Property Prediction System

Abstract

To provide a characteristic prediction system capable of predicting electrical characteristics which are difficult to be measured by a measuring instrument.SOLUTION: A characteristic prediction system 100 comprises a prediction computer 10, a measurement computer 20 and a measuring instrument 30. The measurement computer 20 includes a processor 21 which acquires electrical characteristics of a semiconductor chip measured by the measuring instrument 30. The prediction computer 10 includes: a memory device 12 storing therein a prediction model for predicting out-of-range characteristics which are electrical characteristics exceeding a measurable range of the measuring instrument 30; and a processor 11 for predicting the out-of-range characteristics at least from the electrical characteristics acquired by the processor 21, using the prediction model.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present disclosure relates to a property prediction system. [Background technology]

[0002] Patent Document 1 describes a method for manufacturing a semiconductor device. In the method, a burn-in necessity determination process is performed to determine whether a burn-in test is required for each semiconductor chip based on measurement data from a probe test process. In the method, the semiconductor device is divided into a first lot including packages made up of semiconductor chips determined to require a burn-in test and a second lot including packages made up of semiconductor chips determined not to require a burn-in test, based on the determination result of the burn-in necessity determination process. Then, in the method, a burn-in test is performed only on the packages included in the first lot. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent No. 6310782 Summary of the Invention [Problem to be solved by the invention]

[0004] However, semiconductor devices are sometimes required to operate in high-temperature environments and with large currents. Therefore, semiconductor devices are required to guarantee specifications regarding electrical characteristics under high-temperature environments and large currents. However, when mass-producing semiconductor devices, it may be difficult to measure electrical characteristics using measuring instruments under conditions such as high-temperature environments and large currents.

[0005] One disclosed object is to provide a characteristic prediction system that can predict electrical characteristics that are difficult to measure using a measuring instrument. [Means for solving the problem]

[0006] The property prediction system disclosed herein comprises: a measuring device (21) for acquiring electrical characteristics of the semiconductor device measured by the measuring instrument (30); a memory unit (12) storing a prediction model for predicting out-of-range characteristics, which are electrical characteristics that exceed the measurable range of a measuring instrument; A prediction device (11) that predicts out-of-range characteristics from electrical characteristics acquired by at least a measurement device using a prediction model. 、 The prediction device generates a prediction model through machine learning using training data. It is characterized by:

[0007] The characteristic prediction system disclosed herein includes a storage unit that stores a prediction model for predicting out-of-range characteristics. The characteristic prediction system then uses the prediction model to predict out-of-range characteristics from at least electrical characteristics acquired by a measurement device. This allows the characteristic prediction system to predict out-of-range characteristics that are difficult to measure using a measurement device. The property prediction system disclosed herein also includes: a measuring device (21) for acquiring electrical characteristics of the semiconductor device measured by the measuring instrument (30); a memory unit (12) storing a prediction model for predicting out-of-range characteristics, which are electrical characteristics that exceed the measurable range of a measuring instrument; a prediction device (11) that predicts out-of-range characteristics from at least the electrical characteristics acquired by the measurement device using a prediction model; The prediction device predicts out-of-range characteristics from the electrical characteristics and non-electrical information related to the semiconductor device that is different from the electrical characteristics using a prediction model; The semiconductor device is one of a plurality of chips formed on a semiconductor wafer, the non-electrical information includes at least one of coordinate information of each chip on the semiconductor wafer, configuration information of each chip on the semiconductor wafer, and manufacturing process information of the semiconductor device; The characteristic prediction system, wherein the configuration information includes information on the concentration of the drift layer and the film thickness of the drift layer.

