Analysis method, program, and method for manufacturing a semiconductor device

Abstract

To enable accurately analyzing a defect rate of a semiconductor device.SOLUTION: Provided is an analysis method comprising: a measurement step that acquires measurement values of a first characteristic and a second characteristic of a plurality of semiconductor devices of a measurement group in which concentration of a first impurity and an irradiation amount of a charged particle beam are included in a setting range; a measurement distribution generation step that generates measurement distribution indicating distribution of the measurement value of the first characteristic and the second characteristic in the measurement group; a virtual distribution generation step that simulates the first characteristic and the second characteristic of a plurality of virtual semiconductor devices on the basis of the measurement distribution to generate virtual distribution in which samples of the first characteristic and the second characteristic are distributed in a more wider range than the measurement distribution; and a defect rate calculation step that calculates a defect rate in the virtual distribution.SELECTED DRAWING: Figure 8

Description

[Technical Field]

[0001] This invention relates to an analysis method, a program, and a method for manufacturing a semiconductor device. [Background technology]

[0002] Conventionally, a technique has been known for adjusting the carrier lifetime by irradiating a semiconductor substrate on which a semiconductor device such as an IGBT (Insulated Gate Bipolar Transistor) is formed with charged particles such as helium ions to form carrier recombination centers (see, for example, Patent Documents 1 and 2). Patent Document 1: Japanese Unexamined Patent Publication No. 2019-121657 Patent Document 2: Japanese Unexamined Patent Publication No. 2009-188336 [Overview of the Initiative] [Problems that the invention aims to solve]

[0003] It is preferable to be able to accurately analyze the failure rate of semiconductor devices. [Means for solving the problem]

[0004] To solve the above problems, a first embodiment of the present invention provides an analysis method for analyzing the defect rate of a semiconductor device in which a semiconductor substrate contains a first impurity and the semiconductor substrate is irradiated with a charged particle beam. The analysis method may include a measurement step of acquiring measured values ​​of first and second characteristics of multiple semiconductor devices in a measurement group in which the concentration of the first impurity and the irradiation amount of the charged particle beam fall within a set range. The analysis method may include a measurement distribution generation step of generating a measurement distribution that shows the distribution of measured values ​​of the first and second characteristics in the measurement group. The analysis method may include a virtual distribution generation step of simulating the first and second characteristics of multiple virtual semiconductor devices based on the measurement distribution to generate a virtual distribution in which the samples of the first and second characteristics are distributed over a wider range than in the measurement distribution. The analysis method may include a defect rate calculation step of calculating the defect rate in the virtual distribution.

[0005] The virtual distribution generation step may generate a virtual distribution based on the covariance of the first and second characteristics in the measured distribution.

[0006] In the virtual distribution generation stage, a virtual distribution may be generated for each set range using the covariance of the corresponding measurement distribution.

[0007] During the virtual distribution generation stage, a common covariance may be used to generate virtual distributions for multiple defined ranges.

[0008] The analysis method may include a measurement group distribution generation step in which group representative values ​​of the first and second characteristics are obtained for each measurement group for multiple measurement groups corresponding to the same setting range, and a measurement group distribution showing the distribution of the group representative values ​​is generated. In the virtual distribution generation step, a virtual distribution may be generated based on the measurement distribution and the measurement group distribution.

[0009] During the virtual distribution generation stage, a virtual distribution may be generated for the same setting range as the setting range of the measurement group.

[0010] The analysis method may include a virtual group generation step that generates multiple virtual groups for the same setting range as the measurement group based on the measurement group distribution. In the virtual distribution generation step, the measurement distribution may be applied to the group representative value of each virtual group to generate the virtual distribution.

[0011] In the virtual distribution generation stage, virtual distributions may be generated for each set range. In the defect rate calculation stage, the defect rate for each set range may be calculated using the virtual distributions corresponding to each set range.

[0012] The analysis method may include a relationship information generation step of obtaining a distribution representative value indicating the representative value of each measurement group distribution and generating relationship information indicating the relationship between the values in the set range and the distribution representative value. The analysis method may include a virtual group distribution generation step of generating a virtual group distribution for a set range different from the measurement group distribution based on the relationship information. In the virtual distribution generation step, a virtual distribution for each set range may be generated using the virtual group distribution corresponding to each set range. In the defect rate calculation step, the defect rate of the set range corresponding to the virtual group distribution may be calculated.

[0013] The analysis method may include a shape information acquisition step of obtaining shape information indicating the shape of each measurement group distribution. In the virtual group distribution generation step, the virtual group distribution may be further generated based on the shape information.

[0014] In the defect rate calculation step, the defect rate for each of the first characteristic and the second characteristic may be calculated for each set range.

[0015] In the defect rate calculation step, in each set range, the higher of the defect rate of the first characteristic and the defect rate of the second characteristic may be used as the defect rate of each set range.

[0016] In the defect rate calculation step, based on the value obtained by multiplying the probability that the concentration of the first impurity in the semiconductor substrate and the irradiation amount of the charged particle beam are in each set range by the defect rate in each set range, the average defect rate of the semiconductor device may be calculated.

[0017] The analysis method may include a design step of determining the irradiation amount of the charged particle beam based on the concentration of the first impurity in the semiconductor substrate used in the manufacture of the semiconductor device and the characteristics that the semiconductor device should have. The analysis method may include an average defect rate calculation step of calculating the average defect rate of the semiconductor device based on the irradiation amount of the charged particle beam determined in the design step.

[0018] The first impurity may be carbon. The charged particle beam may be a helium ion.

[0019] In a second embodiment of the present invention, a program is provided for causing a computer to execute the analysis method according to the first embodiment.

[0020] A third embodiment of the present invention provides a method for manufacturing a semiconductor device. The manufacturing method may include a design step in which the amount of charged particle beam irradiation to a semiconductor substrate used for manufacturing the semiconductor device is determined based on the defect rate in each set range calculated by the analysis method according to the first embodiment. The manufacturing method may also include a manufacturing step in which the semiconductor substrate is irradiated with the charged particle beam of the irradiation amount determined in the design step.

[0021] It should be noted that the above summary of the invention does not enumerate all the necessary features of the present invention. Furthermore, subcombinations of these features may also constitute an invention. [Brief explanation of the drawing]

[0022] [Figure 1] This is a cross-sectional view illustrating the overview of the semiconductor device 100. [Figure 2] This figure shows an example of the relationship between the carbon concentration of the semiconductor substrate 10 and the saturation voltage Vce when it is on. [Figure 3] This diagram shows the relationships between the characteristics of the semiconductor device 100. [Figure 4] This figure illustrates an example of a measurement step S1002, a measurement distribution generation step S1004, and a shape information acquisition step S1006 in an analysis method according to one embodiment of the present invention. [Figure 5] This figure shows an example of the virtual distribution generation stage S1018 and the defect rate calculation stage S1022 in the analysis method. [Figure 6] This is a magnified view of region 210 in Figure 5. [Figure 7] This figure shows an example of the distribution of the group representative value 220 for the first characteristic (Eoff) in each measurement group. [Figure 8] This is a flowchart showing one embodiment of the analysis method. [Figure 9]This figure illustrates an example of the measurement group distribution generation stage S1008. [Figure 10] This diagram illustrates the relationship information generation stage S1010. [Figure 11] This diagram illustrates the group distribution acquisition stage S1014. [Figure 12] This diagram illustrates the virtual group generation stage S1016 and the virtual distribution generation stage S1018. [Figure 13] This figure shows an example of the defect rate generated in the defect rate calculation stage S1022. [Figure 14] This figure shows an example of the defect rate generated in the defect rate calculation stage S1022. [Figure 15] This figure shows an example of the defect rate generated in the defect rate calculation stage S1022. [Figure 16] This diagram shows the probability Rij of occurrence for each set range. [Figure 17] This figure shows other examples of analysis methods. [Figure 18] This figure shows an example of a manufacturing method for semiconductor device 100. [Figure 19] This figure shows the relationship between the number of semiconductor devices 100 manufactured and the trend in the defect rate. [Figure 20] An example of a computer 2200 in which multiple aspects of the analysis method may be embodied, either entirely or partially, is shown. [Modes for carrying out the invention]

[0023] The present invention will be described below through embodiments, but these embodiments are not intended to limit the scope of the claims. Furthermore, not all combinations of features described in the embodiments are necessarily essential to the solution of the invention.

