A digital detection system and method for wafer ultrasonic field

Abstract

The present invention relates to the field of semiconductor detection technology, and discloses a digital detection system and method for a wafer ultrasonic field. The digital detection method for a wafer ultrasonic field comprises: configuring an acoustic topological insulator coupling system to construct a scattering-free acoustic wave transmission channel; constructing a heterogeneous metasurface acoustic field control system to obtain depth dimension information; applying an acoustic holographic compensation algorithm to eliminate imaging distortion in a curved area and output a compensated acoustic field distribution; executing a phase recovery and annular reconstruction algorithm to achieve high-quality three-dimensional imaging of an edge area; implementing an adaptive coupling adjustment system to automatically adjust coupling material properties according to the quality of a real-time monitored edge reflection signal. The present invention constructs a scattering-free acoustic wave transmission channel through the acoustic topological insulator coupling system, effectively solving the problem of severe acoustic wave scattering in the edge area, thereby improving the detection sensitivity of the wafer edge area, reducing the defect missed detection rate compared with traditional methods, and improving the integrity of wafer detection.

Description

Technical Field

[0001] The present invention relates to the field of semiconductor detection technology, and more particularly to a digital detection system and method for a wafer ultrasonic field. Background Art

[0002] With the continuous advancement of semiconductor manufacturing processes, wafer sizes have gradually increased, and process precision requirements have continued to rise. The wafer edge area has become a high-incidence area for defects, especially the edge roll-off area, which is particularly challenging to detect. Defects in these edge areas can cause serious problems in subsequent chip manufacturing processes, or even damage the entire wafer, resulting in significant economic losses.

[0003] Ultrasonic testing technology is currently widely used in wafer quality inspection due to its advantages, such as being non-destructive and capable of deep imaging. However, traditional ultrasonic testing methods face several technical challenges at the wafer edge. Firstly, due to the unique geometry of the wafer edge, sound waves scatter significantly in this area, resulting in a low signal-to-noise ratio and insufficient detection sensitivity. Secondly, in the curved transition zone, the ultrasonic wave propagation path is complex, reducing imaging quality and making it difficult to accurately identify small defects. Traditional testing methods, particularly within the 5-10mm radius of the edge and in the curved transition zone, have a high rate of missed defects, failing to meet the full-area, high-precision inspection requirements of modern semiconductor manufacturing.

[0004] Existing technologies attempt to improve edge detection capabilities by refining ultrasonic transducer design and optimizing signal processing algorithms. However, these methods are mostly extensions or improvements to existing technologies, lacking a systematic solution to fundamentally address the unique acoustic field characteristics of edge regions. For example, some technologies employ multi-angle scanning to increase information acquisition in edge regions, but they still cannot overcome the signal quality degradation caused by acoustic wave scattering. Other technologies use high-frequency ultrasonic probes to improve spatial resolution, but imaging distortion still persists in curved edge regions, and coupling stability is poor.

[0005] Therefore, there is an urgent need for an ultrasonic field digital detection method designed for the characteristics of the wafer edge area, which can systematically solve the technical difficulties of edge area detection from multiple dimensions such as sound wave transmission mechanism, sound field control, and imaging algorithm, and provide higher precision and more reliable non-destructive detection capabilities for the entire wafer area. Summary of the Invention

[0006] The present invention provides a digital detection system and method for a wafer ultrasonic field, which solve the technical problems in related technologies of low accuracy and poor reliability of ultrasonic detection of wafer edge areas, especially curved transition areas.

[0007] The present invention provides a digital detection method for a wafer ultrasonic field, comprising:

[0008] Configure an acoustic topological insulator coupling system to build a scattering-free acoustic wave transmission channel;

[0009] Based on the acoustic topological insulator coupling system, a heterogeneous metasurface acoustic field control system is constructed to obtain depth dimension information;

[0010] Based on the depth dimension information, the acoustic holographic compensation algorithm is applied to eliminate the imaging distortion of the curved area. The curved area at the edge of the wafer is virtually flattened into an equivalent plane by calculating the compensation wavefront function, and the sound field distribution after compensation is calculated.

[0011] Using the compensated sound field distribution, phase recovery and annular reconstruction algorithms are performed to achieve high-quality three-dimensional imaging of edge areas.

[0012] Based on the reconstructed three-dimensional imaging results, an adaptive coupling adjustment system is implemented to automatically adjust the coupling material properties according to the quality of the edge reflection signal monitored in real time. The steps of implementing the adaptive coupling adjustment system include:

[0013] An adaptive coupling layer based on phase change material is constructed. Acoustic phase change material is selected as the coupling medium. Its acoustic impedance Z satisfies the following under temperature changes:

[0014] ;

[0015] in For temperature changes The acoustic impedance value under is the reference impedance value, is the impedance variation coefficient, is the temperature response sensitivity coefficient, is the temperature difference relative to the critical temperature of phase transition, is the hyperbolic tangent function;

[0016] Build a real-time monitoring subsystem to continuously evaluate the quality of edge reflection signals and calculate signal quality evaluation indicators based on the signal-to-noise ratio and waveform integrity of the echo signal;

[0017] Apply feedback control algorithms to automatically adjust the temperature distribution of the phase change material based on signal quality;

[0018] Outputs the adaptively optimized coupling state to achieve the best acoustic match between the detection system and the wafer edge.

[0019] Furthermore, the step of configuring the acoustic topological insulator coupling system includes:

[0020] Construct a periodic array of acoustic units, each unit containing a resonant cavity and connecting channels, to form an acoustic topological insulator with a specific band structure;

[0021] An interface layer is configured at the boundary of the acoustic topological insulator to generate localized boundary phonon states, which form a unidirectional topologically protected acoustic wave mode by breaking the time reversal symmetry.

[0022] Precisely align the topologically protected acoustic coupler with the wafer edge area, ensuring maximum coupling efficiency between the two through a precision positioning system;

[0023] A topologically protected acoustic wave transmission channel is formed. The topologically protected acoustic wave transmission channel has anti-scattering and anti-interference properties, and can maintain high fidelity of acoustic wave transmission in the curved area of ​​the wafer edge.

