Method for estimating appropriate process amount

Abstract

Provided is a technique for estimating, on the basis of the intensity of scattered light scattered on a semiconductor substrate, an appropriate amount of process to be performed on the semiconductor substrate in order to obtain a semiconductor substrate satisfying a prescribed reference.　The present invention comprises: a measurement step S20 for causing laser light to be incident on a target substrate, and acquiring an intensity measurement value of scattered light scattered on the target substrate; and an estimation step S70 for, using an estimation model created on the basis of a data set in which an intensity measurement value of scattered light scattered on a reference substrate, as measured by causing laser light to be incident on the reference substrate, is associated with a process amount required for a reference index in the reference substrate to satisfy a prescribed reference, estimating a process amount required for the reference index in the target substrate to satisfy the prescribed reference from the intensity measurement value of the scattered light in the target substrate obtained in the measurement step.

Description

Appropriate method for estimating process quantities

[0001] The present invention relates to a technique for estimating an appropriate process amount when processing a semiconductor substrate.

[0002] A technique for evaluating the state of a surface layer of a semiconductor substrate without destroying the semiconductor substrate is known. Patent Document 1 describes a method for evaluating the surface layer (process-affected layer) of a semiconductor substrate, including scratches and distortions, based on the intensity of scattered light from a laser beam incident on the surface of the semiconductor substrate.

[0003] International Publication No. 2022 / 158394

[0004] According to the above-mentioned prior art, the state of the surface layer can be evaluated without destroying the semiconductor substrate.

[0005] Semiconductor substrates made of compound semiconductor materials, such as silicon carbide, are usually formed by slicing an ingot of the semiconductor material and grinding and polishing the surface. These mechanical processes, such as slicing, grinding, and polishing, can cause scratches and crystal distortion in the surface layer of the semiconductor substrate.

[0006] In this specification, "distortion" refers to the deviation of an actual crystal lattice from the standard when an ideal crystal lattice is used as the standard. The amount of distortion is a value indicating the degree of distortion. These scratches and crystal distortions can reduce device yields in device manufacturing processes such as epitaxial growth, which are performed after the machining process. Furthermore, compound semiconductor materials, such as silicon carbide, are difficult to machine due to their hard and brittle physical properties, and the quality of the surface layer (especially the degree of distortion, distortion distribution, etc.) varies depending on the processing method and processing conditions. Therefore, in actual semiconductor device manufacturing, even when focusing on a specific process, the semiconductor substrates subjected to that process usually do not have a consistent quality. In other words, to achieve certain specified quality conditions, there are additional process amounts required depending on the quality of each semiconductor substrate.

[0007] The present invention has been made in consideration of the above-described circumstances, and an object of the present invention is to provide a technology for estimating an appropriate amount of processing to be performed on a semiconductor substrate in order to obtain a semiconductor substrate that satisfies a predetermined standard, based on the intensity of scattered light scattered in the semiconductor substrate.

[0008] [1] A method for estimating an appropriate process quantity to be performed on a semiconductor substrate, comprising: a measurement step of irradiating a laser beam onto a target substrate and acquiring intensity measurements of light scattered by the target substrate; and an estimation step of estimating a process quantity required for a reference indicator of the target substrate to satisfy a predetermined standard from the intensity measurements of the scattered light on the target substrate obtained in the measurement step, using an estimation model created based on a data set correlating intensity measurements of light scattered by the reference substrate, measured by irradiating a laser beam onto the reference substrate, with a process quantity required for a reference indicator of the reference substrate to satisfy the predetermined standard. [2] The estimation method described in [1], wherein the process quantity includes a non-contact process quantity. [3] The estimation method described in [2], wherein the process quantity includes an etching amount. [4] The estimation method described in any of [1] to [3], wherein the reference indicator includes a measurable indicator of a layer formed by performing an irreversible process on the reference substrate or the target substrate. [5] The estimation method according to any one of [1] to [4], wherein the reference index includes a measurable index in a layer formed by performing an irreversible process on the reference substrate or the target substrate after a non-contact process. [6] The estimation method according to any one of [1] to [5], wherein the reference index includes a measurable index in an epitaxially grown layer formed by epitaxially growing the reference substrate or the target substrate. [7] The estimation method according to any one of [1] to [6], wherein the reference index includes a measurable index in an epitaxially grown layer formed by epitaxially growing the reference substrate or the target substrate after etching. [8] The estimation method according to any one of [1] to [7], wherein the reference index includes a killer defect density in an epitaxially grown layer formed by epitaxially growing the reference substrate or the target substrate. [9] The reference indicator includes an IGSF (In-Grown Stacking Fault) density in an epitaxial growth layer formed by epitaxially growing the reference substrate or the target substrate. [1] The estimation method according to any one of [8] to [9].

[10] The estimation method according to any one of [1] to [9], wherein the reference index includes a BPD (Basal Plane Dislocation) density in an epitaxial growth layer formed by epitaxially growing the reference substrate or the target substrate.

[11] The estimation method according to any one of [1] to

[10] , wherein the reference index includes a BPD (Basal Plane Dislocation) propagation rate in an epitaxial growth layer formed by epitaxially growing the reference substrate or the target substrate.

[12] The estimation method according to any one of [1] to

[11] , further comprising a planarization step of planarizing the surface of the target substrate before the measurement step.

[13] The estimation method according to any one of [1] to

[12] , wherein the target substrate has an arithmetic mean roughness Ra of less than 0.8 nm on its surface.

[14] The estimation method according to any one of [1] to

[13] , wherein the laser light is polarized.

[15] The estimation method according to

[14] , wherein the laser light is P-polarized or S-polarized.

[16] A method for manufacturing a semiconductor substrate, comprising: performing processing on the target substrate based on the appropriate process amount to be performed on the target substrate estimated by the estimation method according to any one of [1] to

[15] .

[17] A method for creating an estimation model, comprising: a first measurement step of irradiating a laser beam onto each of a plurality of reference substrates and acquiring intensity measurement values ​​of scattered light scattered on each of the reference substrates; a processing step of performing processing on each of the reference substrates, including processing with different process amounts; a second measurement step of measuring a reference index of each of the reference substrates; an analysis step of creating a first data set correlating the reference index of each of the reference substrates with the process amount performed on each of the reference substrates, and estimating, based on the first data set, the process amount necessary for the reference index of each of the reference substrates to satisfy a predetermined standard; and an estimation model creation step of creating a second data set correlating the process amount necessary for the reference index of each of the reference substrates, estimated in the analysis step, to satisfy the predetermined standard, with the intensity measurement values ​​of the scattered light measured in the first measurement step, and creating, based on the second data set, an estimation model that estimates an appropriate process amount to be performed on the target substrate from the intensity measurement values ​​of the scattered light on the target substrate.

[18] The estimation method according to any one of [1] to

[15] , wherein the estimation model is an estimation model created using the estimation model creation method according to claim 1.

[0009] According to the invention of [1], it is possible to estimate the process amount required for the reference index on the target substrate to satisfy a predetermined standard from the intensity measurement value of the scattered light scattered on the target substrate using an estimation model created based on a data set that associates the intensity measurement value of the scattered light scattered on the reference substrate with the process amount required for the reference index on the reference substrate to satisfy a predetermined standard, thereby making it possible to non-destructively and quickly estimate the appropriate process amount to be performed on the semiconductor substrate.