[0008] The various aspects disclosed in this specification employ different technical means to achieve their respective objectives. The reference numerals in parentheses in the claims and in this section are intended to exemplify correspondences with the following embodiments and are not intended to limit the technical scope. The objectives, features, and advantages disclosed in this specification will become more apparent by reference to the following detailed description and the accompanying drawings. [Brief explanation of the drawings]

[0009] [Figure 1] FIG. 1 is a block diagram showing a schematic configuration of a characteristic prediction system. [Figure 2] 10 is a flowchart showing a processing operation of the characteristic prediction system. [Figure 3] FIG. 1 is a plan view showing a semiconductor wafer. [Figure 4] 1 is a diagram showing a prediction result in a characteristic prediction system. [Figure 5] 1 is a diagram showing measurement data measured by a measuring device. [Figure 6] 1 is a diagram showing non-measured data. [Figure 7] 1 is a diagram showing training data; [Figure 8] 10 is a diagram showing prediction data. [Figure 9] 1 is a diagram showing the relationship between measurement data and predicted data. [Figure 10] 10 is a flowchart showing the processing operation of a characteristic prediction system according to a modified example. [Figure 11] 10 is a diagram showing classification results. DETAILED DESCRIPTION OF THE INVENTION

[0010] Hereinafter, embodiments of the present disclosure will be described with reference to the drawings. The characteristic prediction system 100 is a system that predicts electrical characteristics of multiple semiconductor chips C1 to Cn formed on a semiconductor wafer 200. In particular, the characteristic prediction system 100 is a system that predicts electrical characteristics that are difficult to measure with a measuring instrument 30 when mass-producing the semiconductor chips C1 to Cn. In the drawings, the characteristic prediction system is referred to as PDIT, the semiconductor wafer as SEM, and the measuring instrument as MES. In the drawings, the measurement computer is referred to as 1COM, the prediction computer as 2COM, the processor as PCS, and the memory device as MEM. Each of the semiconductor chips C1 to Cn corresponds to a semiconductor device.

[0011] <Semiconductor wafer> As shown in Fig. 3, a semiconductor wafer 200 has a plurality of semiconductor chips C1 to Cn formed thereon. More specifically, the semiconductor wafer 200 has components of the semiconductor chips C1 to Cn formed on a substrate mainly composed of Si, SiC, or the like by a semiconductor process. In other words, a plurality of regions that will become the semiconductor chips C1 to Cn are formed on the semiconductor wafer 200. That is, the semiconductor wafer 200 is divided into the semiconductor chips C1 to Cn by dicing. The semiconductor chips C1 to Cn are, for example, MOSFETs or IGBTs. The components include, for example, gate electrodes, drain electrodes, source electrodes, drift layers, buffer layers, trench structures, and the like. n is a natural number greater than or equal to 2.

[0012] <Configuration> First, the configuration of a characteristic prediction system 100 will be described with reference to Fig. 1. The characteristic prediction system 100 includes a measurement computer 20, a prediction computer 10, and a measuring instrument 30.

[0013] The measurement computer 20 includes a processor 21 such as a CPU, a memory device 22 including a volatile memory and a non-volatile memory, etc. The measurement computer 20 is configured to be able to communicate with the measuring device 30 and the prediction computer 10.

[0014] The processor 21 executes a program stored in the memory device 22. The processor 21 executes the program and performs arithmetic processing to perform various controls. For example, the processor 21 controls the measuring device 30. The processor 21 instructs the measuring device 30 to measure the electrical characteristics of each of the semiconductor chips C1 to Cn. The processor 21 acquires the electrical characteristics that are the measurement results measured by the measuring device 30. The processor 21 then inputs the electrical characteristics to the prediction computer 10. The processor 21 corresponds to a measuring device. The processor 21 can also be said to be a measurement processor. The memory device 22 stores the programs executed by the processor 21, the electrical characteristics, and the like.

[0015] The measuring instrument 30 measures the electrical characteristics of each of the semiconductor chips C1 to Cn in response to instructions from the processor 21. The measuring instrument 30 measures the electrical characteristics within a measurable range (condition) of the measuring instrument 30. The measuring instrument 30 inputs measurement data including the measured electrical characteristics to the prediction computer 10.

[0016] The measurable range includes small currents and low-temperature environments. Small currents are currents that can be measured by measuring instrument 30. Low temperatures are temperatures that can be measured by measuring instrument 30. Note that measuring instrument 30 here is not an apparatus that can measure large currents exceeding 1000 A or in high-temperature environments of 200°C. The electrical characteristics measured by measuring instrument 30 can be, for example, electrical characteristics at small currents, electrical characteristics in a low-temperature environment, or electrical characteristics at small currents in a low-temperature environment.