[0024] Furthermore, the steps of constructing the heterogeneous metasurface acoustic field control system include:

[0025] Design and construct a subwavelength unit array structure, where the size of each unit satisfies:

[0026] ;

[0027] in is the unit size, is the working wavelength;

[0028] Apply the gradient phase control algorithm to make the vibration phase between adjacent subwavelength units satisfy:

[0029] ;

[0030] in Indicates the The phase of each unit, Indicates the The phase of each unit, is the design wave number, which represents the desired sound wave propagation characteristics, is the unit spacing;

[0031] Construct a hierarchical structure of heterogeneous metasurfaces and form a three-dimensional sound field control system by stacking metasurface layers with different functions;

[0032] The output is the vertical sound field distribution information after being regulated by the heterogeneous metasurface. The vertical sound field distribution information includes the acoustic characteristics of different depth layers in the wafer edge area.

[0033] Furthermore, the step of applying the acoustic holographic compensation algorithm includes:

[0034] Collect 3D geometric data of the wafer edge bending area and build an accurate digital model;

[0035] Calculate the ultrasonic propagation function in the curved area based on the sound wave propagation theory , the ultrasonic propagation function describes the To the receiving location At angular frequency Sound wave propagation characteristics under ;

[0036] The layered iterative holographic projection algorithm is applied. Based on the principle of solving the inverse problem, the layered iterative holographic projection algorithm calculates the compensation wavefront function based on the difference between the ideal plane wave field and the actual measured wave field.

[0037] The compensated sound field distribution is calculated and output, and the complex curved area is virtually flattened into an equivalent plane to achieve distortion-free reconstruction of the sound field.

[0038] Furthermore, the step of executing the phase recovery and ring reconstruction algorithm includes:

[0039] The improved Gerchberg-Saxton algorithm is applied for phase retrieval, and the complete wavefield information is extracted from the amplitude measurement data through iterative transformation between the frequency domain and the spatial domain.

[0040] Configure and apply a segmented annular array receiving system to capture wafer edge scattered signals. The segmented annular array receiving system includes multiple arc-shaped sensor array modules, each module containing 64 to 128 independently addressable piezoelectric ultrasonic transducer elements;

[0041] Based on the polar coordinate system, a dedicated sound field reconstruction algorithm is applied to process the data collected by the annular array. The sound field reconstruction algorithm uses a combination of Hankel transform and angular harmonic decomposition.

[0042] Integrate phase recovery and annular reconstruction results to output a high-resolution three-dimensional acoustic field distribution map of the wafer edge area.

[0043] Furthermore, the construction of the periodic acoustic unit array includes two types: a "honeycomb rod" structure and a spiral acoustic topological insulator structure. In the honeycomb rod structure, the hexagonal honeycombs form a resonant cavity, and the connecting rods form a waveguide channel. The topological characteristics of the spiral structure are determined by the number of spiral arms. and helical angle Joint decision.

[0044] Furthermore, the subwavelength unit array structure includes three types: resonance type, refraction type and composite type. According to the detection requirements, the "gradient aperture cylindrical array" structure is selected for surface and shallow defect detection, and the "resonance cavity spring" composite structure is selected for deep defect detection inside the wafer.

[0045] Furthermore, the improved GerchbergSaxton algorithm includes two key optimizations: edge enhancement constraint and adaptive step size control. The edge enhancement constraint gives higher weights to the data points in the wafer edge area, and the adaptive step size control introduces a step size factor. , so that the phase update satisfies:

[0046] ;

[0047] in For the The position in the iteration The phase estimate at For the The position in the iteration The phase estimate at For the The adaptive step size factor for the iteration, Represents the complex field converted back to the spatial domain after applying the frequency domain constraint The phase angle, is the spatial position vector.

[0048] The present invention provides a digital detection system for wafer ultrasonic fields, which is used to perform the above-mentioned digital detection method for wafer ultrasonic fields, comprising:

[0049] Acoustic topological insulator coupling module, used to construct a scattering-free acoustic wave transmission channel;

[0050] Heterogeneous metasurface acoustic field control module, used to obtain depth dimension information;

[0051] Acoustic holographic compensation processing module, used to eliminate imaging distortion in curved areas;

[0052] Phase recovery and annular reconstruction processing module, used to achieve high-quality three-dimensional imaging of edge areas;

[0053] Adaptive coupling adjustment module to ensure detection stability and reliability.

[0054] The beneficial effects of the present invention are as follows: by constructing a scattering-free acoustic wave transmission channel through an acoustic topological insulator coupling system, the problem of severe acoustic wave scattering in the edge area is effectively solved, the detection sensitivity of the wafer edge area is improved, the defect missed detection rate is reduced compared with the traditional method, and the integrity of wafer detection is improved;

[0055] The application of acoustic holographic compensation algorithm overcomes the imaging distortion problem caused by the complex ultrasonic propagation path in the curved edge area, making the imaging quality in the curved edge area comparable to that in the flat area of ​​the wafer, and improving the defect location accuracy.

[0056] Through heterogeneous metasurface acoustic field control technology and phase recovery algorithms, the 3D imaging capability of the edge area is enhanced, which can clearly distinguish the vertical layered structure at the edge of the wafer, providing important information for comprehensive defect characterization.

[0057] The adaptive coupling adjustment system responds to changes in wafer edge geometry and contact status in real time, automatically optimizing coupling conditions to ensure consistent inspection results.

[0058] The signal quality fluctuation under different wafer samples and detection conditions is reduced, and the reliability of detection is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 It is a flow chart of a digital detection method of a wafer ultrasonic field in the present invention;

[0060] Figure 2 is a flow chart of step 1 in the present invention;

[0061] Figure 3 It is a flow chart of step 2 in the present invention;

[0062] Figure 4 It is a flow chart of step 3 in the present invention;

[0063] Figure 5 is a flow chart of step 4 in the present invention;

[0064] Figure 6 It is a flow chart of step 5 in the present invention. DETAILED DESCRIPTION

[0065] The subject matter described herein will now be discussed with reference to example embodiments. It should be understood that these embodiments are discussed solely to enable those skilled in the art to better understand and implement the subject matter described herein, and that the functions and arrangements of the elements discussed may be varied without departing from the scope of this specification. Various examples may omit, substitute, or add various processes or components as needed. Furthermore, features described in some examples may be combined in other examples.

