Method for quantitatively detecting stress distribution of damaged layer and intermediate layer on surface of semiconductor wafer

Abstract

The invention discloses a method for quantitatively detecting stress distribution of a damaged layer and a middle layer on the surface of a semiconductor wafer, and belongs to the technical field of semiconductor detection. Comprising the following steps: (1) calculating to obtain a peak position offset and a stress conversion coefficient; (2) a Raman test is carried out on a surface damage layer or a middle layer of the wafer, the diameter of the wafer is 150 mm or above, the surface damage layer is an area extending inwards from the surface of the wafer to be not larger than 30% of the thickness of the wafer, the middle layer is an area extending inwards from the surface of the wafer to be 30%-60% of the thickness of the wafer, a peak position value and a peak position offset are obtained, and the peak position offset is obtained; and calculating surface stress or internal stress of the wafer according to the peak position offset and the peak position offset and stress conversion coefficient to obtain stress distribution of a damaged layer and a middle layer on the surface of the wafer. Quantitative characterization of stress distribution of the damaged layer and the intermediate layer on the surface of the wafer can be realized, and the growth of the crystal and the processing technology of the wafer are fed back according to the detected stress value, so that the growth of the crystal or the technological parameters in the wafer are optimized.

Description

Technical Field

[0001] The present application relates to a method for quantitatively detecting stress distribution of a damaged layer and an intermediate layer on a semiconductor wafer surface, and belongs to the technical field of semiconductor detection. Background Art

[0002] The currently industrialized method for preparing silicon carbide single crystals uses the physical vapor method (PVT) to obtain silicon carbide single crystals, then grinds the crystals flat and rolls them into crystal rods, which are then processed by cutting, grinding, mechanical polishing (MP), chemical mechanical polishing (CMP) and other processing methods with gradually increasing precision to finally obtain conductive or semi-insulating silicon carbide wafers that meet market demand.

[0003] During the growth process of silicon carbide crystals, due to the imbalance of Si / C ratio, the introduction of impurities and temperature gradient changes, various defects such as polycrystalline, polymorphic, micropipes, dislocations, etc. are easily introduced into the crystals. Defects will cause lattice distortion, and more seriously, the crystal form will change, such as the generation of 4H, 6H, 3C, 15R and other polymorphic symbiosis. Lattice distortion releases stress by generating defects, thereby causing internal stress in the silicon carbide crystal. Internal stress will lead to lattice distortion and an increase in defects in the crystal. In severe cases, it will cause the wafer to crack during processing, reduce the yield rate, and be unfavorable for the subsequent processing and production of the crystal.

[0004] On the other hand, when silicon carbide crystals are processed into wafers, they are subjected to external mechanical cutting, grinding, polishing, etc., which will cause incomplete surface lattices, resulting in surface lattice distortion, and then introduce a surface damage layer, forming surface stress. When the wafer is used in epitaxial growth, the surface stress of the surface damage layer will be inherited to the epitaxial layer, resulting in an increase in epitaxial layer defects, increased epitaxial layer stress, and reduced epitaxial uniformity; when the wafer is used as a seed crystal in the growth of SiC crystals, the surface stress will be inherited to the SiC crystal, resulting in increased stress in the SiC crystal body, which is prone to through-type defects, and in severe cases, will cause the crystal to crack during the crystal growth process.

[0005] Therefore, the stress in the wafer is relatively complex. The thermal stress of silicon carbide crystal growth and the residual stress caused by defects in the body are mixed with the surface stress introduced during the silicon carbide wafer processing. It is difficult to distinguish the surface stress of the damaged layer on the wafer surface from the internal stress of the middle layer in the existing technology. At present, there is a lack of a unified quantitative testing method for the stress distribution of the surface damaged layer and the stress distribution of the middle layer of the silicon carbide wafer. Summary of the invention

[0006] In order to solve the above problems, a method for measuring the stress distribution of the surface damage layer and the middle layer of the semiconductor wafer is provided. The method divides the stress in the wafer into surface stress and internal stress, and quantitatively analyzes the surface stress and the internal stress. The surface damage layer and the middle layer stress distribution obtained according to the method can play a feedback role in the adjustment of the parameters of each process of crystal growth and / or wafer processing, thereby optimizing the crystal growth process or the various process parameters of the wafer processing process to prepare high-quality wafers.

[0007] According to one aspect of the present application, a method for quantitatively detecting stress distribution of a damaged layer on a semiconductor wafer surface is provided, comprising the following steps:

[0008] (1) Calculate the peak position shift and stress conversion coefficient;

[0009] (2) Performing Raman testing on a surface damage layer of a wafer, wherein the diameter of the wafer is greater than 150 mm, and the surface damage layer is a region extending inward from the surface of the wafer to no more than 30% of the thickness of the wafer, to obtain a peak value and a peak offset, and calculating the surface stress of the surface damage layer of the wafer according to the peak offset and the peak offset and stress conversion coefficient in step (1), to obtain the stress distribution of the surface damage layer of the wafer.