[0024] In this specification, one side of a semiconductor substrate parallel to its depth direction is referred to as "top," and the other side as "bottom." Of the two main surfaces of a substrate, layer, or other component, one surface is referred to as the top surface, and the other surface as the bottom surface. The directions of "top" and "bottom" are not limited to the direction of gravity or the direction in which the semiconductor device is mounted.

[0025] In this specification, when we refer to items as "identical" or "equal," we may include items that have errors due to manufacturing variations, etc. Such errors are, for example, within 10%.

[0026] In this specification, the conductivity type of a doped region containing impurities is described as P-type or N-type. N-type and P-type are examples of first and second conductivity types. N-type may be the first conductivity type and P-type the second conductivity type, or P-type may be the first conductivity type and N-type the second conductivity type. In this specification, impurities may particularly mean either an N-type donor or a P-type acceptor, and may be referred to as dopants. In this specification, doping means introducing a donor or acceptor into a semiconductor substrate to make it a semiconductor exhibiting an N-type conductivity type or a P-type conductivity type.

[0027] In this specification, when P+ type or N+ type is used, it means that the doping concentration is higher than that of P type or N type, and when P- type or N- type is used, it means that the doping concentration is lower than that of P type or N type.

[0028] Figure 1 is a cross-sectional view illustrating the overview of the semiconductor device 100. The semiconductor device 100 includes at least one of a transistor section 70 and a diode section 80. The transistor section 70 is, for example, an IGBT, but is not limited thereto. The diode section 80 is, for example, a freewheeling diode (FWD) connected in antiparallel to the transistor section 70, but is not limited thereto. The semiconductor device 100 in this example is an RC-IGBT (Reverse Conducting-IGBT) including the transistor section 70 and the diode section 80.

[0029] The semiconductor device 100 in this example has, in the cross-section, a semiconductor substrate 10, an interlayer insulating film 38, an emitter electrode 52, and a collector electrode 24. The semiconductor substrate 10 is a substrate formed of a semiconductor material. As an example, the semiconductor substrate 10 is a silicon substrate. The semiconductor substrate 10 may be a substrate cut from an ingot manufactured by one of the following methods: the Czochralski method (CZ method), the magnetic field applied Czochralski method (MCZ method), or the float zone method (FZ method).

[0030] The semiconductor substrate 10 may contain carbon throughout. Carbon is an impurity that is added intentionally or unintentionally, for example, during the manufacturing of the ingot. The carbon concentration in the semiconductor substrate 10 is, for example, 0 atoms / cm³. 3 The above is 0.6 × 10 16 atoms / cm 3 The following are some examples, but are not limited to these.

[0031] The semiconductor substrate 10 has an upper surface 21 and a lower surface 23. The upper surface 21 and the lower surface 23 are two main surfaces arranged opposite each other. An interlayer insulating film 38 is provided on the upper surface 21 of the semiconductor substrate 10. The interlayer insulating film 38 is a film that includes at least one layer of insulating film such as silicate glass with impurities such as boron or phosphorus added, a thermal oxide film, a nitride film, and other insulating films. The interlayer insulating film 38 is provided with a contact hole 54 that connects the emitter electrode 52 and the semiconductor substrate 10.

[0032] The emitter electrode 52 is provided above the interlayer insulating film 38. The emitter electrode 52 is in contact with the upper surface 21 of the semiconductor substrate 10 through a contact hole 54 in the interlayer insulating film 38. The emitter electrode 52 may also be in contact with the emitter region 12 and the base region 14, which will be described later. The collector electrode 24 is provided on the lower surface 23 of the semiconductor substrate 10. The emitter electrode 52 and the collector electrode 24 are made of a metallic material such as aluminum. In this specification, the direction connecting the emitter electrode 52 and the collector electrode 24 (Z-axis direction) is referred to as the depth direction.

[0033] The semiconductor substrate 10 has an N-type drift region 18. The drift region 18 is provided in both the transistor section 70 and the diode section 80.

[0034] One or more gate trenches 40 and one or more dummy trenches 30 are provided on the upper surface 21 side of the semiconductor substrate 10. The gate trenches 40 function as gate electrodes when a gate voltage is applied, while the dummy trenches 30 do not function as gate electrodes when no gate voltage is applied. In this specification, the gate trenches 40 and dummy trenches 30 may be referred to as trenches. The trenches are provided in the depth direction from the upper surface 21 of the semiconductor substrate 10 to the drift region 18.

[0035] Each of the transistor section 70 and the diode section 80 has one or more trench sections arranged at predetermined intervals in a predetermined arrangement direction. In this example, the transistor section 70 has one or more gate trench sections 40 and one or more dummy trench sections 30 alternately provided along the arrangement direction. In this example, the diode section 80 has multiple dummy trench sections 30 provided along the arrangement direction. In this example, the diode section 80 does not have gate trench sections 40.

[0036] A mesa is provided between each trench. A mesa refers to the region within the semiconductor substrate 10 that is sandwiched between the trenches. For example, the upper end of a mesa is the upper surface 21 of the semiconductor substrate 10. The depth of the lower end of a mesa is the same as the depth of the lower end of a trench. In this example, a mesa 60 is provided in the transistor section 70, and a mesa 61 is provided in the diode section 80. In this specification, when simply referred to as a mesa, it refers to mesa 60 and mesa 61, respectively.

[0037] The mesa portion 60 of the transistor portion 70 has an N+ type emitter region 12 and a P- type base region 14, which are provided in order from the upper surface 21 side of the semiconductor substrate 10. A drift region 18 is provided below the base region 14. A storage region 16 may also be provided in the mesa portion 60. The storage region 16 is located between the base region 14 and the drift region 18.

[0038] The emitter region 12 is exposed on the upper surface 21 of the semiconductor substrate 10 and is provided in contact with the gate trench portion 40. The emitter region 12 may be in contact with the trench portions on both sides of the mesa portion 60. The doping concentration of the emitter region 12 is higher than that of the drift region 18.

[0039] The base region 14 is located below the emitter region 12. In this example, the base region 14 is located in contact with the emitter region 12. The base region 14 may be in contact with the trenches on both sides of the mesa region 60.

[0040] The storage region 16 is located below the base region 14. The storage region 16 is an N+ type region with a higher doping concentration than the drift region 18. By providing a high-concentration storage region 16 between the drift region 18 and the base region 14, the carrier injection promotion effect (IE effect) can be enhanced, and the on-voltage can be reduced. The storage region 16 may be provided so as to cover the entire lower surface of the base region 14 in each mesa portion 60.