[0066] At least one embodiment of the present invention discloses a digital detection method of a wafer ultrasonic field, such as Figures 1 to 6 As shown, the following steps are included:

[0067] Step 1: Configure an acoustic topological insulator coupling system to build a scattering-free acoustic wave transmission channel;

[0068] By applying the principle of acoustic topological insulators and configuring a specially structured acoustic wave coupling system, we can overcome the scattering loss caused by the geometric shape of the wafer edge and achieve efficient transmission of ultrasonic energy along the wafer edge. Specific configuration methods include:

[0069] Step 1.1, construct a periodic acoustic unit array;

[0070] Each unit contains a resonant cavity and a connecting channel, forming an acoustic topological insulator with a specific band structure. It should be noted that the frequency response characteristics of the acoustic unit satisfy:

[0071] ;

[0072] in Indicates the The resonant angular frequency of each unit, is the center angular frequency; is the coupling coefficient, which indicates the coupling strength between adjacent units; For the The wave vector corresponding to each unit is is the distance between adjacent cells, Represents the cosine function.

[0073] Alternatively, in some embodiments, the acoustic unit utilizes a "honeycomb rod" structure, where hexagonal honeycombs form a resonant cavity and connecting rods form a waveguide channel. This structure is particularly well-suited for edge detection of large-diameter 12-inch wafers because it has a wider topological energy band and can support acoustic wave transmission over a wider frequency range. In this configuration, the geometric dimensions of the resonant cavity and channel meet the following constraints:

[0074] ;

[0075] ;

[0076] in is the characteristic size of the resonant cavity, is the channel width; is the central operating wavelength of the system, which is equal to the speed of sound divided by the central angular frequency .

[0077] In another embodiment, when processing 8-inch or smaller wafers, a spiral acoustic topological insulator structure can be used, whose topological characteristics are determined by the number of spiral arms. and helical angle Jointly decided:

[0078] ;

[0079] in The wave vector in the spiral structure is The angular frequency of the sound wave, is the center angular frequency, is the coupling coefficient, is the number of spiral arms, is the sum index (an integer from 1 to N), is the wave vector, is the unit spacing, is the helical angle, is pi.

[0080] This spiral structure can better adapt to the edge of the wafer with a small curvature radius and provide smoother sound wave transmission.

[0081] Step 1.2, configure the interface layer at the boundary of the acoustic topological insulator;

[0082] By configuring an interface layer at the boundary of an acoustic topological insulator to generate localized boundary phonon states, the interface layer breaks the time reversal symmetry to form a unidirectional topologically protected acoustic wave mode, whose propagation characteristics are described by the following expression:

[0083] ;

[0084] in is the boundary Hamiltonian, which represents the energy of the boundary phonon state, is the group velocity, which indicates the energy propagation speed, is the wave vector along the boundary, is the energy gap parameter, which indicates the size of the topological protection band gap.

[0085] Step 1.3: precisely align the topological protection acoustic coupler with the wafer edge area;

[0086] The coupling efficiency between the two is maximized through the precision positioning system. satisfy:

[0087] ;

[0088] in It represents the efficiency of sound energy transmission. and are the acoustic impedances of the coupler and wafer, is the attenuation coefficient, which indicates the energy loss rate of sound waves during transmission. is the coupling length, is the base of natural logarithms.

[0089] For example, when testing wafers of different materials, the acoustic impedance matching can be optimized by adjusting the material composition of the acoustic topological insulator. kg / m²·s), a composite polymer material with an acoustic impedance close to this value can be used to construct the coupling layer. For wafer materials with higher acoustic impedance, such as silicon carbide or gallium nitride, a composite material containing ceramic particles can be used to increase the acoustic impedance of the coupling layer to ensure efficient energy transmission.

[0090] Step 1.4, output the topology protection acoustic wave transmission channel;

[0091] The topologically protected acoustic wave transmission channel has anti-scattering and anti-interference properties, and can maintain high fidelity of acoustic wave transmission in the curved area of ​​the wafer edge, providing a stable acoustic wave transmission environment for subsequent detection.

[0092] In some implementations, the ATI coupling system can further integrate an active noise suppression module, which effectively suppresses external interference by monitoring the ambient sound field in real time and generating anti-phase sound waves. This is particularly beneficial for wafer inspection in high-noise industrial environments, minimizing the impact of background noise on the inspection signal and improving signal quality.

[0093] Step 2: Based on the acoustic topological insulator coupling system, a heterogeneous metasurface acoustic field control system is constructed to obtain depth dimension information;

[0094] After completing the construction of the acoustic wave transmission channel, based on the propagation characteristics of ultrasonic waves in the vertical direction, a multi-layer sub-wavelength heterogeneous metasurface unit array is constructed to achieve precise control of the sound field and enhance the ability to obtain depth dimension information. It should be noted that the specific implementation process includes:

[0095] Step 2.1, design and construct a subwavelength unit array structure;

[0096] The size of each unit meets (in is the unit size, The unit structure adopts three types: resonance type, refraction type and composite type, and is configured according to the detection requirements.

[0097] In some embodiments, different subwavelength unit structures can be selected for different types of wafer defect detection. For example, for surface and shallow defects (such as scratches and microcracks), a "gradient aperture cylindrical array" structure can be used, where the cylindrical unit diameter decreases from the center to the edge, following the following relationship:

[0098] ;

[0099] in For the The diameter of the unit, is the center unit diameter, is the gradient coefficient, which controls the rate of diameter change, For the The normalized distance from each element to the center of the array.

[0100] This structure enables super-resolution focusing and is particularly suitable for detecting tiny surface defects below 10μm.

[0101] For deep defects inside the wafer (such as voids, inclusions, etc.), a "resonance cavity spring" composite structure can be preferably used. It consists of a hollow cylindrical resonant cavity and a coil spring connector. It can produce multi-band acoustic responses and is suitable for layered imaging of defects at different depths.

[0102] Step 2.2, applying the gradient phase control algorithm;

[0103] The vibration phase between adjacent subwavelength units satisfies the following relationship:

[0104] ;

[0105] in Indicates the The phase of each unit, Indicates the The phase of each unit, is the design wave number, which represents the desired sound wave propagation characteristics, is the unit spacing.

[0106] In addition, by precisely controlling the geometric parameters (shape, size, holes, etc.) and material properties of each unit, precise control of the incident sound waves can be achieved.

