Silicon carbide epitaxial substrate

Abstract

Improve the reliability of the silicon carbide epitaxial substrate. 【Solution means】 A silicon carbide epitaxial substrate 10 having a silicon carbide substrate 20 and an epitaxial layer 30 formed on the upper surface of the silicon carbide substrate 20 is used. Here, the upper surface 30a of the epitaxial layer 30 has a plurality of surface defects. The plurality of surface defects include minute surface defects that satisfy L < d / tanθ, where L is the length of the surface defect in plan view, d is the thickness of the epitaxial layer 30, and θ is the off-angle of the silicon carbide substrate 20. In this case, the density of the minute surface defects is 100 pieces / cm 2 Is as follows.

Description

[Technical Field]

[0001] This disclosure relates to silicon carbide epitaxial substrates. [Background technology]

[0002] Silicon carbide (SiC) is extremely stable thermally and chemically, possesses excellent mechanical strength, is resistant to radiation, and has higher dielectric breakdown voltage and thermal conductivity compared to silicon (Si). Semiconductor devices using silicon carbide can achieve performance that is difficult to realize with semiconductor devices using existing semiconductor materials such as silicon. For this reason, silicon carbide is expected to be a next-generation semiconductor material in various fields.

[0003] Patent Document 1 (Japanese Patent Publication No. 2022-151601) describes the production of a silicon carbide epitaxial substrate by depositing an epitaxial layer on the upper surface of a silicon carbide substrate (hereinafter referred to as a silicon carbide substrate). This document describes reducing the density of linear surface defects that occur on the upper surface of the epitaxial layer.

[0004] Non-patent document 1 describes the effect of dislocations present in silicon carbide substrates on semiconductor devices using silicon carbide. [Prior art documents] [Patent Documents]

[0005] [Patent Document 1] Japanese Patent Publication No. 2022-151601 [Non-patent literature]

[0006] [Non-Patent Document 1] Applied Physics, Vol. 82, No. 10, p. 846. [Overview of the project] [Problems that the invention aims to solve]

[0007] When an epitaxial layer is continuously formed on the upper surface of a silicon carbide substrate, the state in the growth chamber of the growth apparatus changes due to deposition of deposits caused by the source gas. The change in the state in the growth chamber of the growth apparatus affects the quality of the silicon carbide epitaxial substrate, and as a result, may affect the reliability of semiconductor devices manufactured using the silicon carbide epitaxial substrate.

[0008] In view of the above circumstances, the present disclosure aims to improve the reliability of silicon carbide epitaxial substrates.

Means for Solving the Problems

[0009] A silicon carbide epitaxial substrate in one embodiment includes a silicon carbide substrate and an epitaxial layer formed on the upper surface of the silicon carbide substrate. Here, the upper surface of the epitaxial layer has a plurality of surface defects. The plurality of surface defects include minute surface defects that satisfy L < d / tanθ, where L is the length of the surface defect in plan view, d is the thickness of the epitaxial layer, and θ is the off-angle of the silicon carbide substrate. The density of the minute surface defects is 100 per cm 2 is as follows.

[0010] The silicon carbide epitaxial substrate in one embodiment has a silicon carbide substrate and an epitaxial layer formed on the upper surface of the silicon carbide substrate. Here, a plurality of surface defects and a plurality of threading dislocations are formed on the upper surface of the epitaxial layer. Among the plurality of surface defects, a defect that satisfies L < d / tanθ when the length of the surface defect in plan view is L, the thickness of the epitaxial layer is d, and the off-angle of the silicon carbide substrate is θ is called a micro surface defect. In a graph with the density of the threading dislocations on the upper surface of the epitaxial layer on the horizontal axis and the density of the micro surface defects on the upper surface of the epitaxial layer on the vertical axis, when linear fitting is performed by the least squares method with the intercept on the vertical axis being 0 using a plurality of data, the slope of the approximate curve is less than 0.05. The plurality of data are obtained by plotting the measurement results of the density of the threading dislocations and the density of the micro surface defects in each of the plurality of regions on the graph when the upper surface of the epitaxial layer is divided into a plurality of regions.

[0011] The manufacturing method of the silicon carbide epitaxial substrate in one embodiment includes: (a) a step of preparing a silicon carbide substrate; (b) a step of forming an epitaxial layer on the upper surface of the silicon carbide substrate in the growth chamber of a film forming apparatus. In the step (b), a silicon-based gas and a carrier gas are supplied into the growth chamber through a first pipe, and a carbon-based gas and a carrier gas are supplied into the growth chamber through a second pipe. In the step (b), the epitaxial layer is continuously formed on the upper surfaces of 25 or more silicon carbide substrates at a film forming rate faster than 40 μm / h. The upper surface of the epitaxial layer has a plurality of surface defects and a plurality of threading dislocations. The plurality of surface defects include micro surface defects that are defects satisfying L < d / tanθ when the length of the surface defect in plan view is L, the thickness of the epitaxial layer is d, and the off-angle of the silicon carbide substrate is θ. In the step (b), the density of the micro surface defects is 100 pieces / cm 2 As follows, the flow rate of the carrier gas in the first pipe and the flow rate of the carrier gas in the second pipe are adjusted.

Advantages of the Invention

[0012] According to one embodiment, the reliability of the silicon carbide epitaxial substrate can be improved.

