Alignment correction method, device and equipment for wafer asymmetric deformation mark and medium

Abstract

The invention provides a wafer asymmetric deformation mark alignment correction method and device, equipment and a medium. The wafer asymmetric deformation mark alignment correction method comprises the steps of obtaining structure parameters of an asymmetric deformation mark on a target wafer; wherein the structure parameters at least comprise a marked geometrical shape, a material attribute and process historical information; inputting the structure parameters into a pre-established asymmetric mark simulation model, and simulating to generate alignment position deviation values under each wavelength; and according to a pre-trained multi-wavelength optimal weight coefficient, performing weighted fusion on the alignment position deviation value corresponding to each wavelength, and outputting a corrected final alignment position of the target wafer to compensate a measurement error caused by asymmetric deformation of the mark. No matter what complex process steps are experienced by the mark, the corresponding corrected final alignment position can be generated in real time only by inputting the actual structural parameters of the mark, and large-scale experimental calibration does not need to be carried out again, so that self-adaptive response to various process conditions is realized, and the production efficiency of wafers is greatly improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of wafer alignment error correction, and particularly relates to a wafer asymmetric deformation mark alignment correction method and device, equipment and medium. BACKGROUND

[0002] In the semiconductor manufacturing process, the photolithography process is a key link to determine the precision of device feature size, and high-precision wafer alignment technology is the basis for realizing good overlay control between multiple layers of patterns. At present, the mainstream wafer alignment system mostly adopts an alignment method based on grating diffraction and interference principle. Ideally, the alignment mark has a symmetrical geometric structure, and the light field amplitudes of the positive and negative diffraction orders are equal and the phases are opposite, so it can be accurately identified. However, in the actual manufacturing process, the alignment mark needs to go through various complex process steps, including epitaxial growth, medium deposition, metallization, chemical mechanical polishing (CMP), dry etching, etc. These processes will cause the mark structure to deform to different degrees. In particular, due to the influence of factors such as wafer in-plane temperature gradient, uneven material stress distribution and edge effect, the deformation of the mark usually presents an asymmetric characteristic, and the closer to the edge of the wafer, the more significant. For example, the side wall tilt, CD offset or bottom asymmetric etching in the etching process will destroy the symmetry of the original grating structure, thereby introducing an additional diffraction phase difference. This phase shift caused by structural asymmetry will directly reflect as a position shift measured by the alignment sensor, i.e. alignment position deviation (APD). Since this deviation cannot be distinguished from the real physical displacement, it causes systematic alignment error.

[0003] To alleviate such problems, the prior art generally adopts a feedback correction strategy based on sampling, also known as Automatic Process Control (APC). Specifically, a small number of sample wafers are selected in a production batch, and the actual overlay error thereof is measured offline using a scanning electron microscope (SEM) or an image-based metrology device (IBO), and then the measurement result is fed back to the alignment system for correcting the alignment parameters of subsequent wafers. However, this method has obvious defects: first, SEM measurement is costly and slow, and it is difficult to cover all wafers and all fields, resulting in insufficient statistical representativeness; second, this method belongs to a post-compensation mechanism, and there is obvious time delay, which cannot realize real-time control; finally, since the deformation degree of each wafer and even different regions in the same wafer is different, it is difficult for a fixed correction value to adapt to complex process fluctuations, resulting in limited correction effect. In summary, the existing wafer alignment technology still faces challenges such as limited alignment accuracy, weak process adaptability, and strong correction lag when facing increasingly complex process environments. There is an urgent need for a new method that can actively predict and real-time correct the alignment position deviation caused by asymmetric marker deformation to realize higher accuracy, stronger robustness and lower cost of alignment control. SUMMARY

[0004] Therefore, the purpose of the present application is to provide a wafer asymmetric deformation marker alignment correction method, device, equipment and medium, which can generate the corresponding corrected final alignment position in real time as long as the actual structure parameters of the marker are inputted, regardless of the complex process steps the marker undergoes, without the need for large-scale experimental calibration, thereby realizing adaptive response to various process conditions and greatly improving the production efficiency of wafers.

[0005] The present application provides a wafer asymmetric deformation marker alignment correction method, which comprises: obtaining the structure parameters of the asymmetric deformation marker on the target wafer; wherein the structure parameters at least include the geometric shape, material properties and process history information of the marker; inputting the structure parameters into a pre-established asymmetric marker simulation model to simulate and generate the alignment position deviation value at each wavelength; weighting and fusing the alignment position deviation values corresponding to each wavelength according to the pre-trained multi-wavelength optimal weight coefficient, and outputting the corrected final alignment position of the target wafer to compensate for the measurement error caused by asymmetric deformation of the marker.

