Defect inspection device

Abstract

To provide a defect inspection device and a defect inspection method capable of discriminating BPD in a SiC epitaxial wafer.SOLUTION: A defect inspection device 1 according to this embodiment includes a radiation optical system 10 that radiates excitation light EL onto a sample 50 including a SiC substrate 51, a buffer layer 52 formed on the SiC substrate 51, and a drift layer 53 formed on the buffer layer 52, a filter section 20 that controls a wavelength band that transmits photoluminescence light PL generated from the sample 50, a detection optical system 30 that detects the photoluminescence light PL that has transmitted through the filter section 20, and an image processing section 40 that forms an image from the detected photoluminescence light PL and discriminates a defect captured in the formed image. The image processing section discriminates the defect on the basis of whether the length of the defect is L1=(D1+D2) / tan θ or L2=D2 / tan θ.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] This disclosure relates to a defect inspection apparatus and a defect inspection method. [Background technology]

[0002] Silicon carbide (SiC) possesses excellent physical and thermal properties, making it useful for manufacturing high-voltage, low-loss semiconductor devices. A SiC epitaxial wafer is a wafer in which an epitaxial layer containing SiC is formed on a SiC substrate. In the manufacturing process of semiconductor devices using SiC epitaxial wafers, detecting defects present in the epitaxial layer and classifying the detected defects are extremely important in order to improve manufacturing yield. In particular, basal plane dislocations (BPDs) are killer defects that have a fatal adverse effect on the performance of semiconductor devices when manufacturing semiconductor devices on SiC epitaxial wafers. Therefore, there is a strong desire to distinguish BPDs from other defects and detect them accordingly.

[0003] In SiC epitaxial wafers, where a SiC-containing drift layer is formed on a SiC substrate, a structure is sometimes adopted in which a buffer layer doped with a high concentration of impurities is sandwiched between the drift layer and the SiC substrate to prevent the generation of BPDs (Block Defects) in the drift layer. With such a structure, BPDs in the buffer layer can be converted into threading edge dislocations (TEDs) at the interface between the buffer layer and the drift layer. Killer defects are generally BPDs in the drift layer. Therefore, by converting BPDs in the buffer layer into TEDs in the drift layer, the generation of killer defects in the drift layer can be suppressed.

[0004] However, considering long-term reliability, it is desirable to detect BPDs that exist only in the buffer layer converted to TED at the interface between the buffer layer and the drift layer. [Prior art documents] [Patent Documents]

[0005] [Patent Document 1] Japanese Patent Publication No. 2015-119056 [Patent Document 2] Patent No. 6999212 [Patent Document 3] Japanese Patent Publication No. 2019-099438 [Overview of the Initiative] [Problems that the invention aims to solve]

[0006] Patent Document 1 describes a defect inspection apparatus that irradiates a SiC epitaxial wafer with ultraviolet light and detects defects from the resulting photoluminescent light. The defect inspection apparatus in Patent Document 1 detects BPD in the drift layer from an image in the near-infrared wavelength band of the photoluminescent light. However, the defect inspection apparatus in Patent Document 1 has difficulty detecting BPD in the buffer layer.

[0007] Patent Document 2 describes a method in which BPDs in the drift layer and buffer layer are converted into Shockley-type stacking faults (SSFs), and the BPDs are detected by photoluminescence light from the converted SSFs. While the method in Patent Document 2 can detect BPDs in the buffer layer, it requires the conversion of BPDs into SSFs, making the inspection process very complicated.

[0008] Patent Document 3 describes detecting BPD in a high-concentration epitaxial layer by photoluminescence light that transmits through a wavelength band of 430 nm or less. The defect inspection apparatus of Patent Document 3 can detect BPD in a high-concentration epitaxial layer as a linear defect that is darker than the surroundings. However, nothing is described about the relationship between the detected BPD and TED in the drift layer.

[0009] In a SiC epitaxial wafer, it is desired to determine whether BPD extends from the buffer layer to the drift layer or exists only in the buffer layer.

[0010] The present disclosure has been made in view of such problems, and provides a defect inspection apparatus and a defect inspection method capable of discriminating BPD in a SiC epitaxial wafer.