[0010] The invention according to [2] makes it possible to non-destructively and quickly estimate an appropriate amount of non-contact processing (etching, epitaxial growth, etc.) to be performed on a semiconductor substrate, particularly when it is desired to process a processed-damaged layer without creating a new processed-damaged layer. Furthermore, because the non-contact process does not introduce a new processed-damaged layer into the substrate, the required amount of non-contact processing can be a value indicating the substrate quality.

[0011] According to the invention of [3], it is possible to non-destructively and quickly estimate the appropriate amount of processing to be performed on a semiconductor substrate, particularly when it is desired to remove an appropriate amount of a process-affected layer by etching.

[0012] The invention according to [4] makes it possible to non-destructively and quickly estimate the required process amount from the intensity measurement value of scattered light scattered in the target substrate to improve the quality of the layer formed by performing irreversible processing on the target substrate.

[0013] The invention according to [5] makes it possible to non-destructively and quickly estimate the amount of non-contact processing required to improve the quality of a layer formed on a target substrate by performing an irreversible process on the target substrate after a non-contact process, based on the intensity measurement value of the scattered light scattered on the target substrate.

[0014] According to the invention described in [6], the present estimation method can be used along the device manufacturing process, including the epitaxial growth process.

[0015] According to the invention of [7], the estimation method can be used along the device manufacturing process, including the epitaxial growth process after the etching process.

[0016] According to the invention of [8], when manufacturing a device from a target substrate, it is possible to estimate the process amount that can treat the processing-affected layer that causes serious characteristic defects in the manufactured device, thereby improving the yield of the manufactured devices.

[0017] According to the invention of [9], it is possible to estimate a process amount that can treat a processing-affected layer that generates IGSF in the epitaxial growth layer of a target substrate after epitaxial growth, in particular, and therefore it is possible to improve the quality of the target substrate after processing and increase the yield of manufactured devices.

[0018] The invention according to

[10] makes it possible to estimate a process amount that can treat a process-affected layer that causes BPDs in the epitaxial growth layer of a target substrate after epitaxial growth, in particular, thereby improving the quality of the target substrate after processing and increasing the yield of manufactured devices.

[0019] According to the invention of

[11] , it is possible to estimate a process amount that can treat the processing-affected layer of the target substrate so as to suppress the propagation of BPD to the epitaxial growth layer of the target substrate, particularly after epitaxial growth, and therefore it is possible to improve the quality of the target substrate after processing and increase the yield of the manufactured devices.

[0020] According to the invention of

[12] , in particular, the amount of laser light incident on the inside of the semiconductor substrate is increased, and the appropriate process amount to be performed on the semiconductor substrate can be non-destructively and quickly estimated from the intensity measurement value of scattered light originating from the processing-affected layer inside the semiconductor substrate.

[0021] According to the invention of

[13] , in particular, the amount of laser light incident inside the semiconductor substrate is increased, and the appropriate process amount to be performed on the semiconductor substrate can be non-destructively and quickly estimated from the intensity measurement value of scattered light originating from the processing-affected layer inside the semiconductor substrate.

[0022] According to the invention of

[14] , it is possible to non-destructively and quickly estimate the appropriate process amount to be performed on a semiconductor substrate from the intensity measurement value of scattered light originating from a processing-affected layer at a specific location on the semiconductor substrate, which corresponds to the polarization state of the laser light.

[0023] According to the invention of

[15] , in particular, the amount of laser light incident inside the semiconductor substrate is increased, and the appropriate process amount to be performed on the semiconductor substrate can be non-destructively and quickly estimated from the intensity measurement value of scattered light originating from the processing-affected layer inside the semiconductor substrate.

[0024] According to the invention of

[16] , the target substrate can be processed by the required process amount, and therefore the quality of the processed target substrate can be improved, and the yield of the manufactured devices can be increased.

[0025] According to the invention of

[17] , an estimation model to be used in a method for estimating an appropriate process amount to be performed on a semiconductor substrate can be created.

[0026] The invention according to

[18] makes it possible to non-destructively and quickly estimate the appropriate process amount to be performed on a semiconductor substrate by using an estimation model created based on a data set that associates the intensity measurement values ​​of scattered light scattered on a plurality of reference substrates with the process amount required for the reference indicator on the reference substrate to satisfy a predetermined standard.

[0027] The present invention provides a technique for estimating, based on the intensity of light scattered in a semiconductor substrate, the appropriate amount of processing to be performed on the semiconductor substrate in order to obtain the semiconductor substrate that satisfies predetermined conditions.

[0028] FIG. 1 is a diagram illustrating a (first) measurement step according to an embodiment of the present invention. FIG. 2 is a diagram illustrating a (first) measurement step according to an embodiment of the present invention. FIG. 3 is a diagram illustrating scattered light intensity measured on a plurality of reference substrates according to an embodiment of the present invention. FIG. 4 is a diagram illustrating an example of a first dataset correlating the etching amount and IGSF density for a plurality of reference substrates according to an embodiment of the present invention. FIG. 5 is a diagram illustrating an example of a first dataset correlating the etching amount and BPD density for a plurality of reference substrates according to an embodiment of the present invention. FIG. 6 is a diagram illustrating an example of a first dataset correlating the etching amount and BPD propagation rate for a plurality of reference substrates according to an embodiment of the present invention. FIG. 7 is a diagram illustrating an etching amount required for an epitaxial growth layer of a reference substrate to satisfy a predetermined standard according to an embodiment of the present invention. FIG. 8 is a diagram illustrating an example of a second dataset and an estimation model according to an embodiment of the present invention. FIG. 9 is a diagram illustrating an example of a second dataset and an estimation model according to an embodiment of the present invention. FIG. 10 is a distribution diagram of scattered light intensity and a required process amount according to an embodiment of the present invention. FIG. 11 is a diagram illustrating a relationship between scattered light intensity and surface roughness of a substrate when S-polarized or P-polarized light is incident on a substrate according to an embodiment of the present invention. FIG. 12 is a diagram illustrating a relationship between scattered light intensity and surface roughness of a substrate when S-polarized or P-polarized light is incident on a substrate according to an embodiment of the present invention.

[0029] The present invention includes a measurement step of irradiating a laser beam onto a target substrate and obtaining intensity measurements of the scattered light scattered on the target substrate, and an estimation step of estimating, using an estimation model described below, the process amount required for the reference indicator on the target substrate to satisfy a predetermined standard from the intensity measurements of the scattered light on the target substrate.

[0030] In the present invention, the process amount is preferably the amount of change in thickness of the semiconductor substrate. More specifically, the process amount is the amount of thickness reduction, i.e., the etching amount, or the amount of thickness increase, i.e., the growth amount. That is, the etching amount can be expressed as "-3 μm" or "-4 μm," and the growth amount can be expressed as "+10 μm" or "+15 μm." Alternatively, after defining the processing method, the process amount can be expressed as the absolute amount of thickness change, such as "3 μm," "4 μm," "10 μm," or "15 μm." Furthermore, in the present invention, the process amount may be a process condition such as the time or temperature of a specific process.

[0031] Furthermore, for example, the process amount may be information indicating both the type of process and its amount. That is, the process amount may be information such as "etching: -3 μm," "thermal etching: 4 μm," "epitaxial growth: +10 μm," or "sublimation: 15 μm." Alternatively, the processing conditions (processing temperature, processing time, etc.) during the process may also be treated as indirect process amounts.