Brief Description of the Drawings

[0013] [Figure 1] FIG. 1 is a cross-sectional view showing a silicon carbide epitaxial substrate in Embodiment 1. [Figure 2] FIG. 2 is a cross-sectional view showing a silicon carbide substrate in Embodiment 1. [Figure 3] FIG. 3 is a schematic cross-sectional view showing two gas supply pipes in Embodiment 1. [Figure 4] FIG. 4 is a graph showing the distribution of carrier concentration in the upper surface of the epitaxial layer in Example 3. [Figure 5] FIG. 5 is a graph showing the distribution of carrier concentration in the upper surface of the epitaxial layer in Comparative Example 1. [Figure 6] FIG. 6 is a graph showing the distribution of carrier concentration in the upper surface of the epitaxial layer in Comparative Example 2. [Figure 7] FIG. 7 is a graph showing the distribution of carrier concentration in the upper surface of the epitaxial layer in Comparative Example 2. [Figure 8] FIG. 8 is a graph showing the variation in carrier concentration of silicon carbide epitaxy when films are continuously formed on a plurality of silicon carbide substrates in Example 3. [Figure 9] FIG. 9 is a graph showing the variation in carrier concentration of silicon carbide epitaxy when films are continuously formed on a plurality of silicon carbide substrates in Comparative Example 1. [Figure 10] FIG. 10 is a graph showing the variation in carrier concentration of silicon carbide epitaxy when films are continuously formed on a plurality of silicon carbide substrates in Comparative Example 2. [Figure 11] FIG. 11 is a differential interference optical microscope image obtained in the detection of minute surface defects. [Figure 12]Figure 12 shows the analysis image obtained during the detection of through-helical dislocations. [Figure 13] Figure 13 is a plan view showing the density distribution of through-helical dislocations and micro-surface defects across the entire upper surface of the silicon carbide epitaxial substrate of Example 3. [Figure 14] Figure 14 is a plan view showing the density distribution of through-helical dislocations and micro-surface defects across the entire upper surface of the silicon carbide epitaxial substrate of Comparative Example 1. [Figure 15] Figure 15 is a graph showing the relationship between the density of through-helic dislocations and the density of micro-surface defects. [Modes for carrying out the invention]

[0014] In all the drawings illustrating the embodiments, the same reference numeral is used for identical components, and repeated explanations are omitted. Hatching may be used even in plan views to improve clarity.

[0015] In this specification, the radial direction refers to the direction along the diameter of the silicon carbide substrate.

[0016] Embodiment 1 The following describes how to improve the reliability of silicon carbide epitaxial substrates and reduce their manufacturing costs by optimizing the manufacturing conditions of the epitaxial layer formed on the upper surface of the silicon carbide substrate.

[0017] Room for improvement The epitaxial layer formed on the upper surface of the silicon carbide substrate requires that the carrier concentration, epitaxial layer thickness, and in-plane density of various surface defects and dislocation defects be within a desired range. Surface defects refer to defects that cause irregularities on the upper surface of the epitaxial layer. Thickness, in the case of the epitaxial layer in Figure 1, refers to the vertical dimension on the plane of the paper in Figure 1. Dislocation defects refer to defects that do not cause irregularities on the upper surface of the epitaxial layer and are detected by methods such as photoluminescence. The above desired range is determined by the design philosophy of the power device. Conditions for a good silicon carbide epitaxial substrate include low carrier concentration, low variation in epitaxial layer thickness, and low density of various defects.

[0018] As described in "Problems the Invention Aims to Solve," when forming an epitaxial layer, the conditions inside the growth chamber of the deposition apparatus change due to the deposition of deposits caused by the raw material gas when deposition is performed continuously. In other words, deposits adhere to the inner walls of the growth chamber of the deposition apparatus. Depending on the manner and location of deposit adhesion, the variation in carrier concentration within the epitaxial layer may become large. When the variation in carrier concentration within the epitaxial layer becomes large, the reliability and yield of the silicon carbide epitaxial substrate decrease.

[0019] From the perspective of improving the reliability of silicon carbide epitaxial substrates, it is desirable to have small variations in carrier concentration within the epitaxial layer. To reduce variations in carrier concentration, it is conceivable to shorten the maintenance cycle for removing deposits from the deposition apparatus and maintain the condition of the apparatus. Shortening the maintenance cycle means reducing the number of silicon carbide substrates that can be continuously deposited in the deposition apparatus, which lowers the operating rate of the deposition apparatus and thus increases the manufacturing cost of silicon carbide epitaxial substrates.

[0020] In addition, the silicon carbide substrate has many through dislocations, which are one of the dislocation-related defects. During epitaxial film formation, through dislocations may be converted into surface defects with unevenness on the upper surface of the epitaxial layer. The types of through dislocations include through edge dislocations and through screw dislocations. These defects can be the cause of various defects in semiconductor devices using silicon carbide, especially silicon carbide power devices (Non-Patent Document 1). A threading screw dislocation (TSD) is a defect that occurs in the epitaxial layer on the upper surface of the silicon carbide substrate. The threading screw dislocation penetrates the epitaxial layer from the lower surface of the epitaxial layer toward the upper surface of the epitaxial layer. Therefore, the threading screw dislocation can be confirmed by observing the upper surface of the epitaxial layer.

[0021] The surface defects include linear surface defects whose length L is expressed as L = d / tanθ when the thickness of the epitaxial layer is d and the off-angle of the silicon carbide substrate is θ. Such linear surface defects are also called carrots. Patent Document 1 discloses a method for reducing such linear surface defects.

[0022] For example, when the thickness of the epitaxial layer is 30 μm, the length L of the above linear surface defect is about 430 μm. The inventors of the present invention have found that in addition to such relatively large linear surface defects, linear surface defects with a smaller length L occur on the upper surface of the epitaxial layer. That is, there are linear surface defects expressed by L = d / tanθ and smaller linear surface defects among the linear surface defects. In this specification, in order to distinguish these linear surface defects from each other, among the linear surface defects, defects that satisfy L < d / tanθ are called minute surface defects. That is, the surface defects occurring on the upper surface of the epitaxial layer include minute surface defects that are defects satisfying L < d / tanθ.

[0023] The presence of minute surface defects can also be a cause of failure in various silicon carbide power devices. An increase in minute surface defects on the upper surface of the epitaxial layer reduces the reliability and yield of the silicon carbide epitaxial substrate. Therefore, from the viewpoint of improving the reliability and yield of silicon carbide epitaxial substrates, it is desirable to reduce the density of minute surface defects on the upper surface of the epitaxial layer.

[0024] Furthermore, a higher deposition rate of the epitaxial layer can reduce the manufacturing cost of silicon carbide epitaxial substrates. In this specification, the deposition rate of the epitaxial layer refers to the speed at which the material constituting the epitaxial layer is deposited on the upper surface of the silicon carbide substrate during the epitaxial layer deposition process. The deposition rate of the epitaxial layer is expressed as the thickness of the material deposited per unit time. The inventors have discovered that increasing the deposition rate of the epitaxial layer increases the number of microsurface defects.