Means for Solving the Problems

[0011] A defect inspection apparatus according to an aspect of the present embodiment includes an irradiation optical system that irradiates a sample including a silicon carbide substrate, a buffer layer formed on the silicon carbide substrate, and a drift layer formed on the buffer layer with excitation light, a filter unit that controls a wavelength band that transmits photoluminescence light generated from the sample, a detection optical system that detects the photoluminescence light that has passed through the filter unit, and an image processing unit that forms an image from the detected photoluminescence light and discriminates defects imaged in the formed image. The filter unit transmits the wavelength band including 420 nm or more and 430 nm or less in the photoluminescence light. When the thickness of the drift layer is D1, the thickness of the buffer layer is D2, the offset angle is θ, and the first length and the second length are L1 and L2 respectively shown below, L1 = (D1 + D2) / tanθ L2 = D2 / tanθ When the length of the defect belongs to a predetermined first range including the first length, the image processing unit determines the defect as the defect including a basal plane dislocation connecting the buffer layer and the drift layer. When the length of the defect is a predetermined second range including the second length and belongs to the second range different from the first range, the image processing unit determines the defect as the defect including the basal plane dislocation in the buffer layer and the edge dislocation in the drift layer.

[0012] In the above defect inspection apparatus, the filter unit changes the wavelength band to be transmitted from the wavelength band including 420 nm or more and 430 nm or less to the wavelength band including near-infrared light. The detection optical system detects the near-infrared light transmitted through the changed filter unit. The image processing unit forms a near-infrared image from the detected near-infrared light. When determining the defect captured in the formed near-infrared image, when the third length is L3 shown below, L3 = D1 / tanθ When the length of the defect belongs to a predetermined third range including the third length and belongs to the third range different from the first range and the second range, the defect may be determined as the defect including the basal plane dislocation in the drift layer.

[0013] In the above defect inspection apparatus, the image processing unit may calculate the accuracy of the determined defect by comparing the number of defects having the length belonging to the first range with the number of defects having the length belonging to the third range.

[0014] In the above defect inspection apparatus, the image processing unit may determine the defect captured in the image by using an algorithm that has been previously machine-learned with a plurality of defects belonging to the first range and a plurality of defects belonging to the second range as learning data.

[0015] In the above defect inspection apparatus, the impurity concentration of the drift layer is 1×10 15 cm -3 or more and 5×10 17 cm -3The impurity concentration in the buffer layer is less than 5 × 10 17 cm -3 The above 1 x 10 19 cm -3 The following is also acceptable.

[0016] A defect inspection method according to one aspect of this embodiment comprises the steps of: irradiating a sample including a silicon carbide substrate, a buffer layer formed on the silicon carbide substrate, and a drift layer formed on the buffer layer with excitation light; transmitting photoluminescent light generated from the sample through a filter unit that controls the wavelength band to be transmitted; detecting the photoluminescent light transmitted through the filter unit; forming an image from the detected photoluminescent light; and determining a defect captured in the formed image, wherein in the step of transmitting the photoluminescent light, the filter unit transmits the wavelength band including 420 nm to 430 nm in the photoluminescent light, and in the step of determining the defect, when the thickness of the drift layer is D1, the thickness of the buffer layer is D2, the offset angle is θ, and the first length and second length are L1 and L2 as shown below, L1 = (D1 + D2) / tanθ L2 = D2 / tanθ If the length of the defect falls within a predetermined first range including the first length, the defect is determined to include a basal plane dislocation connecting the buffer layer to the drift layer. If the length of the defect falls within a predetermined second range including the second length and is different from the first range, the defect is determined to include a basal plane dislocation in the buffer layer and an edge dislocation in the drift layer.

[0017] In the above defect inspection method, after the step of discriminating the defect, a step of changing the wavelength band to be transmitted through the filter unit from the wavelength band including 420 nm or more and 430 nm or less to the wavelength band including near-infrared light, a step of detecting the near-infrared light transmitted through the changed filter unit, a step of forming a near-infrared image from the detected near-infrared light, and a step of discriminating the defect imaged in the formed near-infrared image are further provided. In the step of discriminating the defect imaged in the near-infrared image, when the third length is represented as L3 below, L3 = D1 / tan θ when the length of the defect imaged in the near-infrared image is within a predetermined third range including the third length and different from the first range and the second range, the defect may be discriminated as the defect including the basal plane dislocation in the drift layer.

[0018] The above defect inspection method may further include a step of calculating the accuracy of the discriminated defect by comparing the number of defects having the length belonging to the first range with the number of defects having the length belonging to the third range.

[0019] In the above defect inspection method, the defect imaged in the image may be discriminated by using an algorithm that has been previously machine-learned with a plurality of defects belonging to the first range and a plurality of defects belonging to the second range as learning data.