[0032] In the present invention, the second data set is correlation data that associates the intensity measurement value of scattered light scattered on a reference substrate, measured by irradiating laser light onto the reference substrate, with the process amount required for the reference indicator on the reference substrate to satisfy a predetermined standard.

[0033] The measured intensity of scattered light may be expressed as numerical data such as statistical values, such as position parameters (e.g., including one or more selected from the mean (arithmetic, geometric, harmonic), median (quantile, order statistics), mode, class value, etc.) obtained by statistically processing a plurality of measured values ​​measured over the entire main surface of the substrate, or scale parameters (e.g., including one or more selected from the variance, deviation value, standard deviation, mean deviation, median absolute deviation, range, half-width, etc.). Furthermore, the measured intensity of scattered light may be expressed as image data such as a distribution image in which a plurality of measured values ​​measured over the entire main surface of the substrate are mapped to the substrate, or a histogram of these plurality of measured values.

[0034] The second data set is a data set that associates the scattered light intensity measurement values ​​with the required process variables. Therefore, for example, if the scattered light intensity measurement values ​​are expressed as numerical data such as the above-mentioned statistical values, an estimation model that uses the scattered light intensity measurement values ​​as input values ​​and estimates the required process variables as output values ​​can be created based on regression analysis of the second data set. Furthermore, for example, if the scattered light intensity measurement values ​​are expressed as image data such as the above-mentioned distribution image or histogram of multiple measurement values, an estimation model that uses the scattered light intensity measurement values ​​as input values ​​and estimates the required process variables as output values ​​can be created based on machine learning using the second data set as training data.

[0035] The process amount may be either the amount of a contact process or the amount of a non-contact process. A contact process is a process, such as mechanical polishing (MP) or chemical mechanical polishing (CMP), in which the substrate main surface is processed by bringing another member into contact with the substrate main surface, and can also be called mechanical processing. A non-contact process is a process, such as etching (hydrogen etching, thermal etching, Si vapor pressure etching, wet etching, etc.), epitaxial growth (CVD: chemical vapor deposition method, sublimation method, etc.), or annealing, in which the substrate main surface is processed without bringing another member into contact with the substrate main surface.

[0036] In a preferred embodiment of the present invention, the estimation model is created using non-contact process quantities. This method allows for estimation of the appropriate process quantities when using a non-contact process. In other words, the estimation model created using non-contact process quantities is an estimation model that does not intend for new scratches or distortions to be introduced by subsequent processes. Therefore, it can be used as a model for estimating the depth of a surface layer (process-affected layer) on a target substrate that should be removed and that adversely affects devices. Therefore, in an embodiment using an estimation model created using non-contact process quantities, the present invention can also be used as a method for estimating the depth of a surface layer (process-affected layer) that should be removed and that adversely affects devices. Of course, creating an estimation model using a specific contact process quantity allows for estimation of the extent to which a contact process should remove the surface layer. Therefore, if it is known in advance that the subsequent process will be a contact process, it is preferable to create an estimation model using this method.

[0037] The amount of in-grown stacking faults (IGSFs), which are a type of killer defect, is directly affected by the depth of the surface layer (process-affected layer) to be removed. Therefore, the present invention may be able to estimate the amount of IGSFs that would be produced if the target substrate were subjected to epitaxial growth as is, from the depth of the surface layer (process-affected layer) to be removed on the target substrate, which is estimated using an estimation model created using non-contact process amounts.

[0038] The method for creating an estimation model according to the present invention includes a first measurement step of irradiating a laser beam onto each of a plurality of reference substrates having different qualities and acquiring intensity measurements of scattered light scattered on each reference substrate; a processing step of performing processing on each reference substrate, including processing with different process amounts; a second measurement step of measuring a reference indicator of each reference substrate; an analysis step of creating a first data set that associates the reference indicator of each reference substrate with the process amount performed on each reference substrate, and estimating, based on the first data set, the process amount required for the reference indicator of each reference substrate to satisfy a predetermined standard; and an estimation model creation step of creating a second data set that associates the process amount required for the reference indicator of each reference substrate estimated in the analysis step to satisfy the predetermined standard with the intensity measurements of scattered light, and creating, based on the second data set, an estimation model that estimates an appropriate process amount to be performed on the target substrate from the intensity measurements of scattered light on the target substrate.

[0039] In the present invention, the multiple reference substrates are prepared as multiple substrates of various qualities, and are classified into multiple substrate groups each containing substrates of similar quality based on the similarity of their quality. For example, assume that substrates a1, a2, a3, etc. are classified into substrate group A, and substrates b1, b2, b3, etc. are classified into substrate group B. In this case, substrates a1, a2, a3, etc. have similar quality, and substrates b1, b2, b3, etc. also have similar quality. On the other hand, any of substrates a1, a2, a3, etc. and any of substrates b1, b2, b3, etc. have different quality.

[0040] In an embodiment of the present invention, the plurality of reference substrates are classified into a plurality of substrate groups based on the manufacturers of the respective substrates and the product grades offered by those manufacturers. Alternatively, the plurality of reference substrates may be classified into a plurality of substrate groups based on the measured scattered light intensity of each of the reference substrates as a quality criterion.

[0041] In the present invention, the reference indicator includes indicators (killer defect density, BPD propagation rate, ion concentration, doping concentration, etc.) in a layer formed by further irreversible processing (epitaxial growth, doping, ion implantation, or subsequent annealing, oxide film formation, etc.) when the reference substrate after the contact process or non-contact process is subjected to the relevant processing. In other words, an irreversible processing is defined as a processing that is substantially impossible to redo on the same substrate due to economical efficiency, technical difficulty, or other reasons. For example, the IGSF density in an epitaxial growth layer formed when epitaxial growth is performed on an etched reference substrate can be the reference indicator.

[0042] In the present invention, the first data set is a data set that associates reference indices measured on each reference substrate with process variables that indicate different amounts of processing performed on each reference substrate. Thus, a model that uses desired reference indices as input values ​​and estimates required process variables as output values ​​can be created based on a regression analysis or the like of the first data set.

[0043] In the present invention, the second dataset is a dataset that associates intensity measurements of scattered light scattered on a reference substrate, measured by irradiating the reference substrate with the process variables required for the reference substrate to have a desired reference index, estimated from the first dataset. Therefore, for example, if the intensity measurements of scattered light are expressed as numerical data such as the statistical values ​​described above, an estimation model can be created based on regression analysis of the second dataset, using the intensity measurements of scattered light as input values ​​and estimating the required process variables as output values. Furthermore, for example, if the intensity measurements of scattered light are expressed as image data such as a distribution image or histogram of the above-described multiple measurement values, an estimation model can be created based on machine learning of the second dataset, using the intensity measurements of scattered light as input values ​​and estimating the required process variables as output values.

[0044] Hereinafter, with reference to the accompanying drawings, preferred embodiments of a method for estimating appropriate process quantities to be applied to a semiconductor substrate (hereinafter referred to as a process quantity estimation method), a method for manufacturing a semiconductor substrate, and a method for creating an estimation model according to the present invention will be described in detail. The technical scope of the present invention is not limited to the embodiments shown in the accompanying drawings, and can be modified as appropriate within the scope of the claims. Note that in the following description of the embodiments and the accompanying drawings, similar components are designated by the same reference numerals, and duplicated explanations will be omitted.