[0041] A P-type base region 14 is provided in the mesa portion 61 of the diode portion 80, in contact with the upper surface 21 of the semiconductor substrate 10. A drift region 18 is provided below the base region 14. A storage region 16 may also be provided below the base region 14 in the mesa portion 61.

[0042] In both the transistor section 70 and the diode section 80, an N+ type buffer section 20 may be provided on the lower surface 23 side of the drift section 18. The doping concentration of the buffer section 20 is higher than that of the drift section 18. The buffer section 20 may function as a field stop layer that prevents the depletion layer extending from the lower end of the base section 14 from reaching the P+ type collector section 22 and the N+ type cathode section 82.

[0043] In the transistor section 70, a P+ type collector region 22 is provided below the buffer region 20. The acceptor concentration of the collector region 22 is higher than that of the base region 14.

[0044] In the diode section 80, an N+ type cathode region 82 is provided below the buffer region 20. The donor concentration of the cathode region 82 is higher than that of the drift region 18. The collector region 22 and the cathode region 82 are exposed to the lower surface 23 of the semiconductor substrate 10 and are connected to the collector electrode 24. The collector electrode 24 may be in contact with the entire lower surface 23 of the semiconductor substrate 10. The emitter electrode 52 and the collector electrode 24 are made of a metallic material such as aluminum.

[0045] Each trench extends from the upper surface 21 of the semiconductor substrate 10, through the base region 14, and reaches the drift region 18. In regions where at least one of the emitter region 12 and the storage region 16 is provided, each trench also penetrates these doping regions and reaches the drift region 18.

[0046] The gate trench portion 40 has a groove-shaped gate trench provided on the upper surface 21 of the semiconductor substrate 10, a gate insulating film 42, and a gate conductive portion 44. The gate insulating film 42 is provided covering the inner wall of the gate trench. The gate insulating film 42 may be formed by oxidizing or nitriding the semiconductor on the inner wall of the gate trench. The gate conductive portion 44 is provided inside the gate trench, on the inside of the gate insulating film 42. In other words, the gate insulating film 42 insulates the gate conductive portion 44 from the semiconductor substrate 10. The gate conductive portion 44 is formed of a conductive material such as polysilicon.

[0047] The gate conductive portion 44 may be longer than the base region 14 in the depth direction. The gate trench portion 40 in this cross-section is covered by an interlayer insulating film 38 on the upper surface 21 of the semiconductor substrate 10. The gate conductive portion 44 is electrically connected to the gate wiring. When a predetermined gate voltage is applied to the gate conductive portion 44, a channel formed by an electron inversion layer is formed on the surface layer of the interface of the base region 14 that is in contact with the gate trench portion 40.

[0048] The dummy trench portion 30 may have the same structure as the gate trench portion 40 in that cross-section. The dummy trench portion 30 has a dummy trench, a dummy insulating film 32, and a dummy conductive portion 34 provided on the upper surface 21 of the semiconductor substrate 10. The dummy conductive portion 34 may be electrically connected to the emitter electrode 52.

[0049] In this example, the semiconductor substrate 10 has recombination centers 19 that adjust the carrier lifetime. In Figure 1, the recombination centers 19 are schematically shown with cross marks. The recombination centers 19 are, for example, vacancy defects. The recombination centers 19 are formed by irradiating the semiconductor substrate 10 with charged particles such as helium ions, protons, and electrons.

[0050] The recombination centers 19 may be formed locally in the depth direction. In this case, the recombination centers 19 can be formed by injecting charged particles such as helium ions into a predetermined depth in the semiconductor substrate 10. Concentration of the recombination centers 19 ( / cm³ 3) is the irradiation dose of charged particle beams ( / cm²). 2 It can be controlled by [this method]. The recombination center 19 may be formed on the upper surface 21 side of the semiconductor substrate 10, as shown in Figure 1, or it may be formed on the lower surface 23 side. The recombination center 19 may also be formed throughout the entire depth of the semiconductor substrate 10. In this case, charged particles such as electrons are irradiated so as to penetrate the semiconductor substrate 10.

[0051] Furthermore, the recombination center 19 may be formed locally in a plane parallel to the upper surface 21. In this case, charged particles such as helium ions are selectively irradiated into that plane using a mask or the like. For example, the recombination center 19 may be selectively formed in the region of the diode section 80. The recombination center 19 may be formed in the region of the transistor section 70. Also, the recombination center 19 may be formed in both the diode section 80 and the transistor section 70.

[0052] The characteristics of the semiconductor device 100 can be adjusted by forming a recombination center 19. For example, the carrier lifetime can be adjusted by forming a recombination center 19. Consequently, the switching time during turn-off of the semiconductor device 100 can be adjusted, and the losses during turn-off can be adjusted. In addition, forming a recombination center 19 changes the saturation voltage Vce_sat (sometimes simply referred to as Vce in this specification) when the semiconductor device 100 is in the ON state, and changes the ON loss. These characteristics are affected by the concentration ( / cm³) of the recombination center 19. 3 These characteristics vary depending on the irradiation dose ( / cm²) of the charged particle beam irradiated onto the semiconductor substrate 10. 2 It also varies depending on the carbon concentration ( / cm³) of the semiconductor substrate 10. 3 Since the ease with which recombination centers 19 are formed changes, the characteristics of the semiconductor device 100 also vary depending on the carbon concentration of the semiconductor substrate 10.

[0053] In some cases, the determination of whether a semiconductor device 100 is good or defective is made based on the above-mentioned characteristics of the semiconductor device 100. As described above, the characteristics of the semiconductor device 100 change depending on the carbon concentration of the semiconductor substrate 10 and the amount of charged particle beam irradiation. In the design and manufacture of the semiconductor device 100, it is preferable to be able to accurately estimate the defect rate of the semiconductor device 100 for a given combination of carbon concentration of the semiconductor substrate 10 and the amount of charged particle beam irradiation.

[0054] Figure 2 shows an example of the relationship between the carbon concentration of the semiconductor substrate 10 and the saturation voltage Vce when it is on. Each plot in Figure 2 corresponds to one lot of semiconductor device 100. Each plot shows the relationship between the average carbon concentration of multiple semiconductor devices 100 included in that lot and the average saturation voltage Vce.

[0055] As shown in Figure 2, even if the carbon concentration of the semiconductor substrate 10 is the same, the saturation voltage Vce varies between lots. Each plot in Figure 2 assumes that the design parameters other than carbon concentration are the same. For example, the concentration of impurities injected into the semiconductor substrate 10 and the set value of the charged particle beam irradiation amount are the same. However, due to manufacturing variations, etc., even if the set values ​​of these parameters are the same, variations occur in characteristics such as the saturation voltage Vce. Note that while Figure 2 shows the saturation voltage Vce, the same applies to other characteristics of the semiconductor device 100, such as the carrier lifetime. For this reason, even if the distribution shown in Figure 2 is approximated by a straight line using linear regression analysis, it is difficult to estimate what the defect rate of the semiconductor device 100 will be for a given combination of carbon concentration and charged particle beam irradiation amount.

[0056] Figure 3 shows the relationships between the characteristics of the semiconductor device 100. Figure 3 shows the relationship between the turn-off loss Eoff and saturation voltage Vce of the transistor section 70 (IGBT) at 25°C (in this specification and drawings, 25°C may be referred to as room temperature or RT) and 175°C, as well as the relationship between the forward voltage Vf and reverse recovery loss Err of the diode section 80 (Diode) at 25°C and 175°C. In each distribution diagram, one plot corresponds to one lot of the semiconductor device 100. Furthermore, the settings for manufacturing parameters such as carbon concentration and charged particle beam irradiation dose are the same in each plot.