[0010] The thickness of the surface damage layer is not greater than 30% of the wafer thickness. The thicker the wafer, the smaller the proportion of the surface damage layer to the wafer thickness. The wafer may be a cut wafer, a ground wafer or a polished wafer. Under the same thickness conditions, the thickness of the surface damage layer in the cut wafer is the thickest, the thickness of the surface damage layer in the ground wafer is medium, and the thickness of the surface damage layer in the polished wafer is the thinnest.

[0011] Optionally, the wafer surface stress includes surface absolute stress and surface relative stress;

[0012] The peak position offset of the surface absolute stress is the difference between the actual peak position of the test point and the standard peak position, and the surface relative stress is the difference between the actual peak position of the test point and the actual peak position of the relative point.

[0013] The peak value of the relative point can be a reference value obtained in the entire test plane. The value of each test point is related to the reference value for numerical calculation to obtain the peak value of the relative point. The reference value includes but is not limited to the average value, median, mode and any one of the results calculated by other statistical functions of all stress values ​​in the entire test plane.

[0014] According to different peak offset calculation methods, the surface absolute stress and relative surface stress of the wafer can be obtained. The absolute stress can reflect the difference between the wafer and the perfect SiC crystal without defects, and is used to determine the stress level of the SiC wafer. However, since it is difficult to achieve a perfect SiC wafer with current technology, the relative stress distribution within the wafer surface can be determined by the relative stress.

[0015] For example, when the absolute stress of the wafer is large and the relative stress is small, it means that there are more defects in the surface, but the stress distribution is uniform, the crystal quality is poor, and the uniformity quality in each area of ​​the wafer is poor; if the absolute stress is small and the relative stress is large, the overall wafer quality is good, but there are abnormal quality deviations in some parts.

[0016] Optionally, the thickness of the surface damage layer is obtained by performing Raman axial testing on the wafer.

[0017] Optionally, before step (2), a step of performing a curvature test on the wafer surface is also included. According to the curvature of the wafer, an equidistant focusing test is performed on the damaged layer on the wafer surface in step (2). The equidistant focusing test is to control the distance between the Raman laser spot and the wafer surface at each test point to be equal, and the peak position value and peak position offset are obtained after fitting.

[0018] According to another aspect of the present application, a method for quantitatively detecting stress distribution of an intermediate layer of a semiconductor wafer is provided, comprising the following steps:

[0019] (1) Calculate the peak position shift and stress conversion coefficient;

[0020] (2) Performing Raman testing on the wafer middle layer, wherein the diameter of the wafer is greater than 150 mm, and the wafer middle layer is a region extending inward from the surface of the wafer to 30%-60% of the wafer thickness, to obtain a peak value and a peak offset, and calculating the internal stress of the wafer middle layer based on the peak offset and the peak offset and stress conversion coefficient in step (1), to obtain the stress distribution of the wafer middle layer.

[0021] Optionally, the Raman test in step (2) is an equidistant focusing test and / or an equifocal plane test, wherein the equidistant focusing test is to control the distance between the Raman laser spot at each test point and the wafer surface to be equal, and the equifocal plane test is to control the Raman laser spot at each test point to be in the same horizontal plane.

[0022] Optionally, the difference between the stress values ​​in the wafer calculated by the equal-interval focusing test and the equal-focusing plane test is ≤0.2 MPa.

[0023] Optionally, the peak position difference obtained by the equal spacing focus test and the equal focusing plane test is less than 0.0005 cm -1 .

[0024] Preferably, the difference between the stress values ​​in the wafer calculated by the equal-interval focusing test and the equal-focusing plane test is ≤0.05 MPa.

[0025] Optionally, the peak position difference obtained by the equal spacing focus test and the equal focusing plane test is <0.0001cm -1 .

[0026] Optionally, the thickness inside the wafer is obtained by performing axial Raman testing on the wafer.

[0027] Optionally, the wafer internal stress includes internal absolute stress and internal relative stress;

[0028] The peak position offset of the in-vivo absolute stress is the difference between the actual peak position of the test point and the standard peak position, and the in-vivo relative stress is the difference between the actual peak position of the test point and the actual peak position of the relative point.

[0029] The wafer's internal absolute stress and internal relative stress can be obtained according to different peak offset calculation methods. The internal absolute stress can reflect the relative stress between the wafer and the body, and the surface absolute stress and surface relative stress can reflect the same characteristics of the wafer.

[0030] In the above-mentioned wafer surface damaged layer stress distribution test and wafer intermediate layer stress distribution test, the wafer includes an epitaxial wafer and a substrate wafer.

[0031] Optionally, the diameter of the wafer is greater than 200 mm. It can be understood by those skilled in the art that the larger the size of the wafer, the more detection points involved in its Raman test, the greater the workload for stress detection and calculation, and the greater the stress distribution error obtained. The test method of the present application can maintain a high detection efficiency when detecting an 8-inch wafer, and does not increase the measurement error due to the increase in wafer size.