[0017] 5, the measuring device 30 measures the on-voltage, threshold voltage, breakdown voltage, forward voltage, and switching losses (Eon, Eoff, Err) as electrical characteristics for each of the semiconductor chips C1 to Cn. The measuring device 30 may also measure the saturation current, capacitance (Crss, Ciss, Coss), voltage change (dv / dt), current change (di / dt), recovery current, surge voltage, total gate charge (Qg), etc.

[0018] Eon is the turn-on loss that occurs when the transistor is turned on. Eoff is the turn-off loss that occurs when the transistor is turned off. Err is the recovery loss.

[0019] Crss is the feedback capacitance equivalent to the gate-drain parasitic capacitance. Ciss is the sum of the gate-source parasitic capacitance and the gate-drain parasitic capacitance, and is the input capacitance related to the gate charge / discharge speed. Coss is the sum of the drain-source parasitic capacitance and the gate-drain parasitic capacitance, and is the output capacitance related to the source-drain switching speed.

[0020] dv / dt is the amount of change in drain-source voltage per unit time that occurs during switching transitions. di / dt is the amount of change in the recovery current flowing through the body diode of the MOSFET.

[0021] The measuring device 30 may also measure short circuit resistance, voltage change resistance, recovery resistance, avalanche resistance, etc. as the electrical characteristics. Furthermore, the measuring device 30 may perform a gate screening test, a withstand voltage screening test, a reverse bias test, a high-temperature gate bias test, etc. Any measuring device 30 may be used as long as it measures at least one of the electrical characteristics described above.

[0022] The prediction computer 10 includes a processor 11 such as a CPU, a memory device 12 including volatile memory and non-volatile memory, and the like. The prediction computer 10 is configured to be able to communicate with a measurement computer 20. The prediction computer 10 is configured to be able to input input data and output output data. The input data includes measurement data output from the measurement computer 20 and training data for generating a prediction model. The output data is prediction data predicted by the prediction computer 10.

[0023] The processor 11 executes a program stored in the memory device 12. The processor 11 executes the program and performs arithmetic processing to perform various controls. For example, the processor 11 generates a prediction model through machine learning using training data. As shown in FIG. 4, the processor 11 also predicts prediction data from at least measurement data using the prediction model. The processor 11 corresponds to a prediction device. The processor 11 can also be called a prediction processor. The memory device 12 stores the program executed by the processor 11, the prediction model, the prediction data, and the like.

[0024] The prediction computer 10 may also predict the prediction data from the measurement data and non-measurement data related to each of the semiconductor chips C1 to Cn that is different from the electrical characteristics. The non-measurement data corresponds to non-electrical information.

[0025] An example of non-measurement data is shown in FIG. 6. The non-measurement data includes at least one of coordinate information of each semiconductor chip C1 to Cn on the semiconductor wafer 200, configuration information of each semiconductor chip C1 to Cn on the semiconductor wafer 200, and manufacturing process information of the semiconductor device. The coordinate information is information indicating the X and Y coordinates of each semiconductor chip C1 to Cn. The configuration information includes information on the concentration and film thickness of the drift layer. The manufacturing process information includes information indicating the dimensions of the components of each semiconductor chip C1 to Cn measured during the manufacturing process. The dimensions of the components include the dimensions of the breakdown voltage holding layer, the trench dimensions in a trench MOS, and the film thickness of the gate insulating film. The dimensions of these components can be measured using, for example, a critical dimension scanning electron microscope (CEM). Critical dimension scanning electron microscope (CEM) is an abbreviation for critical dimension-scanning electron microscope.

