JP2015220379A - Integrated circuit device latent defect inspection apparatus, method, and program - Google Patents

Abstract

A device for detecting an accurate outlier regardless of variations in the distance of each point in a scatter diagram is provided. A correlation straight line calculation unit 2 creates a scatter diagram of measurement values of integrated circuit devices designated in advance among the measurement values of the integrated circuit devices stored in the measurement value storage unit 1. A correlation straight line is calculated. The distance calculation unit 3 calculates the distance between the correlation line calculated by the correlation line calculation unit 2 and each point of the scatter diagram. Based on the distance calculated by the distance calculation unit 3, the distribution analysis unit 4 specifies a nearby measurement value for the measurement value of the specified integrated circuit device, and the average value and standard deviation of the distribution of the distance of this nearby measurement value To obtain the Z value of the measured value of the specified integrated circuit device. Further, the distribution analysis unit 4 determines whether or not the measurement value is an outlier based on the Z value, and identifies an integrated circuit device that exhibits the outlier. [Selection] Figure 1

Description

  The present invention relates to an integrated circuit device latent defect inspection apparatus, method and program, and more particularly to an apparatus, method and program for analyzing characteristic measurement values of an integrated circuit device.

  Japanese Patent No. 5080526 (Patent Document 1) discloses a technique for identifying outliers in test data for a plurality of components manufactured on a wafer in a data analysis method and apparatus suitable for semiconductor manufacturing. Is described.

Japanese Patent No. 5080526

  In the technique of Patent Document 1, in order to identify outliers in test data, a correlation chart (scatter diagram) of test data of two different tests is created, and a certain distance from the best fit line (regression line) is created. A method of identifying an existing value as an outlier is used. The problem with this method is that the degree of variation in the distance from the regression line of each point constituting the scatter diagram is not uniform in the scatter diagram but varies depending on the location.

  In the conventional technique such as Patent Document 1, a constant distance from the regression line is used as a standard value regardless of the position in the scatter diagram, and a value exceeding this standard value is determined as an outlier. It is determined that there is a potential defect in the device that indicates However, since the degree of variation in the distance of each point in the scatter diagram varies depending on the position in the scatter diagram, it is difficult to correctly determine an outlier when a uniform standard value is provided for the entire scatter diagram. Therefore, it is necessary to detect an accurate outlier regardless of variations in the distance of each point in the scatter diagram.

  Other problems and novel features will become apparent from the description of the specification and the accompanying drawings.

  (1) An integrated circuit device latent defect inspection apparatus according to an embodiment includes a storage unit, a first calculation unit, a second calculation unit, an analysis unit, a control unit, and a storage unit. The storage unit stores measurement values of the integrated circuit devices. The first calculation unit creates a scatter diagram of the measurement values of the integrated circuit device designated in advance among the measurement values of the integrated circuit devices stored in the storage unit, and calculates a correlation line or a correlation curve from the scatter diagram. To do. The second calculation unit calculates the distance between the correlation line or the correlation curve calculated by the first calculation unit and each point of the scatter diagram. The analysis unit specifies a first measurement value with respect to the measurement value of the designated integrated circuit device based on the distance calculated by the second calculation unit, and an average value and a standard deviation of the distance distribution of the first measurement value To obtain the Z value of the measured value of the specified integrated circuit device. Furthermore, the analysis unit determines whether or not the measured value is an outlier based on the Z value, and identifies an integrated circuit device that exhibits the outlier. The control unit controls the entire apparatus including the storage unit, the first calculation unit, the second calculation unit, and the analysis unit. The storage unit stores a control program and data necessary for controlling the entire apparatus.

  (2) An integrated circuit device latent defect inspection method according to an embodiment includes a first step, a second step, a third step, a fourth step, and a fifth step as information processing steps by a computer system. And have. In the first step, among the measured values of the integrated circuit devices, the measured values of the integrated circuit device designated in advance are collected from the storage unit. In the second step, a scatter diagram of the measurement values collected in the first step is created. In the third step, a correlation line or a correlation curve is calculated from the scatter diagram created in the second step. In the fourth step, the distance between the correlation line or correlation curve calculated in the third step and each point of the scatter diagram is calculated. In the fifth step, based on the distance calculated in the fourth step, the first measurement value for the measurement value of the designated integrated circuit device is specified, and the average value and standard deviation of the distance distribution of the first measurement value are specified. To obtain the Z value of the measured value of the specified integrated circuit device. Further, in the fifth step, it is determined whether or not the measured value is an outlier based on the Z value, and an integrated circuit device exhibiting the outlier is specified.

  (3) An integrated circuit device latent defect inspection program according to an embodiment includes a first routine, a second routine, a third routine, a fourth routine, and a fifth routine that are executed by a computer system. . The processes in the first to fifth routines correspond to the first to fifth steps of the integrated circuit device latent defect inspection method.

  According to one embodiment, an accurate outlier can be detected regardless of variations in the distance of each point in the scatter diagram.