[0045] <<Method for Creating an Estimation Model>> Below, we will explain a method for creating an estimation model using a plurality of reference substrates consisting of substrate groups A to F. In this embodiment, the plurality of reference substrates are classified into substrate groups A to F based on the manufacturers of each substrate and the product grades provided by those companies.

[0046] <Cleaning / Planarization Step S10> The method for creating an estimation model according to the present invention may include a cleaning / planarization step S10, which includes a cleaning step S11 and a planarization step S12, depending on the surface condition of the semiconductor substrate 100 (reference substrate). The cleaning step S11 is a step for eliminating factors that cause the laser light L1 to be scattered by the surface 101 of the semiconductor substrate 100, such as organic contamination, particle contamination, an oxide layer, or ion contamination adhering to the surface of the semiconductor substrate 100. In particular, particles adhering to the surface of the semiconductor substrate 100 cause scattering that is stronger than the scattered light L4 generated by the process-affected layer 102, thereby reducing the correlation between the scattered light intensity and the process amount. Therefore, it is desirable to remove particle contamination and the like before measuring the scattered light intensity.

[0047] The cleaning step S11 has the effect of suppressing the generation of scattered light originating from surface deposits or reducing the intensity of scattered light originating from surface deposits, thereby reducing the proportion of scattered light originating from surface deposits in the scattered light L4 and improving the correlation between the measured intensity of scattered light and the process variable.

[0048] The cleaning step S11 can be performed by any method that can remove at least one of organic contamination, particle contamination, oxide layer, and ion contamination attached to the surface 101 of the semiconductor substrate 100. For example, a general RCA cleaning (NH ４ OH, H ２ O ２ , H ２ Alternatively, chemical cleaning such as acid cleaning (HCl, HF) or physical cleaning using bubbles or a brush can be used. Naturally, any other method can be used as long as it can clean the surface 101 of the semiconductor substrate 100.

[0049] The planarization step S12 is a step of eliminating factors that cause the laser light L1 to be scattered by the surface 101 of the semiconductor substrate 100, such as surface irregularities of the semiconductor substrate 100. In particular, if the surface of the semiconductor substrate 100 is rough, scattering occurs that is stronger than the scattered light L4 generated by the process-affected layer 102, thereby reducing the correlation between the scattered light intensity and the process amount. Therefore, it is desirable that the surface irregularities be sufficiently planarized before measuring the scattered light intensity.

[0050] The planarization step S12 has the effect of suppressing the generation of scattered light due to surface irregularities or reducing the intensity of scattered light due to surface irregularities, thereby reducing the proportion of scattered light due to surface irregularities in the scattered light L4 and improving the correlation between the measured intensity of scattered light and the process amount.

[0051] The planarization step S12 can be performed by any method capable of planarizing the irregularities on the surface 101 of the semiconductor substrate 100. For example, mechanical processing methods such as chemical mechanical polishing, various etching methods, etc. can be used. Naturally, any other method capable of planarizing the irregularities on the surface 101 of the semiconductor substrate 100 can be used.

[0052] The cleaning step S11 and the planarization step S12 according to the embodiment have been described above. Note that the estimation model creation method according to the present invention can also be exemplified as a form that does not include the cleaning step S11 and / or the planarization step S12. That is, if the surface 101 of the semiconductor substrate 100 is sufficiently clean or flat, the cleaning step S11 and / or the planarization step S12 may not be performed, and the method may proceed to the subsequent first measurement step S20.

[0053] <First Measuring Step S20> FIG. 1 is a diagram illustrating the first measuring step S20 performed in the estimation model creating method according to the embodiment of the present invention.

[0054] The first measurement step S20 obtains an intensity measurement value of scattered light L4 scattered by the surface 101 of the semiconductor substrate 100 (reference substrate) or the process-affected layer 102 present below the surface 101. Specifically, laser light L1 irradiated from the light projection system 10 is incident on the surface 101 of the semiconductor substrate 100. Incident light L3 incident on the semiconductor substrate 100 is scattered by the process-affected layer 102, generating scattered light L4. The light receiving system 20 receives and measures this scattered light L4, and obtains an intensity measurement value of the scattered light L4.

[0055] For example, the light projection system 10 used in the first measurement step S20 includes a laser output unit 11 and a wave plate 12. The light projection system 10 is attached so that the laser light L1 is incident at an incident angle θ that is inclined with respect to the normal N to the surface of the semiconductor substrate 100.

[0056] The light projection system 10 may be mounted so that the incident angle θ can be adjusted.

[0057] The laser output unit 11 is a source of the laser light L1, and may be, for example, a gas laser such as a He-Ne laser, a semiconductor laser, or a solid-state laser such as a YAG laser.

[0058] The wave plate 12 performs adjustments such as wavelength separation, fine adjustment, ellipticity adjustment, and polarization rotation on the laser light L1 generated by the laser output unit 11. This wave plate 12 can switch between S-polarized and P-polarized laser light L1, and can control the penetration characteristics of the laser light L1 without changing measurement conditions such as the incident angle θ and wavelength λ. Appropriate materials and configurations for this wave plate 12 are selected depending on the material and conditions of the semiconductor substrate 100 to be measured.

[0059] For example, the light-receiving system 20 used in the first measurement step S20 includes an objective lens 21, an imaging lens 22, a beam splitter 23, a light-receiving sensor 24, and a slit 25. It is desirable that this light-receiving system 20 be attached at a measurement angle φ that is different from the incident angle θ so that reflected light L2 does not enter.

[0060] The light receiving system 20 may be mounted so that the measurement angle φ can be adjusted.

[0061] It is desirable to use a cube-type beam splitter as the beam splitter 23. The light receiving sensor 24 can be configured to convert the intensity of the scattered light L4 into an electrical signal and can use a photomultiplier tube, a photodiode, or the like. The slit 25 is arranged to determine the inspection range of the laser light L1.

[0062] In the method for creating an estimation model according to the embodiment of the present invention, the intensity measurement value of scattered light L4 is obtained for each of a plurality of reference substrates using the above-described measurement system (light projection system, light receiving system) (first measurement step S20).

[0063] The scattered light L4 may include surface scattered light and internal scattered light. The surface scattered light may include scattered light derived from surface deposits, scattered light derived from surface irregularities, and scattered light derived from crystal defects. The internal scattered light may include scattered light derived from crystal defects and distortion.

[0064] The semiconductor substrate 100 is a semiconductor material manufactured through a process that includes at least one of a slicing process, a polishing process, and a grinding process. An example of a semiconductor substrate 100 manufactured through these processes is a silicon (Si) substrate. The slicing process, polishing process, and grinding process described above may introduce a process-affected layer 102 that adversely affects device manufacturing. Recent research has revealed that, in particular, the process-affected layer 102 adversely affects devices in compound semiconductor materials, which are classified as difficult-to-process materials.

[0065] The method for estimating process quantities, the method for manufacturing a semiconductor substrate, and the method for creating an estimation model according to the present invention are applicable to silicon carbide (SiC) substrates, gallium nitride (GaN) substrates, aluminum nitride (AlN) substrates, gallium oxide (Ga ２ O ３ The present invention is suitable for use with compound semiconductor substrates such as silicon dioxide substrates, sapphire substrates, and the like.

