Wafer edge contour extraction method
Through sliding windows and comprehensive analysis models, multi-physical parameters are integrated, and the adaptability of single parameters and fixed windows in wafer edge contour detection is solved, achieving efficient abnormality detection and process optimization.
Patent Information
- Application Number
- CN202510338462.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-03-21
- Publication Date
- 2025-07-18
AI Technical Summary
The prior art has a single parameter acquisition dimension in wafer edge profile detection, lacks the ability to collaborate multi-physics field, and is difficult to quantify cross-domain interactions. The fixed detection window cannot adapt to process changes, resulting in difficulty in attribution of abnormalities and missed detection of transient events.
Using sliding window technology, combining geometric stress, dynamic environment, material interaction and multi-physical field parameters, a comprehensive analysis model is established, and multi-scale monitoring and accurate anomaly detection is achieved through thermal-dynamic correlation analysis and material-energy balance index.
It significantly improves the online control accuracy and process stability of the wafer edge profile, and can quickly locate process parameter drift or equipment status misalignment, avoiding misjudgment of traditional empirical thresholds.
Smart Images
Figure CN120333336A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of semiconductor technology, and more particularly, to a method for extracting the edge profile of a wafer. Background Art
[0002] Wafer manufacturing is a core process in the semiconductor industry, and the precise control of its edge profile directly affects the chip yield and device performance. In the current technical field, methods such as optical detection and mechanical probes are mainly used to analyze the topography of the wafer edge, involving the measurement of basic parameters such as stress distribution and geometric dimensions.
[0003] The prior art usually uses a single detection module to perform segmented scanning on the wafer edge, collects local stress or thickness data through a sensor with a fixed frequency, and then judges the contour abnormality based on an empirical threshold. For example, the thickness fluctuation value is obtained through a laser interferometer, or the high-frequency signal during the polishing process is captured by an acoustic emission sensor, and then the risk of crack or deformation is inferred.
[0004] However, the prior art has significant deficiencies in practical applications: firstly, the parameter acquisition dimension is single, and the synergistic effects of geometric stress, dynamic environment, material interaction, and multi-physical fields are not integrated, resulting in difficulty in attributing process abnormalities; secondly, there is a lack of cross-domain coupling analysis ability, and traditional models are difficult to quantify the interaction between thermal deformation and dynamic vibration, and between material properties and energy fields; thirdly, the real-time monitoring granularity is insufficient, and the fixed detection window cannot adapt to the characteristic scale changes in different process stages, easily missing transient abnormal events. Summary of the Invention
[0005] In order to overcome the above-mentioned defects of the prior art, the present invention provides a method for extracting the edge profile of a wafer, through the following solutions, to solve the problems of single parameter acquisition dimension, lack of multi-physical field collaborative analysis ability, difficulty in quantifying cross-domain interaction, and the fixed detection window cannot adapt to process changes, resulting in difficulty in attributing abnormalities and missing transient events as mentioned in the above background art.
[0006] To achieve the above object, the present invention provides the following technical solution: A method for extracting the edge profile of a wafer, comprising the following steps:
[0007] S1: Sliding window: Establish a sliding window to monitor the target wafer. The window size and sliding step of the sliding window are configured according to the circular edge characteristic scale of the target wafer and the resolution of the measurement system, and each window is sequentially marked as 1, 2... n;
[0008] S2: Data acquisition: Used to collect the geometric stress parameters, dynamic environment parameters, material interaction parameters, and multi-physical field parameters of each sliding window, for monitoring the characteristics of the wafer manufacturing process and the edge profile within each sliding window;
[0009] S3: Data analysis: used to establish mathematical models to analyze the data collected by S2, including thermal deformation comprehensive index analysis model, dynamic coupling coefficient analysis model, cross-domain interaction factor analysis model and multi-field coupling index analysis model;
[0010] S4: Comprehensive analysis: A comprehensive analysis model is established based on the analysis results of S3 to conduct a comprehensive analysis on the target wafer edge profile extraction, including thermal-dynamic correlation analysis and material-energy balance index.
[0011] Preferably, the window size is specifically set to a 500ms time window, corresponding to 3-5 main shaft rotation cycles, and the sliding step size is specifically an update interval of 1 / 5 of the window size.