[0025] Based on the above, when depositing epitaxial layers continuously on the upper surfaces of multiple silicon carbide substrates, there is a challenge in suppressing changes in the state of the growth chamber of the deposition apparatus due to the deposition of deposits caused by the raw material gas. In addition, there are challenges in suppressing variations in carrier concentration within the epitaxial layer and suppressing the occurrence of minute surface defects on the upper surface of the epitaxial layer. On the other hand, reducing the number of continuously deposited epitaxial layers or lowering the deposition rate increases the manufacturing cost of silicon carbide epitaxial substrates. Therefore, it is necessary to solve the challenges of suppressing variations in carrier concentration within the epitaxial layer and suppressing the occurrence of minute surface defects on the upper surface of the epitaxial layer without reducing the number of continuously deposited layers and while maintaining a high deposition rate.

[0026] The inventors have found that the manner and location of deposit adhesion within the growth chamber of the film deposition apparatus are related to the supply conditions of the raw material gas, which will be described later. Therefore, the inventors have devised ways to change the manner and location of deposit adhesion to the inner wall and other surfaces within the growth chamber of the film deposition apparatus. Furthermore, the inventors have found that the variations in carrier concentration and the occurrence of minute surface defects are related to the supply conditions of the raw material gas, which will be described later. Therefore, the inventors have devised ways to maintain a state where variations in carrier concentration are small without reducing the number of films deposited continuously, and to suppress the occurrence of minute surface defects while maintaining a high film deposition rate. These devised ways will be described in detail below.

[0027] Structure of silicon carbide epitaxial substrate Figure 1 is a cross-sectional view showing a silicon carbide epitaxial substrate 10 of Embodiment 1. The silicon carbide epitaxial substrate 10 comprises a silicon carbide substrate 20 and an epitaxial layer 30.

[0028] The silicon carbide substrate 20 has a first surface 20a and a second surface 20b located opposite the first surface 20a, with the epitaxial layer 30 located on the first surface 20a. The silicon carbide substrate 20 is composed of a silicon carbide single crystal. The polytype of the silicon carbide substrate 20 is preferably 4H. There are no restrictions on the size of the silicon carbide substrate 20. From the viewpoint of mass productionability of semiconductor devices made using the silicon carbide epitaxial substrate 10, the silicon carbide substrate 20 preferably has a diameter of 100 mm or more. For a 6-inch SiC substrate, the diameter is preferably 145 mm or more, and more preferably 150 mm or more. For an 8-inch SiC substrate, the diameter is preferably 195 mm or more, and more preferably 200 mm or more. The silicon carbide substrate 20 has a thickness determined according to the diameter according to the standard. For example, if the diameter is 100 mm or 150 mm, the thickness of the silicon carbide substrate 20 is 350 μm ± 25 μm or 500 μm ± 25 μm.

[0029] The first surface 20a of the silicon carbide substrate 20 used in the silicon carbide epitaxial substrate 10 is the Si surface. In other words, silicon is exposed on the first surface 20a. The crystal orientation of the first surface 20a is (0001). On the other hand, the second surface 20b is the C surface. In other words, carbon is exposed on the second surface 20b. The crystal orientation of the second surface 20b is (000-1).

[0030] Figure 2 is a cross-sectional view showing the silicon carbide substrate 20. The silicon carbide substrate 20 is preferably an off-surface substrate having an off-angle θ. Specifically, the first surface 20a, or the normal 20n of the first surface 20a, is preferably inclined by θ from the

[0001] direction to the

[1120] direction. The off-angle θ is preferably 0.5° or more and 8° or less, and more preferably 0.5° or more and 5° or less.

[0031] The properties of the first surface 20a of the silicon carbide substrate 20 are described in detail. The first surface 20a is preferably polished by CMP (Chemical Mechanical Polishing). Specifically, it is preferable that the first surface 20a is polished by CMP until its surface roughness Ra is 1 nm or less. More preferably, the surface roughness Ra of the first surface 20a is 0.2 nm or less. The surface roughness Ra can be measured, for example, by a white light interference microscope. The surface roughness Ra is obtained, for example, by measuring the first surface 20a at three locations over a length of 100 μm and calculating the average of the three measurements. Ideally, the surface roughness Ra is 0 nm, but in reality, it is never 0 nm. Therefore, the lower limit of the preferred range for surface roughness Ra is greater than 0.

[0032] Furthermore, the BPD (Basal Plane Dislocation) density in the first surface 20a is 3000 particles / cm³. 2 The following is preferable: The density of BPD is 2000 particles / cm³. 2 It is more preferable that the following conditions are met: 1000 pieces / cm 2It is more preferable that the following holds. For example, the BPD density can be measured by etching the first surface 20a of the silicon carbide substrate 20 with molten KOH (potassium hydroxide) and counting the number of BPDs that appear as etch pits using an optical microscope. Similar to the surface roughness Ra, the lower limit of the preferable range of the BPD density is greater than 0 pieces / cm 2 is larger.

[0033] The epitaxial layer 30 is formed on the first surface 20a of the silicon carbide substrate 20 by an epitaxial growth method. The epitaxial layer 30 is composed of a single crystal of silicon carbide. The thickness of the epitaxial layer 30 can be arbitrarily set according to the performance required for the semiconductor device manufactured using the silicon carbide epitaxial substrate 10. For example, the thickness of the epitaxial layer 30 is 1 μm or more and 100 μm or less.

[0034] The upper surface 30a of the epitaxial layer 30 has a plurality of linear surface defects. The plurality of linear surface defects includes minute surface defects represented by L < d / tanθ. On the upper surface 30a of the epitaxial layer 30, the density of the minute surface defects is preferably less than 100 pieces / cm 2 It is more preferable that the density of the minute surface defects is less than 20 pieces / cm 2 and even more preferable that it is less than 5 pieces / cm 2 In addition, a plurality of screw dislocations are formed in the epitaxial layer 30. On the upper surface 30a of the epitaxial layer 30, the density of the screw dislocations is preferably 400 pieces / cm 2 or less.

[0035] The minute surface defects have irregularities with respect to the upper surface 30a. When the upper surface 30a of the epitaxial layer 30 is viewed from a direction perpendicular thereto, that is, in plan view, the minute surface defects have an elongated linear shape. The length L in the longitudinal direction of the minute surface defects in plan view depends on the thickness d of the epitaxial layer 30 and the off-angle θ of the silicon carbide substrate 2*0, and is in the relationship of L < d / tanθ. Note that "plan view" refers to viewing the upper surface of the epitaxial layer of the silicon carbide substrate as shown in FIG. 13, for example.