[0032] In the above-mentioned wafer surface damaged layer stress distribution test and wafer intermediate layer stress distribution test, the wafer surface is taken as the zero point, and the test range of the axial test is from -150μm to -30μm inside the wafer to above the wafer surface. According to the axial test, the thickness of the wafer surface damaged layer, the damaged stress extension layer and the intermediate layer is determined.

[0033] Preferably, with the wafer surface as the zero point, the test range of the axial test is from -100 μm to -30 μm inside the wafer to above the wafer surface.

[0034] In the above-mentioned wafer surface damaged layer stress distribution test and wafer intermediate layer stress distribution test, the focus depth (d) and the spot diameter (D) of the Raman spot are determined according to the thickness of the wafer surface damaged layer, the damaged stress extension layer and the intermediate layer, so that the Raman spot is located on the wafer surface damaged layer or the intermediate layer. The calculation formula for selecting the focus depth (d) is:

[0035] Where Dp is the penetration depth of the Raman laser in the sample, α is the absorption coefficient, λ is the laser wavelength, and k is the extinction coefficient. The appropriate focusing depth (d) is selected according to the calculated penetration depth (Dp). The requirement that needs to be met is that the penetration depth (Dp) is greater than the focusing depth (d).

[0036] In the stress distribution test of the wafer surface damage layer, the Raman laser spot is ensured to be located in the surface damage layer. In the stress distribution test of the wafer middle layer, the focusing depth (d) of the Raman laser needs to pass through the surface damage layer and the damage stress extension layer to ensure that the Raman laser spot is located in the wafer middle layer.

[0037] The calculation formula of the spot diameter (D) is:

[0038] Where λ is the laser wavelength, and Na is the numerical aperture of the device. The Raman laser spot is ellipsoidal, with the long axis of the laser spot parallel to the axial direction of the wafer, and the short axis of the laser spot parallel to the radial direction of the wafer. The spot diameter (D) is the short axis diameter of the laser spot. The larger this value is, the stronger the peak intensity of Raman at each test position is, and the more impurity peaks there are, which affects the test accuracy of Raman.

[0039] In the above-mentioned wafer surface damaged layer stress distribution test and wafer intermediate layer stress distribution test, the specific method of the curvature test is:

[0040] S1: Ensure that the test platform is level and select a fixed origin (0,0,0) on the test platform;

[0041] S2: Use Raman to automatically focus on the position with the strongest test signal on the wafer surface, mark this position as (a, b, c), then adjust the Raman laser to move down a distance of H and focus on the position to (a, b, cH);

[0042] S3: Adjust the Raman laser test position to (0,0,cH) to perform an equal focusing plane test to obtain a peak intensity scan of the wafer surface. Calculate the distance from the focusing plane to the wafer surface at different positions to determine the curvature of the wafer surface.

[0043] Optionally, the value range of H is 5-150 μm, preferably 5-100 μm.

[0044] According to the curvature of the wafer surface, an equidistant focusing test can be performed in the stress distribution test of the wafer surface damage layer, and an equidistant focusing test or an equifocusing plane test can be performed in the wafer body stress test. Both can eliminate the influence of the wafer surface curvature on the stress test and accurately test the surface stress or body stress value of the wafer.

[0045] In the above-mentioned wafer surface damaged layer stress distribution test and wafer intermediate layer stress distribution test, the calculation steps of the peak position offset and the stress conversion coefficient are as follows:

[0046] Test and calculate the interplanar spacing dx and interplanar spacing difference Δd at different positions of the wafer, and use Hooke's law or stress-strain formula to calculate the stress values ​​at different positions;

[0047] The peak position value of the wafer is tested by Raman, and the peak position offset at different positions is calculated. The ratio of the peak position offset to the stress value is the peak position offset and stress conversion coefficient.

[0048] Optionally, the wafer is tested using any one or more of an XRD diffractometer, a synchrotron radiation tester, a white light tester, a neutron diffractometer, a scanning electron microscope, and a transmission electron microscope to obtain the interplanar spacing dx at different positions of the wafer, and calculate the interplanar spacing difference Δd.

[0049] When the wafer is tested by an XRD diffractometer, the interplanar spacing dx at different positions of the wafer is calculated using the Bragg diffraction formula.

[0050] In the above-mentioned wafer surface damaged layer stress distribution test and wafer intermediate layer stress distribution test, the peak value and peak offset in step (2) are obtained by fitting, and the fitting uses functions including Gaussian, Lorentz, GaussLor, AGauss, Aloren and AgaussLor.

[0051] Optionally, the wafer is a silicon carbide wafer;

[0052] Preferably, the crystal form of the silicon carbide wafer is one of 2H-SiC, 4H-SiC, 6H-SiC, 3C-SiC, and 15R-SiC.

[0053] The beneficial effects of this application include but are not limited to:

[0054] 1. This application performs axial testing on the wafer, thereby dividing the wafer into a surface damage layer and an intermediate layer, and quantitatively analyzes the stress distribution of the surface damage layer and the intermediate layer to achieve quantitative characterization of different parts of the wafer.

