Analysis device, analysis program, method, semiconductor device, and semiconductor wafer

Abstract

An analysis device (100) is configured to perform a process for analyzing the electromagnetic noise performance of a sample. The analysis device (100) comprises: a model generation unit (102) for generating a model in which a noise performance value pertaining to the electromagnetic noise performance of a first sample (S1) is used as an objective variable and a plurality of product characteristic values pertaining to the first sample (S1) are used as explanatory variables; and a performance prediction unit (103) and a data output unit (104) serving as output units for receiving, as inputs, a plurality of product characteristic values pertaining to a second sample (S2) different from the first sample (S1) corresponding to the plurality of product characteristic values of the first sample (S1) in the explanatory variables of the model, and outputting data related to the analysis of the electromagnetic noise performance of the second sample (S2) using the model.

Description

Analytical device, analytical program, method, semiconductor device, and semiconductor wafer CROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application is based on Patent Application No. 2023-214013 filed in Japan on December 19, 2023, and the contents of the original application are incorporated by reference in their entirety.

[0002] The disclosure of this specification relates to a technique for analyzing the electromagnetic noise performance of electrical and electronic products.

[0003] There are known techniques for analyzing the electromagnetic noise performance of electrical and electronic products, etc. For example, Patent Document 1 discloses a technique for identifying the frequency of electromagnetic noise that causes a product to malfunction.

[0004] Patent No. 7075121

[0005] The technology disclosed in the aforementioned Patent Document 1 can identify the frequency at which a malfunction occurs and the frequency at which a countermeasure should be taken, but cannot analyze and identify the cause of the malfunction, the malfunction mechanism, etc. As a result, there is a concern that analyzing the product will require a large number of man-hours.

[0006] One of the purposes of the disclosure of this specification is to provide an analysis device, an analysis program, a method, a semiconductor device, and a semiconductor wafer that suppress an increase in the number of steps required for analyzing a product.

[0007] One aspect disclosed herein is an analysis device configured to execute an analysis process of the electromagnetic noise performance of a sample, and includes: a model generation unit that generates a model in which a noise performance value of the electromagnetic noise performance of a first sample is a dependent variable and a plurality of product characteristic values ​​of the first sample are explanatory variables; and an output unit that inputs a plurality of product characteristic values ​​of a second sample, different from the first sample, that correspond to the plurality of product characteristic values ​​of the first sample, into the explanatory variables of the model, and outputs data related to the analysis of the electromagnetic noise performance of the second sample using the model.

[0008] Another disclosed aspect is an analysis program configured to execute an analysis process of the electromagnetic noise performance of a sample using at least one processor, and is configured to cause the at least one processor to execute the following steps: generate a model in which the noise performance value of the electromagnetic noise performance of a first sample is used as a dependent variable and multiple product characteristic values ​​of the first sample are used as explanatory variables; input multiple product characteristic values ​​of a second sample, different from the first sample, that correspond to the multiple product characteristic values ​​of the first sample, into the explanatory variables of the model; and output data related to the analysis of the electromagnetic noise performance of the second sample using the model.

[0009] Another disclosed aspect is a method for generating data related to the electromagnetic noise performance of a sample, comprising: generating a model in which a noise performance value of the electromagnetic noise performance of a first sample is used as a dependent variable and a plurality of product characteristic values ​​of the first sample are used as explanatory variables; and inputting a plurality of product characteristic values ​​of a second sample, different from the first sample, that correspond to the plurality of product characteristic values ​​of the first sample into the explanatory variables of the model, and generating data related to an analysis of the electromagnetic noise performance of the second sample using the model.

[0010] According to this aspect, a model is generated using a first sample, and data related to the electromagnetic noise performance of a second sample is obtained using the generated model. That is, it is possible to easily analyze the electromagnetic noise performance of a second sample, which is different from the first sample, based on information about the first sample.

[0011] The generated model uses the noise performance value of the electromagnetic noise performance as the objective variable and the product characteristic values ​​of the first sample as explanatory variables. By inputting the product characteristic values ​​of the second sample into this model instead of the first sample, data relevant to the analysis can be obtained. In other words, by analyzing the basic characteristics that determine the electromagnetic noise performance of the second sample, the cause of malfunction can be easily identified. As a result, the number of steps required for product analysis can be reduced.

[0012] Another disclosed aspect is an analysis device configured to execute an analysis process of the electromagnetic noise performance of a sample, and includes: a model generation unit that generates a model in which a noise performance value of the electromagnetic noise performance of a first sample is a dependent variable and multiple product characteristic values ​​of the first sample are explanatory variables; and an output unit that inputs the multiple product characteristic values ​​of the first sample as explanatory variables of the model and outputs data related to the analysis of the electromagnetic noise performance of the first sample using the model.

[0013] According to this aspect, a model is generated using a first sample, and the product characteristic values ​​of the first sample are input into the generated model to output data related to electromagnetic noise performance. This makes it possible to easily verify whether the generated model is normal. As a result, it is possible to suppress an increase in the number of steps required for product analysis.

[0014] Another disclosed aspect is an analytical device configured to perform an analysis process of the electromagnetic noise performance of a sample, and includes: a memory that stores data related to the analysis of the electromagnetic noise performance of the sample; and a sample extraction unit that extracts, from the samples stored in the memory, samples that have electromagnetic noise performance that satisfies any given condition.

[0015] According to this embodiment, since it is possible to easily extract samples that correspond to given conditions, it is possible to suppress an increase in the number of steps required for analyzing products.

[0016] Another disclosed aspect is a semiconductor device configured to enable measurement of product characteristic values ​​for analyzing electromagnetic noise performance, comprising: a substrate for mounting a circuit; a measurement circuit that is a circuit for measuring product characteristic values ​​and is different from a product circuit that is a circuit for performing the product's functions; and an output terminal that outputs the measurement results from the measurement circuit.

[0017] According to this aspect, a measurement circuit is provided in the semiconductor device, and measurement results of product characteristic values ​​can be obtained from the measurement circuit via the output terminal, so that an increase in the number of steps required for analyzing the product can be suppressed.

[0018] Another disclosed aspect is a semiconductor wafer configured to enable measurements of product characteristic values ​​for analyzing electromagnetic noise performance, comprising: a plurality of dies configured to mount product circuits, which are circuits for performing product functions, on each die, and arranged so as to be spaced apart from each other across scribe areas; and a plurality of measurement circuits, which are circuits for measuring product characteristic values, arranged in the scribe areas so as to form individual correspondences with the plurality of dies.

[0019] According to this embodiment, it is possible to obtain measurement results of product characteristic values ​​through a measurement circuit that has individual correspondences with multiple dies, thereby suppressing an increase in the amount of work required for analyzing the product, including the impact on each die that makes up the wafer.

[0020] Note that the symbols in parentheses included in the claims etc. are intended to exemplify the correspondence with the parts of the embodiments described below, and are not intended to limit the technical scope.

[0021] 11. A schematic configuration diagram of a system for analyzing electromagnetic noise performance. A configuration diagram for explaining the functions of an analysis device. A flowchart showing an example of a model generation process. A flowchart showing an example of a prediction process. A flowchart showing an example of processing for multiple frequencies. A graph showing the occurrence rate as an explanatory variable. A graph showing the forward power and maximum difference for each sample group. A graph showing the forward power and maximum difference for each sample group. A top view showing an example of a semiconductor chip equipped with a measurement circuit. A top view showing an example of a semiconductor chip equipped with a measurement circuit. A diagram showing an example of a semiconductor wafer equipped with a measurement circuit. An enlarged view of part XII in FIG. 11. A diagram corresponding to FIG. 12 showing another example of a semiconductor wafer.

