A non-imaging semiconductor overlay error measurement device and method

Abstract

The application discloses a kind of non-imaging semiconductor overlay error measurement device and method, including building bright field single-pixel imaging light path;Based on single-pixel imaging light path under the premise of non-imaging Fourier frequency domain acquisition;Two-dimensional cross-correlation coefficient distribution is obtained by the calculation of Fourier frequency domain;Profile center subpixel positioning is carried out using two-dimensional cross-correlation coefficient distribution;Overlay error is calculated by the difference calculation of the subpixel positioning of two profiles.The application realizes the premise without imaging, only through light intensity measurement direct positioning detection target subpixel profile center and calculating overlay error, greatly reduce the data amount generated by graph operation, improve measurement efficiency, expand the application of single-pixel imaging technology in the field of semiconductor overlay error measurement.

Description

Technical Field

[0001] This invention relates to a non-imaging semiconductor overlay error measurement device and method, belonging to the field of microelectronics and semiconductor integrated circuit manufacturing. Background Technology

[0002] Semiconductor chip technology is vital to national economic development and information security, and photolithography is a crucial step in semiconductor integrated circuit manufacturing. It involves a series of repetitive processes such as alignment, exposure, and development to transfer the mask pattern onto the wafer. The pattern remaining on the photoresist after exposure and development (the current layer pattern) must be aligned with the existing pattern on the wafer substrate (the previous layer pattern). Overlay error, the offset between the upper and lower layers of the pattern on the wafer, must meet certain requirements to ensure correct connections between different parts of the device. Excessive overlay error is a major cause of short circuits and open circuits in devices; therefore, overlay error measurement is of great significance and application value for improving the yield rate of semiconductor manufacturing. However, currently, the vast majority of overlay error inspection equipment relies on imports, leaving key technologies dependent on foreign sources, seriously threatening the healthy development of my country's semiconductor industry.

[0003] In integrated circuit manufacturing, specialized equipment is typically used to detect the relative position between the current layer pattern and the previous layer pattern on a photolithographically lithographic wafer to determine overlay error. Overlay error quantitatively describes the deviation of the current pattern relative to the previous layer pattern in both the X and Y directions, as well as the distribution of this deviation on the wafer, and is one of the key indicators for evaluating the quality of the photolithography process. The patterns on the wafer specifically used to calculate overlay error are called overlay markers, and these patterns are usually located at the edges of the exposure cells. Currently, commonly used techniques in the field of semiconductor surface overlay error measurement technology include image recognition-based overlay error measurement (IBO) and diffraction-based overlay error measurement (DBO).

[0004] The overlay error measurement based on image recognition technology mainly involves the following steps: Specific overlay marks are fabricated using photolithography on the current and previous layers. These marks do not overlap and their center coordinates are easily calculated. With appropriate selection of camera, lens, and light source, high-precision imaging is performed on the designated overlay mark area of ​​the semiconductor according to a planned route. High-resolution images of a single measurement object are obtained using graphics techniques such as denoising, rotation correction, edge trimming, image segmentation, and image interpolation. Algorithm processing is applied to the double-layer overlay mark image to obtain the center coordinates of the overlay mark contour. The overlay error is obtained by calculating the difference in center coordinates between the two layers of overlay marks. However, this technology has the following problems: 1. The imaging process for obtaining high-resolution local images significantly increases the processing time and reduces detection efficiency; 2. Using intelligent algorithms for overlay mark alignment will generate huge data throughput, which puts enormous pressure on the communication servers used in production. High-end servers are also heavily controlled by foreign patents, which could easily become a new bottleneck problem; 3. As semiconductor manufacturing processes become smaller and camera spatial resolution increases, the problems of insufficient efficiency and excessive data pressure of the original technology will become increasingly serious.

[0005] The main process of diffraction-based overlay error measurement is as follows: Diffraction grating-type photolithographic marks are fabricated on the current layer and reference layer using photolithography; both marks are periodic structures. The overlay mark area is illuminated using a light source. The intensity of the +1 and -1 order diffracted light under the illumination is measured. Utilizing the good linear relationship between the light intensity difference and the overlay error value when overlay error exists, the measurement equipment can calculate the overlay situation between different process layers by measuring the light intensity difference. Compared to IBO overlay marks, DBO marks have the advantages of smaller measurement error and higher measurement repeatability. However, diffraction-based overlay error measurement has the following problems: the relevant patents are monopolized by the Dutch company ASML and are used in their YieldStar measurement instruments. Continuing to use this technology could easily become a new bottleneck problem.

