Methods for determining the carbon concentration of a silicon sample and methods for producing a silicon single-crystal ingot

Abstract

Method for determining the carbon concentration of a silicon sample, which includes: Introducing hydrogen atoms into a silicon sample to be determined; Subjecting the silicon sample to be determined, into which hydrogen atoms have been introduced, to an evaluation by an evaluation method of evaluating a capture level in a silicon band gap without electron beam irradiation; and Determining the carbon concentration of the silicon sample to be determined based on the evaluation result of at least one capture level selected from the group consisting of Ec - 0.10 eV, Ec - 0.13 eV and Ec - 0.15 eV under evaluation results obtained by the evaluation, where the specific carbon concentration is less than 1.0E+16 atoms / cm³ 3 .

Description

Reference to related applications

[0001] This application claims priority over Japanese patent application No. 2016-078579, published as JP 2017 - 191 800 A, filed on April 11, 2016, to which explicit reference is hereby made. Technical field

[0002] The present invention relates to a method for determining the carbon concentration of a silicon sample, a method for producing a silicon single-crystal ingot, a silicon single-crystal ingot and a silicon wafer. State of the art

[0003] Reducing impurity contamination, which degrades the properties of devices, is always necessary for silicon wafers used as semiconductor substrates. In recent years, carbon has gained attention as an impurity in silicon wafers, and the reduction of carbon contamination in silicon wafers has been investigated. To reduce carbon contamination, it is desirable to determine the carbon concentration of a silicon sample and, based on the results, to adjust the manufacturing conditions for a silicon single-crystal ingot, from which a silicon wafer is cut, in such a way as to deplete the carbon introduced during the manufacturing process.

[0004] Traditionally, a method using FT-IR (Fourier-transform infrared spectroscopy) has been widely used to determine the carbon concentration of a silicon sample. Furthermore, methods using SIMS (secondary ion mass spectrometry), photoluminescence, or cathodoluminescence have also been proposed (see, for example, unexamined Japanese patent application, publication JP 2013 - 152 977 A; unexamined Japanese patent application, publication JP 2015 - 101 529 A; and unexamined Japanese patent application, publication JP 2015 - 222 801 A).

[0005] JP 2015-156 420 A addresses the problem of providing a method for evaluating the carbon concentration in a silicon single crystal with high accuracy and simplicity. As a solution to this problem, the document describes a method comprising: a first step in which ions other than carbon and oxygen are implanted into a silicon substrate obtained from a silicon single crystal; a second step in which the concentration of carbon-related defects generated by the first step is measured; and a third step in which the carbon concentration in the silicon single crystal is evaluated from the concentration of carbon-related defects measured by the second step. Summary of the invention

[0006] The lower detection limit of FT-IR is typically on the order of 10 15atoms / cm². Although the method using FT-IR (FT-IR method) is an effective method in cases where the carbon concentration of a silicon sample is comparatively high, the sensitivity for an accurate determination of the carbon concentration of a silicon sample with a low carbon impurity level, in particular a carbon concentration of less than 1E+16 atoms / cm², is 3 , i.e. on the order of 10 15 atoms / cm² 3 or less, is insufficient. Furthermore, even determination using the FT-IR method is difficult if the carbon concentration is equal to or less than 1E+15 atoms / cm³. 3. In recent years, however, silicon wafers with reduced carbon concentrations and low carbon contamination have been required, and therefore it is desirable to quantitatively determine a trace amount of carbon in a silicon sample with a sensitivity that exceeds that of the FT-IR method.

[0007] In contrast, SIMS enables analysis with higher sensitivity than FT-IR. Therefore, the SIMS method allows for the determination of carbon concentrations lower than those of FT-IR for the quantitative determination of trace amounts of carbon, and desirablely, carbon in a silicon sample can be quantitatively determined with a sensitivity that is at least equal to or exceeds that of the SIMS method.

[0008] On the other hand, methods using photoluminescence or cathodoluminescence (luminescence methods), as described in the above publications, allow analysis with a sensitivity even higher than that of the SIMS method. However, the luminescence method described in the above publication requires electron beam irradiation to determine the carbon concentration. This is because the carbon concentration can be determined by measuring the concentration of Ci-Cs generated by activating a substitutional carbon (Cs) to an interstitial carbon (Ci) through electron beam irradiation.However, if oxygen is present in the silicon sample, some of the generated interstitial carbon (Ci) pairs with the interstitial oxygen (Oi) (Ci-Oi), and therefore the concentration of the ultimately generated Ci-C depends on the oxygen concentration. Consequently, the carbon concentration to be determined quantitatively is influenced by the oxygen concentration of the silicon sample.

[0009] Furthermore, electron beam irradiation also presents problems such as the long time required, the need for a large-scale facility, increased costs, the need for heat treatment to form a protective oxide film, and the requirement for a recovery process in addition to electron beam irradiation. The increased number of procedures also makes it easy to encounter interferences, among other issues. Therefore, it is desirable to quantitatively determine a trace amount of carbon in a silicon sample without electron beam irradiation.

[0010] One aspect of the present invention provides a novel means for the quantitative determination of carbon in a silicon sample without electron beam irradiation treatment with a sensitivity equal to or exceeding that of the SIMS method.

[0011] In reproducing studies to solve the above problem, the inventors first discovered that the carbon-related level density in a band gap of silicon, activated by introducing hydrogen atoms into the silicon sample, correlates with the carbon concentration of the silicon sample. Furthermore, as a result of further detailed studies, it was found for the first time that a trace amount of carbon can be quantitatively determined with a sensitivity equal to or exceeding that of the SIMS method, and without electron beam irradiation, according to the following method for determining the carbon concentration of a silicon sample according to an aspect of the present invention.

[0012] One aspect of the present invention was completed based on the above findings.