[0012] Preferably, the geometric stress parameters include laser interference edge thickness fluctuation value, residual stress distribution in the wafer clamping area, real-time measurement value of edge chamfer angle and radial temperature gradient during rotary polishing; the dynamic environmental parameters include edge chemical mechanical polishing acoustic emission spectrum peak, wafer adsorption vacuum pressure fluctuation value, edge photoresist coating thickness discreteness and clean room micro-vibration acceleration; the material interaction parameters include rotary etching instantaneous edge flow velocity, wafer material lattice distortion, edge plasma sheath potential and wafer carrier thermal expansion coefficient; the multi-physical field parameters include edge infrared thermal imaging temperature distribution, nanoindentation hardness gradient, ultraviolet reflectivity change rate and wafer clamping torque fluctuation value.
[0013] Preferably, the laser interference edge thickness fluctuation value in the geometric stress parameter is measured by a NanoMetrics M3000 interferometer in a circular scan at an accuracy of 0.5 μm, and thickness data is collected at intervals of 2°. The residual stress distribution in the wafer clamping area is measured in the whole field by a KLA GX2000 stress testing system in a wafer clamping state, and a Raman spectrum inversion algorithm is used to obtain a stress cloud map. The real-time measurement value of the chamfer angle is measured by a Keyence CV-X100 visual system in combination with a ring-shaped LED light source, and the angle deviation is calculated by an edge contour extraction algorithm. During rotary polishing, the radial temperature gradient is collected by a Fluke TiS75 thermal imager at a rate of 5 frames per second to collect the polishing disk surface temperature, and the radial direction gradient is calculated in combination with the wafer coordinate system.
[0014] Preferably, for the acquisition of the peak value of the acoustic emission spectrum during edge chemical mechanical polishing among the dynamic environmental parameters, a PAC Micro-II sensor array is used to arrange a three-axis array inside the polishing head. The energy characteristics in the frequency band of 200 - 150 kHz are extracted through wavelet packet decomposition technology. The vacuum pressure fluctuation value of wafer adsorption is recorded by a Mensor CPT6000 digital pressure gauge at a sampling rate of 1000 Hz for the pressure pulsation in the adsorption pipeline, and the encoder signal of the polishing machine is triggered synchronously. For the measurement of the thickness dispersion of edge photoresist coating, a Bruker Contour Elite white light interferometer is used for 12-point sampling in the edge annular area, and the process reference line fluctuation is excluded when calculating the thickness standard deviation. For the monitoring of the micro-vibration acceleration in the clean room, a PCB 356A01 three-axis accelerometer is used and installed on the spindle flange of the polishing machine through a magnetic suction base, and the vibration energy integration is carried out according to the ISO 10816 standard.
[0015] Preferably, for the instantaneous edge flow rate during rotary etching among the material interaction parameters, a Siemens MAG5000 electromagnetic flowmeter is used for real-time monitoring at the end of the supply pipeline, and the elbow effect is corrected by combining Computational Fluid Dynamics simulation. The lattice distortion amount of the wafer material is determined by measuring the full width at half maximum of the rocking curve of the crystal plane in the edge area through a Rigaku SmartLab high-resolution XRD system. The edge plasma sheath potential is measured with spatial resolution using an Impedans ALPEN RF probe array at a position 5 mm away from the edge in the reaction chamber, and the 13.56 MHz radio frequency harmonic components are collected synchronously. For the determination of the thermal expansion coefficient of the wafer carrier, a TA Instruments DIL 802 dynamic thermomechanical analyzer is used to measure the linear expansion rate of the carrier material at a rate of 5 °C / min within the process temperature range.
[0016] Preferably, for the nanoindentation hardness gradient among the multi-physical field parameters, a CSM NHT3 nanoindenter is used to perform indentation tests at intervals of 50 μm along the radius direction of the wafer, and the hardness distribution curve is calculated by the Oliver-Pharr method. For the acquisition of the edge infrared thermal imaging temperature distribution, an FLIR A65SC camera is used to perform phase-locked shooting in cooperation with the polishing liquid injection cycle, and the TSR algorithm is used to enhance the thermal gradient resolution. The change rate of ultraviolet reflectivity is sampled for the reflected light intensity by an Ocean Insight QE Pro spectrometer at a wavelength of 280 nm, and the edge circumferential scan is realized through a moving platform. For the monitoring of the torque fluctuation value of wafer clamping, a Kistler 4507B dynamic torque sensor is integrated at the rotating shaft of the vacuum chuck to record the torque impact waveform during the process start and stop phases in real time.