It is a block diagram which shows an example of a structure of the latent defect inspection apparatus of the integrated circuit device in Embodiment 1 of this invention. 3 is a flowchart showing an example of an operation of a latent defect inspection method in the integrated circuit device latent defect inspection apparatus of FIG. 1. FIG. 3 is an explanatory diagram for explaining creation of a scatter diagram in the latent defect inspection method for the integrated circuit device of FIG. 2. FIG. 3 is an explanatory diagram for explaining a method of calculating a distance between a correlation line and each point of a scatter diagram in the latent defect inspection method for the integrated circuit device of FIG. 2. FIG. 3 is an explanatory diagram for explaining a correlation line and a standard value in a scatter diagram in the latent defect inspection method for the integrated circuit device of FIG. 2. 3 is a flowchart for explaining recursive calculation in calculating an average value and a standard deviation with respect to a distribution of distances in the latent defect inspection method for an integrated circuit device in FIG. 2. It is a block diagram which shows an example of a structure of the latent defect inspection apparatus of the integrated circuit device in Embodiment 2 of this invention. It is a flowchart which shows an example of operation | movement of the latent defect inspection method in the latent defect inspection apparatus of the integrated circuit device of FIG. FIG. 9 is an explanatory diagram for explaining a method of calculating a distance between a correlation curve and each point of a scatter diagram in the latent defect inspection method for the integrated circuit device of FIG. 8. FIG. 9 is an explanatory diagram for explaining a definition of a distance between a correlation curve and each point of a scatter diagram in the latent defect inspection method for the integrated circuit device of FIG. 8. 9 is a flowchart for explaining calculation of a correlation curve in the latent defect inspection method for an integrated circuit device of FIG. 8. 9 is a flowchart for explaining recursive calculation in calculating an average value and a standard deviation with respect to a distribution of distances in the latent defect inspection method for an integrated circuit device of FIG. 8. FIG. 8 is an explanatory diagram showing a scatter diagram and a correlation line for explaining the effect in the latent defect inspection apparatus for an integrated circuit device of FIG. 7. FIG. 8 is an explanatory diagram showing a scatter diagram and a correlation curve for explaining the effects in the latent defect inspection apparatus for the integrated circuit device of FIG. 7.

  In the following embodiment, when it is necessary for the sake of convenience, the description will be divided into a plurality of sections or embodiments. However, unless otherwise specified, they are not irrelevant, and one is the other. Some or all of the modifications, details, supplementary explanations, and the like are related.

  Further, in the following embodiments, when referring to the number of elements (including the number, numerical value, quantity, range, etc.), especially when clearly indicated and when clearly limited to a specific number in principle, etc. Except, it is not limited to the specific number, and may be more or less than the specific number.

  Further, in the following embodiments, the constituent elements (including element steps and the like) are not necessarily indispensable unless otherwise specified and apparently essential in principle. Needless to say.

  Similarly, in the following embodiments, when referring to the shape, positional relationship, etc., of components, etc., unless otherwise specified, and in principle, it is considered that this is not clearly the case, it is substantially the same. Including those that are approximate or similar to the shape. The same applies to the above numerical values and ranges.

[Outline of the embodiment]
First, an outline of the embodiment will be described. In the outline of the present embodiment, as an example, the description will be given with parentheses corresponding constituent elements, reference numerals and the like in parentheses.

  (1) An integrated circuit device latent defect inspection apparatus according to an embodiment includes a storage unit (measurement value storage units 1 and 11), a first calculation unit (correlation line calculation unit 2, correlation curve calculation unit 12), A second calculation unit (distance calculation units 3 and 13); an analysis unit (distribution analysis units 4 and 14); a control unit (control units 5 and 15); and a storage unit (storage units 6 and 16). . The storage unit stores measurement values of the integrated circuit devices. The first calculation unit creates a scatter diagram of the measurement values of the integrated circuit device designated in advance among the measurement values of the integrated circuit devices stored in the storage unit, and calculates a correlation line or a correlation curve from the scatter diagram. To do. The second calculation unit calculates the distance between the correlation line or the correlation curve calculated by the first calculation unit and each point of the scatter diagram. The analysis unit specifies a first (neighboring) measurement value for the measurement value of the designated integrated circuit device based on the distance calculated by the second calculation unit, and an average value of the distance distribution of the first measurement value And the standard deviation are obtained to obtain the Z value of the measured value of the designated integrated circuit device. Furthermore, the analysis unit determines whether or not the measured value is an outlier based on the Z value, and identifies an integrated circuit device that exhibits the outlier. The control unit controls the entire apparatus including the storage unit, the first calculation unit, the second calculation unit, and the analysis unit. The storage unit stores a control program and data necessary for controlling the entire apparatus.

  (2) An integrated circuit device latent defect inspection method according to an embodiment includes a first step (S100, S200), a second step (S101, S201), and a third step as information processing steps by a computer system. (S102, S202), a fourth step (S103, S203), and a fifth step (S104, S204). In the first step, among the measured values of the integrated circuit devices, the measured values of the integrated circuit device designated in advance are collected from the storage unit. In the second step, a scatter diagram of the measurement values collected in the first step is created. In the third step, a correlation line or a correlation curve is calculated from the scatter diagram created in the second step. In the fourth step, the distance between the correlation line or correlation curve calculated in the third step and each point of the scatter diagram is calculated. In the fifth step, based on the distance calculated in the fourth step, the first (neighboring) measurement value for the measurement value of the designated integrated circuit device is specified, and the average value of the distance distribution of the first measurement value And the standard deviation are obtained to obtain the Z value of the measured value of the designated integrated circuit device. Further, in the fifth step, it is determined whether or not the measured value is an outlier based on the Z value, and an integrated circuit device exhibiting the outlier is specified.

  (3) An integrated circuit device latent defect inspection program according to an embodiment includes a first routine, a second routine, a third routine, a fourth routine, and a fifth routine that are executed by a computer system. . The processes in the first to fifth routines correspond to the first to fifth steps of the integrated circuit device latent defect inspection method.

  Hereinafter, an embodiment based on the outline of the above-described embodiment will be described in detail with reference to the drawings. Note that components having the same function are denoted by the same reference symbols throughout the drawings for describing the embodiment, and the repetitive description thereof will be omitted.

  In the following embodiment, an example in which a correlation line is calculated from a scatter diagram will be described in Embodiment 1, and an example in which a correlation curve is calculated from a scatter diagram will be described in Embodiment 2. The present invention is not limited to this, and can be applied in combination with each other.

[Embodiment 1]
An integrated circuit device latent defect inspection apparatus, method, and program according to the first embodiment will be described with reference to FIGS.