[0036] The density of minute surface defects can be measured, for example, by a wafer inspection / review system capable of acquiring differential interference contrast microscopy images. Such a wafer inspection / review system can obtain the location and number of regions of linear brightness changes in the acquired differential interference contrast microscopy image and calculate the density of minute surface defects.

[0037] Method for manufacturing silicon carbide epitaxial substrates The manufacturing method for the silicon carbide epitaxial substrate 10 of Embodiment 1 will be described below. First, a silicon carbide substrate 20 is prepared. The silicon carbide substrate 20 has a first surface 20a with the crystal orientation and off-angle described above. The first surface 20a is prepared to have a surface roughness Ra of 1 nm or less by the CMP method. A smaller surface roughness Ra value is preferable. For example, a commercially available silicon carbide substrate 20 may be obtained and the first surface 20a may be further polished by the CMP method.

[0038] Next, an epitaxial layer 30 is formed on the first surface 20a of the silicon carbide substrate 20 using an epitaxial growth method. Here, the epitaxial layer 30 is formed using the CVD (Chemical Vapor Deposition) method.

[0039] For example, a silicon carbide substrate 20 is introduced into the growth chamber of a film deposition apparatus and placed in a holder with the (0001) plane, which is the first surface 20a, facing upwards. The silicon carbide substrate 20 is heated to a temperature of 1500°C to 1800°C, and a carrier gas, a gas that will serve as a carbon source, a gas that will serve as a silicon source, and a gas that will serve as a dopant are introduced into the growth chamber to deposit an epitaxial layer 30. For example, hydrogen (H2) can be used as the carrier gas. A carbon-based gas such as propane (C3H8) can be used as the gas that will serve as the carbon source in the raw material gas. A silicon-based gas such as silane (SiH4) can be used as the gas that will serve as the silicon source in the raw material gas. This can be done. Nitrogen (N2) or the like can be used as the dopant gas. Before depositing the epitaxial layer 30, only the carrier gas may be introduced into the growth chamber to clean the first surface 20a of the silicon carbide substrate 20.

[0040] These gases are supplied into the growth chamber through two gas supply pipes. Figure 3 shows a schematic cross-sectional view of the two gas supply pipes in Embodiment 1.

[0041] As shown in Figure 3, in epitaxial growth, the multiple gas supply pipes that supply the raw material gas and carrier gas into the growth chamber include a silane pipe 41 that supplies silane gas and a propane pipe 42 that supplies propane gas. Through the silane pipe 41 (corresponding to the first pipe), silane gas, which is the raw material gas, and hydrogen gas, which is the carrier gas, are supplied into the growth chamber. Through the propane pipe 42 (corresponding to the second pipe), propane gas and nitrogen gas, which are the raw material gases, and hydrogen gas, which is the carrier gas, are supplied into the growth chamber.

[0042] The ratio of carbon in the introduced carbon source gas to silicon in the silicon source gas (C / Si) is preferably 1 or greater. Specifically, C / Si is preferably 1 or greater and 1.6 or less.

[0043] The pressure inside the growth chamber during the growth of the epitaxial layer 30 is preferably between 10 kPa and 50 kPa.

[0044] As described in the following examples, in the manufacturing method of the silicon carbide epitaxial substrate 10 of Embodiment 1, an epitaxial layer 30 is formed on the (0001) plane of the silicon carbide substrate 20. In Embodiment 1, the flow rate of the carrier gas is adjusted during the film formation process of the epitaxial layer 30. Even in this case, the state of the growth chamber of the film formation apparatus changes due to the deposition of deposits caused by the raw material gas during the film formation process of the epitaxial layer 30. That is, deposits adhere to the inner wall of the growth chamber of the film formation apparatus. However, by adjusting the flow rate of the carrier gas as described above, the manner and location of deposit adhesion change. As a result, the generation of minute surface defects on the upper surface 30a of the epitaxial layer 30 can be suppressed. Furthermore, when forming epitaxial layers 30 continuously on multiple silicon carbide substrates 20 in the film formation apparatus, this adjustment can suppress variations in carrier concentration within the epitaxial layer 30. In other words, by adjusting the flow rate of the carrier gas, the impact on the reliability of the silicon carbide epitaxial substrate 10 due to the deposition of deposits in the growth chamber can be suppressed.

[0045] Furthermore, from the viewpoint of improving the reliability of the silicon carbide epitaxial substrate 10, it is preferable that the temperature difference between the center and edge of the first surface 20a of the silicon carbide substrate 20 during the film formation process of the epitaxial layer 30 be 50°C or less. It is more preferable that this temperature difference be 20°C or less, and even more preferable that it be 10°C or less.

[0046] Examples Sample preparation The following describes the results of measuring the carrier concentration and the density of microsurface defects in a silicon carbide epitaxial substrate 10, which was fabricated using the silicon carbide epitaxial substrate manufacturing method of Embodiment 1.

[0047] Examples 1-3, Comparative Examples 1, 2 A silicon carbide substrate 20 with a diameter of 150 mm was prepared, with the first surface 20a being the (0001) plane and an off-angle θ of 4°. The first surface 20a was polished by CMP until the surface roughness Ra was 1 nm or less. Subsequently, an epitaxial layer 30 was deposited on the first surface 20a of the silicon carbide substrate 20. The growth temperature was set to 1600°C, and the pressure inside the growth chamber during growth was set to 30 kPa. As shown in Table 1, the C / Si ratio in the raw material gas was set to 1.25. Hydrogen gas was used as the carrier gas, and propane gas and silane gas were used as the raw material gases. Nitrogen was used as the dopant source.

[0048] Table 1 shows the flow rate ratios of hydrogen gas in the silane gas pipes between Examples 1-3 and Comparative Examples 1 and 2. When the flow rate ratios of Examples 1-3 and Comparative Example 1 are set to 100, the flow rate of hydrogen gas in Comparative Example 2 was set to 120.

[0049] Table 1 shows the ratio of hydrogen gas flow rates in the propane gas pipes between Examples 1-3 and Comparative Examples 1 and 2. When the flow rate ratio of Comparative Examples 1 and 2 is set to 100, the flow rate of hydrogen gas in the propane gas pipes was set so that the flow rate ratio of Example 1 was 65 and the flow rate ratio of Example 2 was 85.

