METHOD FOR EVALUATING SiC SUBSTRATE, METHOD FOR MANUFACTURING SiC SUBSTRATE, AND DEVICE FOR EVALUATING SiC SUBSTRATE

Abstract

Provided are a method for evaluating a SiC substrate, a method for manufacturing a SiC substrate, and a device for evaluating a SiC substrate. A method for evaluating a SiC substrate according to an embodiment of the present invention comprises: an image acquisition step for acquiring an X-ray topography image of the entirety of a first surface of a SiC substrate; and an estimation step for estimating the base surface dislocation density of the SiC substrate from the X-ray topography image of the entirety of the first surface of the SiC substrate on the basis of the learning result of the deep learning.

Description

Technical Field

[0001] The present invention relates to a method for evaluating a SiC substrate and a method for producing a SiC substrate.

[0002] This application claims priority based on Special Application No. 2023-120078 filed in Japan on July 24, 2023, and Special Application No. 2023-182652 filed in Japan on October 24, 2023, and the contents of which are incorporated herein by reference. Background Art

[0003] Compared with silicon (Si), the insulation breakdown electric field of silicon carbide (SiC) is one order of magnitude larger and the band gap is three times larger. In addition, silicon carbide (SiC) has characteristics such as thermal conductivity being about three times higher than that of silicon (Si). Therefore, it is expected that silicon carbide (SiC) will be applied to power devices, high-frequency devices, high-temperature working devices, etc. Therefore, in recent years, SiC epitaxial wafers are used for substrates of semiconductor devices such as the above.

[0004] SiC epitaxial wafers are obtained by stacking SiC epitaxial layers on the surface of SiC substrates. Hereinafter, the substrate before stacking the SiC epitaxial layer is referred to as SiC substrate, and the substrate after stacking the SiC epitaxial layer is referred to as SiC epitaxial wafer. SiC substrates are made by cutting out SiC ingots. SiC devices such as power devices, high-frequency devices, and high-temperature working devices are obtained by chipping SiC epitaxial wafers after the SiC epitaxial layer of the SiC epitaxial wafer forms the device.

[0005] Basal plane dislocation (BPD) is known as one of the fatal defects of SiC devices in SiC epitaxial wafers. For example, when current flows in the forward direction of a bipolar device, due to the recombination energy of carriers, part of the basal plane dislocation extending from the SiC substrate to the SiC epitaxial layer moves and expands to form a high-resistance stacking fault. The high-resistance part generated in the device becomes the cause of the reduction in device reliability (forward degradation).

[0006] Basal plane dislocations in SiC epitaxial layers are often inherited from basal plane dislocations in SiC substrates. Therefore, methods for evaluating basal plane dislocations in SiC substrates are being sought.

[0007] For example, Patent Document 1 discloses evaluating the in-plane distribution of basal plane dislocations of a SiC substrate using the correlation between the contrast of an X-ray topography image and basal plane dislocations.

[0008] In addition, for example, Patent Document 2 discloses a method of classifying defects one by one using a neural network.

[0009] Prior art literature

[0010] Patent Document 1: Japanese Patent No. 6197722

[0011] Patent Document 2: U.S. Patent Application Publication No. 2020 / 365685 Summary of the invention

[0012] Technical problem to be solved by the invention

[0013] The method described in Patent Document 1 uses only the contrast of the image as a measurement parameter to determine the dislocation density, so the measurement accuracy of the dislocation density is sometimes insufficient. In addition, a SiC substrate with a low dislocation density cannot obtain a sufficient contrast difference, and in the method described in Patent Document 1, the in-plane distribution of basal plane dislocations cannot sometimes be properly evaluated.

[0014] The method described in Patent Document 2 requires classification of defects one by one, and it takes time to evaluate the overall dislocation density.

[0015] The present disclosure has been made in view of the above-mentioned problems, and an object of the present disclosure is to provide a method for evaluating a SiC substrate and a method for producing a SiC substrate, which are capable of simply and quickly measuring the basal plane dislocation density.

[0016] Technical solutions to solve problems

[0017] The inventors of the present invention focused on the overall image of the X-ray topography image instead of focusing on individual defects. It was found that when the overall image of the X-ray topography image is used for deep learning, an analysis that also takes into account the strain of the entire crystal can be performed, and the basal plane dislocation density can be estimated with good accuracy. In addition, by using the overall image of the X-ray topography image instead of focusing on individual defects, the time spent on estimating the basal plane dislocation density can be shortened.