[0020] In the above defect inspection method, the impurity concentration of the drift layer is 1 × 10 15 cm -3 or more and 5 × 10 17 cm -3 less than, and the impurity concentration of the buffer layer may be 5 × 10 17 cm -3 or more and 1 × 10 19 cm -3 or less.

Advantages of the Invention

[0021] This disclosure provides a defect inspection apparatus and a defect inspection method that can determine the BPD in a SiC epitaxial wafer. [Brief explanation of the drawing]

[0022] [Figure 1] This is a diagram illustrating a defect inspection device according to Embodiment 1. [Figure 2] This is a cross-sectional view illustrating a sample to be inspected in the defect inspection apparatus according to Embodiment 1. [Figure 3] This is a schematic diagram illustrating an image formed by the image processing unit in the defect inspection apparatus according to Embodiment 1. [Figure 4] This is a flowchart illustrating a defect inspection method according to Embodiment 1. [Figure 5] This is a diagram illustrating a defect inspection device according to Embodiment 2. [Figure 6] This is a schematic diagram illustrating an image formed by the image processing unit in the defect inspection apparatus according to Embodiment 2. [Figure 7] This figure illustrates how the BPD (Block Point Diagram) appears in defects imaged by the defect inspection apparatus according to Embodiment 1 and the defect inspection apparatus according to Embodiment 2. [Figure 8] This is a flowchart illustrating a defect inspection method according to Embodiment 2. [Modes for carrying out the invention]

[0023] Embodiments of the present disclosure will be described below with reference to the drawings. The following description illustrates preferred embodiments of the present disclosure and does not limit the scope of the present disclosure to the following embodiments. In the following description, the same reference numerals indicate substantially the same thing.

[0024] (Embodiment 1) A defect inspection apparatus and defect inspection method according to Embodiment 1 will be described. Figure 1 is a configuration diagram illustrating a defect inspection apparatus 1 according to Embodiment 1. As shown in Figure 1, the defect inspection apparatus 1 according to this embodiment includes an irradiation optical system 10, a filter unit 20, a detection optical system 30, and an image processing unit 40.

[0025] The irradiation optical system 10 irradiates the sample 50 placed on the stage 60 with excitation light EL. The irradiation optical system 10 may include a light source 11. The irradiation optical system 10 may also include optical components such as filters, mirrors, and lenses. The irradiation optical system 10 irradiates the SiC epitaxial wafer, which will be the sample 50, with excitation light EL. The excitation light EL may include, for example, ultraviolet light that generates photoluminescence light PL from the SiC epitaxial wafer. That is, the excitation light EL has an energy greater than the energy of the band gap of SiC.

[0026] For example, the excitation light EL may be laser light with a wavelength of 320 nm generated from a He-Cd laser, or fourth harmonic laser light with a wavelength of 266 nm generated from a YAG laser. Alternatively, the excitation light EL may be ultraviolet light generated from a mercury xenon lamp. The irradiation optical system 10 may transmit a predetermined wavelength band of the excitation light EL using a filter. For example, if the light source 11 is a mercury xenon lamp, the excitation light EL may include a beam with a wavelength of 313 nm that has been transmitted through the filter. The irradiation optical system 10 may have optical elements such as mirrors and lenses in the optical path from the excitation light EL emitted from the light source 11 to the sample 50. For example, the inspection surface of the sample 50 may be scanned with the excitation light EL by using a galvanometer mirror or the like. Alternatively, the inspection surface of the sample 50 may be scanned with the excitation light EL by moving the stage 60.

[0027] Here, for the sake of explaining the defect inspection device 1, we introduce an XYZ Cartesian coordinate system. The direction perpendicular to the upper surface of the stage 60 is defined as the Z-axis direction, and the planes parallel to the upper surface of the stage 60 are defined as the XY planes. For example, the +Z axis direction is defined as upward, and the -Z axis direction is defined as downward. Note that upward and downward are directions for the sake of explaining the defect inspection device 1 and do not indicate the actual direction in which the defect inspection device 1 will be positioned.

[0028] The filter unit 20 controls the wavelength band through which the photoluminescent light PL generated from the sample 50 is transmitted. Specifically, for example, the filter unit 20 may have a filter 21 that transmits a wavelength band including 420 nm to 430 nm in the photoluminescent light PP. The filter 21 transmits a wavelength band including 420 nm to 430 nm in the photoluminescent light P1. For example, the filter 21 may transmit a wavelength band including a central wavelength of 425 ± 13 nm, or a wavelength band including a central wavelength of 425 ± 5 nm.