[0066] The semiconductor substrate 100 has a surface 101 planarized by chemical mechanical polishing or the like, a process-affected layer 102 introduced by processes including mechanical processing such as slicing, grinding, and polishing, and a bulk layer 103. The process-affected layer 102 and the bulk layer 103 are not clearly separated from each other by a boundary, but rather the process-affected layer 102, in which the degree of strain decreases toward the interior, is considered to be continuously distributed with an unclear boundary between the process-affected layer 102 and the bulk layer 103. In other words, the process-affected layer 102 is a layer having a relatively large amount of strain (enough to affect device fabrication), while the bulk layer 103 is a layer having a relatively small or negligible amount of strain, and it can be said that there is no clear boundary separating the two.

[0067] The surface 101 is preferably planarized by a planarization process such as chemical mechanical polishing (CMP). The preferred flatness of the surface 101 will be described later.

[0068] The laser light L1 enters the semiconductor substrate 100 from the surface 101. That is, the surface 101 is the boundary surface onto which the laser light L1 is incident. Therefore, when the semiconductor substrate 100 has the surface 101 planarized by the planarization process, scattered light caused by the unevenness (roughness) of the surface 101 is less likely to be generated, or the intensity of the scattered light generated at the surface 101 is relatively low. In other words, the planarized surface 101 has the effect of suppressing the generation of scattered light caused by the surface unevenness or reducing the intensity of the scattered light caused by the surface unevenness. As a result, the proportion of scattered light generated in the process-affected layer 102 below the surface 101 in the scattered light L4 increases, and the measured intensity of the scattered light shows a good correlation with the process amount.

[0069] 2A and 2B are explanatory diagrams illustrating the relationship between the polarization of the laser light L1 and the processing-affected layer 102 in the first measurement step S20. Fig. 2A is an explanatory diagram illustrating the case where the semiconductor substrate 100 is irradiated with S-polarized laser light L1. Fig. 2B is an explanatory diagram illustrating the case where the semiconductor substrate 100 is irradiated with P-polarized laser light L1.

[0070] The first measurement step S20 may include a measurement step S21 using S-polarized light, in which S-polarized laser light L1 is incident on the surface 101 of the semiconductor substrate 100 having the processing-affected layer 102, and the intensity of scattered light L4 scattered below the surface 101 is measured (see FIG. 2( a)). Note that the electric field of the S-polarized laser light L1 oscillates perpendicularly to the incident surface, and the S-polarized light has a lower transmittance than P-polarized light irradiated under the same conditions. Therefore, considering the total amount and energy of the incident light L3, S-polarized light is more likely to detect scattered light generated in a shallower portion than P-polarized light, and S-polarized light is therefore suitable for measuring shallow portions of the processing-affected layer 102.

[0071] Specifically, S-polarized light is suitable for measurements in depth ranges of up to 10 μm, up to 8 μm, up to 6 μm, or up to 4 μm from the substrate surface. That is, in measurement step S21 using S-polarized light, S-polarized laser light L1 is incident on surface 101 of semiconductor substrate 100 having processing-affected layer 102, and intensity measurements of scattered light L4 generated below surface 101, mainly in a depth range of about up to 10 μm, can be obtained.

[0072] The first measurement step S20 may also include a measurement step S22 using P-polarized light, in which P-polarized laser light L1 is incident on the surface 101 of the semiconductor substrate 100 and the intensity of scattered light L4 scattered below the surface 101 is measured (see FIG. 2( b)). Note that the electric field of P-polarized laser light L1 oscillates parallel to the plane of incidence, and P-polarized light has a higher transmittance than S-polarized light irradiated under the same conditions. Therefore, considering the total amount and energy of the incident light L3, P-polarized light is more likely to detect scattered light generated in a relatively deeper portion than S-polarized light, and P-polarized light is therefore suitable for measuring deep portions of the process-affected layer 102.

[0073] Specifically, P-polarized light is suitable for measurements within a range of up to 47 μm, up to 30 μm, up to 20 μm, or up to 10 μm from the substrate surface. That is, in measurement step S22 using P-polarized light, P-polarized laser light L1 is incident on surface 101 of semiconductor substrate 100 having processing-affected layer 102, and intensity measurements of scattered light L4 generated within a depth range of approximately up to 47 μm below surface 101 can be obtained.

[0074] In an embodiment of the present invention, the first measurement step S20 as described above obtains scattered light intensity measurements (scattered light intensity) for each reference substrate in substrate groups A to F when P-polarized light and / or S-polarized light is incident thereon ( FIG. 3 ). Note that in this embodiment, the multiple reference substrates are classified into substrate groups A to F based on the manufacturers of each substrate and the product grades provided by those manufacturers, but it is also possible to reclassify each substrate into substrate groups A to F based on the scattered light intensity when P-polarized light and / or S-polarized light is incident on the substrate, for example.

[0075] <Processing Step S30> The processing step S30 includes a first processing step S31 in which a plurality of reference substrates of comparable quality, from which scattered light intensity measurements have been obtained in the first measurement step S20, are subjected to different process amounts for each substrate, and a second processing step S32 in which the same process amount of irreversible processing (epitaxial growth, doping, ion implantation, subsequent annealing, oxide film formation, etc.) is performed for each substrate. The first processing step S31 according to an embodiment of the present invention performs different amounts of etching on each of the plurality of reference substrates. The second processing step S32 according to an embodiment of the present invention performs epitaxial growth under the same conditions on each of the plurality of reference substrates that have undergone different amounts of etching in the first processing step S31, thereby achieving the same amount of epitaxial growth.

[0076] <Second Measurement Step S40> The second measurement step S40 is a step of measuring a reference index for each of a plurality of reference substrates. In the second measurement step S40 according to the embodiment of the present invention, for each of a plurality of reference substrates that have undergone different amounts of etching and the same amount of epitaxial growth in the processing step S30, the killer defect density (including IGSF density, BPD density, etc.), BPD propagation rate, BPD conversion rate, etc. in the epitaxial growth layer are measured as reference indexes. Note that killer defects are known as a general term for defects present in a substrate that cause serious characteristic defects in semiconductor devices fabricated from that substrate.

[0077] <Analysis step S50> The analysis step S50 includes a step S51 of creating a first data set that associates, for each of a plurality of reference substrates, the reference index measured in the second measurement step S40 with the process amount performed in the first processing step S31, and a step S52 of estimating, based on the first data set, the process amount required for the reference index of each reference substrate to satisfy a predetermined standard.

[0078] In step S51 according to an embodiment of the present invention, a first data set is created for each of a plurality of reference substrates that have undergone the second measurement step S40, correlating the killer defect density (including IGSF density and BPD density) in the epitaxial growth layer with the process amount performed in the processing step S30. For example, as shown in FIG. 4, a first data set can be created that correlates the etching amount performed in the first processing step S31 with the SF density. Alternatively, as shown in FIG. 5, a first data set can be created that correlates the etching amount performed in the first processing step S31 with the BPD density. Furthermore, as shown in FIG. 6, a first data set can be created that correlates the etching amount performed in the first processing step S31 with the BPD propagation rate.