[0006] In one possible implementation, the asymmetric marker simulation model is determined by the following steps: An asymmetric grating simulation model is established, and a diffraction light field distribution generated by the asymmetric deformed mark under multiple different wavelength illumination conditions is calculated by using a vector diffraction theory; Based on the diffraction light field distribution and optical system parameters of an alignment sensor, an initial asymmetric mark simulation model is constructed, and a predicted wafer quality value and a predicted alignment position deviation value under each sample wavelength are output based on the initial asymmetric mark simulation model; The initial asymmetric mark simulation model is verified based on actual wafer quality values under each sample wavelength and corresponding predicted wafer quality values, and the initial asymmetric mark simulation model that passes the verification is determined as the asymmetric mark simulation model.

[0007] In one possible implementation, the diffraction light field distribution generated by the asymmetric deformed mark under multiple different wavelength illumination conditions is calculated by using the vector diffraction theory, including: The asymmetric grating corresponding to the asymmetric deformed mark is divided into multiple equal-thickness layers along the depth direction by using a step approximation method; wherein each layer is equivalent to a rectangular grating structure; Maxwell equations are solved in each layer, and a Fourier series form of an incident field, a reflected field and a transmitted field is expanded to obtain electromagnetic field distributions of each diffraction order; Field continuity equation groups are constructed by using electromagnetic boundary conditions between adjacent layers, and an enhanced transmission matrix method is used to transfer electric field information layer by layer to determine the diffraction light field distribution generated by the asymmetric deformed mark under multiple different wavelength illumination conditions.

[0008] In one possible implementation, the alignment position deviation values corresponding to each wavelength are weighted and fused according to pre-trained multi-wavelength optimal weight coefficients, and a corrected final alignment position of the target wafer is output to compensate for measurement errors caused by asymmetric deformation of the mark, including: The alignment position deviation values corresponding to multiple wavelengths are weighted and fused to determine optimal weight coefficients of each wavelength; The measured multi-wavelength alignment position results are weighted and calculated based on the optimal weight coefficients, and a corrected final alignment position is output.

[0009] In one possible implementation, the pre-trained multi-wavelength optimal weight coefficients are determined by the following steps: A Monte Carlo perturbation sample set is constructed, and random changes are applied to the mark depth, side wall angle and asymmetric degree; The asymmetric mark simulation model is used to generate a data set of alignment position deviation values of multiple sample wavelengths corresponding to each perturbation state; A machine learning model is trained based on the dataset to determine the set of optimal multi-wavelength weight coefficients that minimize the alignment position deviation of the weighted samples.

[0010] In one possible implementation, the weighted fusion must satisfy the following constraints:

[0011]

[0012] in, n For the number of different wavelengths, For the first m The optimal weighting coefficients for each wavelength. For the first m Alignment position deviation value corresponding to each wavelength.

[0013] This application embodiment also provides an alignment and correction device for wafer asymmetric deformation marks, the alignment and correction device comprising: The parameter acquisition module is used to acquire the structural parameters of the asymmetric deformation mark on the target wafer; wherein, the structural parameters include at least the geometric shape, material properties and process history information of the mark; The simulation prediction module is used to input the structural parameters into a pre-established asymmetric marker simulation model to simulate and generate alignment position deviation values ​​at each wavelength. The weighted fusion module is used to perform weighted fusion of the alignment position deviation values ​​corresponding to each wavelength according to the pre-trained multi-wavelength optimal weight coefficients, and output the corrected final alignment position of the target wafer to compensate for the measurement error caused by the asymmetric deformation of the marker.

[0014] In one possible implementation, the alignment correction device further includes a simulation model building module, which determines the asymmetric marker simulation model through the following steps: A simulation model incorporating an asymmetric grating was established, and the diffraction field distribution generated by the asymmetric deformation mark under multiple illumination conditions with different wavelengths was calculated using vector diffraction theory. Based on the diffraction field distribution and the optical system parameters of the alignment sensor, an initial asymmetric marking simulation model is constructed. Based on the initial asymmetric marking simulation model, the predicted wafer quality value and the predicted alignment position deviation value at each sample wavelength are output. The initial asymmetric labeling simulation model is verified based on the actual wafer quality value at each sample wavelength and the corresponding predicted wafer quality value. The initial asymmetric labeling simulation model that passes the verification is determined as the asymmetric labeling simulation model.

[0015] The embodiment of the present application also provides an electronic device, comprising a processor, a memory and a bus, the memory stores machine readable instructions executable by the processor, when the electronic device is running, the processor and the memory communicate through the bus, and the machine readable instructions are executed by the processor to perform the steps of the wafer asymmetric deformation mark alignment correction method.

