Calibration curve determination methods, carbon concentration measurement methods and silicon wafer manufacturing methods

Abstract

A calibration curve determination method characterized by: a first step (S11) of providing a plurality of wafer groups, each of which includes a plurality of silicon wafers having the same oxygen concentration and different carbon concentrations, the wafer groups differing from one another in the oxygen concentration of the silicon wafers enclosed therein; a second step (S12) of irradiating a plurality of areas of each of the silicon wafers enclosed in each of the wafer groups with electron beams at different irradiation doses; a third step (S13) of obtaining, by a photoluminescence method, an intensity ratio of a first luminescence intensity derived from silicon to a second luminescence intensity derived from carbon.for each of the areas irradiated with the electron beams; a fourth step (S14) of classifying a plurality of data, each containing the oxygen concentration, the carbon concentration, the irradiation dose, and the intensity ratio, in a plurality of data sets, wherein the data contained in each of the data sets have the same irradiation dose and the same oxygen concentration, and the data sets differ from each other in at least one of the irradiation dose and the oxygen concentration, and determining a first calibration curve representing a relationship between the intensity ratio and the carbon concentration for each of the data sets to obtain a calibration curve group; a fifth step (S15) of selecting one or more combinations from the calibration curve group, wherein each of the one or more combinations represents a pair of the first calibration curves,which are equal to each other in irradiation dose and differ from each other in oxygen concentration, wherein the one or more combinations differ from each other in irradiation dose, and obtaining a difference between the curves of the paired first calibration curves in a double logarithmic plot for each of the one or more combination(s); and a sixth step (S16) which includes selecting a specific combination from the one or more combination(s), wherein the specific combination has a smaller difference between the curves within a predetermined range than the difference between the curves of one or more other combinations, determining the irradiation dose common to the first calibration curves of the specific combination as the electron beam irradiation dose for measurement,and determining a second calibration curve using all data included in the pair or pairs of datasets to determine the pair or pairs of first calibration curves connected in at least the specific combination, wherein the second calibration curve represents a relationship between the intensity ratio and the carbon concentration and can be used for measurement in combination with the electron beam irradiation dose.

Description

REGIONThe present invention relates to a calibration curve determination method, a carbon concentration measurement method, and a silicon wafer production method.BACKGROUNDIn recent years, energy savings and shifts have been driven toward clean energies. In line therewith, the scope of the market of power electronics has been multiplied.A silicon wafer, for example, is used for the production of power electronics. Such a silicon wafer preferably contains almost no impurities such as carbon or oxygen. Contamination of the silicon wafer with contaminants is, however, unavoidable in the manufacturing process. For this reason, it is important to know the concentration of impurities contained in the silicon wafer.A silicon single crystal having desirable characteristics for a substrate of a power device has a carbon concentration equal to or less than a lower determination limit in secondary ion mass spectrometry (SIMS), which is a known measurement method. It has therefore been difficult to accurately determine the carbon concentration in such a silicon single crystal. Under these circumstances, a measurement method using a photoluminescence (PL) method has been proposed as a method for measuring low carbon concentrations.The PL method is a method including irradiating a material with excitation light and observing the emitted light when the excited electrons transition to the ground state.Japanese Patent Application Laid-Open No. 2013-152 977 describes a method for measuring a concentration of an impurity in a semiconductor wafer using the PL method. According to this method, the semiconductor wafer is irradiated with an electron beam to make the impurity contained in the semiconductor wafer luminescently active, and then the luminescence intensity is measured by the PL method. A ratio of the intensity at 1280 nm to the intensity at 1570 nm is obtained from the spectrum. A calibration curve representing the relationship between the luminescence intensity ratio thus obtained and the concentration of the impurity contained in the semiconductor wafer is then determined. Thereafter, the calibration curve is extrapolated to a lower carbon concentration range, and measurement of the luminescence intensity is performed on a wafer containing an impurity at a low concentration. This luminescence intensity is referred to as a calibration curve to obtain the impurity concentration.Japanese Patent Application Laid-Open No. 2015-101 529 A describes a carbon oxygen measurement method by which the influence of the oxygen concentration on the measurement result becomes small. According to this method, luminescence intensity ratios obtained by the PL method and concentration ratio values, which are the ratios of carbon concentrations and oxygen concentrations, are obtained for a plurality of silicon single crystals having different carbon and oxygen concentrations. A calibration curve representing the relationship between the photoluminescence ratios and the concentration ratio values is determined. Thereafter, the luminescence intensity measurement is carried out on a silicon single crystal sample, and the measurement results are referred to the calibration curve to obtain the concentration ratio value. Subsequently, the oxygen concentration of a silicon single crystal sample is measured, and the carbon concentration is calculated from the oxygen concentration and the concentration ratio value.Furthermore, the laid-open specification US 2015 / 0 338 276 A1 describes a method for generating a calibration curve. The method includes ion implanting different doses of an impurity in a plurality of first samples, measuring an intensity of photoluminescence originating from the impurity by photoluminescence spectroscopy for the first samples, and measuring an intensity of photoluminescence by photoluminescence spectroscopy for a second sample made of the same semiconductor.SUMMARYWhen the intensity of a luminescence derived from carbon is obtained by the PL method, the concentration of oxygen contained in a silicon wafer has an influence on the intensity. For this reason, in order to determine carbon concentrations of a plurality of silicon wafers containing oxygen at different oxygen concentrations using the method described in Japanese Patent Application Laid-Open No. 2013-152,977 A, calibration curves for respective oxygen concentrations must be determined.Japanese Patent Application Laid-Open No. 2015-101 529 A uses, as a parameter of the calibration curve, a concentration ratio value that is the ratio of a carbon concentration to an oxygen concentration. This makes it possible to obtain the carbon concentrations even in the case where the silicon single crystals have different oxygen concentrations.However, the inventors have found that when the carbon concentration of a silicon wafer containing oxygen in high concentration is to be measured using the calibration curve obtained in Japanese Patent Application Laid-Open No. 2015-101 529, the influence of the oxygen concentration cannot be eliminated. For example, a blank produced by the Czochralski method tends to contain oxygen in high concentration. A silicon wafer obtained from such a ingot contains oxygen in a higher concentration as compared with the