[0055] 2. The wafer surface stress and internal stress values ​​obtained according to the present application can play a feedback role in the adjustment of the parameters of each process of crystal growth and / or wafer processing, thereby optimizing the crystal growth process or the various process parameters of the wafer processing process to prepare high-quality wafers.

[0056] 3. Currently, only qualitative characterization has been achieved for SiC stress, and quantitative characterization has not yet been achieved. In addition, qualitative characterization is mainly for stress caused by defects, while stress that has not yet been released through defects due to lattice distortion cannot be characterized. The present application provides a method for measuring the stress distribution of the damaged layer and the intermediate layer on the surface of a silicon carbide wafer, which solves the technical problem that only qualitative characterization can be achieved for silicon carbide wafers. In addition, it can also quantitatively test the stress that has not yet been released through defects due to lattice distortion in the wafer.

[0057] 4. The measurement method of the present application can eliminate the influence of wafer surface bending and warping, realize quantitative testing of wafer surface stress and internal stress, and improve the testing accuracy of the testing method.

[0058] 5. This application adopts two testing methods: equal-interval focusing and equal-focusing plane testing, which can verify the accuracy of the method of measuring the stress distribution of the damaged layer and the intermediate layer on the surface of the semiconductor wafer in this application, and the accuracy of the measurement method can be further improved according to the setting of the focusing depth and the spot diameter. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0060] Figure 1 This is a diagram showing the calculation principle of the peak position offset and the stress conversion coefficient involved in Example 1 of the present application.

[0061] Figure 2 This is a flow chart of the Raman stress test involved in Example 1 of the present application.

[0062] Figure 3 Schematic diagram of stress distribution in the middle layer of the Raman equidistant focusing test substrate involved in Example 4 of the present application and stress distribution test results.

[0063] Figure 4 Schematic diagram of stress distribution in the middle layer of the Raman isofocusing plane test substrate involved in Example 4 of the present application and stress distribution test results.

[0064] Figure 5 Schematic diagram of stress distribution in the surface damaged layer of the Raman equidistant focusing test substrate involved in Example 5 of the present application and stress distribution test results. DETAILED DESCRIPTION

[0065] The present application is described in detail below in conjunction with the embodiments, but the present application is not limited to these embodiments. The quantitative test method of the present application can be used to detect the surface stress of the surface damage layer of the wafer and the internal stress of the intermediate layer. In the following embodiments, 4H-SiC single crystal substrate is used as an example for explanation. It is understood by those skilled in the art that this method is applicable to any type of wafer, and the type and crystal form of the wafer do not constitute a limitation to this test method.

[0066] The principle of using Raman to test the stress of the substrate in the following embodiments is as follows: the Raman spectrometer passes a monochromatic laser of a certain wavelength through a series of optical paths and then converges the laser through an objective lens to irradiate the substrate surface, and the laser photons interact with the substrate lattice to cause photon scattering. The occurrence of Raman scattering is related to the lattice vibration of the substrate itself. When there is residual stress in the sample, the tensile and compressive stresses will cause the atomic bond length to be elongated or shortened accordingly, thereby causing the atomic vibration frequency to decrease or increase, which is reflected in the Raman spectrum as the characteristic peak of the substrate moves to a low frequency or a high frequency, that is, a peak position offset occurs.

[0067] Embodiment 1 Method for quantitatively detecting stress of semiconductor substrate

[0068] (1) Calculation of peak position shift and stress conversion coefficient

[0069] First, the silicon carbide substrate is tested using a single crystal XRD diffractometer and the substrate is analyzed based on the reciprocal lattice space. The parameters of the standard 4H-SiC substrate are calculated and compared with the standard parameters, such as Figure 1 (a) is shown. Then, according to the Bragg diffraction equation: 2dsinθ=nλ, where d is the crystal plane spacing, θ is the diffraction half angle, n is the diffraction order, and λ is the wavelength, the θ and d of different substrates are tested, as shown in Figure 1 As shown in (b), it can be seen that the tensile and compressive stresses inside the crystal cause the interplanar spacing d to expand and contract accordingly. x , d x It is divided into d1 and d2, where d0 is the theoretical crystal plane spacing, d1 is the crystal plane spacing after tensile stress, and d1 is greater than d0, and d2 is the crystal plane spacing after compressive stress, and d2 is less than d0. The difference in crystal plane spacing Δd at different positions is calculated, Δd=d1-d0 or Δd=d0-d2.

[0070] The actual peak position of the substrate is obtained according to the Raman test, and the peak position offset Δv at different test points is calculated compared with the standard peak position; Figure 1 (c) The stress value σ of the silicon carbide substrate at different test points is calculated by combining Hooke's law or the stress-strain formula. The ratio of the peak offset to the stress value at different test points is the peak offset to stress conversion coefficient μ, which ranges from -125 to -2500.