[0050] In Example 3, when the hydrogen gas flow rate ratio in the propane gas pipe in Comparative Examples 1 and 2 was set to 100, the hydrogen gas flow rate ratio in the propane gas pipe was adjusted between 65 and 85. In Example 3, the carrier concentration on the upper surface 30a of the epitaxial layer 30 after film formation was measured, and the hydrogen gas flow rate in the propane gas pipe in the subsequent film formation process was adjusted according to the measurement results.

[0051] Table 1 shows the temperature difference between the center and edge of the first surface 20a of the silicon carbide substrate 20 in Examples 1-3 and Comparative Examples 1 and 2. In Examples 1-3, the temperature difference is set to be 50°C or less. Specifically, in Examples 1 and 2, the temperature difference is 20°C. In Example 3, the temperature difference is between 10 and 40°C. A temperature difference of 10°C or less is even more desirable. In Comparative Examples 1 and 2, the temperature difference is thought to be greater than 50°C, but it is not constant, so specific numerical values ​​for the temperature difference are not listed in Table 1.

[0052] In Examples 1 and 2 and Comparative Example 1, the epitaxial layer 30 was deposited on only one silicon carbide substrate 20. In Example 3, the epitaxial layer 30 was deposited continuously on 28 silicon carbide substrates 20. In Comparative Example 2, the epitaxial layer 30 was deposited continuously on 15 silicon carbide substrates 20. As used herein, "continuously depositing epitaxial layers" means performing the process of depositing epitaxial layers 30 on multiple silicon carbide substrates 20 without performing maintenance on the growth chamber of the deposition apparatus. In other words, "continuously depositing epitaxial layers" means performing the process of depositing epitaxial layers 30 on multiple silicon carbide substrates 20 without releasing the vacuum state inside the growth chamber of the deposition apparatus.

[0053] In Examples 1-3, the deposition rate of the epitaxial layer 30 was set to 48 μm / h, while in Comparative Examples 1 and 2, the deposition rates were set to 45 μm / h and 43 μm / h, respectively. Thus, in Examples 1-3, the deposition rate of the epitaxial layer 30 was set faster than 40 μm / h.

[0054] [Table 1]

[0055] Table 2 shows the analysis results for each of Examples 3-1, 3-2, 3-3, and Comparative Example 1-1. In Examples 3-1, 3-2, and 3-3, analysis was performed on separate silicon carbide epitaxial substrates 10 arbitrarily selected from among those prepared under the same conditions as in Example 3 described above. In Comparative Example 1-1, analysis was performed on a silicon carbide epitaxial substrate 10 arbitrarily selected from among those prepared under the same conditions as in Comparative Example 1 described above.

[0056] Furthermore, through-helical dislocations were detected in the silicon carbide epitaxial substrate 10 for each of Examples 3-1, 3-2, 3-3, and Comparative Example 1-1. From the detection results, the average value (average number) of through-helical dislocations within the upper surface 30a of the epitaxial layer 30 was calculated, as shown in Table 2.

[0057] Furthermore, based on the detection results of micro-surface defects on the upper surface 30a of the epitaxial layer 30, the average value of micro-surface defects within the upper surface 30a of the epitaxial layer 30 was calculated, as shown in Table 2.

[0058] Furthermore, as shown in Table 2, the slope of the approximation curve representing the relationship between the density of through-helical dislocations and the density of micro-surface defects was calculated for each of Examples 3-1, 3-2, 3-3, and Comparative Example 1-1.

[0059] [Table 2]

[0060] measurement The film deposition rates shown in Table 1 were calculated from the thickness of the epitaxial layer after deposition and the deposition time. The thickness of the epitaxial layer was measured using a non-contact film thickness measuring device. The thickness of the epitaxial layer was measured using the "UTS-2000" film thickness measuring device from JASCO Corporation.

[0061] The carrier concentrations shown in Table 1 were measured at multiple locations in the X and Y directions within the upper surface 30a of the epitaxial layer 30 using the CV (Capacitance-Voltage) method. Carrier concentrations were measured using a CV measuring instrument "CVmap92A" manufactured by Four Dimensions. The X and Y directions are along the upper surface 30a of the epitaxial layer 30 and are orthogonal to each other in a plan view. The X direction is the direction of the orientation flat (the direction along the

[1120] direction), and the Y direction is the direction orthogonal to the orientation flat. Among the measurement results of carrier concentrations at multiple locations on the upper surface 30a of a single silicon carbide epitaxial substrate 10, the maximum carrier concentration is defined as Max, the minimum carrier concentration as Min, and the average of the measurement results of all carrier concentrations at multiple locations is defined as Ave. In this case, the variation in carrier concentration within the upper surface 30a of the epitaxial layer 30 is expressed by the formula 0.5 × (Max - Min) / Ave. In determining the carrier concentration variability shown in Table 1, a variability of 20% or less was considered OK, and a variability greater than 20% was considered NG.

[0062] The detection of minute surface defects shown in Table 1 was performed by observing the surface of the epitaxial layer 30 using a differential interference contrast lens of a confocal microscope, and analyzing the obtained differential interference contrast optical microscopy images by dividing them into 5 mm × 5 mm regions (chips). This surface observation was performed using the "SICA" (registered trademark) defect inspection device manufactured by Lasertec Corporation. Because minute surface defects have irregularities on the surface, their brightness appears different in the image compared to the flat, normal crystal region. The average value of minute surface defects shown in Table 1 was calculated by averaging the number of minute surface defects in each of the multiple regions mentioned above and expressing it as the number per unit area. In the determination of minute surface defects in Table 1, the average value of minute surface defects is 100 defects / cm². 2 If it is less than 100 pieces / cm, it will be judged as OK. 2 If the above conditions were met, it was judged as NG (Not Good). The average in-plane value of microsurface defects shown in Table 2 was calculated by averaging the number of microsurface defects in each of the multiple above-mentioned regions and expressing it as the number per unit area.

[0063] The detection of through-helical dislocations in the silicon carbide epitaxial substrate 10 shown in Table 1 was performed by measuring the reflection image of the

[0008] plane (0008 reflection image) using X-ray topography, and then dividing the obtained image into 5 mm × 5 mm regions for analysis. The 0008 reflection image was measured using the topographic measurement system "XRTMicron" (registered trademark) manufactured by Rigaku Corporation. The in-plane average value of the through-helical dislocation density shown in Table 2 was calculated by determining the number of through-helical dislocations per unit area in each of the multiple regions mentioned above, and then calculating the average value of these values.