[0029] Furthermore, the filter unit 20 may include a filter 22 that transmits wavelengths including near-infrared light. For example, filter 22 transmits wavelengths including 700 nm to 1100 nm. The filter unit 20 may select either filter 21 or filter 22 depending on the wavelength band to be detected.

[0030] The detection optical system 30 includes a detector 31. The detection optical system 30 detects the photoluminescent light P1 that has passed through the filter section 20. In addition to the detector 31, the detection optical system 30 may also include optical elements such as lenses and mirrors that guide the photoluminescent light P1 that has passed through the filter section 20 to the detector 31.

[0031] The detector 31 may include an image sensor in which solid-state image sensors such as CCDs (Charged-coupled devices) and CMOSs ​​(Complementary metal-oxide-semiconductors) are arranged in a two-dimensional array, or it may include a line sensor in which multiple image sensors are arranged in a line.

[0032] The image processing unit 40 forms an image from the photoluminescent light PL detected by the detection optical system 30. The image processing unit 40 then identifies defects captured in the formed image. The image processing unit 40 is, for example, a computer having a CPU, memory, storage, etc. The image processing unit 40 may also have input / output means including a display for displaying the image.

[0033] Figure 2 is a cross-sectional view illustrating a sample 50 to be inspected in a defect inspection apparatus 1 according to Embodiment 1. As shown in Figure 2, the sample 50 includes a SiC substrate 51, a buffer layer 52, and a drift layer 53. The sample 50 has a structure in which the SiC substrate 51, buffer layer 52, and drift layer 53 are stacked. The buffer layer 52 is formed on the SiC substrate 51, and the drift layer 53 is formed on the buffer layer 52. Therefore, the buffer layer 52 is positioned between the SiC substrate 51 and the drift layer 53.

[0034] The SiC substrate 51 serves as a base for the fabrication of thin films of the buffer layer 52 and drift layer 53 by epitaxial growth. The SiC substrate 51 is a substrate containing a SiC single crystal (4H-SiC). The method for fabricating the SiC substrate 51 is not particularly limited. For example, the SiC substrate 51 can be obtained by slicing a SiC ingot obtained by sublimation or the like. The BPD may be present in the SiC substrate 51 along the (0001) plane (c plane).

[0035] The buffer layer 52 and the drift layer 53 are thin films containing a SiC single crystal (4H-SiC) deposited on the SiC substrate 51 by epitaxial growth. Specifically, the epitaxial layer including the buffer layer 52 and the drift layer 53 may be formed on the SiC substrate 51, for example, on a growth surface having an offset angle from the (0001) plane in the <11-20> direction. The epitaxial layer may be 4H-SiC grown on the SiC substrate 51 by step flow growth (lateral growth from atomic steps).

[0036] The buffer layer 52 is a layer formed by epitaxially growing an epitaxial layer containing a high concentration of impurities on the SiC substrate 51. The impurities used for doping may include nitrogen, boron, titanium, vanadium, aluminum, gallium, phosphorus, etc. The buffer layer 52 functions, for example, in a power device to suppress the arrival of minority carriers to the drift layer 53. By stacking the buffer layer 52, when a forward current is passed through a bipolar device having a BPD, it is possible to prevent the minority carriers from reaching the BPD present on the SiC substrate 51.

[0037] The drift layer 53 is a layer formed by epitaxially growing an epitaxial layer containing a low concentration of impurities on the buffer layer 52. The impurities used for doping may be nitrogen, boron, titanium, vanadium, aluminum, gallium, phosphorus, etc., similar to those used in the buffer layer 52. The drift layer 53 is doped with impurities at a lower concentration than the buffer layer 52 and functions, for example, as a drift layer in a SiC-MOSFET.

[0038] The impurity concentration in buffer layer 52 is 1 × 10⁻⁶. 18 cm -3 The above is preferable. For example, the impurity concentration of the buffer layer 52 is 5 × 10 17 cm -3 The above 1 x 10 19 cm -3 The following applies: The impurity concentration in the drift layer is 1 × 10⁻⁶. 16 cm -3 A certain degree is preferable. For example, the impurity concentration in the drift layer should be 1 × 10⁻⁶. 15 cm -3 The above 5 x 10 17 cm -3 It is less than.