[0071] The calculation formula of Hooke's law is σ / S=E×(Δd / d0), where S is the area affected by σ, which is a fixed value. When Hooke's law is applicable to microscopic calculations, S is the microscopic area, which can be calculated as unit 1. E is the Young's modulus of the substrate, which is a fixed parameter of the substrate. Δd is the difference in crystal plane spacing at different positions, d0 is the theoretical crystal plane spacing, and σ is the stress value that needs to be calculated.

[0072] The calculation formula for stress and strain is σ=E×Δd, where E is the Young's modulus of the substrate, which is a fixed parameter of the substrate, Δd is the difference in crystal plane spacing at different positions, and σ is the stress value to be calculated.

[0073] (2) Quantitative test method of substrate stress

[0074] First, set the Raman test points. You can perform single-point testing or Mapping scanning of the entire substrate. Therefore, single-point testing is a sporadic coordinate point setting, and Mapping scanning is 10x10~50x50 and more points. You can choose the appropriate test method according to the properties of the substrate, such as Figure 2 (a) shows the Raman test coordinate diagram of the silicon carbide substrate.

[0075] Then the test results are output, such as Figure 2 As shown in (b), the results include but are not limited to editable peak position, peak intensity data and Mapping.

[0076] Then Figure 2 (c) The output data is subjected to peak position fitting, and the fitting functions include but are not limited to Gaussian, Lorentz, GaussLor, AGauss, Aloren and AGaussLor to obtain accurate peak position values ​​and peak position offsets. The peak position is accurate to more than one decimal place. The peak position value and peak offset results are shown in Figure 2 (d) as shown.

[0077] Finally, according to the peak position shift and stress conversion coefficient μ calculated in (1) and the peak position shift Δv obtained in the Raman test, the stress value of silicon carbide is calculated. The calculation formula is as follows: Figure 2 As shown in (e), σ(MPa)=μ×Δv(cm -1 ), the final stress test image of the substrate is as follows Figure 2 (f) as shown.

[0078] Example 2 Test of curvature of substrate surface

[0079] Since SiC is subjected to a lot of external mechanical stress during the process of preparing wafers from crystals through a series of processes, it is difficult to ensure that the surface is completely level, that is, there is a certain degree of curvature. In order to eliminate the influence of substrate surface curvature on substrate surface stress and internal stress test, it is necessary to first perform a curvature test on the substrate. According to the curvature, the substrate surface stress is tested with equal spacing focus, and the substrate internal stress is tested with equal spacing focus or equal focus plane.

[0080] After the curvature test, the tester needs to be calibrated using single crystal silicon. First, a Z-axis mapping test is performed on the single crystal Si to confirm that it is a Gaussian distribution. The surface of the single crystal Si substrate is taken as the zero point, and an axial mapping test is performed from -100μm inside the single crystal Si substrate to +20μm above the single crystal Si substrate to analyze the change in Raman signal intensity. The single crystal Si substrate is a non-transparent material, so its peak intensity presents a Gaussian distribution, with the surface zero point position as the symmetry axis; then a surface focusing single point test is performed on the single crystal Si, and the single crystal Si Raman peak position is made to be 520.70cm by the correction instrument. -1 , at this time, it means that the instrument optical path has been calibrated.

[0081] The 4H-SiC substrate was tested in the Z direction using a calibrated tester. The Raman laser was emitted from -100μm inside the 4H-SiC substrate to +20μm above the 4H-SiC substrate. The 4H-SiC substrate and other silicon carbides are transparent materials, so the Raman laser can penetrate into the silicon carbide substrate, and the feedback signal gradually weakens as the Raman laser incident depth increases. Other silicon carbides include semi-insulating, conductive, and 3C, 4H, 15R, 6H and other silicon carbide substrates with different crystal forms. Therefore, the Raman peak intensity of the silicon carbide substrate gradually decreases as the Raman laser penetrates deeper into the silicon carbide substrate.

[0082] The Raman device’s autofocus function can be used to automatically find the strongest peak of the test signal on the substrate surface, so that the curvature of the substrate surface can be measured with the help of this function. The specific method is as follows:

[0083] S1: First, ensure that the test platform is level, select a fixed origin (0,0) on the test platform, and mark the point as the three-dimensional coordinate (0,0,0);

[0084] S2: Raise the objective lens to a height that does not touch the substrate according to the thickness of the substrate, and automatically focus on any point on the substrate surface where the test signal is the strongest. Mark this position as (a, b, c), then adjust the Raman laser downward by a distance of H and focus on the position to (a, b, cH);

[0085] S3: Adjust the Raman laser test position to (0,0,cH) to perform an iso-focal plane test to obtain a two-dimensional peak intensity scan of the substrate surface. According to the peak intensity distribution relationship of the silicon carbide substrate and the two-dimensional peak intensity scan, the distance between the focal plane of the Raman laser and the surface of the silicon carbide substrate can be obtained, thereby obtaining the in-plane curvature of the substrate and obtaining the in-plane curvature distribution map of the substrate.