[0018] (1) The evaluation method of the SiC substrate involved in the first technical solution includes: an image acquisition step, obtaining an X-ray morphology image of the entire first surface of the SiC substrate; and an estimation step, based on the learning result of deep learning, estimating the basal plane dislocation density of the SiC substrate from the X-ray morphology image of the entire first surface of the SiC substrate.

[0019] (2) The evaluation method of the SiC substrate involved in the above technical solution may also include an image compression step of compressing the X-ray topography image obtained in the image acquisition step. In the image compression step, the pixel size of the X-ray topography image is set to 0.1 mm. 2The estimating step is performed using the compressed X-ray topography image.

[0020] (3) The evaluation method of the SiC substrate involved in the above technical solution may also include a preliminary step of measuring the X-ray topography images of each of the plurality of SiC substrates and the basal plane dislocation density of each of the plurality of SiC substrates. In the preliminary step, the relationship between the X-ray topography images and the basal plane dislocation density of each SiC substrate is learned by the deep learning.

[0021] (4) In the SiC substrate evaluation method according to the above aspect, the number of SiC substrates measured in the preliminary step may be 100 or more.

[0022] (5) In the SiC substrate evaluation method according to the above-mentioned aspect, an image obtained by inverting or rotating the X-ray topography image acquired in the previous step may be used for the learning.

[0023] (6) In the SiC substrate evaluation method according to the above technical solution, the X-ray topography image may be a diffraction image including diffraction vectors in the <11-20> direction.

[0024] (7) In the SiC substrate evaluation method according to the above technical solution, the X-ray topography image may be a diffraction image including diffraction vectors in the <1-100> direction.

[0025] (8) In the SiC substrate evaluation method according to the above technical solution, the diameter of the SiC substrate may be 145 mm or more.

[0026] (9) In the SiC substrate evaluation method according to the above technical solution, the diameter of the SiC substrate may be greater than or equal to 195 mm.

[0027] (10) In the SiC substrate evaluation method according to the above technical solution, the diameter of the SiC substrate measured in the preliminary step and the diameter of the SiC substrate of which the X-ray topography image is obtained in the image obtaining step may be different.

[0028] (11) In the evaluation method of the SiC substrate according to the above technical solution, the basal plane dislocation density of the SiC substrate may be 1.0×10 4 / cm 2 the following.

[0029] (12) A second invention relates to a method for manufacturing a SiC substrate including the method for evaluating a SiC substrate according to the above invention.

[0030] (13) The evaluation device for a SiC substrate according to the third technical solution comprises an input unit and an output unit. The input unit is configured to receive an X-ray topography image of the entire first surface of the SiC substrate. The output unit is configured to output a basal plane dislocation density of the SiC substrate.

[0031] (14) The SiC substrate evaluation device according to the above aspect may further include a learning unit that learns the correlation between the X-ray topography images of each of the plurality of SiC substrates and the basal plane dislocation density of each of the plurality of SiC substrates.

[0032] (15) In the SiC substrate evaluation device according to the above-mentioned technical solution, the learning unit may be configured to perform learning using deep learning.

[0033] (16) The evaluation device for SiC substrates involved in the above technical solution may also include a memory. The memory is located inside the evaluation device or in a server outside the evaluation device. The memory stores X-ray topography images of the first surface of multiple SiC substrates as a whole and measurement data of basal plane dislocation density of the multiple SiC substrates.

[0034] (17) In the SiC substrate evaluation device according to the above technical solution, the learning unit may include a learning program that learns the correlation between the X-ray topography images of the plurality of SiC substrates and the basal plane dislocation density of the plurality of SiC substrates by deep learning.

[0035] (18) In the SiC substrate evaluation device according to the above-mentioned technical solution, the learning unit may perform learning using an image obtained by inverting or rotating at least one of the X-ray topography images of each of the plurality of SiC substrates.

[0036] (19) In the SiC substrate evaluation device according to the above-mentioned technical solution, the diameter of at least one of the plurality of SiC substrates used for learning by the learning unit may be different from the diameter of the SiC substrate input to the input unit.

[0037] Effects of the Invention

[0038] When the evaluation method of the SiC substrate according to the above-mentioned aspect is used, the basal plane dislocation density can be measured rapidly in a non-destructive manner. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 It is a top view of the SiC substrate according to the present embodiment.