[0013] One aspect of the present invention relates to a method for determining the carbon concentration of a silicon sample (hereinafter simply described as the "determination method"), wherein the method comprises: Introducing hydrogen atoms into a silicon sample to be determined; Subjecting the silicon sample to be determined, into which hydrogen atoms have been introduced, to an evaluation by an evaluation method of evaluating a capture level in a silicon band gap without electron beam irradiation; and

[0014] Determining the carbon concentration of the silicon sample to be determined based on the evaluation result of at least one capture level selected from the group consisting of Ec-0.10 eV, Ec-0.13 eV and Ec-0.15 eV under evaluation results obtained by the evaluation, where the specific carbon concentration is less than 1.0E+16 atoms / cm³ 3 .

[0015] "E+" represents an exponent, as is widely known. For example, "1.0E+16" means, as is widely known, "1.0 x 10 16 “The same applies to the other notations using E+.”

[0016] According to one embodiment, the determined carbon concentration is equal to or less than 1.0E+15 atoms / cm³. 3 .

[0017] In one embodiment, the oxygen concentration of the silicon sample to be determined by the FT-IR method is equal to or greater than 1.0E+17 atoms / cm³. 3 The oxygen concentrations described below are values ​​determined by the FT-IR method, unless otherwise stated.

[0018] In one embodiment, the carbon concentration of the silicon sample to be determined is determined based on an evaluation result at Ec-0.15 eV.

[0019] In one embodiment, hydrogen atoms are introduced into the silicon sample to be determined by immersing the silicon sample in a solution.

[0020] In one embodiment, the solution is hydrofluoric acid.

[0021] In one embodiment, the carbon concentration of the silicon sample to be determined is determined by using a calibration curve based on the above evaluation result.

[0022] In one embodiment, the determination method comprises: Introducing hydrogen atoms into a plurality of silicon samples to generate a calibration curve in which carbon concentrations, measured by an evaluation method other than the evaluation method, are known,

[0023] Subjecting the majority of silicon samples to generate a calibration curve, into which hydrogen atoms have been introduced, to evaluation by the same evaluation procedure as for the silicon sample to be determined, and generating the calibration curve by using the evaluation result at the same capture level as the capture level used to determine the carbon concentration of the silicon sample to be determined, as well as the known carbon concentrations described above.

[0024] In one embodiment, the evaluation method is a DLTS (Deep-Level Transient Spectroscopy) method.

[0025] In one embodiment, the width Wa of a depletion layer formed on the silicon sample for generating the calibration curve during the evaluation of the silicon sample to be determined by the DLTS method, and the width Wb of a depletion layer formed on the silicon sample for generating the calibration curve during the evaluation of the silicon sample to be determined by the DLTS method satisfy the following equation: |Wa−Wb|≤2.0 μm

[0026] Another aspect of the present invention relates to a method for producing a silicon single-crystal ingot, which comprises the following: Growing a silicon single crystal using the Czochralski method;

[0027] Determining the carbon concentration of a sample cut from the silicon single-crystal ingot using the above determination method;

[0028] Determining the manufacturing conditions for a silicon single-crystal ingot based on the determined carbon concentration of the silicon sample; as well as

[0029] Growth of the silicon single-crystal ingot by the Czochralski process under the specified manufacturing conditions.

[0030] The term "production conditions" in the present invention and description includes the pulling device used, the purity of the raw material polysilicon, the growth conditions (pulling rate, gas flow rate, and the like), and the like. Furthermore, cases of modification of the pulling device include a case in which the pulling device itself is the same, but the configuration of an element in the device is changed, a case in which the installation position of an element is changed, and the like. In one embodiment, the silicon single-crystal ingot grown from the tip under the specified production condition has a carbon concentration equal to or less than 1.0 × 10⁻¹⁵ atoms / cm³. 3 , determined by the above determination procedure.

[0031] In one embodiment, the silicon sample cut from the tip has an oxygen concentration equal to or greater than 1.0E+17 atoms / cm³. 3 , determined by the FT-IR method.

[0032] In one embodiment, the silicon single-crystal ingot grown under the specified manufacturing conditions has a carbon concentration of equal to or less than 1.0E+15 atoms / cm³. 3 on, determined by the above determination method of the carbon sample cut from this silicon single crystal ingot, over the entire area from the tip to the bottom.

[0033] In one embodiment, the silicon single crystal grown under the specified manufacturing conditions has an oxygen concentration of equal to or greater than 1.0E+17 atoms / cm³. 3on, determined by FT-IR, for the silicon sample cut out from this silicon single crystal ingot over the entire area from the tip to the bottom.

[0034] According to one aspect of the present invention, a trace amount of carbon contained in the silicon sample can be quantitatively determined with a sensitivity equal to or greater than that of the SIMS method, without electron beam irradiation. Since carbon can be quantitatively determined without electron beam irradiation, the trace amount of carbon in the silicon sample can also be quantitatively determined independently of the oxygen concentration. Brief description of the drawings Fig. Figure 1 is an explanatory drawing showing the setup of a silicon single crystal pulling device used in the examples. Fig. 2 explains an example of DLTS spectra (spectrum before fitting and three spectra after fitting), obtained according to the examples. Fig. Figure 3 shows a calibration curve generated on the basis of an evaluation result (capture level density, determined from the DLTS signal intensity) at Ec-0.10 eV according to Example 1. Fig. Figure 4 shows a calibration curve generated on the basis of an evaluation result (capture level density, determined from the DLTS signal intensity) at Ec-0.13 eV according to Example 1. Fig. Figure 5 shows a calibration curve generated on the basis of an evaluation result (capture level density, determined from the DLTS signal intensity) at Ec-0.15 eV according to Example 1. Fig. Figure 6 shows a calibration curve generated on the basis of the DLTS signal intensity, determined during the setting of a blocking voltage according to Example 2. Fig. Figure 7 shows a calibration curve generated on the basis of the DLTS signal intensity determined during the changing of a blocking voltage to set the widths of the depletion layers to the same level, according to Example 2. Description of embodiments [Method for determining the carbon concentration of a silicon sample]

[0035] The method for determining the carbon concentration of a silicon sample according to one aspect of the present invention comprises the following: introducing hydrogen atoms into a silicon sample to be determined; subjecting the silicon sample to be determined, into which hydrogen atoms have been introduced, to an evaluation by an evaluation method of evaluating a capture level in a silicon band gap without electron beam irradiation; and determining the carbon concentration of the silicon sample to be determined based on the evaluation result of at least one capture level selected from the group consisting of Ec-0.10 eV, Ec-0.13 eV and Ec-0.15 eV, with evaluation results obtained by the evaluation, wherein the determined carbon concentration is less than 1.0 E+16 atoms / cm³. 3 .