[0086] Example 3 Axial test to distinguish between substrate surface damaged layer and intermediate layer

[0087] Since the substrate is affected by the temperature field and the introduction of inclusions during the crystal growth process, thermal stress and residual stress exist inside the crystal, and the crystal is obtained after a series of cutting, grinding and polishing. The processing process of the crystal to form the substrate will cause surface lattice distortion, thereby generating surface stress. In order to further distinguish the surface damage layer and the intermediate layer of the substrate, the substrate is subjected to axial testing, namely, Z-direction Mapping, according to the light transmittance characteristics of silicon carbide. The substrate surface is taken as the zero point, and the test range of the axial test is from -150μm to -30μm inside the substrate to above the substrate surface.

[0088] During the test, the laser spot gradually focuses upward from the substrate body to the damaged stress extension layer, further to the surface damaged layer, and then continues to focus upward into the air to determine the thickness of the substrate surface damaged layer and the damaged stress extension layer. With the substrate surface as the zero point, the test range of the axial test is that the Raman laser spot gradually moves upward from -100μm inside the substrate to 30μm above the substrate surface. Five points of the substrate are tested up and down, left and right. The five points on the substrate show the same trend, and there is only a difference in the maximum negative stress at the surface damage.

[0089] The laser moves from -100μm inside the substrate to -30μm. At this time, the stress fluctuates up and down along the horizontal line, indicating that the internal stress is uniform. When the laser spot is focused on the damaged stress extension layer, the stress curve begins to show a downward trend at -30μm, and the downward trend of stress gradually increases. When it approaches the zero point on the substrate surface, it reaches the maximum negative stress value, which is the stress value of the surface damaged layer. Finally, the laser spot continues to move upward and gradually leaves the substrate surface to the air. The stress shows a rapid rise to chaotic fluctuations. Therefore, the thickness of the surface damaged layer and the damaged stress extension layer of the substrate can be determined based on the axial test results of the substrate.

[0090] Example 4 Quantitative detection of stress distribution in the substrate intermediate layer

[0091] During the PVT crystal growth process, silicon carbide is affected by temperature gradient changes and the intrusion of inclusions, resulting in residual stress in the crystal. Different from the surface processing stress, the internal stress is located at a certain depth from the surface, reflecting the thermal stress and residual stress of crystal growth. In addition, silicon carbide undergoes a large amount of external mechanical stress in the process of preparing the substrate from the crystal through a series of processing, and it is difficult to ensure that the surface is completely horizontal, that is, there is a certain degree of curvature on the surface of the substrate. According to the measurement method of the curvature of the substrate surface obtained by Raman testing in Example 2, this embodiment, combined with the physical properties of silicon carbide, involves a method for quantitatively characterizing the internal stress in its intermediate layer that is not affected by the surface curvature.

[0092] According to the physical property of silicon carbide being light-transmissive, the method of using Raman to measure the stress distribution in the substrate intermediate layer is divided into equal-spacing focusing test and equal-focusing plane test, as follows:

[0093] 1. Equal spacing focus test

[0094] The principle of equidistant focusing test is as follows Figure 3 As shown, Figure 3 (b) Schematic diagram of the focus of the Raman laser on the middle layer of the substrate. Figure 3 (a) is the test principle diagram. Under this test mode, the focal depth (d) of the Raman laser focused on the inside of the substrate at different test points remains unchanged. First, the penetration depth (Dp) of the Raman laser and the spot diameter (D) of the Raman laser are calculated according to the formula:

[0095] According to the characteristics of silicon carbide materials, the peak intensity reaches the highest when the laser is focused on the substrate surface. Therefore, the automatic focusing function of the Raman tester can be used to focus the laser on the substrate surface. Then, according to the focus depth (d) determined above, the equidistant focus test is performed on different test points of the substrate, that is, the distance between the focus plane and the substrate surface is always equal. This method is not affected by the bending and warping of the substrate surface, so that the obtained Raman peak intensity is always maintained at the maximum axial value of the test point. According to the stress quantitative test method of Example 1, the in vivo stress test results of the substrate are obtained as follows: Figure 3 (c) as shown.

[0096] 2. Equifocal plane test

[0097] The principle of isofocal plane test is as follows Figure 4 As shown, Figure 4 (b) Schematic diagram of the focus of the Raman laser on the middle layer of the substrate. Figure 4(a) is the test principle diagram. Under this test mode, the focus depth (d) of the Raman laser spot focused on the inside of the substrate changes at different test points. The laser spot at different test points is focused on the same plane inside the substrate. The penetration depth (Dp) of the Raman laser is calculated according to the same formula as the equidistant focusing test, and the appropriate focus depth (d) is selected in the same selection method.