[0022] Hereinafter, several embodiments will be described with reference to the drawings. Note that corresponding components in each embodiment are given the same reference numerals, and redundant description may be omitted. When only a portion of the configuration is described in each embodiment, the configuration of another embodiment described previously can be applied to the remaining portion of the configuration. Furthermore, in addition to the combinations of configurations explicitly stated in the description of each embodiment, configurations of several embodiments can also be partially combined together even if not explicitly stated, as long as there is no particular problem with the combination.

[0023] First Embodiment The system 1 shown in FIG. 1 is a system for analyzing the electromagnetic noise performance of products such as electrical and electronic products. The analysis here may include prediction of the product's electromagnetic noise performance. The analysis here may include sampling. The electrical or electronic products to be analyzed here refer to products that have one or more external power supply points or one or more internal power sources, and may broadly include, for example, automobiles, automotive parts, home appliances, industrial products, agricultural products, and semiconductor products such as integrated circuits and various sensors that use such integrated circuits.

[0024] Electrical and electronic products, including automotive parts, are required to have sufficient electromagnetic noise performance as one of their quality specifications. That is, the product must have a sufficiently low level of electromagnetic noise emission (EMI: Electro Magnetic Interference) and a sufficiently high level of immunity (EMS: Electro Magnetic Susceptibility) to prevent malfunction due to external electromagnetic noise.

[0025] The electromagnetic noise performance of these electrical or electronic products is determined by the complex interactions of multiple factors, such as the circuit configuration, materials used, and housing shape. Therefore, product design that ensures electromagnetic noise performance is highly challenging, and predicting electromagnetic noise performance is difficult. For this reason, if it is determined that the electromagnetic noise performance of an electrical or electronic product does not meet requirements, a significant amount of development work, such as redesign, may be required.

[0026] Furthermore, immunity performance often ultimately leads to malfunctions of semiconductor integrated circuits. Multiple circuit blocks within an integrated circuit interact with each other when electromagnetic noise enters from outside, and the weakest circuit block determines the rate of the malfunction. However, it takes a great deal of time and effort to identify which circuit block caused the malfunction, i.e., to identify the malfunction mechanism.

[0027] Generally, electromagnetic noise performance is evaluated in accordance with evaluation methods defined by international standards depending on the application of the electrical or electronic product. For example, the ISO 11452 series, which applies to automotive parts, requires that immunity performance be confirmed in accordance with specified evaluation methods over an extremely wide frequency band from 0.01 MHz to 18 GHz. If the injected frequency range changes significantly, the malfunction mechanism will also change accordingly, leading to further increases in development man-hours.

[0028] The number of samples for evaluating electromagnetic noise performance is often one, for example, in the case of automotive parts, as a matter of commercial practice. In this disclosure, the number of first samples (S1) is deliberately set to two or more, or one or more samples are evaluated under different conditions, to obtain "variation in electromagnetic noise performance." This focuses on the relationship between this and "variation in multiple product characteristic values" of the first samples, and introduces a statistical analysis method that has not been applied in this technical field. This method makes it possible to easily analyze the relationship between electromagnetic noise performance and product characteristic values ​​using mathematical techniques, reducing the amount of work required to analyze malfunction mechanisms and take countermeasures.

[0029] The system 1 provides an analysis method that is effective in reducing development man-hours by identifying the cause of malfunction. The system 1 may include a measurement device 200, a data server 300, and an analysis device 100.

[0030] The measuring device 200 is configured to be capable of electrically connecting to a product. The measuring device 200 is configured to be capable of measuring the electrically connected product. As shown in Fig. 2, the measuring device 200 may include a noise performance value measuring unit 201 and a product characteristic value measuring unit 202.

[0031] The noise performance value measurement unit 201 is configured to be able to actually measure the electromagnetic noise performance of a product and output a noise performance value based on the actual measurement to the outside of the measurement device 200. The noise performance value will be described in detail below, but it may be the actual measured value of the electromagnetic noise performance itself, a calculated value calculated based on the actual measured value, a calculated value obtained by simulation, etc. When a calculated value obtained by simulation is used, the measurement device 200 or the noise performance value measurement unit 201 may be replaced by a simulation device that performs a simulation. The noise performance value measurement unit 201 may output the noise performance value to the data server 300, or may output it directly to the analysis device 100.

[0032] The product characteristic value measurement unit 202 is configured to measure product characteristic values ​​of the product and output the product characteristic values ​​to the outside of the measurement device 200. The product characteristic values, which will be described in detail later, may be actual measured values ​​of current or voltage values ​​measured from the product, or may be calculated values ​​calculated based on the actual measured values, calculated values ​​calculated by simulation, etc. When calculated values ​​calculated by simulation are used, the measurement device 200 or the product characteristic value measurement unit 202 may be replaced by a simulation device that executes a simulation. The simulation may be, for example, a Monte Carlo simulation that takes into account manufacturing variations based on circuit design information. The product characteristic value measurement unit 202 may output the product characteristic values ​​to the data server 300 or directly to the analysis device 100.

[0033] The data server 300 is mainly composed of a storage medium and a computer that controls the storage medium. The data server 300 accumulates data measured by the measuring device 200. The data server 300 may be communicably connected to the measuring device 200 via wired or wireless communication, and may sequentially acquire data from the measuring device 200 without user operation. The data server 300 may store the ID of a sample individual in a measured product in association with a noise performance value and a product characteristic value.

[0034] Furthermore, the data server 300 may be communicably connected to the analysis device 100 via wired or wireless communication. The data server 300 may output the noise performance value and the product characteristic value to the analysis device 100. At this time, the ID of the individual sample may also be output. The data server 300 may output the noise performance value and the product characteristic value for each product sample, or may output the noise performance value and the product characteristic value for multiple samples together.

[0035] Furthermore, a user of the system 1 may save the noise performance values ​​and product characteristic values ​​accumulated in the data server 300 to a terminal 310 including a storage medium such as a USB memory, connect the terminal 310 to the analysis device 100, and input the noise performance values ​​and product characteristic values ​​into the analysis device 100.

[0036] <Configuration of the Analysis Apparatus> The analysis apparatus 100 is primarily configured as a computer and includes at least one memory 10 and one processor 20. The memory 10 may be at least one type of non-transient physical storage medium, such as a semiconductor memory such as a flash memory, a magnetic medium such as a hard disk drive (HDD), or a non-volatile storage medium such as an optical medium, that non-temporarily stores computer programs and data that can be read by the processor 20. Furthermore, the memory 10 may be provided with a rewritable volatile storage medium such as a random access memory (RAM). The computer program may be an analysis program that analyzes electromagnetic noise performance. The processor 20 includes at least one type of core, such as a central processing unit (CPU), a graphics processing unit (GPU), or a reduced instruction set computer (RISC)-CPU.

[0037] The computer in the analysis device 100 may be configured by installing an analysis program on a personal computer equipped with a general-purpose operating system, or may be a dedicated device specialized for analysis.