[0057] As shown in Figure 2, even with the same carbon concentration and charged particle beam irradiation dose, variations occur in the respective characteristic values. Therefore, as shown in Figure 3, each plot is distributed two-dimensionally. For these plots, for example, items inside range 200 may be classified as good products, and items outside of it as defective products. As shown in Figure 3, when the shape of range 200 is a polygon such as a pentagon or hexagon, it becomes difficult to accurately estimate the defect rate of the semiconductor device 100 using the linear regression analysis described above.

[0058] Figure 4 illustrates an example of a measurement step S1002, a measurement distribution generation step S1004, and a shape information acquisition step S1006 in an analysis method according to one embodiment of the present invention. The analysis method in this example analyzes the defect rate of a semiconductor device 100 in which the semiconductor substrate 10 contains a first impurity and the semiconductor substrate 10 is irradiated with a charged particle beam. The first impurity is, for example, carbon, oxygen, or nitrogen, but is not limited thereto. The charged particle beam is, for example, helium ions, hydrogen ions, or electron beams, but is not limited thereto. In this example, the first impurity is carbon, and the charged particle beam is helium ions.

[0059] In measurement step S1002, measured values ​​of the first and second characteristics of multiple semiconductor devices 100 are obtained. In measurement step S1002, the characteristics of the semiconductor devices 100 may be measured, or already measured values ​​may be obtained. Preferably, the first and second characteristics are correlated with each other. The first and second characteristics may be characteristics that are trade-offs with each other. The first and second characteristics may be any two of the following, as described in Figure 3: the turn-off loss Eoff of the transistor section 70 at 25°C or 175°C, the saturation voltage Vce of the transistor section 70 at 25°C or 175°C, the forward voltage Vf of the diode section 80 at 25°C or 175°C, and the reverse recovery loss Err of the diode section 80 at 25°C or 175°C. In the example in Figure 4, the first and second characteristics are the saturation voltage Vce_RT at room temperature and the turn-off loss Eoff_RT at room temperature. In this specification and drawings, when each characteristic is described without specifying temperature, such as saturation voltage and turn-off loss, it refers to the characteristic at room temperature. Room temperature can be any temperature between 15°C and 30°C, for example, 27°C.

[0060] In measurement step S1002, measured values ​​of the characteristics of multiple semiconductor devices are obtained, where the carbon concentration per unit volume of the semiconductor substrate 10 and the helium ion irradiation dose per unit area of ​​the semiconductor substrate 10 fall within a predetermined set range. The carbon concentration and helium ion irradiation dose are examples of manufacturing condition variables. The carbon concentration of each semiconductor device 100 used in measurement step S1002 may be the measured carbon concentration of the actual semiconductor substrate 10, or it may be the specified carbon concentration value set by the manufacturer of the semiconductor substrate 10. Furthermore, the helium ion irradiation dose used in measurement step S1002 may be the set value of the helium ion irradiation dose actually irradiated onto the semiconductor substrate 10.

[0061] In the measurement stage S1002, measurement values of a plurality of semiconductor devices 100 with equivalent carbon concentration and helium irradiation amount as manufacturing condition variables are acquired. In the measurement stage S1002, the width of the setting range is determined such that the carbon concentration and helium irradiation amount of the plurality of semiconductor devices 100 included in one measurement group are equivalent. In the analysis method, as shown in FIG. 13 etc., a defective rate is calculated for one or a plurality of setting ranges. Each setting range may be set such that the range of the carbon concentration and the range of the helium irradiation amount do not overlap with other setting ranges. More specifically, in three adjacent setting ranges, if the central value of the range of the carbon concentration increases by 0.05×10 16 / cm 3 、0.1×10 16 / cm 3 、0.15×10 16 / cm 3 like this, the magnitude of the range of the carbon concentration in each setting range may also be 0.05×10 16 / cm 3 each. The range of the carbon concentration in each setting range in this example is 0.025×10 16 / cm 3 ~0.075×10 16 / cm 3 、0.075×10 16 / cm 3 ~0.125×10 16 / cm 3 、0.125×10 16 / cm 3 ~0.175×10 16 / cm 3 ~0.175×10 16 / cm 3 may be.

[0062] In the measurement distribution generation step S1004, a measurement distribution 208 is generated that shows the distribution of measured values ​​of the first and second characteristics (saturation voltage Vce and turn-off loss Eoff in this example) for one measurement group. In the measurement distribution generation step S1004, a scatter plot with the first and second characteristics as two axes may be generated, as shown in Figure 4. One plot 202 in Figure 4 represents one semiconductor device 100. In the measurement distribution generation step S1004, a distribution 204 for the first characteristic (Vce) and a distribution 206 for the second characteristic (Eoff) may be generated, as shown in Figure 4. Distributions 204 and 206 are distributions in which the histogram is approximated by a Gaussian distribution. The value of the first characteristic that shows a peak in distribution 204 is denoted as Vc, and the value of the second characteristic that shows a peak in distribution 206 is denoted as Ec.

[0063] In the shape information acquisition stage S1006, shape information showing the shape of the measurement distribution 208 is acquired. In the shape information acquisition stage S1006, shape information showing the covariance of the first and second characteristics in the measurement distribution 208 may be generated. In the shape information acquisition stage S1006, shape information showing the major axis vector and minor axis vector of the distribution of the measurement distribution 208 may be generated. In this example, the shape information acquisition stage S1006 generates a variance-covariance matrix of the first and second characteristics. The variance-covariance matrix is ​​represented by equation 1. σ1 is the standard deviation of the distribution 204 of the first characteristic, σ2 is the standard deviation of the distribution 206 of the second characteristic, σ 12 This is the covariance of the first and second characteristics.

number

[0064] Figure 5 shows an example of the virtual distribution generation stage S1018 and the defect rate calculation stage S1022 in the analysis method. In the virtual distribution generation stage S1018, the first and second characteristics of multiple virtual semiconductor devices 100 are simulated based on the measurement distribution 208 to generate a virtual distribution 308 in which the samples of the first and second characteristics are distributed over a wider range than in the measurement distribution 208.

[0065] In the virtual distribution generation stage S1018, the reference point 301 on a plane with the first and second characteristics as two axes may be determined using representative values ​​from distributions 204 and 206. The representative values ​​are, for example, the values ​​Vc and Ec that indicate the peaks of distributions 204 and 206, but are not limited to these. The representative values ​​should be such that the position of distribution 204 on the axis of the first characteristic and the position of distribution 206 on the axis of the second characteristic can be identified.

[0066] In the virtual distribution generation stage S1018, a distribution corresponding to the shape of the measured distribution 208 is applied to the reference point 301 to generate a virtual distribution 308 in which the sample is distributed over a wider range than the measured distribution 208. For example, the distribution 304 for the first characteristic (Eoff) and the distribution 306 for the second characteristic (Vce) in the virtual distribution 308 may be distributions that extend the tail portions of distributions 204 and 206 further outwards.

[0067] In the virtual distribution generation stage S1018, information indicating the covariance of the measured distribution 208 (e.g., the variance-covariance matrix) is taken into consideration, and each sample point that follows a bivariate normal distribution is generated by simulation. As an example, in the virtual distribution generation stage S1018, a Monte Carlo simulation is performed based on a virtual distribution formed from parameters such as variance, covariance, and mean. In the virtual distribution generation stage S1018, a virtual distribution 308 with a larger number of samples than the measured distribution 208 may be generated. This makes it easier for the virtual distribution 308 to distribute samples over a wider range than the measured distribution 208.