[0006] The industry now has increasingly higher requirements for controlling overlay error, and the measurement of overlay error also faces enormous challenges. How to improve the measurement speed of overlay error, improve accuracy, reduce data throughput, and solve the bottleneck problems of key technologies are the main problems in integrated circuit manufacturing processes. Summary of the Invention

[0007] This invention aims to address the shortcomings of existing technologies by providing a non-imaging semiconductor overlay error measurement device and method. It is a novel non-imaging method for achieving high measurement efficiency, low data throughput, and high precision overlay error measurement.

[0008] Technical solution of the present invention:

[0009] Firstly, a dual-modal single-pixel imaging optical path is proposed, including an active single-pixel bright-field imaging module and a passive single-pixel dark-field imaging module. A beam splitter switches between the active single-pixel bright-field imaging module and the passive single-pixel dark-field imaging module. The active single-pixel bright-field imaging module performs single-pixel Fourier frequency domain acquisition on the detected target to obtain the bright-field Fourier frequency domain of the detected target. The passive single-pixel dark-field imaging module performs single-pixel Fourier frequency domain acquisition on the detected target to obtain the dark-field Fourier frequency domain of the detected target.

[0010] The active single-pixel bright-field imaging module includes:

[0011] light source;

[0012] Spatial light modulator;

[0013] Imaging lens;

[0014] Concave mirror;

[0015] Photoelectric sensors;

[0016] Spectrometer;

[0017] In the active bright-field single-pixel imaging module, light is emitted from the light source and illuminates the outer inclined surface of the beam splitter at a 45-degree angle. Half of the light component passes through the inclined surface without changing direction, and is converged and reflected by the concave radiating mirror before illuminating the spatial light modulator. The spatial light modulator modulates the light to obtain a structured light field. The structured light field is reflected to the imaging lens. The structured light field is projected onto the surface of the detection target after passing through the imaging lens. The surface of the detection target reflects the structured light field, which then illuminates the target surface of the photoelectric sensor. The photoelectric sensor measures the total intensity of the structured light field, and by calculating the correlation between the structured light field intensity and the structured light field, the Fourier frequency value of the target at a specified position in the bright-field Fourier frequency domain is obtained. The bright-field Fourier frequency domain of the detection target is obtained through frequency domain scanning.

[0018] The passive single-pixel dark-field imaging module includes:

[0019] light source;

[0020] Spatial light modulator;

[0021] Imaging lens;

[0022] Concave mirror;

[0023] Photoelectric sensors;

[0024] Spectrometer;

[0025] In the passive dark-field single-pixel imaging module, light is emitted from the light source and illuminates the surface of the target, where it undergoes diffuse reflection. The resulting diffuse light passes through an imaging lens and is imaged onto a spatial light modulator. The spatial light modulator modulates the light to obtain a structured light field. The structured light field is reflected onto the surface of a concave mirror. After being converged and reflected by the concave mirror, the structured light field is incident on the inner inclined surface of a beam splitter at a 45-degree angle. After being reflected by the inner inclined surface of the beam splitter, the structured light field illuminates the target surface of a photodetector. The photodetector measures the total intensity of the structured light field, and by calculating the correlation between the intensity and the structured light field, the Fourier frequency value at a specified position in the dark-field Fourier frequency domain of the target is obtained. Through frequency domain scanning, the dark-field Fourier frequency domain of the target is obtained.

[0026] Secondly, a dual-modal single-pixel imaging method is proposed, including: active single-pixel bright-field imaging and passive single-pixel dark-field imaging; switching between active single-pixel bright-field imaging and passive single-pixel dark-field imaging is achieved through a beam splitter; active single-pixel bright-field imaging performs single-pixel Fourier frequency domain acquisition on the target to obtain the bright-field Fourier frequency domain of the target; passive single-pixel dark-field imaging performs single-pixel Fourier frequency domain acquisition on the target to obtain the dark-field Fourier frequency domain of the target.