[0017] Preferably, the comprehensive analysis model of the thermal deformation index is specifically expressed as: G1 represents the comprehensive index of thermal deformation, ΔT represents the fluctuation value of the laser interference edge thickness, σ_T = 0.5μm represents the reference thickness, S_max represents the peak value of residual stress, S_0 = 50MPa represents the reference stress, θ_avg represents the average chamfer angle, θ_ref = 22° represents the reference angle, represents the radial temperature gradient, and R = 150mm represents the wafer radius.
[0018] Preferably, the dynamic coupling coefficient analysis model is specifically expressed as: G2 = (f peak / f c ) a ×exp(P vac / P0)+(σ thk ×a vib ) / (1 + η), where G2 represents the dynamic coupling coefficient, f_peak represents the main frequency of acoustic emission, f_c = 120kHz represents the cut-off frequency, α = 0.8 is the material coefficient, P_vac represents the amplitude of vacuum pressure fluctuation, P_0 = 10Pa represents the reference pressure, σ_thk represents the standard deviation of photoresist thickness, η = 3nm represents the process tolerance, and a_vib represents the effective value of micro-vibration acceleration.
[0019] Preferably, the cross-domain interaction factor analysis model is specifically expressed as: G3 = (v flow ×ε lat )(1 / 3) / (Φ plasma ×a exp ×T β ), where v_flow represents the etching solution edge flow rate, ε_lat represents the lattice distortion amount, β = 0.25 represents the temperature coefficient, Φ_plasma represents the plasma sheath potential, α_exp represents the vehicle thermal expansion coefficient, and T is the process temperature.
[0020] Preferably, the multi-field coupling index analysis model is specifically expressed as: G4 represents the multi-field coupling index, ΔH represents the difference in nano-hardness gradient, H_0 = 12GPa represents the reference hardness, T_edge represents the infrared thermal imaging edge temperature, ΔR_UV represents the change rate of ultraviolet reflectivity, τ_max represents the peak value of clamping torque, and E_c = 200N·m represents the critical torque.
[0021] Preferably, the thermal-dynamic correlation analysis is specifically expressed as: K1 = G1 / G2 × ln(1 + |G1 - G2|). When K1 > 1.2, it indicates that thermal deformation dominates the anomaly. When K1 < 0.8, it indicates that dynamic vibration dominates the anomaly.
[0022] Preferably, the material-energy balance index is specifically expressed as: K2 = (G3 × G4) (1 / 3) / (1 + e (G3-G4) ). When K2 > 0.75, it indicates a mismatch between material properties and the energy field. When K2 < 0.35, it indicates a drift in process parameters.
[0023] Technical effects and advantages of the present invention:
[0024] 1. Through the dynamic configuration of the sliding window, the present invention realizes multi-scale monitoring of the entire process chain of wafer manufacturing. The window size and step size are adaptively adjusted according to the edge feature scale, which can not only capture the transient fluctuations within the spindle rotation period but also take into account the long-term process stability analysis, significantly improving the timeliness and coverage of anomaly detection;
[0025] 2. By integrating four types of parameters, namely geometric stress, dynamic environment, material interaction, and multi-physical fields, the present invention establishes a comprehensive analysis model covering thermal deformation, dynamic coupling, cross-domain interaction, and multi-field coupling. This solution breaks through the limitations of traditional single-parameter analysis and can accurately analyze the combined effects of cross-domain factors such as polishing fluid flow rate, plasma sheath potential, and carrier thermal expansion on the edge profile, providing a multi-dimensional decision-making basis for process optimization;
[0026] 3. The innovative thermal-dynamic correlation analysis and material-energy balance index model of the present invention can clearly distinguish anomalies dominated by thermal deformation from those dominated by dynamic vibration and quantify the matching degree between material properties and the energy field. This technology avoids the misjudgment risk of traditional empirical thresholds, can quickly locate process parameter drift or equipment state misalignment problems, and significantly improves the online control accuracy and process stability of the wafer edge profile. Description of the Drawings
[0027] Figure 1 It is a schematic diagram of the overall structure of the present invention. Detailed Embodiments
[0028] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.