[0068] In the defect rate calculation stage S1022, it is determined whether each sample point included in the virtual distribution 308 is within a predetermined range 200. Then, the sample points outside the range 200 are treated as defective samples, and the defect rate is calculated. In Figure 5, the virtual distribution 308 is schematically shown as an ellipse, but the virtual distribution 308 is a collection of many sample points.

[0069] Figure 6 is an enlarged view of region 210 in Figure 5. In Figure 6, sample points 302 of the virtual distribution 308 are shown as black circles. In the defect rate calculation stage S1022, sample points 302 that are outside the range 200 are determined to be defective samples. In this example, a virtual distribution 308 with a wider range than the measured distribution 208 is generated and the defect rate is calculated. Therefore, the defect rate can be calculated accurately even from the measured distribution 208, which has a relatively small number of samples.

[0070] Figures 4 to 6 illustrate the method for calculating the defect rate within a single set range. In the analysis method, the defect rate may be calculated for multiple set ranges. In this case, the shape of the measurement distribution 208 (e.g., the variance-covariance matrix) in one set range may be applied to the virtual distribution 308 of another set range. Alternatively, the analysis method may use the measured values ​​of multiple measurement groups to calculate the defect rate for each set range.

[0071] Figure 7 shows an example of the distribution of the group representative value 220 of the first characteristic (Eoff) in each measurement group. As described above, one measurement group includes multiple semiconductor devices 100 in which the carbon concentration and the irradiation amount of charged particles are within a predetermined set range. Note that there may be multiple measurement groups even for the same set range. Measurement groups may be grouped by manufacturing date, manufacturing lot, or other indicators. In Figure 7 and subsequent figures, the processing content may be explained using the group representative value 220 of the first characteristic, but the same processing as described in Figure 7, etc., is performed on the group representative value of the second characteristic (Vce).

[0072] The group representative value 220 may be the average value of the first characteristic of each semiconductor device 100 within the measurement group. However, the group representative value 220 is not limited to the average value. The group representative value 220 should be able to define the position of the measurement distribution 208 of the measurement group on the axis of the first characteristic. For example, the group representative value 220 may be the value Ec of the first characteristic at which the distribution 204 shows a peak.

[0073] As shown in Figure 7, the group representative value 220 does not exist for all combinations of carbon concentration and helium irradiation dose as manufacturing condition variables, but may exist for specific combinations of carbon concentration and helium irradiation dose. In other words, when a plane with carbon concentration and helium irradiation dose as two axes is divided into multiple setting ranges, there are setting ranges in which the group representative value 220 exists and setting ranges in which the group representative value 220 does not exist. The measurement group may be a collection of semiconductor devices 100 manufactured in the past. In this case, since there is no measurement group for combinations of carbon concentration and helium irradiation dose that were not used in manufacturing, the group representative value 220 is not obtained. In the analysis method of this example, the defect rate is estimated even for setting ranges in which the group representative value 220 is not obtained.

[0074] Figure 8 is a flowchart illustrating one embodiment of the analysis method. The analysis method comprises a preparation stage S1050 and a processing stage S1060. The preparation stage S1050 comprises a measurement value acquisition stage S1002, a measurement distribution generation stage S1004, a shape information acquisition stage S1006, a measurement group distribution generation stage S1008, and a relationship information generation stage S1010.

[0075] The measurement value acquisition stage S1002, the measurement distribution generation stage S1004, and the shape information acquisition stage S1006 are the same as those described in Figure 4. In this example, the measurement value acquisition stage S1002, the measurement distribution generation stage S1004, and the shape information acquisition stage S1006 are performed for at least one set range in which the group representative value 220 exists. The measurement value acquisition stage S1002, the measurement distribution generation stage S1004, and the shape information acquisition stage S1006 may also be performed for multiple or all set ranges in which the group representative value 220 exists.

[0076] Figure 9 illustrates an example of the measurement group distribution generation stage S1008. In the measurement group distribution generation stage S1008, group representative values ​​220 of the first and second characteristics are obtained for each measurement group for multiple measurement groups corresponding to the same setting range 230. For example, as shown in Figure 7, group representative values ​​220 belonging to the setting range 230, which includes combinations of carbon concentration C1 and helium irradiation amount He1, are extracted.

[0077] In the measurement group distribution generation stage S1008, a measurement group distribution 222 is generated that shows the distribution of group representative values ​​220 belonging to the same set range 230. In this example, the measurement group distribution 222 is a distribution obtained by approximating the histogram of the group representative values ​​220 with a Gaussian distribution.

[0078] In the measurement group distribution generation stage S1008, a representative value of the measurement group distribution 222 is obtained. As an example, the representative value is the value of the first characteristic (Eoff_pm) at which the measurement group distribution 222 peaks, but it is not limited to this. The representative value of the distribution should be a value that can define the position of the measurement group distribution 222 on the axis of the first characteristic. The representative value of the distribution may also be the mean value of the measurement group distribution 222.

[0079] Furthermore, in the measurement group distribution generation stage S1008, shape information indicating the shape of the measurement group distribution 222 may be acquired. The shape information is, for example, the standard deviation σm of the measurement group distribution 222, but is not limited to this. The shape information should be information that can reproduce the shape of the distribution. In the measurement group distribution generation stage S1008, the representative value of the distribution and shape information are acquired for multiple setting ranges 230 in which the group representative value 220 exists.

[0080] Figure 10 illustrates the relationship information generation stage S1010. The relationship information generation stage S1010 generates relationship information 240 that shows the relationship between values ​​within a set range (for example, the median carbon concentration and median helium irradiation dose within the set range) and a representative distribution value (Eoff_pm). The relationship information 240 may be a regression plane fitted to multiple sample points that show these relationships. These sample points are points in a three-dimensional space determined by the carbon concentration values, helium irradiation dose values, and representative distribution values ​​within the set range. The relationship information 240 shows an approximate value of the representative distribution value within an arbitrary set range.

[0081] In the relationship information generation stage S1010, relationship information may be further generated that shows the relationship between the set range values ​​(manufacturing condition variables, carbon concentration and helium irradiation amount) and the shape information (σm) of the measurement group distribution 222. This relationship information shows approximate values ​​of the shape information within an arbitrary set range.

[0082] In the relationship information generation stage S1010, relationship information may be further generated that shows the relationship between the values ​​within the set range (carbon concentration and helium irradiation dose) and information indicating the shape of the measurement distribution 208 (for example, each element of the variance-covariance matrix). This relationship information shows an approximate value of the shape of the measurement distribution 208 within an arbitrary set range.

[0083] As shown in Figure 8, the processing step S1060 includes a setting range determination step S1012, a group distribution acquisition step S1014, a virtual group generation step S1016, a virtual distribution generation step S1018, and a defect rate calculation step S1022. In the setting range determination step S1012, the setting range for which the defect rate should be calculated is selected.

[0084] Figure 11 illustrates the group distribution acquisition stage S1014. In the group distribution acquisition stage S1014, a virtual group distribution 322 is acquired within the selected setting range 320. The setting range 320 may be different from the setting range in which the measurement group distribution 222 exists, or it may be the same setting range in which the measurement group distribution 222 exists. If a group representative value 220 of a measurement group exists within the setting range 320, the measurement group distribution 222 may be used instead of the virtual group distribution 322 in the processing for that setting range 320.

