Device and method for rapid detection of three-dimensional defects in micro-nano devices

Abstract

The present invention provides a device and method for rapid detection of three-dimensional defects in micro-nano devices in the field of optical measurement. The device and method are based on the principle of through-focus scanning optical microscopy (TSOM), and specifically relate to a TSOM device without mechanical axial scanning and a TSOM method for rapid measurement of geometric dimensions and detection of defects. The TSOM device utilizes a combination of a dispersive objective lens and a diffraction grating to achieve optical axial scanning instead of mechanical axial scanning. While improving the working speed, it fully utilizes the light flux of the imaging system and solves the effects of lateral offset and vibration interference caused by mechanical scanning. The method establishes a relationship between the magnification of the imaging system in the TSOM device, the position of light of different wavelengths on the detector surface, and the spectral bandwidth, and the parameters of the system components. The system component parameters can be adjusted to meet the measurement requirements, retaining the advantage of a large field of view of microscopic detection, and having a fast working speed, which is suitable for online detection and mass production applications.

Description

Technical Field

[0001] The present invention belongs to the field of optical micro-nano dimensional measurement, and more specifically, relates to a device and method for rapid three-dimensional defect detection in micro-nano devices based on the principle of through-focus scanning optical microscopy (TSOM). Specifically, the invention relates to a device for rapid three-dimensional defect detection in micro-nano devices using a combination of a dispersive objective lens and a diffraction grating to perform TSOM optical axial scanning, and a method for rapid three-dimensional defect detection in micro-nano devices that can rapidly construct TSOM images. Background Art

[0002] Through-focus scanning optical microscopy (TSOM) is an emerging micro- and nano-metrology and defect detection technology, a recommended technique in the semiconductor industry. With the development of new materials and technologies, integrated circuits are moving towards smaller sizes and more 3D architectures, entering the post-Moore era. This poses new challenges to nano-metrology and defect detection technologies.

[0003] The TSOM method is based on the principle of light scattering. When light is irradiated on a material and scattered, the profile shape and numerical value of the scattered light intensity depend on the surface statistical parameters of the object. When the scattered light and reflected light leave the sample surface, there is a certain scattering field above the sample. The scattering field contains important three-dimensional dimensions and defect information of the sample. The TSOM method acquires TSOM images by collecting a series of defocused images, and inverts and matches the three-dimensional morphological information such as line width, line height and sidewall inclination of the sample to be measured. The TSOM method uses the idea of ​​"what you see is not what you get" to bypass the optical diffraction limit. At the same time, the TSOM method retains the advantages of traditional optical measurement with a large field of view, and has the potential for fast scanning to measure dimensions and detect defects.

[0004] However, in practical applications, the mechanical axial scanning of the TSOM method requires multiple image acquisitions, which is slow and susceptible to lateral offset and vibration of the scanning mechanism. This results in lateral shifts in the acquired TSOM images, seriously affecting the accuracy of the TSOM method. Furthermore, only a very small portion of the through-focus images is used to construct the TSOM image, resulting in very low light flux utilization of the imaging system. Therefore, there is an urgent need to expand and improve the TSOM method to enhance its accuracy and efficiency. Summary of the Invention

[0005] In response to the above-mentioned defects or improvement needs of the prior art, the present invention provides a TSOM device and a TSOM method that combine the dispersion function of a dispersive objective lens with the spectroscopic function of a diffraction grating, and replaces the TSOM mechanical axial scanning with TSOM optical axial scanning, which not only improves the working speed, but also fully utilizes the light flux of the imaging system, and completely eliminates the lateral deviation and vibration interference of the mechanical axial scanning structure.

[0006] The technical solution adopted by the present invention is: a device and method for rapid detection of three-dimensional defects in micro-nano devices, comprising the following steps:

[0007] S1. Build the TSOM device. The TSOM device includes: an illumination module, which utilizes a combination of optical components such as a bandwidth-limited LED light source and a cylindrical mirror to generate a beam of collimated light in a strip structure, which serves as a collimated light source to illuminate the dispersion module; a dispersion module, in which the key component is the dispersion lens in the dispersion objective, which is placed after the illumination module. The dispersion function of the dispersion lens is used to establish a linear relationship between light of different wavelengths and the position of the axial dispersion focal point. Therefore, multiple focal planes corresponding to different wavelengths are conjugate to each other, achieving defocused illumination and simultaneously realizing TSOM optical axial scanning instead of TSOM mechanical axial scanning; an imaging module, in which the key component is the diffraction grating. The diffraction grating's spectroscopic function is used to spread light according to wavelength onto the entire imaging surface of the detector. Light of different wavelengths is imaged onto different columns of the detector, and each column corresponds to the dispersion focus of light of different wavelengths in the dispersion objective, thereby achieving rapid imaging of TSOM optical axial scanning.