[0040] Figure 2 This is a flowchart of the SiC substrate evaluation method according to the present embodiment.

[0041] Figure 3 The measurement results of Example 1 are shown.

[0042] Figure 4 The measurement results of Example 2 are shown.

[0043] Figure 5 It is a schematic diagram of an evaluation device for a SiC substrate according to the present embodiment.

[0044] Description of symbols

[0045] 10SiC substrate, N positioning groove (notch), S1 preliminary step, S2 image acquisition step, S3 image compression step, S4 estimation step DETAILED DESCRIPTION

[0046] Hereinafter, the evaluation method and manufacturing method of the SiC substrate involved in the present embodiment will be described in detail with appropriate reference to the accompanying drawings. In order to make the features of the present embodiment easier to understand, the accompanying drawings used in the following description may enlarge the characteristic parts for convenience, and the size ratios of the components may be different from the actual ones. The materials, dimensions, etc. illustrated in the following description are examples, and the present disclosure is not limited to those materials, dimensions, etc., and can be implemented by appropriately changing them within the scope of not changing its purpose.

[0047] In this specification, individual orientations are indicated by [], and collective orientations are indicated by <>. Regarding negative indices, "-" (bar) is placed above the numbers in crystallography, but in this specification, a minus sign is placed before the numbers.

[0048] In the evaluation method of the SiC substrate according to the present embodiment, the basal plane dislocation density of the SiC substrate is obtained.

[0049] Figure 1 It is a top view of the SiC substrate 10 according to the present embodiment. Figure 1 This is a view of the SiC substrate viewed from the <0001> direction. The SiC substrate 10 is formed of, for example, n-type SiC. The polytype of the SiC substrate 10 is not particularly limited, and may be any of 2H, 3C, 4H, and 6H. The SiC substrate 10 is, for example, 4H-SiC.

[0050] The SiC substrate 10 has a substantially circular shape in a plan view. The SiC substrate 10 may have a positioning groove N for grasping the direction of the crystal axis. In addition, the SiC substrate 10 may have an orientation flat instead of the positioning groove N.

[0051] The diameter of the SiC substrate 10 is, for example, 145 mm or more, preferably 149 mm or more, more preferably 155 mm or less, and further preferably 151 mm or less. The diameter of the SiC substrate 10 is, for example, 195 mm or more, preferably 199 mm or more, more preferably 205 mm or less, and further preferably 201 mm or less. The SiC substrate 10 may be, for example, a 6-inch substrate or an 8-inch substrate.

[0052] Basal plane dislocation is a dislocation existing in the (0001) plane (= c-plane) which is the basal plane of the SiC substrate 10. The basal plane dislocation density is obtained by dividing the number of basal plane dislocations existing in the SiC substrate 10 by the area of ​​the SiC substrate 10. The SiC substrate 10 can be either a just substrate having no offset angle or an offset substrate having an offset angle. The offset angle is, for example, not less than 0.1 degrees and not more than 8 degrees.

[0053] Figure 2 2 is a flowchart of the evaluation method of the SiC substrate according to the present embodiment. The evaluation method of the SiC substrate according to the present embodiment includes, for example, a preliminary step S1, an image acquisition step S2, an image compression step S3, and an estimation step S4.

[0054] In the preliminary step S1, the X-ray topography images of each of the multiple SiC substrates and the basal plane dislocation density of each of the multiple SiC substrates are measured. The measured data are stored in the memory of the evaluation device. The memory can also be stored in a server located outside the evaluation device. The learning program of the evaluation device uses the stored data for learning. The learning program learns the relationship between the X-ray topography images and the basal plane dislocation density of each SiC substrate through deep learning. The data collected in the preliminary step S1 corresponds to the teacher data for deep learning. In the preliminary step S1, the learning process of deep learning is performed. When the actual SiC substrate is evaluated using the learned evaluation device, it is not necessary to perform the preliminary step S1 during the evaluation.

[0055] In the previous step S1, the relationship between the X-ray topography image of the SiC substrate and the basal plane dislocation density in the SiC substrate of the X-ray topography image is obtained as teacher data. The X-ray topography image of the entire SiC substrate includes not only dislocations but also overall strain information.

[0056] The X-ray topography image is a topography image of the entire surface of the first surface of the SiC substrate. By using the entire image of the SiC substrate instead of the partial image of the defective portion for learning, the classification of the partial image is not required in the estimation process of the estimation step S4, and the time taken for estimating the basal plane dislocation density can be shortened. In addition, the captured X-ray topography image can be compressed to a pixel size of 0.1 mm. 2 The pixel size indicates the size of one pixel. For example, an image obtained by compressing a captured X-ray topography image to 250×250 pixels may be used for learning.

