Methods, apparatus, epitaxial wafers and their preparation methods for measuring damage depth

By preparing and pre-processing test samples and using multiple detection methods to obtain the damage index, the accuracy and repeatability issues of silicon wafer damage depth detection in existing technologies have been solved. This has enabled high-precision quantitative measurement of shallow diffuse damage, improving the quality and yield of epitaxial wafers.

CN123124758APending Publication Date: 2026-11-17XIAN ESWIN MATERIAL TECHNOLOGY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611229706.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-13
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing technologies have poor accuracy and repeatability when detecting the depth of damage on silicon wafers. They are particularly insensitive to shallow and diffuse damage and are prone to introducing secondary damage, making it difficult to meet the requirements of high-precision process control.

Method used

By preparing a set of test samples from the same silicon wafer, pre-processing is performed to amplify the difference in physical state between the damaged and undamaged areas. The damage index is obtained using methods such as Raman spectroscopy, photoluminescence detection, and X-ray diffraction, and the damage depth is determined in combination with the material removal depth.

Benefits of technology

It achieves highly sensitive, high-precision, and quantitative measurement of silicon wafer damage depth, improving the repeatability and accuracy of detection, especially for the assessment of shallow, diffuse damage, ensuring the precise removal of the damaged layer in subsequent processes.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123124758A_ABST
    Figure CN123124758A_ABST
Patent Text Reader

Abstract

This disclosure provides a method, apparatus, epitaxial wafer, and preparation method for measuring damage depth. The measurement method includes: acquiring a set of test samples, wherein the test samples originate from the same silicon wafer under test, and each test sample corresponds to a different material removal depth; pre-processing the test samples to amplify the difference in physical state between damaged and undamaged areas within the same test sample under the same detection method; detecting the pre-processed test samples to obtain a damage index for each test sample, wherein the damage index characterizes the degree of damage to the test sample; and determining the damage depth of the silicon wafer under test based on the material removal depth and damage index corresponding to the test sample. This detection method can achieve high-precision, quantitative, and highly repeatable measurement of damage depth.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This disclosure relates to the field of semiconductor manufacturing technology, and more specifically, to a method, apparatus, dielectric, device, epitaxial wafer, and method for preparing the same for measuring damage depth. Background Technology

[0002] In semiconductor manufacturing, silicon wafers, especially those used as substrates for epitaxial growth, typically undergo a series of mechanical processing steps, including slicing, grinding, and polishing. These processes inevitably introduce mechanical damage layers, such as lattice distortions and dislocations, into the surface and subsurface of the silicon wafer. To ensure high-quality crystal layers are obtained during subsequent epitaxial growth, the depth of these damage layers must be accurately assessed to achieve their complete removal.

[0003] Currently, the industry typically employs angle polishing to detect damage depth. However, because this method relies on complex polishing angle control and etching operations, the accuracy and repeatability of the results are easily affected by the operator's skill level and fluctuations in process conditions. More importantly, this destructive testing method itself may introduce secondary interference, and its detection sensitivity and universality are limited for damage layers with different characteristics produced at different processing stages, such as fine grinding and polishing.

[0004] Therefore, there is an urgent need to provide a new measurement method to address the shortcomings of the angle polishing method in terms of detection accuracy, repeatability, and universality. Summary of the Invention

[0005] This disclosure provides a method, apparatus, medium, device, epitaxial wafer, and preparation method for measuring damage depth. The measurement method achieves highly sensitive, high-precision, and quantitative measurement of silicon wafer damage depth by amplifying the physical state differences between damaged and undamaged areas and quantitatively characterizing the damage index. This provides a reliable basis for accurately removing damaged layers to improve the quality of semiconductor substrates such as epitaxial wafers.

[0006] The first aspect of this disclosure provides a method for measuring the damage depth of a silicon wafer. The method includes: acquiring a set of test samples, wherein the test samples originate from the same silicon wafer under test, and each test sample corresponds to a different material removal depth; preprocessing the test samples to amplify the physical state difference between damaged and undamaged areas in the same test sample under the same detection method; detecting the preprocessed test samples to obtain a damage index for each test sample, wherein the damage index characterizes the degree of damage to the test sample; and determining the damage depth of the silicon wafer under test based on the material removal depth and damage index corresponding to the test sample.

[0007] In one specific embodiment of the first aspect of this disclosure, the test sample is preprocessed to amplify the difference in physical state between the damaged and undamaged areas of the same test sample under the same detection method. This includes: selecting a target detection method from preset detection methods; determining preprocessing rules based on the response characteristics of the target detection method, wherein the response characteristics of the target detection method refer to the response law of the detection signal characteristic parameters corresponding to the target detection method and the corresponding physical state change in the damaged area of ​​the silicon wafer under test; and preprocessing the test sample based on the preprocessing rules.

[0008] In one specific embodiment of the first aspect of this disclosure, the pre-treated test samples are tested to obtain a damage index for each test sample, wherein the damage index is used to characterize the degree of damage to the test sample, including: testing the pre-treated test samples based on a target detection method to obtain a detection signal of the test sample; analyzing and processing the detection signal to determine detection signal characteristic parameters, wherein the detection signal characteristic parameters are quantified values ​​used to characterize the corresponding physical state of the damaged area of ​​the test sample; and obtaining the damage index based on the detection signal characteristic parameters.

[0009] In one specific embodiment of the first aspect of this disclosure, the target detection method includes any one of Raman spectroscopy, photoluminescence detection, X-ray diffraction, and minority carrier lifetime detection.

[0010] In one specific embodiment of the first aspect of this disclosure, the target detection method includes Raman spectroscopy detection, and the corresponding detection signal characteristic parameters include Raman characteristic peak shift and Raman characteristic peak half-width at half-maximum.

[0011] In one specific embodiment of the first aspect of this disclosure, obtaining a damage index based on the characteristic parameters of the detection signal includes: obtaining the absolute displacement of the Raman characteristic peak relative to the standard characteristic peak, and the absolute value of the actual half-width at half-maximum (HWHM) relative to the standard HWHM, based on the displacement of the Raman characteristic peak and the HWHM of the Raman characteristic peak; substituting the absolute displacement of the Raman characteristic peak relative to the standard characteristic peak and the absolute value of the actual HWHM relative to the standard HWHM into a preset calculation model to obtain the damage index, wherein the preset calculation model is: DI = ,in, This represents the absolute displacement of the Raman characteristic peak from the standard characteristic peak w. This is the absolute difference between the actual half-width and the standard half-width r. These are the weighting coefficients.

[0012] In one specific embodiment of the first aspect of this disclosure, the damage depth of the silicon wafer under test is determined based on the material removal depth and damage index corresponding to the test sample, including: establishing the relationship between the damage index and the material removal depth; identifying the material removal depth threshold corresponding to the first stable plateau of the damage index according to the relationship; and determining the material removal depth threshold as the damage depth of the silicon wafer under test.

[0013] A second aspect of this disclosure provides a device for measuring the damage depth of a silicon wafer. The device includes an acquisition module, a preprocessing module, a detection module, and a determination module. The acquisition module is configured to acquire a set of test samples, wherein the test samples originate from the same silicon wafer under test, and each test sample corresponds to a different material removal depth. The preprocessing module is configured to preprocess the test samples to amplify the physical state difference between damaged and undamaged areas in the same test sample under the same detection method. The detection module is configured to detect the preprocessed test samples to obtain a damage index for each test sample, wherein the damage index characterizes the degree of damage to the test sample. The determination module is configured to determine the damage depth of the silicon wafer under test based on the material removal depth and damage index corresponding to the test sample.

[0014] A third aspect of this disclosure provides a computing device including a processor and a memory, the processor being configured to execute instructions stored in the memory to implement the measurement method of the first aspect described above.

[0015] The fourth aspect of this disclosure provides a computer storage medium storing at least one instruction, which is executed by a processor to implement the measurement method of the first aspect described above.

[0016] The fifth aspect of this disclosure provides a method for preparing an epitaxial wafer, the method comprising: obtaining a target silicon wafer for which an epitaxial layer is to be grown; obtaining the damage depth of a damage layer on the target silicon wafer, wherein the damage depth is obtained by the measurement method described in the first aspect; removing the damage layer on the target silicon wafer based on the damage depth; and growing an epitaxial layer on the target silicon wafer with the damage layer removed to obtain an epitaxial wafer.

[0017] The sixth aspect of this disclosure provides an epitaxial wafer obtained by the preparation method in the fifth aspect above, wherein the epitaxial wafer does not contain stacking fault defects in the epitaxial layer, and the number of step defects with a lateral feature size greater than or equal to 10 micrometers and a depth greater than or equal to 5 nanometers on the surface of the epitaxial wafer is less than 5. Attached Figure Description

[0018] Figure 1 This is a schematic flowchart illustrating a method for measuring the damage depth of a silicon wafer according to an embodiment of this disclosure.

[0019] Figure 2 This is a partial flowchart illustrating a method for measuring the damage depth of a silicon wafer according to an embodiment of this disclosure.

[0020] Figure 3 This is a schematic flowchart illustrating a method for measuring the damage depth of a silicon wafer according to an embodiment of this disclosure.

[0021] Figure 4 This is a partial flowchart illustrating a method for measuring the damage depth of a silicon wafer according to an embodiment of this disclosure.