[0027] In active bright-field single-pixel imaging, light is emitted from the light source and illuminates the outer inclined surface of the beam splitter at a 45-degree angle. Half of the light component passes through the inclined surface without changing direction, and is converged and reflected by the concave radiating mirror before illuminating the spatial light modulator. The spatial light modulator modulates the light to obtain a structured light field. The structured light field is reflected to the imaging lens. The structured light field is projected onto the surface of the target being detected through the imaging lens. The target surface reflects the structured light field, which then illuminates the target surface of the photoelectric sensor. The photoelectric sensor measures the total intensity of the structured light field, and by calculating the correlation between the structured light field intensity and the structured light field, the Fourier frequency value of the target at a specified position in the bright-field Fourier frequency domain is obtained. Through frequency domain scanning, the bright-field Fourier frequency domain of the target is obtained.

[0028] In passive dark-field single-pixel imaging, light is emitted from the light source and illuminates the surface of the target, where it undergoes diffuse reflection. The resulting diffuse light passes through an imaging lens and is imaged onto a spatial light modulator. The spatial light modulator modulates the light to obtain a structured light field. This structured light field is reflected onto the surface of a concave mirror. After being converged and reflected by the concave mirror, the structured light field is incident on the inner inclined surface of a beam splitter at a 45-degree angle. After being reflected by the inner inclined surface of the beam splitter, the structured light field illuminates the target surface of a photodetector. The photodetector measures the total intensity of the structured light field and calculates the correlation between the intensity and the structured light field to obtain the Fourier frequency value at a specified position in the dark-field Fourier frequency domain of the target. Through frequency domain scanning, the dark-field Fourier frequency domain of the target is obtained.

[0029] Thirdly, a non-imaging semiconductor overlay error measurement device is proposed, comprising a dual-mode single-pixel imaging optical path and a computing unit; the Fourier frequency domain of the target is acquired non-imagingly through the dual-mode single-pixel imaging optical path; the Fourier frequency domain is processed by the computing unit to obtain a two-dimensional cross-correlation coefficient distribution; the center coordinates of the target contour with sub-pixel accuracy are calculated based on the two-dimensional cross-correlation coefficient distribution; and the semiconductor overlay error is calculated based on the center coordinates of the target contour with sub-pixel accuracy.

[0030] Fourthly, a non-imaging semiconductor overlay error measurement method is proposed, which is implemented as follows:

[0031] S1: Using a dual-modal single-pixel imaging optical path, combined with the Fourier single-pixel imaging principle, the Fourier frequency domain distribution of the target under test is obtained without imaging.

[0032] S2: Calculate the two-dimensional cross-correlation distribution from the Fourier frequency domain distribution of the target under test;

[0033] S3: Based on the two-dimensional cross-correlation coefficient distribution, obtain the sub-pixel precision coordinates of the center of the target contour;

[0034] S4: Calculate the sub-pixel precision coordinates of the center of the target contour to obtain the semiconductor overlay error.

[0035] Compared with the prior art, the present invention has the following advantages:

[0036] (1) Existing image recognition-based interpolation error (IBO) measurement methods rely on high spatial resolution imaging equipment and image post-processing algorithms for IBO measurement. With the rapid increase in pixel density of modern cameras, this generates huge data throughput, which limits further improvement in efficiency. However, this invention utilizes the imaging optical path to calculate the IBO error from the bright-field Fourier frequency domain of the detected target without image reconstruction, thus effectively reducing data throughput and improving measurement efficiency.

[0037] (2) Existing diffraction-based overlay error measurement (DBO) methods, while accurate due to the use of diffraction principles, are highly sensitive to environmental vibrations; while the method proposed in this invention uses a dual-modal single-pixel imaging optical path, thus exhibiting higher environmental robustness.

[0038] (3) In addition, the device proposed in this invention has the advantages of compact structure, low manufacturing cost, high dynamic range and low computational complexity. Attached Figure Description

[0039] Figure 1 This is a schematic diagram of the dual-modal single-pixel imaging optical path of the present invention;

[0040] Figure 2 This is a flowchart of the non-imaging semiconductor overlay error measurement method of the present invention;

[0041] Figure 3 This is the target to be tested used in Embodiment 2 of the present invention;

[0042] Figure 4 This is the normalized result of the two-dimensional cross-correlation distribution calculated in Embodiment 2 of the present invention;

[0043] Figure 5 This is the interpolation result for one row of the two-dimensional cross-correlation. Detailed Implementation

[0044] This invention provides a non-imaging semiconductor overlay error measurement device and technology. The device has a compact structure, high dynamic range, and low data throughput. The technology has low computational complexity, high detection efficiency, and high measurement accuracy.