[0038] The analysis device 100 may include a data input unit 101, a model generation unit 102, a performance prediction unit 103, a data output unit 104, an internal memory unit 105, a sample extraction unit 106, and a display unit 107 as processing units that realize functions by the processor 20 executing an analysis program.

[0039] Noise performance values ​​and product characteristic values ​​are input to the data input unit 101. Here, the noise performance values ​​may be actual measured values ​​of electromagnetic noise performance, as described above, or may be calculated values ​​of electromagnetic noise performance. Here, the electromagnetic noise performance may be noise performance related to electromagnetic compatibility (Electromagnetic Compatibility). The electromagnetic noise performance may include, for example, electromagnetic emission performance and electromagnetic immunity performance, as well as malfunction immunity performance in electrostatic testing. The electromagnetic immunity performance may be a malfunction immunity value against electromagnetic noise obtained through evaluation in accordance with evaluation methods such as international standards.

[0040] The measured value of the electromagnetic noise performance may be a value obtained by evaluating the electromagnetic noise performance through actual measurement in accordance with one test method specified in an international standard or the like.

[0041] Test methods based on international standards include, for example, the IEC 61967 series (integrated circuits, evaluation of electromagnetic emissions), the IEC 62132 series (integrated circuits, evaluation of electromagnetic immunity), CISPR 25 (Vehicles, small craft, and internal combustion engines - Radio disturbance characteristics - Limits and methods of measurement for the protection of on-board receivers), the ISO 11451 series (Road vehicles - Vehicle test methods for electrical disturbances due to narrowband radiated electromagnetic energy), the ISO 11452 series (Road vehicles - Component test methods for electrical disturbances due to narrowband radiated electromagnetic energy), CISPR 32 (Electromagnetic compatibility of multimedia equipment - Emission requirements), CISPR 35 (Electromagnetic compatibility of multimedia equipment - Immunity requirements), ANSI C63.4 (American National Standard for Measurement of Radio Noise Emissions from Low-Voltage Electrical and Electronic Equipment in the Frequency Range of 9 kHz to 40 GHz), and evaluation methods defined in IEC 61000, evaluation methods that apply or quote them, or similar test evaluation methods that partially modify the tests.

[0042] The calculated value of electromagnetic noise performance may be, for example, a result of modeling an internal integrated circuit as an equivalent circuit and calculating the electromagnetic noise performance value through simulation using a computer or the like. The calculated value of electromagnetic noise performance may be a result of simulating modeling the integrated circuit as an equivalent circuit. The calculated value of electromagnetic noise performance may be a result of simulating a model including a test setup specified in the aforementioned international standards or the like.

[0043] The product characteristic values ​​may be, for example, evaluation results of various tests, such as a wafer acceptance test (WAT), during the manufacturing process. The product characteristic values ​​may be evaluation results, regardless of whether they are included in normal manufacturing process tests, and may be evaluation results of circuit characteristics of electrical or electronic products. The evaluation results of the circuit characteristic values ​​may include at least one of the following: oscillation frequency, output value of an analog-to-digital converter (ADC) or a digital-to-analog converter (DAC), output value of an amplifier circuit, gain of an amplifier circuit, output value of a constant voltage or constant current source circuit, output values ​​of various digital circuits, various resistance values, various capacitance values, and output value of a filter circuit including an electromagnetic noise filter. The product characteristic values ​​may be predicted values ​​of inspections during the manufacturing process, circuit characteristics, etc., calculated by simulating manufacturing variations using a computer or the like.

[0044] The model generation unit 102 generates a model for predicting electromagnetic noise performance based on the noise performance value and the product characteristic value. Below, a method for generating the model will be described using multiple regression analysis as an example, but it is not limited to multiple regression analysis, and machine learning techniques including simple regression analysis, Bayesian linear regression, Gaussian process regression, neural networks, etc. may also be used.

[0045] When multiple regression analysis is used, a regression equation such as that shown in equation (A) is formulated.

[0046] Y = a0 + a1X1 + a2X2 + ... + anXn (A), where an is the regression coefficient, X is the explanatory variable, Y is the response variable, and n is the number of explanatory variables used. Here, the response variable is the noise performance value, and the explanatory variables are the product characteristic values. Note that the response variable and the explanatory variables may be calculated based on actual measurements and transformed values ​​obtained by converting each value. Examples of transformations include logarithmic transformation, exponential transformation, power calculation, and taking a power root. Depending on the malfunction mechanism, a regression model using transformed values ​​can be expected to improve the accuracy of electromagnetic noise performance prediction. For example, by converting values ​​so that the variation of multiple samples is closer to a normal distribution, a model with high prediction accuracy can be generated. When using transformed values, the transformation process may be performed in advance by the measurement device 200, the data server 300, the data input unit 101, or other units other than the model generation unit 102.

[0047] The explanatory variables used in the regression model and their number n are arbitrary. When there are many explanatory variables, it is also possible to determine the priority of the explanatory variables using the variance ratio between the objective variable and the explanatory variables, the residual sum of squares, etc. The number of explanatory variables can also be determined based on a certain R2 score or using an evaluation function, etc. Examples of such evaluation functions include the Akaike Information Criterion (AIC) and the Bayesian Information Criterion (BIC). In this way, it is possible to improve the generative model by comparing an already generated model with a newly generated model.

[0048] The performance prediction unit 103 extracts the item names of product characteristic values ​​corresponding to one or more objective variables used in model generation by the model generation unit 102. Furthermore, the performance prediction unit 103 calculates a predicted value of electromagnetic noise performance using the generation model generated by the model generation unit 102.

[0049] Here, the performance prediction unit 103 may calculate a predicted value of electromagnetic noise performance for product characteristic values ​​of a sample (referred to as a first sample S1) whose electromagnetic noise performance is known and which was used in model generation by the model generation unit 102. When the product characteristic values ​​of the first sample S1 are used, the predicted value of electromagnetic noise performance can be used to confirm whether the generated model is normal or has the expected prediction accuracy.

[0050] Furthermore, the performance prediction unit 103 may calculate a predicted value of electromagnetic noise performance for product characteristic values ​​of a sample (second sample S2) that is different from the first sample S1 and whose electromagnetic noise performance is unknown. When the product characteristic values ​​of the second sample S2 are used, the predicted value of electromagnetic noise performance can be used for product verification and development.

[0051] The data output unit 104 outputs data in an externally usable format. The data may include the generative model itself. The data may include the names of one or more product characteristic values ​​used in model generation, all or part of the predicted values ​​of electromagnetic noise performance, or a graph visualizing these. The externally usable format may be, for example, a table data format such as CSV or XLSX, an image data format such as PNG, or a document data format such as WORD, PPTX, or PDF. Note that the output unit may be defined as a processing unit combining both the performance prediction unit 103 and the data output unit 104.

[0052] The internal storage unit 105 stores various variable values ​​in the memory 10 and performs input and output to and from each unit. Specifically, the variable values ​​include input noise performance values ​​and product characteristic values, objective variables, explanatory variables, a list of objective variables to be used, a generative model, calculated values ​​of evaluation functions, and predicted values ​​of electromagnetic noise performance. These variable values ​​may also be referred to as data related to the analysis of electromagnetic noise performance. The internal storage unit 105 may store data related to the analysis of electromagnetic noise performance for a large number of samples acquired by the data input unit 101 and to be extracted by the sample extraction unit 106 (described later).

