becoming more and more complex, especially after the measurement structure develops from 2D to 3D, the application scenarios have many challenges such as complex measurement structure, low sensitivity of key dimension parameters, high correlation between key dimension parameters, and the need to measure multiple key dimension parameters at the same time. Therefore, the traditional film thickness measurement based on SiO2 wafers and the adjustment of equipment system parameters often cannot meet the matching accuracy requirements of more complex optical critical dimension measurement equipment.

[0005] A recently proposed optical critical dimension measurement equipment matching method requires the design of a set of two-dimensional and three-dimensional structures to cover the main optical critical dimension measurement application scenarios. According to the structure of this set of samples, an optical critical dimension measurement database is established for each sample. The full-band spectral signal is collected for each sample on the golden machine. According to the optical critical dimension database, the corresponding critical dimension is verified and interpolated to obtain the corresponding critical dimension. This set of critical dimension values ​​will be used as the matching target value for the remaining machines. The same full-band spectral signal is collected for each sample on the matching machine. Through the regression algorithm, the system parameters and the floating parameters in the optical critical dimension model are adjusted so that each critical dimension of each sample matches the measurement value of the golden machine as much as possible. According to the system parameter values ​​obtained by the above regression method, the system parameters of the optical critical dimension measurement equipment are adjusted, with the goal of matching its measurement value with the golden machine. At present, the regression algorithm used in the industry for optical critical dimension measurement is mainly the LM algorithm for solving nonlinear least squares problems, and the LMF (Levenberg-Marquardt-Fletcher) algorithm extended on this basis. With the increase of two-dimensional and three-dimensional structures used for matching optical critical dimension measurement equipment, the structures of advanced processes are becoming more and more complex, and the measurement parameters are becoming more and more challenging. Sometimes, problems such as non-convergence and low goodness of fit are encountered, and it is difficult to obtain the global optimal solution for the system adjustment parameters, resulting in the optical critical dimension measurement values ​​of some two-dimensional or three-dimensional structures being unable to meet the matching accuracy requirements. Summary of the invention

[0006] In order to at least partially solve the above problems in the prior art, the present invention provides a spectrum matching method for an optical critical dimension measurement device, comprising the following steps:

[0007] Provide multiple 2D and 3D samples for optical critical dimension measurement;

[0008] Collecting a gold machine spectral signal of the sample on a gold machine, and converting the gold machine spectral signal into a spectral image and then performing high-order quantization to generate a gold machine high-order quantized spectral image;

[0009] Collecting a matching machine spectral signal of the sample on a matching machine, converting the matching machine spectral signal into a spectral image and then performing high-order quantization to generate a matching machine high-order quantized spectral image; and

[0010] The system parameters of the matching machine are adjusted using a preprocessing regression algorithm so that the matching machine high-order quantized spectral image matches the golden machine high-order quantized spectral image.

[0011] In one embodiment of the present invention, it is stipulated that the two-dimensional and three-dimensional samples include one or more of the following items: a two-dimensional shallow trench isolation structure, a two-dimensional post-etching detection structure, a two-dimensional sigma etching structure, a three-dimensional contact hole structure, a two-dimensional gate sidewall structure, a two-dimensional silicon germanium epitaxial structure, a three-dimensional post-etching detection structure, a three-dimensional gate sidewall structure, a three-dimensional source and drain shape structure, a three-dimensional silicon germanium epitaxial structure, an active area etching structure, a buried word line etching structure, a bit line etching structure, a capacitor etching structure and a channel through hole etching structure.

[0012] In one embodiment of the present invention, it is provided that the measurement point of the matching machine spectrum signal of the sample collected on the matching machine is the same as the measurement point of the golden machine spectrum signal of the sample collected on the golden machine.

[0013] In one embodiment of the present invention, it is provided that the preprocessing regression algorithm comprises:

[0014] Use low truncation error order to calculate the optical critical dimension positive problem and determine the size of the residual in the Gauss-Newton equation of the optical critical dimension positive problem;

[0015] When the residual amount or the degree of linearity is greater than the judgment value, solving the problem using the Levenberg-Marquardt-Fletcher algorithm in a trust region format; and

[0016] When the residual amount or the degree of linearity is less than or equal to the judgment value, the quasi-Newton method is used to approximately calculate the second-order Hessian matrix at the local optimal solution to correct the Gauss-Newton equation.