[0036] The above measurement procedure is described in further detail below. <Zu bestimmende Siliciumprobe>

[0037] The silicon sample to be determined in the above method is, for example, a silicon sample cut from a silicon single-crystal ingot. For instance, a portion of the sample, cut into a wafer shape, can be further excised from the silicon single-crystal ingot for determination. Furthermore, the sample to be determined can be a silicon sample cut from various types of silicon wafers (for example, a polished wafer or an epitaxial wafer) used as semiconductor substrates. Additionally, the silicon wafer can be one that has undergone various processing treatments (for example, polishing, etching, washing, and the like) commonly performed on silicon wafers. The silicon sample can be either n-type or p-type silicon.

[0038] The oxygen concentration of the silicon sample to be determined can, for example, be equal to or higher than 1.0E+17 atoms / cm³. 3 be (for example, 1.0E+17 to 27.5E+17 atoms / cm³) 3 The oxygen concentration is a value determined using the FT-IR method.

[0039] For example, a sample derived from a silicon single crystal grown using the Czochralski method (CZ method) typically contains oxygen. On the other hand, as described above, in the luminescence method, which requires electron beam irradiation, the carbon concentration to be quantified depends on the oxygen concentration. Therefore, the higher the oxygen concentration in the silicon sample, the lower the accuracy of the carbon concentration determination tends to be.

[0040] In contrast, the electron beam irradiation treatment is not performed in the above measurement method during the evaluation. Therefore, the carbon concentration can be determined independently of the oxygen concentration. Accordingly, the carbon concentration of a silicon sample with a comparatively high oxygen concentration can also be determined with high accuracy within the above range using the measurement method for a silicon sample with a high oxygen concentration. <Einführen von Wasserstoffatomen in Siliciumproben>

[0041] As described above, the inventors have for the first time found a correlation between the carbon-related level density in a band gap of silicon, activated by introducing hydrogen atoms into a silicon sample, and the carbon concentration of the silicon sample. The introduction of hydrogen atoms can be carried out by a dry treatment (dry type) or a wet treatment (wet type, namely the use of a solution). For example, the introduction of hydrogen atoms by dry treatment can be carried out by ion implantation, hydrogen plasma, or the like. The introduction of hydrogen atoms according to the present invention and description comprises an embodiment in which hydrogen atoms are introduced in an ionic or plasma state.

[0042] The introduction of hydrogen atoms by wet treatment can be carried out by contacting a silicon sample with a solution (for example, by immersion). The solution used can be either an acidic or a basic solution, as long as it contains hydrogen atoms in an ionized state (ion), a salt state, or the like. For example, the acidic solution could include hydrofluoric acid, a mixture of hydrofluoric acid and nitric acid (nitric acid), a mixture of sulfuric acid and hydrogen peroxide, a mixture of hydrochloric acid and hydrogen peroxide, and the like.

[0043] Furthermore, the base solution can comprise a sodium hydroxide solution, a potassium hydroxide solution, a mixture of ammonia water and hydrogen peroxide, and the like. The above-mentioned various solutions are preferably water-based (water-containing solutions), more preferably aqueous solutions. The acid concentration of the acid solution and the base concentration of the base solution are not particularly limited. For example, the introduction of hydrogen atoms with hydrofluoric acid can be carried out by immersing the silicon sample to be determined in 1 to 25 wt% hydrofluoric acid for 1 to 10 minutes. After immersion, the sample to be determined can be subjected to post-treatments, such as washing with water and drying, as required. <Evaluierung der Siliciumprobe, in welche Wasserstoffatome eingeführt worden sind>

[0044] The carbon-related level in the band gap of silicon activated by the introduction of hydrogen atoms into a silicon sample correlates with the carbon concentration in the silicon. This finding was first discovered by the inventors as a result of intensive investigations. Therefore, the carbon-related density in the band gap of the silicon sample into which hydrogen atoms have been introduced can be determined without electron beam irradiation. The expression "without electron beam irradiation" in the present invention and description means that no treatment, in particular irradiation of the silicon sample with an electron beam, is carried out, provided that electron beam irradiation, which inevitably occurs under sunlight, lighting, or the like, is acceptable.

[0045] In the above determination method, a capture level of Ec-0.10 eV, Ec-0.13 eV, or Ec-0.15 eV is used as the carbon-related level. One or more of these capture levels (carbon-related levels) are activated by the introduction of hydrogen atoms, and therefore the carbon concentration can be determined based on the capture level densities without electron beam irradiation. The determination of the capture level density can be performed by various evaluation methods capable of evaluating the capture level in the silicon band gap. Such evaluation methods may include a DLTS method, a half-life method, an ICTS (isothermal capacitance transient spectroscopy) method, a low-temperature photoluminescence (PL) method, a cathodoluminescence (CL) method, and the like.Conventional carbon concentration determinations using the PL and CL methods required electron beam irradiation. In contrast, the determination method described above activates one or more capture levels by introducing hydrogen atoms, thus enabling carbon concentration measurement based on capture level densities without electron beam irradiation. Existing techniques can be applied to these determination methods through various evaluation procedures without any limitations.