[0057] The X-ray topography image of the SiC substrate measured in the prior step S1 may be a diffraction image including diffraction vectors in the <11-20> direction or a diffraction image including diffraction vectors in the <1-100> direction. The X-ray topography image of the SiC substrate measured in the prior step S1 may be both a diffraction image including diffraction vectors in the <11-20> direction and a diffraction image including diffraction vectors in the <1-100> direction.

[0058] The number of basal plane dislocations contained in the SiC substrate measured in the preliminary step S1 can be obtained from the X-ray topography image, by photoluminescence, by using a mirror electron microscope, or by measuring the etch pits after KOH etching. In order to obtain highly reliable data as teacher data, in the preliminary step S1, it is preferred to evaluate the etch pits after KOH etching and classify the basal plane dislocations.

[0059] In the prior step S1, it is preferred to measure the relationship between the X-ray topography images of more than 100 SiC substrates and the basal plane dislocation density as teacher data. The more the number of teacher data, the higher the estimation accuracy of the basal plane dislocation density in the estimation step S4. In the case where it is not possible to prepare a sufficient number of SiC substrates, the image after reversing or rotating the X-ray topography image of one SiC substrate can also be used as a part of the teacher data for learning. For example, the combination of the X-ray topography image of the first substrate and the basal plane dislocation density of the first substrate can be used as the first data, the combination of the image after reversing the X-ray topography image of the first substrate and the basal plane dislocation density of the first substrate can be used as the second data, and the combination of the image after rotating the X-ray topography image of the first substrate and the basal plane dislocation density of the first substrate can be used as the third data for learning. Here, the use of both the inverted image and the rotated image as part of the teacher data is exemplified, but only one of the inverted image and the rotated image can be used.

[0060] The diameters of the plurality of SiC substrates measured in the prior step S1 may not necessarily be the same. For example, the plurality of SiC substrates measured in the prior step S1 may include both a 6-inch substrate and an 8-inch substrate. In addition, the diameters of the plurality of SiC substrates measured in the prior step S1 may also be the same. For example, the plurality of SiC substrates measured in the prior step S1 may be either a 6-inch substrate or an 8-inch substrate.

[0061] In addition, the position of the SiC substrate in the SiC ingot can also be added to the teacher data. The SiC substrate is obtained by cutting out from the SiC ingot. For the SiC ingot, sometimes the tendency of the basal plane dislocation density to occur is different between the tail side at the beginning of crystal growth and the head side at the end of crystal growth. By adding the data of the position of the SiC ingot from which the SiC substrate was cut out to the teacher data, the estimation accuracy of the basal plane dislocation density in the estimation step S4 can be improved.

[0062] The SiC substrate evaluation apparatus performs the preliminary step S1 to learn the correlation between the X-ray topography image of the entire SiC substrate and the basal plane dislocation density.

[0063] Next, a SiC substrate to be actually evaluated is prepared. It is preferred that the basal plane dislocation density of the SiC substrate to be actually evaluated be 1.0×10 4 / cm 2 The following. The diameter of the SiC substrate that is actually the object of evaluation may be the same as or different from the diameter of the SiC substrate measured in the prior step S1. For example, the diameter of the SiC substrate that is actually the object of evaluation may be larger than the diameter of the SiC substrate measured in the prior step S1. When the diameter of the SiC substrate that is actually the object of evaluation is the same as the diameter of the SiC substrate measured in the prior step S1, the evaluation accuracy is improved.

[0064] In the image acquisition step S2, an X-ray topography image of the entire first surface of the SiC substrate to be actually evaluated is acquired. The first surface of the SiC substrate for which the X-ray topography image is acquired may be a chemical mechanical polishing (CMP) surface or a ground surface before CMP polishing.

[0065] The X-ray topography image of the SiC substrate measured in the image acquisition step S2 may be either a diffraction image including diffraction vectors in the <11-20> direction or a diffraction image including diffraction vectors in the <1-100> direction. The X-ray topography image of the SiC substrate measured in the image acquisition step S2 may be either a diffraction image including diffraction vectors in the <11-20> direction or a diffraction image including diffraction vectors in the <1-100> direction. The X-ray topography image of the SiC substrate measured in the image acquisition step S2 is set as a diffraction image including diffraction vectors in the same direction as the X-ray topography image of the SiC substrate measured in the previous step S1.