[0008] S2. Calibrate the TSOM device. During the calibration process, the dispersive objective lens in the already constructed TSOM device is replaced with a high-precision back-polished plane mirror. The mirror is then placed on a high-precision translation stage. A dual-frequency laser rangefinder is used to measure the distance the plane mirror moves. A detector collects wavelength images corresponding to different distances. This allows the magnification of the imaging system, the position on the detector surface where light of different wavelengths falls after passing through the optical system, and the relationship between parameters such as spectral bandwidth and the grating constant, slit width, angle of incidence, spectral order, and collimator focal length to be calibrated. This completes the calibration of the TSOM device.

[0009] S3, Rapid Image Acquisition. Using a calibrated TSOM device, a single imaging operation acquires depth-direction scattering information for a column of plane positions. Based on the calibration results, the scattering information is mapped and superimposed by column, yielding an actual TSOM image of the workpiece at a particular column of plane positions. Simultaneously, by relatively moving the workpiece in a direction perpendicular to the column planes, scattering information for a column position perpendicular to different column planes can be obtained. Actual TSOM images for different column positions can then be output, rapidly constructing the actual acquired TSOM image.

[0010] S4, Dimension Measurement and Defect Detection. Data processing is performed on the actual TSOM images, including TSOM image library matching and TSOM image machine learning data processing methods. Inversion matching is used to obtain 3D micro-nano geometric dimensions or defects, thereby completing rapid 3D defect detection of micro-nano devices.

[0011] The TSOM image library matching method, one of the data processing methods, mainly includes: constructing a TSOM simulation image library; TSOM image comparison and size and defect extraction. When constructing the TSOM simulation image library, computer simulation is used to generate TSOM simulation images of different sizes and defects based on the principle of light scattering, and the different TSOM simulation images are combined to form a TSOM simulation image library. When performing TSOM image comparison and size and defect extraction, the TSOM device is used to quickly obtain TSOM actual images of the geometric surface area of ​​interest of the workpiece to be measured. The images are compared with standard TSOM simulation images corresponding to different geometric size values ​​in the stored matching library. The geometric size values ​​and defects corresponding to the TSOM simulation image that is closest to the actual TSOM image are used as the data processing results.

[0012] TSOM image machine learning is another data processing method, which mainly includes: building a TSOM machine learning model; evaluating the machine learning model and size and defect extraction. When building the TSOM learning model, first select representative TSOM images from the actual TSOM image sequence collected, and select an appropriate number of representative TSOM images as training data sets and test data sets; then perform feature extraction on the training data set, and select several appropriate machine learning algorithms based on the selected features; after selecting the algorithm, use the training data set to train the model separately to obtain different machine learning models. When evaluating the machine learning model and size and defect extraction, first use the test data set to analyze and compare the results of different learning models, and select the model with the best performance as the optimal model; then perform feature extraction on the TSOM image to be tested, input the extracted feature vector into the optimal model, and use the geometric size values ​​and defects obtained from the training as the data processing results.

[0013] Preferably, the dispersive objective lens focuses light of different wavelengths at different axial dispersion focal positions to generate defocused illumination. When the position of the workpiece to be measured is fixed, the scattered light of each wavelength corresponds to a defocused image with a certain defocus degree.

[0014] Preferably, the diffraction grating spreads the multi-wavelength scattered light scattered from the surface of the workpiece to be measured to different column positions of the entire imaging surface of the detector according to the length of the wavelength, and each column position corresponds to the dispersion focus of light of different wavelengths of the dispersive objective lens, thereby achieving one-time imaging to obtain defocused images with different defocus degrees, and realizing rapid imaging of TSOM optical axial scanning.

[0015] Preferably, the detector is a high-resolution CCD camera.