[0046] For example, with regard to the DLTS method, the lower limit of the detection of the capture density according to the DLTS method is generally about 10 -4 - up to 10 -5 - times the concentration of the carrier, and therefore a quantitative determination of carbon is also possible at a concentration on the order of 1013 Atoms / cm 3or lower. As described above, the DLTS method is a preferred evaluation method from the perspective of enabling the quantitative determination of carbon with higher sensitivity. When using the DLTS method as an evaluation method, a DLTS spectrum obtained as the sum of the corresponding peaks obtained by the DLTS method is subjected to fitting using a known method, whereby a DLTS spectrum with a capture level at Ec-0.10 eV, Ec-0.13 eV, or Ec-0.15 eV can be separated. For example, in DLTS measurements at a frequency of 250 Hz, the capture level density at Ec-0.15 eV, the capture level density at Ec-0.13 eV and the capture level density at Ec-0.15 eV can be determined based on the peak intensities (DLTS signal intensity) at peaks in the vicinity of 76 K, 87 K and 101 K, respectively.The peak used to determine the carbon concentration is at least one of the three peaks mentioned above, or two or three peaks may be used. Generally, the higher the peak intensity, the higher the carbon concentration that can be determined. For higher accuracy in determining the carbon concentration, it is preferred to determine the carbon concentration of the silicon sample based on the evaluation results at Ec-0.13 eV and / or Ec-0.15 eV. For example, the higher the peak intensity of the DLTS spectrum (DLTS signal intensity) with a capture level at Ec-0.15 eV, separated by fitting, the higher the carbon concentration that can be determined.

[0047] Similarly, when using any method as an evaluation procedure, it is preferred to determine the carbon concentration of the silicon sample to be analyzed using a calibration curve. The calibration curve shows the correlation between the evaluation result obtained by the evaluation procedure (for example, the capture level density determined from the peak intensity (DLTS signal intensity) determined by the DLTS method) and the carbon concentration. The expression for the relationship to determine the capture level density from the DLTS signal intensity is known.Preferably, hydrogen atoms are introduced into a plurality of silicon samples to generate a calibration curve with known carbon concentrations. A plurality of silicon samples, into which hydrogen atoms have been introduced, are subjected to evaluation using the same evaluation procedure as the silicon sample to be determined. A calibration curve can then be generated using an evaluation result at the same capture level as that used to determine the carbon concentration of the silicon sample to be determined and the known carbon concentration of the silicon sample for generating the calibration curve. The known carbon concentration of the silicon sample for generating the calibration curve is determined using a different evaluation procedure than the one used for evaluating the silicon sample to be determined.The evaluation procedure is exemplified by well-known evaluation methods such as SIMS method, FT-IR method, luminescence method, and the SIMS method is preferred.

[0048] Various silicon samples, such as the example above for the silicon sample to be determined, can be used as the silicon sample for generating the calibration curve. To further improve the accuracy of the carbon concentration determination, the silicon sample for generating the calibration curve is preferably either a silicon sample cut from the same silicon sample as the silicon sample to be determined, or a silicon sample obtained by the same manufacturing process as that used for the silicon sample to be determined.

[0049] In addition, the DLTS method involves alternately and periodically applying a blocking voltage to form a depletion layer and a low voltage close to 0 V to trap the carriers in the depletion layer to a semiconductor junction (Schottky junction or pn junction) formed on a silicon sample to be evaluated. The width (determination width) of the formed depletion layer changes depending on the magnitude of the applied blocking voltage. If the width of the depletion layer is represented as W, then W can be calculated according to the following equation. In the equation, N represents DThe dopant concentration is represented, and therefore W is inversely proportional to the dopant concentration. Therefore, in a case where there is a difference in the dopant concentration (i.e., the resistance value) between the silicon sample used to generate the calibration curve and the sample to be determined, when the blocking voltage V is applied with the same value as that used in the measurement by the DLTS method, there is a change in the width W of the depletion layer formed between the silicon sample used to generate the calibration curve and the sample to be determined. From the point of view of determining the carbon concentration with higher accuracy, the width Wa of the depletion layer formed in the silicon sample used to generate the calibration curve is preferably substantially equal to the width Wb of the depletion layer formed in the sample to be determined.For example, the absolute value of the difference between Wa and Wb is preferably equal to or less than 2.0 µm. This means that Wa and Wb preferably satisfy equation 2 below. |Wa−Wb|≤2.0 μm

[0050] To determine the widths of the depletion layers formed at the same level in both samples, a blocking voltage can be applied based on the dopant concentration of each sample and W, calculated from the following equation. W={2Ksε0qND(Vbi−V)}12 K s : relative permittivity of silicon q: Elementary charge V bi Built-in potential ε0: Permitivity of the vacuum

[0051] By using the calibration curve generated in this way, the carbon concentration of the silicon sample to be determined can be calculated from the evaluation result obtained using the evaluation method described above. The carbon concentration of the silicon sample to be determined, as measured by the above method, is less than 1.0 × 10⁻¹⁶ atoms / cm³. 3 According to the above determination method, the carbon concentration of a carbon-containing silicon sample can be measured using such an FT-IR method in a concentration range where high-precision determination is difficult. Furthermore, according to the above measurement method, the carbon concentration within a range where measurement is difficult can be determined using the FT-IR method. From this point, the carbon concentration of the sample to be determined is preferably equal to or less than 1.0 × 10⁻¹⁵ atoms / cm³. 3, as determined according to the above determination method. Furthermore, according to the above determination method, the carbon concentration can be determined to an order of 10 14 atoms / cm² 3 can also be determined, and furthermore, carbon can also be detected at low concentrations, for example on the order of 10 13 atoms / cm² 3 or lower, can be determined quantitatively. Therefore, the carbon concentration of the silicon sample to be determined can be, for example, 1.0 E + 14 atoms / cm³. 3 up to 1.0E+15 atoms / cm³ 3 or 1.0E+13 atoms / cm² 3 up to 1.0E+15 atoms / cm³ 3 as a value determined by the determination procedure described above. [Method for the production of a silicon single-crystal ingot, silicon single-crystal ingot and silicon wafer]

[0052] One aspect of the present invention relates to a method for producing a silicon single-crystal ingot, wherein the method comprises: Growth of a silicon single-crystal ingot by the Czochralski method;

[0053] Determining the carbon concentration of a silicon sample cut from the silicon single crystal ingot by the above measurement method according to one aspect of the present invention;

[0054] Establishing manufacturing conditions for the silicon single-crystal ingot based on the determined carbon concentration of the silicon sample; as well as

[0055] Growth of the silicon single-crystal ingot by the Czochralski process under the specified manufacturing conditions.