[0053] The sample extraction unit 106 has a function of extracting samples of electrical or electronic products having electromagnetic noise performance values ​​that satisfy a given condition. For example, when a condition related to a predicted value, such as a predicted value or a range of predicted values, is provided by a user's operation of the analysis device 100, the sample extraction unit 106 can extract one or more samples that satisfy the condition by comparing the condition related to the predicted value with the predicted value calculated by the performance prediction unit 103. Furthermore, when a condition related to an actual measurement, such as an input noise performance value or a range of noise performance values, is provided by a user's operation of the analysis device 100, the sample extraction unit 106 can extract one or more samples that satisfy the condition by comparing the condition related to the actual measurement with the predicted value calculated by the performance prediction unit 103. The sample extraction unit 106 may also have a function of extracting samples that satisfy a given condition, such as the median, mode, or average of a histogram of the product characteristic values ​​extracted by the performance prediction unit 103, or samples shifted by a standard deviation from the average. Furthermore, in cases where key product characteristic values ​​are already known from design information for an electrical or electronic product or the above-described series of model generation and analysis for identical or similar electrical or electronic products, samples having electromagnetic noise performance that satisfies any conditions for the product characteristic values ​​may be extracted from the samples stored in memory 10 without performing the above-described series of model generation. In this way, sample extraction can be performed without newly obtaining multiple noise performance values ​​for the samples, leading to further reduction in development man-hours. When multiple performance characteristic values ​​are extracted, samples corresponding to each product characteristic value may be extracted.

[0054] The display unit 107 displays data to be displayed relating to the analysis processing of electromagnetic noise performance using the display device 30. The data to be displayed is a portion of the data processed by the data input unit 101, the model generation unit 102, the performance prediction unit 103, the data output unit 104, and the sample extraction unit 106.

[0055] <Processing by Analysis Device> An example of processing by an analysis method using multiple regression analysis focusing on variance ratios and regression residuals will be described below using the flowcharts in Figures 3 to 5. This processing may be realized by processor 20 of analysis device 100 executing an analysis program stored in memory 10.

[0056] First, FIG. 3 shows a process of extracting a specific explanatory variable from a plurality of explanatory variables and outputting the extracted explanatory variable or its generative model.

[0057] In S101, the data input unit 101 writes the noise characteristic value of the input first sample S1 into the memory 10 as the objective variable V1. The data input unit 101 writes the product characteristic values ​​of the input first sample S1 into the memory 10 as explanatory variables V2 and V3. Here, the initial values ​​for V2 and V3 are set to the same value. The data input unit 101 also initializes the higher-level item V4 of the variance ratio and the calculated value V6 of the evaluation function. After S101, the process proceeds to S102.

[0058] In S102, the model generation unit 102 reads out the dependent variable V1 and the explanatory variable V3 from the memory 10, and calculates the variance ratio of the explanatory variable V3 to the dependent variable V1. After the processing of S102, the process proceeds to S103.

[0059] In S103, the model generation unit 102 extracts the item with the largest variance ratio (hereinafter referred to as V4d) from the variance ratios calculated in S102. After the process of S103, the process proceeds to S104.

[0060] In S104, the model generation unit 102 formulates a regression equation V5 as a generation model capable of predicting electromagnetic noise performance, using V4d and the explanatory variable V3 specified by V4 stored in the memory 10. That is, each regression coefficient of the regression equation V5 is set. After processing S104, the process proceeds to S105.

[0061] In S105, the model generation unit 102 calculates the difference between the response variable V1 and the regression equation V5, that is, the regression residual. After the process of S105, the process proceeds to S106.

[0062] In S106, the model generation unit 102 evaluates the regression residual calculated in S105 using the above-mentioned evaluation function. The model generation unit 102 determines whether the calculated value V6d of the evaluation function based on the current regression residual is smaller than the calculated value V6 of the evaluation function stored in the memory 10. If the answer is Yes, the process proceeds to S107 to improve the model. If the answer is No, the model generation unit 102 determines that further improvement of the model is unnecessary and proceeds to S108.

[0063] In S107, the model generation unit 102 deletes the explanatory variable item V4d with the largest variance ratio from the explanatory variables V3 stored in the memory 10. Furthermore, the model generation unit 102 adds V4d to V4 stored in the memory 10. Then, the model generation unit 102 overwrites the calculated value V6 of the evaluation function stored in the memory 10 with V6d calculated in S106. After processing S107, the process returns to S102. That is, the results of the current model generation are reflected in the improvement of the next model, and processing continues.

[0064] In S108, the data output unit 104 outputs the higher-level items V4 of the variance ratio and the regression equation V5 formulated in the most recent processing of S104 to an external file. In S109 after processing of S108, the display unit 107 displays the numerical values ​​used in the processing, the results, etc. (for example, external files) on the screen of the display device 30, and the processing ends.

[0065] FIG. 4 shows the process of predicting electromagnetic noise performance, including the processes of S101 to S108.

[0066] In S201, the data input unit 101 writes the product characteristic value P1 of the product to be predicted into the memory 10. As described above, the product to be predicted may be the first sample S1 used in the model generation (S101 to S108), or may be a second sample S2 different from the first sample S1. After processing S201, the process proceeds to S202.

[0067] S202 is a model generation process similar to S101 to S108. However, in S108, the top item V4 of the variance ratio and the regression equation V5 are written to the memory 10 for subsequent processing. After processing S202, the process proceeds to S203.

[0068] In S203, the performance prediction unit 103 applies the higher-level items V4 of the variance ratio and the regression equation V5 output in S202 to the product characteristic value P1 to calculate a predicted value P2 of the electromagnetic noise performance of the prediction target product. After processing S203, the process proceeds to S204.

[0069] In S204, the data output unit 104 outputs the top items V4 of the variance ratio, the regression formula V5, and the predicted value P2 to an external file as data related to the analysis. In S205 after the processing of S204, the display unit 107 displays the numerical values ​​used in the processing, the results, etc. (e.g., external files) on the screen of the display device 30, and then the processing ends. Note that the order of the processing of S201 and the processing of S202 may be reversed.

[0070] While the flowcharts in FIGS. 3 and 4 (S101 to S108, S201 to S205) have described the model generation process and the electromagnetic noise performance prediction process for one frequency, the flowchart in FIG. 5 (S301 to S309) will describe the process for multiple frequencies, i.e., the entire analysis process.

[0071] In S301, the data input unit 101 writes V1 to V6 as initial values ​​into the memory 10, similar to S101 and S201. Here, explanatory variables V2, V3, etc. must be prepared for each frequency to be analyzed. In S301, the product characteristic values ​​of the first sample S1 are used. After processing S301, the process proceeds to S302.

[0072] Steps S302 to S304 are performed for one frequency. In step S302, the model generation unit 102 reads out the response variable V1 and explanatory variables V2 and V3 for the frequency being analyzed from the memory 10. After step S302, the process proceeds to step S303.

[0073] In S303, the model generation unit 102 executes the same model generation process as in S101 to S108 for the frequency under analysis using the variables V1 and V2 read out in S302. After the process of S303, the process proceeds to S304.

[0074] In S304, the model generation unit 102 adds the top item V4 of the variance ratio to the top item P3 of the variance ratio at angular frequency as the calculated value of the frequency under analysis, and adds the regression equation V5 to the regression equation P4 at each frequency. After processing S304, the process proceeds to S305.