[0039] The thickness D2 of the buffer layer 52 is, for example, 1 μm. However, the thickness D2 of the buffer layer 52 is not limited to 1 μm. The thickness D1 of the drift layer 53 is, for example, 10 μm. However, the thickness D1 of the drift layer 53 is not limited to 10 μm. It is desirable that the thickness D1 of the drift layer 53 be different from the thickness D2 of the buffer layer 52. For example, it is desirable that the thickness D1 of the drift layer 53 be greater than the thickness D2 of the buffer layer 52, but this is not a requirement.

[0040] In sample 50, a portion of the BPD is carried over to the drift layer 53. That is, the defect K1 formed in sample 50 includes, for example, a BPD that extends from the buffer layer 52 to the drift layer 53. Specifically, defect K1 includes a BPD that extends from the SiC substrate 51 through the buffer layer 52 to the drift layer 53. In such a defect K1, the angle between the direction in which the BPD extends and the upper surface (called the base plane) of the SiC substrate 51 is the offset angle θ. That is, the BPD extends in the direction of the offset angle θ with respect to the base plane. The BPD carried over to the drift layer 53 emits photoluminescent light PL with a wavelength of 710 nm due to the recombination of carriers generated by the irradiation of the sample 50 with excitation light EL.

[0041] On the other hand, BPD may be converted to TED at the interface 54 between the buffer layer 52 and the drift layer 53. In this case, the defect K2 formed on the sample 50 includes BPD in the buffer layer 52 and TED in the drift layer 53. Specifically, defect K2 includes BPD in the buffer layer 52 and TED converted from BPD at the interface 54 between the buffer layer 52 and the drift layer 53. In defect K2, the angle between the direction in which the BPD in the buffer layer 52 extends and the base plane is the offset angle θ. The BPD in the buffer layer 102 before conversion at interface 54 does not emit photoluminescent light PL even when the sample 50 is irradiated with excitation light EL because the carrier lifetime in the buffer layer 52 is short.

[0042] Figure 3 is a schematic diagram illustrating an image formed by the image processing unit 40 in the defect inspection apparatus 1 according to Embodiment 1. Figure 3 is an image captured from the +Z axis direction, formed from the photoluminescence light PL of the sample 50. For example, Figure 3 shows one field of view during inspection, which is 1.75 mm square or 1.5 mm square.

[0043] As shown in Figure 3, when an image is formed from photoluminescent light (PL) transmitted in a wavelength band including 420 nm to 430 nm, the image shows darker (blacker) and longer linear defects and shorter linear defects compared to the surrounding area. For example, the black linear defects extend in the Y-axis direction. In the Y-axis direction, the length of the long linear defects is called the first length L1, and the length of the short linear defects is called the second length L2.

[0044] Here, let D1 be the thickness of the drift layer 53, D2 be the thickness of the buffer layer 52, θ be the offset angle, and L1 and L2 be the first and second lengths, respectively, as shown in equations (1) and (2) below. It is preferable that the thickness D1 of the drift layer 53 and the thickness D2 of the buffer layer 52 be different, but they may be the same thickness.

[0045] L1 = (D1 + D2) / tanθ (1) L2 = D2 / tanθ (2)

[0046] In this case, a defect with a first length L1 corresponds to a defect K1 that includes a BPD connecting the buffer layer 52 to the drift layer 53. A defect with a second length L2 corresponds to a defect K2 that includes a BPD in the buffer layer 52 and a TED in the drift layer 53. Therefore, when the length of a defect captured in an image is the first length L1, the image processing unit 40 determines that the defect is a defect K1 that includes a BPD connecting the buffer layer 52 to the drift layer 53. On the other hand, when the length of a defect captured in an image is the second length, the image processing unit 40 determines that the defect is a defect K2 that includes a BPD in the buffer layer 52 and a TED in the drift layer 53.

[0047] The length of defect K1 is not limited to the first length L1, but may also include the first length L1 which includes unavoidable errors such as variations in thickness that occur when forming the buffer layer 52 and the drift layer 53. Similarly, the length of defect K2 is not limited to the second length L2, but may also include the second length L2 which includes unavoidable errors. The unavoidable errors depend on conditions such as the thickness of the buffer layer 52 and the drift layer 53 when they are formed, but may be, for example, ±10% or ±5% of the first length L1 and the second length L2.