[0056] In the process of producing a silicon single-crystal ingot according to one aspect of the present invention, a known process relating to the Czochralski process (CZ process) can be used to grow the silicon single-crystal ingot by the CZ process. Carbon may be present in the silicon single-crystal ingot grown by the CZ process due to carbon mixed into the raw polysilicon, CO gas generated during growth, and the like. To produce the silicon single-crystal ingot in which the mixing of carbon is suppressed, it is preferred to determine a concentration of such added carbon with high accuracy in order to establish production conditions based on the determination results.For this reason, the determination method according to one aspect of the present invention is suitable as a method for determining the concentration of the added carbon.

[0057] Regarding the silicon single-crystal ingot, if the upper end in the direction of growth is referred to as the tip and the other end as the base, the carbon concentration typically tends to increase towards the base (segregation property). Consequently, even if the silicon sample cut from the base contains carbon at a concentration that allows for highly accurate measurement by FT-IR, the carbon concentration of the silicon sample cut from the tip will be lower than that of the base, and therefore highly accurate determination may be difficult, or determination may be difficult using the FT-IR method.However, to produce a silicon single-crystal ingot in which the carbon concentration is reduced across the entire region from tip to base, the carbon concentration at the tip is preferably determined with high accuracy, and the conditions for producing the silicon single-crystal ingot are then defined based on these determined carbon concentrations to further reduce the carbon concentration. The term "tip" refers to the region from the seeding area of ​​the single crystal to the straight body portion, and the term "base" refers to the region from the straight body portion of the silicon single-crystal ingot to a part where the crystal diameter narrows to a conical shape.With regard to the above points, the determination method according to one aspect of the present invention is suitable as a method for the quantitative determination of a trace amount of carbon in the silicon sample cut from the tip, since the method allows a highly accurate determination of the carbon concentration within a concentration range in which highly accurate measurement using the FT-IR method is difficult. The carbon concentration of the silicon sample cut from the tip can be less than 1.0 × 10¹⁶ atoms / cm³. 3 its density is preferably equal to or less than 1.0E+15 atoms / cm³ 3 , than the carbon concentration determined by the above method. Furthermore, the carbon concentration of the silicon sample cut from the tip can, for example, be within a range of 1.0 × 10¹⁴ atoms / cm³. 3 up to 1.0E+15 atoms / cm³ 3 It can amount to 1.0E+13 atoms / cm². 3up to 1.0E+15 atoms / cm³ 3 The amount is greater than the carbon concentration determined by the above determination method.

[0058] Furthermore, as described above, the silicon single-crystal ingot grown by the CZ process normally contains oxygen. As described above, the luminescence method requires electron beam irradiation, and therefore the determined carbon concentration is influenced by the oxygen concentration of the silicon sample. In contrast, in the determination method according to one aspect of the present invention as described above, the carbon concentration can be determined without electron beam irradiation, and therefore the carbon concentration can be determined independently of the oxygen concentration. For this reason, the carbon concentration of a silicon sample containing oxygen at a relatively high concentration, for example, a silicon sample containing oxygen at a concentration of 1.0 × 10¹⁷ atoms / cm³ or greater, can be determined without dependence on the oxygen concentration. 3 (for example, 1.0E+17 to 27.5E+17 atoms / cm²) 3) contains, can also be determined with high accuracy. Therefore, even if the oxygen concentration of a silicon sample cut from the silicon single-crystal ingot grown by the CZ method, for example, a silicon sample cut from the bottom, is within the above range, the carbon concentration can be determined with high accuracy according to the above determination method.

[0059] The silicon sample cut from the silicon single-crystal ingot grown by the CZ process can be a sample from any part of the silicon single-crystal ingot (bottom, tip, or transition region between them). Preferably, the sample is a silicon sample cut from the tip, which tends to contain carbon in a lower concentration. By growing the silicon single-crystal ingot as required under production conditions determined by applying a carbon-reducing agent based on the carbon concentration of the silicon sample cut from the tip, it is possible to produce a silicon single-crystal ingot in which the carbon contamination is reduced across the entire region from the tip to the bottom. For example, one or more of the following agents can be used as carbon-reducing agents: (1) Use of a high-purity product in which a smaller amount of carbon is added to the raw material polysilicon. (2) Appropriate adjustment of the pulling rate and / or the argon (Ar) gas flow rate at the time of crystal pulling to suppress the dissolution of CO in a polysilicon melt. (3) Making changes to the structure and mounting position of a carbon element contained in the drawing device, and the like.

[0060] Silicon single-crystal ingot produced under such specified manufacturing conditions can have a carbon concentration equal to or less than 1.0E+15 atoms / cm³. 3 , can also exhibit a carbon concentration of 1.0E+14 atoms / cm³ 3 up to 1.0E+15 atoms / cm³ 3 sufficient, exhibit, or can also have a carbon concentration of 1.0E+13 atoms / cm³ 3 up to 1.0E+15 atoms / cm³ 3The silicon single-crystal ingot produced in this manner can exhibit an oxygen concentration equal to or greater than 1.0 × 10¹⁷ atoms / cm³, as determined by the method of determination according to one aspect of the present invention. 3 (for example, 1.0E+17 to 27.5E+17 atoms / cm²) 3 ) across the entire area from the top to the bottom.

[0061] The silicon single-crystal ingot obtained by the above manufacturing process can have a carbon concentration equal to or less than 1.0E+15 atoms / cm³. 3 , can also have a carbon concentration of 1.0E+14 atoms / cm³ 3 up to 1.0E+15 atoms / cm³ 3 sufficient, or can also have a carbon concentration of 1.0E+13 atoms / cm³ 3 up to 1.0E+15 atoms / cm³ 3The amount of oxygen in the silicon sample cut from this silicon single-crystal ingot is determined according to the determination methods according to one aspect of the present invention, over the entire area from the tip to the bottom. Furthermore, the oxygen concentration of the same can, for example, be equal to or greater than 1.0 × 10¹⁷ atoms / cm². 3 (for example, 1.0E+17 to 27.5E+17 atoms / cm²) 3 ) across the entire area from the top to the bottom.