[0107] Optionally, for detecting the curved area at the wafer edge, an adaptive phase control method is provided to measure the geometric curvature of the wafer edge in real time and dynamically adjust the phase gradient based on the measurement results:

[0108] ;

[0109] in For location The wave number at is the base wave number, is the wave number adjustment, For and location Curvature Related adjustment functions.

[0110] Through this adaptive phase control, the system can automatically optimize detection parameters for different wafer edge shapes and improve adaptability.

[0111] Step 2.3, constructing the hierarchical structure of the heterogeneous metasurface;

[0112] By stacking metasurface layers with different functions, a three-dimensional sound field control system is formed. Therefore, the mutual coupling relationship between each layer is described by the transfer matrix method:

[0113] ;

[0114] in and denote the input and output sound pressures, respectively, and denote the input and output particle velocities, respectively. 、 、 and are the transfer matrix elements of the sound pressure transfer coefficient, the sound pressure-velocity coupling coefficient, the velocity-sound pressure coupling coefficient, and the velocity transfer coefficient, respectively.

[0115] In practical wafer inspection applications, specific heterogeneous metasurface hierarchical configurations have been developed for wafers of varying thicknesses and materials. For example, for standard 300mm silicon wafers (725-775μm thick), a three-layer heterogeneous metasurface structure is employed: the first layer is an acoustic wave focusing layer, used to improve spatial resolution; the second layer is a spectrum separation layer, which decomposes different frequency components to enhance depth information; and the third layer is a phase compensation layer, which corrects for phase distortion caused by acoustic wave propagation within the wafer.

[0116] The transfer matrix of this three-layer structure can be expressed as the product of three basic transfer matrices:

[0117] ;

[0118] in is the total transfer matrix, 、 and are the transfer matrices of the first, second and third layers respectively.

[0119] Through this hierarchical design, the system can achieve optimal depth resolution performance when inspecting silicon wafers, and can distinguish adjacent layer defects with a pitch as small as 10μm.

[0120] Step 2.4, outputting the vertical sound field distribution information after being regulated by the heterogeneous metasurface;

[0121] The vertical acoustic field distribution information after the heterogeneous metasurface control includes the acoustic characteristics of different depth layers in the wafer edge area, providing basic data for the depth positioning and three-dimensional imaging of defects. It can be seen that the vertical resolution reaches the wavelength. , which is superior to the depth resolution capability of traditional ultrasonic testing.

[0122] In some implementations, the acoustic field distribution information can be further subdivided into multiple frequency bands, each corresponding to the acoustic characteristics of a specific depth range. For semiconductor wafer inspection, the inspection frequency range (e.g., 10-100 MHz) can be divided into 5-8 sub-bands. The low-frequency band (10-30 MHz) is primarily used to detect deep defects, the mid-frequency band (30-60 MHz) is suitable for detecting mid-layer defects, and the high-frequency band (60-100 MHz) targets surface and shallow defects. This multi-band analysis method improves the accuracy of defect depth location.

[0123] Step 3: Based on the depth dimension information, an acoustic holographic compensation algorithm is applied to eliminate the imaging distortion of the curved area. The curved area at the edge of the wafer is virtually flattened into an equivalent plane by calculating the compensation wavefront function, and the compensated sound field distribution is calculated.

[0124] After constructing the above-mentioned acoustic field control system, an accurate digital model is constructed for the curved area of ​​the wafer edge to calculate the propagation function of the ultrasonic wave in the curved area of ​​the wafer edge, and acoustic holographic compensation technology is used to achieve distortion-free imaging of the curved area. It should be noted that the specific implementation process includes:

[0125] Step 3.1, build an accurate digital model;

[0126] Collect three-dimensional geometric data of the wafer edge bending area, and use high-precision laser scanning or optical interferometry technology to obtain microscopic morphology data of the edge bending area with a resolution of micron level.

[0127] Step 3.2, calculate the ultrasonic propagation function of the curved area;

[0128] The calculation is based on the sound wave propagation theory, and the ultrasonic propagation function describes the position of the sound source. To the receiving location At angular frequency The sound wave propagation characteristics under:

[0129] ;

[0130] in is the propagation function, is the receiving position vector, is the sound source position vector, is the angular frequency, is the wave number, equal to Divided by the speed of sound, is the distance from the sound source to the receiving point, is the base of natural logarithms, is the imaginary unit, is pi, The correction term to consider boundary conditions and scattering effects is calculated using the finite element or boundary element method.

[0131] Step 3.3, calculate the compensation wavefront function;

[0132] It should be understood that based on the principle of solving the inverse problem, from the ideal plane wave field and the actual measured wave field Starting from the difference, the compensation wavefront function is calculated :

[0133] ;

[0134] in To compensate for the wavefront function (which contains amplitude and phase information and is used to offset the distortion of sound wave propagation caused by the curved area), is an ideal plane wave field, Represents the propagation function The complex conjugate of represents the square of the propagation function, is the regularization parameter, is the receiving position vector, is the sound source position vector, is the angular frequency, is the propagation function.

[0135] In addition, through iterative optimization, gradually reduce and The error between .

[0136] Step 3.4, output the sound field distribution after compensation;

[0137] The complex curved area is virtually flattened into an equivalent plane to achieve distortion-free reconstruction of the sound field. The calculated compensation wavefront function is applied to the original sound field. By multiplying the two, the corrected sound field distribution is output, achieving virtual mapping of the curved area to the equivalent plane.

[0138] The compensated acoustic field distribution eliminates the distortion in sound wave propagation caused by the wafer edge's curved geometry, rendering the acoustic field characteristics at the edge as if acquired on a flat surface. This provides distortion-free acoustic data for subsequent defect detection. Consequently, after holographic compensation, the imaging resolution of the curved edge is comparable to that of a flat surface, improving defect detection capabilities.