[0064] The slope of the approximation curve shown in Table 2 was calculated by graphing the density of through-helical dislocations in the silicon carbide epitaxial substrate 10 on the horizontal axis and the density of minute surface defects within the upper surface 30a of the silicon carbide epitaxial substrate 10 on the vertical axis, and performing linear fitting using the least squares method with the intercept of the vertical axis set to 0.

[0065] Results and Discussion Figures 4, 5, 6, and 7 show the carrier concentration distribution within the upper surface 30a of the epitaxial layer 30. Figure 4 is a graph showing the carrier concentration at each position within the upper surface 30a of the epitaxial layer 30 formed under the same conditions as in Example 1. Figure 5 is a graph showing the carrier concentration at each position within the upper surface 30a of the epitaxial layer 30 formed under the same conditions as in Comparative Example 1. Figure 6 is a graph showing the carrier concentration at each position within the upper surface 30a of the epitaxial layer 30 constituting the first silicon carbide epitaxial substrate 10, of which 15 films were deposited consecutively under the same conditions as in Comparative Example 2. Figure 7 is a graph showing the carrier concentration at each position within the upper surface 30a of the epitaxial layer 30 constituting the fifth silicon carbide epitaxial substrate 10, of which 15 films were deposited consecutively under the same conditions as in Comparative Example 2.

[0066] In Figures 4, 5, 6, and 7, the horizontal axis represents the radial position of the upper surface 30a of the silicon carbide epitaxial substrate 10, and the vertical axis represents the carrier concentration within the upper surface 30a of the epitaxial layer 30.

[0067] The silicon carbide epitaxial substrate 10 is a disc-shaped semiconductor substrate. The diameter of the silicon carbide substrate 20 is 150 mm. In other words, the silicon carbide epitaxial substrate 10 has a diameter of 6 inches. The values ​​on the horizontal axis in Figures 4, 5, 6, and 7 represent the radial distance from the center of the top surface 30a of the circular silicon carbide epitaxial substrate 10. In Figures 4, 5, 6, and 7, the diamond-shaped plots represent the carrier concentrations measured at each of the multiple measurement points aligned in the X direction, which is the radial direction of the top surface 30a of the silicon carbide epitaxial substrate 10. In Figures 4, 5, 6, and 7, the circular plots represent the carrier concentrations measured at each of the multiple measurement points aligned in the Y direction, which is the radial direction of the top surface 30a of the silicon carbide epitaxial substrate 10. Figures 4, 5, 6, and 7 show the relative carrier concentrations at other locations, with the carrier concentration at the center of the upper surface 30a of the silicon carbide epitaxial substrate 10 set to 100.

[0068] As shown in Figure 4, the variation in carrier concentration within the upper surface 30a of the epitaxial layer 30 formed under the same conditions as in Example 1 was calculated using the formula 0.5 × (Max-Min) / Ave., and the variation in carrier concentration was 5.5%. In contrast, as shown in Figure 5, the variation in carrier concentration within the upper surface 30a of the epitaxial layer 30 formed under the same conditions as in Comparative Example 1 was 21.7%.

[0069] Furthermore, as shown in Figures 6 and 7, the variation in carrier concentration within the upper surface 30a of the epitaxial layer 30, which was formed continuously under the same conditions as in Comparative Example 2, was 5.7% for the first silicon carbide epitaxial substrate 10 and 8.7% for the fifth silicon carbide epitaxial substrate 10. Although graphs are not shown for the sixth and subsequent silicon carbide epitaxial substrates 10, as shown in Figures 6 and 7, the variation in carrier concentration increased with the number of continuously deposited layers. As a result, although graphs are not shown here, the variation in carrier concentration within the upper surface 30a of the epitaxial layer 30 constituting the 15th silicon carbide epitaxial substrate 10 exceeded 20%.

[0070] Based on the above, as shown in Table 1, Example 1 was judged as OK, and similarly, Examples 2 and 3 were judged as OK. In contrast, Comparative Examples 1 and 2 were judged as NG. Thus, in Examples 1 to 3, the variation in carrier concentration on the upper surface 30a of the epitaxial layer 30 can be suppressed compared to Comparative Examples 1 and 2. Even when film deposition is performed continuously on 25 or more silicon carbide substrates 20, as in Example 3, the variation in carrier concentration can be suppressed.

[0071] In this process, multiple silicon carbide substrates 20 are sometimes transported in sets of 25 in a container called a cassette or carrier. As a result, processes such as the deposition of the epitaxial layer 30 can be carried out by treating the 25 silicon carbide substrates 20 as one lot and performing deposition on these 25 substrates 20 continuously. In practice, deposition may be carried out continuously on as many as 100 to 150 silicon carbide substrates 20. Since the silicon carbide substrates 20 are processed in sets of 25, if the number of silicon carbide substrates 20 to be deposited on continuously is less than 25, the manufacturing cost of the silicon carbide epitaxial substrate 10 increases. Therefore, performing deposition on 25 or more silicon carbide substrates 20 continuously contributes to improving the efficiency of the manufacturing process of the silicon carbide epitaxial substrate 10. In other words, the manufacturing cost of the silicon carbide epitaxial substrate 10 can be reduced.

[0072] Furthermore, the silicon carbide substrates 20 are sometimes processed in sets of 10, including the deposition process. Therefore, continuously depositing films onto 10 or more silicon carbide substrates 20 contributes to improving the efficiency of the silicon carbide epitaxial substrate manufacturing process 10.

[0073] Figures 8, 9, and 10 show the variation in carrier concentration within the upper surface 30a of each epitaxial layer 30 when epitaxial layers 30 are continuously deposited on the upper surfaces of multiple silicon carbide substrates 20. Figure 8 is a graph showing the carrier concentration within the upper surface 30a of each epitaxial layer 30 when 28 epitaxial layers 30 are continuously deposited under the same conditions as in Example 3. Figure 9 is a graph showing the carrier concentration within the upper surface 30a of each epitaxial layer 30 when 25 epitaxial layers 30 are continuously deposited under the same conditions as in Comparative Example 1. Figure 10 is a graph showing the carrier concentration within the upper surface 30a of each epitaxial layer 30 when 5 epitaxial layers 30 are continuously deposited under the same conditions as in Comparative Example 2.