[0026] An example of the predicted data is shown in FIG. 8. The predicted data is the electrical characteristics of each semiconductor chip C1 to Cn that are beyond the measurable range of the measuring instrument 30. The predicted data is a predicted value of the electrical characteristics of each semiconductor chip C1 to Cn when a large current exceeding 1000 A flows in a high-temperature environment exceeding 200°C, for example. The predicted data may also be a predicted value of the electrical characteristics of each semiconductor chip C1 to Cn in a high-temperature environment exceeding 200°C, for example, or a predicted value of the electrical characteristics of each semiconductor chip C1 to Cn when a large current exceeding 1000 A flows. Therefore, the predicted data can be considered electrical characteristics that are difficult to measure with the measuring instrument 30. The predicted data corresponds to out-of-range characteristics. The predicted data can be considered, for example, electrical characteristics at the time of a large current, electrical characteristics in a high-temperature environment, or electrical characteristics at the time of a large current in a high-temperature environment.

[0027] The prediction computer 10 predicts, for example, values of the electrical characteristics measured by the measuring instrument 30 under high-temperature conditions and large currents as prediction data. The prediction computer 10 may also predict, as prediction data, results of a gate screening test, a withstand voltage screening test, a reverse bias test, and a high-temperature gate bias test under high-temperature conditions and large currents. Furthermore, the prediction computer 10 may predict durability performance and a market failure rate as prediction data. Any prediction computer 10 may be employed as long as it predicts at least one of a plurality of electrical characteristics as prediction data.

[0028] The prediction model is used to predict the electrical characteristics of each of the semiconductor chips C1 to Cn. In particular, in this embodiment, a prediction model is employed that predicts predicted data, which is electrical characteristics of each of the semiconductor chips C1 to Cn that are beyond the measurable range of the measuring instrument 30, among the electrical characteristics of each of the semiconductor chips C1 to Cn. The prediction model can also be said to be a mechanism or algorithm for predicting predicted data from measurement data as input data. Therefore, it can be said that the processor 11 uses the prediction model to predict the electrical characteristics of each of the semiconductor chips C1 to Cn under conditions that are difficult to measure with the measuring instrument 30.

[0029] The prediction model is a machine learning model generated by a well-known method. Machine learning techniques include deep learning and simple perceptron. In the case of deep learning, it may be based on a simple perceptron or a multilayer perceptron. If the training data is an imbalanced data set, an oversampling or undersampling process may be included. Figure 9 shows an example of a forward voltage predicted (estimated) by a multilayer perceptron. In Figure 9, the horizontal axis represents the measurement data (actual measured value) and the vertical axis represents the predicted data. A machine learning model can also be called a trained model.

[0030] As shown in FIG. 7, the training data includes measurement data that is the result of measurements at multiple low currents, electrical characteristics in a high-temperature environment or at a high current, and non-measurement data. FIG. 7 illustrates only the training data for the first wafer of the first lot. However, the training data is not limited to this. The training data may include measurement data and non-measurement data for multiple wafers generated from multiple lots. The electrical characteristics at a high current in the training data are, for example, electrical characteristics measured at a high current for several semiconductor chips among the multiple semiconductor chips C1 to Cn. The electrical characteristics in a high-temperature environment in the training data are, for example, electrical characteristics measured in a high-temperature environment for several semiconductor chips among the multiple semiconductor chips C1 to Cn.

[0031] <Processing operation> Here, the prediction process of the prediction data will be described with reference to Fig. 2. First, in the chip preparation step S10, a semiconductor wafer 200 on which a plurality of semiconductor chips C1 to Cn are formed is prepared. That is, in the chip preparation step S10, the semiconductor wafer 200 to be predicted is prepared.

[0032] In the electrical test step S11, the processor 21 of the measurement computer 20 issues a measurement instruction to the measuring instrument 30. Then, in response to the instruction from the processor 21, the measuring instrument 30 measures the electrical characteristics of each of the semiconductor chips C1 to Cn.

[0033] In the measurement data input step S12, the measuring instrument 30 inputs measurement data, which is the measured electrical characteristics, to the measurement computer 20, as shown in Fig. 5. This allows the measurement computer 20 to acquire measurement data for predicting prediction data. The measurement data input step S12 can also be considered a measurement data acquisition step in which the measurement computer 20 acquires measurement data.

[0034] In the non-measurement data input step S13, non-measurement data as shown in Fig. 6 is input to the measurement computer 20. As a result, the measurement computer 20 acquires non-measurement data for predicting prediction data. The non-measurement data input step S13 can also be said to be a non-measurement data acquisition step in which the measurement computer 20 acquires non-measurement data.