[0079] In step S52, a process amount required for the reference index of each reference substrate to satisfy a predetermined standard is estimated. The predetermined standard according to this embodiment is a process amount required for the SF density (reference index) in the epitaxial growth layer to satisfy a predetermined value (0.5 cm in this embodiment). －２ ) or less, and the BPD propagation rate (reference index) in the epitaxial growth layer is a predetermined value (0.1% in this embodiment) or less. Note that the above-mentioned predetermined criteria are merely examples. The predetermined criteria, including the reference index and the predetermined value that the reference index should aim for, can be freely determined depending on the desired substrate quality. For example, if the SF density (reference index) in the epitaxial growth layer is 0.1 cm －２ The predetermined standard may be set to be equal to or less than this.

[0080] In step S52 according to an embodiment of the present invention, a process amount required for the reference indicators of each reference substrate to satisfy a predetermined standard is estimated from the exponential relationship between the killer defect density (IGSF density, BPD density) or BPD propagation rate and the etching amount, as shown in Figures 4 to 6. In this embodiment, in Figures 4 to 6, the killer defect density (IGSF density, BPD density) and the BPD propagation rate show an exponential decrease with respect to the etching amount performed in the first processing step S31. Therefore, the process amount required for the reference indicators of each reference substrate to satisfy a predetermined standard can be estimated from an approximation curve such as that shown by the dashed line in each figure or an approximation formula showing the approximation curve.

[0081] By the analysis step S50 according to the embodiment of the present invention, the SF density in the epitaxial growth layer of each reference substrate in the substrate groups A to F was adjusted to 0.5 cm －２ The etching amount required to make the BPD density or other killer defect density in the epitaxial growth layer of each reference substrate in substrate groups A to F can also be estimated in a similar manner.

[0082] <Estimation model creation process S60> The estimation model creation process S60 includes a process S61 of creating a second data set that associates the process amount required for the reference indicator of each reference substrate to satisfy a predetermined standard, as estimated in the analysis process S50, with the scattered light intensity measurement value for each reference substrate obtained in the first measurement process S20, and a process S62 of creating an estimation model that estimates the required process amount from the scattered light intensity measurement value based on the second data set.

[0083] In step S61 according to the embodiment of the present invention, as shown in FIG. 8, the IGSF density in the epitaxial growth layer of each reference substrate is set to 0.5 cm －２ A second data set is created that correlates the amount of etching required to achieve the following with the intensity measurements of scattered light when P-polarized and S-polarized light is incident on each reference substrate:

[0084] In step S62 according to the embodiment of the present invention, the IGSF density in the epitaxial growth layer of each reference substrate shown in FIG. 8 is set to 0.5 cm －２ An estimated model curve is created by regression analysis from multiple data points indicating the etching amount required to achieve a BPD propagation rate of 0.1% or less in the epitaxial growth layer of each reference substrate and the measured scattered light intensity when P-polarized light and S-polarized light are incident on each reference substrate (dashed line in FIG. 8 ). Alternatively, in step S61 according to an embodiment of the present invention, as shown in FIG. 9 , a second data set may be created that associates the etching amount required to achieve a BPD propagation rate of 0.1% or less in the epitaxial growth layer of each reference substrate with the measured scattered light intensity when P-polarized light and S-polarized light are incident on each reference substrate.

[0085] In step S62 according to an embodiment of the present invention, an estimated model curve may be created by regression analysis from multiple data points showing the etching amount required to make the BPD propagation rate in the epitaxial growth layer of each reference substrate 0.1% or less, and the measured scattered light intensity when P-polarized light and S-polarized light are incident on each reference substrate, as shown in FIG. 9 (dashed line in FIG. 9 ).

[0086] Similarly, an estimated model curve can be created by regression analysis from a data set indicating the etching amount required to reduce the BPD density or other killer defect density in the epitaxial growth layer of the reference substrate to a predetermined value or less and the scattered light intensity measurements when P-polarized light and S-polarized light are incident on each reference substrate. Furthermore, the scattered light intensity measurements may be either one or both of the measurements when P-polarized light or S-polarized light is incident on each reference substrate. However, since S-polarized light has better measurement sensitivity for shallower portions of the substrate, when using either one of the measurements when P-polarized light or S-polarized light is incident on each reference substrate as the scattered light intensity measurement, it is preferable to use the measurement when S-polarized light is incident.

[0087] By using the estimation model created by the above method, it is possible to estimate the process amount required for the reference indicator of a certain substrate to satisfy a predetermined standard from the measured value of the scattered light intensity of the substrate.

[0088] <<Process Amount Estimation Method>> <Cleaning / Planarization Step S10>> The process amount estimation method according to the present invention may include a cleaning / planarization step S10 including the cleaning step S11 and / or the planarization step S12 described above, depending on the surface state of the semiconductor substrate 100, which is the target substrate for which the required process amount is to be estimated.

[0089] <Measurement Step S20> The method for estimating a process quantity according to the present invention includes a measurement step S20 for obtaining a measurement value of the intensity of scattered light L4 scattered by surface 101 of semiconductor substrate 100, which is a target substrate for which a required process quantity is to be estimated, or by process-affected layer 102 present below surface 101. The specific method is similar to the first measurement step S20 described above, except that semiconductor substrate 100 is the target substrate, and therefore a description thereof will be omitted.

[0090] <Estimation step S70> The estimation step S70 is a step of estimating the process amount required for the reference indicator on the target substrate to satisfy a predetermined standard from the intensity measurement value of scattered light on the target substrate obtained in the measurement step S20, based on the estimation model created in advance by the estimation model creation method (S10 to S60) described above.

[0091] In the estimation step S70 according to the embodiment of the present invention, the IGSF density in the epitaxial growth layer of the target substrate is estimated to be 0.5 cm 3 from the intensity measurement of scattered light when P-polarized light and / or S-polarized light is incident on the target substrate. －２ An appropriate amount of etching to be performed on the target substrate is estimated to make the BPD propagation rate in the epitaxial growth layer of the target substrate 0.1% or less.

[0092] That is, the predetermined standard according to the embodiment of the present invention is "IGSF density (standard index) in the epitaxial growth layer of the target substrate is 0.5 cm －２ or less" or "the BPD propagation rate (reference index) in the epitaxial growth layer of the target substrate is 0.1% or less." The above-mentioned predetermined criteria can be selected from those set in the analysis step S50 in the method for creating the estimation model.

[0093] For example, "the IGSF density in the epitaxial growth layer of the target substrate is set to 0.5 cm －２For the purpose of achieving "less than" the above, the process amount required for the target substrate can be estimated based on the estimation model shown in FIG.

[0094] 8, when the scattered light intensity measurement value when P-polarized light is incident on the target substrate is about 0.002 mV, the process amount required for the target substrate can be estimated to be about 0.5 μm (FIG. 8(a)). Also, when the scattered light intensity measurement value when S-polarized light is incident on the target substrate is about 0.015 mV, the process amount required for the target substrate can be estimated to be about 0.5 μm (FIG. 8(b)).

[0095] Furthermore, for example, with the objective of "keeping the BPD propagation rate in the epitaxial growth layer of the target substrate at 0.1% or less," the amount of processing required for the target substrate can be estimated based on the estimation model shown in FIG.

[0096] 9, when the scattered light intensity measurement value when P-polarized light is incident on the target substrate is about 0.002 mV, the process amount required for the target substrate can be estimated to be about 1.0 μm (FIG. 9(a)). Also, when the scattered light intensity measurement value when S-polarized light is incident on the target substrate is about 0.015 mV, the process amount required for the target substrate can be estimated to be about 1.0 μm (FIG. 9(b)).