[0029] This embodiment discloses a method for extracting the wafer edge profile. Referring to Figure 1 as shown, it includes the following steps:
[0030] S1: Sliding window: A sliding window is established to monitor the target wafer. The window size and sliding step size of the sliding window are configured according to the circular edge feature scale of the target wafer and the resolution of the measurement system, and each window is sequentially marked as 1, 2... n.
[0031] The window size is specifically set to a 500ms time window, corresponding to 3-5 spindle rotation cycles, and the sliding step is specifically set to an update interval of 1 / 5 of the window size.
[0032] S2: Data acquisition: used to collect the geometric stress parameters, dynamic environment parameters, material interaction parameters and multi-physical field parameters of each sliding window, and to monitor the wafer manufacturing process characteristics and edge profile in each sliding window.
[0033] The geometric stress parameters include the laser interference edge thickness fluctuation value, the residual stress distribution in the wafer clamping area, the real-time measurement value of the edge chamfer angle, and the radial temperature gradient during rotary polishing; the dynamic environmental parameters include the edge chemical mechanical polishing acoustic emission spectrum peak, the wafer adsorption vacuum pressure fluctuation value, the edge photoresist coating thickness discreteness, and the clean room micro-vibration acceleration; the material interaction parameters include the instantaneous edge flow velocity of rotary etching, the lattice distortion of the wafer material, the edge plasma sheath potential, and the thermal expansion coefficient of the wafer carrier; the multi-physical field parameters include the edge infrared thermal imaging temperature distribution, the nanoindentation hardness gradient, the ultraviolet reflectivity change rate, and the wafer clamping torque fluctuation value.
[0034] The laser interference edge thickness fluctuation value in the geometric stress parameters is measured by a NanoMetrics M3000 interferometer in a circular scan at an accuracy of 0.5μm, and thickness data is collected at intervals of 2°. The residual stress distribution in the wafer clamping area is measured in the whole field by a KLAGX2000 stress testing system in the wafer clamping state, and the stress cloud map is obtained by a Raman spectrum inversion algorithm. The real-time measurement value of the chamfer angle is obtained by a Keyence CV-X100 visual system with a circular LED light source, and the angle deviation is calculated by an edge profile extraction algorithm. During rotary polishing, the radial temperature gradient is collected by a Fluke TiS75 thermal imager at a rate of 5 frames per second to collect the polishing disk surface temperature, and the radial direction gradient is calculated in combination with the wafer coordinate system.
[0035] For the peak acquisition of acoustic emission spectrum during edge chemical mechanical polishing in dynamic environmental parameters, a three-axis array is arranged inside the polishing head using a PAC Micro-II sensor group. The energy characteristics of the 200-150kHz frequency band are extracted by wavelet packet decomposition technology. The wafer adsorption vacuum pressure fluctuation value is recorded by a Mensor CPT6000 digital pressure gauge at a sampling rate of 1000Hz to record the adsorption pipeline pressure pulsation and synchronously trigger the polishing machine encoder signal. The edge photoresist coating thickness dispersion is measured by a Bruker ContourElite white light interferometer to perform 12-point sampling in the edge annular area. The process baseline fluctuation is excluded when calculating the thickness standard deviation. The clean room micro-vibration acceleration monitoring uses a PCB 356A01 three-axis accelerometer, which is installed on the main shaft flange of the polishing machine through a magnetic base, and the vibration energy integration is performed according to the ISO 10816 standard.
[0036] For the rotational etching instantaneous edge flow rate in the material interaction parameters, it is monitored in real time at the end of the supply pipeline using a Siemens MAG5000 electromagnetic flowmeter. The elbow effect is corrected by combining Computational Fluid Dynamics simulations. The lattice distortion of the wafer material is measured by determining the full width at half maximum of the rocking curve of the crystal plane in the edge region using a Rigaku SmartLab high-resolution XRD system. The edge plasma sheath potential is measured with spatial resolution using an Impedans ALPEN RF probe set at a distance of 5 mm from the edge in the reaction chamber, and the 13.56 MHz radio frequency harmonic components are synchronously collected. The coefficient of thermal expansion of the wafer carrier is measured using a TA Instruments DIL 802 dynamic thermomechanical analyzer, and the linear expansion rate of the carrier material is measured at a rate of 5 °C / min within the process temperature range.