[0098] Since the focus planes in this test method are equal, combined with the characteristics of silicon carbide, when the laser is focused on the substrate surface, the peak intensity reaches the highest and the signal gradually decays with the increase of the focus depth. And according to the effective signal-to-noise ratio tested by the Raman tester, it is found that the effective signal-to-noise ratio is inversely correlated with the Raman laser focus depth (d). Therefore, in this test method, the focus depth (d) at different test points needs to be combined with the curvature of the substrate to obtain the effective peak position offset through the stress quantitative test method of Example 1, thereby reducing the error of the in vivo stress test. The test method obtains the in vivo stress test results of the substrate as shown in FIG. Figure 4 (c) as shown.

[0099] Through the above two testing methods of equal spacing focusing and equal focusing plane, the peak position offset of the substrate middle layer is obtained respectively, and the internal stress distribution of the substrate is calculated. The difference of the internal stress value of the substrate calculated by the two methods of equal spacing focusing test and equal focusing plane test is ≤0.2Mpa, and the peak position difference obtained by the two methods of equal spacing focusing test and equal focusing plane test is <0.0005cm -1 .

[0100] The internal stress includes the internal absolute stress and the internal relative stress, which is illustrated by taking 4H-SiC as an example:

[0101] The absolute stress in the body is calculated by subtracting the measured 4H-SiC peak position from the standard peak position to determine the simplified wave vector peak position, such as FTO (2 / 4). The absolute stress in the body is calculated as:

[0102] σabsolute=μ×(v 实际 -V FTO(2 / 4) ), σ is in MPa, v 实际 and V FTO(2 / 4) The unit is cm -1 , where v 实际 is the actual FTO(2 / 4) peak position at the substrate test point, V FTO(2 / 4) is the standard peak position of FTO (2 / 4) of the substrate;

[0103] The relative stress in the body is calculated by subtracting the measured values ​​of the same simplified wave vector peak position at any two points in the SiC substrate surface, such as the FTO (2 / 4) peak position. The calculation formula for the relative stress in the body is:

[0104] σrelative=μ×(v (x1,y1) -V(x2,y2) ), σ is in MPa, v (x1,y1) and V (x2,y2) The unit is cm -1 , where v (x1,y1) is the actual FTO(2 / 4) peak position at the substrate test point, V (x2,y2) is the actual FTO (2 / 4) peak position of the substrate relative point. The relative stress value of the test point is calculated according to the above calculation formula. If the relative value of σ is positive, it means that the stress of the test point relative to the relative point is tensile stress. If the relative value of σ is negative, it means that the stress of the test point relative to the relative point is compressive stress.

[0105] Example 5 Quantitative detection of stress distribution of damaged layer on substrate surface

[0106] According to the axial test of Example 3, the thickness of the surface damage layer of the substrate is determined. Since the thickness of the surface damage layer of the substrate is less than the curvature of the substrate surface, the laser spot falls into the interior of the substrate in the equal-focus plane test method, resulting in errors in the test results of the substrate surface stress. Therefore, the equal-space focusing test method is mainly used in the quantitative test of the surface stress. The test principle is as follows Figure 5 As shown, Figure 5 (b) Schematic diagram of the focus of the Raman laser on the damaged layer on the substrate surface. Figure 5 (a) is the test principle diagram. Under this test mode, the focal depth (d) of the Raman laser focused on the damaged layer on the substrate surface at different test points remains unchanged.

[0107] The penetration depth (Dp) of the Raman laser and the spot diameter (D) of the Raman laser are calculated according to the formula. The focus depth (d) and the spot diameter (D) are determined according to the above. By debugging the parameters of the Raman tester, the laser is focused on the substrate surface using the automatic focusing function so that the surface damage layer of the substrate falls within the range of the laser spot, thereby achieving equal-spaced focusing tests at different test points of the substrate, that is, the distance between the focusing plane and the substrate surface is always equal. This method is not affected by the bending and warping of the substrate surface, so that the obtained Raman peak intensity is always maintained at the axial maximum value of the test point. According to the stress quantitative test method of Example 1, the surface stress test results of the substrate are obtained as follows: Figure 5 (c) as shown.

[0108] The surface stress includes surface absolute stress and surface relative stress. The calculation formula of surface absolute stress is the same as that of body absolute stress. The calculation formula of surface relative stress is the same as that of body relative stress. The surface absolute stress and surface relative stress of the substrate can be measured according to different calculation formulas.

[0109] This application analyzes and simulates the crystal growth stress based on the characteristics of SiC crystals obtained by PVT method. Figure 1 As shown in (d), the in-plane radial relative stress actually measured by the above quantitative test method and the simulated radial relative stress results both show a symmetrical distribution with high edge stress and low center stress, and the simulation results are consistent with the actual measured results; the simulated axial absolute stress and the actual measured in-plane axial absolute stress show a distribution with positive front stress and negative rear stress, and the simulation results are also consistent with the actual measured results. The values ​​of absolute stress and relative stress calculated by the calculation formula of this embodiment are consistent with the magnitude of the simulation results, both of which are 10~10 2 MPa. Since the SiC crystal is thick and the penetration depth of Raman laser is limited, the SiC crystal is cut into multiple substrates. The actual measured axial absolute stress is obtained by testing the in vivo absolute stress of different substrates on the same axis.