[0062] A silicon wafer cut from the silicon single-crystal ore can have a carbon concentration equal to or less than 1.0E+15 atoms / cm³. 3 , can also have a carbon concentration of 1.0E+14 atoms / cm³ 3 up to 1.0E+15 atoms / cm³ 3 sufficient, or can also have a carbon concentration of 1.0E+13 atoms / cm³ 3 up to 1.0E+15 atoms / cm³ 3sufficient, as determined by the determination method according to one aspect of the present invention. Furthermore, the oxygen concentration of the same can, for example, be equal to or greater than 1.0 E + 17 atoms / cm³. 3 (for example, 1.0E+17 to 27.5E+17 atoms / cm²) 3 ).

[0063] As a result, according to one aspect of the present invention, it will be possible to provide a silicon single-crystal ingot and a carbon in a silicon wafer containing a concentration that is difficult to determine by the FT-IR method. Examples

[0064] The present invention will be further explained below with reference to the examples. However, the present invention is not limited to the embodiments shown in the examples. [Example 1] 1. Growing silicon single-crystal ingot by the CZ process

[0065] A plurality of silicon single-crystal ingots with different carbon concentrations were grown by changing one or more of the following production conditions, selected from the group consisting of the purity of the raw material polysilicon, the growing device, and the growth conditions, using the silicon single-crystal growing device with the in Fig. setup shown.

[0066] The following details of the in Fig. 1 silicon single crystal pulling device shown is described.

[0067] One in Fig. 1 The silicon single-crystal drawing apparatus 10 shown comprises a chamber 11, a bearing rotation shaft 12 arranged vertically and penetrating the center of the bottom of the chamber 11, a graphite susceptor 13 attached to the upper end section of the bearing rotation shaft 12, a quartz crucible 14 housed in the graphite susceptor 13, a heating device 15 arranged around the graphite susceptor 13, a bearing shaft drive mechanism 16 for raising / lowering and rotating the bearing rotation shaft 12, a seed crystal clamping device 17 for securing the seed crystals, a drawing wire 18 for suspending the seed crystal clamping device, a wire winding mechanism 19 for winding the drawing wire 18, and a heat shield 22 to prevent the silicon single-crystal ingot 20 from being heated by radiant heat from the heating device 15 and the quartz crucible 14. as well as to suppress the temperature fluctuation of the silicon melt 21,as well as a control unit 23 for controlling each unit.

[0068] A gas supply 24 for introducing argon gas into chamber 11 is arranged above chamber 11. Argon gas is introduced from the gas inlet 24 into chamber 11 through a gas line 25, and the amount of gas introduced is controlled by a conductance valve 26.

[0069] A gas outlet 27 for venting argon gas in chamber 11 is located at the bottom of chamber 11. Argon gas in the sealed chamber 11 is discharged from the gas outlet 27 via a gas outlet line 28. A conductance valve 29 and a vacuum pump 30 are installed in the middle of the gas outlet line 28. By controlling the flow rate of the argon gas through the conductance valve 29 during the intake of argon gas into chamber 11 by means of the vacuum pump 30, the pressureless condition in chamber 11 is maintained.

[0070] Furthermore, a magnetic field generator 31 is provided to apply a magnetic field to the silicon melt 21 outside the chamber 11. The magnetic field provided by the magnetic field generator 31 can be a horizontal magnetic field or a reverse magnetic field. 2. Cutting out from the silicon sample

[0071] Each silicon single-crystal ingot pulled according to point 1 above was sectioned, and a wafer-shaped sample was cut from the tip of the ingot. A silicon sample for DLTS determination and a silicon sample for SIMS determination were obtained from the same sample. The oxygen concentration of each sample determined by the FT-IR method was 2.0 × 10⁻¹⁷ to 12.0 × 10⁻¹⁷ atoms / cm³. 3 The silicon single-crystal ingot was of n-type silicon (value of resistivity: 10 to 100 Ω·cm). 3. Determination by DLTS procedure

[0072] The silicon sample cut from each silicon single-crystal ingot for DLTS determination was successively subjected to the following procedures (A), (B), and (C), forming a Schottky barrier layer on one side of each silicon sample and an ohmic layer (Ga layer) on the other. Hydrogen atoms were introduced into the silicon sample for DLTS determination by the following procedure (A) (wet treatment). (A) Immersion in 5 wt% hydrofluoric acid for 5 minutes and then washing with water for 10 minutes. (B) Formation of a Schottky electrode (Au electrode) by vacuum deposition. (C) Formation of a back-side ohmic layer by rubbing with gallium.

[0073] A reverse voltage to form a depletion layer and a pulsed voltage to trap the carriers on the depletion layer were applied alternately and periodically to the Schottky junction of the silicon sample, which had been subjected to the above procedures (A) to (C). The transient response of the electrical capacitance of a generated diode corresponding to the above voltage was measured.

[0074] The voltage was applied and the capacitance measured while the sample temperature was monitored within a predetermined temperature range. The DLTS signal intensity ΔC was plotted against the temperature, yielding the DLTS spectra. The measurement frequency was 250 Hz. The silicon sample used for DLTS determination was not subjected to electron beam irradiation.

[0075] The obtained DLTS spectra were subjected to true-shape fitting processing using a program manufactured by Semilab Inc., and were separated into DLTS spectra with a capture level at Ec-0.10 eV (peak position: temperature 76 K), a capture level at Ec-0.13 eV (peak position: temperature 87 K), and a capture level at Ec-0.15 eV (peak position: temperature 101 K). Hereinafter, the DLTS spectrum with a capture level at Ec-0.10 eV is referred to as "E1 Fit.", the DLTS spectrum with a capture level at Ec-0.13 eV as "E2 Fit.", and the DLTS spectrum with a capture level at Ec-0.15 eV as "E3 Fit."