[0045] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below through specific implementations and in conjunction with the accompanying drawings.

[0046] Example 1

[0047] like Figure 1 As shown, the dual-modal single-pixel imaging optical path provided in this embodiment of the invention includes:

[0048] Light source 101; In this embodiment of the invention, a high-brightness broadband white LED bead is used as the light source for bright-field imaging optical path.

[0049] Light source 102; In this embodiment of the invention, a broadband white LED panel is used as the light source for dark field imaging.

[0050] Photoelectric sensor 103; used to measure the total intensity of modulated bright-field structured light;

[0051] Photoelectric sensor 104; used to measure the total light intensity of modulated dark field structured light;

[0052] Beam splitter 105; located between the bright field light source and the dark field detector, forming a 45° beam splitting path with the bright field and dark field optical paths at the beam splitting interface; in this embodiment, the beam splitter is used to integrate the bright field optical path and the dark field optical path into one device to achieve optical path multiplexing.

[0053] Spatial light modulator 106; used to modulate the amplitude of the light field focused on its surface; in this embodiment, a digital micromirror device is used as the spatial light modulator, which can perform reflective binarized amplitude modulation.

[0054] Concave mirror 107; used to adjust the position of the focal plane and change the direction of the optical path, and to improve the compactness of the device;

[0055] Imaging lens 108; used to accurately project an image of an object onto a spatial light modulation surface or to accurately focus projected structured light onto the surface of the object under test;

[0056] Target 109; in semiconductor overlay measurement processes, it is usually an overlay mark.

[0057] This embodiment uses a beam splitter to switch between bright field and dark field lighting modes.

[0058] Among them, the bright field mode is as follows Figure 1 As shown by the solid arrow, the light source 101 provides illumination. The illumination light passes directly through the beam splitter 105 and is focused by the reflection of the concave mirror 107, projecting the light onto the surface of the spatial light modulator 106. Here, the light field is binarized, the light field amplitude is encoded, and reflected. The reflected structured light field is projected onto the detection target 109 through the imaging lens 108. The reflected light from the object's surface reaches the target surface of the photoelectric sensor 103, and the total light intensity is measured.

[0059] Among them, the dark field mode is as follows Figure 1 As indicated by the dashed arrow, the target 109 is illuminated in dark by the light source 102. The light diffusely reflected from the target surface is collected by the imaging lens 108 and focused onto the surface of the spatial light modulator 106. The spatial light modulator 106 encodes the amplitude of the light field and reflects it. The reflected structured light field is reflected and converged by the concave mirror 107 and reflected at the beam splitter interface of the beam splitter 105. The reflected light reaches the target surface of the photoelectric sensor 104, and the total light intensity is measured.

[0060] According to embodiments of the present invention, a single-pixel bright and dark field dual-modal imaging device is proposed, which can effectively utilize the high dynamic characteristics of a single-pixel sensor to solve the common problems of local strong reflection light and insufficient contrast in semiconductor defect detection, enabling the device to adapt to more different surface imaging conditions in the semiconductor overlay detection process.

[0061] Example 2

[0062] like Figure 2 As shown, a method for calculating non-imaging semiconductor overlay errors based on the single-pixel imaging principle is implemented. The specific steps are as follows:

[0063] S1: Utilizing a dual-modal single-pixel imaging optical path, combined with the Fourier single-pixel imaging principle, the Fourier frequency domain distribution of the target under test is obtained without imaging. Based on the device provided in Embodiment 1, a suitable optical path is selected according to the reflectivity distribution characteristics of the imaging target to obtain the optimal contrast. In this embodiment, a bright-field optical path will be used for explanation, and its optical path diagram is as follows: Figure 1As shown, it includes a light source 101, a photoelectric sensor 103, a beam splitter 105, a spatial light modulator 106, a concave mirror 107, an imaging lens 108, and a detection target 109.

[0064] The object under test is placed in a light-proof environment. Light emitted from the light source emitter is modulated by a spatial light modulator and reflected onto the object under test. A Fourier modulation template P is loaded onto the spatial light modulator.