[0016] The embodiment of the present application also provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to perform the steps of the wafer asymmetric deformation mark alignment correction method.

[0017] The wafer asymmetric deformation mark alignment correction method, device, equipment and medium provided by the embodiment of the present application, the alignment correction method comprises: obtaining the structure parameters of the asymmetric deformation mark on the target wafer; wherein the structure parameters at least include the geometric shape, material attribute and process history information of the mark; inputting the structure parameters into the pre-established asymmetric mark simulation model, simulating to generate the alignment position deviation value under each wavelength; according to the pre-trained multi-wavelength optimal weight coefficient, the alignment position deviation values corresponding to each wavelength are weighted and fused, and the corrected final alignment position of the target wafer is output, so as to compensate the measurement error caused by the asymmetric deformation of the mark. No matter what complex process steps the mark experiences, as long as the actual structure parameters are input, the corresponding corrected final alignment position can be generated in real time, without the need of large-scale experimental calibration, so that the adaptive response to various process conditions is realized, and the production efficiency of the wafer is greatly improved.

[0018] In order to make the above objectives, characteristics and advantages of the present application more apparent and easy to understand, the following preferred embodiments are specifically described below, and the accompanying drawings are described in detail as follows. BRIEF DESCRIPTION OF DRAWINGS

[0019] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments, and it should be understood that the following drawings only show some embodiments of the present application, and therefore should not be regarded as a limitation to the scope, and for those skilled in the art, other related drawings can also be obtained without creative labor on the basis of these drawings.

[0020] Figure 1 A flow chart of a wafer asymmetric deformation mark alignment correction method provided by the embodiment of the present application; Figure 2 One of the structure schematic diagrams of a wafer asymmetric deformation mark alignment correction device provided by the embodiment of the present application; Figure 3Figure 2 is a structural schematic diagram of a wafer asymmetric deformation mark alignment correction device provided by an embodiment of the present application; Figure 4 Figure 4 is a structural schematic diagram of an electronic device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0021] In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions of the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, not all the embodiments. The components of the embodiments of the present application described and shown in the drawings can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the claimed present application, but only represents selected embodiments of the present application. Based on the embodiments of the present application, every other embodiment obtained by a person skilled in the art without creative work belongs to the scope of protection of the present application.

[0022] Firstly, the application scenarios applicable to the present application are introduced. The present application can be applied to the field of wafer alignment error correction.

[0023] It is found through research that the existing technology generally adopts a sampling-based feedback correction strategy, also known as automatic process control (APC). The specific method is to select a small number of sample wafers in the production batch, measure the actual overlay error of the sample wafers offline using a scanning electron microscope (SEM) or an image-based metrology device (IBO), and then feed back the measurement results to the alignment system for correction of the alignment parameters of subsequent wafers. However, this method has obvious defects: firstly, the SEM measurement is high in cost and slow in speed, and it is difficult to cover all wafers and all fields, resulting in insufficient statistical representativeness; secondly, this method belongs to a post-compensation mechanism and has obvious time delay, and cannot realize real-time control; finally, since the deformation degree of the marks of each wafer and even different regions in the same wafer is different, it is difficult for a fixed correction value to adapt to complex process fluctuations, resulting in limited correction effect. In summary, the existing wafer alignment technology still faces challenges such as limited alignment accuracy, weak process adaptability and strong correction lag when facing increasingly complex process environments. There is an urgent need for a new method that can actively predict and correct the alignment position deviation caused by asymmetric mark deformation to realize higher accuracy, stronger robustness and lower cost of alignment control.

[0024] Based on this, the embodiment of the present application provides a wafer asymmetric deformation mark alignment correction method, no matter what complex process steps the mark experiences, as long as the actual structure parameters are input, the corresponding corrected final alignment position can be generated in real time, without re-performing large-scale experimental calibration, thereby realizing adaptive response to various process conditions, and greatly improving the production efficiency of the wafer.

[0025] Please refer to Figure 1 , Figure 1 The embodiment of the present application provides a wafer asymmetric deformation mark alignment correction method. As shown in Figure 1 The alignment correction method provided by the embodiment of the present application comprises the following steps. S101: Obtain the structure parameters of the asymmetric deformation mark on the target wafer; wherein the structure parameters at least include the geometric shape, material attribute and process history information of the mark.

[0026] It should be noted that the structure parameters include grating period, top line width, bottom line width, complex refractive index of each layer of material, and other parameters such as stack layer number and interface position.

[0027] S102: Input the structure parameters into the pre-established asymmetric mark simulation model, and simulate to generate the alignment position deviation value under each wavelength.

[0028] In this step, the structure parameters are input into the pre-established asymmetric mark simulation model, and the alignment position deviation value under each wavelength is simulated.