[0022] Figure 5 This is a schematic flowchart illustrating a method for measuring the damage depth of a silicon wafer according to an embodiment of this disclosure.

[0023] Figure 6 This is a flowchart illustrating a method for measuring the damage depth of a silicon wafer, provided as another embodiment of this disclosure.

[0024] Figure 7 This is a partial flowchart illustrating a method for measuring the damage depth of a silicon wafer according to an embodiment of this disclosure.

[0025] Figure 8 This is a schematic flowchart illustrating a method for measuring the damage depth of a silicon wafer according to an embodiment of this disclosure.

[0026] Figure 9 This is a partial flowchart illustrating a method for measuring the damage depth of a silicon wafer according to an embodiment of this disclosure.

[0027] Figure 10 This is a schematic flowchart illustrating a method for preparing an epitaxial wafer according to an embodiment of the present disclosure.

[0028] Figure 11 A graph showing the relationship between the material removal depth and the damage index of silicon wafer A provided in an embodiment of this disclosure.

[0029] Figure 12 This is a polishing observation image of the subsurface damage layer of silicon wafer A at an angle.

[0030] Figure 13 A graph showing the relationship between the material removal depth and the damage index of silicon wafer B provided in an embodiment of this disclosure.

[0031] Figure 14 This is a polishing observation image of the subsurface damage layer of silicon wafer B at an angle.

[0032] Figure 15 A graph showing the relationship between the material removal depth and the damage index of a silicon wafer C provided in an embodiment of this disclosure.

[0033] Figure 16A schematic block diagram of a silicon wafer damage depth measuring device provided in another embodiment of this disclosure.

[0034] Figure 17 A schematic block diagram of a computing device provided in an embodiment of this disclosure. Detailed Implementation

[0035] The technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this disclosure, and not all embodiments. Based on the embodiments of this disclosure, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this disclosure.

[0036] In semiconductor manufacturing, accurately determining the damage depth of silicon wafers is a crucial step in ensuring the quality of subsequent epitaxial growth. To achieve quantitative measurement of surface and subsurface damage depth on silicon wafers, a combination of angled polishing and chemical etching is commonly used. Specifically, this method first polishes the silicon wafer sample into a bevel with a predetermined angle, such as 10°. Then, chemical etching exposes defects in the subsurface damage region. Finally, an optical microscope is used to observe and measure the extension length of the defects on the bevel, and the damage depth is calculated based on trigonometric relationships. The basic principle is that by constructing a bevel, damage information distributed along the depth direction is converted into a scale that is easier to observe and measure in the horizontal direction. This method is widely used in related technologies, mainly because it can transform some types of crystal defects into visualized morphologies, thus providing direct morphological evidence for process evaluation.

[0037] However, the angle polishing method, due to the inherent characteristics of its principle and implementation, has many inherent limitations in practical applications. First, in terms of geometric measurement, this method relies on the trigonometric relationship of converting vertical depth into slope length. Any slight deviation in the polishing angle will introduce systematic cumulative errors into this conversion relationship. For example, when the nominal polishing angle is 10°, a deviation of ±0.5° in the actual angle will result in an error of more than 5% in the final calculated damage depth. In addition, if the flatness and straightness of the slope are not ideal, the measurement of the projected length of the defect on the slope will also be inaccurate, further increasing the measurement uncertainty.

[0038] Secondly, in terms of process control, the chemical etching step in this method is crucial for revealing defects, but its process is difficult to reproduce precisely. Even slight fluctuations in the concentration, temperature, and flow state of the etching solution can lead to uneven distribution of the etching rate in both time and space. This non-uniformity causes a shift in the depth reference of the defects exposed on the slope; that is, damage of the same true depth may appear at different slope positions on samples from different batches or even different regions of the same batch. This directly impairs the repeatability and comparability of measurements.

[0039] Furthermore, regarding the sample state, the polishing process in this method is itself a mechanical action. This process may introduce new lattice damage or stress, i.e., secondary damage, onto the newly formed surface or subsurface of the sample while preparing the bevel. This can interfere with the original damage state of the sample, causing the subsequently observed defect morphology to be mixed with artifacts introduced by the preparation process, thereby affecting the accuracy of the assessment of the original processing damage depth.

[0040] Finally, in terms of defect detection, this method relies on the morphological contrast between the etched damaged area and the intact lattice region. For shallow, diffuse damage generated by grinding and polishing processes, it often manifests as lattice distortion, stress concentration, or clusters of point defects, rather than extended dislocations or cracks sufficient to cause significant selective etching. Therefore, such damage may not form sufficient optical contrast after etching, making it difficult to effectively identify and measure under an optical microscope, resulting in insufficient detection sensitivity or even missed detections.

[0041] Therefore, it is evident that when this method is applied to scenarios with more stringent requirements for measurement accuracy, repeatability, and applicability, such as evaluating shallow, diffuse damage introduced by subsequent precision grinding and polishing processes, its limitations become particularly pronounced. Specifically, when detecting shallow damage formed during grinding and polishing, the shallow damage layer and low defect density make it difficult for chemical etching to produce sufficient morphological contrast at the bevel. Simultaneously, minute deviations in the polishing angle, spatial non-uniformity of the etching rate, and human interpretation differences during microscopic observation collectively lead to large dispersion and poor measurement repeatability in the final measured damage depth data. These factors collectively limit the method's ability to provide high-precision and high-reliability quantitative evaluation of silicon wafer damage depth in advanced semiconductor manufacturing processes.

[0042] In view of this, the present disclosure provides a method for measuring the damage depth of a silicon wafer. The method includes: acquiring a set of test samples, wherein the test samples originate from the same silicon wafer under test, and each test sample corresponds to a different material removal depth; pre-processing the test samples to amplify the difference in physical state between damaged and undamaged areas in the same test sample under the same detection method; detecting the pre-processed test samples to obtain a damage index for each test sample, wherein the damage index characterizes the degree of damage to the test sample; and determining the damage depth of the silicon wafer under test based on the material removal depth and damage index corresponding to the test sample. Thus, the measurement method provided by this disclosure, compared to traditional techniques that rely on bevel polishing and chemical etching to reveal defect morphology, fundamentally overcomes the inherent deficiencies of the latter in terms of detection sensitivity, quantification, and operational repeatability by systematically reconstructing the measurement paradigm. Its technical advantages are particularly significant when evaluating shallow, diffuse damage introduced by advanced polishing processes.

[0043] Specifically, the measurement method provided in this disclosure has several improvements that systematically solve the limitations of traditional methods mentioned above, such as the angle polishing method, which result in large dispersion and poor repeatability of measurement results, and cannot meet the requirements of high-precision process control for quantitative and stable measurement.

[0044] First, this measurement method prepares a set of test samples from the same silicon wafer with known material removal depths, abandoning the preparation method of single-point, inclined surface samples. This enables systematic and discrete sampling of the complete cross-section of the damaged layer from the surface to the interior, transforming depth measurement into a lateral comparison of the damage state of samples with different known depths, and completely eliminating the conversion error caused by inclined surface geometry.

[0045] Secondly, a targeted preprocessing step is introduced, the core function of which is to amplify the difference in physical state between the damaged and undamaged areas under subsequent detection methods. This is not achieved through geometric amplification, but rather through physical or chemical means, such as wet oxygen oxidation, to actively modulate the microscopic state of the sample. This selectively introduces more significant stress, defect recombination, or changes in electrical properties in the damaged area, thereby enhancing the signal response intensity and signal-to-noise ratio of subsequent non-destructive or micro-destructive testing techniques, such as Raman spectroscopy. Consequently, the detection method provided in this disclosure can generate clear, distinguishable, and quantifiable detection signals even for very shallow or weak lattice distortions, effectively overcoming the bottleneck of insufficient sensitivity of traditional methods for shallow damage detection.

[0046] Finally, the damage index is obtained through quantitative detection, and its relationship with the material removal depth is analyzed to objectively determine the damage depth. This process transforms the determination of damage depth from a subjective visual interpretation of microscopic morphology to a mathematical analysis of objective quantitative data trends. This eliminates the uncertainty of human observation, achieves objectivity and standardization of the determination criteria, and fundamentally ensures the high repeatability and reliability of the measurement results.

[0047] Therefore, the measurement method provided in this disclosure achieves highly sensitive, precise, quantitative, and repeatable measurements of surface and subsurface damage depths on silicon wafers, particularly shallow, diffuse damage that is difficult to accurately assess using traditional methods, through a complete overhaul of the sample preparation method, the mechanism enhancing the difference in physical states between damaged and undamaged areas, and the logic for result determination. This measurement method provides a reliable basis for the precise removal of damaged layers in subsequent processes and is of significant value for improving the crystal quality and device fabrication yield of high-end semiconductor substrates such as epitaxial wafers.

[0048] The following describes, with reference to the accompanying drawings, a method, apparatus, medium, device, epitaxial wafer and its preparation method for measuring silicon wafer damage depth according to at least one embodiment of the present disclosure.

[0049] The damage depth in the silicon wafer damage depth measurement method mentioned in this disclosure refers to the maximum vertical thickness of the region extending from the silicon wafer surface inwards where the crystal structure integrity is damaged due to the processing. This damage includes, but is not limited to, lattice distortion, dislocation networks, microcracks, and stress concentration areas. This damage depth is used to characterize the minimum surface material thickness that needs to be removed to ensure the quality of subsequent processes (such as epitaxial growth).