[0037] For the nanoindentation hardness gradient in the multi-physical field parameters, a CSM NHT3 nanoindenter is used to perform indentation tests at 50-μm intervals along the radius direction of the wafer, and the hardness distribution curve is calculated by the Oliver-Pharr method. The edge infrared thermal imaging temperature distribution is collected using a FLIR A65SC camera with phase-locked shooting in conjunction with the polishing fluid injection period, and the TSR algorithm is used to enhance the thermal gradient resolution. The change rate of ultraviolet reflectivity is sampled for the reflected light intensity by an Ocean Insight QE Pro spectrometer at a wavelength of 280 nm, and circumferential edge scanning is achieved through a moving platform. The monitoring of the wafer clamping torque fluctuation value uses a Kistler 4507B dynamic torque sensor integrated at the rotating shaft of the vacuum chuck, and the torque impact waveform during the process start and stop phases is recorded in real time.
[0038] S3: Data analysis: Used to establish mathematical models to analyze the data collected in S2, including a comprehensive thermal deformation index analysis model, a dynamic coupling coefficient analysis model, a cross-domain interaction factor analysis model, and a multi-field coupling index analysis model.
[0039] The comprehensive thermal deformation index analysis model is specifically expressed as: G1 represents the comprehensive thermal deformation index, ΔT represents the laser interference edge thickness fluctuation value, σ_T = 0.5 μm represents the reference thickness, S_max represents the peak residual stress, S_0 = 50 MPa represents the reference stress, θ_avg represents the average chamfer angle, θ_ref = 22° represents the reference angle, represents the radial temperature gradient, and R = 150 mm represents the wafer radius.
[0040] The dynamic coupling coefficient analysis model is specifically expressed as: G2 = (f peak / fc ) a × exp(P vac / P0)+(σ thk × a vib ) / (1 + η), where G2 represents the dynamic coupling coefficient, f_peak represents the main frequency of acoustic emission, f_c = 120 kHz represents the cut-off frequency, α = 0.8 is the material coefficient, P_vac represents the amplitude of vacuum pressure fluctuation, P_0 = 10 Pa represents the reference pressure, σ_thk represents the standard deviation of photoresist thickness, η = 3 nm represents the process tolerance, and a_vib represents the effective value of micro-vibration acceleration.
[0041] The cross-domain interaction factor analysis model is specifically expressed as: G3 = (v flow × ε lat )(1 / 3) / (Φ plasma × a exp × T β ), where v_flow represents the etching solution edge flow rate, ε_lat represents the lattice distortion amount, β = 0.25 represents the temperature coefficient, Φ_plasma represents the plasma sheath potential, α_exp represents the vehicle thermal expansion coefficient, and T is the process temperature.
[0042] The multi-field coupling index analysis model is specifically expressed as: G4 represents the multi-field coupling index, ΔH represents the nano-hardness gradient difference, H_0 = 12 GPa represents the reference hardness, T_edge represents the infrared thermal imaging edge temperature, ΔR_UV represents the ultraviolet reflectance change rate, τ_max represents the clamping torque peak, and E_c = 200 N·m represents the critical torque.
[0043] S4: Comprehensive analysis: Based on the analysis results of S3, establish a comprehensive analysis model to conduct a comprehensive analysis on the extraction of the target wafer edge profile, including thermo-dynamic correlation analysis and material-energy balance index.
[0044] The thermo-dynamic correlation analysis is specifically expressed as: K1 = G1 / G2 × ln(1 + |G1 - G2|). When K1 > 1.2, it indicates that thermal deformation dominates the abnormality; when K1 < 0.8, it indicates that dynamic vibration dominates the abnormality.
[0045] The material-energy balance index is specifically expressed as: K2 = (G3 / G4) (1 / 3) / (1 + e(G3 - G4)). When K2 > 0.75, it indicates a mismatch between material properties and the energy field; when K2 < 0.35, it indicates a drift in process parameters.