[0016] The computer-readable storage medium of the present invention stores executable instructions thereon, and when the instructions are executed by one or more processors, the one or more processors can execute the device and method for rapid detection of three-dimensional defects of micro-nano devices described in claims 4 to 8.

[0017] Compared with the prior art, the present invention has the following advantages:

[0018] (1) The present invention utilizes the dispersion effect to realize TSOM optical axial scanning to obtain defocused images, effectively avoiding lateral deviation and vibration interference of the mechanical axial scanning structure and improving the stability of the device.

[0019] (2) The present invention retains the advantages of a large field of view for microscopic measurement and detection, is conducive to rapid scanning requirements, and effectively improves efficiency.

[0020] (3) The dimensional measurement and defect detection results of the present invention are highly accurate and work at a fast speed, and are suitable for online use and mass production applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 Schematic diagram of a device model according to an embodiment of the present invention.

[0022] Figure 2 This is an overall flow chart of the machine learning method according to an embodiment of the present invention.

[0023] Figure 3 It is an overall flow chart of the library matching method according to an embodiment of the present invention.

[0024] Figure 4 is a flowchart for constructing TSOM images.

[0025] Figure 5 This is the schematic diagram of the TSOM calibration device.

[0026] Figure 6 It is a calibration flow chart. DETAILED DESCRIPTION

[0027] Figure 1 Figure 1 is a schematic diagram of a device model according to an embodiment of the present invention. The system consists of three components: an illumination unit 11, a dispersion unit 12, and an imaging unit 13. Key components of the illumination unit include a bandwidth-limited LED light source 111, a precision pinhole 112, and a cylindrical mirror 113. Key components of the dispersion unit include a dispersion objective lens 121. Key components of the imaging unit include a diffraction grating 131 and a detector 132.

[0028] When size measurement and defect detection begin, the illumination light emitted by the limited bandwidth light source 111 is filtered by the lens group and the precision pinhole 112 to become a relatively ideal point light source, and then filtered by the cylindrical mirror 113 and other optical elements to become a beam of collimated light with a strip structure. After passing through the dispersion part 12, light of different wavelengths is focused at different axial dispersion focal positions. When the position of the workpiece 21 to be measured is fixed, the scattered light of each wavelength corresponds to a defocused image with a certain degree of defocus, thereby achieving different degrees of defocus. The light at different focal points depending on the wavelength is scattered by the workpiece 21 to be measured, and then passes through the dispersion objective lens 121 and other optical elements to be imaged in the imaging light path, realizing axial scanning of a strip target point of the workpiece to be measured in the direction of the optical axis.

[0029] In the imaging section 13, light of different wavelengths strikes the diffraction grating 131 at the same angle. Optical components then spread the light across the entire imaging surface of the detector 132, dispersing it according to wavelength. Light of different wavelengths is imaged onto different columns of the detector, each corresponding to a different wavelength dispersion focus of the dispersive objective. This utilizes the dispersion function of the dispersive objective and the spectroscopic function of the diffraction grating to achieve rapid imaging during TSOM optical axial scanning.

[0030] In the process of size measurement and defect detection, there are two methods for data processing of the actually collected TSOM images: TSOM image machine learning and TSOM image library matching data processing methods, as shown in the following example. Figure 2 、 3 shown.

[0031] Figure 2 This is an overall flow chart of the machine learning method of the present invention. The TSOM device needs to be calibrated before use. After calibration, the important parameters of the device, such as the wavelength of the illumination light, the polarization state, and the numerical aperture of the objective lens, are kept unchanged. After that, the position of the workpiece to be measured is fixed, and the calibrated TSOM device is used to quickly capture the defocused image and construct the actual TSOM image. The TSOM image machine learning method is selected to first construct a training data set and a test data set, then select a suitable machine learning algorithm to create and train the model, and then extract the features of the workpiece to be measured and input them into the optimal model. The geometric dimension values ​​and defects obtained through training are used as the results of data processing.