[0050] This disclosure provides a method for measuring the damage depth of a silicon wafer, such as... Figure 1 As shown, the measurement method includes the following steps S100 to S400.

[0051] Step S100: Obtain a set of test samples, wherein the test samples are derived from the same silicon wafer to be tested, and each test sample corresponds to a different material removal depth.

[0052] For example, each test sample can be processed using a controllable material removal process to correspond to different material removal depths. For instance, using chemical etching, multiple test samples from the same silicon wafer are immersed in an etching solution with constant composition and temperature. By precisely controlling different etching times and utilizing a calibrated etching rate, a series of samples with preset removal depths can be obtained. For example, using a mixed solution of hydrofluoric acid, nitric acid, and water as the etching solution, the etching rate is determined through pre-experiments, and by setting different immersion times, a group of test samples with removal depths exhibiting an arithmetic progression or a specific gradient distribution can be obtained. This measurement method benefits from the good repeatability of the etching rate under stable conditions, facilitating accurate and consistent depth control. Furthermore, a gradient series of removal depths means that the material removal depth values ​​of each test sample are different from each other, and these depth values ​​cover a numerical range sufficient to characterize the trend of damage layer changes; their distribution can be an arithmetic sequence or other preset rules.

[0053] It should be noted that the material removal methods in this disclosure are not limited to the chemical etching method mentioned in the above examples. Other applicable methods can be selected according to process conditions and equipment availability. For example, controllable mechanical grinding or polishing can be used, and materials of different thicknesses can be removed on different samples by adjusting the grinding pressure, time, or using stepped abrasives; or dry etching processes such as plasma etching or reactive ion etching can be used, and gradient removal of depth can be achieved by controlling the etching time, power, or gas flow rate. The common goal of these methods is to remove material to a preset depth from the surface of each test sample, thereby constructing a depth variable. This disclosure does not limit the specific number, size, etching process parameters, and depth gradient interval of the test samples, which can be designed according to actual needs. For example, the number of test samples can be selected in the range of 5 to 15 pieces, and the depth gradient interval can be set in a preset value, such as 0.5 micrometers to 1.2 micrometers. All of these can be designed according to actual needs and will not be elaborated here.

[0054] Step S200: Pre-process the test sample to amplify the difference in physical state between the damaged and undamaged areas in the same test sample under the same detection method.

[0055] Specifically, step S200 involves preprocessing the test sample obtained in step S100. Its core objective is to selectively enhance the difference in a certain physical property between the damaged and undamaged areas within the same test sample, which is sensitive to subsequent detection methods, before actual testing. This preprocessing is not intended to further remove material to alter the different material removal depths established in step S100, but rather as a signal modulation or contrast enhancement method to actively amplify the microscopic physical state changes caused by damage. This makes the potentially weak damage-related signals significant in subsequent detection, ultimately improving the detection signal-to-noise ratio and the accuracy of damage depth determination.

[0056] Step S300: The pretreated test samples are tested to obtain the damage index of each test sample, wherein the damage index is used to characterize the degree of damage to the test sample.

[0057] For example, the damage index is a dimensionless or dimensionless numerical value that is positively correlated with the severity of residual damage in the test sample. It is a mathematical transformation of the characteristic parameters of the detection signal, with the aim of constructing a single, comparable scalar to concisely characterize the damage state of the sample.

[0058] By performing this step, for each test sample with a material removal depth obtained in step S100, a corresponding damage index characterizing the degree of remaining damage at that depth can be obtained. This series of depth-index datasets provides a precise quantitative basis for the next step of objectively determining the damage depth.

[0059] Step S400: Determine the damage depth of the silicon wafer to be tested based on the material removal depth and damage index corresponding to the test sample.

[0060] Specifically, this step is the final step in completing the entire measurement method process and outputting quantitative results. Its core is to identify the depth threshold corresponding to the boundary between the damaged layer and the underlying intact crystal substrate from the damage index of the system variation.

[0061] In this method, the silicon wafer under test is at least one representative wafer randomly selected from a batch of silicon wafers produced with the same batch and process parameters. The damage depth obtained in step S400 can be used to characterize the average damage state of the batch of silicon wafers. It should be noted that the silicon wafer under test is a sacrificial sample and is no longer used in product production after testing, but its measurement data has key guiding significance for the quality control of the entire batch of silicon wafers.

[0062] For example, when selecting a silicon wafer to be tested, the damage depth value of that wafer measured by this method can be directly used to characterize the average damage state of the corresponding batch of silicon wafers. For instance, a dual control approach can be adopted, ensuring representativeness before wafer selection and verifying consistency after selection, effectively avoiding batch misjudgment caused by anomalies in a single wafer. Specifically, representative screening sampling is performed, and by excluding visible defects and fixing the sampling location, the probability of sampling known atypical individuals is reduced. In this way, even if only one wafer is still tested, the reliability of that wafer as a statistically valid representative of the batch is greatly improved. At the same time, a review is performed, using historical data or process knowledge to set reasonableness criteria for the single-wafer test results. For example, after obtaining the damage depth value, it is compared with the historical baseline. If it is an outlier, at least two more wafers are retested, and the maximum damage depth of the retest is used as the batch characterization value.

[0063] For example, when selecting multiple silicon wafers, such as 2 to 5 wafers to be tested, the test method can be performed on each wafer to obtain multiple damage depth values. The calculation results based on the multiple damage depth values, such as the arithmetic mean, median, or range, can then be used as the damage depth reference value for the batch of silicon wafers to further improve statistical reliability.

[0064] Regardless of whether a single-wafer or multi-wafer approach is used, the damage depth data obtained is used to guide the setting of removal amounts in subsequent processes, such as chemical mechanical polishing, thereby improving the processing accuracy and yield of the entire batch of silicon wafers.

[0065] This disclosure does not limit the specific selection method for the representative silicon wafer in the batch, i.e., the silicon wafer to be tested. Selection can be made based on the principle that the selected silicon wafer can reasonably reflect the average or typical level of damage status of silicon wafers in the same batch, avoiding data deviation caused by a single extreme selection. Those skilled in the art can select the corresponding sampling strategy based on factors such as the process stability of the actual growth line, quality control requirements, and testing costs. For example, under stable process conditions, a silicon wafer can be randomly selected as the sample to be tested; when it is necessary to evaluate the range of process fluctuations, several silicon wafers at specific intervals in the processing sequence, such as the first, middle, and last, can be selected. All of these can be designed according to actual needs and will not be elaborated upon here.

[0066] Step S100 aims to prepare a baseline sample for establishing a quantitative correlation between damage signals and material removal depth. Its core is ensuring that all test samples have a consistent initial damage state, while systematically introducing material removal depth as the sole variable. To this end, a set of test samples are all derived from the same silicon wafer under test, ensuring they have the same crystallographic orientation, initial doping concentration, and damage substrate introduced by the previous processing. By performing controlled-depth material removal on each test sample, different samples have different material removal depths, thus forming a gradient depth series covering a range from zero surface removal to exceeding the estimated damage depth.

[0067] The preprocessing operation in step S200 is functionally independent of, yet sequentially coordinated with, the test samples with different material removal depths in step S100. Step S100 establishes the system variable of different material removal depths, while step S200, based on this, uniformly applies processing conditions to each test sample with an existing material removal depth to enhance its internal signal contrast. Together, they constitute a complete sample pre-preparation process oriented towards high-sensitivity detection, laying a crucial foundation for the subsequent accurate acquisition of quantitative indicators characterizing the degree of damage.

[0068] Step S300 involves inspecting the test samples pretreated in step S200 to obtain a damage index for each sample. This step is the core of quantitative measurement. Its purpose is to detect test samples where the physical state difference between damaged and undamaged areas is amplified. The detection signal corresponding to the specific damaged area has been enhanced. Through detection, a numerical index that can comprehensively quantify the current damage state of the sample, namely the damage index, is extracted or calculated. This index will serve as key data for subsequent correlation analysis with known material removal depth to determine the damage boundary.

[0069] The reliable execution of step S400 relies entirely on the accurate and highly discriminative damage index data provided in step S300. The differences in physical state between the test samples with different material removal depths in step S100 and the damaged and undamaged areas in step S200 collectively ensure the significance and resolvability of the correlation between the damage index data and the corresponding material removal depth. Thus, steps S100 to S400 constitute a complete logical closed loop, from the introduction of a variable system, signal selective enhancement, quantitative characterization of the damage state, to the objective determination of the depth threshold. This closed-loop method, through the coordinated design and control of each link in the measurement chain, ultimately achieves high-precision, quantitative, and highly repeatable measurement of the subsurface damage depth of silicon wafers, especially shallow, diffuse damage that is difficult to measure accurately using traditional methods.

[0070] In at least one embodiment of this disclosure, such as Figure 2As shown, step S200 preprocesses the test sample to amplify the difference in physical state between the damaged and undamaged areas in the same test sample under the same detection method, including steps S210 to S230.

[0071] Step S210: Select a target detection method from the preset detection methods. Specifically, target detection methods refer to any physical or chemical detection means that can respond to microstructural defects, stress, or disorder in crystalline materials and output analyzable signals, thereby providing a basis for inspecting the depth of damage.