[0065] P(x,y|f x ,f y ,φ)=A+B+cos[2π(f x +f y )+φ], (1)

[0066] Where (x,y) represents the two-dimensional coordinates in the spatial domain, A represents the average intensity or DC term of the pattern, and B represents the contrast. Assuming the intensity distribution of the two-dimensional image is I(x,y), then the modulation and backscattering energy E φ (f x ,f y The inner product of I(x,y) and P(x,y) can be expressed as follows:

[0067] E φ (f x ,f y )=∫∫I(x,y){A+B·cos2π(f x x+f y y)+φ}dxdy (2)

[0068] If D φ (f x ,f y ) represents the detector's response to the light field, and it contains two parts as shown in the following formula.

[0069] D φ (f x ,f y )=m·E φ (f x ,f y )+D n (3)

[0070] One is the sensor's modulation of the light field E φ (f x ,f y The linear response of the sensor is given by the linear coefficient m, where m is an inherent property of the sensor itself. nThis represents ambient noise, which can be considered a short-term constant between two short modulation intervals. Next, the Fourier frequency domain coefficients are calculated using the four-step phase-shift method. One Fourier coefficient is obtained for every four spatial light illuminations. These four structured light templates share the same frequency pair (f...). x ,f y The four phases are 0, π / 2, π, and 3π / 2, respectively. The Fourier series C(f) obtained by this technique is... x ,f y This can be represented as:

[0071]

[0072] This allows obtaining any Fourier frequency domain for any specified region; this embodiment uses the above-described technique to obtain, for example... Figure 3 The Fourier frequency domain of the overlay marker shown within the dashed box indicates that the pixel size of the overlay target is M pixels wide and N pixels high, and the selection box used is K pixels wide. Since this invention directly obtains partial frequency values ​​from the Fourier frequency domain of the one-dimensional electrical signal, subsequent calculations can be performed without imaging, thus effectively avoiding the data acquisition and computational burden caused by imaging and post-processing algorithms.

[0073] S2: Calculate the two-dimensional cross-correlation distribution from the Fourier frequency domain distribution of the target under test;

[0074] This embodiment uses the Fourier single-pixel imaging principle to obtain the Fourier coefficients of the left and right blocks without imaging. The Fourier distributions of these two blocks are respectively calculated using f1(f x ,f y f2(f) x ,f y This is represented by (). Based on the classical graphics operation that frequency domain multiplication is equivalent to spatial domain convolution, this embodiment obtains the two-dimensional cross-correlation distribution CC of the two regions. 2D (x,y), its mathematical representation is:

[0075] CC 2D (x,y)=F -1 { f1(f x ,f y )·f2(f x ,f y (5)

[0076] In this embodiment, the gray-level normalization result of the two-dimensional cross-correlation distribution obtained by the above technique is as follows: Figure 4As shown, black represents 0 and white represents 1; it can be clearly observed that the correlation reaches its maximum value under specific displacement conditions. Since the calculation process only involves simple two-dimensional fast Fourier transform and sorting, the computational complexity is extremely low, which can effectively improve the efficiency of overlay error calculation and reduce computational and communication pressure.

[0077] S3: Based on the two-dimensional cross-correlation coefficient distribution, obtain the sub-pixel precision coordinates of the center of the target contour;

[0078] Based on the gray-level normalization result of the obtained two-dimensional cross-correlation distribution, the contour center coordinates of the target pattern with sub-pixel precision are further calculated.

[0079] For each row of the two-dimensional cross-correlation distribution, its one-dimensional correlation distribution can be extracted. The extraction result for the i-th row is as follows: Figure 5 As shown, cubic spline interpolation is performed on it to find the sub-pixel correlation peak position, denoted as P. f (i). The final average sub-pixel correlation peak position P is shown in the following formula:

[0080]

[0081] Where N is the pixel height of the detected object. By using cubic splines, interpolation, and averaging, this algorithm fully utilizes all acquired two-dimensional correlation data, thus achieving high accuracy.

[0082] Final horizontal profile center position Cor L It is given by the following formula:

[0083]

[0084] Where P is the peak position of the average subpixel correlation in the horizontal direction, M is the horizontal pixel width of the detected object, and K is the horizontal pixel width of the selection box used.

[0085] For each column of the two-dimensional cross-correlation distribution, its one-dimensional correlation distribution can also be extracted, and cubic spline interpolation can be performed on it to find the sub-pixel correlation peak position, denoted as P. f (j). The final average sub-pixel correlation peak position P is shown in the following equation:

[0086]

[0087] Where N is the pixel height of the detected object. By using cubic splines, interpolation, and averaging, this algorithm fully utilizes all acquired two-dimensional correlation data, thus achieving high accuracy.