[0029] Here, the structure parameters are first input into the asymmetric grating simulation model of the asymmetric mark simulation model, the diffraction field distribution of the asymmetric deformation mark under multiple different wavelength illumination conditions is calculated by using the vector diffraction theory, and the alignment position deviation value (APD) under each wavelength is simulated according to the diffraction field distribution and the optical system model of the alignment sensor.

[0030] In one possible implementation, the asymmetric mark simulation model is determined by the following steps: A: Establish an asymmetric grating simulation model, and calculate the diffraction field distribution of the asymmetric deformation mark under multiple different wavelength illumination conditions by using the vector diffraction theory.

[0031] Here, the asymmetric grating simulation model is established, and the diffraction field distribution of the asymmetric deformation mark under multiple different wavelength illumination conditions is calculated by using the vector diffraction theory.

[0032] In one possible implementation, the diffraction field distribution of the asymmetric deformation mark under multiple different wavelength illumination conditions is calculated by using the vector diffraction theory, comprising: (1) : using the ladder approximation method to divide the asymmetric grating corresponding to the asymmetric deformation mark into multiple equal-thickness layers along the depth direction; wherein each layer is equivalent to a rectangular grating structure.

[0033] Here, the entire grating structure is divided into N equal-thickness thin layers along the direction perpendicular to the wafer surface, each layer has a thickness, each layer is regarded as a uniform rectangular grating, and the width is taken as the actual transverse dimension at the midpoint of the layer; for the case of inclined and asymmetric side wall, the left and right boundaries of each layer are determined by linearly decreasing or increasing the left and right inclination angles, respectively.

[0034] (2) : solving Maxwell's equations in each layer, expanding the Fourier series form of the incident field, the reflected field and the transmitted field to obtain the electromagnetic field distribution of each diffraction order.

[0035] Here, after completing the structure discretization, for each rectangular sublayer, the Rigorous Coupled-Wave Analysis (RCWA) is used to solve the local electromagnetic field to obtain the electromagnetic field distribution of each diffraction order.

[0036] (3) : constructing a field continuity equation set by simultaneously considering the electromagnetic boundary conditions between adjacent layers, using the enhanced transmission matrix method to transfer the electric field information layer by layer, and determining the diffraction field distribution generated by the asymmetric deformation mark under multiple different wavelength illumination conditions.

[0037] Here, the field continuity equation set is constructed by simultaneously considering the electromagnetic boundary conditions between adjacent layers, and the enhanced transmission matrix method is used to transfer the electric field information layer by layer to determine the diffraction field distribution generated by the asymmetric deformation mark under multiple different wavelength illumination conditions.

[0038] It should be noted that after the eigenmodes of all sublayers are obtained, the layers are connected by the continuity condition of the electromagnetic field at the interface (i.e. the continuity of the tangential field) to form a complete multi-layer structure response.

[0039] B: based on the diffraction field distribution and the optical system parameters of the alignment sensor, an initial asymmetric mark simulation model is constructed, and the initial asymmetric mark simulation model is used to output the predicted wafer quality value and the predicted alignment position deviation value under each sample wavelength.

[0040] Here, according to the diffraction field distribution and the optical system parameters of the alignment sensor, an initial asymmetric mark simulation model is constructed, and the initial asymmetric mark simulation model is used to output the predicted wafer quality value and the predicted alignment position deviation value under each sample wavelength.

[0041] C: The initial asymmetric labeling simulation model is verified based on the actual wafer quality value at each sample wavelength and the corresponding predicted wafer quality value. The initial asymmetric labeling simulation model that passes the verification is determined as the asymmetric labeling simulation model.

[0042] Here, if the error between the actual wafer quality value and the corresponding predicted wafer quality value at each sample wavelength is within a preset range, the initial asymmetric labeling simulation model passes the verification; if it is not within the preset range, the initial asymmetric labeling simulation model fails the verification and needs to be adjusted.

[0043] In this application, a high-precision asymmetric marker simulation model (based on rigorous coupled-wave analysis (RCWA)) is constructed to accurately reflect the structural deformation of wafer markers caused by epitaxial growth, etching, chemical mechanical polishing, and other processes under different manufacturing conditions, and their impact on the phase of the diffraction field. The established mapping relationship between "marker deformation—diffraction response—alignment deviation" allows the system to predict alignment position deviations for asymmetric markers with any known geometric parameters without requiring additional measured data. Therefore, even in complex situations where the degree of marker deformation varies in different regions (such as the center and edges) within the same wafer, accurate compensation can still be achieved, significantly improving the adaptability to various process flows and material systems.