[0074] The horizontal axis in Figures 8, 9, and 10 shows the number of silicon carbide substrates 20 when film deposition is performed continuously on the upper surface of the silicon carbide substrate 20. The vertical axis in Figures 8, 9, and 10 shows the variation in carrier concentration. The values ​​on the vertical axis are numerical values ​​representing the variation, calculated using the formula 0.5 × (Max-Min) / Ave.

[0075] As shown in Figure 8, when film deposition was performed under the same conditions as in Example 3, the variation in carrier concentration was small in all cases, and even when film deposition was performed continuously, the variation in carrier concentration remained small.

[0076] In contrast, as shown in Figure 9, when film deposition was performed under the same conditions as in Comparative Example 1, the variation in carrier concentration increased with increasing numbers of silicon carbide substrates 20 to which films were deposited consecutively. For example, in the epitaxial layer 30 deposited on the 15th substrate, the variation in carrier concentration exceeded 20%. From the graph in Figure 8, it can be predicted that the variation in carrier concentration will exceed 20% when the number of substrates to which films are deposited consecutively exceeds approximately 30.

[0077] Furthermore, as shown in Figure 10, when film deposition was performed under the same conditions as in Comparative Example 2, the variation in carrier concentration increased with increasing numbers of silicon carbide substrates 20 to which films were deposited consecutively. From the graph in Figure 10, it can be predicted that when the number of films deposited consecutively exceeds approximately 20, the variation in carrier concentration will exceed 20%.

[0078] Thus, in Example 3, the variation in carrier concentration on the upper surface 30a of the epitaxial layer 30 can be suppressed compared to Comparative Examples 1 and 2. Even when film deposition is performed continuously on 25 or more silicon carbide substrates 20, as in Example 3, the variation in carrier concentration can be suppressed. Therefore, the manufacturing cost of the silicon carbide epitaxial substrate 10 can be reduced, and the variation in carrier concentration can be suppressed.

[0079] Figure 11 shows differential interference contrast (OCT) microscopic images obtained during the detection of minute surface defects. As indicated by the arrows in Figure 11, the minute surface defects appear as linear lines on the upper surface 30a of the epitaxial layer 30. In this image, light is irradiated from the top of Figure 11. Areas that are linearly recessed from the upper surface 30a can be seen as black lines in the shaded areas. Areas that are linearly protruding from the upper surface 30a can be seen as white lines where light is shining.

[0080] Figure 12 shows an analysis image (0008 reflection image) obtained using X-ray topography in the detection of through-helical dislocations. As indicated by the arrows in Figure 12, through-helical dislocations appear as black spots on the upper surface 30a of the epitaxial layer 30. Figures 11 and 12 are images obtained by observing almost the same location on the same silicon carbide epitaxial substrate 10. By comparing Figures 11 and 12, it can be considered that the locations where minute surface defects occur and the locations where through-helical dislocations occur correspond to each other.

[0081] Figures 13 and 14 are plan views showing the density distribution of through-helical dislocations and micro-surface defects across the entire upper surface 30a of the silicon carbide epitaxial substrate 10. The left side of Figure 13 shows the density distribution of through-helical dislocations across the entire upper surface 30a of the silicon carbide epitaxial substrate 10 fabricated under the same conditions as in Example 3. The right side of Figure 13 shows the density distribution of micro-surface defects across the entire upper surface 30a of the silicon carbide epitaxial substrate 10 fabricated under the same conditions as in Example 3. The left side of Figure 14 shows the density distribution of through-helical dislocations across the entire upper surface 30a of the silicon carbide epitaxial substrate 10 fabricated under the same conditions as in Comparative Example 1. The right side of Figure 14 shows the density distribution of micro-surface defects across the entire upper surface 30a of the silicon carbide epitaxial substrate 10 fabricated under the same conditions as in Comparative Example 1.

[0082] Figures 13 and 14 show the density of through-helical dislocations and microsurface defects on the upper surface 30a of the silicon carbide epitaxial substrate 10 in various brightness levels within each partitioned 5mm × 5mm region. As shown in Figures 13 and 14, when the manufacturing conditions of Comparative Example 1 are applied, the silicon carbide epitaxial substrate 10 used has a lower density of through-helical dislocations compared to when the manufacturing conditions of Example 3 are applied. Thus, when the manufacturing conditions of Example 3 are applied, even though the silicon carbide epitaxial substrate 10 has a relatively high density of through-helical dislocations, the density of microsurface defects on the entire upper surface 30a of the silicon carbide epitaxial substrate 10 can be reduced compared to when the manufacturing conditions of Comparative Example 1 are applied.

[0083] As shown in Figure 14 and Table 2, the silicon carbide epitaxial substrate 10 fabricated under the same conditions as Comparative Example 1 appears to have high quality when focusing solely on the density distribution of through-helical dislocations. However, when the silicon carbide epitaxial substrate 10 is fabricated under the same conditions as Comparative Example 1, minute surface defects tend to occur on the upper surface 30a of the epitaxial layer 30. From the viewpoint of improving the reliability and yield of the silicon carbide epitaxial substrate 10, it is important to suppress the density of minute surface defects, as is the case with the silicon carbide epitaxial substrate 10 fabricated under the same conditions as Example 3.

[0084] As shown on the right side of Figure 13, the density of microsurface defects is 100 defects / cm² in any region of the upper surface 30a of the silicon carbide epitaxial substrate 10. 2 The following can be observed. Based on these measurement results, as shown in Table 1, the determination of micro-surface defects was OK for Examples 1-3 and Comparative Example 2. In contrast, the determination of micro-surface defects was NG for Comparative Example 1.

[0085] Thus, in Examples 1 to 3, the occurrence of minute surface defects on the upper surface 30a of the epitaxial layer 30 can be suppressed compared to Comparative Example 1. In Embodiment 1, by adjusting the hydrogen gas flow rate, the occurrence of minute surface defects can be suppressed while maintaining a film deposition rate faster than 40 μm / h. Therefore, the manufacturing cost of the silicon carbide epitaxial substrate 10 can be reduced, and variations in carrier concentration can be suppressed.