  The first embodiment solves the first problem of the prior art. The first problem is that the degree of variation in the distance from the regression line of each point constituting the scatter diagram is not uniform in the scatter diagram but varies depending on the location. In the conventional technology, a constant distance from the regression line is used as a standard value regardless of the position in the scatter diagram, and a value exceeding this standard value is determined as an outlier, which can cause a potential failure in a device that indicates this value. It is judged that there is. However, since the degree of variation in the distance of each point in the scatter diagram varies depending on the position in the scatter diagram, it is difficult to correctly determine an outlier when a uniform standard value is provided for the entire scatter diagram. Therefore, in the first embodiment, as described below, the outlier is correctly determined regardless of the variation in the distance of each point in the scatter diagram.

<Configuration of integrated circuit device latent defect inspection apparatus>
First, the configuration of the integrated circuit device latent defect inspection apparatus according to the first embodiment will be described with reference to FIG. FIG. 1 is a block diagram showing an example of the configuration of an integrated circuit device latent defect inspection apparatus according to the first embodiment.

  The integrated circuit device latent defect inspection apparatus according to the first embodiment includes a measurement value storage unit 1, a correlation line calculation unit 2, a distance calculation unit 3, a distribution analysis unit 4, a control unit 5, and a storage unit 6.

  The measured value storage unit 1 stores the position of each integrated circuit device on the semiconductor wafer, measured values of at least two types of characteristics of the integrated circuit device, and information on non-defective products of the integrated circuit device.

  The correlation line calculation unit 2 is connected to the measurement value storage unit 1 and creates a scatter diagram regarding one characteristic value and another type of characteristic value from the characteristic measurement value information of the integrated circuit device. A correlation line representing the correlation between the measured values is calculated.

  The distance calculation unit 3 is connected to the correlation line calculation unit 2 and calculates the distance between each point in the scatter diagram and the calculated correlation line.

  The distribution analysis unit 4 is connected to the distance calculation unit 3 and analyzes the distribution state of the distance. As a result of the analysis, if there is anything judged to be an outlier, an integrated circuit device having a characteristic measurement value indicating the value is specified, and an indication that the integrated circuit device has a potential failure is output. In the distribution analysis, based on the distance between each point in the scatter diagram and the calculated correlation line, a nearby measurement value for the measurement value of the specified integrated circuit device is specified, and the average of the distance distribution of this neighborhood measurement value is determined. The Z value of the measured value of the specified integrated circuit device is obtained by obtaining the value and the standard deviation. Further, in the distribution analysis, it is determined whether or not the measured value is an outlier based on the Z value, and an integrated circuit device exhibiting the outlier is specified.

  The control unit 5 and the storage unit 6 are connected to each other, and the control program and various data stored in the storage unit 6 can be transmitted to the control unit 5. The control unit 5 is connected to each of the measurement value storage unit 1, the correlation line calculation unit 2, the distance calculation unit 3, and the distribution analysis unit 4, and performs data transmission and control of each unit.

  This integrated circuit device latent defect inspection apparatus is constructed using a computer system. For example, the measured value storage unit 1 is a storage unit such as a hard disk. The control unit 5 is a control unit such as a CPU. The storage unit 6 is a storage unit such as a memory. The correlation line calculation unit 2, the distance calculation unit 3, and the distribution analysis unit 4 are software calculation means (calculation unit) realized by the CPU of the control unit 5 executing a control program stored in the memory of the storage unit 6. 7). The measurement value storage unit 1 can be input from the input unit 8 such as a keyboard or a mouse. From the distribution analysis unit 4, an output operation to the output unit 9 such as display on a display or printing on a printer is possible.

<Operation of the integrated circuit device latent defect inspection apparatus>
Next, the operation of the integrated circuit device latent defect inspection apparatus according to the first embodiment will be described with reference to FIGS. 2 and 3 to 4. FIG. 2 is a flowchart showing an example of the operation of the latent defect inspection method in the above-described latent defect inspection apparatus for the integrated circuit device of FIG.

  The operation of the integrated circuit device in the latent defect inspection apparatus is a latent defect inspection method comprising steps realized by the control unit 5 executing a control program stored in the storage unit 6. Each step of this latent defect inspection method corresponds to the processing of each routine by the control program.

  First, in the correlation line calculation unit 2, two types of characteristic measurement values relating to a plurality of designated semiconductor wafers or an integrated circuit device fabricated on one designated wafer are extracted from the measurement value storage unit 1. It is. At this time, the integrated circuit device may extract characteristic measurement values only for the integrated circuit device determined to be non-defective based on the non-defective product information determined by another means. The value may be retrieved. Furthermore, only characteristic measurement values of some integrated circuit devices in the wafer, which are categorized by a predetermined method, may be extracted. All of these characteristic measurement values are extracted in accordance with a method designated in advance (step S100).

  Then, the correlation line calculation unit 2 creates a scatter diagram from the two types of characteristic measurement values that have been taken out. As shown in FIG. 3, the scatter diagram shows one of two different characteristic measurement values of the same integrated circuit device having a value in the X-axis direction and the other in the Y-axis direction. This scatter diagram is obtained by performing this operation on specific measurement values of all integrated circuit devices extracted based on a method designated in advance (step S101). The example shown in FIG. 3 shows an example in which the characteristic measurement value A of one semiconductor device is A1, the characteristic measurement value B is B1, the value in the X-axis direction is A1, and the value in the Y-axis direction. Plots a point at position B1. A scatter diagram is created by plotting the measured characteristic values of all target semiconductor devices.

Next, the correlation line calculation unit 2 obtains a correlation line from this scatter diagram (step S102). The correlation line is a scatter diagram composed of two types of characteristic measurement values x _i and y _i of n integrated circuit devices in which i is 1 to n, where x and y are the characteristic measurement values of the integrated circuit device. It is given by equations (1) and (2).