[0066] Next, in the image compression step S3, the X-ray topography image obtained in the image acquisition step is compressed. In the image compression step S3, the pixel size of the X-ray topography image is set to 0.1 mm. 2 When the diameter of the SiC substrate measured in the previous step S1 is different from the diameter of the SiC substrate to be actually evaluated, the X-ray topography images obtained by measuring the respective SiC substrates are compressed to the same size (eg, 250×250 pixels).

[0067] The resolution of each X-ray topography image is reduced by performing the image compression step S3. By performing the image compression step S3, the calculation in the estimation step S4 can be performed based on the low-resolution image, and the calculation load of the evaluation device is reduced. In the case where the calculation processing capacity of the evaluation device is high enough, the image compression step S3 may not be performed.

[0068] Next, an estimation step S4 is performed. In the estimation step S4, the basal plane dislocation density of the SiC substrate to be evaluated is estimated from the X-ray topography image of the entire first surface of the SiC substrate to be evaluated based on the learning result of the deep learning.

[0069] The X-ray topography image used in the image analysis of the estimation step S4 is the X-ray topography image acquired in the image acquisition step S2. When the image compression step S3 is performed, the X-ray topography image compressed in the image compression step S3 is used for the image analysis in the estimation step S4.

[0070] The evaluation device that has been learned by deep learning in the previous step S1 takes the X-ray topography image of the entire first surface of the SiC substrate as input and outputs the basal plane dislocation density of the SiC substrate. The correlation between the X-ray topography image and the basal plane dislocation density is learned by deep learning, so the evaluation device can estimate the basal plane dislocation density of the SiC substrate to be evaluated with high accuracy.

[0071] As described above, the evaluation method of the SiC substrate involved in the present embodiment uses the topographic image of the entire surface of the first surface of the SiC substrate to estimate the basal plane dislocation density of the SiC substrate. When the defects are classified one by one to obtain the basal plane dislocation density, it is necessary to classify the defects one by one, and the measurement of the basal plane dislocation density will take time. In contrast, the evaluation method of the SiC substrate involved in the present embodiment does not classify the defects one by one, but can obtain the basal plane dislocation density of the SiC substrate from the overall image, so that the time spent on the evaluation of the basal plane dislocation density can be shortened.

[0072] In addition, the topography image of the entire first surface of the SiC substrate includes information on defects and strains of the SiC substrate. The evaluation method of the SiC substrate involved in the present embodiment is based on the topography image including information on defects as microscopic information in the SiC substrate and information on strains as macroscopic information in the SiC substrate for learning, so the estimation accuracy of the basal plane dislocation density is high.

[0073] In addition, the evaluation method of the SiC substrate involved in the present embodiment can reduce the computational load of the evaluation device for estimating the basal plane dislocation density by performing the image compression step S3. When it is desired to estimate the basal plane dislocation density with good accuracy, it is generally necessary to increase the resolution of the image, but in the present embodiment, it was found that the basal plane dislocation density can be estimated with high accuracy even if the resolution of the image is reduced.

[0074] In addition, by specifying the pixel size of the image in the image compression step S3, even if the diameter of the SiC substrate used in the measurement of the prior step S1 is different from the diameter of the SiC substrate of the evaluation object, the basal plane dislocation density of the SiC substrate of the evaluation object can be estimated. For example, it is difficult to prepare an 8-inch substrate, and it is difficult to prepare many samples in the prior step S1. Even in this case, the prior step S1 can be performed with a 6-inch substrate, and the basal plane dislocation density of the 8-inch substrate can be estimated based on the learning results of the 6-inch substrate.

[0075] In addition, the evaluation method of the SiC substrate involved in the present embodiment can be incorporated into a step of the manufacturing method of the SiC substrate and the SiC epitaxial wafer. The evaluation method of the SiC substrate involved in the present embodiment can easily and quickly evaluate the basal plane dislocation density of the SiC substrate, so that defective products with high basal plane dislocation density can be removed in the middle of the manufacturing process, thereby improving the yield rate of the SiC substrate and the SiC epitaxial wafer. The evaluation method of the SiC substrate involved in the present embodiment can also be applied to the evaluation of the basal plane dislocation density as low as 1.0×10 4 / cm 2 The following situation.