oxygen concentration range (1·10 17 to 8·10 17 atoms / cm 3), as used for the determination of a calibration curve in Japanese Patent Application Laid-Open No. 2015-101 529 A. In this case, even by using the method described in Japanese Patent Application Laid-Open No. 2015-101 529 A, the influence of the oxygen concentration on the result of measurement performed for the carbon concentration cannot be eliminated, and calibration curves must be determined for respective oxygen concentrations. In view of the above situation, an object of the present invention is to make it possible to measure a carbon concentration using a small number of calibration curves also for a silicon wafer containing oxygen in high concentration.The object of the invention is achieved by the subject matter according to claim 1 of the present invention. Further embodiments are evident from the dependent claims.According to a first aspect of the present invention, there is provided a calibration curve determination method characterized by comprising a first step of providing a plurality of wafer groups each including a plurality of silicon wafers having the same oxygen concentration and different carbon concentrations, the wafer groups differing from each other in oxygen concentration of the silicon wafers included therein; a second step of irradiating a plurality of regions of each of the silicon wafers included in each of the wafer groups with electron beams having different irradiation doses; a third step of obtaining, by a photoluminescence method, an intensity ratio of a first luminescence intensity derived from silicon to a second luminescence intensity derived from carbon for each of the regions irradiated with the electron beams; a fourth step of classifying a plurality of data each including the oxygen concentration, the carbon concentration, the irradiation dose, and the intensity ratio in a plurality of sets of data, wherein the data included in each of the sets of data have the same irradiation dose and the same oxygen concentration, and the sets of data are different from each other at least in one of the irradiation dose and the oxygen concentration, and determining a first calibration curve representing a relationship between the intensity ratio and the carbon concentration for each of the sets of data to obtain a calibration curve group; a fifth step of selecting one or more combinations from the calibration curve group, each of the one or more combinations representing a pair of the first calibration curves that are equal to each other in irradiation dose and different from each other in oxygen concentration, the one or more combinations being different from each other in irradiation dose, and obtaining a difference between curves of the paired first calibration curves in a double logarithmic representation for each of the one or more combinations; and a sixth step including selecting a specific combination from the one or more combinations, the specific combination having a smaller difference between the curves within a predetermined range than the difference between the curves of one or more other combinations, determining the irradiation dose common to the first calibration curves of the specific combination as an electron beam irradiation dose for measurement, and determining a second calibration curve using all data included in the (the) pair / s of the data sets to determine the pair or pairs of first calibration curves connected in at least the specific combination, the second calibration curve representing a relationship between the intensity ratio and the carbon concentration and being usable in combination with the electron beam irradiation dose for measurement.According to a second aspect of the present invention, there is provided a carbon concentration measurement method, characterized by comprising: determining the electron beam irradiation dose for measurement and determining the second calibration curve by the determination method according to the first aspect; irradiating a silicon wafer for measurement whose carbon concentration is unknown with an electron beam at the electron beam irradiation dose for measurement; obtaining, by a photoluminescence method, an intensity ratio of a first luminescence intensity derived from silicon to a second luminescence intensity derived from carbon for each of the regions irradiated with the electron beams; and referring the thus obtained intensity ratio to the second calibration curve to obtain a carbon concentration of the silicon wafer for measurement.According to a third aspect of the present invention, there is provided a silicon wafer manufacturing method characterized by comprising measuring a carbon concentration in the silicon wafer for measurement by the measurement method according to the second aspect.BRIEF DESCRIPTION OF THE DRAWINGSFIG. 1 is a flowchart showing a calibration curve determination method according to an embodiment of the present invention; FIG. 2 is a flow chart showing a silicon wafer manufacturing method according to an embodiment of the present invention; FIG. 3 is a graph showing an effect of a carbon concentration measurement method according to an embodiment of the present invention; FIG. 4 is a double logarithmic graph showing an example of a relationship between a carbon concentration and a luminescence intensity ratio when a calibration curve determination sample is irradiated with an electron beam at an irradiation dose of 1·10 15 electrons / cm 2 ; FIG. 5 is a double logarithmic graph showing an example of a relationship between a carbon concentration and a luminescence intensity ratio when a calibration curve determination sample is irradiated with an electron beam at an irradiation dose of 7·10 15 electrons / cm 2 ; FIG. 6 is a double logarithmic graph showing an example of a relationship between a carbon concentration and a luminescence intensity ratio when a calibration curve determination sample is irradiated with an electron beam at an irradiation dose of 2·10 16 electrons / cm 2 ; FIG. 7 is a diagram showing an example of a relationship between a carbon concentration and a luminescence intensity ratio when a calibration curve determination sample is irradiated with an electron beam at an irradiation dose of 7·10 16 electrons / cm 2 ; FIG. 8 is a half logarithmic graph showing an example of a relationship between an electron beam irradiation dose and a difference in the curve between a pair of first calibration curves; FIG. 9 is a double logarithmic representation showing an example of a second calibration curve according to Example 1; and FIG. 10 is a double logarithmic representation showing an example of a second calibration curve according to Comparative Example 1.DETAILED DESCRIPTIONEmbodiments of the present invention will be described below. The following description is not intended to limit the present invention.Calibration curve determination methodFIG. 1 is a flowchart showing a calibration curve determination method according to an embodiment of the present invention. In this method, a first step S 11, a second step S 12, a third step S 13, a fourth step S 14, a fifth step S 15, and a sixth step S 16 are executed in this order.In the method shown in FIG. 1, a calibration curve is determined using the PL method. Before describing the method in FIG. 1 in detail, the principle of the PL method is described below.The PL method is a known method including irradiating a material with excitation light to cause excitation of electrons and observing light emitted with the transition of the excited electrons to the ground state. Specifically, electron-hole pairs are generated when a silicon wafer is irradiated with light having an energy higher than that of the forbidden band. Here, an excess amount of electron-hole pairs with respect to a thermal equilibrium state is generated in silicon crystals. The excess electron-hole pairs recombine to restore an equilibrium state. Light is emitted in this recombination process. A defect or defect in the silicon wafer can influence the spectrum of this emitted light. Therefore, information concerning impurities in the silicon wafer can