[0097] When estimating the process quantity using both P-polarized light and S-polarized light, the average of the process quantities estimated separately using P-polarized light and S-polarized light may be used as the estimated value of the process quantity. For example, according to the estimation model shown in FIG. 8 , if the mode of the P-polarized scattered light intensity measurements on the target substrate is approximately 0.005 mV, the required process quantity is estimated to be approximately 1.0 μm. Furthermore, if the mode of the S-polarized scattered light intensity measurements on the target substrate is approximately 0.015 mV, the required process quantity for the target substrate is estimated to be approximately 0.5 μm. In this case, the final estimated value of the process quantity may be approximately 1.25 μm, calculated by averaging the required process quantity for the target substrate estimated from the mode of the P-polarized scattered light intensity measurements on the target substrate (approximately 1.0 μm) and the required process quantity for the target substrate estimated from the mode of the S-polarized scattered light intensity measurements on the target substrate (approximately 0.5 μm). This allows for more accurate estimation of the process quantity required for the target substrate.

[0098] Furthermore, when estimating the process quantity using both P-polarized light and S-polarized light, the smaller of the process quantities estimated separately using P-polarized light and S-polarized light may be used as the estimated process quantity. For example, according to the estimation model shown in FIG. 8 , if the most frequent value of the scattered light intensity measurements of P-polarized light on the target substrate is approximately 0.005 mV, the process quantity required for the target substrate is estimated to be approximately 1.0 μm. Furthermore, if the most frequent value of the scattered light intensity measurements of S-polarized light on the target substrate is approximately 0.015 mV, the process quantity required for the target substrate is estimated to be approximately 0.5 μm. In this case, the final estimated process quantity may be approximately 0.5 μm. This minimizes the process quantity for the target substrate, improving yield.

[0099] Furthermore, when estimating the process amount using both P-polarized light and S-polarized light, the larger of the process amounts estimated separately using P-polarized light and S-polarized light may be used as the estimated process amount. For example, according to the estimation model shown in FIG. 8 , if the most frequent value of the P-polarized scattered light intensity measurement value on the target substrate is approximately 0.005 mV, the process amount required for the target substrate is estimated to be approximately 1.0 μm. Furthermore, if the most frequent value of the S-polarized scattered light intensity measurement value on the target substrate is approximately 0.015 mV, the process amount required for the target substrate is estimated to be approximately 0.5 μm. In this case, the final estimated process amount may be approximately 1.0 μm. This allows the process amount for the target substrate to be slightly excessive, thereby more reliably removing killer defects on the target substrate.

[0100] As described with reference to FIG. 2 , P-polarized light and S-polarized light have different suitable measurement depth ranges. Therefore, when estimating a process amount using both P-polarized light and S-polarized light, if the process amounts estimated separately using P-polarized light and S-polarized light differ, either one of the process amounts estimated using P-polarized light or S-polarized light may be prioritized as the estimated value. For example, as in this embodiment, if the required etching depth is in the range of 10 μm or less, measurement using S-polarized light is suitable, and therefore the process amount estimated using S-polarized light may be prioritized as the estimated value.

[0101] FIG. 10 shows, by color coding, the scattered light intensity ((a), (b)) and the required process amount ((e) to (f)) for the entire surface of each target substrate manufactured by each manufacturer, Vender-A and Vender-B, according to an embodiment of the present invention.

[0102] As shown in Figures 10(a) and 10(b), the measured scattered light intensity is not uniform even for a single substrate. In other words, it is thought that the process amount required to satisfy a predetermined standard differs for each part of the substrate. Therefore, as shown in Figures 10(e) and 10(f), a distribution map of the process amount estimated in the estimation step S70 can be created (mapping step S71).

[0103] The mapping step S71 is a step of creating a distribution map of the required process amounts estimated for each portion of the substrate based on a plurality of threshold values, thereby making it possible to visually and stepwise identify the required process amounts estimated for each portion of the substrate.

[0104] For example, the mapping step S71 can display, for each target substrate, the distribution of etching amounts required for the target substrate to make the IGSF density in the epitaxial growth layer equal to or less than a predetermined value or the distribution of etching amounts required for the target substrate to make the BPD propagation rate in the epitaxial growth layer equal to or less than a predetermined value, along with a scale bar that divides the target substrate into unit thicknesses within a predetermined range of processing thicknesses. Note that the process amount may be process conditions such as the time and temperature of a specific process, and the upper and lower limit ranges and division units of the scale bar may be arbitrarily determined in accordance with the process conditions.

[0105] According to this embodiment, for each target substrate manufactured by each manufacturer, Vender-A and Vender-B, the IGSF density in the epitaxial growth layer is set to 0.5 cm －２ The distribution of the etching amount required for the target substrate to achieve a BPD propagation rate of 0.1% or less (FIGS. 10(c) and (d)) and the distribution of the etching amount required for the target substrate to achieve a BPD propagation rate of 0.1% or less in the epitaxial growth layer (FIGS. 10(e) and (f)) can be displayed together with a scale bar divided into 0.5 μm units in the processing thickness range of the target substrate from 0 to 4 μm. Alternatively, ...c) and (d)) can be displayed together with a scale bar divided into 0.5 μm units in the processing thickness range of the target substrate from 0 to 4 μm. －２ It is also possible to set the predetermined standard as being equal to or less than 0.1 cm and perform the above estimation model creation method and process amount estimation method (S10 to S70). In this case, the IGSF density in the epitaxial growth layer is set to 0.1 cm －２ The distribution of the etching amount required for the target substrate to achieve the following (FIGS. 10(g) and (h)) can be displayed together with a scale bar divided into 0.5 μm increments within the processing thickness range of the target substrate, 0 to 4 μm.

[0106] <Semiconductor Substrate Manufacturing Method> <Manufacturing Process S80> The manufacturing process S80 according to the present invention is a process of processing a target substrate based on the appropriate process amount to be applied to the target substrate, which is estimated by the above-described method for estimating the process amount.

[0107] In the manufacturing process S80 according to the embodiment of the present invention, the IGSF density in the epitaxial growth layer estimated in the estimation process S70 is set to 0.5 cm －２ The target substrate is etched based on the amount of etching that is appropriate for the target substrate to have a BPD propagation rate of 0.1% or less in the epitaxially grown layer.

[0108] For example, the manufacturing process S80 can process the target substrate by the same amount as the appropriate process amount to be performed on the target substrate estimated in the estimation process S70. That is, if the estimation process S70 estimates that the etching amount required for the target substrate is about 1 μm based on the estimation model shown in FIG. 7, the manufacturing process S80 can etch the target substrate by about 1 μm.

[0109] Furthermore, for example, the manufacturing process S80 can process the target substrate by an amount that exceeds the appropriate process amount to be performed on the target substrate, as estimated in the estimation process S70. That is, if the estimation process S70 estimates that the etching amount required for the target substrate is approximately 1 μm based on the estimation model shown in FIG. 8 or 9, the manufacturing process S80 can etch the target substrate by, for example, approximately 1.5 μm (estimated value + 0.5 μm), approximately 2.0 μm (estimated value + 1.0 μm), or approximately 2.5 μm (estimated value + 1.5 μm). In this way, killer defects in the target substrate can be more reliably removed.