[0035] In the teacher data input step S14, teacher data as shown in Fig. 7 is input to the prediction computer 10. As a result, the prediction computer 10 acquires teacher data for generating a prediction model. The teacher data input step S14 can also be considered a teacher data acquisition step in which the prediction computer 10 acquires teacher data.

[0036] In the trained model generation step S15, the prediction computer 10 generates a prediction model as a trained model using training data such as that shown in Fig. 7. The prediction computer 10 generates a prediction model by machine learning from the acquired training data, which are measurement results at multiple small currents, and non-measurement data, and electrical characteristics in a high-temperature environment or at a large current.

[0037] In the prediction step S16, the prediction computer 10 predicts prediction data. The prediction computer 10 uses a prediction model to make predictions from the measurement data input in step S12 and the non-measurement data input in step S13. In this way, the prediction computer 10 obtains the prediction data shown in FIG. 8. In this way, the prediction computer 10 can obtain the electrical characteristics of each of the semiconductor chips C1 to C2, which are difficult to measure with the measuring instrument 30.

[0038] In the prediction step S16, it is preferable that the prediction computer 10 uses two or more measurement data, which allows the prediction computer 10 to improve prediction accuracy, leading to reduced tolerances and test costs.

[0039] In the chip shipping step S17, the semiconductor chips C1 to C2 for which the measurement of the predicted data has been completed are shipped. Note that in the chip shipping step S17, only the semiconductor chips whose predicted data meet the reference value may be shipped from among the semiconductor chips C1 to C2.

[0040] The means and / or functions provided by the computers 10, 20 can be provided by software recorded in the tangible memory devices 12, 22 and a computer that executes the software, software alone, hardware alone, or a combination thereof. For example, when the computers 10, 20 are provided by electronic circuits that are hardware, the means and / or functions can be provided by digital circuits including a large number of logic circuits, or analog circuits.

[0041] <Effects> The characteristic prediction system 100 disclosed herein includes a memory device 12 that stores a prediction model for predicting prediction data. The characteristic prediction system 100 uses the prediction model to predict prediction data from at least electrical characteristics (measurement data) acquired by a measuring instrument 30. Therefore, the characteristic prediction system 100 can predict prediction data (electrical characteristics) that are difficult to measure by the measuring instrument 30.

[0042] In particular, semiconductor chips C1 to Cn, which are primarily made of SiC, require temperatures exceeding 200°C and large currents exceeding 1000A. Even for such semiconductor chips C1 to Cn, specifications must be guaranteed through electrical testing. Furthermore, in mass production of semiconductor chips C1 to Cn, measurements in high-temperature environments or at high currents require expensive measuring instruments 30, which may increase measurement costs or even make measurements impossible. In contrast, the characteristic prediction system 100 uses a prediction model to predict the electrical characteristics of each semiconductor chip C1 to Cn that exceed the measurable range of the measuring instrument 30. Therefore, the characteristic prediction system 100 can guarantee specifications while suppressing increases in measurement costs.

[0043] It is also possible to predict electrical characteristics beyond the measurable range of the measuring instrument 30 using a function. However, for semiconductor chips C1 to Cn, it is necessary to consider tolerances that reduce prediction accuracy due to variations in epitaxial concentration and film thickness, and process variations. Furthermore, the function becomes nonlinear, making it difficult to achieve accuracy. In such cases, the tolerances must be incorporated into the chip performance, which increases chip costs. In contrast, the characteristic prediction system 100 predicts prediction data using a prediction model, allowing for highly accurate predictions. This makes it possible to reduce tolerances and measurement costs.

[0044] Incidentally, variations in the concentration and film thickness of the drift layer tend to be concentric, making positional information useful. Furthermore, the concentration and film thickness of the drift layer strongly affect the DC (direct current) characteristics and AC (alternating current) characteristics. Therefore, it is preferable for the characteristic prediction system 100 to use the concentration and film thickness of the drift layer as non-measurement data. In other words, the characteristic prediction system 100 can improve prediction accuracy by incorporating the concentration and film thickness of the drift layer as non-measurement data into machine learning. It can also be said that the characteristic prediction system 100 can generate a prediction model that enables accurate prediction by incorporating the concentration and film thickness of the drift layer as non-measurement data into machine learning. Furthermore, the configuration information as non-measurement data may include information indicating the defect state of the semiconductor wafer 200 and information indicating the concentration and film thickness of the buffer layer.