[0139] Step 4: Using the compensated sound field distribution, perform phase recovery and annular reconstruction algorithms to achieve high-quality three-dimensional imaging of the edge area;

[0140] After completing the acoustic holographic compensation, the complete wave field information is extracted from the amplitude measurement data, and combined with the ring acquisition strategy, high-quality 3D acoustic field reconstruction of the wafer edge area is achieved. The specific implementation process includes:

[0141] Step 4.1, extract complete wavefield information;

[0142] The improved Gerchberg-Saxton algorithm is used for phase recovery to extract complete wavefield information from the amplitude measurement data. It should be noted that the improved Gerchberg-Saxton algorithm gradually recovers the phase information of the wavefield by iteratively transforming between the frequency domain and the spatial domain:

[0143] In the specific scenario of wafer ultrasonic field testing, the traditional Gerchberg-Saxton algorithm is improved by adding constraints and convergence strategies suitable for the acoustic field characteristics of the wafer edge. Specifically, the detailed steps are as follows:

[0144] Initialization phase: Input is the amplitude distribution data measured by the ultrasound receiving system First, set the initial phase estimate by physical constraints , either plane wave assumption or initial estimation based on geometric relationships can be used;

[0145] Transform from spatial domain to frequency domain: transform the complex amplitude field of the current iteration step Convert to the frequency domain using the two-dimensional fast Fourier transform algorithm:

[0146] ;

[0147] in represents the Fourier transform operation; For spatial location The amplitude distribution at is the spatial position vector, For the The position in the iteration The phase estimate at is the imaginary unit, represents the complex exponential function; For the The number of waves in the iteration is The frequency domain representation of is the wave number, represents the base of natural logarithms;

[0148] Frequency domain constraint application: Apply frequency domain constraints to maintain the calculated The phase remains unchanged, while the amplitude is replaced by the frequency domain amplitude obtained by system transfer function estimation or prior knowledge :

[0149] ;

[0150] in represents the frequency domain representation after applying the frequency domain constraint, is the frequency domain amplitude, represents the phase angle of the complex number, is the imaginary unit, represents the base of natural logarithms, For the The number of waves in the iteration is Frequency domain representation of ;

[0151] Frequency domain to spatial domain transformation: The constrained frequency domain representation is converted back to the spatial domain through inverse Fourier transform:

[0152] ;

[0153] in represents the inverse Fourier transform operation; It is the complex field converted back to the spatial domain after applying the frequency domain constraint; After applying the frequency domain constraint, The number of waves in the iteration is Frequency domain representation of ;

[0154] Spatial domain constraint application: Apply spatial domain constraints to maintain the calculated The phase remains unchanged, and the amplitude is replaced by the actual measured spatial domain amplitude distribution :

[0155] ;

[0156] in is the complex field after applying spatial domain constraints; For spatial location Amplitude distribution at ; It is the complex field converted back to the spatial domain after applying the frequency domain constraint; is an imaginary unit; represents the phase angle of the complex number, represents the base of natural logarithms;

[0157] Update the phase estimate: Extract a new phase estimate from the updated complex field:

[0158] ;

[0159] in is the updated phase estimate; is the complex field after applying spatial domain constraints; represents the phase angle of the complex number;

[0160] Iteration termination judgment: Calculate the phase change between two adjacent iterations:

[0161] ;

[0162] in is the phase change, represents the L2 norm; is the updated phase estimate; For the The position in the iteration The phase estimate at ;

[0163] like Less than the preset threshold (Usually set to arrive If the maximum number of iterations is reached (usually 50-100 for the wafer edge area), the iteration is terminated and the final recovered phase distribution is output. ; Otherwise, return to the spatial domain to frequency domain transform and continue the iteration.

[0164] To address the special acoustic field characteristics of the wafer edge area, the improved Gerchberg-Saxton algorithm in the embodiments of this application adds two key optimizations:

[0165] Edge enhancement constraint: In the spatial domain constraint step, data points corresponding to the wafer edge area are given higher weights to improve the accuracy of phase recovery in the edge area;

[0166] Adaptive step size control: Introducing step size factor , so that each phase update is:

[0167] ;

[0168] in For the The position in the iteration The phase estimate at For the The position in the iteration The phase estimate at For the The adaptive step size factor for the iteration, Represents the complex field converted back to the spatial domain after applying the frequency domain constraint The phase angle, is the spatial position vector, Represents the difference between the current phase estimate and the target phase.

[0169] Through the above implementation steps, the improved Gerchberg-Saxton algorithm of this application can accurately recover phase information from amplitude measurement data in the wafer edge area, providing complete complex field data for subsequent sound field reconstruction. Experimental verification shows that the improved Gerchberg-Saxton algorithm improves phase recovery accuracy and convergence speed when processing sound field data in the wafer edge area.

[0170] Step 4.2, capturing the wafer edge scattered signal;

[0171] Configure and apply a segmented annular array receiver system to capture wafer edge scattered signals, including the following components and implementation details:

[0172] Physical Structure: The annular array receiving system of this application is composed of multiple high-sensitivity piezoelectric ultrasonic transducer units, which are arranged in an arc shape and segmented around the edge of the wafer to form a complete annular coverage. Specifically, the segmented annular array receiving system includes:

[0173] Multiple arc-shaped sensor array modules, each module contains 64-128 independently addressable piezoelectric ultrasonic transducer elements, operating in the frequency range of 10-100MHz, and the element size is ,in is the wavelength corresponding to the center frequency;

[0174] Multi-channel parallel data acquisition unit, with a sampling rate of no less than 500MS / s (megasamples / second) for each channel and a data quantization accuracy of 12-16 bits, ensuring high-fidelity signal acquisition;

[0175] A precision positioning subsystem, including a three-axis micro-motion stage and a rotation mechanism, with positioning accuracy better than 5μm, is used to precisely position each arc array to the optimal detection position at the edge of the wafer;

[0176] The data integration and synchronization control module ensures that the data acquisition of multiple arc arrays is strictly synchronized in time and space, with the error controlled within 5ns.

[0177] Working Principle: The ring array receiving system operates in the following way:

[0178] During wafer edge inspection, 8-12 arc arrays are evenly distributed around the wafer edge, with each array covering an angle range of 30° to 45°, forming a complete 360° annular coverage.

[0179] Each arc array is adjusted to the optimal detection distance and angle through a precision positioning subsystem, so that each sensor element can receive the maximum scattered signal from the edge area of ​​the wafer;

[0180] The acoustic wave excitation unit (usually integrated with the acoustic topological insulator coupling system) transmits ultrasonic waves to the edge of the wafer, and the annular array receiving system synchronously collects all scattered and transmitted signals;

[0181] The data integration module aligns and merges the signals from different arc arrays according to spatial position and acquisition time to form a complete annular data set.