[0044] S103: Based on the pre-trained multi-wavelength optimal weight coefficients, the alignment position deviation values ​​corresponding to each wavelength are weighted and fused to output the corrected final alignment position of the target wafer, so as to compensate for the measurement error caused by the asymmetric deformation of the marker.

[0045] In this step, the alignment position deviation values ​​corresponding to each wavelength are weighted and fused according to the pre-trained multi-wavelength optimal weight coefficients, and the corrected final alignment position of the target wafer is output to compensate for the measurement error caused by the asymmetric deformation of the marker.

[0046] Existing technologies rely on scanning electron microscopy (SEM) sampling and automatic process control (APC), a typical "post-correction" approach with significant time delays, making it difficult to handle rapid process disturbances between batches or within a single wafer. This invention, however, performs simulation prediction and weighted fusion calculations during the alignment process, achieving real-time online correction across the entire wafer—a feedforward control approach. This allows the alignment system to output compensated, correct position information immediately, avoiding batch yield losses due to error accumulation and improving the stability and controllability of the production process.

[0047] In one possible implementation, the pre-trained multi-wavelength optimal weight coefficient is used to weight and fuse the alignment position deviation values corresponding to each wavelength, and a corrected final alignment position of the target wafer is output to compensate for the measurement error caused by the asymmetric deformation of the mark. The alignment position deviation values corresponding to multiple wavelengths are weighted and fused to determine the optimal weight coefficient of each wavelength. The measured multi-wavelength alignment position results are weighted and calculated based on the optimal weight coefficient, and a corrected final alignment position is output.

[0048] Here, the alignment position deviation values corresponding to multiple wavelengths are weighted and fused to determine the optimal weight coefficient of each wavelength. The measured multi-wavelength alignment position results are weighted and calculated based on the optimal weight coefficient, and a corrected final alignment position is output.

[0049] Wherein, the corrected final alignment position is determined by the following formula x :

[0050] Wherein, if the number of wavelength types is 4, n = 4, is the alignment position measured by the mark grating without deformation, is the alignment position deviation value corresponding to the m th wavelength.

[0051] In one possible implementation, the weighted fusion needs to satisfy the following constraint conditions:

[0052]

[0053] Wherein, n is the number of different wavelengths, is the optimal weight coefficient of the m th wavelength, is the alignment position deviation value corresponding to the m th wavelength.

[0054] In one possible implementation, the pre-trained multi-wavelength optimal weight coefficient is determined by the following steps: a: Construct a Monte Carlo perturbation sample set and apply random changes to the mark depth, side wall angle and asymmetric degree.

[0055] Here, the probability distribution type of each key parameter is defined, for example: the side wall inclination angle, the groove depth, the CD deviation and the asymmetric degree (such as the difference between the left and right wall angles) all follow the normal distribution. The total sample number is set, and N groups of independent parameter vectors are extracted from the pseudo-random number generator to form the Monte Carlo perturbation sample set.

[0056] b: generating a dataset of alignment position deviation values of multiple samples at each wavelength under each perturbation state using the asymmetric mark simulation model.

[0057] Here, a dataset of alignment position deviation values of multiple samples at each wavelength under each perturbation state is generated using the asymmetric mark simulation model.

[0058] c: training a machine learning model based on the dataset to determine a set of optimal weight coefficients of multiple wavelengths that minimizes the weighted sample alignment position deviation values.

[0059] Here, a supervised learning framework is selected, and the goal is to train a mapping function $f: \mathbf{x} \mapsto \mathbf{w}$ that maps the current mark structure features $\mathbf{x}$ to a set of optimal wavelength weight coefficients $\mathbf{w} = [w_1, w_2, w_3, w_4]$ so that the weighted alignment position is closest to the true position.

[0060] Wherein, a set of training samples and a set of validation samples are first created by Monte Carlo simulation, so that the alignment mark depth and mark asymmetry are perturbed. Then, a machine learning algorithm is used to calculate the weight coefficients of each wavelength and verify.

[0061] In this application, by utilizing the difference in sensitivity of the alignment signal under multi-wavelength illumination to the asymmetric deformation, the optimal wavelength weight optimization mechanism is introduced, and by weighting and fusing the alignment position measurement results under multiple wavelengths, the final alignment value closer to the true position is obtained. This method not only suppresses the systematic error caused by asymmetric phase shift, but also dynamically adjusts the weight coefficient through machine learning algorithm, further enhancing the generalization ability of the correction model.