[0072] In at least one embodiment of this disclosure, the target detection method includes any one of Raman spectroscopy, photoluminescence detection, X-ray diffraction, and minority carrier lifetime detection.

[0073] In at least one embodiment of this disclosure, the target detection method includes Raman spectroscopy detection, and the corresponding detection signal characteristic parameters include Raman characteristic peak shift and Raman characteristic peak full width at half maximum (FWHM). The peak shift mainly reflects the stress state of the lattice, while the FWHM reflects the degree of dispersion of lattice vibration modes and is related to the lattice order (defect density). Combining the two can provide a more comprehensive assessment of damage.

[0074] For example, step S210 involves selecting a target detection method from a set of preset detection methods. This step forms the basis for determining subsequent preprocessing rules and specific detection schemes. The preset detection methods include, but are not limited to, Raman spectroscopy, photoluminescence spectroscopy, X-ray diffraction, and minority carrier lifetime testing, all of which can specifically respond to structural defects, stress states, or carrier behavior in silicon crystals. Based on the characteristics of the silicon wafer under test, the type of damage, and the required detection accuracy, one method is selected as the target detection method for this measurement. This selection allows for a clearly targeted design of subsequent preprocessing processes, aiming to maximize the amplification of physical state differences detectable by the specific detection method.

[0075] Step S220: Determine preprocessing rules based on the response characteristics of the target detection method, wherein the response characteristics of the target detection method refer to the response law of the detection signal characteristic parameters corresponding to the target detection method and the physical state change in the damaged area of ​​the silicon wafer under test.

[0076] Specifically, detection signal characteristic parameters refer to any quantitative characteristic value extracted from the raw signal data obtained by detecting the test sample, which can reflect the physical state or material properties of a specific region of the sample. For example, these may include, but are not limited to: peak position, wavelength or wavenumber, peak intensity, and full width at half maximum (FWHM) of a specific characteristic peak in spectroscopic detection; angular position, FWHM, and integrated intensity of a specific diffraction peak in diffraction detection; and minority carrier lifetime and photoluminescence intensity in electrical or optical detection. Specifically, in Raman spectroscopy, detection signal characteristic parameters include the characteristic peak corresponding to the silicon crystal (located at approximately 520 cm⁻¹). -1 Peak displacement values ​​(in cm) near -1 (in units) and the full width at half maximum (FWHM) of the characteristic peak (in cm) -1 (unit)

[0077] Response characteristics refer to the correlation between the characteristic parameters of the detection signal corresponding to the target detection method, such as Raman peak shift, photoluminescence intensity, and diffraction peak half width at half maximum, and the specific physical state changes in the damaged area of ​​the silicon wafer, such as increased lattice stress, increased defect density, and increased non-radiative recombination centers.

[0078] Preprocessing rules refer to a set of pre-defined process conditions, sequence of steps, or parameter sets based on the response characteristics of the selected target detection method, designed to amplify the detectable physical state differences between damaged and undamaged regions. The core function of these rules is to guide how to implement active physical or chemical modulation on the test sample, thereby selectively enhancing the physical state differences between damaged and undamaged regions—that is, selectively enhancing the specific signal characteristics of the damaged region that will be quantified in subsequent detection—while maintaining the material removal depth.

[0079] For example, if the target detection method is Raman spectroscopy, which is sensitive to lattice stress, the pretreatment rule can be designed as a thermal oxidation process that introduces additional stress; if the target detection method is photoluminescence, which is sensitive to carrier recombination behavior, or minority carrier lifetime detection, the pretreatment rule can be designed as surface passivation or defect chemical decoration.

[0080] S230: Preprocess the test samples based on preprocessing rules.

[0081] Step S230 then preprocesses the test samples with different material removal depths obtained in step S100 according to the preprocessing rules determined in step S220. The function of this preprocessing is to perform predetermined physical or chemical modulation, rather than to further change the material removal depth of the sample.

[0082] For example, when the target detection method is Raman spectroscopy, the preprocessing rule can be defined as performing a thermal oxidation treatment for a preset time, such as 30 minutes to 180 minutes, within a specific temperature range, such as 800°C to 1150°C, and under a specific atmosphere, such as humid oxygen or dry oxygen. Subsequently, the generated oxide layer is completely removed using an etchant with specific components, such as a buffered oxide etchant. After this treatment, the lattice stress difference between the damaged and undamaged areas within the same test sample is significantly amplified, thereby greatly improving the contrast of the shift signals of characteristic peaks in the subsequent Raman spectrum, effectively enhancing the detection sensitivity and signal-to-noise ratio.

[0083] For example, when the target detection method is minority carrier lifetime detection, the pretreatment rule can be defined as chemical passivation of the sample surface under specific conditions to differentially modulate the carrier recombination rate between the damaged region and the bulk region. Specific process parameters can be designed according to actual needs and will not be elaborated here.

[0084] Therefore, steps S210, S220, and S230 constitute a coherent process from selecting the detection method to designing pretreatment rules and implementing the process. This process ensures a high degree of synergy between the pretreatment operation and the final detection method. Its fundamental purpose is to create optimal conditions for subsequent accurate quantification of damage degree by enhancing the detection signal of the targeted damaged area while keeping the removal depth of different materials corresponding to the test sample constant.

[0085] It should be noted that the process parameters included in the pretreatment rules, such as temperature, time, gas flow rate, and solution concentration, can be specifically adjusted according to the material characteristics and the required signal amplification. The design aims to significantly improve the difference between the detection signal characteristic parameters (such as Raman peak shift) corresponding to the damaged area and the corresponding parameters in the undamaged area of ​​the pretreated test sample compared to before treatment, for example, by more than 50%, thereby effectively improving the signal-to-noise ratio and sensitivity of subsequent detection. These can be designed according to actual needs and will not be elaborated upon here.

[0086] At least one embodiment of this disclosure provides another method for measuring the damage depth of a silicon wafer, such as Figure 3 As shown, the measurement method includes the following steps S100a to S400a.

[0087] Step S100a: Obtain a set of test samples, wherein the test samples are derived from the same silicon wafer to be tested, and each test sample corresponds to a different material removal depth.

[0088] Step S210a: Select a target detection method from the preset detection methods. Step S220a: Determine preprocessing rules based on the response characteristics of the target detection method, wherein the response characteristics of the target detection method refer to the response law of the detection signal characteristic parameters corresponding to the target detection method and the physical state change corresponding to the damaged area of ​​the silicon wafer under test.

[0089] Step S230a: Preprocess the test sample based on the preprocessing rules.

[0090] Step S300a: The pretreated test samples are tested to obtain the damage index of each test sample, wherein the damage index is used to characterize the degree of damage to the test sample.

[0091] Step S400a: Determine the damage depth of the silicon wafer to be tested based on the material removal depth and damage index corresponding to the test sample.

[0092] It should be noted that the specific implementation methods of steps S100a to S400a above can be referred to the above embodiments, and will not be repeated here.

[0093] In at least one embodiment of this disclosure, such as Figure 4 As shown, step S300 involves testing the pretreated test samples to obtain a damage index for each test sample. The damage index is used to characterize the degree of damage to the test samples, and includes steps S310 to S330.

[0094] Step S310: Detect the pre-processed test sample based on the target detection method to obtain the detection signal of the test sample.

[0095] For example, when the target detection method is Raman spectroscopy, a laser Raman spectrometer is used, employing a laser of a specific wavelength, such as 488 nm, as the excitation source, and scanning the sample surface at a preset power, such as 5 mW, to obtain a Raman scattering spectrum containing silicon characteristic peaks as the raw detection signal. Alternatively, when the target detection method is photoluminescence spectroscopy, a photoluminescence spectrometer is used to collect the fluorescence spectrum emitted by the sample under a specific excitation light.

[0096] Step S320: Analyze and process the detection signal to determine the characteristic parameters of the detection signal, wherein the characteristic parameters of the detection signal are quantified values ​​used to characterize the corresponding physical state of the damaged area of ​​the test sample.

[0097] For example, specific implementation methods for analyzing and processing detection signals include, but are not limited to, signal denoising, peak position fitting, intensity integration, or lifetime calculation. For instance, for Raman spectroscopy signals, fitting analysis can determine the exact peak position and full width at half maximum (FWHM) of silicon characteristic peaks; for photoluminescence spectroscopy, the intensity or FWHM of specific emission peaks can be extracted; and for minority carrier lifetime detection, the carrier lifetime value is directly obtained. These characteristic parameters of the detection signals provide direct input data for subsequent calculation of a comprehensive damage index.

[0098] Step S330: Obtain the damage index based on the characteristic parameters of the detection signal.

[0099] Specifically, step S310, based on the target detection method selected in the preprocessing stage, performs actual detection on the test sample processed in step S200 to obtain the original detection signal. The purpose of this step is to obtain raw data that reflects the internal physical state of the sample in a non-destructive or minimally destructive manner. Depending on the target detection method, appropriate detection equipment and parameter settings are used to acquire the signal.

[0100] Step S320 analyzes and processes the detection signal obtained in step S310 to extract one or more quantized detection signal feature parameters. For a detailed description of the detection signal feature parameters, please refer to the above embodiments, which will not be repeated here.