[0075] In S305, the model generation unit 102 determines whether model generation for all frequencies to be analyzed has been completed. If the answer is No, the process proceeds to S306. If the answer is Yes, the model generation for each frequency is terminated and the process proceeds to S307.

[0076] In S306, the model generating unit 102 changes the frequency. After the process of S306, the process returns to S302 and the processes of S302 to S304 are performed on the next frequency.

[0077] In S307, the performance prediction unit 103 applies the higher-order item P3 of the variance ratio and the regression equation P4 to the product characteristic value P1 of the prediction target product to calculate a prediction value P2 of the prediction target product for all frequencies. The prediction target product may be the first sample S1 as in S201, or may be the second sample S2. After processing S307, the process proceeds to S308.

[0078] In S308, the data output unit 104 outputs the predicted value P2, the top items of the variance ratio P3, and the regression equation P4 to an external file as data related to the analysis. In S309 after the processing of S308, the display unit 107 displays the numerical values ​​used in the processing, the results, etc. (for example, the external file) on the screen of the display device 30, and the processing ends.

[0079] As described above, in order to predict electromagnetic noise performance, actual measured or calculated values ​​of electromagnetic noise performance measurements imposed on electrical or electronic products and data on variations in product characteristic values ​​are used.

[0080] An example of a product characteristic value is a measurement of the circuit characteristics of each individual product for the purpose of improving yield and eliminating products with abnormal performance in the manufacture of electrical or electronic products. These measurements are necessary for product manufacturing and are therefore available at low cost regardless of the implementation of the present disclosure. In semiconductor circuit manufacturing, WAT is known as an example of such a measurement.

[0081] These test items are generally measured in a steady state, and therefore, although they can confirm basic circuit characteristics, they do not directly measure electromagnetic noise performance containing frequency information. In this embodiment, we have discovered that electromagnetic noise containing frequency information can be estimated by using the results of measuring these basic circuit characteristics on a large number of samples and analyzing their variations. By using this method, it is possible to predict electromagnetic noise performance without the need to build a physical model required for electromagnetic noise performance prediction. Furthermore, if the basic characteristics of a circuit that determine electromagnetic noise performance can be identified through statistical analysis, etc., design support such as changing the design policy of that circuit portion becomes possible.

[0082] According to the first embodiment described above, a model is generated using the first sample S1, and the generated model is used to calculate a predicted value for the electromagnetic noise performance of the second sample S2. That is, it is possible to easily predict the electromagnetic noise performance of the second sample S2, which is different from the first sample S1, from information about the first sample S1.

[0083] The generated model uses the noise performance value of the electromagnetic noise performance as the objective variable and the product characteristic value of the first sample S1 as the explanatory variable. A predicted value can be obtained by inputting the product characteristic value of the second sample S2 into this model instead of the first sample S1. In other words, by analyzing the basic characteristics that determine the electromagnetic noise performance of the second sample S2, the cause of malfunction can be easily identified. As a result, the number of steps required for product analysis can be reduced.

[0084] Furthermore, according to the first embodiment, a model is generated using the first sample S1, and the product characteristic values ​​of the first sample S1 are input into the generated model to calculate a predicted value for electromagnetic noise performance. This makes it possible to easily verify whether the predictions of the generated model are normal. As a result, it is possible to suppress an increase in the number of steps required for product analysis.

[0085] Furthermore, according to the first embodiment, the product characteristic values ​​include circuit characteristic values ​​of circuits included in the sample. Since the circuit characteristic values ​​are used as explanatory variables, it becomes easier to identify the cause of a malfunction by analyzing the cause related to the problematic circuit characteristic values.

[0086] Furthermore, according to the first embodiment, at least one of the variance ratio and the sum of squared residuals of the product characteristic value relative to the noise performance value is calculated, and a regression equation for the electromagnetic noise performance is generated using at least one of the variance ratio and the sum of squared residuals, allowing the variation between samples to be reflected in the regression equation.

[0087] Furthermore, according to the first embodiment, the model is improved by comparing an already generated model with a newly generated model. By improving the model, it is possible to increase the prediction accuracy of electromagnetic noise performance.

[0088] Furthermore, according to the first embodiment, individual models are generated for multiple frequencies, and predicted values ​​of electromagnetic noise performance for each of the multiple frequencies are calculated using the models generated for each of the multiple frequencies. By predicting electromagnetic noise performance for multiple frequencies, it becomes easier to identify the cause of malfunction.

[0089] Furthermore, according to the first embodiment, it is possible to extract samples having noise performance values ​​that satisfy arbitrary conditions from among multiple samples for which the predicted values ​​are to be calculated, based on arbitrary conditions and predicted values ​​of electromagnetic noise performance. This function makes it easy to identify samples that should be prioritized for verification, and prevents an increase in the number of steps required for product analysis.

[0090] Furthermore, according to the first embodiment, the data relating to the analysis process can be displayed, and the user can easily check the analysis results by visually checking the display.

[0091] Second Embodiment The second embodiment is a modification of the first embodiment. The second embodiment will be described focusing on the differences from the first embodiment.

[0092] The method of calculating the variance ratio and the method of determining the explanatory variables used in Equation (A), which is the regression equation shown in the first embodiment, are, for example, as follows. Here, a matrix is ​​prepared in which the noise performance values, etc. of m samples are taken as vertical vectors and the vectors of l product characteristic values ​​corresponding to each sample are combined in the horizontal direction. For this matrix, the covariance matrix S is as shown in Equation 1 below. Here, i is an integer that satisfies 0≦i≦l, and j is an integer that satisfies 0≦j≦l.

[0093] Any element Sij in the above satisfies the following equation 2. Note that Xik represents the kth element of Xi (1≦k≦m). Xi with a bar represents the average of the elements in Xi. When i=j, Sii represents the variance as shown in equation 3. Here, Sij=Sji. Si0 or S0j represents the covariance between the objective variable and the ith or jth explanatory variable, respectively. In the following equations, Si0 may be changed to S0j, and other i and j may be interchanged.

[0094] Here, SRi, Sei corresponding to the i-th objective variable, and Ri, which is its variance ratio, are defined as in the following Equations 4, 5, and 6.

[0095] In the above method, the covariance matrix shown in Equation 1 is used as an example for the purpose of explanation, but the covariance coefficient shown in Equation 2 may be calculated directly for each objective variable and each explanatory variable. In that case, formulating Equation 1 may be unnecessary. Furthermore, Equation 6 may be obtained by multiplying or dividing various values ​​such as the degree of freedom. Xi, which maximizes Ri, may be added as an explanatory variable to be used in Equation (A).

[0096] Next, a method for calculating the sum of squared residuals used in Equation (A), the regression equation shown in the first embodiment, will be described. A regression equation is prepared in which an arbitrary (n+1)th explanatory variable Xi is added to n explanatory variables already selected using the aforementioned method for determining a response variable based on a variance ratio or this method for calculating the sum of squared residuals. Xi may be selected by calculating the sum of squared residuals from all unselected explanatory variables, as shown below. Using Equation (A), a predicted value Y obtained using n explanatory variables is used, and a predicted value Ypred obtained in the same manner as Equation (A) using n+1 explanatory variables obtained by adding Xi to the explanatory variables is used. Sei, VRi, and Ri can be obtained as shown in Equation 7, Equation 8, and Equation 9 below.