[0062] In a specific embodiment, Step 1: Obtain precise marking parameters under the current process conditions: Before the wafer enters the alignment process, the alignment mark structure parameters related to the current field are first obtained. These parameters reflect the geometric deformation and material changes of the mark caused by previous processes (such as etching, deposition, chemical mechanical polishing, etc.). Step 2: Call the pre-built asymmetric mark simulation model: The mark parameters obtained in Step 1 are input into the pre-built and verified asymmetric grating simulation model, and the electromagnetic field is solved using Rigorous Coupled-Wave Analysis (RCWA) to obtain the diffraction characteristics under different wavelengths of illumination. The specific execution process includes: 1. Establishing a two-dimensional periodic grating model, using the step approximation method to discretize the non-rectangular sidewalls into multiple layers of equal-thickness sheets, each layer being considered a homogeneous dielectric layer; 2. Solving the eigenmodes of Maxwell's equations at each wavelength to obtain the complex amplitude of each order of diffracted light; 3. Calculating the phase difference between the positive and negative first-order diffracted light; 4. Inputting this phase difference into the alignment sensor signal model to simulate the interference signal waveform received by the detector; 5. Extracting the alignment position deviation caused by asymmetric deformation at that wavelength by fitting the signal peak position. The final output is a four-dimensional vector representing the alignment position offset measured at each of the four working wavelengths. Step 3: Calling the machine learning model to calculate the optimal wavelength weight coefficients: After obtaining the alignment position deviation at each wavelength, the next step is to determine a set of optimal weights so that the weighted overall alignment position is closest to the true unbiased position.

[0063] This application implements a complete technical chain of "input marker parameters → prediction deviation → dynamic weighting → output correction position", which has the following significant advantages: 1. High process adaptability: Regardless of the form of asymmetric deformation of the marker (such as tilting, offset, or aspect ratio change), its impact can be accurately predicted based on the physical model; 2. No additional measurement cost: Eliminates the dependence on SEM sampling inspection, avoiding destructive testing and time delay; 3. Supports feedforward control: Error prediction and correction are completed during the alignment process, which is a real-time active compensation mechanism.

[0064] This application provides an alignment correction method for asymmetric deformation marks on a wafer. The method includes: acquiring structural parameters of the asymmetric deformation marks on a target wafer; wherein the structural parameters include at least the mark's geometry, material properties, and process history information; inputting the structural parameters into a pre-established asymmetric mark simulation model to simulate and generate alignment position deviation values ​​at various wavelengths; and weighting and fusing the alignment position deviation values ​​corresponding to each wavelength according to pre-trained multi-wavelength optimal weighting coefficients to output the corrected final alignment position of the target wafer, thereby compensating for measurement errors caused by the asymmetric deformation of the marks. Regardless of the complexity of the process steps the marks undergo, as long as the actual structural parameters are input, the corresponding corrected final alignment position can be generated in real time without the need for large-scale experimental calibration, thus achieving adaptive response to various process conditions and significantly improving wafer production efficiency.

[0065] Please see Figure 2 , Figure 3 , Figure 2 This is one of the structural schematic diagrams of an alignment and correction device for wafer asymmetric deformation marks provided in an embodiment of this application; Figure 3 This is a second schematic diagram of a wafer asymmetric deformation mark alignment and correction device provided in an embodiment of this application. Figure 2 As shown, the alignment correction device 200 includes: The parameter acquisition module 210 is used to acquire the structural parameters of the asymmetric deformation mark on the target wafer; wherein, the structural parameters include at least the geometric shape, material properties and process history information of the mark; The simulation prediction module 220 is used to input the structural parameters into a pre-established asymmetric marker simulation model and simulate and generate alignment position deviation values ​​at each wavelength. The weighted fusion module 230 is used to perform weighted fusion of the alignment position deviation values ​​corresponding to each wavelength according to the pre-trained multi-wavelength optimal weight coefficients, and output the corrected final alignment position of the target wafer to compensate for the measurement error caused by the asymmetric deformation of the marker.

[0066] Furthermore, such as Figure 3 As shown, the alignment correction device 200 further includes a simulation model construction module 240, which determines the asymmetric mark simulation model through the following steps: A simulation model incorporating an asymmetric grating was established, and the diffraction field distribution generated by the asymmetric deformation mark under multiple illumination conditions with different wavelengths was calculated using vector diffraction theory. Based on the diffraction field distribution and the optical system parameters of the alignment sensor, an initial asymmetric marking simulation model is constructed. Based on the initial asymmetric marking simulation model, the predicted wafer quality value and the predicted alignment position deviation value at each sample wavelength are output. The initial asymmetric labeling simulation model is verified based on the actual wafer quality value at each sample wavelength and the corresponding predicted wafer quality value. The initial asymmetric labeling simulation model that passes the verification is determined as the asymmetric labeling simulation model.