[0101] Step S330, based on one or more detection signal characteristic parameters determined in step S320, obtains a single quantitative index, namely the damage index, to characterize the overall damage level of the test sample through a predetermined algorithm or calculation model. This index aims to integrate characteristic parameters reflecting different physical aspects into a comparable scalar.

[0102] Furthermore, the specific form of the predetermined algorithm or calculation model, the types of feature parameters used, and their weighting coefficients can be adjusted according to the characteristics of the damage mechanism and the detection method. The fundamental purpose is to ensure that the calculated damage index maintains a monotonic and stable positive correlation with the actual damage degree of the sample.

[0103] At least one embodiment of this disclosure provides another method for measuring the damage depth of a silicon wafer, such as Figure 5 As shown, the measurement method includes the following steps S100b to S400b.

[0104] Step S100b: Obtain a set of test samples, wherein the test samples are derived from the same silicon wafer to be tested, and each test sample corresponds to a different material removal depth.

[0105] Step S200b: Pre-process the test sample to amplify the difference in physical state between the damaged and undamaged areas in the same test sample under the same detection method.

[0106] Step S310b: Detect the pre-processed test sample based on the target detection method to obtain the detection signal of the test sample.

[0107] Step S320b: Analyze and process the detection signal to determine the characteristic parameters of the detection signal, wherein the characteristic parameters of the detection signal are quantified values ​​used to characterize the corresponding physical state of the damaged area of ​​the test sample.

[0108] Step S330b: Obtain the damage index based on the characteristic parameters of the detection signal.

[0109] Step S400b: Determine the damage depth of the silicon wafer to be tested based on the material removal depth and damage index corresponding to the test sample.

[0110] It should be noted that the specific implementation methods of steps S100b to S400b above can be referred to the above embodiments, and will not be repeated here.

[0111] At least one embodiment of this disclosure provides another method for measuring the damage depth of a silicon wafer, such as Figure 6 As shown, the measurement method includes the following steps S100c to S400c.

[0112] Step S100c: Obtain a set of test samples, wherein the test samples are derived from the same silicon wafer to be tested, and each test sample corresponds to a different material removal depth.

[0113] Step S210c: Select a target detection method from the preset detection methods. Step S220c: Determine preprocessing rules based on the response characteristics of the target detection method, wherein the response characteristics of the target detection method refer to the response law of the detection signal characteristic parameters corresponding to the target detection method and the physical state change corresponding to the damaged area of ​​the silicon wafer under test.

[0114] Step S230c: Preprocess the test sample based on the preprocessing rules.

[0115] Step S310c: Detect the pre-processed test sample based on the target detection method to obtain the detection signal of the test sample.

[0116] Step S320c: Analyze and process the detection signal to determine the characteristic parameters of the detection signal, wherein the characteristic parameters of the detection signal are quantified values ​​used to characterize the corresponding physical state of the damaged area of ​​the test sample.

[0117] Step S330c: Obtain the damage index based on the characteristic parameters of the detection signal.

[0118] Step S400c: Determine the damage depth of the silicon wafer to be tested based on the material removal depth and damage index corresponding to the test sample.

[0119] It should be noted that the specific implementation methods of steps S100c to S400c can be referred to the above embodiments, and will not be repeated here.

[0120] In at least one embodiment of this disclosure, the target detection method includes Raman spectroscopy detection, and the corresponding detection signal characteristic parameters include Raman characteristic peak shift and Raman characteristic peak half width at half maximum (FWHM).

[0121] Based on the above embodiments, in at least one embodiment of this disclosure, such as Figure 7 As shown, step S330 obtains the damage index based on the characteristic parameters of the detection signal, including the following steps S331 to S332.

[0122] Step S331: Based on the Raman characteristic peak displacement and the Raman characteristic peak half-width at half-maximum (FWHM), obtain the absolute displacement of the Raman characteristic peak and the standard characteristic peak, as well as the absolute value of the actual FWHM and the standard FWHM.

[0123] Specifically, the standard characteristic peak position corresponds to the bulk characteristic peak of intact single-crystal silicon, typically located at approximately 520 cm⁻¹, and the standard full width at half maximum (FWHM) corresponds to the reference FWHM of an intact crystal under the same testing conditions. The purpose of calculating these absolute values ​​is to eliminate the sign effect caused by stress direction, such as compressive or tensile stress, thereby focusing on the deviation of the physical state from the intact crystal benchmark, providing purely quantitative input parameters for subsequent comprehensive assessment of the degree of damage.

[0124] Step S332: Substitute the absolute displacements of the Raman characteristic peak and the standard characteristic peak, and the absolute values ​​of the actual half-width at half-maximum (HWHM) and the standard HWHM into the preset calculation model to obtain the damage index. The preset calculation model is: DI = ,in, This represents the absolute displacement of the Raman characteristic peak from the standard characteristic peak w. This is the absolute difference between the actual half-width and the standard half-width r. is the weighting coefficient, and e is a natural number, approximately 2.718.

[0125] Since mechanical damage is often dominated by lattice distortion and stress accumulation, resulting in a relatively low defect density, in at least one embodiment of this disclosure, the weighting coefficient is 0.3. This indicates that in this calculation model, the contribution weight of peak displacement variation to the damage index is higher than that of half-width variation.

[0126] Thus, the damage index DI calculated by this preset calculation model can simultaneously respond to changes in lattice stress and order, and its value is positively correlated with the severity of residual damage in the test sample. The damage index obtained thus provides accurate and comparable quantitative data points for subsequently establishing its correlation curve with the material removal depth and accurately determining the damage depth.

[0127] At least one embodiment of this disclosure provides another method for measuring the damage depth of a silicon wafer, such as Figure 8 As shown, the measurement method includes the following steps S100d to S400d.

[0128] Step S100d: Obtain a set of test samples, wherein the test samples are derived from the same silicon wafer to be tested, and each test sample corresponds to a different material removal depth.

[0129] Step S210d: Select the target detection method from the preset detection methods, wherein the target detection method is Raman spectroscopy detection method.

[0130] Step S220d: Determine the preprocessing rules based on the response characteristics of Raman spectroscopy detection, where the response characteristics of Raman spectroscopy detection refer to the response law of the physical state change in the damaged area of ​​the silicon wafer under test, corresponding to the characteristic parameters of the detection signal of Raman spectroscopy detection.

[0131] Step S230d: Preprocess the test sample based on the preprocessing rules.

[0132] Specifically, the pretreatment of the test sample in this step can be achieved by growing a layer of silicon oxide on the silicon surface through high-temperature wet oxygen oxidation. The principle is that the molar volume of silicon oxide is greater than that of silicon, and during oxidation, the oxide layer exerts compressive stress on the underlying silicon lattice. In the damaged region, due to the pre-existing distortions and defects in the lattice, the oxidation process is more likely to occur or become uneven, leading to more significant local stress concentration, thus amplifying the lattice stress difference between the damaged and undamaged regions. This stress difference directly corresponds to the peak shift change of the silicon characteristic peak in the Raman spectrum.

[0133] The pretreatment also includes removing the oxide layer after oxidation to restore the silicon surface for Raman testing.

[0134] Step S310d: Detect the pretreated test sample based on Raman spectroscopy to obtain the detection signal of the test sample, wherein the detection signal includes at least the shift of Raman characteristic peaks and the full width at half maximum (FWHM) of Raman characteristic peaks.

[0135] Step S320d: Analyze and process the detection signal to determine the characteristic parameters of the detection signal, wherein the characteristic parameters of the detection signal are quantified values ​​used to characterize the corresponding physical state of the damaged area of ​​the test sample.

[0136] Step S331d: Based on the Raman characteristic peak displacement and the Raman characteristic peak half-width at half-maximum (FWHM), obtain the absolute displacement of the Raman characteristic peak and the standard characteristic peak, as well as the absolute value of the actual FWHM and the standard FWHM.

[0137] Step S332d: Substitute the absolute displacements of the Raman characteristic peak and the standard characteristic peak, and the absolute values ​​of the actual half-width at half-maximum (HWHM) and the standard HWHM into the preset calculation model to obtain the damage index. The preset calculation model is: DI = ,in, This represents the absolute displacement of the Raman characteristic peak from the standard characteristic peak w. This is the absolute difference between the actual half-width and the standard half-width r. These are the weighting coefficients.

[0138] Step S400d: Determine the damage depth of the silicon wafer to be tested based on the material removal depth and damage index corresponding to the test sample.

[0139] In at least one embodiment of this disclosure, such as Figure 9 As shown, step S400 determines the damage depth of the silicon wafer to be tested based on the material removal depth and damage index corresponding to the test sample, including the following steps S410 to S332.

[0140] Step S410: Establish the relationship between the damage index and the material removal depth.

[0141] For example, establishing the relationship between the damage index and the material removal depth can be achieved by plotting data points with the material removal depth on the x-axis and the damage index on the y-axis, and then fitting a trend line or curve to form a DI-d relationship curve that intuitively reflects the evolution of the damage state with depth. This step transforms discrete measurement data into a continuous functional relationship or graphical trend, providing an analytical basis for the subsequent objective determination of the damage boundary.

[0142] Step S420: Based on the change relationship, identify the material removal depth threshold corresponding to the first entry of the damage index into the stable platform.