[0017] In one embodiment of the present invention, it is stipulated that the system parameters include: the initial azimuth angle of the polarizer, the initial azimuth angle of the first compensator, the initial azimuth angle of the second compensator, the initial azimuth angle of the analyzer, the incident angle, the initial azimuth angle of the moving stage and the initial height offset in the Z direction.

[0018] In one embodiment of the present invention, it is provided that using a preprocessing regression algorithm to adjust the system parameters of a matching machine so that the matching machine high-order quantized spectral image matches the golden machine high-order quantized spectral image includes:

[0019] The matching degree between the matching machine high-order quantization spectral image and the golden machine high-order quantization spectral image is determined according to the structural similarity, wherein the structural similarity SSIM is determined based on the brightness, contrast and structure between the matching machine high-order quantization spectral image sample x and the golden machine high-order quantization spectral image sample y, and is expressed as the following formula:

[0020]

[0021] Where μ x represents the average value of x, μ y represents the mean value of y, σ x 2 represents the variance of x, σ y 2 represents the variance of y, σ xy represents the covariance of x and y, and c1 and c2 represent constants that maintain stability.

[0022] In one embodiment of the present invention, it is provided that the method further comprises:

[0023] Perform matching verification on production line quality control wafers, wherein when the matching error of the wafer is within the index range, perform periodic tracking verification; when the matching error of the wafer exceeds the index range, use the wafer as a sample to match the high-order quantized spectral image of the matching machine with the high-order quantized spectral image of the golden machine.

[0024] The present invention also provides an optical critical dimension measurement device, which uses the method to perform device matching.

[0025] In one embodiment of the present invention, it is stipulated that the optical critical dimension measurement equipment includes: an optical critical dimension measurement equipment of a dual rotating compensator, an optical critical dimension measurement equipment of a rotating polarizer-rotating compensator, an optical critical dimension measurement equipment of a rotating compensator-rotating analyzer, an optical critical dimension measurement equipment of a rotating polarizer, an optical critical dimension measurement equipment of a rotating analyzer, and a reflectometer-type optical critical dimension measurement equipment.

[0026] The present invention has at least the following beneficial effects:

[0027] (1) The two-dimensional and three-dimensional sample structures provided by the present invention can cover the main optical critical dimension measurement application scenarios.

[0028] (2) The high-order quantitative spectral image of the sample obtained on the golden machine is used as the matching target. Through the preprocessing regression algorithm, the global optimal value of the system parameters can be found.

[0029] (3) The use of a preprocessing regression algorithm combined with a structural similarity criterion can quickly and accurately achieve high-order quantitative spectral image matching, greatly improving the matching efficiency and matching accuracy of optical critical dimension measurement equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] To further illustrate the advantages and features of various embodiments of the present invention, a more detailed description of various embodiments of the present invention will be presented with reference to the accompanying drawings. It will be understood that these drawings only depict typical embodiments of the present invention and are therefore not to be considered as limiting the scope thereof. In the accompanying drawings, for clarity, identical or corresponding parts will be represented by identical or similar reference numerals.

[0031] Figure 1 A schematic flow chart of a spectrum matching method for an optical critical dimension measurement device in one embodiment of the present invention is shown.

[0032] Figure 2A -F shows a schematic diagram of the structure of two-dimensional and three-dimensional samples in one embodiment of the present invention.

[0033] Figure 3 A schematic diagram of a spectrum image of a gold machine in one embodiment of the present invention is shown.

[0034] Figure 4 A schematic diagram of a high-order quantitative spectral image of a golden machine in one embodiment of the present invention is shown. DETAILED DESCRIPTION

[0035] It should be noted that the components in the figures may be shown exaggeratedly for the sake of illustration and are not necessarily true to scale. In the figures, identical or functionally identical components are provided with the same reference numerals.