[0182] Signal processing functions: The system also includes a dedicated signal processing unit that performs the following functions:

[0183] Signal filtering and enhancement: Apply adaptive filters to remove environmental noise and system interference, and improve the signal-to-noise ratio;

[0184] Beamforming: Utilize multi-channel data to achieve dynamic focusing beamforming and improve spatial resolution;

[0185] Scattered signal separation: Use waveform analysis algorithms to separate different components such as direct waves, surface scattered waves, and volume scattered waves to highlight potential defect signals.

[0186] Through the above-mentioned structure and function, the annular array receiving system of the present application can simultaneously capture the scattered signals of the edge area of ​​the wafer from a 360° full angle, and its spatial coverage and detection sensitivity are improved, providing high-quality original data for subsequent polar coordinate sound field reconstruction.

[0187] Step 4.3, applying a dedicated sound field reconstruction algorithm to process the data collected by the annular array;

[0188] Based on the polar coordinate system, a dedicated sound field reconstruction algorithm is applied to process the data collected by the annular array. In addition, the sound field reconstruction algorithm uses a combination of Hankel transform and angular harmonic decomposition to express the sound field as:

[0189] ;

[0190] in is the sound field distribution, is the radial distance, is the polar coordinate angle, is the axial distance, is the spectral domain representation, for Bessel function of order, is the radial wave number, is the axial wave number, is the angular harmonic order, is the imaginary unit, and are the angular propagation phase factor and the axial propagation phase factor, respectively. represents the base of natural logarithms, represents the sum operation, represents the integral operation, is the differential of the radial wave number.

[0191] Step 4.4, integrating phase recovery and ring reconstruction results;

[0192] Output high-resolution three-dimensional acoustic field distribution map of the wafer edge area. Therefore, the three-dimensional acoustic field distribution map clearly shows the tiny defects in the wafer edge area with a resolution of 1 / 100 of the working wavelength. , the detection sensitivity has been improved.

[0193] Step 5: Based on the reconstructed three-dimensional imaging results, an adaptive coupling adjustment system is implemented to automatically adjust the coupling material properties according to the edge reflection signal quality monitored in real time;

[0194] Based on the above steps, a temperature-sensitive acoustic phase change material is used to construct an adaptive coupling layer, which automatically adjusts the material properties according to the quality of the edge reflection signal monitored in real time to ensure the stability of the coupling state and the reliability of the detection results. It should be noted that the specific implementation process includes:

[0195] Step 5.1, constructing an adaptive coupling layer based on phase change materials;

[0196] The acoustic phase change material is selected as the coupling medium, and its acoustic impedance The following relationship is satisfied under temperature changes:

[0197] ;

[0198] in For temperature changes The acoustic impedance value under is the reference impedance value, is the impedance variation coefficient, is the temperature response sensitivity coefficient, is the temperature difference relative to the critical temperature of phase transition, is the hyperbolic tangent function.

[0199] In addition, by controlling the temperature distribution, precise adjustment of the acoustic impedance of the coupling layer can be achieved.

[0200] Step 5.2: Build a real-time monitoring subsystem to continuously evaluate the quality of edge reflection signals;

[0201] Calculate signal quality evaluation indicators based on the signal-to-noise ratio and waveform integrity of the echo signal :

[0202] ;

[0203] in is the signal quality evaluation index, is the signal-to-noise ratio, is the cross-correlation coefficient with the reference waveform, is the spectral entropy (reflecting the uniformity of spectrum distribution), 、 、 are the weight coefficients of signal-to-noise ratio, cross-correlation coefficient and spectral entropy respectively.

[0204] Step 5.3, applying a feedback control algorithm to automatically adjust the temperature distribution of the phase change material according to the signal quality;

[0205] In addition, the control algorithm is based on the gradient descent principle to optimize the temperature distribution to maximize the signal quality index :

[0206] ;

[0207] in Indicates the The temperature distribution of the iteration, Indicates the The temperature distribution of the iteration, is the learning rate, is the gradient of the signal quality index relative to the temperature distribution, is the number of iterations.

[0208] Step 5.4, output the coupling state after adaptive optimization;

[0209] Achieve optimal acoustic matching between the detection system and the wafer edge, and respond in real time to changes in the wafer edge's geometric characteristics and contact status, ensuring that the acoustic energy transmission efficiency is maximized during the detection process and that the signal quality is always maintained at the best state.

[0210] A digital detection system for wafer ultrasonic fields, used to perform the above-mentioned digital detection method for wafer ultrasonic fields, comprising:

[0211] Acoustic topological insulator coupling module, used to construct a scattering-free acoustic wave transmission channel;

[0212] Heterogeneous metasurface acoustic field control module, used to obtain depth dimension information;

[0213] Acoustic holographic compensation processing module, used to eliminate imaging distortion in curved areas;

[0214] Phase recovery and annular reconstruction processing module, used to achieve high-quality three-dimensional imaging of edge areas;

[0215] Adaptive coupling adjustment module to ensure detection stability and reliability.

[0216] Here, the present invention provides an implementation example:

[0217] This embodiment has been verified and applied in an actual semiconductor manufacturing production line. The following is a practical application example based on defect detection in the edge area of ​​12-inch (300mm) silicon wafers.

[0218] This example is applied to wafer edge defect detection in a 14nm process at a semiconductor foundry. The inspection targets 300mm silicon wafers, with a particular focus on defects within 5mm of the edge, including identifying subtle defects in the edge roll-off region. In this scenario, inspection of the wafer edge, particularly the roll-off transition region, is extremely challenging. Traditional ultrasonic inspection methods have low signal-to-noise ratios in this region, resulting in poor imaging quality and a high rate of missed defects, severely impacting product yield and quality.

[0219] After detailed research, it was found that the defects in the wafer edge area in this scenario mainly include three categories:

[0220] Edge cracks (edgecrack), width less than 5μm, length 5-50μm;

[0221] Delamination of the edges, with an area of ​​10-100 μm² and a thickness of less than 1 μm;

[0222] Foreign matter inclusions in the edge area, with a diameter between 3-20μm.

[0223] These three types of defects have serious impacts on subsequent wafer processing and device performance. In particular, microcracks may expand during high-temperature processing and cause wafer breakage.