[0143] For example, by analyzing the DI-d curve, the inflection point from the continuous decline of the damage index to the entry into a stable plateau can be identified. The depth value corresponding to this inflection point is defined as the material removal depth threshold. This identification can be accomplished through a combination of graphical observation and numerical calculation, such as calculating the rate of change or moving average of the damage index between adjacent data points. When the rate of change is consistently lower than a preset threshold, such as 0.1 / micrometer, it is determined that the material has entered a stable plateau.

[0144] Step S430: Determine the material removal depth threshold as the damage depth of the silicon wafer under test.

[0145] Specifically, step S410 aims to establish a quantitative relationship between the damage index and the material removal depth. Specifically, the known material removal depth corresponding to each test sample prepared in step S100 is used as the independent variable, and the corresponding damage index obtained through detection and calculation in step S300 is used as the dependent variable. A data set is then constructed and correlation analysis is performed.

[0146] Step S420, based on the relationship established in step S410, identifies the material removal depth threshold corresponding to when the damage index first enters and remains within a stable fluctuation range. The underlying principle is that when the material removal depth has not yet reached the bottom of the damage layer, each increase in removal depth causes significant changes in the residual lattice defects or stress state in the sample, leading to a continuous decrease in the damage index. When the removal depth reaches or completely penetrates the damage layer, the subsequently removed material is an intrinsically intact crystal, and its damage index will no longer exhibit a trend change with increasing depth. Instead, it will fluctuate randomly within a small range around a low background value, thus presenting a "stable plateau" on the curve.

[0147] Step S430 finalizes the material removal depth threshold identified in step S420 as the damage depth of the silicon wafer under test. This threshold objectively characterizes the vertical distance from the surface to the bottom of the damaged layer, i.e., the extension boundary of the subsurface defects introduced by processing in the depth direction.

[0148] Thus, by executing the above three sub-steps in sequence, the measurement method provided in this disclosure realizes the entire process of objective quantitative analysis from data association modeling, feature inflection point identification to final depth value determination, effectively avoiding the errors introduced by relying on subjective morphological interpretation in traditional methods, thereby ensuring high accuracy and high repeatability of measurement results.

[0149] This disclosure also provides a method for preparing an epitaxial wafer, such as... Figure 10 As shown, the preparation method includes the following steps S10 to S40.

[0150] Step S10: Obtain the target silicon wafer for epitaxial layer growth.

[0151] Specifically, the target silicon wafer refers to the substrate silicon wafer obtained from a single-crystal silicon ingot through mechanical processing such as wire cutting. The damage layer of the target silicon wafer mainly originates from the wire cutting process, and also includes mechanical damage that may remain or be newly introduced in subsequent processes after wire cutting. This method obtains the cumulative damage depth of the target silicon wafer before the formation of the epitaxial layer through actual measurement, and uses this as the thickness target for subsequent precise removal of the damage layer.

[0152] Step S20: Obtain the damage depth of the damage layer of the target silicon wafer, wherein the damage depth is obtained by the measurement method in the above embodiment.

[0153] Specifically, the damage depth of the target silicon wafer in this step is not detected in real time on the target silicon wafer itself, but is obtained in advance using the measurement method described in the above embodiments before the epitaxial layer preparation is performed. This measurement method uses one or more representative silicon wafers produced in the same batch and under the same process conditions as the target silicon wafers as sacrificial samples, and quantitatively determines the extension depth of the damage layer using the measurement method described in the above embodiments. Therefore, before any damage layer removal processing is performed on the batch of target silicon wafers, its damage depth value has been obtained as a clear process input parameter, thus providing a quantitative target for subsequent precise removal processing. The specific measurement method can be referred to in the above embodiments and will not be repeated here.

[0154] Step S30: Remove the damaged layer of the target silicon wafer based on the damage depth.

[0155] Specifically, based on the damage depth measured in step S20, the target silicon wafer is subjected to material removal to a corresponding depth to completely remove its damage layer.

[0156] In at least one of the present disclosures, the material removal process corresponding to the removal of the damaged layer may include chemical mechanical polishing, chemical etching, or a combination of both, with the removal depth precisely controlled to be equal to or slightly greater than the measured damage depth.

[0157] Step S40: An epitaxial layer is grown on the target silicon wafer after the damaged layer has been removed to obtain an epitaxial wafer.

[0158] Specifically, in step 40, an epitaxial layer is grown on the clean silicon wafer surface where the damaged layer has been removed, thereby obtaining the target epitaxial wafer.

[0159] Thus, the above epitaxial wafer fabrication method not only ensures the complete elimination of damage structures that lead to epitaxial defects, providing a substrate surface with a complete crystal lattice structure for epitaxial growth, but also minimizes the excessive removal of intact bulk silicon material, achieving efficient material utilization and precise control of substrate thickness. This fabrication method, by combining precise damage depth measurement with customized removal processes, fundamentally guarantees the excellent crystal quality and surface integrity of the prepared epitaxial wafers.

[0160] This disclosure also provides an epitaxial wafer obtained by the preparation method described in the above embodiments. This epitaxial wafer exhibits high crystal quality. Specifically, because the damaged layer of the substrate is completely removed based on precisely measured depth, the epitaxial wafer does not contain stacking fault defects originating from substrate damage. Furthermore, the number of step-like defects with a lateral feature size greater than or equal to 10 micrometers and a depth greater than or equal to 5 nanometers on the surface of the epitaxial wafer is less than 5. This quantitative indicator further confirms that the substrate prepared based on the method of this invention can support high-quality epitaxial growth.

[0161] For example, when the surface of the epitaxial wafer is inspected using a high-resolution differential interferometry technique, such as 5 nm resolution, the number of stepped defects with a lateral feature size greater than or equal to 10 micrometers and a depth greater than or equal to 5 nanometers observed is at least one of 4, 3, 2, 1, or 0.

[0162] It should be noted that the epitaxial wafers provided in the embodiments of this disclosure are not limited to the examples above. The specific quality indicators of the epitaxial wafers, such as dislocation density and surface roughness, may vary depending on the device requirements. However, the epitaxial wafers obtained by the preparation method of this disclosure ensure the complete removal of substrate damage, which is the common basis for improving various crystal quality indicators.

[0163] To verify the effectiveness, accuracy, and advantages of the silicon wafer damage depth measurement method provided in this disclosure embodiment compared to related technologies, a comparative verification experiment was conducted in conjunction with the actual semiconductor silicon wafer manufacturing process. The experiment used three 12-inch silicon wafers with different processing degrees and representative damage depths as test objects, labeled as wafer A, wafer B, and wafer C, respectively. For each wafer, the damage depth was simultaneously measured using the method of this invention (experimental group) and the traditional angle polishing method (control group) to compare and analyze the measurement results, operational characteristics, and applicable scope of the two methods. The characteristics and rationality of the core process parameters in the implementation of the method of this invention are explained below to provide a basis for subsequent effectiveness verification.

[0164] First, obtaining a set of test samples includes the following steps: obtaining multiple test samples of a preset size from the 12-inch silicon wafer to be tested, for example, by laser cutting. Specifically, the test samples of the preset size are 8 square samples with a size of 4 cm × 4 cm. These test samples are cleaned to remove surface contaminants. Subsequently, using a predetermined ratio of acidic etching solution, such as HF:HNO3:H2O = 1:3:6 (volume ratio), at a calibrated etching rate, a set of corresponding material removal depths were obtained by setting different etching times for each test sample. For example, with an etching rate pre-calibrated to 0.5±0.05μm / min, a gradient series of removal depths of 0μm, 1μm, 2μm, 3μm, 4μm, 5μm, 6μm, and 7μm were obtained. After etching, the sample surface was immediately rinsed with ultrapure water to remove residual etching solution and dried with nitrogen.

[0165] Then, the test sample is pretreated to amplify the difference in physical state between the damaged and undamaged areas in the same test sample under the same detection method. The specific steps are as follows: First, a wet oxygen treatment is performed. The test samples, after gradient etching, are placed sequentially into a quartz boat and pushed into the constant temperature zone of a tube furnace. Pure oxygen is first introduced, for example, at a flow rate of 2L / min, for a first preset time of, for example, 5 minutes, to remove impurities from the furnace tube. Then, the temperature is raised to a preset temperature of 1050℃, and the atmosphere is switched to a wet oxygen atmosphere, for example, O2, at a flow rate of 1L / min + 95℃ water vapor at a flow rate of 1L / min. The oxidation is maintained at this temperature for a second preset time of, for example, 90 minutes. After the oxidation is completed, the water vapor source is turned off, and pure oxygen is introduced again for a third preset time of, for example, 5 minutes, to replace the residual water vapor in the furnace tube.

[0166] Then, the oxide layer is removed: prepare BOE etching solution, for example, with a volume ratio of 49% HF aqueous solution: 40% NH4F aqueous solution = 1:6. Immerse the sample after wet oxygen treatment in BOE etching solution and soak it at room temperature for a fourth preset time, for example, 3 minutes, to remove the silicon dioxide oxide layer generated on the surface. After etching, rinse the sample surface with ultrapure water until the pH value is neutral and blow it dry with nitrogen.