[0036] In the present invention, unless otherwise specified, "arranged on...", "arranged above..." and "arranged above..." do not exclude the existence of an intermediate between the two. In addition, "arranged on or above..." merely indicates the relative positional relationship between two components, and in certain cases, such as after reversing the product direction, it can also be converted into "arranged below or below...", and vice versa.

[0037] In the present invention, each embodiment is only intended to illustrate the aspects of the present invention and should not be construed as limiting.

[0038] In the present invention, unless otherwise specified, the quantifiers "a", "an" and "an" do not exclude the presence of a plurality of elements.

[0039] It should also be noted that in the embodiments of the present invention, for the sake of clarity and simplicity, only a portion of the parts or components may be shown, but those of ordinary skill in the art will understand that under the teachings of the present invention, the required parts or components may be added according to the needs of the specific scenario. In addition, unless otherwise specified, the features in different embodiments of the present invention may be combined with each other. For example, a feature in the second embodiment may be used to replace a corresponding or functionally identical or similar feature in the first embodiment, and the resulting embodiment also falls within the disclosure scope or recorded scope of the present application.

[0040] It should also be noted that within the scope of the present invention, the terms "same", "equal", "equal to" and the like do not mean that the values ​​of the two are absolutely equal, but allow a certain reasonable error, that is, the terms also cover "substantially the same", "substantially equal", "substantially equal to". By analogy, in the present invention, the terms "perpendicular to", "parallel to" and the like indicating directions also cover the meanings of "substantially perpendicular to" and "substantially parallel to".

[0041] In addition, the numbering of the steps of the methods of the present invention does not limit the execution order of the method steps. Unless otherwise specified, the method steps can be executed in different orders.

[0042] The present invention will be further described below in conjunction with specific embodiments with reference to the accompanying drawings.

[0043] Figure 1 FIG. 2 is a flow chart showing a spectrum matching method of an optical critical dimension measurement device in one embodiment of the present invention. Figure 1 As shown, the method comprises the following steps:

[0044] Step 101: Provide a plurality of two-dimensional and three-dimensional samples for optical critical dimension measurement.

[0045] Step 102: Collect the golden machine spectrum signal of the sample on the golden machine, convert the golden machine spectrum signal into a spectrum image, and then perform high-order quantization to generate a golden machine high-order quantized spectrum image.

[0046] Step 103: collecting a matching machine spectral signal of the sample on a matching machine, converting the matching machine spectral signal into a spectral image and then performing high-order quantization to generate a matching machine high-order quantized spectral image.

[0047] Step 104: Use a preprocessing regression algorithm to adjust the system parameters of the matching machine so that the matching machine high-order quantized spectral image matches the golden machine high-order quantized spectral image.

[0048] The steps of the method are described in detail below with reference to the accompanying drawings.

[0049] Figure 2A -F shows a schematic diagram of the structure of a two-dimensional and three-dimensional sample in one embodiment of the present invention. Figure 2A As shown in FIG. 101 , in step 101 , a set of two-dimensional and three-dimensional samples can be provided, which cover the main optical critical dimension measurement application scenarios. Figure 2A As shown, the first sample 201 is a two-dimensional shallow trench isolation (STI) structure, and the line-to-space ratio of the structure is 1:1. Figure 2B As shown, the second sample 202 is a two-dimensional STI structure with a line-space ratio of 1:2. Figure 2C As shown, the third sample 203 is a two-dimensional STI structure with a line-space ratio of 1:4. Figure 2D As shown, the fourth sample 204 is a two-dimensional poly AEI (After Etch Inspection) structure. The fifth sample 205 is a two-dimensional sigma etch (Sigma Etch) structure. The sixth sample 206 is a three-dimensional contact hole structure. In addition, those skilled in the art should understand that the above two-dimensional and three-dimensional sample structures are only examples and should not be used as limitations. Those skilled in the art can select appropriate sample structures according to actual needs. For example, the two-dimensional poly AEI structure is as follows Figure 2DThe line-to-space ratio shown is 1:1, and it can also be 1:2 or 1:4. It is also possible to add a two-dimensional gate spacer structure, a silicon germanium epitaxial (SiGeEPI) structure, etc. The three-dimensional structure can also add a three-dimensional PolyAEI, a three-dimensional Spacer, a three-dimensional source and drain shape (SSD), and a three-dimensional SiGe EPI structure. For the matching of process measurement equipment for dynamic random access memory (DRAM), active area etching structures (AAET, ActiveArea Etch), buried wordline etching structures (BWET, Buried Wordline Etch), bitline etching structures (BLET, Bitline Etch), and capacitor etching structures (CapET, Capacitor Etch) can also be used. For the matching of 3D NAND process measurement equipment, channel through hole etching structures (CHET, Channel Hole Etch) can be added.