  After the correlation line is obtained, information on the correlation line and the scatter diagram is taken into the distance calculation unit 3. Then, the distance calculation unit 3 obtains the distance between the correlation line and each point in the scatter diagram. FIG. 4 shows a method for calculating the distance between the correlation line and each point of the scatter diagram. The distance between the point (Xi, Yi) on the scatter diagram and the correlation line Y = F (X) is defined as Yi-F (Xi). The distance distribution is obtained by calculating the distance from the correlation line for all points in the scatter diagram (step S103). At this time, the distance distribution is obtained separately according to the position in the scatter diagram. Information on the distribution of the distance is taken into the distribution analysis unit 4.

  Next, the distribution analysis unit 4 obtains an average value and a standard deviation from the distance distributions separately obtained according to the positions in the scatter diagram, so that the distance (deviation) from the average value of the distance distribution with respect to the distance of each point is calculated. ) Divided by the standard deviation, that is, the Z value is obtained. If the Z value is outside the range of the predetermined standard value, the Z value is determined to be an outlier, and the integrated circuit device that indicates this value is identified as a device having a potential defect and output. (Step S104).

<Correlation line and standard value in scatter diagram>
Next, correlation lines and standard values in a scatter diagram in the above-described latent defect inspection method for an integrated circuit device will be described with reference to FIG. FIG. 5 is an explanatory diagram for explaining the correlation line and the standard value in the scatter diagram.

  FIG. 5 is an example of a scatter diagram, and a scatter diagram created based on two types of characteristic measurement values (X and Y) of all semiconductor devices determined to be non-defective among the semiconductor devices fabricated on the same wafer. FIG. The correlation line is calculated based on all the data shown in the scatter plot.

  For example, standard value straight lines (upper standard value and lower standard value) are straight lines having a certain distance from the correlation line, and are used for determination of outliers by a conventional method. It can be seen that the variation in the data points around the regression line is small in the vicinity of A in FIG. As described above, although the variation in data points varies depending on the location in the scatter diagram, it is not appropriate to make a determination as an outlier using a uniform standard value in the scatter diagram.

Therefore, in the present embodiment, in order to determine whether or not this data is an outlier with respect to the data indicated by the semiconductor device i, data variation in the vicinity of the data point of the semiconductor device i is evaluated. First, for each piece of data in the vicinity of data (x _i , y _i ), a distribution of distances from the correlation line is obtained. At this time, as a definition of the neighborhood, the value x of the characteristic measurement value X of the data included in the neighborhood may be a range satisfying | x−x _i | ≦ c. At this time, c is a predetermined constant. Alternatively, the data _(x i, _{y i)} data _(x i, _{y i)} of the data located on the left side of the up to n data from close to, the data _(x i of the data located on the right, y _It is possible to select a maximum of n data from those close to _i ). At this time, n is a predetermined constant. For each piece of data in the vicinity selected in this way, each distance from the correlation line is obtained, and a distance distribution is obtained. At this time, this distribution may include the distance of the data (x _i , y _i ) itself.

Next, an average value and a standard deviation of the obtained distance distribution are obtained. Then, the Z value for the distance distribution of the data (x _i , y _i ), that is, (the distance between the data (x _i , y _i ) and the correlation line—the average value of the distribution) / the standard deviation of the distribution is obtained. If the Z value does not fall within the predetermined standard value range, it is determined that the data (x _i , y _i ) is an outlier.

  The standard value in this embodiment determined in this way is determined as an outlier using a standard value that is not uniform in the scatter diagram corresponding to the variation in data points, for example, as shown by the solid line in FIG. To do. In other words, the distance from the correlation line is short at a place where the variation of the data points is small as in the vicinity of A in FIG. 5, and the distance from the correlation line is long at the place where the dispersion of the data points is large as in the vicinity of B. Set.

<Recursive calculation in calculating average and standard deviation for distance distribution>
Next, recursive calculation in calculating the average value and standard deviation for the distance distribution in the above-described latent defect inspection method for an integrated circuit device will be described with reference to FIG. FIG. 6 is a flowchart for explaining recursive calculation in calculating the average value and the standard deviation for the distance distribution.

  In the present embodiment, recursive calculation is performed in calculating the average value and standard deviation for the distance distribution. First, the distribution of distance is set as an initial value, and the entire data is set as a population (step S300). Next, in order to make the initial value of the set A the same as the mother set, the mother set is copied to the set A (step S301). Further, an average value and a standard deviation of the set A are obtained (step S302). The population is analyzed and evaluated using the obtained average value and standard deviation, and a set B is newly defined by excluding elements whose Z value is larger than 3 (step S303).

  Then, the set A and the set B are compared (step S304). If the set A and the set B are not the same (S304-No), the set B is set in the set A (step S305), and the process returns to step S302. The operation for obtaining the average value and standard deviation of the set A is repeated again. If the set A and the set B are the same in step S304 (S304-Yes), the average value and standard deviation of the mother set are determined as the average value and standard deviation (step S306).

  As described above, recursive calculation is performed in calculating the average value and the standard deviation for the distance distribution.

  According to the first embodiment described above, a correlation line is calculated from a scatter diagram of measurement values of integrated circuit devices designated in advance, and a distance between this correlation line and each point of the scatter diagram is calculated. Further, based on the calculated distance, a nearby measurement value for the measurement value of the designated integrated circuit device is specified, and the average value and the standard deviation of the distance distribution of the neighborhood measurement value are obtained to obtain the designated integration value. Obtain the Z value of the measured value of the circuit device. Then, based on the Z value, it can be determined whether or not the measured value is an outlier, and an integrated circuit device exhibiting the outlier can be specified.

[Embodiment 2]
An integrated circuit device latent defect inspection apparatus, method, and program according to the second embodiment will be described with reference to FIGS.