[0045] The dimensions of components, such as the dimensions of the breakdown voltage holding layer, the trench dimensions in a trench MOS, and the thickness of the gate insulating film, have a strong effect on the DC and AC characteristics. Therefore, the characteristic prediction system 100 can improve prediction accuracy by incorporating the dimensions of the components into machine learning as non-measurement data. It can also be said that the characteristic prediction system 100 can generate a prediction model that enables accurate predictions by incorporating the dimensions of the components into machine learning.

[0046] Measurement data such as forward voltage, saturation current, capacitance, switching loss, voltage change, current change, recovery current, surge voltage, and total gate charge are nonlinear with respect to current and voltage values. Furthermore, because the coefficients also depend on the drift layer concentration, drift layer film thickness, and process variations, high prediction accuracy cannot be achieved even if a unique function is determined. Therefore, the characteristic prediction system 100 can improve prediction accuracy by applying machine learning to non-measurement data in addition to measurement data. It can also be said that the characteristic prediction system 100 can generate a prediction model that enables accurate predictions by applying machine learning to non-measurement data in addition to measurement data.

[0047] For tolerances such as short-circuit tolerance, voltage change tolerance, recovery tolerance, and avalanche tolerance, it is difficult to measure (electrically test) the semiconductor chips C1 to Cn under a load equivalent to that of actual operation due to ringing caused by the parasitic inductance of the measuring instrument 30 and the measurement limits of the measuring instrument 30. Therefore, it is conceivable to conduct sampling tests, tests after module assembly, or incorporate tests under a load lower than that of actual operation into the electrical test. Furthermore, the number of parameters that determine the tolerances is large and complex. Therefore, it has been difficult to predict the tolerances when a load equivalent to that of actual operation is applied from the electrical test of the semiconductor chips C1 to Cn. In contrast, the characteristic prediction system 100 uses machine learning, which can handle a large number of parameters and build complex prediction models, and therefore can predict the tolerance when a desired load is applied.

[0048] The preferred embodiments of the present disclosure have been described above. However, the present disclosure is not limited to the above embodiments, and various modifications are possible within the scope of the present disclosure. The present disclosure is not limited to the combinations shown in the embodiments, and can be implemented using various combinations.

[0049] (Variation) A modified characteristic prediction system 100 will be described with reference to Figures 10 and 11. The characteristic prediction system 100 may perform a classification step in addition to the processing of the above embodiment. In Figure 10, the same step numbers are assigned to the same processing as in the above embodiment. Steps S14 and S15 are the same as steps S14 and S15 in Figure 2. However, the names have been changed to distinguish them from steps S20 and S21.

[0050] The characteristic prediction system 100 performs a second teacher data input step S20, a second trained model generation step S21, and a classification step S22 between the prediction step S16 and the chip shipping step S17. However, steps S20 and S21 can also be performed before step S16 is completed.

[0051] In the second teacher data input step S20, second teacher data is input to the prediction computer 10. The second teacher data is teacher data for generating a second trained model. In other words, the second teacher data is different from teacher data for generating a prediction model. The prediction computer 10 acquires second teacher data for generating a classification model, which is the second trained model. The second teacher data input step S20 can also be said to be a second teacher data acquisition step in which the prediction computer 10 acquires the second teacher data.

[0052] In the second trained model generation step S21, the prediction computer 10 generates a classification model. The second trained model is a machine learning model for performing classification, unlike the prediction model. The prediction computer 10 generates the classification model by machine learning using the second training data.

[0053] The classification model may include, for example, logistic regression and linear support vector machines as linear classifiers, or k-nearest neighbors, decision trees, random forests, nonlinear support vector machines, and deep learning as nonlinear classifiers. If the dataset is imbalanced, oversampling or undersampling may be performed.