[0076] Figure 5 2 is a schematic diagram of an evaluation device 20 for a SiC substrate according to the present embodiment. The evaluation device 20 for a SiC substrate includes, for example, an input unit 21, an output unit 22, a learning unit 23, and a memory 24. The evaluation device 20 performs learning and estimation.

[0077] The learning of the evaluation device 20 is performed in the learning unit 23 and the memory 24. The memory 24 stores the X-ray topography images of each of the plurality of SiC substrates and the measurement data of the basal plane dislocation density of each of the plurality of SiC substrates. The data stored in the memory 24 is the data obtained by the measurement in the previous step S1. The memory 24 stores the data in the previous step S1. Figure 5 In FIG. 1 , an example is shown in which the memory 24 is built into the evaluation device 20 . However, the memory 24 may be located in a server outside the evaluation device 20 .

[0078] The learning unit 23 learns the correlation between the X-ray topography images of each of the multiple SiC substrates and the basal plane dislocation density of each of the multiple SiC substrates based on the data stored in the memory 24. A learning program is stored in the learning unit 23. The learning unit 23 performs learning by operating the learning program. The learning is performed, for example, by deep learning. The data used for learning are the relationship between the X-ray topography images of the SiC substrates and the basal plane dislocation density of the SiC substrates. The more data used for learning, the higher the accuracy of learning. In the case where a sufficient number of SiC substrates cannot be prepared, at least one image after inverting or rotating the X-ray topography images of each of the multiple SiC substrates can also be used for learning.

[0079] The evaluation device 20 performs inference in the input unit 21, the learning unit 23, and the output unit 22. The input unit 21 is configured to receive an X-ray topography image of the entire first surface of the SiC substrate. Here, the input unit 21 receives an X-ray topography image of the SiC substrate to be evaluated whose basal plane dislocation density is unknown.

[0080] The learning unit 23 estimates the basal plane dislocation density of the SiC substrate to be evaluated from the input X-ray topography image based on the result of learning during learning. The learning unit 23 does not perform learning during estimation, but estimates based on the result of learning.

[0081] The output unit 23 is configured to output the basal plane dislocation density of the SiC substrate to be evaluated. The evaluation device 20 estimates the basal plane dislocation density of the SiC substrate to be evaluated from the X-ray topography image based on the learning result.

[0082] The X-ray topography image input to the input unit 21 and the basal plane dislocation density output from the output unit 23 may be stored as one of the teacher data in the memory 24. By including the inference result in the teacher data, the amount of teacher data increases, and the learning accuracy of the evaluation device 20 can be improved.

[0083] As mentioned above, although the preferred embodiment of the present invention is described in detail, the present invention is not limited to the specific embodiment, and various modifications and changes can be made within the scope of the gist of the present invention described in the claims.

[0084] [Example]

[0085] First, 800 X-ray topography images of a SiC substrate with a diameter of 150 mm were prepared. In addition, a portion of these images were inverted and rotated to prepare 4,000 X-ray topography images. Each of the 4,000 X-ray topography images was compressed to 250×250 pixels.

[0086] Next, the evaluation device was used to learn the correlation between 4,000 X-ray morphology images and the number of basal plane dislocations of the SiC substrate in each X-ray morphology image through deep learning. The program language used for learning used a Python-based language. For the program, the GUI application was made with tkinter, and the deep learning module was made with Tensorflow-Keras. In addition, the information about the position of the SiC ingot from which the SiC substrate was cut was also added to the teacher data.

[0087] In Example 1, the basal plane dislocation density of the SiC substrate at the tail side (initial stage of crystal growth) of the SiC ingot was obtained. For the basal plane dislocation density, both the estimated value and the measured value using the evaluation method of the SiC substrate involved in this embodiment were obtained. The measured value is the value of the number of basal plane dislocations obtained by classifying the etch pits after KOH etching. For the estimated value, an X-ray topography image of the SiC substrate to be measured was obtained, and the estimated value was obtained using an image compressed to 250×250 pixels. Figure 3 The evaluation results of Example 1 are shown. Correlation coefficient R of Example 1 2 The value was 0.6886, and a high correlation was confirmed between the measured value and the estimated value.