[0076] As an example, one of the obtained DLTS spectra is in Fig. 2 shown. Fig. Figure 2 shows a DLTS spectrum before fitting processing, as well as each DLTS spectrum of the E1 Fit., E2 Fit., and E3 Fit. obtained by fitting. In the figure, the unit of the vertical axis is an arbitrary unit (AU). Determination using SIMS

[0077] The carbon concentration of the silicon sample cut out from each silicon single crystal ingot for SIMS determination was determined by performing SIMS measurements (determination of carbon concentration by a scanning change method). 4. Generation of the calibration curve

[0078] A calibration curve was determined using the carbon concentration determined by SIMS according to point 3 above and the capture level density, determined from the DLTS signal intensity at a peak position of each DLTS spectrum (according to Fitting), determined for the silicon sample for DLTS determination, obtained from the same sample as the silicon sample for SIMS determination. In particular, a Fig. The calibration curve shown in Figure 3 was generated, in which the capture level density, determined from a DLTS signal intensity at a peak position (temperature: 76 K) in the DLTS spectrum of the E1 Fit., was plotted on the vertical axis, and the carbon concentration determined by SIMS was plotted on the horizontal axis. Fig. The calibration curve shown in Figure 4 was generated, in which the capture level density, determined from a DLTS signal intensity at a peak position (temperature: 87 K) in the DLTS spectrum of the E2 Fit., was plotted on the vertical axis, and a carbon concentration determined by SIMS was plotted on the horizontal axis. Fig. The calibration curve shown in Figure 5 was generated in which the capture level density, determined from a DLTS signal intensity at a peak position (temperature: 101 K) in the DLTS spectrum of the E3 Fit., was plotted on the vertical axis, and a carbon concentration determined by SIMS determination was plotted on the horizontal axis.

[0079] A capture level density Nt at Ec-0.10 eV is determined from the DLTS signal intensity at the peak position (temperature: 76 K) in the DLTS spectrum of the E1 Fit.

[0080] A capture level density Nt at Ec-0.13 eV is determined from the DLTS signal intensity at the peak position (temperature: 87 K) in the DLTS spectrum of the E2 Fit.

[0081] A capture level density Nt at Ec-0.15 eV is determined from the DLTS signal intensity at the peak position (temperature: 101 K) in the DLTS spectrum of the E3 Fit.

[0082] As in Fig. 3, Fig. 4 to Fig. As shown in Figure 5, all three calibration curves exhibited positive slopes, and therefore a positive correlation between the capture level density of each capture level determined from the DLTS signal intensity and the carbon concentration can be confirmed. Therefore, a carbon concentration of less than 1.0 E+16 atoms / cm³ could be assumed. 3 be determined, and a carbon concentration equal to or less than 1.0E+15 atoms / cm³ 3 , which is difficult to determine using the FT-IR method, can be determined by using the in Fig. 3, Fig. 4 to Fig. The calibration curves shown in section 5 can also be determined. Furthermore, a lower carbon concentration can also be achieved by using the curves shown in the diagram. Fig. 3, Fig. 4 to Fig. The correlation equations shown in the 5 diagrams can be determined.

[0083] Among these, the one for the capture level at Ec-0.15 eV showed Fig. The calibration curve shown in section 5 shows a strong correlation, where the square of the correlation coefficient R 2 0.8 or higher, and thus it was also confirmed that the carbon concentration can be determined with higher accuracy by using such a calibration curve.

[0084] Furthermore, among the respective capture levels, the DLTS spectrum with a capture level at Ec-0.13 eV is suitable for the quantitative determination of a further trace amount of carbon, as the peak shape is sharper. [Example 2]

[0085] In a similar manner to Example 1, the DLTS determination and SIMS determination were carried out using silicon samples cut from the silicon single crystal ingots with different carbon concentrations. Fig. Figure 6 shows a calibration curve generated using the DLTS spectrum obtained by setting the blocking voltage to -2 V, in which the capture level density Nt, determined from the DLTS signal intensity at the peak position (temperature: 101 K) in the DLTS spectrum of the E3 Fit., is plotted on the vertical axis during the DLTS measurement, and the carbon concentration determined by the SIMS determination is plotted on the horizontal axis.

[0086] On the other hand, the in Fig. The calibration curve shown in Figure 7 is a calibration curve created using the DLTS spectrum obtained during the switching of the blocking voltage according to the dopant concentration of the silicon sample by using Equation 1, such that the width of the depletion layer (depth from the surface) is in a range of 3.0 to 4.5 µm (i.e., |Wa-Wb| ≤ 1.5 µm) in each sample, with the capture level density Nt, determined from the DLTS signal intensity at the peak position (temperature: 101 K) in the DLTS spectrum of the E3 Fit., plotted on the vertical axis, and the carbon concentration, determined by the SIMS determination, plotted on the horizontal axis. The blocking voltage applied to each silicon sample was a blocking voltage calculated using Equation 1 shown above, such that the width W of the depletion layer was in a range of 3.0 to 4.5 µm according to the dopant concentration of each silicon sample.

[0087] When making a comparison between the in Fig. 6 shown calibration curve and the one in Fig. The calibration curve shown in section 7 had a squared correlation coefficient in the Fig. The calibration curve shown in Figure 7 is closer to Figure 1, and it can therefore be confirmed that a calibration curve allowing a determination of the carbon concentration with higher accuracy can be created by determining the applied blocking voltage such that the widths of the depletion layers are set to the same level. To perform a determination with higher accuracy, it is preferred to set the blocking voltage so that the width of the depletion layer is set to the same level as the width of the depletion layer in the DLTS determination used to generate the calibration curve, even in the DLTS determination of the silicon sample to be determined.

[0088] Using the calibration curve generated as in Example 1 and Example 2, a carbon trace concentration can be determined with high accuracy by using the DLTS spectrum measured after introducing hydrogen atoms into the silicon sample to be determined, without the use of an electron beam irradiation treatment. [Example 3]

[0089] In the silicon single-crystal ingots evaluated in Example 2, the production conditions for the silicon single-crystal ingot were set to 1.0E+15 atoms / cm³. 3 exceeding carbon concentration, determined from the in Fig. The calibration curve shown in 7 was modified by changing the purity of the raw polysilicon or the like, so that the amount of added carbon was reduced, and then the silicon single-crystal ingot was grown under the modified production conditions. Each silicon sample was cut from the top, bottom, and the region between the two parts of the grown ingot, and the DLTS determination was performed in a manner similar to that in Example 1. In the determination, the blocking voltage was varied according to the dopant concentration of the silicon sample using Equation 1, so that the width of the depletion layer was in a range of 3.0 to 4.5 µm. When the carbon concentration in the sample taken from the top, bottom, and region between the grown ingot was determined using Equation 1, the DLTS determination was performed in a manner similar to that in Example 1. Fig.The calibration curve shown in Figure 7, based on the DLTS signal intensity at the peak position (temperature: 101 K) in the DLTS spectrum of the E3 Fit, showed that all carbon concentrations were equal to or less than 1.0 E+15 atoms / cm³. 3 It was confirmed that the addition of carbon was suppressed by changing the production conditions, and a silicon single-crystal ingot with a lower carbon concentration was obtained. The oxygen concentrations in this silicon sample, determined by the FT-IR method, ranged from 2.0 × 10¹⁷ to 12.0 × 10¹⁷ atoms / cm³. 3 .