[0085] In the group distribution acquisition stage S1014, an approximate value of the representative value of the distribution (Eoff_pm) in the set range 320 is obtained from the relational information 240. Similarly, an approximate value of the shape information (σm) in the set range 320 is obtained. In the group distribution acquisition stage S1014, a virtual group distribution 322 with the shape indicated by the shape information (σm) is generated at the position on the first characteristic axis indicated by the acquired representative value of the distribution. The virtual group distribution 322 may be a Gaussian distribution with a standard deviation σm. In the group distribution acquisition stage S1014, a virtual group distribution 322 is generated for both the first and second characteristics. The virtual group distribution 322 is a virtual distribution of the representative value of the group that is assumed to exist if measurement groups exist in each set range 320.

[0086] Figure 12 illustrates the virtual group generation stage S1016 and the virtual distribution generation stage S1018. In the virtual group generation stage S1016, multiple virtual groups are generated for each setting range 320 based on the virtual group distribution 322. In the virtual group generation stage S1016, multiple virtual groups may be generated for setting ranges where no measurement groups exist, or for setting ranges that are the same as measurement groups.

[0087] In the virtual group generation stage S1016, the group representative values ​​(Eoff_a, Vce_a) for each virtual group are determined according to the probability of occurrence of each value of the first and second characteristics, as shown by the virtual group distribution 322-1 for the first characteristic and the virtual group distribution 322-2 for the second characteristic. In other words, the distribution of the group representative values ​​for the multiple virtual groups generated in the virtual group generation stage S1016 is the same as that of the virtual group distribution 322.

[0088] In the virtual group generation stage S1016, the group representative values ​​for the first and second characteristics may be determined independently based on their respective virtual group distributions 322. In another example, in the virtual group generation stage S1016, the combination of group representative values ​​for the first and second characteristics may be determined based on the shape information acquired in the shape information acquisition stage S1006. For example, in the virtual group generation stage S1016, the combination of group representative values ​​for the first and second characteristics may be generated using the variance-covariance matrix acquired in the shape information acquisition stage S1006. This allows for the determination of the group representative values ​​for virtual groups while considering the correlation between the first and second characteristics and the probability of occurrence in each virtual group distribution 322. The group representative values ​​for each virtual group may be determined such that the shape of the scatter plot of the group representative values ​​of multiple virtual groups, with the first and second characteristics as axes, is similar to the shape of the measurement distribution 208.

[0089] In the virtual distribution generation stage S1018, a virtual distribution 308, as explained in Figure 5, is generated for each virtual group generated in the virtual group generation stage S1016. That is, the virtual distribution 308 is applied using the group representative values ​​(Eoff_a, Vce_a) of the virtual group as the reference point 301. In the virtual distribution generation stage S1018, a virtual distribution 308 containing a predetermined number (e.g., 5000) samples may be generated using Monte Carlo simulation. The shape of the virtual distribution 308 can be determined based on information showing the covariance of the first and second characteristics in the measured distribution 208 (e.g., a variance-covariance matrix). As described above, the information showing the covariance of the first and second characteristics may be defined for each set range. In this case, as explained in Figure 11, relational information showing the relationship between each set range and the covariance may be generated. Alternatively, common covariance information may be used for multiple set ranges. For example, common covariance information may be generated from the measurement distribution 208 in one setting range selected as representative, or common covariance information may be generated from a unified distribution obtained by integrating the measurement distributions 208 in multiple setting ranges. Among the setting ranges, the setting range with the largest number of measurement groups may be selected as the representative as described above.

[0090] In the determination stage S1020, it is determined whether the virtual group generation stage S1016 and the virtual distribution generation stage S1018 have been performed on a predetermined number of virtual groups. If the predetermined number of virtual groups has not reached the predetermined number, a new virtual group is generated in the virtual group generation stage S1016, and a virtual distribution 308 is generated in the virtual distribution generation stage S1018.

[0091] When the number of processed virtual groups reaches a predetermined number (for example, 1000 times), in the defect rate calculation stage S1022, the defect rate in the set range is calculated using the virtual distribution corresponding to each set range. In addition, in the defect rate calculation stage S1022, the defect rate in the virtual distribution 308 may be calculated each time a virtual distribution 308 is generated in the virtual distribution generation stage S1018. The defect rate calculation stage S1022 may calculate the average of the calculated defect rates and use that as the defect rate in the set range. In another example, each time a new virtual distribution 308 is generated in the virtual distribution generation stage S1018, an integrated distribution may be generated by integrating it with the virtual distribution 308 already generated for that virtual group. The defect rate calculation stage S1022 may calculate the defect rate in the integrated distribution.

[0092] In the determination stage S1024, it is determined whether the defect rate has been calculated for all setting ranges for which the defect rate should be calculated. If there are setting ranges for which the defect rate has not been calculated, a new setting range is selected in the setting range determination stage S1012, and the process from S1014 onwards is performed. If the defect rate has been calculated for all setting ranges, the process is terminated.

[0093] In this example, a virtual distribution can be generated even for setting ranges different from the setting range in which measurement groups exist. Therefore, the defect rate can be estimated accurately even for setting ranges in which no measurement groups exist. Furthermore, a virtual group distribution can be set and a virtual distribution generated for setting ranges identical to the setting range of measurement groups. This allows for an increased sample size and more accurate estimation of the defect rate even for those setting ranges.

[0094] Figure 13 shows an example of the defect rate generated in defect rate calculation step S1022. In defect rate calculation step S1022, the defect rate Xij is calculated for each set range determined by the combination of carbon concentration and helium irradiation amount, which are manufacturing condition variables. i represents the range of carbon concentration, and j represents the range of helium irradiation amount. In defect rate calculation step S1022, a matrix-shaped table as shown in Figure 13 may be generated. Each cell in Figure 13 corresponds to a set range. Note that Figure 13 shows the defect rate based on the first characteristic (Eoff), but in defect rate calculation step S1022, the defect rate for both the first characteristic and the second characteristic is calculated for each set range.

[0095] Figure 14 shows an example of the defect rate generated in the defect rate calculation stage S1022. Figure 14 shows the defect rate Yij for the second characteristic in each set range.

[0096] Figure 15 shows an example of the defect rate generated in the defect rate calculation step S1022. In the defect rate calculation step S1022 of this example, a table of integrated defect rates is generated by combining the table of defect rates for the first characteristic and the table of defect rates for the second characteristic. In this example as well, the integrated defect rate Zij is calculated for each set range. In this example, for each set range, the higher of the defect rate Xij of the first characteristic and the defect rate Yij of the second characteristic (i.e., the one with a higher probability of being judged as defective) may be used as the integrated defect rate Zij for each set range.

[0097] Figure 16 shows the probability Rij of each set range occurring. The probability Rij of each set range represents the probability that the carbon concentration in the semiconductor substrate 10 used in the manufacture of the semiconductor device 100, and the helium ion irradiation dose in the manufacturing process of the semiconductor device 100, fall within their respective set ranges. Note that the error in the helium ion irradiation dose in the manufacturing process is small. Therefore, the probability of the set range occurring, excluding the set value of the helium ion irradiation dose, may be set to 0. The probability distribution of the carbon concentration in the semiconductor substrate 10 may be supplied by the manufacturer of the semiconductor substrate 10, or it may be generated from past measurement values.