[0110] Furthermore, for example, the manufacturing process S80 can process the target substrate by an amount less than the appropriate process amount to be performed on the target substrate, as estimated in the estimation process S70. That is, if the etching amount required for the target substrate is estimated to be approximately 1 μm in the estimation process S70 based on the estimation model shown in FIG. 8 or 9, the manufacturing process S80 can etch the target substrate by, for example, approximately 1.5 μm (estimated value -0.5 μm), approximately 2.0 μm (estimated value -1.0 μm), or approximately 2.5 μm (estimated value -1.5 μm). In this way, the process amount for the target substrate can be minimized, thereby improving yield.

[0111] <<Flatness of Semiconductor Substrate>> A preferred flatness of the surface 101 of a semiconductor substrate 100 used as a reference substrate in the above-described method for creating an estimation model or as a target substrate in the method for estimating a process quantity will be described with reference to Figures 11 and 12. Note that the arithmetic mean roughness (Ra) is defined here as a value measured by atomic force microscopy (AFM) over a range of approximately 10 µm in length at the center and / or periphery of the semiconductor substrate 100, but it can also be defined using other measurement regions or measurement methods.

[0112] Fig. 11 is a diagram showing the surface roughness Ra of a substrate and the scattered light intensity when S-polarized or P-polarized light is incident on the substrate. As can be seen from Fig. 11, there is a positive correlation between the surface roughness Ra of a substrate and the scattered light intensity when S-polarized or P-polarized light is incident on the substrate.

[0113] Fig. 12 is a diagram showing the relationship between the scattered light intensity measured using S-polarized light or P-polarized light and the roughness Ra of the surface 101, and corresponds to an enlarged view of a portion of Fig. 11. Note that (a) and (b) correspond to the scattered light intensity when S-polarized light is incident and when P-polarized light is incident, respectively.

[0114] As shown in Figure 12, in the range where the surface roughness Ra is greater than 0.8 nm, the scattered light intensity when S-polarized light is incident is relatively large, exceeding 4 mV, and the scattered light intensity when P-polarized light is incident is relatively large, exceeding 0.4 mV.

[0115] Considering that the range of scattered light intensity in the second data set shown in Figures 8 and 9 is less than 0.1 mV for S-polarized light incidence and less than 0.01 mV for P-polarized light incidence, and the positive correlation between Ra and scattered light intensity described above, it is considered preferable that the surface 101 of the semiconductor substrate 100 used as a reference substrate or target substrate has an Ra of at least 0.8 nm or less in order to suppress the effect of roughness on scattered light intensity.

[0116] Therefore, the surface 101 of the semiconductor substrate 100 preferably has an arithmetic mean roughness (Ra) of 0.8 nm or less, and more preferably, the surface 101 of the semiconductor substrate 100 has an arithmetic mean roughness (Ra) of 0.5 nm or less.

[0117] The configurations and functions shown in the above embodiments are merely examples and can be modified in various ways based on design requirements and the like.

[0118] REFERENCE SIGNS LIST 10 Light projection system 11 Laser output unit 12 Wave plate 20 Light receiving system 21 Objective lens 22 Imaging lens 23 Beam splitter 24 Light receiving sensor 25 Slit 100 Semiconductor substrate 101 Surface 102 Process-affected layer 103 Bulk layer L1 Laser light L2 Reflected light L3 Incident light L4 Scattered light N Normal θ Incident angle φ Measurement angle

Claims

1. A method for estimating an appropriate process amount to be performed on a semiconductor substrate, comprising: a measurement step of irradiating a laser light onto a target substrate and acquiring intensity measurement values ​​of light scattered on the target substrate; and an estimation step of estimating a process amount required for the reference index of the target substrate to satisfy a specified standard from the intensity measurement value of the scattered light on the target substrate obtained in the measurement step, using an estimation model created based on a data set correlating the intensity measurement value of the scattered light on the reference substrate measured by irradiating a laser light onto the reference substrate and the process amount required for the reference index of the reference substrate to satisfy the specified standard.

2. The estimation method according to claim 1, wherein the process variable includes a non-contact process variable.

3. The estimation method according to claim 2, wherein the process amount includes an etching amount.

4. The estimation method according to claim 1, wherein the reference indicator includes an indicator that is measurable in a layer formed by performing an irreversible process on the reference substrate or the target substrate.

5. The estimation method according to claim 1, wherein the reference indicator comprises an indicator measurable in a layer formed by performing an irreversible process on the reference substrate or the target substrate after a non-contact process.

6. The estimation method according to claim 1, wherein the reference indicators include indicators that can be measured in an epitaxially grown layer formed by epitaxially growing the reference substrate or the target substrate.

7. The estimation method according to claim 1, wherein the reference indicators include indicators that can be measured in an epitaxially grown layer of the reference substrate or the target substrate after etching.

8. The estimation method according to claim 1, wherein the reference indicator includes a killer defect density in an epitaxial growth layer formed by epitaxially growing the reference substrate or the target substrate.

9. The estimation method according to claim 1, wherein the reference indicator includes an IGSF (In-Grown Stacking Fault) density in an epitaxial growth layer formed by epitaxially growing the reference substrate or the target substrate.

10. The estimation method according to claim 1, wherein the reference indicator includes a BPD (Basal Plane Dislocation) density in an epitaxial growth layer formed by epitaxially growing the reference substrate or the target substrate.

11. The estimation method according to claim 1, wherein the reference indicator includes a BPD (Basal Plane Dislocation) propagation rate in an epitaxial growth layer formed by epitaxially growing the reference substrate or the target substrate.

12. The estimation method according to claim 1, further comprising a planarization step of planarizing the surface of the target substrate prior to the measuring step.

13. The estimation method according to claim 1, wherein the target substrate has an arithmetic mean roughness Ra of less than 0.8 nm on its surface.

14. The estimation method of claim 1, wherein the laser light is polarized.

15. The estimation method according to claim 14, wherein the laser light is P-polarized or S-polarized.

16. A method for manufacturing a semiconductor substrate, comprising the step of processing the target substrate based on the appropriate process amount to be performed on the target substrate estimated by the estimation method according to any one of claims 1 to 15.

17. A method for creating an estimation model, comprising: a first measurement step of irradiating a laser beam onto each of a plurality of reference substrates and acquiring intensity measurement values ​​of scattered light scattered in each of the reference substrates; a processing step of performing processing on each of the reference substrates including processing different process amounts; a second measurement step of measuring a reference index of each of the reference substrates; an analysis step of creating a first data set correlating the reference index of each of the reference substrates with the process amount performed on each of the reference substrates, and estimating, based on the first data set, a process amount necessary for the reference index of each of the reference substrates to satisfy a specified standard; and an estimation model creation step of creating a second data set correlating the process amount necessary for the reference index of each of the reference substrates to satisfy a specified standard, estimated in the analysis step, with the intensity measurement values ​​of the scattered light measured in the first measurement step, and creating, based on the second data set, an estimation model that estimates an appropriate process amount to be performed on the target substrate from the intensity measurement values ​​of the scattered light at the target substrate.

18. An estimation method according to any one of claims 1 to 15, wherein the estimation model is an estimation model created using the estimation model creation method according to claim 17.

Images