[0054] In the classification step S22, the prediction computer 10 classifies the semiconductor chips C1 to Cn. The prediction computer 10 classifies (sorts, identifies) the semiconductor chips C1 to Cn using the classification model and at least one prediction data acquired in step S16. That is, the prediction computer 10 determines a boundary line and classifies the semiconductor chips C1 to Cn according to the prediction data.

[0055] 11 shows an example in which semiconductor chips C1 to Cn are classified by power loss, which is a combination of conduction loss and switching loss, including one or more pieces of data on on-resistance, turn-on loss, turn-off loss, and recovery loss predicted by a prediction model. The characteristic prediction system 100 classifies the chips into Class A, which is hatched diagonally, and Class B, which is hatched dotted, with a boundary line as the dividing line. As shown in FIGS. 11(a) and 11(b), the classification in the classification step S22 can be changed depending on the purpose. In other words, the boundary line can be changed depending on the purpose, etc.

[0056] In this way, the prediction computer 10 obtains classification data for classifying the semiconductor chips C1 to Cn, as shown in Fig. 11. That is, the characteristic prediction system 100 predicts out-of-range characteristics of the semiconductor chips C1 to Cn, and classifies the semiconductor chips C1 to Cn based on the prediction data for each of the semiconductor chips C1 to Cn. By performing the classification step S22, the characteristic prediction system 100 can reduce the tolerance for each item, which is expected to improve inverter performance and chip yield.

[0057] Furthermore, for example, in an inverter circuit, a configuration in which semiconductor chips are driven in parallel is conceivable. Such an inverter circuit may be configured with multiple semiconductor modules in which multiple semiconductor chips are connected in parallel. In this case, the characteristic prediction system 100 may classify the semiconductor chips C1 to Cn using the threshold voltage at high temperature predicted by the prediction model. This allows multiple semiconductor chips classified into the same class to be used in one module and one inverter circuit. Therefore, the characteristic prediction system 100 can suppress unbalanced operation of semiconductor chips within or between modules.

[0058] Although the present disclosure has been described with reference to the embodiments, it is understood that the present disclosure is not limited to the embodiments or structures. The present disclosure also encompasses various modifications and modifications within the scope of equivalents. In addition, although various combinations and forms are shown in the present disclosure, other combinations and forms including only one element, more, or less than one element are also within the scope and spirit of the present disclosure.

[0059] (Disclosure of technical ideas) This specification discloses multiple technical ideas described in the following multiple clauses. Some clauses may be written in a multiple dependent form, with the subsequent clause referring to the preceding clause as an alternative. Furthermore, some clauses may be written in a multiple dependent form, referring to another multiple dependent clause. These multiple dependent clauses define multiple technical ideas.

[0060] (Technical thought 1) a measuring device (21) for acquiring electrical characteristics of the semiconductor device measured by the measuring device (30); a storage unit (12) storing a prediction model for predicting out-of-range characteristics, which are electrical characteristics that exceed the measurable range of the measurement device; a prediction device (11) that uses the prediction model to predict the out-of-range characteristics from at least the electrical characteristics acquired by the measurement device.

[0061] (Technical thought 2) The characteristic prediction system described in Technical Idea 1, wherein the prediction device generates the prediction model through machine learning using training data.

[0062] (Technical Thought 3) The characteristic prediction system described in Technical Idea 1 or 2, wherein the prediction device uses the prediction model to predict the out-of-range characteristics from the electrical characteristics and non-electrical information about the semiconductor device that is different from the electrical characteristics.

[0063] (Technical Thought 4) the semiconductor device is one of a plurality of chips formed on a semiconductor wafer, The characteristic prediction system described in Technical Idea 3, wherein the non-electrical information includes at least one of coordinate information of each chip on the semiconductor wafer, configuration information of each chip on the semiconductor wafer, and manufacturing process information of the semiconductor device.

[0064] (Technical Thought 5) The characteristic prediction system according to Technical Concept 4, wherein the configuration information includes information on the concentration of the drift layer and the film thickness of the drift layer.

[0065] (Technical Thought 6) The characteristic prediction system according to Technical Idea 4, wherein the manufacturing process information includes information indicating dimensions of components of the semiconductor device measured during the manufacturing process.