[0098] In the defect rate calculation step S1022, a value obtained by multiplying the defect rate (e.g., Zij) by the probability of occurrence Rij may be calculated for each set range. Alternatively, in the defect rate calculation step S1022, the average defect rate of the semiconductor device 100, when the semiconductor device 100 is manufactured using the semiconductor substrate 10 and with the helium ion irradiation dose, may be calculated by accumulating the values ​​obtained by multiplying the defect rate and the probability of occurrence.

[0099] Figure 17 shows another example of the analysis method. The analysis method comprises a failure rate analysis stage S1102, a semiconductor device design stage S1104, and an average failure rate calculation stage S1106. The processing in the failure rate analysis stage S1102 is the same as the analysis method described in Figures 1 to 16. In other words, in the failure rate analysis stage S1102, the failure rate is calculated for each set range.

[0100] In the semiconductor device design stage S1104, the amount of helium ion irradiation is determined based on the carbon concentration in the semiconductor substrate 10 used to manufacture the semiconductor device 100 and the characteristics that the semiconductor device 100 should have. For example, in the semiconductor device design stage S1104, the amount of helium ion irradiation is determined according to the value of the carrier lifetime that the semiconductor device 100 should have. The carbon concentration of the semiconductor substrate 10 may be information supplied by the manufacturer of the semiconductor substrate 10, or it may be obtained by measuring the semiconductor substrate 10.

[0101] In the average failure rate calculation step S1106, the average failure rate of the semiconductor device 100 is calculated based on the helium ion irradiation dose determined in the semiconductor device design step S1104. For example, in the average failure rate calculation step S1106, as explained in Figure 16, the probability of occurrence in the set range that does not include the determined helium ion irradiation dose may be set to 0, and the average failure rate may be calculated. By performing such a process, it is possible to estimate the failure rate of the semiconductor device 100 after changing the helium ion irradiation dose.

[0102] The analysis method may also include a design step in which the permissible range of helium ion irradiation is determined based on the carbon concentration in the semiconductor substrate 10 used in the manufacture of the semiconductor device 100 and the defect rate in each set range. In this case, a value that is permissible as the average defect rate of the semiconductor device 100 may be set. In the design step, a permissible range of setting values ​​for the helium ion irradiation amount may be presented such that, when the helium ion irradiation amount is set to a certain value, the average defect rate explained in Figure 17 is below the permissible value.

[0103] Figure 18 shows an example of a manufacturing method for semiconductor device 100. The manufacturing method comprises a defect rate analysis stage S1202, a design stage S1204, and a manufacturing stage S1206. The defect rate analysis stage S1202 is the same as the analysis method described in Figures 1 to 17.

[0104] In design stage S1204, the amount of helium ions irradiated onto the semiconductor substrate 10 used in the manufacture of the semiconductor device 100 is determined based on the defect rate in each set range. Design stage S1204 may be the same as the design stage described in Figure 17.

[0105] In the manufacturing stage S1206, the semiconductor substrate 10 is irradiated with the amount of helium ions determined in the design stage S1204. Also in the manufacturing stage S1206, each component of the semiconductor device 100, as explained in Figure 1, is formed. By this method, the defect rate can be accurately estimated, and the semiconductor device 100 can be designed and manufactured.

[0106] Figure 19 shows the relationship between the number of semiconductor devices 100 manufactured and the change in the defect rate. In this example, when a defect occurs in a manufactured semiconductor device 100, this information is fed back to the manufacturing process to suppress the occurrence of defects. In this case, when the number of semiconductor devices 100 manufactured exceeds a certain level, the accuracy of the feedback improves and the defect rate gradually decreases.

[0107] In contrast, by using the analysis method described in the embodiment, it is possible to generate numerous characteristics of a virtual semiconductor device 100 and analyze the defect rate. This makes it easier to appropriately set parameters such as the helium ion irradiation dose, and accelerates the suppression of defect occurrence.

[0108] Figure 20 shows an example of a computer 2200 in which multiple embodiments of the analysis method may be fully or partially implemented. The computer 2200 is equipped with a program that allows it to execute the analysis method described in Figures 1 to 19.

[0109] A program installed on computer 2200 can cause computer 2200 to function as an operation or one or more sections of an apparatus according to an embodiment of the present invention, or to execute such operation or one or more sections, and / or to cause computer 2200 to execute a method or a step of such method according to an embodiment of the present invention. Such a program may be executed by CPU 2212 to cause computer 2200 to perform a particular operation associated with some or all of the blocks of the flowcharts and block diagrams described herein.

[0110] The computer 2200 according to this embodiment includes a CPU 2212, RAM 2214, a graphics controller 2216, and a display device 2218, which are interconnected by a host controller 2210. The computer 2200 also includes input / output units such as a communication interface 2222, a hard disk drive 2224, a DVD-ROM drive 2226, and an IC card drive, which are connected to the host controller 2210 via an input / output controller 2220. The computer also includes legacy input / output units such as a ROM 2230 and a keyboard 2242, which are connected to the input / output controller 2220 via an input / output chip 2240.

[0111] The CPU 2212 operates according to programs stored in the ROM 2230 and RAM 2214, thereby controlling each unit. The graphics controller 2216 retrieves image data generated by the CPU 2212 from a frame buffer provided in RAM 2214 or from itself, and displays the image data on the display device 2218.

[0112] The communication interface 2222 communicates with other electronic devices via a network. The hard disk drive 2224 stores programs and data used by the CPU 2212 in the computer 2200. The DVD-ROM drive 2226 reads programs or data from the DVD-ROM 2201 and provides them to the hard disk drive 2224 via the RAM 2214. The IC card drive reads programs and data from the IC card and / or writes programs and data to the IC card.

[0113] The ROM 2230 stores boot programs and / or programs that depend on the computer 2200's hardware, which are executed by the computer 2200 when activated. The input / output chip 2240 may also connect various input / output units to the input / output controller 2220 via parallel ports, serial ports, keyboard ports, mouse ports, etc.

[0114] The program is provided on a computer-readable medium such as a DVD-ROM 2201 or an IC card. The program is read from the computer-readable medium and installed on a hard disk drive 2224, RAM 2214, or ROM 2230, which are also examples of computer-readable medium, and executed by the CPU 2212. The information processing described within these programs is read by the computer 2200, resulting in coordination between the program and the various types of hardware resources described above. The apparatus or method may be configured to realize the manipulation or processing of information in accordance with the use of the computer 2200.

[0115] For example, when communication is performed between a computer 2200 and an external device, the CPU 2212 may execute a communication program loaded into RAM 2214 and, based on the processing described in the communication program, instruct the communication interface 2222 to perform communication processing. Under the control of the CPU 2212, the communication interface 2222 reads transmission data stored in a transmission buffer processing area provided in a recording medium such as RAM 2214, a hard disk drive 2224, a DVD-ROM 2201, or an IC card, transmits the read transmission data to the network, or writes received data received from the network to a reception buffer processing area provided on the recording medium.

[0116] Furthermore, the CPU 2212 may read all or necessary parts of files or databases stored on external storage media such as the hard disk drive 2224, DVD-ROM drive 2226 (DVD-ROM 2201), or IC card into the RAM 2214, and perform various types of processing on the data in the RAM 2214. The CPU 2212 then writes the processed data back to the external storage media.