[0067] Furthermore, the simulation model building module 240 is used to calculate the diffraction light field distribution generated by the asymmetric deformation mark under multiple different wavelength illumination conditions using vector diffraction theory: The asymmetric grating corresponding to the asymmetric deformation mark is divided into multiple layers of equal thickness along the depth direction using the stepped approximation method; each layer is equivalent to a rectangular grating structure. Solve Maxwell's equations within each layer and expand the Fourier series forms of the incident, reflected, and transmitted fields to obtain the electromagnetic field distribution for each diffraction order. By establishing a set of field continuity equations based on the electromagnetic boundary conditions between adjacent layers, and using the enhanced transmission matrix method to transmit electric field information layer by layer, the diffraction field distribution generated by the asymmetric deformation mark under multiple different wavelength illumination conditions is determined.

[0068] Furthermore, the weighted fusion module 230 is used to perform weighted fusion of the alignment position deviation values ​​corresponding to each wavelength according to the pre-trained multi-wavelength optimal weight coefficients, and output the corrected final alignment position of the target wafer to compensate for the measurement error caused by the asymmetric deformation of the marker: The alignment position deviation values ​​corresponding to multiple wavelengths are weighted and fused to determine the optimal weight coefficient for each wavelength. The measured multi-wavelength alignment position results are weighted and calculated based on the optimal weighting coefficients, and the corrected final alignment position is output.

[0069] Furthermore, such as Figure 3 The alignment correction device 200 shown also includes a machine learning module 250, which determines the pre-trained multi-wavelength optimal weight coefficients through the following steps: Construct a Monte Carlo perturbation sample set by applying random variations to the label depth, sidewall angle, and degree of asymmetry; The asymmetric labeling simulation model is used to generate a dataset of alignment position deviation values ​​for multiple sample wavelengths corresponding to each perturbation state. A machine learning model is trained based on the dataset to determine the set of optimal multi-wavelength weight coefficients that minimize the alignment position deviation of the weighted samples.

[0070] This application provides an alignment and correction device for asymmetric deformation marks on a wafer. The device includes: a parameter acquisition module for acquiring structural parameters of the asymmetric deformation marks on a target wafer; wherein the structural parameters include at least the mark's geometry, material properties, and process history information; a simulation prediction module for inputting the structural parameters into a pre-established asymmetric mark simulation model to simulate and generate alignment position deviation values ​​at various wavelengths; and a weighted fusion module for weighted fusion of the alignment position deviation values ​​corresponding to each wavelength based on pre-trained multi-wavelength optimal weight coefficients, outputting the corrected final alignment position of the target wafer to compensate for measurement errors caused by the asymmetric deformation of the marks. Regardless of the complexity of the process steps the marks undergo, as long as the actual structural parameters are input, the corresponding corrected final alignment position can be generated in real time without the need for large-scale experimental calibration, thereby achieving adaptive response to various process conditions and significantly improving wafer production efficiency.

[0071] Please see Figure 4 , Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Figure 4 As shown, the electronic device 400 includes a processor 410, a memory 420, and a bus 430.

[0072] The memory 420 stores machine-readable instructions executable by the processor 410. When the electronic device 400 is running, the processor 410 communicates with the memory 420 via the bus 430. When the machine-readable instructions are executed by the processor 410, they can perform the operations described above. Figure 1 The specific implementation of the alignment and correction method for wafer asymmetric deformation marks in the method embodiment shown can be found in the method embodiment, and will not be repeated here.

[0073] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, can perform the above-described actions. Figure 1 The specific implementation of the alignment and correction method for wafer asymmetric deformation marks in the method embodiment shown can be found in the method embodiment, and will not be repeated here.

[0074] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0075] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the shown or discussed mutual couplings, direct couplings, or communication connections may be through some communication interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0076] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0077] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0078] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a processor-executable, non-volatile, computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0079] Finally, it should be noted that the above-described embodiments are merely specific implementations of this application, used to illustrate the technical solutions of this application, and not to limit them. The scope of protection of this application is not limited thereto. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some of the technical features, within the scope of the technology disclosed in this application. Such modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for aligning and correcting wafer asymmetric deformation marks, characterized in that, The alignment correction method includes: Obtain the structural parameters of the asymmetric deformation mark on the target wafer; wherein, the structural parameters include at least the mark's geometry, material properties, and process history information; The structural parameters are input into a pre-established asymmetric marker simulation model to simulate and generate alignment position deviation values ​​at each wavelength. Based on the pre-trained multi-wavelength optimal weight coefficients, the alignment position deviation values ​​corresponding to each wavelength are weighted and fused to output the corrected final alignment position of the target wafer, in order to compensate for the measurement error caused by the asymmetric deformation of the marker.