[0048] Therefore, the image processing unit 40 determines that a defect captured in an image is a defect K1 including a BPD connecting the buffer layer 52 to the drift layer 53 if the length of the defect is within a predetermined first range including a first length L1. The image processing unit 40 also determines that a defect is a defect K2 including a BPD in the buffer layer 52 and a TED in the drift layer 53 if the length of the defect captured in an image is within a predetermined second range including a second length L2, and is different from the first range. The first range is the range of the first length L1 including unavoidable errors, for example, from (first length - unavoidable errors) to (first length L1 + unavoidable errors). The second range is the range of the second length L2 including unavoidable errors, for example, from (second length L2 - unavoidable errors) to (second length L2 + unavoidable errors).

[0049] The length of a defect may be calculated from the number of pixels in the region detected as a defect by the image processing unit 40. Alternatively, the image processing unit 40 may identify defects captured in an image by using an algorithm that has been pre-trained using a plurality of defects K1 belonging to the first range and a plurality of defects K2 belonging to the second range as training data.

[0050] Next, the defect inspection method of this embodiment will be described. Figure 4 is a flowchart illustrating the defect inspection method according to Embodiment 1. As shown in Figure 4, the defect inspection method of this embodiment includes the steps of: irradiating the sample 50 with excitation light EL in step S11; transmitting photoluminescence light PL through the filter unit 20 in step S12; detecting the photoluminescence light PL in step S13; forming an image from the photoluminescence light PL in step S14; and identifying defects captured in the image in step S15.

[0051] First, as shown in step S11, the sample 50 is irradiated with excitation light EL. The sample 50 includes a SiC substrate 51, a buffer layer 52 formed on the SiC substrate 51, and a drift layer 53 formed on the buffer layer 52. The impurity concentration of the drift layer is, for example, 1 × 10⁻⁶. 15 cm -3 The above 5 x 10 17 cm -3 The impurity concentration in the buffer layer is less than, for example, 5 × 10 17 cm -3 The above 1 x 10 19 cm -3 The following applies. Therefore, the impurity concentration in the buffer layer 52 is greater than the impurity concentration in the drift layer 53. The excitation light EL is irradiated by the irradiation optical system 10. The irradiation optical system 10 may scan the sample 50 with the excitation light EL. The sample 50 may also be scanned with the excitation light EL by moving the stage 60. The sample 50 irradiated with the excitation light EL generates photoluminescence light PL.

[0052] Next, as shown in step S12, the photoluminescent light PL generated from the sample 50 is transmitted through the filter section 20. Specifically, the photoluminescent light PL generated from the sample 50 is transmitted through the filter section 20, which controls the wavelength band to be transmitted. For example, the filter section 20 transmits a wavelength band including 420 nm to 430 nm in the photoluminescent light PL through the filter 21.

[0053] Next, as shown in step S13, the photoluminescent light P1 that has passed through the filter section 20 is detected. For example, the detector 31 in the detection optical system 30 detects the photoluminescent light PL that has passed through the filter 21.

[0054] Next, as shown in step S14, an image is formed from the detected photoluminescent light PL. Specifically, the image processing unit 40 forms an image from the photoluminescent light PL detected by the detection optical system 30, which includes the detector 31.

[0055] Next, as shown in step S15, defects captured in the formed image are identified. For example, if the length of a defect captured in the image belongs to a predetermined first range including a first length L1, the image processing unit 40 may identify the defect as defect K1, which includes a BPD connecting the buffer layer 52 to the drift layer 53. Alternatively, if the length of a defect captured in the image belongs to a predetermined second range including a second length L2, which is different from the first range, the image processing unit 40 may identify the defect as defect K2, which includes a BPD in the buffer layer 52 and a TED in the drift layer 53. In this way, defects in the SiC epitaxial wafer can be inspected.

[0056] Next, the effects of this embodiment will be described. The defect inspection apparatus 1 of this embodiment forms an image from photoluminescent light P1 having a wavelength band including 420 nm to 430 nm that has been transmitted through the filter 21. Therefore, the image processing unit 40 can detect BPDs formed in the epitaxial layer as black linear defects in the formed image. Furthermore, the image processing unit 40 can identify the defect based on the length of the defect captured in the image. This makes it possible to identify defects in a SiC epitaxial wafer.

[0057] Furthermore, the defect inspection device 1 can identify the BPD of the buffer layer 52. Therefore, at the interface 54 between the buffer layer 52 and the drift layer 53, it is possible to identify the BPD that exists only in the buffer layer 52 that has been converted to TED, thereby improving the device characteristics of the drift layer 53.