[0167] Based on the pre-processed test samples, Raman spectroscopy was used to detect the samples. The specific steps are as follows: All processed samples were precisely scanned using a laser Raman spectrometer. The test parameters were set as follows: excitation wavelength 488 nm, laser power 5 mW (care should be taken to avoid laser thermal damage to the silicon lattice); 9 test points were selected for each sample, for example, in a 3×3 array with a spacing of 1 cm, and the peak position ω and full width at half maximum (FWHM) of the silicon characteristic peaks at each test point were recorded; By detecting the test samples using Raman spectroscopy, the following characteristic parameters of the detection signal can be obtained: Raman characteristic peak, and the absolute shift of the Raman characteristic peak relative to the standard characteristic peak. Actual half-height and width, the absolute difference between actual half-height and width and standard half-height and width. And the ratio of that absolute difference to the standard half-width. r.

[0168] Then, data processing and damage index calculation: average the data from the 9 test points of each test sample to obtain the average peak position ω and average half width at half maximum (FWHM); substitute them into the damage index formula to calculate the damage index DI for each sample.

[0169] Specifically, DI = ,in, This represents the absolute displacement of the Raman characteristic peak from the standard characteristic peak w. This is the absolute difference between the actual half-width and the standard half-width r. is the weighting coefficient, which is 0.3, and e is a natural number, approximately 2.718.

[0170] Based on the damage index obtained above, the damage depth of the silicon wafer is obtained. Specifically, a correlation curve is plotted with the removal depth of each test sample as the abscissa and the damage index DI as the ordinate. As the removal depth increases, the damage index DI gradually decreases. When the removal depth increases to a certain critical value, DI tends to stabilize, for example, with a fluctuation range of ≤0.1 micrometers. The corresponding removal depth at this time is the damage depth of the silicon wafer to be tested.

[0171] Control group: Traditional angle polishing method was used. Samples from other parts of the same three silicon wafers were polished at a 10° angle, etched with Secco etching solution, and the starting position of the etched stripes was measured under a metallographic microscope to calculate the damage depth.

[0172] Based on the above process parameters, the complete process of the method of this invention was implemented on silicon wafers A, B, and C respectively, and angle polishing tests were performed simultaneously. The damage depth results measured by the two methods are summarized and compared in the table below.

[0173] Table 1. Relevant data on the measurement methods for silicon wafer A. like Figure 11 As shown, with DI in Table 1 as the vertical axis and etching depth (i.e., material removal depth) as the vertical axis, a graph showing the relationship between the material removal depth and the damage index of the test sample was obtained. The graph shows that the damage index of silicon wafer A at a damage depth of 8 micrometers entered a stable plateau for the first time, that is, the damage depth of silicon wafer A is 8 micrometers. Figure 12 Here is a polishing observation image of the subsurface damage layer of silicon wafer A, as shown below. Figure 12 As shown, the corresponding results indicate that the damage depth of silicon wafer A measured is 7.59 micrometers, which is the average of the two.

[0174] Therefore, the test results provided in this disclosure are quite close to the results provided in the comparative example, thus verifying the effectiveness of this method.

[0175] Table 2. Relevant data on the measurement methods for silicon wafer B. like Figure 13 As shown in Table 2, with DI as the vertical axis and etching depth (i.e., material removal depth) as the horizontal axis, a graph showing the relationship between the material removal depth and the damage index of the test sample was obtained. This graph indicates that the damage index of silicon wafer B first enters a stable plateau at a damage depth of 10 micrometers, meaning the damage depth of silicon wafer B is 10 micrometers. Figure 14 The observation diagram of the subsurface damage layer angle polishing of silicon wafer B is shown. The corresponding result shows that the damage depth of silicon wafer A is 10.51 micrometers, which is the average of the two.

[0176] Therefore, the test results provided in this disclosure are quite close to the results provided in the comparative example, thus verifying the effectiveness of this method.

[0177] Table 3. Relevant data on measurement methods for silicon wafer C. The silicon wafer C corresponding to Table 3 has relatively small damage. The angle polishing method, due to insufficient etching contrast, cannot form a clearly identifiable damage cutoff line, making measurement difficult or the results unreliable. However, the method of this invention amplifies the weak stress difference through wet oxygen pretreatment, still showing a clear DI-depth variation trend in Raman spectroscopy, such as... Figure 15 As shown, the material removal depth threshold at which the damage index first enters the stable platform is 1.5 micrometers, meaning the damage depth of silicon wafer C is 1.5 micrometers. This demonstrates the significant advantage of the method of this invention in terms of detection sensitivity compared to traditional methods, effectively evaluating shallow damage introduced by processes such as grinding and polishing. Throughout the experiment, the operation process of the method of this invention (gradient etching, batch wet oxygen treatment, automated Raman scanning) is easier to standardize and automate, and is far less affected by human operating skills than the angle polishing method, which requires precise control of polishing angle and etching conditions. Therefore, it has potential advantages in terms of measurement consistency and repeatability.

[0178] The embodiments of this disclosure next employ differential interferometry to inspect the surface morphology of the epitaxial wafer, in order to detect the number of step-like defects on the epitaxial wafer with a lateral feature size greater than or equal to 10 micrometers and a depth greater than or equal to 5 nanometers; and to use transmission electron microscopy to detect the number of corresponding epitaxial layer stacking fault defects on the epitaxial wafer. The specific detection methods and parameters can be adjusted according to actual needs, and will not be elaborated here.

[0179] Specifically, the epitaxial wafers include the following three types: one is an epitaxial wafer obtained by the preparation method in this embodiment, which has high crystal quality and is called the example sample; another is an epitaxial wafer formed based on a silicon wafer whose damaged layer depth was measured and removed using an angle polishing method, called the comparative sample; and the last is an epitaxial wafer formed based on a silicon wafer without any treatment, called the blank control group sample. Under the same production process conditions, samples obtained by different detection and processing methods were tested, and the number of stacking fault defects in the epitaxial layer and step defects with a lateral feature size greater than or equal to 10 micrometers and a depth greater than or equal to 5 nanometers were counted. The results are compared in Table 4.

[0180] Table 4 Comparison of defects in epitaxial wafers Based on the experimental results of the blank control group, comparative example, and examples, under the completely identical production process, only the method for measuring damage depth was changed, leading to the following conclusions: Comparison of the three sets of data shows that, under the premise of consistent basic processes, the improvement of the damage depth measurement method directly determines the surface quality of the final epitaxial wafer. This embodiment of the present disclosure achieves precise quantitative detection of damage depth through targeted preprocessing of signal differences. This result provides accurate depth information for the material removal step in the epitaxial fabrication process, thereby enabling the removal of the damaged layer while preserving the intact substrate material to the greatest extent possible. The final result is that the prepared epitaxial wafer has a complete crystal structure (no stacking faults), and the shallow step-like defects on the surface are effectively controlled to an extremely low level (less than 5). This result fully demonstrates that the method of the present invention possesses high-precision detection capabilities and high process applicability in semiconductor manufacturing, and has clear practical application value for improving the quality of high-end epitaxial wafers.

[0181] Based on the physical principles and detection architecture revealed by the silicon wafer damage depth measurement method provided in this disclosure, its core detection logic can also be transformed and directly applied to the rapid and non-destructive batch monitoring of the crystal quality of the epitaxial substrate, i.e., the silicon wafer that has completed all processing and is ready for epitaxial growth.

[0182] For example, this disclosure also provides a method for monitoring the quality of an epitaxial substrate. The monitoring method includes: acquiring an epitaxial substrate to be monitored as a test sample; preprocessing the test sample to amplify the physical state differences between its damaged and undamaged areas; detecting the preprocessed test sample to obtain corresponding detection parameters; determining a quality evaluation index for the test sample based on the detection characteristic parameters; and comparing the quality evaluation index with a preset quality threshold to determine whether the crystal quality of the epitaxial substrate is qualified.

[0183] Thus, this monitoring method enables rapid and consistent quality screening of epitaxial substrates produced in different batches and on different process lines. Substrates with quality evaluation indicators exceeding the threshold range can be judged as unqualified and controlled to prevent them from flowing into subsequent expensive epitaxial growth processes, thereby effectively improving the overall yield and crystal quality consistency of the epitaxial wafer production line. This monitoring method transforms offline, destructive depth measurement technology into a quality monitoring tool that can be used for online or random inspection, expanding the application scenarios and value of the original invention.

[0184] For example, the test sample corresponding to the epitaxial substrate can omit the etching step in the measurement method provided in this disclosure, because its purpose is not to measure the damage depth, but to directly evaluate the overall crystal quality after the substrate is processed.

[0185] For example, the pretreated test sample is subjected to Raman spectroscopy to obtain at least one Raman spectral characteristic parameter of the test sample, including the peak shift and / or full width at half maximum (FWHM) of the silicon characteristic peak.

[0186] For example, the quality evaluation index can be directly adopted using Raman spectral characteristic parameters, such as requiring the peak shift to be within 520 ± Δω0 cm. -1 Within the range, and half-width is less than Alternatively, it can be a comprehensive index calculated based on these parameters (e.g., a calculated value similar to the damage index DI). The preset quality threshold is obtained by calibrating a reference substrate known to be of high quality (e.g., one that has been verified by the process to grow a defect-free epitaxial layer) after performing the same process test.

[0187] For example, the steps for pretreating the test sample can be referred to the specific operation of the measurement method provided in the embodiments of this disclosure, and will not be repeated here.