[0097] Here, Sen is a value already defined by Equation 5 or Equation 7 through the process of selecting the nth explanatory variable. In Equation 8, VRi is defined as the difference between the aforementioned Sen and Sei when Xi is added as the (n+1)th explanatory variable. Note that Equation 9 may be obtained by multiplying or dividing various values ​​such as the degrees of freedom. Xi that maximizes Ri may be added as the explanatory variable used in Equation (A). In that case, Sen may be redefined by Equation 7 corresponding to the newly added Xi and used to determine the (n+2)th explanatory variable. Note that Equations 1 to 9 described above may also be rephrased as Equations 1 to 9.

[0098] 6 to 13, the third embodiment is a modification of the first embodiment. The third embodiment will be described focusing on the differences from the first embodiment.

[0099] In the third embodiment, the WAT evaluation results for an analog sensor product with a differential amplifier circuit are used, with the forward wave power at the time of malfunction obtained using the IEC 62132-4 DPI method as the objective variable and the explanatory variable being the WAT evaluation item. Figure 6 shows the occurrence rates of the top three explanatory variables obtained by analyzing the variance ratio and sum of squared residuals at each frequency, grouped by WAT evaluation item. These results indicate that for the target product, the WAT evaluation items related to various frequencies and the WAT evaluation item related to amplification factor contribute significantly to the immunity performance. In this way, the cause of the malfunction, such as the malfunction mechanism, can be inferred simply by analyzing the evaluation results, significantly reducing the labor required for product analysis.

[0100] By using the oscillation frequency and amplitude gain as representative WAT evaluation items with high contributions as described above, samples that meet any desired conditions can be automatically selected. Examples of extracted samples are shown in Table 1.

[0101] The results of evaluating these samples using the IEC 62132-4 DPI method are shown in Figures 7 and 8. Among samples A1, A2, and A3, which have the same oscillation frequency but different amplitude gains, the maximum difference between each sample is approximately 2 dB. In contrast, among samples A1, B1, and C1, which have the same amplitude gain but different oscillation frequencies, there is a difference of approximately 8 dB. This shows that immunity performance is significantly affected by oscillation frequency. From the above, it can be seen that by performing the proposed electromagnetic noise performance analysis, it can be inferred that the design items that contribute to malfunction are oscillation frequency and amplitude gain, enabling the selection of appropriate evaluation samples.

[0102] To perform these analyses, it is necessary to obtain product characteristic values ​​for multiple samples, which can be obtained by measuring each characteristic value of the product.

[0103] Furthermore, in order to perform the analysis more easily or effectively, the semiconductor device as a sample may be configured as shown in Fig. 9. The semiconductor device 500 in Fig. 9 is configured in the shape of a chip having a rectangular outer periphery by cutting a portion of a semiconductor wafer corresponding to a die.

[0104] The semiconductor device 500 includes a substrate 510, a product circuit 520, a measurement circuit 530, an output switching circuit 540, and multiple pads (PADs) 550. The substrate 510 is formed in a plate shape primarily made of silicon or the like. The product circuit 520 and the measurement circuit 530 are arranged to divide the central area of ​​the semiconductor device 500, and coexist on the same chip. The product circuit 520 is a circuit mounted in a block on the substrate 510 to perform the product's functions. The product circuit 520 is configured by combining multiple types of electronic components according to their functions.

[0105] The measurement circuit 530 is a circuit for measuring product characteristic values, mounted in a block on the substrate 510 so as to be separated from the product circuit 520. The measurement circuit 530 is configured with fewer electronic components than a product circuit solely for measuring product characteristic values. For example, the measurement circuit 530 may be configured with only an oscillator circuit used for measurement. For example, the measurement circuit 530 may be used as a so-called TEG (test element group).

[0106] The measuring circuit 530 may be capable of measuring a plurality of types of product characteristic values. For example, the measuring circuit 530 may be capable of measuring one or a plurality of types of product measurement values ​​among the oscillation frequency, the frequencies of various control signals generated from the oscillation frequency, measurement values ​​related to time such as the oscillation frequency and the pulse widths of various control signals, output values ​​of an ADC or a DAC, output values ​​of an amplifier circuit, gain of an amplifier circuit, output values ​​of a constant voltage or constant current source circuit, output values ​​of various digital circuits, various resistance values, various capacitance values, and output values ​​of filter circuits including electromagnetic noise filters.

[0107] The multiple pads 550 are input / output terminals arranged around the entire periphery of the semiconductor device 500. The multiple pads 550 may be divided into pads dedicated to the product circuit 520 and pads dedicated to the measurement circuit 530. Alternatively, at least some of the pads 550 may be shared between the product circuit 520 and the measurement circuit 530. Each pad 550 functions as an output terminal that outputs the measurement results of the product characteristic values ​​to the measurement circuit 530.

[0108] The output switching circuit 540 is a circuit that is disposed separately from the measurement circuit 530, for example, in an area between the measurement circuit 530 and a pad 550 that functions as an output terminal for the measurement result, and switches the type of product characteristic value output to the pad 550. The output switching circuit 540 may be configured using, for example, an analog switch, a multiplexer, or the like. The presence of such an output switching circuit 540 makes it possible to measure many types of product characteristic values ​​without measuring a large number of pads 550. In other words, by suppressing the probe movement time required to change the pad 550 to be connected during measurement, the evaluation time can be reduced. Note that the output switching circuit 540 may be integrated with the measurement circuit 530 without being separated from it in a separate area.

[0109] Alternatively, a semiconductor device 600 may be configured as shown in FIG. 10 . The semiconductor device 600 includes a substrate 610, a measurement circuit 630, an output switching circuit 640, and a plurality of pads 650. That is, the example of FIG. 10 differs from the example of FIG. 9 in that no product circuit is mounted thereon and the device is dedicated to measurement. The measurement circuit 630 has the same function as the example of FIG. 9 , but can be mounted in the entire central area of ​​the semiconductor device 600. This allows measurement of a wide variety of product characteristic values. The output switching circuit 640 may be arranged to surround the entire circumference of the measurement circuit 630 to accommodate a higher performance measurement circuit 630, and may be capable of switching between a wide variety of product characteristic values.

[0110] 11 , the semiconductor wafer SCW may be configured so that the measurement circuit is not disposed inside the semiconductor device, but rather the measurement circuit 630 is disposed outside. The semiconductor wafer SCW includes a substrate 710, a plurality of dies 700, and a plurality of measurement circuits 730. The substrate 710 has a partially circular outer periphery including a linearly extending flat zone, and is formed in a flat plate shape primarily made of silicon or the like. Note that the semiconductor wafer SCW may have a notch instead of the flat zone.

[0111] The die 700 is a unit that will be cut into chips in a later process. Each die 700 is configured so that a product circuit can be mounted therein. The multiple dies are arranged in a two-dimensional grid pattern, spaced apart from each other by regions called scribe lines SL. The arrangement direction of the multiple dies 700 may be substantially perpendicular or substantially horizontal to the extension direction of the flat zone, or may be oblique at an acute or obtuse angle to the extension direction.