[0058] (Embodiment 2) Next, a defect inspection apparatus according to Embodiment 2 will be described. In the defect inspection apparatus of this embodiment, the wavelength band of the photoluminescence light P1 transmitted by the filter unit 20 is changed from a wavelength band including 420 nm to 430 nm to a wavelength band including near-infrared light. Figure 5 is a configuration diagram illustrating the defect inspection apparatus according to Embodiment 2. As shown in Figure 5, the defect inspection apparatus 2 of this embodiment is modified from the aforementioned defect inspection apparatus 1 by changing the filter 21 in the filter unit 20 to a filter 22.

[0059] The detection optical system 30 detects photoluminescent light PL, which includes near-infrared light, that has passed through the modified filter section 20. The image processing unit 40 forms an image from the detected photoluminescent light PL, which includes near-infrared light. The image formed from the photoluminescent light PL, which includes near-infrared light, is called a near-infrared image. The image processing unit 40 identifies defects captured in the near-infrared image.

[0060] Figure 6 is a schematic diagram illustrating an image formed by the image processing unit 40 in the defect inspection apparatus 2 according to Embodiment 2. Figure 6 shows an image taken at the same position on the sample 50 as in Figure 3. As shown in Figure 6, the image processing unit 40 forms a near-infrared image from photoluminescent light PL transmitted in a wavelength band including near-infrared light. In this case, linear defects are observed in the near-infrared image. The length of the linear defects is called the third length L3. In Figure 6, for the sake of explanation, the linear defects in the near-infrared image are shown as black lines, similar to Figure 3.

[0061] BPD in buffer layer 52 is not captured in the near-infrared image. TED in drift layer 53 is not captured in the near-infrared image. On the other hand, BPD in drift layer 53 is captured in the near-infrared image.

[0062] Here, let D1 be the thickness of the drift layer 53, D2 be the thickness of the buffer layer 52, θ be the offset angle, and L3 be the third length as shown in equation (3) below. It is desirable that the thickness D1 of the drift layer 53 and the thickness D2 of the buffer layer 52 are different.

[0063] L3 = D1 / tanθ (3)

[0064] In this case, a defect with a length of the third length L3 corresponds to a defect K3 in the drift layer 53 that includes the BPD. The image processing unit 40 determines that a defect is a defect K3 in the drift layer 53 that includes the BPD when the length of the defect is the third length L3. The length of defect K3 is not limited to the third length L3, but may also include the third length L3 that includes unavoidable errors such as variations in thickness that occur when forming the buffer layer 52 and the drift layer 53. Therefore, the image processing unit 40 determines that a defect captured in a near-infrared image is a defect K3 in the drift layer 53 that includes the BPD when the length of the defect is within a predetermined third range that includes the third length L3 and belongs to a third range that is different from the first and second ranges.

[0065] Figure 7 illustrates how the BPDs appear in defects K1 and K2 as captured by the defect inspection apparatus 1 according to Embodiment 1 and the defect inspection apparatus 2 according to Embodiment 2. As shown in Figures 3 and 7, when a wavelength band including 420 nm to 430 nm is transmitted, both defects K1 and K2 are captured in the image. On the other hand, as shown in Figures 6 and 7, when a wavelength band including near-infrared light is transmitted, defect K3 is captured in the near-infrared image. Defects K1 and K2 are not captured in the near-infrared image. Note that defect K3 is obtained by subtracting the portion of defect K2 from defect K1. Therefore, the third length L3 is the difference between the first length L1 and the second length L2.

[0066] Furthermore, the image processing unit 40 may calculate the accuracy of the identified defects by comparing the number of defects K1 belonging to the first range with the number of defects K3 belonging to the third range. The closer the number of defects K1 and defects K3 are, the higher the accuracy of the defect identification.

[0067] Next, the defect inspection method of this embodiment will be described. Figure 8 is a flowchart illustrating the defect inspection method according to Embodiment 2. As shown in Figure 8, in addition to the steps S11 to S15 described above, the defect inspection method of this embodiment includes a step S16 in which the wavelength band transmitted to the filter unit 20 is changed to near-infrared light, a step S17 in which near-infrared light is detected, a step S18 in which a near-infrared image is formed from the near-infrared light, and a step S19 in which defects displayed in the image are identified.

[0068] As shown in step S16, after step S15 in which defects are identified, the wavelength band transmitted to the filter unit 20 is changed from a wavelength band including 420 nm to 430 nm to a wavelength band including near-infrared light. Specifically, in the filter unit 20, filter 21 is changed to filter 22.