[0224] For the above application scenarios, a complete wafer ultrasonic field digital detection system is constructed according to this embodiment. The specific implementation details are as follows:

[0225] Acoustic topological insulator coupling system realization:

[0226] An acoustic topological insulator (ATI) was designed based on a "honeycomb rod" structure, consisting of 192 hexagonal resonant cavity units, each 250μm in size and operating at 50MHz. The interface layer uses a gradient acoustic impedance design, with seven layers to achieve a smooth transition from coupler to wafer.

[0227] Coupling coefficient between acoustic units in the system It is set to 0.12, the characteristic size of the resonant cavity is about 75μm, and the width of the connecting channel is about 38μm.

[0228] Experimental measurements show that the acoustic wave transmission efficiency of the acoustic topological insulator coupling system in the wafer edge area is 3.8 times higher than that of traditional planar couplers, and the scattering loss is reduced by 78%.

[0229] A three-layer heterogeneous metasurface structure was constructed, with the unit size controlled at 10μm (about 1 / 6 of the wavelength):

[0230] The first layer is a "gradient aperture cylindrical array" consisting of 512 radially distributed cylindrical units with diameters decreasing from the center to the edge according to the following pattern:

[0231] ;

[0232] in For the The diameter of the unit, For the The normalized distance from each element to the center of the array;

[0233] The second layer is the spectrum separation layer, which uses 12 groups of units with different resonance frequencies;

[0234] The third layer is the phase compensation layer, which contains 128 units with independently controllable phases.

[0235] The entire metasurface system achieves precise control of sound waves in the 10-100MHz frequency band, with a depth resolution of 5μm.

[0236] Acoustic holographic compensation algorithm implementation:

[0237] First, a high-precision laser confocal scanning system is used to obtain accurate three-dimensional morphological data of the wafer edge bending area with a resolution of 0.5μm and a scanning range covering a 10mm area of ​​the edge. Based on the acquired geometric data, the acoustic wave propagation function of this area is calculated using the finite element method, and the grid density is set to The layered iterative holographic projection algorithm sets the upper limit of the number of iterations to 50 times, and the regularization parameter is 0.01, and the convergence threshold is .

[0238] Experimental results show that after compensation, the difference in imaging quality between the curved edge area and the flat area is reduced to less than 10%.

[0239] Improved Gerchberg-Saxton algorithm and ring reconstruction implementation:

[0240] Ten arc sensor arrays are configured, each containing 96 piezoelectric transducer elements, with an operating frequency of 10-100 MHz, a sampling rate of 800 MS / s, and a quantization accuracy of 14 bits. In the edge enhancement constraint, the weight of the data points in the wafer edge area is set to 2.5 times that of the center area, and the adaptive step factor is set to 1.5. according to set up.

[0241] The Bessel function expansion in the polar coordinate system is truncated to The radial wavenumber integration uses the adaptive Gaussian quadrature method.

[0242] The system imaging resolution has been measured to be 1 / 3.2 of the wavelength.

[0243] Adaptive coupling adjustment system implementation:

[0244] A composite material containing phase change polymer (polyvinylidene fluoride, PVDF) is used as the coupling medium, and the parameters of its acoustic impedance change with temperature are: baseline impedance value ; Impedance variation coefficient ; Temperature response sensitivity coefficient .

[0245] The weight coefficient in the signal quality evaluation index is set as (Signal-to-noise ratio), (cross-correlation coefficient), (spectral entropy).

[0246] Learning rate of feedback control algorithm The temperature control accuracy is ±0.1℃.

[0247] Experiments have verified that the adaptive coupling adjustment system can respond to changes in the edge shape of different wafers in real time, and always maintain the acoustic energy transmission efficiency at the optimal state.

[0248] This implementation was tested in a six-month comparison environment in an actual production environment, processing over 5,000 wafer samples and collecting a large amount of test data. The following are the verification results of the two key technologies:

[0249] Comparison of edge area defect detection rates:

[0250] The present embodiment is compared with the traditional ultrasonic detection method at different areas of the wafer edge. The detection results are shown in Table 1:

[0251] Table 1: Comparison of test results at different areas of the wafer edge with traditional ultrasonic testing methods

[0252]

[0253] The data shows that in the most challenging 0-2mm edge region, this implementation achieves a 96.8% defect detection rate, a 2.28-fold improvement over traditional methods. Even in farther-flung edge regions (5-10mm), the detection rate remains improved. Overall, the overall defect missed detection rate in edge regions has been reduced from an average of 22.5% with traditional methods to 1.9% with this implementation, significantly improving detection reliability.

[0254] The imaging quality and detection accuracy of different types of defects are compared and evaluated, as shown in Table 2:

[0255] Table 2: Comparison of imaging quality and detection accuracy for different types of defects

[0256]

[0257] Data shows that this implementation method has achieved improvements in all types of defect detection, particularly significantly increasing the accuracy of detecting tiny defects. For example, for microcracks, detection accuracy increased from 12μm with traditional methods to 3.5μm, with an 8.6dB improvement in signal-to-noise ratio. More importantly, this implementation method achieves detection accuracy comparable to that of flat areas in curved edge regions, overcoming the limitations of traditional methods in edge regions.

[0258] In actual production line applications, the application of this implementation has reduced the subsequent wafer breakage rate due to undetected edge defects from 0.8% to 0.12%, saving the company significant production costs and increasing product yield by approximately 1.5 percentage points. This improvement is particularly important for high-end chip manufacturing processes and has generated significant economic value.

[0259] The above describes an embodiment of the present invention, but this embodiment is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Ordinary technicians in this field can also make more forms of equivalent embodiments based on the inspiration of this embodiment, all of which are protected by this embodiment.