[0117] Various types of information, such as various types of programs, data, tables, and databases, may be stored on the recording medium and subjected to information processing. The CPU 2212 may perform various types of processing on the data read from RAM 2214, including various types of operations, information processing, conditional judgments, conditional branching, unconditional branching, information retrieval / replacement, etc., as described throughout this disclosure and specified by the program instruction sequence, and write the results back to RAM 2214. The CPU 2212 may also retrieve information in files, databases, etc., within the recording medium. For example, if multiple entries are stored in the recording medium, each having an attribute value of a first attribute associated with an attribute value of a second attribute, the CPU 2212 may search among the multiple entries for an entry that matches the condition for which the attribute value of the first attribute is specified, read the attribute value of the second attribute stored in that entry, and thereby obtain the attribute value of the second attribute associated with the first attribute that satisfies a predetermined condition.

[0118] The programs or software modules described above may be stored on or near computer 2200 on a computer-readable medium. Alternatively, recording media such as hard disks or RAM provided within a server system connected to a dedicated communication network or the Internet can be used as computer-readable media, thereby providing programs to computer 2200 via the network.

[0119] Although the present invention has been described above using embodiments, the technical scope of the present invention is not limited to the scope described in the above embodiments. It will be apparent to those skilled in the art that various modifications or improvements can be made to the above embodiments. It will be clear from the claims that such modified or improved forms may also be included in the technical scope of the present invention.

[0120] It should be noted that the execution order of operations, procedures, steps, and stages in the apparatus, systems, programs, and methods shown in the claims, specifications, and drawings is not explicitly stated as "before," "prior to," etc., and that these can be implemented in any order unless the output of a previous process is used in a later process. Even if the operation flow in the claims, specifications, and drawings is described using phrases such as "first," "next," etc. for convenience, it does not mean that it is essential to perform the operations in that order. [Explanation of symbols]

[0121] 10...Semiconductor substrate, 12...Emitter region, 14...Base region, 16...Storage region, 18...Drift region, 19...Recombination center, 20...Buffer region, 21...Top surface, 22...Collector region, 23...Bottom surface, 24...Collector electrode, 30...Dummy trench section, 32...Dummy insulating film, 34...Dummy conductive section, 38...Interlayer insulating film, 40...Gate trench section, 42...Gate insulating film, 44...Gate conductive section, 52...Emitter electrode, 54...Contact hole, 60, 61... • Mesa section, 70... Transistor section, 80... Diode section, 82... Cathode region, 100... Semiconductor device, 200... Range, 202... Plot, 204... Distribution, 206... Distribution, 208... Measured distribution, 210... Region, 220... Group representative value, 222... Measured group distribution, 230... Setting range, 240... Related information, 301... Reference point, 302... Sample point, 304... Distribution, 306... Distribution, 308... Virtual distribution, 320... Setting range, 322... Virtual group distribution

Claims

1. An analysis method for analyzing the defect rate of a semiconductor device in which a semiconductor substrate contains a first impurity and the semiconductor substrate is irradiated with a charged particle beam, A measurement step in which the concentration of the first impurity and the irradiation dose of the charged particle beam are within a set range is obtained for a group of semiconductor devices, and a measurement step in which the first and second characteristics are obtained. A measurement distribution generation step of generating a measurement distribution that shows the distribution of the measured values ​​of the first and second characteristics in the measurement group, A virtual distribution generation step is performed to simulate the first and second characteristics of a plurality of virtual semiconductor devices based on the measurement distribution, using a combination of the concentration of the first impurity and the irradiation amount of the charged particle beam, thereby generating a virtual distribution in the plurality of virtual semiconductor devices in which the samples of the first and second characteristics are distributed over a wider range than in the measurement distribution. A defect rate calculation step for calculating the defect rate in the aforementioned virtual distribution An analysis method comprising the following features.

2. The virtual distribution generation step generates the virtual distribution based on the covariance of the first and second characteristics in the measured distribution. The analysis method according to claim 1.

3. In the virtual distribution generation step, the virtual distribution is generated for each set range using the covariance of the corresponding measurement distribution. The analysis method according to claim 2.

4. In the virtual distribution generation step, the virtual distribution is generated using a common covariance for a plurality of defined ranges. The analysis method according to claim 2.

5. The system further comprises a measurement group distribution generation step, which involves acquiring group representative values ​​of the first and second characteristics for each of the multiple measurement groups corresponding to the same setting range, and generating a measurement group distribution that shows the distribution of the group representative values. In the virtual distribution generation step, the virtual distribution is generated based on the measurement distribution and the measurement group distribution. The analysis method according to any one of claims 1 to 4.

6. In the virtual distribution generation step, the virtual distribution is generated for the same setting range as the setting range of the measurement group. The analysis method according to claim 5.

7. The system further comprises a virtual group generation step, which generates a plurality of virtual groups for the same setting range as the setting range of the measurement group, based on the measurement group distribution. In the virtual distribution generation step, the measurement distribution is applied to the group representative value of each of the plurality of virtual groups to generate the virtual distribution. The analysis method according to claim 6.

8. In the virtual distribution generation step, the virtual distribution is generated for each of the set ranges, In the defect rate calculation step, the defect rate for each of the set ranges is calculated using the virtual distribution corresponding to each of the set ranges. The analysis method according to any one of claims 5 to 7.

9. A relationship information generation step involves obtaining representative distribution values ​​that represent the representative values ​​of each of the aforementioned measurement group distributions, and generating relationship information that shows the relationship between the values ​​in the set range and the representative distribution values. A virtual group distribution generation step that generates a virtual group distribution for a set range different from the measurement group distribution based on the aforementioned related information. Furthermore, In the virtual distribution generation step, the virtual distribution for each of the setting ranges is generated using the virtual group distributions corresponding to each of the setting ranges. In the defect rate calculation step, the defect rate for the set range corresponding to the virtual group distribution is calculated. The analysis method according to claim 8.

10. The system further includes a shape information acquisition step for acquiring shape information that shows the shape of each of the aforementioned measurement group distributions, In the virtual group distribution generation step, the virtual group distribution is generated based on the shape information. The analysis method according to claim 9.

11. In the defect rate calculation step, the defect rate for each of the first and second characteristics is calculated for each of the set ranges. The analysis method according to any one of claims 1 to 10.

12. In the defect rate calculation step, within each of the setting ranges, the higher of the defect rate of the first characteristic and the defect rate of the second characteristic is set as the defect rate for each setting range. The analysis method according to claim 11.

13. In the defect rate calculation step, the average defect rate of the semiconductor device is calculated based on a value obtained by multiplying the probability that the concentration of the first impurity in the semiconductor substrate and the irradiation amount of the charged particle beam fall within the respective set ranges by the defect rate within each of the set ranges. The analysis method according to any one of claims 1 to 12.

14. A design step in which the irradiation amount of the charged particle beam is determined based on the concentration of the first impurity in the semiconductor substrate used in the manufacture of the semiconductor device and the characteristics that the semiconductor device should have, An average defect rate calculation step in which the average defect rate of the semiconductor device is calculated based on the irradiation dose of the charged particle beam determined in the design stage. The analysis method according to claim 13, further comprising the above.

15. The first impurity is carbon, and the charged particle beam is helium ions. The analysis method according to any one of claims 1 to 14.

16. A program for causing a computer to perform the analysis method described in any one of claims 1 to 15.

17. A design step in which the amount of charged particle beam irradiation applied to the semiconductor substrate used in the manufacture of the semiconductor device is determined based on the defect rate in each of the set ranges calculated by the analysis method described in any one of claims 1 to 15, A manufacturing step in which the semiconductor substrate is irradiated with the charged particle beam at the irradiation dose determined in the design stage. A method for manufacturing a semiconductor device comprising the same equipment.

Images