be obtained by spectroscopic analysis of light emission in detail.In the carbon impurity evaluation, when a silicon wafer is irradiated with an electron beam before irradiation with excitation light, the carbon impurity in this crystal is rendered luminescent-active. More specifically, when the silicon wafer is irradiated with the electron beam, radiation damage or defects occur in the silicon crystal as follows: Si(s) + e → Si(i) + V Si(i) + C(s) → Si(s) + C(i) C(i) + C(s) → Ci - Cs or C(i) + O(i) → Ci - Oi, wherein V represents a impurity and e represents an electron. The suffix (s) represents that the atom immediately preceding this suffix is positioned at a lattice point (substituting). The suffix (i) represents that the atom immediately before this suffix is positioned between lattice points (interstitial). That is, C(i) represents interstitial carbon, C(s) represents substituting or replacing carbon, and O(i) represents interstitial oxygen. Further, Ci - Cs represents a complex defect of interstitial carbon and substitutional carbon, and C(i) - O(i) represents a complex defect of interstitial carbon and interstitial oxygen.Thus, introduction of the primary defect results in the generation of the complex Ci-Oi and Ci-Cs defects. These Ci-Oi and Ci-Cs defects emit light when irradiated with excitation light, particularly a laser beam such as in the visible light or UV region. The intensity or intensity ratio of the light emitted by these defects is affected by the number of Ci-Oi and Ci-Cs defects, for example. Therefore, analyzing the luminescence intensity or intensity ratio can provide information about carbon impurities in the silicon wafer.The method shown in FIG. 1 will be described in the order of steps.First stepThe first step S 11 is a step of providing a plurality of wafer groups.Each of the wafer groups includes a plurality of silicon wafers having the same oxygen concentration and different carbon concentrations. The wafer groups differ from each other in the oxygen concentrations of the silicon wafers contained therein.Each of the wafer groups is provided as follows. A plurality of silicon wafers are cut from one or more silicon single crystal blanks produced by the Czochralski method, for example. The carbon concentrations and the oxygen concentrations of these silicon wafers are determined. Secondary ion mass spectrometry (SIMS) can be used as a method for determining carbon concentration. Fourier transform infrared spectroscopy (FT-IR) can be used as a method for determining the oxygen concentration. Based on these measurement results, the silicon wafers are classified into the wafer groups.The number of silicon wafers included in each of the wafer groups is two or more. The accuracy of the calibration curves is increased as the number of silicon wafers having different carbon concentrations is increased. The number of silicon wafers contained in each of the wafer groups is preferably three or more, and more preferably four or more.The carbon concentrations of the silicon wafers contained in each of the wafer groups are preferably in the range of 4·10 14 to 4·10 15 atoms / cm 3. When the carbon concentration is too low, the carbon concentration cannot be determined by a measurement method other than the PL method. If the carbon concentration is too high, the proportion of carbon contained in the silicon wafer is excessively high, and the electron beam irradiation cannot sufficiently cause luminescence activation. Note that the luminescence activation by the electron beam irradiation will be described in detail below.The number of wafer groups is two or more. As the number of wafer groups different in oxygen concentration of the silicon wafers is increased, the accuracy of the calibration curves is improved. The number of wafer groups is preferably three or more, and more preferably four or more.The oxygen concentration of the silicon wafers contained in each wafer group preferably falls within the range of 1·10 17 to 2·10 13 atoms / cm 3( ASTM '79). When the oxygen concentration is too low and a silicon wafer for measurement, i.e., a silicon wafer on which the oxygen concentration measurement is to be performed using a calibration curve determined by the method described herein has a relatively high oxygen concentration, the generation balance between a G-line photoluminescence derived from a complex Ci-Cs defect and a C-line photoluminescence derived from a complex Ci-Oi defect is significantly different between the measurement by the PL method performed to determine the calibration curve and the measurement by the PL method performed on the silicon wafer for measurement.As a result, there is a possibility that calibration curves must be determined for respective oxygen concentrations. In the case where the oxygen concentration is too high, the amount of oxygen precipitates increases, which reduces the tendency to generate complex Ci-Oi defects as compared with the case where oxygen atoms are uniformly distributed in the silicon wafer. In this case, therefore, when the oxygen concentration of the silicon wafer for measurement is relatively low, the generation balance between G-line photoluminescence and C-line photoluminescence greatly differs between the measurement for the PL method performed to determine the calibration curve and the measurement by the PL method performed on the silicon wafer for measurement. As a result, there is a possibility that the carbon concentration of the silicon wafer cannot be accurately measured for measurement.The silicon wafers for determining the calibration curves and the silicon wafers for measuring may contain a dopant at a concentration falling within the range in which a first luminescence intensity is measurable, which will be described later. The dopant is added to adjust the resistivity of a semiconductor. Examples of conventional dopants are boron (B), phosphorus (P), aluminum (Al) and arsenic (As). These dopants can be determined as bound exciton luminescence (BE luminescence) in the measurement by means of the PL method. If the dopant concentration is too high, a peak derived from the first photoluminescence disappears in the tail of a peak derived from the BE luminescence. This may make it difficult or impossible to detect the first photoluminescence.Second stepThe second step S 12 is a step of irradiating an electron beam to a silicon wafer to render the carbon contained in the silicon luminescently active. Specifically, a plurality of regions of each silicon wafer included in each of the wafer groups provided in the first step S 11 are irradiated with electron beams at different irradiation doses.The electron beam irradiation doses preferably fall within the range of 1·10 14 to 1·10 8 electrons / cm 2. The electron beam irradiation doses more preferably fall within the range of 1·10 15 to 1·10 17 electrons / cm 2. If the irradiation dose of the electron beams is too low, the carbon contained in the silicon wafer cannot be sufficiently luminescently activated. In contrast, when the electron beam irradiation dose is too high, radiation damage in the silicon wafer becomes large, increasing the number of non-radiation active centers that do not generate luminescence although they have a defect level. As a result, too low a value can be detected as the luminescence intensity.In the case where the electron beam irradiation is performed at a sample temperature not exceeding 100° C., annealing of the radiation defect such as a void introduced by the electrode irradiation, which causes recovery of the defect, can be suppressed as compared with the case where the electron beam irradiation is performed at a higher temperature. Examples of the irradiation method include a method in which irradiation is performed under air cooling and a method in which irradiation is performed in the atmosphere in which an irradiation table is water cooled.A particle beam other than an electron beam may be used for luminescence activation of an impurity such as carbon. For example, the irradiation may be performed with a high-energy particle beam such as protons or various ion species. In this case, however, not only atomic