[0112] The plurality of measurement circuits 730 are circuits for measuring product characteristic values, similar to the measurement circuits 530 and 630 in FIGS. 9 and 10 . The plurality of measurement circuits 730 are arranged on the scribe line SL. This arrangement eliminates restrictions on the area within the die 700 for mounting the product circuits. As shown enlarged in FIG. 12 , the plurality of measurement circuits 730 form individual correspondences with the plurality of dies 700. For example, the individual correspondences are one-to-one relationships between the dies 700 and the measurement circuits 730. For this reason, the same number of measurement circuits 730 as the dies 700 are provided, and the measurement circuits 730 are arranged adjacent to one side of the die 700 with which they individually correspond one-to-one.

[0113] 13, the individual correspondence is a multiple-to-one (e.g., four-to-one) relationship between the dies 800 and the measurement circuits 830. In the example of FIG. 13, the measurement circuits 830 are arranged adjacent to the vertices of the four dies 800 arranged in a grid pattern, so as to be surrounded by the four dies 800. In this case, the number of measurement circuits 830 may be approximately one-fourth of that shown in FIG. 12. The individual correspondence may also be two-to-one.

[0114] In this way, by configuring the measurement circuits 730, 830 adjacent to all of the multiple dies 700, 800 on the semiconductor wafer SCW, it is possible to obtain the variation in product characteristic values ​​for each die 700, 800 on the semiconductor wafer SCW. Analysis of this variation is expected to provide an improvement indicator for the semiconductor process. Such improvements can realize product designs that are less prone to malfunctions and can also reduce the amount of work required for analysis and countermeasures.

[0115] Furthermore, by carrying out evaluation and analysis of the measuring circuits 530, 630, 730, and 830 for each manufacturing line before product design, the measuring circuits 530, 630, 730, and 830 can be utilized in noise countermeasure design for product design.

[0116] According to the third embodiment described above, the semiconductor devices 500 and 600 are provided with measuring circuits 530 and 630, and measurement results of product characteristic values ​​can be obtained from the measuring circuits 530 and 630 via pads 550 and 650 serving as output terminals, thereby suppressing an increase in the number of steps required for product analysis.

[0117] Furthermore, according to the third embodiment, it is possible to acquire the measurement results of the product characteristic values ​​through the measurement circuits 730, 830 that have individual correspondences with the multiple dies 700, 800, and therefore it is possible to suppress an increase in the amount of work required for analyzing the product, including the impact on each of the dies 700, 800 that make up the semiconductor wafer SCW.

[0118] (Other Embodiments) Although multiple embodiments have been described above, the present disclosure should not be construed as being limited to these embodiments, and can be applied to various embodiments within the scope of the gist of the present disclosure.

[0119] In another embodiment, the system 1 may be configured not to include the data server 300 and the terminal 310. The data measured by the measuring device 200 may be stored in the memory 10 of the analysis device 100.

[0120] In another embodiment, the numerical values, results, etc. used in the processing may not be displayed in the analysis device 100. In this case, the analysis device 100 may be configured not to include the display device 30. The analysis device 100 may output the analysis results to an external file so that they can be processed and displayed by another device.

[0121] In another embodiment, the electromagnetic immunity performance may be the power value that leads to malfunction due to electromagnetic noise, i.e., the malfunction tolerance value, obtained through evaluation in accordance with evaluation methods such as international standards, etc. The electromagnetic immunity performance may also be the amount of output fluctuation when a predetermined power or the like is applied.

[0122] In another embodiment, the measured value of electromagnetic noise performance may be a value obtained by combining evaluations of actual measurements of electromagnetic noise performance in accordance with multiple types of test methods specified in international standards, etc. For example, an analysis may be performed that combines the evaluation results of conducted emissions and radiated emissions, or the evaluation results of conducted immunity and radiated immunity. These are effective for estimating the emission generation and propagation mechanisms and the immunity malfunction mechanisms.

[0123] In another embodiment, the test method may include applying different disturbance waves to the same test system. For example, the test method may include applying a disturbance wave modulation method such as CW (Continuous Wave) modulation, AM (Amplitude Modulation) modulation, or FM (Frequency Modulation), and may also include using a noise model such as Additive White Gaussian Noise (AWGN).

[0124] Furthermore, the test system may be changed to perform evaluations using different noise propagation modes. For example, the test method may include injecting the various disturbance waves mentioned above into the product under evaluation using an evaluation system with different propagation modes, such as common mode and differential mode. Combining these multiple evaluation conditions is expected to improve the efficiency of estimating immunity malfunction mechanisms.

[0125] In another embodiment, the evaluation results of the circuit characteristic values ​​may include at least one of the following: oscillation frequency, frequencies of various control signals generated from the oscillation frequency, pulse widths of various control signals, rise times, fall times, and other time-related measurements; their duty ratios; output values ​​of ADCs and DACs; output values ​​of amplifier circuits; gains of amplifier circuits; output values ​​of constant voltage or constant current source circuits; output values ​​of various digital circuits; various resistance values; various capacitance values; and output values ​​of filter circuits including electromagnetic noise filters.

[0126] In another embodiment, the predicted value of the electromagnetic noise performance calculated by the performance prediction unit 103 may be used as a behavior model of the IC. Specifically, models such as the ICIM-CI model (IEC 62433-4) and the ICIM-CPI (IEC 62433-6) are conceivable.

[0127] In another embodiment, the predicted value of the electromagnetic noise performance may be used as the condition used in the sample extraction by the sample extraction unit 106. Furthermore, one or more explanatory variables, which are the item names of the product characteristic values ​​extracted by the performance prediction unit 103, may be used as the condition.

[0128] In another embodiment, the range of predicted values ​​used as a condition for sample extraction by the sample extraction unit 106 may be a range in which the difference between the actual measured value or predicted value of a specific sample and the predicted value of a sample other than the specific sample is less than a certain value.

[0129] The controller and methods described herein may be implemented by a special-purpose computer comprising a processor programmed to perform one or more functions embodied in a computer program. Alternatively, the apparatus and methods described herein may be implemented by special-purpose hardware logic circuitry. Alternatively, the apparatus and methods described herein may be implemented by one or more special-purpose computers comprising a processor executing a computer program in combination with one or more hardware logic circuits. Furthermore, the computer program may be stored as instructions executed by a computer on a computer-readable non-transitory storage medium.

[0130] (Disclosure of Technical Ideas) This specification discloses multiple technical ideas described in the following multiple clauses. Some clauses may be described in a multiple dependent form, where the subsequent clause alternatively cites the preceding clause. These multiple dependent clauses define multiple technical ideas.

[0131] <Technical Idea 1> An analysis device configured to execute an analysis process of electromagnetic noise performance of a sample, the analysis device comprising: a model generation unit (102) that generates a model in which a noise performance value of the electromagnetic noise performance of a first sample (S1) is used as a dependent variable and a plurality of product characteristic values ​​of the first sample are used as explanatory variables; and an output unit (103, 104) that inputs a plurality of product characteristic values ​​of a second sample (S2) different from the first sample that correspond to the plurality of product characteristic values ​​of the first sample as the explanatory variables of the model, and outputs data related to the analysis of the electromagnetic noise performance of the second sample using the model.