  The second embodiment solves the second problem of the prior art. The second problem is that the shape of the scatter diagram may not be distributed around the regression line, and in this case, an appropriate outlier cannot be determined. In other words, if there is a linear correlation between the measured values, the scatter plot will be a scatter plot distributed evenly around the regression line, but if it is a curve correlation, it will not be a scatter plot distributed evenly around the regression line. Therefore, it is impossible to determine an outlier appropriately. As a result, an integrated circuit device having a latent defect cannot be accurately identified. Therefore, in the second embodiment, as described below, an outlier is appropriately determined in the case of a curve correlation.

  The second embodiment is an example in which a correlation curve is calculated instead of the correlation straight line calculated in the first embodiment described above. In the following, differences from the first embodiment will be mainly described.

<Configuration of integrated circuit device latent defect inspection apparatus>
First, the configuration of the integrated circuit device latent defect inspection apparatus according to the second embodiment will be described with reference to FIG. FIG. 7 is a block diagram illustrating an example of the configuration of the integrated circuit device latent defect inspection apparatus according to the second embodiment.

  The integrated circuit device latent defect inspection apparatus according to the second embodiment includes a measurement value storage unit 11, a correlation curve calculation unit 12, a distance calculation unit 13, a distribution analysis unit 14, a control unit 15, and a storage unit 16.

  The correlation curve calculation unit 12 is connected to the measurement value storage unit 11, creates a scatter diagram from the characteristic measurement value information of the integrated circuit device and the like, and calculates a correlation curve representing the correlation between the characteristic measurement values. The correlation curve calculation unit 12 is calculation means (calculation unit 17) by software realized by the CPU of the control unit 15 executing a control program stored in the memory of the storage unit 16.

  The other measured value storage unit 11, distance calculation unit 13, distribution analysis unit 14, control unit 15, and storage unit 16 have the same functions as the respective units of the first embodiment described above. The measurement value storage unit 11 can be input from the input unit 18. An output operation to the output unit 19 is possible from the distribution analysis unit 14.

<Operation of the integrated circuit device latent defect inspection apparatus>
Next, the operation of the integrated circuit device latent defect inspection apparatus according to the second embodiment will be described with reference to FIGS. 8 and 9 to 10. FIG. 8 is a flowchart showing an example of the operation of the latent defect inspection method in the above-described latent defect inspection apparatus for the integrated circuit device of FIG.

  First, in the correlation curve calculation unit 12, the integrated circuit device characteristic measurement values stored in the measurement value storage unit 11 are integrated on a plurality of designated semiconductor wafers or on one designated wafer. Characteristic measurements for the circuit device are retrieved. At this time, the integrated circuit device may extract characteristic measurement values only for the integrated circuit device determined to be non-defective based on the non-defective product information determined by another means. The value may be retrieved. In addition, only the characteristic measurements of some designated integrated circuit devices in the wafer may be retrieved. All of these characteristic measurement values are extracted in accordance with a method designated in advance (step S200).

  Then, the correlation curve calculation unit 12 creates a scatter diagram from the extracted characteristic measurement values. As shown in FIG. 3 described above, the scatter diagram is obtained by taking one of the two different characteristic measurement values of the same integrated circuit device as a value in the X-axis direction and the other as a value in the Y-axis direction. This scatter diagram is obtained by performing this operation on specific measurement values of all integrated circuit devices extracted based on a method designated in advance (step S201).

  Next, the correlation curve calculation unit 12 obtains a correlation curve from this scatter diagram. If the correlation curve is given by equation (3), the correlation curve is given as a value that minimizes equation (4) from the scatter diagram (step S202).

  Here, Xi and Yi represent each point of the scatter diagram, that is, the characteristic measurement value of the same integrated circuit device, and the characteristic measurement values of n integrated circuit devices from i = 1 to n are shown on the scatter diagram. Shall.

  After the correlation curve is obtained, information on the correlation curve and the scatter diagram is taken into the distance calculation unit 13. Then, the distance calculation unit 13 obtains the distance between the correlation curve and each point of the scatter diagram. FIG. 9 shows a method for calculating the distance between the correlation curve and each point of the scatter diagram. The distance between the point (Xi, Yi) on the scatter diagram and the correlation curve is defined as Yi-F (Xi). The distance distribution is obtained by calculating the distance from the correlation line for all points in the scatter diagram (step S203). Information on the distribution of this distance is taken into the distribution analysis unit 14.

  Next, in the distribution analysis unit 14, an average value and a standard deviation are obtained from the distance distribution, and a value obtained by dividing a distance (deviation) from the average value of each value by the standard deviation, that is, a Z value is obtained. If the Z value is outside the range of the predetermined standard value, the Z value is determined to be an outlier, and the integrated circuit device that indicates this value is identified as a device having a potential defect and output. (Step S204).

<Correlation curve>
Next, the correlation curve in the above-described latent defect inspection method for an integrated circuit device will be described. In the second embodiment, a p-th order curve (p: real number) is used as the correlation curve. Here, examples of a quadratic curve and a cubic curve are shown.

  For example, a quadratic curve represented by Expression (5) is used as the correlation curve. At this time, a, b, and c are constants. From the scatter diagram, a, b, and c are determined so that Equation (6) is minimized. Specifically, it is obtained by equation (7).

  Further, a cubic curve represented by Expression (8) is used as the correlation curve. At this time, a, b, c, and d are constants. From the scatter diagram, a, b, c, and d are determined so that Equation (9) is minimized. Specifically, it is obtained by the equation (10).

<Distance between correlation curve and each point of scatter diagram>
Next, the distance between the correlation curve and each point of the scatter diagram in the above-described latent defect inspection method for an integrated circuit device will be described with reference to FIG. FIG. 10 is an explanatory diagram for explaining the definition of the distance between the correlation curve and each point of the scatter diagram.

  In the second embodiment, the distance between the correlation curve and each point of the scatter diagram is not the method shown in FIG. 9 described above, but the distance is set with the shortest distance between the correlation curve and each point as shown in FIG. Define. That is, although it defined with the distance of the Y-axis direction in FIG. 9, in FIG. 10, it defines with the shortest distance irrespective of the Y-axis direction and the X-axis direction.