vacancies (V) are introduced, but generation of secondary defects such as vacancy clusters increases as compared with the case where irradiation with an electron beam is performed. Accordingly, when a silicon wafer containing a carbon impurity is irradiated with protons, photoluminescence other than that originating from the carbon impurity increases. Examples of such photoluminescence not found in electron beam irradiation are photoluminescence resulting from interstitial silicon, photoluminescence resulting from crystal stresses, and photoluminescence resulting from irradiation damage. In addition, the number of non-radiating centers increases. As a result, it becomes difficult to derive an accurate correlation between the impurity concentration and the luminescence intensity.Therefore, when PL measurement for carbon impurity is performed in a silicon wafer, an electron beam that suppresses photoluminescence irrelevant to carbon concentration evaluation and decreases the number of non-radiating centers is most suitable for luminescence activation. Electron beam irradiation is also preferred in that it enables uniform introduction of defects in the depth direction.Depending on the luminescence activation state, the luminescence intensity obtained by the PL method after the luminescence activation process of the impurity may be weaker than when a certain period of time has elapsed after the luminescence activation process. This is because immediately after the luminescence activation process, there are many unstable irradiation defects which are restored at room temperature. In this case, it will increase the luminescence intensity by holding the wafer at room temperature for a long period of time or annealing the wafer at a low temperature. Note that annealing must be performed in a state where defects contributing to carbon-related photoluminescence are not recovered.It is possible that the silicon wafer is singulated into pieces in advance and then the electron beam irradiation is performed on the pieces. In the case of dicing the silicon wafer, each piece is prepared to have a size of, for example, 5 mm·10 mm.Third stepThe third step S 13 is a step of obtaining an intensity ratio of the first luminescence intensity derived from silicon to a second luminescence intensity derived from carbon for each of the regions irradiated with the electron radiation.Preferably, the first luminescence intensity is of the luminescence type of the free excitons (FE luminescence), and the second luminescence intensity is derived from photoluminescence derived from a complex defect of interstitial carbon and replacing (substitutional) carbon. As the excitation light, for example, an excitation laser beam having a wavelength of 532 nm can be used.In the method shown in FIG. 1, for example, the G-line luminescence intensity is measured as the first luminescence intensity. Although the C-line luminescence also exists as a peak representing the carbon concentration, the C-line is derived from Ci-Oi formed by carbon and oxygen. In contrast, the G line is derived from Ci-Cs formed solely by carbon. For this reason, the G-line luminescence shows a larger response change with a change in carbon concentration. Therefore, measuring the G-line luminescence makes it possible to more accurately measure the concentration of the carbon impurity.The luminescence intensity is measured in a state in which, for example, a silicon wafer is immersed in a cooling liquid such as liquid helium and the temperature is maintained constant. This makes it possible to suppress generation of heat by a sample due to irradiation with the excitation laser beam and to avoid changes in balance between various kinds of luminescence such as FE luminescence due to variations in the sample temperature, and measurement can be performed more stably.Fourth stepThe fourth step S 14 is a step of determining a first calibration curve representing the relationship between the carbon concentration and the luminescence intensity obtained by the PL method, and obtaining a calibration curve group including a plurality of first calibration curves. Specifically, the plurality of data each including an oxygen concentration, a carbon concentration, an irradiation dose, and an intensity ratio are classified into a plurality of data sets. The data contained in each of the data sets have the same irradiation dose and the same oxygen concentration. The data sets differ from one another in at least one of the irradiation dose or the oxygen concentration. A first calibration curve representing the relationship between the intensity ratio and the carbon concentration is determined for each data set to obtain a calibration curve group.The "same oxygen concentration" means that the oxygen concentration falls within the range between 0.5·C 0( Av) and 1.5·C 0( Av), where Cβ(Av) is the average value of the oxygen concentrations. "Different values of the oxygen concentration" means that the oxygen concentrations are outside the above range.The first calibration curve is a straight line obtained by plotting the logarithm of the luminescence intensity ratio over the logarithm of the carbon concentration.Let [G PL / FE PL] be the ratio between the FE luminescence intensity of the G-line intensity [C] be the carbon concentration, a be the graph of the luminescence intensity with respect to the carbon concentration, and A be the intercept of the double logarithmic representation, the relationship therebetween is given by the following equation (1):The first calibration curve represented by the equation (1) is determined for each of the sets of data, thereby obtaining a calibration curve group including the first calibration curves.Fifth stepThe fifth step S 15 is a step of obtaining a difference between curves for at least one combination selected from the calibration curve group. In particular, one or more combinations are selected from the calibration curve group. Here, each of the one or more combinations is a pair of first calibration curves that are identical to each other in the irradiation dose differ from each other in the oxygen concentration, and the one or more combinations differ from each other in the irradiation dose. A difference between the slopes of the paired first calibration curves in a double logarithmic representation is obtained for each of the one or more combinations.The course of the calibration curve is the "a" mentioned in the fourth step S14. A pair of first calibration curves for obtaining the curve a can be selected as desired. When two or more combinations different from each other in the irradiation dose are selected, this selection is also arbitrary.The pair of first calibration curves is preferably selected such that the difference in the course or increase is 0.5 or less. When selecting the two or more combinations different in the irradiation dose, the first calibration curves included in each combination are preferably selected such that the difference in the curves is 0.5 or less. If the variation difference is too large, there is a possibility that the influence of the oxygen concentration on the result of the carbon concentration measurement cannot be eliminated, and thus it may become difficult to determine the carbon concentration with a small number of calibration curves.Sixth stepIn the sixth step S 16, an electron beam irradiation dose for measurement is determined, and a second calibration curve is determined.Specifically, assume that the number of combinations is 1. In this case, the history difference is checked whether it falls within a predetermined range, for example, the range described in the fifth step S 15. The irradiation dose common to the first calibration curves of this combination is determined as the electron beam irradiation dose for measurement. Using all data included in the pair or pairs of data sets used for determining the pair or pairs of first calibration curves included in at least this combination, a second calibration curve representing a relationship between the intensity ratio and the carbon concentration and usable for measurement under the electron beam irradiation dose is