[0050] Figure 3 A schematic diagram of a spectrum image of a gold machine in one embodiment of the present invention is shown. Figure 4 FIG. 1 is a schematic diagram of a high-order quantitative spectrum image of a gold machine in one embodiment of the present invention. In step 102, a full-band spectrum signal is collected for each sample wafer on the gold machine. The original spectrum is a one-dimensional analog signal curve graph. After the spectrum is loaded using a common analysis tool, Figure 3 As shown in the figure, it is not easy to distinguish intuitively because there are many overlaps. Figure 3 The high-order quantized spectral image of the golden machine generated after high-order quantization of the spectral image shown in the figure is as follows Figure 4 As shown. In this embodiment, since there are only 101 sampling points of the spectrum, the graphics obtained by quantization of 128 orders and above are actually exactly the same, that is, when the image bit number N=7, all the features of the original spectrum are obtained. In this step, different quantization step sizes (or encoding methods) are used to obtain full spectrum images under different step sizes. A quantization step size greater than 64 orders is recommended. Higher-order quantization can obtain more spectral details, but it will increase the amount of calculation. Furthermore, image recognition technology can be used to easily perform accurate quantitative calculations and analysis on a large number of spectra. In subsequent steps, this high-order quantized spectrum image will be used as the spectrum matching target of the matching machine.

[0051] In step 103, the matching machine can use the same settings as the golden machine to collect full-band spectral signals for the entire set of sample wafers, and the measurement points are exactly the same as those of the golden machine. They are also converted into corresponding spectral accurate images and subjected to high-order quantization.

[0052] In step 104, a pre-processing regression algorithm (Regression) is first provided, in which a low truncation error order (TO, Truncation Order) is used to calculate the optical critical dimension positive problem, and the size of the residual in the Gauss-Newton equation of the optical critical dimension positive problem is determined to determine the subsequent processing method. The Levenberg-Marquardt-Fletcher (LMF) algorithm in the trust region format is used to solve the case where the residual is small or the linearity is high. For the case of large residuals or high nonlinearity, the quasi-Newton method is used to approximate the second-order Hessian matrix at the local optimal solution, and the Gauss-Newton equation is modified to improve the convergence speed.

[0053] In step 105, on the optical critical dimension modeling platform, the pre-processing regression algorithm is used to adjust the relevant parameters using the high-order quantized spectral image measured by the golden machine as the matching target value. The optical critical dimension model used by the golden machine is used, the same geometric parameters are floated, and the system parameters used for equipment calibration are floated, and the system parameters include the initial azimuth angle (P0) of the polarizer, the initial azimuth angle (CA0) of the compensator A, the initial azimuth angle (CB0) of the compensator B, the initial azimuth angle (A0) of the analyzer, the angle of incidence (AOI), the initial azimuth angle (AZ) of the motion stage, and the initial height offset (Zoffset) in the Z direction. The high-order quantized spectral image obtained on the matching machine is made to match the high-order quantized spectral image obtained on the golden machine as much as possible. In other words, for this batch of sample wafers, a set of global optimal values ​​of system parameters are found so that the spectrum obtained by the matching machine under the adjustment of these system parameters can well match the golden machine. When the spectra obtained on the matching machine and the golden machine are close enough, or the residual is small enough, the calculated optical critical dimension value will also be well matched in conjunction with the optical critical dimension database.