<Calculation of correlation curve>
Next, calculation of a correlation curve in the above-described latent defect inspection method for an integrated circuit device will be described with reference to FIG. FIG. 11 is a flowchart for explaining the calculation of the correlation curve.

  First, a set composed of all points in the scatter diagram is defined as a mother set (step S400). Next, the same set as the mother set is defined as an initial value of the set A (step S401). Then, a correlation curve is calculated from information on each point included in the set A (step S402). The distance between the calculated correlation curve and all points in the scatter diagram included in the population is calculated (step S403).

  Further, an average value and a standard deviation of the distribution are obtained from the calculated distance distribution (step S404). At this time, an average value and a standard deviation may be obtained based on a recursive method. Then, the Z value is calculated from the distance to each point in the scatter diagram using the average value and the standard deviation. The Z value is a value obtained by dividing the value obtained by subtracting the average value from the distance by the standard deviation. When the magnitude of the Z value calculated in this way is larger than 3, a point on the scatter diagram showing that value is not selected, and a point with a magnitude of 3 or less is extracted to define a new set B (Step S405).

  Then, the previously created set A is compared with the set B defined this time (step S406). If the set A and the set B are not identical (S406-No), the set B is set in the set A ( Step S407), the process returns to step S402, and the repeated operation is continued. If the set A and the set B are the same in step S406 (S406-Yes), the repetitive operation is terminated, and the finally obtained correlation curve is output (step S408).

<Recursive calculation in calculating average and standard deviation for distance distribution>
Next, in the above-described latent defect inspection method for an integrated circuit device, recursive calculation in calculating the average value and standard deviation for the distance distribution will be described with reference to FIG. FIG. 12 is a flowchart for explaining recursive calculation in calculating the average value and the standard deviation for the distance distribution.

  In the present embodiment, recursive calculation is performed in calculating the average value and standard deviation for the distance distribution. First, the distribution of the distance from the correlation curve for each point in the scatter diagram is obtained, and the entire data is used as a population (step S500). Next, the initial value of the set A is set as a set having the same elements as the mother set (step S501). Further, the average value and standard deviation of the set A are obtained (step S502). When the value of each element of the population is analyzed by this average value and standard deviation, and the Z value (the value obtained by subtracting the average value from each value divided by the standard deviation) is greater than 3, If the element is not selected and the size is 3 or less, a new set B is defined by selecting the element (step S503).

  Then, the set A and the set B are compared (step S504). If the set A and the set B are not identical (S504-No), the set B is set in the set A (step S505), and the process returns to step S502. , Continue the operation repeatedly. In step S504, if set A and set B are the same (S504-Yes), the repetitive operation is terminated, and the average value and standard deviation obtained at the end are determined as the average value and standard deviation of the distance distribution ( Step S506).

  As described above, recursive calculation is performed in calculating the average value and the standard deviation for the distance distribution.

  According to the second embodiment described above, the correlation curve is calculated from the scatter diagram of the measurement values of the integrated circuit device designated in advance, and the distance between the correlation curve and each point of the scatter diagram is calculated. Further, based on the calculated distance, a nearby measurement value for the measurement value of the designated integrated circuit device is specified, and the average value and the standard deviation of the distance distribution of the neighborhood measurement value are obtained to obtain the designated integration value. Obtain the Z value of the measured value of the circuit device. Then, based on the Z value, it can be determined whether or not the measured value is an outlier, and an integrated circuit device exhibiting the outlier can be specified.

[Effect of Embodiment 1 and Embodiment 2]
The effects of the first embodiment and the second embodiment will be described with reference to FIGS. 5 and 13 to 14 described above.

  FIG. 5 described above shows two types of characteristic measurement values (measurement value X and measurement value Y) of all the semiconductor devices determined to be non-defective devices among the integrated circuit devices fabricated on one semiconductor wafer. It is an example of the scatter diagram by.

  The correlation straight line shown in the figure is obtained from the data of each point in the scatter diagram. In the conventional method, a standard value for determining an outlier is set based on the distribution of the distance between each point of the scatter diagram and the correlation line, and this standard value is uniform for all data. For this reason, a straight line indicating a standard value is defined, and this straight line is a straight line having a certain distance from the correlation straight line parallel to the correlation straight line. A point located outside the straight line (in the case of both side standard values of the upper standard value and the lower standard value) is determined as an outlier.

  However, as can be seen from the example of FIG. 5, the variation in the distance from the correlation line of the data points is not uniform, and greatly varies depending on the position in the scatter diagram. For example, the variation in distance from the correlation line of each data point is small in the vicinity of A, but the variation is large in the vicinity of B. Thus, when the degree of variation differs depending on the location in the scatter diagram, it is inappropriate to determine an outlier with a uniform standard value. In other words, data determined to be outliers in areas with large variations are not actually outliers, but within normal variations, or conversely, outliers were not determined within standard values in regions with small variations. It is assumed that the data is actually outliers. Therefore, providing an appropriate standard value corresponding to the degree of variation at each location is important for correctly determining an outlier.

  Therefore, in the present embodiment, an appropriate outlier is determined by setting a dedicated standard value for each piece of data in the scatter diagram by the configuration and operation of the above-described integrated circuit device latent defect inspection apparatus. Has been implemented. For example, as indicated by a solid line in FIG. 5, a standard value is defined such that the distance from the correlation line is short at a place where the variation is small such as the vicinity of A and the distance from the correlation line is long at a place where the variation is large such as the vicinity of B. That is, the average value and the standard deviation are obtained based on the distribution of the distance from the correlation line of the data located in the vicinity of the data to be determined, and the Z value of the target data is obtained. And if this Z value is within the range of a predetermined standard value (this standard value is constant for all data), it is not an outlier, and conversely, if it exceeds the standard value, an outlier Judge.