[0046] The present invention first dynamically configures the sliding window parameters based on the edge feature scale of the target wafer and the resolution of the measurement system, sets a 500 ms time window, covering 3 - 5 spindle rotation periods, and updates it with a step size of 1 / 5 of the window size to achieve multi-scale process monitoring; then obtains the laser interference edge thickness fluctuation value through the annular scanning of the NanoMetrics M3000 interferometer, measures the residual stress distribution in the clamping area using the KLA GX2000 stress test system, combines the Keyence CV-X100 vision system to monitor the chamfer angle deviation in real time, and synchronously collects the radial temperature gradient data of the Fluke TiS75 thermal imager to construct a geometric stress parameter set; at the same time, deploys a PAC Micro-II acoustic emission sensor group to capture the polishing acoustic emission spectrum in the 200 - 150 kHz frequency band, records the vacuum pressure fluctuation through a Mensor CPT6000 pressure gauge, calculates the photoresist thickness dispersion using a Bruker Contour Elite white light interferometer, and monitors the micro-vibration acceleration using a PCB 356A01 accelerometer to form a dynamic environment parameter set; further uses a Siemens MAG5000 electromagnetic flowmeter to measure the edge flow rate of the etching solution, analyzes the lattice distortion amount through a Rigaku SmartLab XRD system, obtains the plasma sheath potential using an Impedans ALPEN RF probe, and combines a TA Instruments DIL 802 thermomechanical analyzer to determine the thermal expansion coefficient of the carrier to construct a material interaction parameter set; at the same time, integrates the hardness gradient data of a CSM NHT3 nanoindenter, the infrared thermal imaging temperature distribution of a FLIR A65SC camera, the ultraviolet reflectivity change of an Ocean Insight QE Pro spectrometer, and the clamping torque fluctuation value of a Kistler 4507B torque sensor to form a multi-physical field parameter set. In the data analysis stage, a comprehensive thermal deformation index model, a dynamic coupling coefficient model, a cross-domain interaction factor model, and a multi-field coupling index model are respectively established. By calculating the index values of each model in real time, finally, a thermal-dynamic correlation analysis and a material-energy balance index calculation are performed. When K1 > 1.2, it is determined that the thermal deformation is the dominant anomaly; when K1 < 0.8, it is determined that the dynamic vibration is abnormal; when K2 > 0.75, it indicates a mismatch between the material-energy fields; when K2 < 0.35, it indicates a process parameter drift, thereby achieving accurate anomaly attribution and process control.
[0047] Through the dynamic configuration of the sliding window, the present invention realizes multi-scale monitoring of the entire process chain of wafer manufacturing. The window size and step length are adaptively adjusted according to the edge feature scale, which can not only capture the transient fluctuations within the spindle rotation period but also take into account the long-term process stability analysis, significantly improving the timeliness and coverage of anomaly detection. By integrating four types of parameters, namely geometric stress, dynamic environment, material interaction, and multi-physical fields, the present invention establishes a comprehensive analysis model covering thermal deformation, dynamic coupling, cross-domain interaction, and multi-field coupling. This solution breaks through the limitations of traditional single-parameter analysis, can accurately analyze the combined effects of cross-domain factors such as polishing fluid flow rate, plasma sheath potential, and carrier thermal expansion on the edge profile, and provides a multi-dimensional decision-making basis for process optimization. The innovative thermal-dynamic correlation analysis and material-energy balance index model of the present invention can clearly distinguish anomalies dominated by thermal deformation from those dominated by dynamic vibration and quantify the matching degree between material properties and the energy field. This technology avoids the misjudgment risk of traditional empirical thresholds, can quickly locate problems such as process parameter drift or equipment state inaccuracy, and significantly improves the online control accuracy and process stability of the wafer edge profile.
[0048] Secondly, in the accompanying drawings of the disclosed embodiments of the present invention, only the structures related to the disclosed embodiments are involved. Other structures can refer to the general design. Without conflict, the same embodiment and different embodiments of the present invention can be combined with each other.
[0049] Finally, the above are only the preferred embodiments of the present invention and are not used to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.
Claims
1. A method for extracting the edge profile of a wafer, characterized in that, It includes the following steps: S1: Sliding window: A sliding window is established to monitor the target wafer. The window size and sliding step of the sliding window are configured according to the circular edge feature scale of the target wafer and the resolution of the measurement system, and each window is sequentially marked as 1, 2... n; S2: Data acquisition: It is used to collect the geometric stress parameters, dynamic environment parameters, material interaction parameters, and multi-physical field parameters of each sliding window, and is used to monitor the wafer manufacturing process characteristics and edge profiles within each sliding window; S3: Data analysis: It is used to establish a mathematical model to analyze the data collected in S2, including a comprehensive thermal deformation index analysis model, a dynamic coupling coefficient analysis model, a cross-domain interaction factor analysis model, and a multi-field coupling index analysis model; S4: Comprehensive analysis: A comprehensive analysis model is established based on the analysis results of S3 to comprehensively analyze the extraction of the edge profile of the target wafer, including thermal-dynamic correlation analysis and material-energy balance index.