[0054] When matching high-order quantized spectral images through preprocessing regression algorithms, structural similarity (SSIM) is used as an evaluation indicator for the quality of the match. Structural similarity is an indicator that measures the similarity between two images. It is often used to measure the similarity between images before and after distortion, and is also used to measure the authenticity of images generated by models, such as image de-raining, image de-fogging, image harmonization, etc. SSIM compares and measures based on the brightness, contrast, and structure between samples x and y, and is expressed as the following formula:

[0055]

[0056] Among them, μ x represents the average value of x, μ y represents the mean value of y, σ x 2 represents the variance of x, σ y 2 represents the variance of y, σ xy represents the covariance of x and y, c1 and c2 represent constants that maintain stability. c1=(k1L) 2 ,c2=(k2L) 2 , k1=0.01, k2=0.03,L.

[0057] Specifically, the image can be divided into N areas according to the band and sample type (for example, the band type can be divided into 190-250nm, 250-500nm, 500-1000nm, 1000-1500nm, 1500-2000nm, and the sample type can be divided into logic chip, DRAM chip, NAND chip, etc.), and the structural similarity index SSIM between the machine to be matched and the golden machine is calculated based on the image in each small area. i ,i=1,…,N.

[0058] Then calculate the average similarity index with regularization term as the evaluation function RMSSIM(x,y):

[0059]

[0060] Where α represents the regularization coefficient. The maximum value of the objective function is 1, and the matching problem becomes the optimization problem maxRMSSIM(x,y).

[0061] The specific iteration process is as follows:

[0062] Step 0: Assume that the image of the golden machine is y, and the initial value of the image of the machine to be matched is x0, set k:=0,

[0063] Step 1. Calculation , if it is greater than the threshold , the iteration is terminated with As the best match with the golden machine.

[0064] Step 2: Calculation exist The numerical derivative of the key component of x (such as a specially selected wavelength) at and , and the directional derivative .

[0065] Step 3: Do a one-dimensional line search in the direction to get the best correction step size λ k .

[0066] Step 4: ,k:=k+1, go to step 1.

[0067] The range of structural similarity is -1 to 1. When two images are exactly the same, the value of SSIM is equal to 1. As an implementation of the structural similarity theory, the structural similarity index defines structural information from the perspective of image composition as an attribute that is independent of brightness and contrast, reflecting the structure of objects in the scene, and models distortion as a combination of three different factors: brightness, contrast, and structure. The mean is used as an estimate of brightness, the standard deviation is used as an estimate of contrast, and the covariance is used as a measure of structural similarity. Due to its outstanding performance, SSIM has become a widely used method for measuring video quality in broadcasting and cable television. It is widely used in super-resolution and image deblurring.

[0068] Based on the system parameter values ​​obtained by the preprocessing regression method, the system parameters of the optical key dimension measurement equipment can be adjusted in the machine host computer calibration software to make the measured spectrum match the golden machine.

[0069] Furthermore, the matching verification of the production line quality control wafers can be performed. If the matching error is within the index range, periodic tracking verification can be performed later. If the matching error of some structures exceeds the index range, the wafers of this part of the structure can be added to the sample, and steps 101 to 104 can be repeated.

[0070] This method can be applied to the matching of various optical critical dimension measurement equipment, including optical critical dimension measurement equipment of dual rotating compensators, optical critical dimension measurement equipment of rotating polarizer and rotating compensator, optical critical dimension measurement equipment of rotating compensator and rotating analyzer, optical critical dimension measurement equipment of rotating polarizer, optical critical dimension measurement equipment of rotating analyzer, reflectometer-type optical critical dimension measurement equipment, etc.

[0071] Although various embodiments of the present invention are described above, it should be understood that they are presented as examples only and not as limitations. It is obvious to those skilled in the relevant art that various combinations, modifications and changes can be made thereto without departing from the spirit and scope of the present invention. Therefore, the breadth and scope of the present invention disclosed herein should not be limited by the exemplary embodiments disclosed above, but should only be defined according to the attached claims and their equivalents.