  For the selection of the proximity data of the target data, the data located within a certain distance from the target data, for example, the X-axis value of the target data and the X-axis of the data that is close to the X-axis value of the target data The neighborhood data is selected so that the difference from the direction value is equal to or less than a certain value. Alternatively, regarding the value in the X-axis direction of each data, a predetermined number of data is selected in order from the closest to the value in the X-axis direction of the target data, and is set as the neighborhood data.

  In addition, when appropriate outliers are determined by setting a dedicated standard value for each piece of data in a scatter diagram, the same applies to correlation curves as well as correlation lines. Can do. Also in the case of the correlation curve, a standard value is determined where the distance from the correlation curve is short at a place where the variation is small and the distance from the correlation curve is long at a place where the variation is large. The neighborhood data can be selected by selecting the neighborhood data of the target data as in the case of the correlation line.

  Next, the effect of this embodiment will be described. FIG. 13 is an explanatory diagram showing a scatter diagram and a correlation line based on the measurement values A and B of the integrated circuit device. In a conventional integrated circuit device latent defect inspection method, in order to identify an outlier in the scatter diagram, quantification is performed based on the distance from the correlation line of the scatter diagram. That is, a correlation line (regression line) is obtained from the scatter diagram, the distance between each point of the scatter diagram and the correlation line is calculated, and the Z value of each value is calculated in the distance distribution to quantify the outlier. When the outlier is larger (smaller) than the specified value, the integrated circuit device showing the value is determined to be defective. In the example of FIG. 13, an integrated circuit device indicating a point in the scatter diagram existing outside the standard value line drawn above and below the correlation line is determined to be defective.

  This concept quantifies the degree of deviation from the correlation when there is a correlation between two measured values in a set of measured values, and defines it as an outlier with respect to a measured value that shows a certain value or more. . In this method, when the correlation is a linear correlation, the degree of deviation from the correlation can be quantified in the scatter diagram by the distance from the correlation line. However, when the correlation is a curved correlation, the degree of deviation from the correlation line. It is not appropriate to quantify. Therefore, in the present embodiment, when the correlation is a curve correlation, the quantification is performed with the distance from the correlation curve as a quantification method of the degree of deviation from the correlation.

  FIG. 14 is an explanatory diagram showing a scatter diagram and a correlation curve for explaining the effect of the present embodiment, and is the same scatter diagram as FIG. 13. A correlation curve is defined from this scatter diagram, and the distance between each point of the scatter diagram and the correlation curve is obtained. Outliers are defined from the average value and standard deviation of the distance distribution, and a defect determination area is defined in the scatter diagram based on the standard value. The boundary between the defect determination area and the non-defective product determination area is equal to a curve obtained by moving the correlation curve up and down (Y-axis direction) by a certain amount. When the relationship between the measurement value A and the measurement value B is not linear correlation but curve correlation, the distance between each point of the scatter diagram and the correlation curve is used as in the latent defect inspection method according to the present embodiment. By defining an outlier, the degree of deviation from the correlation can be appropriately quantified, and an accurate outlier can be calculated.

Next, an example in which a curve correlation occurs between measured values of an integrated circuit device is presented. For example, most of the leakage current of the integrated circuit device is a sub-threshold current, and therefore can be written as Equation (11). Here, I is a leakage current, A and B are constants related to the manufacturing process of the integrated circuit device, and T is temperature. When the leakage current of the semiconductor device is measured while changing the temperature, if the leakage current at the high temperature T _H is I _H and the leakage current at the low temperature T _L is I _L , the equations (12) and (13) are obtained. If there is a relationship of T _H = αT _L (α is a constant), Equation (14) is obtained.

  From this, it can be seen that there is a curve correlation when a scatter diagram is drawn using the leakage current at low temperature of the semiconductor device and the leakage current at high temperature as measured values. For example, when α = 2, there is a quadratic curve correlation between the leakage current value at a low temperature and the leakage current value at a high temperature.

  According to the present embodiment described above, an accurate outlier can be detected regardless of variations in the distances between the points in the scatter diagram. That is, in the present embodiment, by using the distance from the correlation line or the correlation curve and setting the standard value according to the variation of each point in the scatter diagram, it is possible to accurately detect the outlier. In the present embodiment, in the scatter diagram, when the correlation curve and the distance from the correlation curve are used as means for quantifying the outlier, compared to the outlier quantification using the correlation line, Accurate quantification of outliers is possible. As a result, according to the present embodiment, the potential defect of the integrated circuit device can be accurately inspected.

  As mentioned above, the invention made by the present inventor has been specifically described based on the embodiment. However, the invention is not limited to the embodiment, and various modifications can be made without departing from the scope of the invention. Needless to say.

  For example, in the first embodiment, the example in which the correlation line is calculated from the scatter diagram has been described, and in the second embodiment, the example in which the correlation curve is calculated from the scatter diagram has been described. is there. In this case, both the effect when using the correlation line and the effect when using the correlation curve can be obtained.