determined.Alternatively, it is assumed that the number of combinations is 2 or more. Among these combinations, a specific combination having a rise difference falling within the predetermined range, for example, the range described for the fifth step S 15, or smaller than that in one or more other combinations is selected. The irradiation dose is common to the first calibration curves of the specific combination and is determined as the electron beam irradiation dose for measurement. By using all data included in the pair or pairs of data sets to form the pair or pairs of first calibration curves included in at least this specific combination, a second calibration curve representing the relationship between the intensity ratio and the carbon concentration and usable for measurement under the electron beam irradiation dose is determined.For determining the second calibration curve, it is possible to use, in addition to the above-mentioned pairs of data sets (which will be referred to as irradiation dose determination data sets hereinafter), all the data included in one or more data sets (which will be referred to as auxiliary data set hereinafter), the irradiation dose of which is equal to that of the irradiation dose determination data sets and the oxygen concentration of which is between the oxygen concentration of the one of the irradiation dose determination data sets and the oxygen concentration of the other of the irradiation dose determination data sets. This makes it possible to increase the accuracy of the calibration curve.The sixth step S 16 may further include calculating a correlation coefficient between the carbon concentration and the intensity ratio for all the data included in the irradiation dose determination data sets, and confirming based on the correlation coefficient that the second calibration curve may be used. Alternatively, in cases where the auxiliary data set has been used for determining the second calibration curve, the sixth step S 16 may further include calculating the correlation coefficient between the carbon concentration and the intensity ratio of all the irradiation dose determination data and the auxiliary data set included in the data set, and confirming that the second calibration curve can be used based on the correlation coefficient.The correlation coefficient can be obtained by the following equation (2):R 2 is preferably 0.9 or greater. In the case that R 2 is too small, the correlation between the carbon concentration and the intensity ratio is weak, and it may become difficult to measure a carbon concentration with a smaller number of calibration curves. This checking or confirmation process can be carried out before or after the second calibration curve is determined.If the magnitude of R 2 does not meet the standard, a calibration curve is determined for each oxygen concentration. Alternatively, the process may return to the fifth step S 15 to again select sets of irradiation dose determination data and again determine a second calibration curve.The sixth step S 16 may also include a process of extrapolating the determined second calibration curve to a lower carbon concentration region. This makes it possible to measure the carbon concentration of a silicon wafer for measurement that contains carbon at a lower concentration than the carbon concentration used to form the calibration curve.The carbon concentration measurement method and the silicon wafer production method of FIG. 2 is a flowchart showing an example of a silicon wafer production method including a method of measuring a carbon concentration using the calibration curve obtained by the method of FIG. 1. In this manufacturing method, the seventh step S 17, the eighth step S 18, the ninth step S 19, and the tenth step are executed in this order. Subsequently, in this manufacturing method, based on the result of the tenth step S 20, the eleventh step S 21 is executed, or the twelfth step S 22 and the thirteenth step S 23 are executed in this order.The seventh step S 17 is a step of irradiating a silicon wafer for measurement whose carbon concentration is unknown with an electron beam at the electron beam irradiation dose for measurement determined in the sixth step S 16. The methods etc. of the electron beam irradiation are preferably the same as the methods etc. used to determine the calibration curve.The eighth step S 18 is a step of acquiring the luminescence intensity ratio of a first luminescence intensity derived from silicon, acquired by the PL method, to a second luminescence intensity derived from carbon for a silicon wafer for measurement irradiated with the electron beam. This first luminescence is the same as the first luminescence measured at the time of determining the calibration curve. The first luminescence is, for example, an FE luminescence. The second luminescence is the same as the second luminescence measured at the time of determining the calibration curve. The second luminescence is, for example, a G-line luminescence. The wavelength or the like of the excitation light used for the measurement according to the PL method is preferably the same as the wavelength or the like of the excitation light used for obtaining the calibration curve.The ninth step S 19 is a step of referring the luminescence intensity ratio obtained in the eighth step S 18 to the second calibration curve to obtain the carbon concentration in the silicon wafer for measurement.By executing the seventh step S 17 to the ninth step S 19, the concentration of the carbon contained in the silicon wafer for measurement that is unknown is obtained. This measurement method does not require complex concentration conversion. The tenth step S 20 is a step of determining whether the carbon concentration of the silicon wafer for measurement falls within a predetermined allowable range, for example, a predetermined range allowable for a silicon wafer as a product. The silicon wafer as a product is, for example, a power electronic substrate, a CMOS (Complementary Metal Oxide Semiconductor) image sensor substrate, or the like. A detailed example for determining whether a silicon wafer is defective or not defective will be described as a case used as an example in which this silicon wafer is used for a semiconductor device including a collector and an emitter.First, the allowable range of carbon concentration is determined. The allowable range of the carbon concentration is determined on the basis of an allowable range of, for example, device characteristics such as switching characteristics and the collector-emitter saturation voltage. The carbon concentration of the silicon wafer affects the device characteristics over the lifetime characteristics. For example, the relationship between the carbon concentration of the silicon wafer and the device characteristics such as the collector-emitter saturation voltage is examined in advance. This makes it possible to determine the allowable range of the carbon concentration.The carbon concentration of the silicon wafer for measurement is then obtained by the above measurement method. This carbon concentration is compared with the predetermined allowable carbon concentration range to determine whether the silicon wafer is defect-free or defective for measurement. For example, when the carbon concentration falls within the allowable range, the silicon wafer is classified as a non-defective wafer, and the eleventh step S 21 is executed. For example, the batch of such silicon wafers is delivered.When the carbon concentration is outside the allowable range, the wafer is classified as a defective wafer, and the twelfth step S 22 and the thirteenth step S 23 are executed. For example, by analyzing the silicon wafer to measure and verify the manufacturing process conditions, the cause of the defect is analyzed. For example, the silicon wafer manufacturing process is optimized.As described above, in this measurement method, the electron beam irradiation dose can be optimized. An example of the influence of optimizing the electron beam irradiation dose is shown in FIG. 3.FIG. 3 is a graph showing the effect of the method for measuring the carbon concentration according to the embodiment of the present invention. In FIG. 3, the abscissa represents the carbon concentration, and the ordinate represents the luminescence intensity ratio. In FIG. 3, the data given by [irradiation dose 1·10 15, low [Oi]] was obtained for silicon wafers having oxygen concentrations equal to or greater than 3.5·10 17 atoms / cm 3 and lower than 8.0·10 17 atoms / cm 3( ASTM '79) with the electron irradiation dose of 1·10 15 electrons / cm 2. The data represented by [irradiation dose 1·10 15, high [Oi]] were obtained for silicon wafers having oxygen concentrations of 8.0·10 17 atoms / cm 3 to 1.1·10 13 atoms / cm 3( ASTM '79) with the electron irradiation dose of 1·10 15 electrons / cm 2. The data denoted by [irradiation dose 7·10 16, [low Oi]] became to be obtained for silicon wafers having oxygen concentrations equal to or higher than 3.5·10 17 atoms / cm 3 and lower than 8.0. 