2. The alignment correction method according to claim 1, characterized in that, The asymmetric labeling simulation model is determined through the following steps: A simulation model incorporating an asymmetric grating was established, and the diffraction field distribution generated by the asymmetric deformation mark under multiple illumination conditions with different wavelengths was calculated using vector diffraction theory. Based on the diffraction field distribution and the optical system parameters of the alignment sensor, an initial asymmetric marking simulation model is constructed. Based on the initial asymmetric marking simulation model, the predicted wafer quality value and the predicted alignment position deviation value at each sample wavelength are output. The initial asymmetric labeling simulation model is verified based on the actual wafer quality value at each sample wavelength and the corresponding predicted wafer quality value. The initial asymmetric labeling simulation model that passes the verification is determined as the asymmetric labeling simulation model.

3. The alignment correction method according to claim 2, characterized in that, The calculation of the diffraction field distribution generated by the asymmetric deformation mark under multiple different wavelength illumination conditions using vector diffraction theory includes: The asymmetric grating corresponding to the asymmetric deformation mark is divided into multiple layers of equal thickness along the depth direction using the stepped approximation method; each layer is equivalent to a rectangular grating structure. Solve Maxwell's equations within each layer and expand the Fourier series forms of the incident, reflected, and transmitted fields to obtain the electromagnetic field distribution for each diffraction order. A set of field continuity equations is constructed by combining the electromagnetic boundary conditions between adjacent layers. The electric field information is transmitted layer by layer using the enhanced transmission matrix method, and the diffraction field distribution generated by the asymmetric deformation mark under multiple different wavelength illumination conditions is determined.

4. The alignment correction method according to claim 1, characterized in that, The step of weighting and fusing the alignment position deviation values ​​corresponding to each wavelength according to the pre-trained multi-wavelength optimal weight coefficients, and outputting the corrected final alignment position of the target wafer to compensate for the measurement error caused by the asymmetric deformation of the marker, includes: The alignment position deviation values ​​corresponding to multiple wavelengths are weighted and fused to determine the optimal weight coefficient for each wavelength. The measured multi-wavelength alignment position results are weighted and calculated based on the optimal weighting coefficients, and the corrected final alignment position is output.

5. The alignment correction method according to claim 1, characterized in that, The optimal weight coefficients for the pre-trained multi-wavelength system are determined through the following steps: Construct a Monte Carlo perturbation sample set by applying random variations to the label depth, sidewall angle, and degree of asymmetry; The asymmetric labeling simulation model is used to generate a dataset of alignment position deviation values ​​for multiple sample wavelengths corresponding to each perturbation state. A machine learning model is trained based on the dataset to determine the set of optimal multi-wavelength weight coefficients that minimize the alignment position deviation of the weighted samples.

6. The alignment correction method according to claim 1, characterized in that, The weighted fusion must meet the following constraints: in, n For the number of different wavelengths, For the first m The optimal weighting coefficients for each wavelength. For the first m Alignment position deviation value corresponding to each wavelength.

7. An alignment and correction device for wafer asymmetric deformation marks, characterized in that, The alignment correction device includes: The parameter acquisition module is used to acquire the structural parameters of the asymmetric deformation mark on the target wafer; wherein, the structural parameters include at least the geometric shape, material properties and process history information of the mark; The simulation prediction module is used to input the structural parameters into a pre-established asymmetric marker simulation model to simulate and generate alignment position deviation values ​​at each wavelength. The weighted fusion module is used to perform weighted fusion of the alignment position deviation values ​​corresponding to each wavelength according to the pre-trained multi-wavelength optimal weight coefficients, and output the corrected final alignment position of the target wafer to compensate for the measurement error caused by the asymmetric deformation of the marker.

8. The alignment and correction device according to claim 7, characterized in that, The alignment correction device further includes a simulation model construction module, which determines the asymmetric mark simulation model through the following steps: A simulation model incorporating an asymmetric grating was established, and the diffraction field distribution generated by the asymmetric deformation mark under multiple illumination conditions with different wavelengths was calculated using vector diffraction theory. Based on the diffraction field distribution and the optical system parameters of the alignment sensor, an initial asymmetric marking simulation model is constructed. Based on the initial asymmetric marking simulation model, the predicted wafer quality value and the predicted alignment position deviation value at each sample wavelength are output. The initial asymmetric labeling simulation model is verified based on the actual wafer quality value at each sample wavelength and the corresponding predicted wafer quality value. The initial asymmetric labeling simulation model that passes the verification is determined as the asymmetric labeling simulation model.

9. An electronic device, characterized in that, include: The device includes a processor, a memory, and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor communicates with the memory via the bus, and the machine-readable instructions are executed by the processor to perform the steps of the alignment correction method for wafer asymmetric deformation marks as described in any one of claims 1 to 6.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, performs the steps of the alignment and correction method for wafer asymmetric deformation marks as described in any one of claims 1 to 6.

Images