[0188] This disclosure also provides a device for measuring the depth of damage on a silicon wafer, such as... Figure 16 As shown, the measuring device 100 includes an acquisition module 110, a preprocessing module 120, a detection module 130, and a determination module 140. The acquisition module 110 is configured to acquire a set of test samples, wherein the test samples originate from the same silicon wafer under test, and each test sample corresponds to a different material removal depth. The preprocessing module 120 is configured to preprocess the test samples to amplify the physical state difference between damaged and undamaged areas in the same test sample under the same detection method. The detection module 130 is configured to detect the preprocessed test samples to obtain a damage index for each test sample, wherein the damage index characterizes the degree of damage to the test sample. The determination module 140 is configured to determine the damage depth of the silicon wafer under test based on the material removal depth and damage index corresponding to the test sample.

[0189] In at least one embodiment of this disclosure, the preprocessing module 120 is further configured to select a target detection method from preset detection methods; determine preprocessing rules based on the response characteristics of the target detection method, wherein the response characteristics of the target detection method refer to the response law of the detection signal feature parameters corresponding to the target detection method and the physical state change corresponding to the damaged area of ​​the silicon wafer under test; and preprocess the test sample based on the preprocessing rules.

[0190] In at least one embodiment of this disclosure, the detection module 130 is further configured to detect the pre-processed test sample based on a target detection method to obtain a detection signal of the test sample; analyze and process the detection signal to determine detection signal characteristic parameters, wherein the detection signal characteristic parameters are quantified values ​​used to characterize the corresponding physical state of the damaged area of ​​the test sample; and obtain a damage index based on the detection signal characteristic parameters.

[0191] In at least one embodiment of this disclosure, the detection module 130 is further configured to obtain the absolute displacement of the Raman characteristic peak relative to the standard characteristic peak, and the absolute value of the actual half-width at half-maximum (HWHM) relative to the standard HWHM, based on the displacement of the Raman characteristic peak and the HWHM of the Raman characteristic peak. The absolute displacement of the Raman characteristic peak relative to the standard characteristic peak and the absolute value of the actual HWHM relative to the standard HWHM are then substituted into a preset calculation model to obtain the damage index, wherein the preset calculation model is: DI = ,in, This represents the absolute displacement of the Raman characteristic peak from the standard characteristic peak w. This is the absolute difference between the actual half-width and the standard half-width r. These are the weighting coefficients.

[0192] In at least one embodiment of this disclosure, the determining module 140 is further configured to establish a relationship between the damage index and the material removal depth; based on the relationship, identify the material removal depth threshold corresponding to the first stable platform of the damage index; and determine the material removal depth threshold as the damage depth of the silicon wafer under test.

[0193] This disclosure also provides a computing device, such as... Figure 17 As shown, the computing device 200 includes a processor 210 and a memory 220. The processor 210 is used to execute instructions stored in the memory 220 to implement the measurement method in the above embodiment.

[0194] In at least one embodiment of this invention, the computing device 200 includes one or more processors 210. The application programs stored in the memory 220 may include one or more modules, each corresponding to a set of instructions. Furthermore, the processors 210 are configured to execute instructions to perform the measurement methods described in the above embodiments.

[0195] In at least one embodiment of this invention, the computing device 200 may further include a power supply component configured to perform power management of the computing device 200, a wired or wireless network interface configured to connect the computing device 200 to a network, and an input / output (I / O) interface. The computing device 200 may operate on an operating system stored in memory 220, such as Windows Server™, Mac OS X™, Unix™, Linux™, FreeBSD™, or similar.

[0196] This disclosure provides a computer storage medium storing at least one instruction, which is executed by the processor 210 of the computing device 200 to implement the measurement method described in the above embodiments.

[0197] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0198] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0199] In the embodiments provided in this disclosure, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0200] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0201] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0202] If the functionality is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program verification codes, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0203] Furthermore, it should be noted that the combination of the various technical features in this case is not limited to the combination methods described in the claims of this case or the combination methods described in the specific embodiments. All technical features described in this case can be freely combined or combined in any way, unless they contradict each other.

[0204] It should be noted that the above examples are merely specific embodiments of the present invention, and the present invention is obviously not limited to the above embodiments, with many similar variations. All modifications that can be directly derived or conceived by those skilled in the art from the content disclosed in this invention should fall within the protection scope of this invention.

[0205] It should be understood that the terms "first," "second," etc., mentioned in the embodiments of the present invention are merely for the purpose of more clearly describing the use of the technical solutions in the embodiments of the present invention, and are not intended to limit the scope of protection of the present invention.

[0206] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method of measuring the damage depth of a silicon wafer, characterized by, include: A set of test samples is obtained, wherein the test samples are derived from the same silicon wafer under test, and each test sample corresponds to a different material removal depth; The test sample is pretreated to amplify the difference in physical state between the damaged and undamaged areas in the same test sample under the same detection method; The pretreated test samples are tested to obtain a damage index for each test sample, wherein the damage index is used to characterize the degree of damage to the test sample. The damage depth of the silicon wafer under test is determined based on the material removal depth and the damage index corresponding to the test sample.

2. The measurement method according to claim 1, characterized in that, The pretreatment of the test sample to amplify the difference in physical state between damaged and undamaged areas in the same test sample under the same detection method includes: Select the target detection method from the preset detection methods; Preprocessing rules are determined based on the response characteristics of the target detection method, wherein the response characteristics of the target detection method refer to the response law of the detection signal feature parameters corresponding to the target detection method and the physical state change corresponding to the damaged area of ​​the silicon wafer under test. The test samples are preprocessed based on the aforementioned preprocessing rules.

3. The measurement method according to claim 2, characterized in that, The pretreated test samples are then tested to obtain a damage index for each test sample, wherein the damage index characterizes the degree of damage to the test sample, including: The pre-processed test sample is detected based on the target detection method described above to obtain the detection signal of the test sample; The detection signal is analyzed and processed to determine the detection signal characteristic parameters, wherein the detection signal characteristic parameters are quantified values ​​used to characterize the corresponding physical state of the damaged area of ​​the test sample; The damage index is obtained based on the characteristic parameters of the detection signal.

4. The measurement method according to claim 2, characterized by, The target detection method includes any one of Raman spectroscopy, photoluminescence, X-ray diffraction, and minority carrier lifetime detection.

5. The measurement method according to claim 4, characterized in that, The target detection method includes Raman spectroscopy detection, and the corresponding detection signal characteristic parameters include Raman characteristic peak shift and Raman characteristic peak half width at half maximum (FWHM).

6. The measurement method according to claim 5, characterized in that, The process of obtaining the damage index based on the detection signal feature parameters includes: Based on the Raman characteristic peak displacement and the Raman characteristic peak half-width at half-maximum (FWHM), the absolute displacement of the Raman characteristic peak relative to the standard characteristic peak, and the absolute values ​​of the actual FWHM and the standard FWHM are obtained. The damage index is obtained by substituting the absolute displacements of the Raman characteristic peaks and the standard characteristic peaks, and the absolute values ​​of the actual half-width at half-maximum (HWHM) and the standard HWHM, into a preset calculation model. The preset calculation model is as follows: DI = 0 , wherein, is an absolute displacement amount of the Raman characteristic peak from the standard characteristic peak w, is an absolute difference value of the actual half-height width from the standard half-height width r, is a weight coefficient.

7. The measuring method according to any one of claims 1 to 6, characterized in that, The determination of the damage depth of the silicon wafer under test based on the material removal depth and the damage index corresponding to the test sample includes: Establish the relationship between the damage index and the material removal depth; Based on the aforementioned relationship, identify the material removal depth threshold corresponding to the first time the damage index enters the stable platform; The material removal depth threshold is determined as the damage depth of the silicon wafer under test.

8. A device for measuring the damage depth of a silicon wafer, characterized by include: The acquisition module is configured to acquire a set of test samples, wherein the test samples are derived from the same silicon wafer under test, and each test sample corresponds to a different material removal depth; The preprocessing module is configured to preprocess the test sample to amplify the difference in physical state between the damaged and undamaged areas in the same test sample under the same detection method. The detection module is configured to detect the pre-processed test samples to obtain a damage index for each test sample, wherein the damage index is used to characterize the degree of damage to the test sample. The determination module is configured to determine the damage depth of the silicon wafer under test based on the material removal depth and the damage index corresponding to the test sample.

9. A computing device, comprising: It includes a processor and a memory, the processor being configured to execute instructions stored in the memory to implement the measurement method as described in any one of claims 1 to 7.

10. A computer storage medium, characterized in that, The storage medium stores at least one instruction, which is executed by a processor to implement the measurement method as described in any one of claims 1 to 7.

11. A method for preparing an epitaxial wafer, characterized in that, include: Obtain the target silicon wafer for epitaxial layer growth; The damage depth of the damaged layer of the target silicon wafer is obtained, wherein the damage depth is obtained by the measurement method according to any one of claims 1 to 7; The damaged layer of the target silicon wafer is removed based on the damage depth; An epitaxial layer is grown on the target silicon wafer after the damaged layer is removed to obtain an epitaxial wafer.

12. An epitaxial wafer, characterized in that, The epitaxial wafer is obtained by the preparation method according to claim 11, wherein the epitaxial wafer does not contain stacking fault defects in the epitaxial layer, and On the surface of the epitaxial wafer, the number of stepped defects with a lateral feature size greater than or equal to 10 micrometers and a depth greater than or equal to 5 nanometers is less than 5.