10 17 atoms / cm 3( ASTM '79) with the electron irradiation dose of 7·10 16 electrons / cm 2 are obtained. The data denoted by [irradiation dose 7·10 16, high [Oi]] were obtained for silicon wafers having oxygen concentrations of 8.0·10 17 atoms / cm 3 to 1.1·10 18 atoms / cm 3( ASTM '78), with the electron irradiation dose of 7·10 16 electrons / cm 2.As shown in FIG. 3, the data denoted by [irradiation dose 1·10 15, low [Oi]], and the data denoted by [irradiation dose 1·10 15, high [Oi]] were obtained when the silicon wafers were irradiated with the electron beams at irradiation doses that were not optimized. On the other hand, the data denoted by [irradiation dose 7·10 16, low [Oi]] and the data denoted by [irradiation dose 7·10 16, high [Oi]] were obtained when the silicon wafers were irradiated with electron beams at irradiation doses optimized. As is clear from FIG. 3, when the irradiation doses were not optimized, the straight line trace obtained from data for the silicon wafers having low oxygen concentrations and the straight line trace obtained from data only for silicon wafers having a high oxygen concentration have a relatively large difference. On the other hand, the difference between the curves is very small when the irradiation doses have been optimized. That is, in the latter case, the influence of the oxygen concentration on the relationship between the carbon concentration and the luminescence intensity ratio is small. Therefore, the carbon concentration can be measured with a smaller number of calibration curves.As described in Japanese Patent Application Laid-Open 2015-101529, both the luminescence intensity ratio and the oxygen concentration must be measured in order to obtain the carbon concentration of the silicon wafer for measurement. In contrast, in the present measurement method, the influence of the oxygen concentration on the measurement result of the carbon concentration can be eliminated. For this reason, when the luminescence intensity ratio is detected by the PL method, the carbon concentration can be obtained without measurement of the oxygen concentration or the like.In this manufacturing method, the above-mentioned erroneous / non-defective determination may be performed at a predetermined frequency. In this case, when a wafer product whose impurity concentration exceeds an upper limit is manufactured, investigations of the causes of a high carbon concentration and abnormalities in the manufacturing process can be immediately started. This makes it possible to reduce the possibility of manufacturing a large number of wafer products whose carbon concentration exceeds the allowable range, thereby increasing the yield.Working examples of the present invention will now be described.Determination of the calibration curveExample 1Silicon wafers Nos. 1 to 12 were prepared from a plurality of silicon single crystal blanks prepared by the Czochralski method.The carbon concentrations and the oxygen concentrations of these silicon wafers No. 1 to No. 12 were measured. Carbon concentrations were measured by the SIMS method and oxygen concentrations were measured by the FT-IR method. The measurement results are shown in Table 1 below. Table 1 Table 111,3 · 10 155,6 · 10 1725,5 · 10 143,8 · 10 1736,4 · 10 143,6 · 10 1741,0 · 10 153,0 · 10 1752,7 · 10 153,3 · 10 1762,8 · 10 158,5 · 10 1771,8 · 10 151,0 · 10 1881,6 · 10 151,0 · 10 1891,4 · 10 151,0 · 10 18105,6 · 10 141,5 · 10 18116,6 · 10 141,3 · 10 18126,5 · 10 141,1 · 10 18Based on these measurement results above, the silicon wafers No. 1 to No. 12 were classified into a first wafer group including the silicon wafers No. 1 to No. 5 having low oxygen concentrations and a second wafer group including the silicon wafers No. 6 to No. 12 having high oxygen concentrations.Four calibration curve determination samples each having a size of 5 mm·10 mm were cut out from the central portion of each of the silicon wafers No. 1 to No. 12.These calibration curve determination samples were irradiated with electron beams. From the four calibration curve determination samples obtained from each of the silicon wafers No. 1 to No. 12, the first sample was irradiated with an electron beam at an irradiation dose of 1·10 15 electrons / cm 2 the second sample was irradiated with an electron beam at an irradiation dose of 7·10 15 electrons / cm 2 the third sample was irradiated with an electron beam at an irradiation dose of 2·10 16 electrons / cm 2 and the fourth sample was irradiated with an electron beam at an irradiation dose of 7·10 16 electrons / cm 2.The electron beam irradiation with the irradiation dose of 1·10 15 electrons / cm 2 was carried out at a voltage of 2 MeV under air cooling in the atmosphere such that the sample temperature did not exceed 100° C. The electron beam irradiation at the irradiation dose of 7·10 15 electrons / cm 2 was carried out at a voltage of 1 MeV with water cooling of the irradiation stage in the atmosphere such that the sample temperature did not exceed 100° C. Thereby, carbon in each calibration curve determination sample was luminescence-activated.Then, the luminescence intensities were measured by the PL method. This measurement was carried out for all calibration curve determination samples under the same conditions. Specifically, an excitation laser beam having a wavelength of 532 nm was used, and the intensity on each sample surface was set to 100 mW. Each calibration curve determination sample was immersed in liquid helium, and the FE luminescence intensity and the G-line luminescence intensity of each sample were measured with each sample maintained at a temperature of 4.2 K. Each ratio of these luminescence intensities (G / FE luminescence intensity) was determined. The measurement results are shown in Table 2 below. The luminescence intensity ratios thus obtained were classified into a plurality of data sets having different electron beam irradiation doses for each wafer group. A first calibration curve representing the relationship between the luminescence intensity ratio and the carbon concentration was determined for each data set, whereby the calibration curve groups shown in Figs. 4 to 7 were obtained.FIG. 4 is a double logarithmic graph showing an example of the relationship between the carbon concentration and the luminescence intensity ratio when each calibration curve determination sample was irradiated with an electron beam at an irradiation dose of 1·10 15 electrons / cm 2.FIG. 5 is a double logarithmic graph showing an example of the relationship between the carbon concentration and the luminescence intensity ratio when each calibration curve determination sample was irradiated with an electron beam at an irradiation dose of 7·10 15 electrons / cm 2.FIG. 6 is a double logarithmic graph showing an example of the relationship between the carbon concentration and the luminescence intensity ratio when each calibration curve determination sample was irradiated with an electron beam at an irradiation dose of 2·10 16 electrons / cm 2.FIG. 7 is a diagram showing an example of the relationship between the carbon concentration and the luminescence intensity ratio when each calibration curve determination sample was irradiated with an electron beam at an