2. A method for extracting a wafer edge profile according to claim 1, characterized in that: The window size is specifically set to a 500ms time window, corresponding to 3-5 main axis rotation periods, and the sliding step is specifically the update interval of 1 / 5 of the window size.
3. A method for extracting the edge profile of a wafer according to claim 1, characterized in that: The geometric stress parameters include the laser interference edge thickness fluctuation value, the residual stress distribution in the wafer clamping area, the real-time measurement value of the edge chamfer angle, and the radial temperature gradient during rotary polishing; the dynamic environment parameters include the peak value of the acoustic emission spectrum during edge chemical mechanical polishing, the vacuum pressure fluctuation value of wafer adsorption, the thickness dispersion of edge photoresist coating, and the micro-vibration acceleration in the clean room; the material interaction parameters include the instantaneous edge flow rate during rotary etching, the lattice distortion amount of the wafer material, the edge plasma sheath potential, and the thermal expansion coefficient of the wafer carrier; the multi-physical field parameters include the edge infrared thermal imaging temperature distribution, the nano-indentation hardness gradient, the change rate of ultraviolet reflectivity, and the fluctuation value of the wafer clamping torque.
4. A method for extracting a wafer edge profile according to claim 1, characterized in that: The comprehensive thermal deformation index analysis model is specifically expressed as: G1 represents the comprehensive thermal deformation index, ΔT represents the laser interference fringe thickness fluctuation value, σ_T = 0.5μm represents the reference thickness, S_max represents the peak residual stress, S_0 = 50MPa represents the reference stress, θ_avg represents the average chamfer angle, θ_ref = 22° represents the reference angle, represents the radial temperature gradient, and R = 150mm represents the wafer radius.
5. A method for extracting the edge profile of a wafer according to claim 1, characterized in that: The dynamic coupling coefficient analysis model is specifically expressed as: G2 = (f peak / f c ) a × exp(P vac / P0)+(σ thk × a vib ) / (1 + η), where G2 represents the dynamic coupling coefficient, f_peak represents the main frequency of acoustic emission, f_c = 120 kHz represents the cut-off frequency, α = 0.8 is the material coefficient, P_vac represents the amplitude of vacuum pressure fluctuation, P_0 = 10 Pa represents the reference pressure, σ_thk represents the standard deviation of photoresist thickness, η = 3 nm represents the process tolerance, and a_vib represents the effective value of micro-vibration acceleration.
6. A method for extracting a wafer edge profile according to claim 1, characterized in that: The cross-domain interaction factor analysis model is specifically expressed as: G3 = (v flo w × ε lat ) (1 / 3) / (Φ plasma × a exp × T β ), where v_flow represents the etching solution edge flow rate, ε_lat represents the lattice distortion variable, β = 0.25 represents the temperature coefficient, Φ_plasma represents the plasma sheath potential, α_exp represents the vehicle thermal expansion coefficient, and T is the process temperature.
7. A method for extracting the edge profile of a wafer according to claim 1, characterized in that: The multi-field coupling index analysis model is specifically expressed as: G4 represents the multi-field coupling index, ΔH represents the difference in nano-hardness gradient, H_0 = 12 GPa represents the reference hardness, T_edge represents the infrared thermal imaging edge temperature, ΔR_UV represents the change rate of ultraviolet reflectivity, τ_max represents the peak clamping torque, and E_c = 200 N·m represents the critical torque.
8. A method for extracting a wafer edge profile according to claim 1, characterized in that: The thermal-dynamic correlation analysis is specifically expressed as: K1 = G1 / G2 × ln(1 + |G1 - G2|). When K1 > 1.2, it indicates that thermal deformation dominates the anomaly. When K1 < 0.8, it indicates that dynamic vibration dominates the anomaly.
9. A method for extracting a wafer edge profile according to claim 1, characterized in that: The material-energy balance index is specifically expressed as: K2 = (G3 × G4) (1 / 3) / (1 + e (G3-G4) ). When K2 > 0.75, it indicates a mismatch between the material properties and the energy field. When K2 < 0.35, it indicates a drift in the process parameters.