Claims

1. A digital detection method for wafer ultrasonic field, characterized in that: include: Configure an acoustic topological insulator coupling system to build a scattering-free acoustic wave transmission channel; Based on the acoustic topological insulator coupling system, a heterogeneous metasurface acoustic field control system is constructed to obtain depth dimension information; Based on the depth dimension information, the acoustic holographic compensation algorithm is applied to eliminate the imaging distortion of the curved area. The curved area at the edge of the wafer is virtually flattened into an equivalent plane by calculating the compensation wavefront function, and the sound field distribution after compensation is calculated. Using the compensated sound field distribution, phase recovery and annular reconstruction algorithms are performed to achieve high-quality three-dimensional imaging of edge areas. Based on the reconstructed three-dimensional imaging results, an adaptive coupling adjustment system is implemented to automatically adjust the coupling material properties according to the quality of the edge reflection signal monitored in real time. The steps of implementing the adaptive coupling adjustment system include: An adaptive coupling layer based on phase change material is constructed. Acoustic phase change material is selected as the coupling medium. Its acoustic impedance Z satisfies the following under temperature changes: ； in For temperature changes The acoustic impedance value under is the reference impedance value, is the impedance variation coefficient, is the temperature response sensitivity coefficient, is the temperature difference relative to the critical temperature of phase transition, is the hyperbolic tangent function; Build a real-time monitoring subsystem to continuously evaluate the quality of edge reflection signals and calculate signal quality evaluation indicators based on the signal-to-noise ratio and waveform integrity of the echo signal; Apply feedback control algorithms to automatically adjust the temperature distribution of the phase change material based on signal quality; Outputs the adaptively optimized coupling state to achieve the best acoustic match between the detection system and the wafer edge.

2. The digital detection method of wafer ultrasonic field according to claim 1, characterized in that: The step of configuring the acoustic topological insulator coupling system comprises: Construct a periodic array of acoustic units, each unit containing a resonant cavity and connecting channels, to form an acoustic topological insulator with a specific band structure; An interface layer is configured at the boundary of the acoustic topological insulator to generate localized boundary phonon states, which form a unidirectional topologically protected acoustic wave mode by breaking the time reversal symmetry. Precisely align the topologically protected acoustic coupler with the wafer edge area, ensuring maximum coupling efficiency between the two through a precision positioning system; A topologically protected acoustic wave transmission channel is formed. The topologically protected acoustic wave transmission channel has anti-scattering and anti-interference properties, and can maintain high fidelity of acoustic wave transmission in the curved area of ​​the wafer edge.

3. The digital detection method of wafer ultrasonic field according to claim 1, characterized in that: The steps of constructing the heterogeneous metasurface acoustic field control system include: Design and construct a subwavelength unit array structure, where the size of each unit satisfies: ； in is the unit size, is the working wavelength; Apply the gradient phase control algorithm to make the vibration phase between adjacent subwavelength units satisfy: ； in Indicates the The phase of each unit, Indicates the The phase of each unit, is the design wave number, which represents the desired sound wave propagation characteristics, is the unit spacing; Construct a hierarchical structure of heterogeneous metasurfaces and form a three-dimensional sound field control system by stacking metasurface layers with different functions; The output is the vertical sound field distribution information after being regulated by the heterogeneous metasurface. The vertical sound field distribution information includes the acoustic characteristics of different depth layers in the wafer edge area.

4. The digital detection method of wafer ultrasonic field according to claim 1, characterized in that: The step of applying the acoustic holographic compensation algorithm comprises: Collect 3D geometric data of the wafer edge bending area and build an accurate digital model; Calculate the ultrasonic propagation function in the curved area based on the sound wave propagation theory , the ultrasonic propagation function describes the To the receiving location At angular frequency Sound wave propagation characteristics under ; The layered iterative holographic projection algorithm is applied. Based on the principle of solving the inverse problem, the layered iterative holographic projection algorithm calculates the compensation wavefront function based on the difference between the ideal plane wave field and the actual measured wave field. The compensated sound field distribution is calculated and output, and the complex curved area is virtually flattened into an equivalent plane to achieve distortion-free reconstruction of the sound field.

5. The digital detection method of wafer ultrasonic field according to claim 1, characterized in that: The steps of performing the phase recovery and ring reconstruction algorithm include: The improved Gerchberg-Saxton algorithm is applied for phase retrieval, and the complete wavefield information is extracted from the amplitude measurement data through iterative transformation between the frequency domain and the spatial domain. Configure and apply a segmented annular array receiving system to capture wafer edge scattered signals. The segmented annular array receiving system includes multiple arc-shaped sensor array modules, each module containing 64 to 128 independently addressable piezoelectric ultrasonic transducer elements; Based on the polar coordinate system, a dedicated sound field reconstruction algorithm is applied to process the data collected by the annular array. The sound field reconstruction algorithm uses a combination of Hankel transform and angular harmonic decomposition. Integrate phase recovery and annular reconstruction results to output a high-resolution three-dimensional acoustic field distribution map of the wafer edge area.

6. The digital detection method of wafer ultrasonic field according to claim 2, characterized in that: The periodic acoustic unit array is constructed by using two types of structures: a "honeycomb rod" structure and a spiral acoustic topological insulator structure. In the honeycomb rod structure, the hexagonal honeycombs form a resonant cavity, and the connecting rods form a waveguide channel. The topological characteristics of the spiral structure are determined by the number of spiral arms. and helical angle Joint decision.

7. The digital detection method of wafer ultrasonic field according to claim 3, characterized in that: The subwavelength unit array structure includes three types: resonance type, refraction type and composite type. According to the detection requirements, the "gradient aperture cylindrical array" structure is selected for surface and shallow defect detection, and the "resonance cavity spring" composite structure is selected for deep defect detection inside the wafer.

8. The digital detection method of wafer ultrasonic field according to claim 5, characterized in that: The improved GerchbergSaxton algorithm includes two key optimizations: edge enhancement constraint and adaptive step size control. The edge enhancement constraint gives higher weight to the data points in the edge area of ​​the wafer, and the adaptive step size control introduces a step size factor. , so that the phase update satisfies: ； in For the The position in the iteration The phase estimate at For the The position in the iteration The phase estimate at For the The adaptive step size factor for the iteration, Represents the complex field converted back to the spatial domain after applying the frequency domain constraint The phase angle, is the spatial position vector.

9. A digital detection system for wafer ultrasonic field, characterized in that: A digital detection method for a wafer ultrasonic field according to any one of claims 1 to 8, comprising: Acoustic topological insulator coupling module, used to construct a scattering-free acoustic wave transmission channel; Heterogeneous metasurface acoustic field control module, used to obtain depth dimension information; Acoustic holographic compensation processing module, used to eliminate imaging distortion in curved areas; Phase recovery and annular reconstruction processing module, used to achieve high-quality three-dimensional imaging of edge areas; Adaptive coupling adjustment module to ensure detection stability and reliability.

Images