[0088] Final vertical profile center position Cor VIt is given by the following formula:

[0089]

[0090] Where P is the calculated peak position of the average subpixel correlation in the vertical direction, N is the vertical pixel width of the detected object, and K is the vertical pixel height of the selection box used.

[0091] Repeat step S3 to measure the center coordinates of the outlines of the previous layer graphic and the current layer graphic respectively.

[0092] S4: Calculate the subpixel precision coordinates of the center of the target contour, that is, subtract the center coordinates of the contours of the previous layer graphic and the current layer graphic to obtain the semiconductor overlay error.

[0093] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0094] In the description of this specification, the terms "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine the different embodiments or examples described in this specification and the features of the different embodiments or examples.

[0095] Combination and combination.

[0096] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A non-imaging semiconductor overlay error measurement device, characterized in that, It includes a dual-modal single-pixel imaging optical path and a computing unit. Through the dual-modal single-pixel imaging optical path, it acquires the Fourier frequency domain of a designated region of the detected target without imaging, directly obtaining partial frequency values ​​of the Fourier frequency domain from the one-dimensional electrical signal. This allows for subsequent calculations without imaging, effectively avoiding the data acquisition and computational burden caused by imaging and post-processing algorithms. The computing unit processes the Fourier frequency domain of the designated region to obtain a two-dimensional cross-correlation coefficient distribution. The correlation of the two-dimensional cross-correlation coefficient distribution reaches its maximum value under specific displacement conditions. The calculation process only involves a simple two-dimensional fast Fourier transform and sorting. The computational complexity is extremely low, effectively improving the efficiency of overlay error calculation and reducing computational and communication pressure. Based on the two-dimensional cross-correlation coefficient distribution, cubic spline interpolation is performed on the cross-correlation peaks to find the sub-pixel correlation peak positions. The sub-pixel peak positions are averaged to obtain the sub-pixel precision coordinates of the target contour center. Since cubic splines, interpolation, and averaging are used, all acquired two-dimensional correlation data are fully utilized, thus achieving high accuracy. The sub-pixel precision coordinates of the target contour center are calculated by subtracting the contour center coordinates of the previous and current layers of the image to obtain the semiconductor overlay error.

2. A method for measuring non-imaging semiconductor overlay error, characterized in that, The implementation is as follows: S1: Utilizing a dual-mode single-pixel imaging optical path and combining the Fourier single-pixel imaging principle, the Fourier frequency domain of a specified region of the target under test can be obtained without imaging. Partial frequency values ​​of the Fourier frequency domain can be directly obtained from the one-dimensional electrical signal, enabling subsequent calculations to be performed without imaging, effectively avoiding the pressure of data acquisition and calculation caused by imaging and post-processing algorithms. S2: Based on the Fourier frequency domain of the specified region of the target to be measured, the two-dimensional cross-correlation coefficient distribution is obtained. The correlation of the two-dimensional cross-correlation coefficient distribution reaches its maximum value under specific displacement conditions. The calculation process only involves simple two-dimensional fast Fourier transform and sorting, with extremely low computational complexity, effectively improving the efficiency of overlay error calculation and reducing computational and communication pressure. S3: Based on the two-dimensional cross-correlation coefficient distribution, cubic spline interpolation is performed on the cross-correlation peak to find the sub-pixel correlation peak position. The sub-pixel peak positions are averaged to obtain the sub-pixel precision coordinates of the target contour center. Since cubic spline interpolation and averaging are used, all the collected two-dimensional correlation data are fully utilized, thus achieving high accuracy. S4: Calculate the sub-pixel precision coordinates of the center of the target contour, that is, calculate the difference between the center coordinates of the contours of the previous layer graphic and the current layer graphic to obtain the semiconductor overlay error.