[0088] In Example 2, the basal plane dislocation density of the SiC substrate on the head side (latter half of crystal growth) of the SiC ingot was obtained. For the basal plane dislocation density, both the estimated value and the measured value using the evaluation method of the SiC substrate involved in this embodiment were obtained. The measured value is the value of the number of basal plane dislocations obtained by classifying the etch pits after KOH etching. For the estimated value, an X-ray topography image of the SiC substrate to be measured was obtained, and the estimated value was obtained using an image compressed to 250×250 pixels. Figure 4 The evaluation results of Example 2 are shown. The correlation coefficient R of Example 2 2 The value was 0.7254, and a high correlation was confirmed between the measured value and the estimated value.

[0089] As shown in Examples 1 and 2, when the evaluation method of the SiC substrate according to the present embodiment is used, the basal plane dislocation density of the SiC substrate can be easily estimated. In addition, even when a compressed image is used, the basal plane dislocation density can be obtained with sufficient accuracy.

Claims

1. A method for evaluating a SiC substrate, comprising: An image acquisition step of acquiring an X-ray topography image of the entire first surface of the SiC substrate; and The estimating step estimates the basal plane dislocation density of the SiC substrate from an X-ray topography image of the entire first surface of the SiC substrate based on a learning result of deep learning.

2. The method for evaluating a SiC substrate according to claim 1, The method further comprises an image compression step for compressing the X-ray topography image obtained in the image acquisition step. In the image compression step, the pixel size of the X-ray topography image is set to 0.1 mm. 2 The above dimensions, The estimating step is performed using the compressed X-ray topography image.

3. The method for evaluating a SiC substrate according to claim 1, The method further includes the step of measuring the X-ray topography images of each of the plurality of SiC substrates and the basal plane dislocation density of each of the plurality of SiC substrates. In the preliminary step, the relationship between the X-ray topography image and the basal plane dislocation density of each SiC substrate is learned by the deep learning.

4. The method for evaluating a SiC substrate according to claim 3, The number of SiC substrates measured in the preliminary step is 100 or more.

5. The method for evaluating a SiC substrate according to claim 3, The X-ray topography image obtained in the previous step is inverted or rotated and used for the learning.

6. The method for evaluating a SiC substrate according to claim 1, The X-ray topography image is a diffraction image including diffraction vectors in the <11-20> direction.

7. The method for evaluating a SiC substrate according to claim 1, The X-ray topography image is a diffraction image including diffraction vectors in the direction of <1-100>.

8. The method for evaluating a SiC substrate according to claim 1, The SiC substrate has a diameter of 145 mm or more.

9. The method for evaluating a SiC substrate according to claim 1, The SiC substrate has a diameter of 195 mm or more.

10. The method for evaluating a SiC substrate according to claim 3, The diameter of the SiC substrate measured in the preliminary step is different from the diameter of the SiC substrate of which the X-ray topography image is acquired in the image acquisition step.

11. The method for evaluating a SiC substrate according to claim 1, The basal plane dislocation density of the SiC substrate is 1.0×10 4 / cm 2 the following.

12. A method for manufacturing a SiC substrate, A method for evaluating a SiC substrate according to any one of claims 1 to 11.

13. An evaluation device for a SiC substrate, It has an input part and an output part, The input unit is configured to receive an X-ray topography image of the entire first surface of the SiC substrate, The output unit is configured to output a basal plane dislocation density of the SiC substrate.

14. The SiC substrate evaluation device according to claim 13, There is also a learning department. The learning unit learns the correlation between the X-ray topography images of each of the plurality of SiC substrates and the basal plane dislocation density of each of the plurality of SiC substrates.

15. The SiC substrate evaluation device according to claim 14, The learning unit is configured to perform learning using deep learning.

16. The SiC substrate evaluation device according to claim 13, It also has memory. The storage device is located inside the evaluation device or in a server outside the evaluation device. The memory stores X-ray topography images of the entire first surfaces of a plurality of SiC substrates and measurement data of basal plane dislocation densities of the plurality of SiC substrates.

17. The SiC substrate evaluation device according to claim 14, The learning unit has a learning program, The learning program learns the correlation between the X-ray topography images of each of the plurality of SiC substrates and the basal plane dislocation density of each of the plurality of SiC substrates through deep learning.

18. The SiC substrate evaluation device according to claim 14, The learning unit performs learning using an image obtained by inverting or rotating at least one of the X-ray topography images of each of the plurality of SiC substrates.

19. The SiC substrate evaluation device according to claim 13, A diameter of at least one of the plurality of SiC substrates used for learning by the learning section is different from a diameter of the SiC substrate input to the input section.

Images