[0069] Next, as shown in step S17, the near-infrared light transmitted through the modified filter section 20 is detected. Specifically, the detector 31 in the detection optical system 30 detects the near-infrared light transmitted through the filter 22 in the filter section 20.

[0070] Next, as shown in step S18, the image processing unit 40 forms a near-infrared image from the near-infrared light detected by the detection optical system 30.

[0071] Next, as shown in step S19, the image processing unit 40 identifies the defects captured in the near-infrared image. Specifically, the image processing unit 40 identifies the defect as a defect K3 containing a BPD in the drift layer 53 if the length of the defect captured in the near-infrared image belongs to a predetermined third range including the third length L3. By identifying defects in this way, the SiC epitaxial wafer can be inspected.

[0072] Furthermore, after step S19, the process may include a step of calculating the accuracy of the identified defects by comparing the number of defects K1 with lengths belonging to the first range with the number of defects K3 with lengths belonging to the third range.

[0073] According to this embodiment, the defect inspection device 2 can identify the defect K3 captured in the near-infrared image. This allows for the detection of the BPD in the drift layer 53. Furthermore, it is possible to compare the defect K1 captured in the image transmitted through a wavelength band including 420 nm to 430 nm with the defect K3 captured in the near-infrared image, thereby improving the defect detection rate. Other configurations and effects are described in Embodiment 1.

[0074] While embodiments of this disclosure have been described above, this disclosure includes appropriate modifications that do not impair its purpose and advantages, and is not limited by the embodiments described above. Furthermore, combinations of the configurations of Embodiments 1 and 2 also fall within the scope of the technical concept of this disclosure. [Explanation of Symbols]

[0075] 1, 2 Defect Inspection Equipment 10 Irradiation optical system 11 Light source 20 Filter section 21, 22 filters 30 detection optics 31 detectors 40 Image Processing Unit 50 samples 51 SiC substrate 52 Buffer Layers 53 Drift Layer 54 Interface 60 stages D1, D2 thickness EL excitation light K1, K2, K3 defects L1 First length L2 Second length PL photoluminescence light

Claims

1. An irradiation optical system for irradiating a sample, which includes a silicon carbide substrate, a buffer layer formed on the silicon carbide substrate, and a drift layer formed on the buffer layer, with excitation light. A detection optical system for detecting photoluminescent light generated from the sample, An image processing unit that forms an image from the detected photoluminescence light and identifies defects captured in the formed image, An optical element that selectively guides light in the wavelength band including 420 nm to 430 nm in the aforementioned photoluminescent light to the detection optical system, Equipped with, The image processing unit determines, based on a comparison of the length of the defect with a reference length based on the thickness and offset angle of the buffer layer, whether the defect includes a basal plane dislocation connecting the buffer layer to the drift layer, or whether the defect includes the basal plane dislocation in the buffer layer and an edge dislocation in the drift layer. Defect inspection device.

2. When the thickness of the buffer layer is D2, the offset angle is θ, and the reference length is L2 as shown below, L² = D² / tanθ The image processing unit determines that the defect includes the basal plane dislocation in the buffer layer and the edge dislocation in the drift layer if the length of the defect is the same as the length of the reference length. The defect inspection apparatus according to claim 1.

3. When the thickness of the drift layer is D1, the thickness of the buffer layer is D2, the offset angle is θ, and the reference length is L1 as shown below, L1=(D1+D2) / tanθ The image processing unit determines that the defect is a defect that includes a basal plane dislocation connecting the buffer layer to the drift layer, if the length of the defect is the same as the length of the reference length. The defect inspection apparatus according to claim 1.

4. An irradiation optical system for irradiating a sample, which includes a silicon carbide substrate, a buffer layer formed on the silicon carbide substrate, and a drift layer formed on the buffer layer, with excitation light. A detection optical system for detecting photoluminescent light generated from the sample, An image processing unit that forms an image from the detected photoluminescence light and identifies defects captured in the formed image, An optical element that selectively guides light in the wavelength band including 420 nm to 430 nm in the aforementioned photoluminescent light to the detection optical system, Equipped with, When the thickness of the buffer layer is D2, the offset angle is θ, and the reference length is L2 as shown below, L² = D² / tanθ The image processing unit determines that the defect includes a basal plane dislocation in the buffer layer and an edge dislocation in the drift layer when the length of the defect is the same as the length of the reference length. Defect inspection device.

Images