[0086] Figure 15 is a graph showing the relationship between the density of through-helical dislocations and the density of micro-surface defects. The horizontal axis of Figure 15 shows the density of through-helical dislocations in the silicon carbide epitaxial substrate 10. The vertical axis of Figure 15 shows the density of micro-surface defects on the upper surface 30a of the epitaxial layer 30. Each of the multiple triangular plots shown in Figure 15 represents the density of through-helical dislocations and micro-surface defects measured in each region when the upper surface 30a of the silicon carbide epitaxial substrate 10 of Comparative Example 1-1 is divided into multiple 5mm × 5mm regions. Each of the multiple circular plots shown in Figure 15 represents the density of through-helical dislocations and micro-surface defects measured in each region when the upper surface 30a of the silicon carbide epitaxial substrate 10 of Example 3-1 is divided into multiple 5mm × 5mm regions.

[0087] Figure 15 shows the approximate curve obtained by linear fitting using the least squares method with multiple data points representing multiple measurement results from Comparative Example 1-1, as shown by the solid line. In other words, the graph shown in Figure 15 was created using these multiple data points. The intercept of the vertical axis of the approximate curve shown by the solid line is 0 (items / cm²). 2 ) is stated. Figure 15 also shows the approximate curve obtained by linear fitting using the least squares method with multiple data points, which are multiple measurement results from Example 3-1, as a dashed line. The intercept of the vertical axis of the approximate curve shown as a dashed line is 0 (pieces / cm). 2 The data in question consists of graphs plotting the measurement results of the density of through-helical dislocations and the density of micro-surface defects in each of several regions on the upper surface 30a of the silicon carbide epitaxial substrate 10.

[0088] A large slope value in the approximation curve indicates a high probability that through-helical dislocations in the substrate are converted into micro-surface defects. From the viewpoint of improving the reliability and yield of the silicon carbide epitaxial substrate 10, it is desirable that the density of micro-surface defects be less than 0.05. As shown in Figure 15, the slope of the approximation curve for Comparative Example 1-1 is represented by y = 1.744x. The slope of the approximation curve for Example 3-1 is represented by y = 0.024x. In other words, as shown in Table 2, the slope of the approximation curve for Comparative Example 1-1 is 1.744, and the slope of the approximation curve for Example 3-1 is 0.024. In the silicon carbide epitaxial substrate 10 of Example 3-1, the density of micro-surface defects on the upper surface 30a of the epitaxial layer 30 is kept low.

[0089] Thus, in Example 3-1, the occurrence of minute surface defects on the upper surface 30a of the epitaxial layer 30 can be suppressed compared to Comparative Example 1-1. In Embodiment 1, by adjusting the hydrogen gas flow rate, the occurrence of minute surface defects can be suppressed while maintaining a relatively fast film deposition rate of 40 μm / h. Therefore, the manufacturing cost of the silicon carbide epitaxial substrate 10 can be reduced, and variations in carrier concentration can be suppressed.

[0090] In Embodiment 1, when the flow rate ratio of Comparative Examples 1 and 2 is set to 100, the flow rate ratio of hydrogen gas in the propane gas pipe is adjusted between 65 and 85. By adjusting the gas amount in this way, it is possible to suppress the occurrence of minute surface defects and variations in carrier concentration on the upper surface 30a of the epitaxial layer 30 while maintaining a film deposition rate faster than 40 μm / h. Therefore, it is possible to improve the reliability and yield of the silicon carbide epitaxial substrate 10 while reducing the manufacturing cost of the silicon carbide epitaxial substrate 10. Furthermore, even when film deposition is performed continuously on 25 or more silicon carbide substrates 20, variations in carrier concentration on the upper surface 30a of the epitaxial layer 30 can be suppressed. In other words, it is possible to suppress variations in carrier concentration while reducing the manufacturing cost of the silicon carbide epitaxial substrate 10.

[0091] Although the present inventors have described the invention in detail based on its embodiments, it goes without saying that the present invention is not limited to the embodiments described above and can be modified without departing from the spirit of the invention.

[0092] For example, another epitaxial layer may be formed between the silicon carbide substrate and the epitaxial layer described above. In that case, the epitaxial layer described herein as being formed on the upper surface of the silicon carbide substrate is interpreted as being formed on the silicon carbide substrate. [Explanation of Symbols]

[0093] 10 Silicon carbide epitaxial substrate 20 Silicon carbide substrate 20a Page 1 20b 2nd side 20n normal 30 Epitaxial Layer 30a top surface 41 Silane pipe 42 Propane Pipes

Claims

1. A silicon carbide epitaxial substrate having an epitaxial layer formed on a silicon carbide substrate, The epitaxial layer has minute surface defects on its upper surface, The length L of the aforementioned minute surface defect satisfies L < d / tanθ in a plan view. The density of the aforementioned micro-surface defects is 100 defects / cm³. ２ The following: When the upper surface of the epitaxial layer is divided into multiple regions, and the maximum carrier concentration in each region is Max, the minimum is Min, and the average is Ave., The variation in the carrier concentration, expressed by the formula 0.5 × (Max - Min) / Ave., is 20% or less. Silicon carbide epitaxial substrate. d: Thickness of the epitaxial layer θ: Off-angle of the silicon carbide substrate

2. The density of the aforementioned micro-surface defects is 20 defects / cm³. ２ The following is: The silicon carbide epitaxial substrate according to claim 1.

3. The density of the aforementioned micro-surface defects is 5 defects / cm³. ２ The following is: The silicon carbide epitaxial substrate according to claim 1.

4. The variation in the carrier concentration is 10% or less. The silicon carbide epitaxial substrate according to claim 1.

5. A silicon carbide epitaxial substrate having an epitaxial layer formed on a silicon carbide substrate, The epitaxial layer has minute surface defects and through-helical dislocations on its upper surface. The length L of the aforementioned minute surface defect satisfies L < d / tanθ in a plan view. The upper surface of the epitaxial layer is divided into multiple regions, and the slope of the approximation line passing through the origin, determined from the relationship between the density of through-helical dislocations and the density of micro-surface defects in each of the regions, is less than 0.05 when plotted on a graph with the density of through-helical dislocations on the horizontal axis and the density of micro-surface defects on the vertical axis. Silicon carbide epitaxial substrate. d: Thickness of the epitaxial layer θ: Off-angle of the silicon carbide substrate

Images