[0132] <Technical Idea 2> An analysis device configured to execute an analysis process of electromagnetic noise performance of a sample, the analysis device comprising: a model generation unit (102) that generates a model in which a noise performance value of the electromagnetic noise performance of a first sample (S1) is a dependent variable and a plurality of product characteristic values ​​of the first sample are explanatory variables; and an output unit (103, 104) that inputs the plurality of product characteristic values ​​of the first sample as the explanatory variables of the model, and outputs data related to the analysis of the electromagnetic noise performance of the first sample using the model.

[0133] <Technical Concept 3> The analytical device according to Technical Concept 1 or 2, wherein the product characteristic value includes a circuit characteristic value of a circuit included in the sample.

[0134] <Technical Idea 4> The analysis device according to any one of Technical Ideas 1 to 3, wherein the model generation unit calculates at least one of a variance ratio and a sum of squared residuals of the product characteristic value with respect to the noise performance value, and generates a regression equation for the electromagnetic noise performance using the at least one of the variance ratio and the sum of squared residuals of the product characteristic value with respect to the noise performance value.

[0135] <Technical Idea 5> The analysis device according to any one of Technical Ideas 1 to 4, wherein the model generation unit compares the model that has already been generated with the model that has been newly generated, and improves the model.

[0136] <Technical Idea 6> The analysis device according to any one of Technical Ideas 1 to 5, wherein the model generation unit generates the models individually for a plurality of frequencies, and the output unit calculates predicted values ​​of the electromagnetic noise performance for each of the plurality of frequencies using the models individually generated for the plurality of frequencies.

[0137] <Technical Idea 7> The analysis device according to any one of Technical Ideas 1 to 6, further comprising a sample extraction unit (106) that extracts, based on an arbitrary condition, from a plurality of samples to be output by the output unit, the sample having the noise performance value that satisfies the arbitrary condition.

[0138] <Technical Concept 8> The analytical device according to any one of Technical Concepts 1 to 7, further comprising a display unit (107) configured to be able to display data related to the analytical processing.

[0139] <Technical Idea 9> The analytical device according to any one of Technical Ideas 1 to 8, wherein a measuring device (200) that measures the noise performance value and the product characteristic value is communicably connected to a data server (300) that accumulates measured data, and the noise performance value and the product characteristic value are acquired via the data server.

Claims

1. An analytical device configured to execute an analysis process of the electromagnetic noise performance of a sample, comprising: a model generation unit (102) that generates a model in which a noise performance value of the electromagnetic noise performance of a first sample (S1) is used as a response variable and multiple product characteristic values ​​of the first sample are used as explanatory variables; and an output unit (103, 104) that inputs multiple product characteristic values ​​of a second sample (S2) different from the first sample that correspond to the multiple product characteristic values ​​of the first sample as the explanatory variables of the model, and outputs data related to the analysis of the electromagnetic noise performance of the second sample using the model.

2. An analytical device configured to execute an analysis process of the electromagnetic noise performance of a sample, comprising: a model generation unit (102) that generates a model in which a noise performance value of the electromagnetic noise performance of a first sample (S1) is a dependent variable and multiple product characteristic values ​​of the first sample are explanatory variables; and an output unit (103, 104) that inputs the multiple product characteristic values ​​of the first sample into the explanatory variables of the model, and outputs data related to the analysis of the electromagnetic noise performance of the first sample using the model.

3. The analytical device according to claim 1 or 2, wherein the product characteristic values ​​include circuit characteristic values ​​of a circuit contained in the sample.

4. The analysis device according to claim 1 or 2, wherein the model generation unit calculates at least one of the variance ratio and the sum of squared residuals of the product characteristic value relative to the noise performance value, and generates a regression equation for the electromagnetic noise performance using the at least one of them.

5. The analysis device according to claim 1 or 2, wherein the model generation unit compares the already generated model with the newly generated model and improves the model.

6. The analysis device according to claim 1 or 2, wherein the model generation unit generates the models individually for a plurality of frequencies, and the output unit calculates predicted values ​​of the electromagnetic noise performance for each of the plurality of frequencies using the models individually generated for the plurality of frequencies.

7. The analysis device of claim 1 or 2, further comprising a sample extraction unit (106) that extracts, based on an arbitrary condition, from a plurality of samples to be output by the output unit, the sample having the noise performance value that satisfies the arbitrary condition.

8. The analysis device according to claim 1 or 2, further comprising a display unit (107) configured to be able to display data relating to the analysis process.

9. An analytical device as described in claim 1 or 2, wherein a measuring device (200) that measures the noise performance value and the product characteristic value is communicatively connected to a data server (300) that accumulates the measured data, and the noise performance value and the product characteristic value are obtained via the data server.

10. An analysis program configured to execute an analysis process of the electromagnetic noise performance of a sample using at least one processor (20), the analysis program configured to cause the at least one processor to execute the following: generate a model in which a noise performance value of the electromagnetic noise performance of a first sample (S1) is a response variable and multiple product characteristic values ​​of the first sample are explanatory variables; input multiple product characteristic values ​​of a second sample (S2) different from the first sample that correspond to the multiple product characteristic values ​​of the first sample as the explanatory variables of the model; and output data related to the analysis of the electromagnetic noise performance of the second sample using the model.

11. A method for generating data related to the electromagnetic noise performance of a sample, comprising: generating a model in which a noise performance value of the electromagnetic noise performance of a first sample (S1) is used as a dependent variable and multiple product characteristic values ​​of the first sample are used as explanatory variables; and inputting multiple product characteristic values ​​of a second sample (S2) different from the first sample that correspond to the multiple product characteristic values ​​of the first sample into the explanatory variables of the model, and generating data related to an analysis of the electromagnetic noise performance of the second sample using the model.

12. An analytical device configured to perform an analysis process of the electromagnetic noise performance of a sample, comprising: a memory (10) that stores data related to the analysis of the electromagnetic noise performance of the sample; and a sample extraction unit (106) that extracts samples having electromagnetic noise performance that satisfies any given condition from among the samples stored in the memory.

13. A semiconductor device configured to enable measurements of product characteristic values ​​for analyzing electromagnetic noise performance, comprising: a substrate (510, 610) for mounting a circuit; a measurement circuit (530, 630) for measuring the product characteristic values, which is different from a product circuit, which is a circuit for performing the function of the product; and an output terminal (550, 650) for outputting the measurement results by the measurement circuit.

14. The semiconductor device according to claim 13, further comprising an output switching circuit (540, 640) for switching the type of the product characteristic value output to the output terminal.

15. The semiconductor device according to claim 13 or 14, wherein the substrate is mounted with the measurement circuit out of the product circuit and the measurement circuit.

16. The semiconductor device according to claim 13 or 14, wherein both the product circuit and the measurement circuit are mounted on the substrate so as to divide the substrate area.

17. A semiconductor wafer configured to enable measurements of product characteristic values ​​for analyzing electromagnetic noise performance, comprising: a plurality of dies (700, 800) each configured to be capable of mounting a product circuit, which is a circuit for performing a product function, and arranged so as to be spaced apart from each other across a scribe area (SL); and a plurality of measurement circuits (730, 830), which are circuits for measuring the product characteristic values, and which are arranged in the scribe area so as to form individual corresponding relationships with the plurality of dies.

18. The semiconductor wafer according to claim 17, wherein the individual correspondence is a one-to-one correspondence between the die and the measurement circuitry.

19. The semiconductor wafer according to claim 17, wherein the individual correspondence is a multiple-to-one correspondence between the dies and the measurement circuits.

Images