Claims

1. A spectrum matching method for optical critical dimension measurement equipment, characterized in that: The following steps are involved: Provide multiple 2D and 3D samples for optical critical dimension measurement; Collecting a gold machine spectral signal of the sample on a gold machine, converting the gold machine spectral signal into a spectral image and then performing high-order quantization to generate a gold machine high-order quantized spectral image; Collecting a matching machine spectral signal of the sample on a matching machine, converting the matching machine spectral signal into a spectral image and then performing high-order quantization to generate a matching machine high-order quantized spectral image; as well as The system parameters of the matching machine are adjusted using a preprocessing regression algorithm to match the matching machine high-order quantization spectral image with the golden machine high-order quantization spectral image, wherein the degree of matching between the matching machine high-order quantization spectral image and the golden machine high-order quantization spectral image is determined based on the brightness, contrast and structure between the matching machine high-order quantization spectral image sample x and the golden machine high-order quantization spectral image sample y, and is expressed as the following formula: Among them, μ x represents the average value of x, μ y represents the mean value of y, represents the variance of x, represents the variance of y, σ xy represents the covariance of x and y, c1 and c2 represent constants that maintain stability; The preprocessing regression algorithm includes: Use low truncation error order to calculate the optical critical dimension positive problem and determine the size of the residual in the Gauss-Newton equation of the optical critical dimension positive problem; When the residual amount or the degree of linearity is greater than the judgment value, solving the problem using the Levenberg-Marquardt-Fletcher algorithm in a trust region format; and When the residual amount or the degree of linearity is less than or equal to the judgment value, the quasi-Newton method is used to approximately calculate the second-order Hessian matrix at the local optimal solution to correct the Gauss-Newton equation.

2. The spectrum matching method of optical critical dimension measurement equipment according to claim 1, characterized in that: The two-dimensional and three-dimensional samples include one or more of the following: two-dimensional shallow trench isolation structure, two-dimensional post-etching detection structure, two-dimensional sigma etching structure, three-dimensional contact hole structure, two-dimensional gate sidewall structure, two-dimensional silicon germanium epitaxial structure, three-dimensional post-etching detection structure, three-dimensional gate sidewall structure, three-dimensional source and drain shape structure, three-dimensional silicon germanium epitaxial structure, active area etching structure, buried word line etching structure, bit line etching structure, capacitor etching structure and channel through hole etching structure.

3. The spectrum matching method of optical critical dimension measurement equipment according to claim 1, characterized in that: The measurement points of the matching machine spectrum signal collected on the matching machine for the sample are the same as the measurement points of the golden machine spectrum signal collected on the golden machine for the sample.

4. The spectrum matching method of optical critical dimension measurement equipment according to claim 1, characterized in that: The system parameters include: an initial azimuth angle of the polarizer, an initial azimuth angle of the first compensator, an initial azimuth angle of the second compensator, an initial azimuth angle of the analyzer, an incident angle, an initial azimuth angle of the moving stage, and an initial height offset in the Z direction.

5. The spectrum matching method of optical critical dimension measurement equipment according to claim 1, characterized in that: Also includes: Perform matching verification of wafers for production line quality control, and perform periodic tracking verification when the wafer matching error is within the indicator range; When the matching error of the wafer exceeds the index range, the wafer is used as a sample to match the high-order quantized spectrum image of the matching machine with the high-order quantized spectrum image of the golden machine.

6. An optical key dimension measurement device, characterized in that: Device matching is performed using the method described in any one of claims 1-5.

7. The optical critical dimension measuring device according to claim 6, characterized in that: include: Optical critical dimension measuring equipment of dual rotating compensators, optical critical dimension measuring equipment of rotating polarizer and rotating compensator, optical critical dimension measuring equipment of rotating compensator and rotating analyzer, optical critical dimension measuring equipment of rotating polarizer, optical critical dimension measuring equipment of rotating analyzer and reflectometer type optical critical dimension measuring equipment.

Images