irradiation dose of 7·10 16 electrons / cm 2.In Figs. 4 to 7, the abscissa represents the carbon concentration, and the ordinate represents the luminescence intensity ratio. Further, in FIGS. 4 to 7, a solid line represents the first calibration curve obtained with the first wafer group. A broken line represents the first calibration curve obtained for the second wafer group.A slope was determined for each of the resulting first calibration curves. The result is shown in Table 3. In addition, the obtained results by plotting the curves of the first calibration curves with respect to the logarithms of the electron beam irradiation doses are shown in FIG. 8. Table 3 Table 3Electron beam irradiation dose [electrons / cm 2]1 · 10 157 · 10 152 · 10 167 · 10 16Course of the first calibration curve (first wafer group)0,821,21,72,3Course of the first calibration curve (second wafer group)2,32,12,32,7FIG. 8 is a half logarithmic graph showing an example of the relationship between the electron beam irradiation dose and the difference in increase between the pair of first calibration curves. In FIG. 8, the abscissa represents the logarithm of the electron beam dose, and the ordinate represents the rise of the calibration curve.For each of the calibration curve groups each including a pair of first calibration curves that are equal to each other in electron beam dose and different from each other in oxygen concentration, a difference between the slopes of the paired first calibration curves included in the calibration curve group was determined using the data in Table 3 and FIG. 8. As can be seen from Table 3 and FIG. 8, the aforementioned difference was 0.5 or less when the electron beam irradiation dose was 7·10 16 electrons / cm 2. Thus, the irradiation dose of 7·10 16 electrons / cm 2 was determined as the electron beam irradiation dose for measurement. Note that the difference between the slopes of the first calibration curves obtained when each sample was irradiated with an electron beam at the irradiation dose of 7·10 16 electrons / cm 2 was 0.38.From the data sets used for determining the first calibration curves, the data sets whose electron beam irradiation dose was 7·10 16 electrons / cm 2 were selected, and using all the data included in the selected data sets, the relationship between the luminescence intensity ratios and the carbon concentration was plotted in a double logarithmic representation. The second calibration curve shown in FIG. 9 was obtained from the plotted data.FIG. 9 is a double logarithmic representation showing an example of the second calibration curve according to Example 1. In FIG. 9, the abscissa represents the carbon concentration, and the ordinate represents the luminescence intensity ratio.The second calibration curve shown in FIG. 9 is a straight line represented by y=5·10 -38 x 2,50, where x represents the carbon concentration and y represents the luminescence intensity ratio.The correlation coefficient between the carbon concentration and the luminescence intensity ratio was obtained from all the data to derive the second calibration curve. As a result, R was 20,91, and thus a high correlation was found between the carbon concentration and the luminescence intensity ratio.(Comparative Example 1)The electron beam irradiation dose for measurement was determined to be 1·10 15 electrons / cm 2. Using all data sets whose electron beam irradiation dose was 1·10 15 electrons / cm 2 in Example 1, the relationship between the luminescence intensity ratios and the carbon concentration was plotted as a double logarithmic plot. The second calibration curve shown in Fig. 10 was obtained from these data. FIG. 10 is a double logarithmic representation showing an example of the second calibration curve according to Comparative Example 1. In FIG. 10, the abscissa represents the carbon concentration, and the ordinate represents the luminescence intensity ratio. The second calibration curve is a straight line represented by the same y=6·10 -26 x 1,64.When the correlation coefficient between the carbon concentration and the luminescence intensity ratio was obtained from all the data used to derive this second calibration curve, R was 20,53.(Comparative Example 2)The electron beam irradiation dose for measurement was determined to be 1·10 15 electrons / cm 2. Using all the data sets obtained for the first wafer group from the data sets whose electron beam irradiation dose was 1·10 15 electrons / cm 2 in Example 1, the relationship between the luminescence intensity ratios and the carbon concentration was plotted as a double logarithmic plot. The second calibration curve for a low oxygen concentration shown by a solid line in FIG. 4 was obtained.The electron beam irradiation dose for measurement was determined to be 1·10 15 electrons / cm 2. Using all the data sets obtained for the first wafer group from the data sets whose electron beam irradiation dose was 1·10 15 electrons / cm 2 in Example 1, the relationship between the luminescence intensity ratios and the carbon concentration was plotted as a double logarithmic plot. The second calibration curve for a high oxygen concentration shown by a broken line in FIG. 4 was obtained.The second calibration curve for the low oxygen concentration is a straight line represented by y=3·10 -13 x 0,821. The second calibration curve for the high oxygen concentration is a straight line represented by y=1·10 -36 x 2,33.EvaluationSilicon wafers A-F whose carbon concentrations were unknown were prepared from a plurality of silicon single crystal blanks prepared by the Czochralski method. Of the silicon wafers A-F, the silicon wafers A-C contained oxygen in low concentration, while the wafers D-F contained oxygen in high concentrations. Three measurement samples each having a size of 5 mm·10 mm were cut out from the central portion of each of the silicon wafers A-F.One of the three measurement samples obtained from each wafer was irradiated with an electron beam of an irradiation dose of 7·10 16 electrons / cm 2 and the luminescence intensity ratio was determined using the PL method under the same conditions as in the determination of the first calibration curve in Example 1. The luminescence intensity ratio thus obtained was referred to as the second calibration curve determined in Example 1, whereby a carbon concentration was obtained.Another one of the three measurement samples obtained from each of the silicon wafers A-F was irradiated with an electron beam at the irradiation dose of 1·10 15 electrons / cm 2 and the luminescence intensity ratio was determined using the PL method under the same conditions as in the determination of the first calibration curve in Example 1. For each of the silicon wafers A-F, the luminescence intensity ratio thus obtained was referred to as the second calibration curve determined in Comparative Example 1, whereby a carbon concentration was obtained.For each of the silicon wafers A-C, the luminescence intensity ratio was referred to as the second calibration curve for the low oxygen concentration determined in Comparative Example 2, thereby obtaining a carbon concentration. In addition, for each of the silicon wafers D-F, the luminescence intensity ratio thus obtained was referred to as the second calibration curve for high oxygen concentration determined in Comparative Example 2, thereby obtaining a carbon concentration.For the remaining one of the three measurement samples obtained from each wafer, the carbon concentration was measured by the SIMS method, and an oxygen concentration was measured by the FT-IR method.The results are shown in Table 4. Table 4 Table 4G / FE luminescence intensity at irradiation dose of 7·10 16 electrons / cm23,4 · 10 01,7 · 10 14,0 · 10 -16,0 · 10 -19,3 · 10 02,5 · 10 1Carbon concentration obtained using the calibration curve of Example 1 (atoms / cm 3)1,4 · 10 152,6 · 10 155,8 · 10 146,8 · 10 142,0 · 10 153,0 · 10 15G / FE luminescence intensity at irradiation dose of 1·10 15 electrons / cm21,0 · 10 01,2 · 10 03,1 · 10 -14,1 · 10 -25,3 · 10 -11,5 · 10 0Carbon concentration obtained by using the calibration curve of Comparative Example 1 (atoms / cm 3)2,4 · 10 