[0032] Figure 3This is the overall flow chart of the library matching method of the present invention. The TSOM device needs to be calibrated before use. After the calibration, the important parameters of the device, such as the wavelength of the illumination light, the polarization state, and the numerical aperture of the objective lens, are kept unchanged. After that, the position of the workpiece to be measured is fixed, and the calibrated TSOM device is used to quickly capture the defocused image and construct the TSOM actual image. The TSOM image library matching method is selected. First, a TSOM simulation image library is constructed using computer simulation. At the same time, the TSOM actual image of the workpiece to be measured is obtained by using the device. By performing differential comparison with the TSOM simulation image library, the TSOM simulation image closest to the TSOM actual image of the workpiece to be measured is found, and the corresponding geometric dimensions are used as the measurement results.

[0033] Figure 4 This is a flowchart for constructing the actual TSOM image captured using the image output by the detector 132. A single detector image captures depth-direction scattering information 31 of the workpiece under test at a column plane position. This scattering information 31, acquired by the detector, is remapped and superimposed column by column based on the calibration results of the imaging system to obtain an actual TSOM image 32 of the workpiece under test at a specific column plane position, as well as a pseudo-color rendered actual TSOM image 33. The workpiece under test is relatively moved in a direction perpendicular to the column plane, and push-scanning is used to obtain scattering information 34 of the workpiece under test at different column positions.

[0034] The magnification M of the imaging system needs to be calibrated t , light of different wavelengths at position u on the surface of the detector 132 i , and spectral bandwidth δλ and other parameters. The relationship between the above parameters and the grating constant, slit width, incident angle, spectral order and collimator focal length are as follows:

[0035]

[0036]

[0037]

[0038] Among them, f4, f D1 , f5, f6, f7 and f8 are the focal lengths of the lens group in the optical path, i is the grating incident angle, λ0 is the main wavelength, λ is the illumination wavelength, d is the grating constant, q is the detector pixel size, and m is the diffraction order.

[0039] Figure 5 It is a schematic diagram of a calibration imaging system, in which the dispersion objective lens 121 is replaced by a high-precision plane reflector 141, which is placed on a two-dimensional adjustment frame 142 that can be translated along the optical axis, and a dual-frequency laser rangefinder 143 is placed behind it.

[0040] Figure 6This is a calibration flow chart. The plane reflector 141 is used as the target to be measured. Before calibration, the direction of the plane reflector 141 is first adjusted so that its surface is strictly perpendicular to the optical axis of the imaging system. The adjustment frame 142 then drives the plane reflector to move along the optical axis of the measurement system. Every time the plane reflector 141 moves a position, the dual-frequency laser rangefinder 143 measures the distance moved once, and simultaneously controls the detector 132 to capture an image. Since the distances between the various points on the surface of the plane reflector 141 are basically the same, the image captured by the detector 132 is an image of a specific wavelength, imaged at a specific position on the detection surface, and multiple sets of translation distances and imaged positions are recorded. The required parameters can then be calibrated using the multivariate regression method 45.

[0041] The accompanying drawings show some block diagrams and / or flow charts. It should be understood that some blocks in the block diagrams and / or flow charts, or combinations thereof, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when these instructions are executed by the processor, they can create a device for implementing the functions / operations described in the block diagrams and / or flow charts.

[0042] Therefore, the technology of the present disclosure can be implemented in the form of hardware and / or software (including firmware, microcode, etc.). In addition, the technology of the present disclosure can take the form of a computer program product on a computer-readable medium having instructions stored thereon, which can be used by or in conjunction with an instruction execution system (e.g., one or more processors). In the context of the present disclosure, a computer-readable medium can be any medium that can contain, store, convey, propagate, or transmit instructions. For example, a computer-readable medium can include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, device, or propagation medium. Specific examples of computer-readable media include: magnetic storage devices, such as magnetic tape or hard disks (HDDs); optical storage devices, such as compact disks (CD-ROMs); memories, such as random access memory (RAM) or flash memory; and / or wired / wireless communication links."

[0043] Although exemplary embodiments of the present invention have been shown and described, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the scope and spirit of the invention as defined in the claims and their equivalents.