1, 11 Measurement value storage unit 2 Correlation line calculation unit 12 Correlation curve calculation unit 3, 13 Distance calculation unit 4, 14 Distribution analysis unit 5, 15 Control unit 6, 16 Storage unit 7, 17 Calculation unit 8, 18 Input unit 9 , 19 Output section

Claims (18)

A storage for storing the measurement values of each integrated circuit device;
A first calculation for creating a scatter diagram of the measurement values of the integrated circuit device designated in advance among the measurement values of the integrated circuit devices stored in the storage unit and calculating a correlation line or a correlation curve from the scatter diagram. And
A second calculation unit for calculating a distance between the correlation straight line or the correlation curve calculated by the first calculation unit and each point of the scatter diagram;
Based on the distance calculated by the second calculation unit, a first measurement value for the measurement value of the designated integrated circuit device is specified, and an average value and a standard deviation of the distance distribution of the first measurement value are obtained. Obtaining a Z value of the measured value of the designated integrated circuit device, determining whether the measured value is an outlier based on the Z value, and identifying an integrated circuit device exhibiting the outlier An analysis unit;
A control unit that controls the entire apparatus including the storage unit, the first calculation unit, the second calculation unit, and the analysis unit;
A storage unit for storing a control program and data necessary for controlling the entire apparatus;
An apparatus for inspecting potential defects of an integrated circuit device. In the integrated circuit device latent defect inspection apparatus according to claim 1,
When the analysis unit identifies the first measurement value, the analysis unit obtains the measurement value of the integrated circuit device having a measurement value within a predetermined range from the measurement value of the designated integrated circuit device. An apparatus for inspecting a latent defect of an integrated circuit device, wherein one measured value is set. In the integrated circuit device latent defect inspection apparatus according to claim 1,
When the analysis unit specifies the first measurement value, a predetermined number of measurement values in order from the smallest difference from the measurement value of the designated integrated circuit device are used as the first measurement value. , Integrated circuit device latent defect inspection device. In the integrated circuit device latent defect inspection apparatus according to claim 1,
The said analysis part is a latent defect test | inspection apparatus of an integrated circuit device which performs a recursive calculation, when calculating | requiring the average value and standard deviation of distance distribution of a said 1st measured value. In the integrated circuit device latent defect inspection apparatus according to claim 1,
The first calculation unit uses the p-th order curve (p: real number) as the correlation curve when calculating the correlation curve. In the integrated circuit device latent defect inspection apparatus according to claim 1,
The first calculation unit is a latent defect inspection apparatus for an integrated circuit device, which uses a recursive method when calculating the correlation curve. As a step of information processing by computer system,
A first step of collecting, from the storage unit, pre-designated integrated circuit device measurement values among the measurement values of the integrated circuit devices;
A second step of creating a scatter plot of the measurements collected in the first step;
A third step of calculating a correlation line or a correlation curve from the scatter diagram created in the second step;
A fourth step of calculating a distance between the correlation straight line or the correlation curve calculated in the third step and each point of the scatter diagram;
Based on the distance calculated in the fourth step, a first measurement value for the measurement value of the designated integrated circuit device is specified, and an average value and a standard deviation of the distance distribution of the first measurement value are obtained. Determining the Z value of the measured value of the designated integrated circuit device, determining whether the measured value is an outlier based on the Z value, and identifying an integrated circuit device exhibiting the outlier 5 steps,
A method for inspecting potential defects of an integrated circuit device, comprising: The method for inspecting a potential defect of an integrated circuit device according to claim 7,
In the fifth step, when the first measurement value is specified, the measurement value of the integrated circuit device having a measurement value within a predetermined range is different from the measurement value of the designated integrated circuit device. A method for inspecting a latent defect of an integrated circuit device, which is a first measurement value. The method for inspecting a potential defect of an integrated circuit device according to claim 7,
In the fifth step, when the first measurement value is specified, a predetermined number of measurement values in order from the smallest difference from the measurement value of the designated integrated circuit device are referred to as the first measurement value. A method for inspecting a latent defect of an integrated circuit device. The method for inspecting a potential defect of an integrated circuit device according to claim 7,
In the fifth step, the method of inspecting a latent defect of an integrated circuit device, wherein recursive calculation is performed when obtaining an average value and a standard deviation of the distance distribution of the first measurement value. The method for inspecting a potential defect of an integrated circuit device according to claim 7,
In the third step, a latent defect inspection method for an integrated circuit device, wherein a p-order curve (p: real number) is used as the correlation curve when calculating the correlation curve. The method for inspecting a potential defect of an integrated circuit device according to claim 7,
In the third step, a method for inspecting potential defects in an integrated circuit device, wherein a recursive method is used when calculating the correlation curve. A first routine for collecting, from the storage unit, a measurement value of an integrated circuit device designated in advance among measurement values of each integrated circuit device;
A second routine for creating a scatter diagram of the measured values collected in the first routine;
A third routine for calculating a correlation line or a correlation curve from the scatter diagram created in the second routine;
A fourth routine for calculating a distance between the correlation straight line or the correlation curve calculated in the third routine and each point of the scatter diagram;
Based on the distance calculated in the fourth routine, a first measurement value for the measurement value of the designated integrated circuit device is specified, and an average value and a standard deviation of the distance distribution of the first measurement value are obtained. Determining the Z value of the measured value of the designated integrated circuit device, determining whether the measured value is an outlier based on the Z value, and identifying an integrated circuit device exhibiting the outlier 5 routines,
A program for inspecting potential defects of an integrated circuit device, which causes a computer system to execute. In the integrated circuit device latent defect inspection program according to claim 13,
In the fifth routine, when the first measurement value is specified, the measurement value of the integrated circuit device having a measurement value within a predetermined range is different from the measurement value of the designated integrated circuit device. An integrated circuit device latent defect inspection program as a first measurement value. In the integrated circuit device latent defect inspection program according to claim 13,
In the fifth routine, when the first measurement value is specified, a predetermined number of measurement values in order from the smallest difference from the measurement value of the designated integrated circuit device are used as the first measurement value. An integrated circuit device latent defect inspection program. In the integrated circuit device latent defect inspection program according to claim 13,
In the fifth routine, a latent defect inspection program for an integrated circuit device that performs recursive calculation when obtaining an average value and a standard deviation of the distance distribution of the first measurement value. In the integrated circuit device latent defect inspection program according to claim 13,
In the third routine, when calculating the correlation curve, a p-th order curve (p: real number) is used as the correlation curve. In the integrated circuit device latent defect inspection program according to claim 13,
In the third routine, a recursive defect inspection program for an integrated circuit device, which uses a recursive method when calculating the correlation curve.

Images