[0066] (Technical Thought 7) The characteristic prediction system according to any one of Technical Ideas 1 to 6, wherein the electrical characteristics acquired by the measuring device include at least one of on-voltage, threshold voltage, breakdown voltage, forward voltage, saturation current, capacitance, switching loss, voltage change, current change, recovery current, surge voltage, and total gate charge.

[0067] (Technical Thought 8) The characteristic prediction system according to any one of Technical Ideas 1 to 6, wherein the electrical characteristics acquired by the measuring device include at least one of short circuit resistance, voltage change resistance, recovery resistance, and avalanche resistance.

[0068] (Technical Thought 9) the semiconductor device is one of a plurality of chips formed on a semiconductor wafer, The characteristic prediction system described in any one of technical ideas 1 to 8, wherein the prediction device predicts the out-of-range characteristics of multiple chips and classifies the multiple chips based on the out-of-range characteristics of each chip. [Explanation of symbols]

[0069] 10... prediction computer, 20... measurement computer, 30... measuring instrument, 11, 21... processor, 12, 22... memory device, 100... characteristic prediction system, 200... semiconductor wafer

Claims

1. a measuring device (21) for acquiring electrical characteristics of the semiconductor device measured by the measuring instrument (30); a storage unit (12) storing a prediction model for predicting out-of-range characteristics, which are electrical characteristics that exceed the measurable range of the measuring instrument; and a prediction device (11) that predicts the out-of-range characteristics from the electrical characteristics acquired by at least the measurement device using the prediction model, The prediction device is a characteristic prediction system that generates the prediction model through machine learning using training data.

2. a measuring device (21) for acquiring electrical characteristics of the semiconductor device measured by the measuring instrument (30); a storage unit (12) storing a prediction model for predicting out-of-range characteristics, which are electrical characteristics that exceed the measurable range of the measuring instrument; and a prediction device (11) that predicts the out-of-range characteristics from the electrical characteristics acquired by at least the measurement device using the prediction model, the prediction device predicts the out-of-range characteristic from the electrical characteristic and non-electrical information related to the semiconductor device that is different from the electrical characteristic using the prediction model; the semiconductor device is one of a plurality of chips formed on a semiconductor wafer, the non-electrical information includes at least one of coordinate information of each chip on the semiconductor wafer, configuration information of each chip on the semiconductor wafer, and manufacturing process information of the semiconductor device; The configuration information includes information on the concentration of a drift layer and the film thickness of the drift layer.

3. 2. The characteristic prediction system according to claim 1, wherein the prediction device predicts the out-of-range characteristic using the prediction model from the electrical characteristic and non-electrical information about the semiconductor device that is different from the electrical characteristic.

4. the semiconductor device is one of a plurality of chips formed on a semiconductor wafer, 4. The characteristic prediction system according to claim 3, wherein the non-electrical information includes at least one of coordinate information of each chip on the semiconductor wafer, configuration information of each chip on the semiconductor wafer, and manufacturing process information of the semiconductor device.

5. The characteristic prediction system according to claim 4 , wherein the configuration information includes information on the concentration of a drift layer and the film thickness of the drift layer.

6. 5. The characteristics prediction system according to claim 4, wherein said manufacturing process information includes information indicating dimensions of components of said semiconductor device measured during said manufacturing process.

7. 2. The characteristic prediction system according to claim 1, wherein the electrical characteristics acquired by the measuring device include at least one of an on-voltage, a threshold voltage, a breakdown voltage, a forward voltage, a saturation current, a capacitance, a switching loss, a voltage change, a current change, a recovery current, a surge voltage, and a total gate charge.

8. 2. The characteristic prediction system according to claim 1, wherein the electrical characteristics acquired by the measuring device include at least one of a short circuit resistance, a voltage change resistance, a recovery resistance, and an avalanche resistance.

9. the semiconductor device is one of a plurality of chips formed on a semiconductor wafer, The characteristic prediction system according to claim 1 , wherein the prediction device predicts the out-of-range characteristics of a plurality of the chips and classifies the plurality of chips based on the out-of-range characteristics of each chip.

Images