[0110] In Example 1 of the present application, the peak position offset and stress conversion coefficient of the substrate are obtained according to the different characteristics of the substrate, which provides a basis for the quantitative test of the substrate stress. The surface damage layer and the intermediate layer of the substrate are distinguished according to the axial test method of Example 3, and the curvature of the substrate is determined according to the above-mentioned Example 2. The stress distribution test in the surface damage layer selects an equidistant focusing test, and the stress distribution in the intermediate layer selects an equidistant focusing test and / or an equifocusing plane test. The focusing depth (d) and the spot diameter (D) of the Raman laser spot are calculated according to the selected test method, thereby eliminating the influence of the substrate surface curvature on the stress in the substrate surface damage layer and the intermediate layer, and providing a unified standard for the quantitative characterization of the stress distribution in the substrate surface damage layer and the intermediate layer.

[0111] The above is only the embodiment of the present application, and the protection scope of the present application is not limited by these specific embodiments, but is determined by the claims of the present application. For those skilled in the art, the present application can have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the technical ideas and principles of the present application should be included in the protection scope of the present application.

Claims

1. A method for quantitatively detecting stress distribution of a damaged layer on a semiconductor wafer surface, characterized in that: The steps include: (1) Calculate the peak position shift and stress conversion coefficient; (2) Performing Raman testing on a surface damage layer of a wafer, wherein the diameter of the wafer is greater than 150 mm, and the surface damage layer is a region extending inward from the surface of the wafer to no more than 30% of the thickness of the wafer, to obtain a peak value and a peak offset, and calculating the surface stress of the surface damage layer of the wafer according to the peak offset and the peak offset and stress conversion coefficient in step (1), to obtain the stress distribution of the surface damage layer of the wafer.

2. The method according to claim 1, characterized in that The surface stress includes surface absolute stress and surface relative stress; The peak position offset of the surface absolute stress is the difference between the actual peak position of the test point and the standard peak position, and the surface relative stress is the difference between the actual peak position of the test point and the actual peak position of the relative point; and / or The wafer comprises an epitaxial wafer and a substrate wafer, and the diameter of the wafer is greater than 200 mm.

3. The method according to claim 1, characterized in that The thickness of the surface damage layer is obtained by performing axial Raman testing on the wafer.

4. The method according to claim 1, characterized in that: Before step (2), the method also includes the step of performing a curvature test on the wafer surface. According to the curvature of the wafer, in step (2), an equidistant focusing test is performed on the damaged layer on the wafer surface. The equidistant focusing test is to control the distance between the Raman laser spot and the wafer surface at each test point to be equal. The peak position value and the peak position offset are obtained after fitting.

5. A method for quantitatively detecting stress distribution in an intermediate layer of a semiconductor wafer, characterized in that: The steps include: (1) Calculate the peak position shift and stress conversion coefficient; (2) Performing Raman testing on the wafer middle layer, wherein the diameter of the wafer is greater than 150 mm, and the wafer middle layer is a region extending inward from the surface of the wafer to 30%-60% of the wafer thickness, to obtain a peak value and a peak offset, and calculating the internal stress of the wafer middle layer based on the peak offset and the peak offset and stress conversion coefficient in step (1), to obtain the stress distribution of the wafer middle layer.

6. The method according to claim 5, characterized in that The Raman test in step (2) is an equidistant focusing test and / or an equifocal plane test, wherein the equidistant focusing test is to control the distance between the Raman laser spot at each test point and the wafer surface to be equal, and the equifocal plane test is to control the Raman laser spot at each test point to be in the same horizontal plane; and / or The wafer comprises an epitaxial wafer and a substrate wafer, and the diameter of the wafer is greater than 200 mm.

7. The method according to claim 6, characterized in that The difference between the stress values ​​in the wafer body calculated by the equal spacing focus test and the equal focusing plane test is ≤0.2Mpa; or The peak position difference obtained by fitting the equal spacing focus test and the equal focusing plane test is less than 0.0005cm -1 .

8. The method according to claim 5, characterized in that The thickness inside the wafer is obtained by performing axial Raman testing on the wafer.

9. The method according to claim 5, characterized in that The wafer internal stress includes internal absolute stress and internal relative stress; The peak position offset of the in-vivo absolute stress is the difference between the actual peak position of the test point and the standard peak position, and the in-vivo relative stress is the difference between the actual peak position of the test point and the actual peak position of the relative point.

10. The method according to any one of claims 1 to 4 or any one of claims 5 to 9, characterized in that: The calculation steps of the peak position offset and the stress conversion coefficient are as follows: Test and calculate the interplane spacing dx and interplane spacing difference △d at different positions of the wafer, and use Hooke's law or stress-strain formula to calculate the stress values ​​at different positions; The peak position value of the wafer is tested by Raman, and the peak position offset at different positions is calculated. The ratio of the peak position offset to the stress value is the peak position offset and stress conversion coefficient.

Images