152,7 · 10 151,2 · 10 153,4 · 10 141,6 · 10 153,1 · 10 15Carbon concentration obtained by using the calibration curve for low oxygen concentration (atoms / cm 3)1,8 · 10 152,2 · 10 154,2 · 10 14---Carbon concentration obtained by using the calibration curve for high oxygen concentration (atoms / cm 3)---7,2 · 10 142,2 · 10 153,4 · 10 15Carbon concentration by SIMS (atoms / cm 3)1,4 · 10 152,6 · 10 155,5 · 10 147,5 · 10 141,7 · 10 153,0 · 10 15Oxygen concentration (atoms / cm 3) (ASTM '79)3,2 · 10 175,4 · 10 176,0 · 10 179,3 · 10 171,2 · 10 181,5 · 10 18As shown in Table 4, the carbon concentrations obtained by the process including irradiation with an electron beam at an irradiation dose as determined in Example 1 and relating to the second calibration curve determined in Example 1 almost coincide with the carbon concentrations determined by the SIMS method irrespective of the oxygen concentrations. This shows that the carbon concentration could be measured using only the calibration curve regardless of the oxygen concentration when the second calibration curve was used.On the other hand, the carbon concentrations obtained by the process including irradiation with an electron beam at the irradiation dose mentioned in Comparative Example 1 which was based on the second calibration curve determined in Comparative Example 1 are significantly different from the carbon concentrations determined by the SIMS method. This shows that it was difficult to determine the carbon concentration of a silicon wafer containing oxygen at a high concentration, for example, by using only the second calibration curve of Comparative Example 1.When the carbon concentrations of the silicon wafers A-C were determined with reference to, for example, the second calibration curve and the low oxygen concentration, the carbon concentrations thus obtained exhibited small deviations from the results of the carbon concentrations measured by SIMS. On the other hand, when the carbon concentration of the silicon wafer D was determined with reference to the second calibration curve for the low oxygen concentration, the obtained value had a large deviation from the result of the carbon concentration measurement carried out by SIMS.When the performed electron beam irradiation was not optimized, a calibration curve for each oxygen concentration was to be determined as the second calibration curve determined in Comparative Example 2 to measure the carbon concentration precisely.

Claims

A calibration curve determination method characterized by comprising: a first step (S11) of providing a plurality of wafer groups each including a plurality of silicon wafers having the same oxygen concentration and different carbon concentrations, the wafer groups differing from each other in oxygen concentration of the silicon wafers included therein; a second step (S12) of irradiating a plurality of regions of each of the silicon wafers included in each of the wafer groups with electron beams having different irradiation doses; a third step (S13) of obtaining, by a photoluminescence method, an intensity ratio of a first luminescence intensity derived from silicon to a second luminescence intensity derived from carbon for each of the regions irradiated with the electron beams; a fourth step (S14) of classifying a plurality of data each including the oxygen concentration, the carbon concentration, the irradiation dose and the intensity ratio into a plurality of sets of data, wherein the data included in each of the sets of data have the same irradiation dose and the same oxygen concentration and the sets of data are different from each other in at least one of the irradiation dose and the oxygen concentration, and determining a first calibration curve representing a relationship between the intensity ratio and the carbon concentration for each of the sets of data to obtain a calibration curve group; a fifth step (S 15) of selecting one or more combinations from the calibration curve group, each of the one or more combinations representing a pair of the first calibration curves that are equal to each other in irradiation dose and different from each other in oxygen concentration, the one or more combinations being different from each other in irradiation dose, and obtaining a difference between curves of the paired first calibration curves in a double logarithmic representation for each of the one or more combination / s; and a sixth step (S16) including selecting a specific combination from the one or more combination / s, the specific combination having a smaller difference between the curves within a predetermined range than the difference between the curves of one or more other combinations, determining the irradiation dose common to the first calibration curves of the specific combination as an electron beam irradiation dose for measurement, and determining a second calibration curve using all data included in the pair or pairs of data sets to determine the pair or pairs of first calibration curves connected in at least the specific combination, the second calibration curve representing a relationship between the intensity ratio and the carbon concentration and being usable in combination with the electron beam irradiation dose for measurement.The determination method according to claim 1, wherein the carbon concentration of the silicon wafers in each of the wafer groups falls within a range of 4·10 14 to 4·10 15 atoms / cm 3.The determination method according to claim 1 or 2, wherein the oxygen concentrations of the silicon wafers fall within a range between 1·10 17 and 2·10 18 atoms / cm 3( ASTM '79) in each of the wafer groups.The determination method according to any one of claims 1 to 3, wherein the irradiation dose of the electron beams in the second step falls within the range of 1·10 14 to 1·10 18 electrons / cm 2.The determination method according to any one of claims 1 to 4, wherein the differential amount in the sixth step is 0.5 or less.The determination method according to any one of claims 1 to 5, wherein the sixth step further includes calculating a correlation coefficient between the carbon concentration and the intensity ratio from all the data included in the pair of data sets used or to be used to determine the pair of first calibration curves included in the specific combination, and confirming that the calibration curve can be used based on the correlation coefficient.The determination method according to any one of claims 1 to 6, wherein in the sixth step, the second calibration curve is determined using all the data included in the pair of data sets, and further using all the data included in one or more data sets, wherein the irradiation dose of the one or more data sets is the same as that of the pair of data sets, and the oxygen concentration of the one or more data sets is between the oxygen concentration of the one of the pair of data sets and the oxygen concentration of the other of the pair of data sets.A carbon concentration measurement method characterized by comprising: determining the electron beam irradiation dose for measurement and determining the second calibration curve by the determination method according to any one of claims 1 to 7; irradiating a silicon wafer for measurement whose carbon concentration is unknown with an electron beam at the electron beam irradiation dose for measurement; obtaining, by a photoluminescence method, an intensity ratio of a first luminescence intensity derived from silicon to a second luminescence intensity derived from carbon for each of the regions irradiated with the electron beams; and referring the intensity ratio thus obtained to the second calibration curve to obtain a carbon concentration of the silicon wafer for measurement.The measurement method according to claim 8, wherein known concentrations of the plurality of silicon wafers and the silicon wafer for measurement fall within a range in which the first luminescence intensity can be measured.A silicon wafer manufacturing method characterized by comprising measuring a carbon concentration in the silicon wafer for measurement by the measurement method according to claim 8 or 9.

Images