3. The non-imaging semiconductor overlay error measurement device according to claim 1, characterized in that: The dual-modal single-pixel imaging optical path includes: An active single-pixel bright-field imaging module; a passive single-pixel dark-field imaging module; switching between the active single-pixel bright-field imaging module and the passive single-pixel dark-field imaging module via a beam splitter; the active single-pixel bright-field imaging module performs single-pixel Fourier frequency domain acquisition on the detected target to obtain the bright-field Fourier frequency domain of the detected target; the passive single-pixel dark-field imaging module performs single-pixel Fourier frequency domain acquisition on the detected target to obtain the dark-field Fourier frequency domain of the detected target; The imaging optical path further includes: a bright-field light source, a dark-field light source, a spatial light modulator, an imaging lens, a concave mirror, a beam splitter, a first photoelectric sensor, and a second photoelectric sensor; the beam splitter is located between the bright-field light source and the second photoelectric sensor, forming a 45° beam splitting path with the bright-field optical path and the dark-field optical path at the beam splitting interface; the concave mirror is used to adjust the focal plane position and change the direction of the optical path to improve the compactness of the device; In the active bright-field single-pixel imaging module: bright-field light is emitted from a bright-field light source and incident on a concave mirror through a beam splitter. The concave mirror converges and reflects the bright-field light to a spatial light modulator. The spatial light modulator modulates the light to obtain a structured light field. The structured light field is projected onto the surface of the target being detected through an imaging lens. The structured light field reflected from the target surface reaches the target surface of the first photoelectric sensor. The first photoelectric sensor measures the total intensity of the structured light field and calculates the Fourier frequency value at a specified position in the bright-field Fourier frequency domain based on the correlation between the intensity and the structured light field. The bright-field Fourier frequency domain of the target being detected is obtained through frequency domain scanning. In the passive dark-field single-pixel imaging module: dark-field light is emitted from the dark-field light source to illuminate the target and diffusely reflect off the target surface; the diffuse light is imaged onto the spatial light modulator by the imaging lens; the spatial light modulator modulates the light to obtain a structured light field; the structured light field is converged and reflected by the concave mirror and then incident on the beam splitter interface and reflected by the beam splitter interface to the target surface of the second photoelectric sensor; the second photoelectric sensor measures the total intensity of the structured light field and calculates the Fourier frequency value at a specified position in the dark-field Fourier frequency domain by the correlation between the intensity of the structured light field and the structured light field; the dark-field Fourier frequency domain of the target is obtained by frequency domain scanning.

4. The method for measuring non-imaging semiconductor overlay error according to claim 2, characterized in that: The dual-modal single-pixel imaging optical path includes: An active single-pixel bright-field imaging module; a passive single-pixel dark-field imaging module; switching between the active single-pixel bright-field imaging module and the passive single-pixel dark-field imaging module via a beam splitter; the active single-pixel bright-field imaging module performs single-pixel Fourier frequency domain acquisition on the detected target to obtain the bright-field Fourier frequency domain of the detected target; the passive single-pixel dark-field imaging module performs single-pixel Fourier frequency domain acquisition on the detected target to obtain the dark-field Fourier frequency domain of the detected target; The imaging optical path further includes: a bright-field light source, a dark-field light source, a spatial light modulator, an imaging lens, a concave mirror, a beam splitter, a first photoelectric sensor, and a second photoelectric sensor; the beam splitter is located between the bright-field light source and the second photoelectric sensor, forming a 45° beam splitting path with the bright-field optical path and the dark-field optical path at the beam splitting interface; the concave mirror is used to adjust the focal plane position and change the direction of the optical path to improve the compactness of the device; In the active bright-field single-pixel imaging module: bright-field light is emitted from a bright-field light source and incident on a concave mirror through a beam splitter. The concave mirror converges and reflects the bright-field light to a spatial light modulator. The spatial light modulator modulates the light to obtain a structured light field. The structured light field is projected onto the surface of the target being detected through an imaging lens. The structured light field reflected from the target surface reaches the target surface of the first photoelectric sensor. The first photoelectric sensor measures the total intensity of the structured light field and calculates the Fourier frequency value at a specified position in the bright-field Fourier frequency domain based on the correlation between the intensity and the structured light field. The bright-field Fourier frequency domain of the target being detected is obtained through frequency domain scanning. In the passive dark-field single-pixel imaging module: dark-field light is emitted from the dark-field light source to illuminate the target and diffusely reflect off the target surface; the diffuse light is imaged onto the spatial light modulator by the imaging lens; the spatial light modulator modulates the light to obtain a structured light field; the structured light field is converged and reflected by the concave mirror and then incident on the beam splitter interface and reflected by the beam splitter interface to the target surface of the second photoelectric sensor; the second photoelectric sensor measures the total intensity of the structured light field and calculates the Fourier frequency value at a specified position in the dark-field Fourier frequency domain by the correlation between the intensity of the structured light field and the structured light field; the dark-field Fourier frequency domain of the target is obtained by frequency domain scanning.

Images