Claims

1. A device for rapid detection of three-dimensional defects in micro-nano devices, characterized by: Through-focus scanning optical microscopy (TSOM) equipment includes: The illumination module uses a light source and a cylindrical mirror to generate a beam of collimated light in a strip structure to illuminate the dispersion module; The dispersion module uses the dispersion function of the dispersion lens in the dispersion objective to focus light of different wavelengths at different focal positions, realizing TSOM optical axial scanning; The imaging module uses the spectroscopic function of the diffraction grating to spread light across the entire imaging surface of the detector according to wavelength. Light of different wavelengths is imaged onto different columns of the detector, and each column corresponds to a different wavelength of the dispersive objective lens. The single image on the detector contains the focus and defocus information of the structure to be measured; TSOM device calibration module, which uses a laser rangefinder to measure distance and is used to calibrate the TSOM device; The fast image acquisition module acquires the image on the detector and processes it to obtain the actual acquired TSOM image; The size measurement and defect detection module processes the actual collected TSOM images and performs inversion matching to obtain the three-dimensional geometric dimensions or defects of micro-nano devices.

2. A method for rapid detection of three-dimensional defects in micro-nano devices according to the apparatus of claim 1, characterized in that: include: Steps for building a TSOM device, using a combination of an illumination module, a dispersion module, and an imaging module to build a TSOM device; Steps for calibrating the TSOM device: Use a laser rangefinder to calibrate the built TSOM device; The steps for rapid image acquisition use the calibrated TSOM device to quickly construct the actual acquired TSOM image; The steps of size measurement and defect detection are to process the data of the actually collected TSOM images, including the data processing methods of TSOM image library matching and TSOM image machine learning, and inversion matching to obtain the three-dimensional geometric dimensions or defects of micro-nano devices.

3. The method for rapid detection of three-dimensional defects in micro-nano devices according to claim 2, characterized in that: In the step of calibrating the TSOM device, it is only necessary to replace the dispersive objective lens in the assembled TSOM device with a high-precision plane mirror. By translating the high-precision plane mirror, the output results of the laser rangefinder can be used to calibrate the magnification of the imaging system, the position where light of different wavelengths falls on the surface of the detector after passing through the optical system, and the relationship between the spectral bandwidth parameter and the grating constant, slit width, incident angle, spectral order and collimator focal length, thereby completing the calibration of the TSOM device.

4. The method for rapid detection of three-dimensional defects in micro-nano devices according to claim 2, characterized in that: In the step of rapid image acquisition, the calibrated TSOM device obtains the depth-direction scattering information of the workpiece to be measured at a column plane position through a single imaging operation. Then, based on the calibration results, the scattering information is mapped and superimposed by columns to obtain the actual TSOM image of the workpiece to be measured at a certain column plane position. At the same time, by relatively moving the workpiece to be measured along the extension direction perpendicular to the column plane, the scattering information of the workpiece to be measured at a certain column position perpendicular to the extension direction of different column planes can be obtained, and then the actual TSOM images of different column positions can be output.

5. The method for rapid detection of three-dimensional defects in micro-nano devices according to claim 2, characterized in that: The data processing method for TSOM image library matching in the steps of dimension measurement and defect detection includes: The steps of constructing a TSOM simulation image library are to use a computer to simulate and generate a TSOM simulation image library; The steps of TSOM image comparison and size and defect extraction are as follows: using the TSOM device to quickly obtain a TSOM actual image of the geometric surface area of ​​interest of the workpiece to be measured, and comparing it with the standard TSOM simulation images corresponding to different geometric size values ​​in the stored matching library, and taking the geometric size values ​​and defects corresponding to the TSOM simulation image closest to the TSOM actual image as the data processing result.

6. The method for rapid detection of three-dimensional defects in micro-nano devices according to claim 2, characterized in that: The data processing methods for TSOM image machine learning in the steps of dimension measurement and defect detection include: The steps to build a TSOM machine learning model are to use a suitable number of representative TSOM real-world images to create training and test datasets, and then select an appropriate machine learning algorithm to create and train the model; Evaluate the machine learning model and the steps of dimension and defect extraction, evaluate the constructed machine learning model, select the optimal model and use the geometric dimension values ​​and defects obtained by its training as the data processing results.

7. A computer-readable storage medium having executable instructions stored thereon, wherein when the instructions are executed by one or more processors, the one or more processors execute the method for rapid detection of three-dimensional defects in micro-nano devices according to any one of claims 2 to 6.

8. A computer program product comprising a computer program, characterized in that: When the computer program is processed and executed, the method for rapid detection of three-dimensional defects of micro-nano devices according to any one of claims 2 to 6 is implemented.

Images