[0090] Although the DLTS method was used as the evaluation method for evaluating the capture level in the silicon band gap in the examples above, the carbon concentration can be determined without electron beam irradiation by using different evaluation methods for evaluating the capture levels in the silicon band gap, based on the evaluation results of at least one capture level, selected from the group consisting of Ec-0.10 eV, Ec-0.13 eV and Ec-0.15 eV, obtained for the silicon sample in which hydrogen atoms have been introduced. Commercial applicability

[0091] One aspect of the present invention can be used in the technical field of silicon single-crystal ingots and silicon wafers.

Claims

[1] Method for determining the carbon concentration of a silicon sample, comprising: Introducing hydrogen atoms into a silicon sample to be determined; Subjecting the silicon sample to be determined, into which hydrogen atoms have been introduced, to an evaluation by an evaluation method of evaluating a capture level in a silicon band gap without electron beam irradiation; and Determining the carbon concentration of the silicon sample to be determined based on the evaluation result of at least one capture level selected from the group consisting of Ec - 0.10 eV, Ec - 0.13 eV and Ec - 0.15 eV under evaluation results obtained by the evaluation, where the specific carbon concentration is less than 1.0E+16 atoms / cm³ 3 . [2] Method for determining the carbon concentration of a silicon sample according to claim 1, wherein the determined carbon concentration is equal to or less than 1.0E+15 atoms / cm³ 3 . [3] Method for determining the carbon concentration of a silicon sample according to claim 1 or 2, wherein the oxygen concentration of the silicon sample to be determined, as determined by the FT-IR method, is equal to or greater than 1.0E+17 atoms / cm³ 3 . [4] Method for determining the carbon concentration of a silicon sample according to any one of claims 1 to 3, wherein the carbon concentration of the silicon sample to be determined is determined on the basis of an evaluation result at Ec - 0.15 eV. [5] Method for determining the carbon concentration of a silicon sample according to any one of claims 1 to 4, wherein the introduction of hydrogen atoms into the silicon sample to be determined is carried out by immersing the silicon sample to be determined in a solution. [6] Method for determining the carbon concentration of a silicon sample according to claim 5, wherein the solution is hydrofluoric acid. [7] Method for determining the carbon concentration of a silicon sample according to any one of claims 1 to 6, wherein the carbon concentration of the silicon sample to be determined is determined by using a calibration curve based on the evaluation result. [8] Method for determining the carbon concentration of a silicon sample according to claim 7, comprising: Introducing hydrogen atoms into a plurality of silicon samples to generate a calibration curve in which carbon concentrations, determined by an evaluation method other than the evaluation method, are known, Subjecting the majority of silicon samples to generate a calibration curve, into which hydrogen atoms have been introduced, to evaluation by the same evaluation procedure as that used for the silicon sample to be determined, and generating the calibration curve by using the evaluation result at the same capture level as the capture level used to determine the carbon concentration of the silicon sample to be determined, as well as the known carbon concentrations. [9] Method for determining the carbon concentration of a silicon sample according to claim 8, wherein the evaluation method is the DLTS method. [10] Method for determining the carbon concentration of a silicon sample according to claim 9, wherein the width Wa of a depletion layer formed on the silicon sample for generating the calibration curve during the evaluation of the silicon sample to be determined by the DLTS method, and the width Wb of a depletion layer formed on the silicon sample for generating the calibration curve during the evaluation of the silicon sample to be determined by the DLTS method satisfy the following equation: |Wa−Wb|=2.0 [11] Method for producing a silicon single-crystal ingot comprising: Growing a silicon single crystal using the Czochralski method; Determining the carbon concentration of a sample cut from the silicon single-crystal ingot by the method according to one of claims 1 to 10; Determining the manufacturing conditions for a silicon single-crystal ingot based on the determined carbon concentration of the silicon sample; as well as Growth of the silicon single-crystal ingot by the Czochralski process under the specified manufacturing conditions. [12] Method for producing a silicon single-crystal ingot according to claim 11, wherein a silicon sample cut from the tip of the silicon single-crystal ingot grown under the specified production conditions has a carbon concentration of equal to or less than 1.0E+15 atoms / cm². 3 exhibits, determined by the method according to any one of claims 1 to 10. [13] Method for producing a silicon single-crystal ingot according to claim 11 or 12, wherein the silicon sample cut from the tip has an oxygen concentration of equal to or greater than 1.0E+17 atoms / cm³ 3 , determined by the FT-IR method, exhibits. [14] Method for producing a silicon single-crystal ingot according to any one of claims 11 to 13, wherein the silicon single-crystal ingot, grown under the specified production conditions, has a carbon concentration of equal to or less than 1.0E+15 atoms / cm³ 3 , is determined by the method according to one of claims 1 to 10, of a sample cut from the silicon single crystal ingot, wherein the measurement is carried out over an entire area from the tip to the bottom of the silicon single crystal ingot. [15] A method for producing a silicon single-crystal ingot according to any one of claims 11 to 14, wherein the silicon single-crystal ingot, grown under the specified production conditions, has an oxygen concentration of equal to or greater than 1.0E+17 atoms / cm². 3exhibits, determined by the FT-IR method, a silicon sample cut from the silicon single crystal ingot, the measurement being carried out over the entire area from the tip to the bottom of the silicon single crystal ingot.

Images