JP3573678B2 - Inspection / analysis device, method, and recording medium for semiconductor application device - Google Patents

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to an inspection / analysis technique for a semiconductor application device, particularly a semiconductor application device having a structure in which semiconductor elements such as a memory LSI are arranged in a certain direction.
[0002]
[Prior art]
The cause of the failure of semiconductor application equipment is that when it is not miniaturized as it is now, troubles in manufacturing, especially relatively large dimensions, except for cases where the dimensions and arrangement of elements and wiring are obviously inappropriate Garbage etc. were the main. For this reason, if an appearance inspection device such as an optical microscope is used, the cause of the defect can be found with the naked eye, and the size of the semiconductor application device is not as large as it is now, so that the time required for the analysis is short.
[0003]
Since the range affected by the dust is limited to the vicinity of the dust, particularly in the case of a semiconductor application device having a redundant circuit such as a memory LSI, a defective element and a defective circuit affected by the dust are replaced with a redundant circuit. , It was easy to convert it to a good electrical product. For this reason, when the main cause of the defect is dust, a memory LSI having a redundant circuit has a higher yield compared to a logic LSI having no similar remedy means.
[0004]
However, in recent years, with the miniaturization of element dimensions and the increase in the size of semiconductor applied devices, slight design and manufacturing variations and deviations that cannot be detected by appearance inspection have become problems as causes of defects. For example, when some wiring is connected by through-holes and as a result, the resistance increases, etc. Is more likely to occur at a circuit or mask position affected by the above.
[0005]
In these failures, the characteristics of the failure distribution caused only by manufacturing variations and deviations are random, but if a failure factor enters at the design stage, it becomes a regular distribution. Generally, in a semiconductor application device, a defect having a random distribution and a defect having a regular distribution are mixed.
[0006]
If the defect factor entered at the design stage is so strong that it causes a single failure, the failure distribution shows a clear regularity failure distribution, but if the effect is too weak to cause a failure by itself, a random A defect occurs only at a location where the position of the distribution defect factor and the position of the defect factor entered at the design stage overlap. For this reason, it may look like a random distribution at first glance. In this case, it is difficult to notice that a failure factor that has entered at the design stage is included, and as a result, measures in the design department are delayed.
[0007]
In recent semiconductor application devices in which the margin in the manufacturing process is decreasing, random distribution failures tend to increase, and accordingly, failure factors entered at the design stage are hidden in the random distribution failures, making it difficult to understand. In addition, there is a problem that it becomes more difficult to analyze the cause of the defect. In addition, unlike the case of dust, the adverse effect of the cause of failure covers a wide range such as the entire semiconductor application device, so that it is difficult to rescue the circuit with a redundant circuit, and the yield of the memory LSI is larger than that of the case due to dust. Was.
[0008]
For example, if the technology described in Japanese Patent Application Laid-Open No. 7-72206 is used, an electrical defect is determined as a block-like defect when all or most of the elements in the block in which the memory elements of the memory LSI are two-dimensionally arranged are defective. Column failure if all or most of the elements arranged in the column direction of the block are defective; row failure if all or most of the elements arranged in the row direction of the block are defective; If two elements are defective, a two-bit or pair-bit defect, and if one element in the block alone is defective, a one-bit or single-bit defect, and so on, a defect distribution determined in a fixed form in appearance. , Block driving circuit, column driving circuit or column driving circuit, row driving wiring or row driving wiring, insulation failure between two bits or poor connection between two bits, By associating such a defect can be constructed an algorithm associating failure distribution and failure cause.
[0009]
In other words, by applying the technique described in Japanese Patent Application Laid-Open No. 7-72206, when two bits are short-circuited due to dust shorting between wirings or disconnection of wirings, defective distribution and defect The cause can be accurately correlated.
[0010]
However, 1-bit defects that are not caused by dust but are caused by manufacturing variations or deviations are distributed throughout the semiconductor application device, and defects such as blocks, columns, or rows, which do not have a fixed shape in appearance. In the case of distribution, it is difficult for the technique to associate a failure distribution with a failure cause. For example, even if the cause of failure is the same, the number of defective elements is not constant, and the shape of the outer periphery of the defect distribution is not constant.
[0011]
On the other hand, the inventor of the present application categorized the defect distribution into a regular distribution and an irregular distribution in Japanese Patent Application No. 11-1680 previously filed as a related art of the present application. A technique has been proposed that enables the latter to be estimated as a design-induced failure.
[0012]
[Problems to be solved by the invention]
However, as described above, in recent years, slight design and manufacturing variations and deviations have become important as failure factors, and the frequency of appearance of seemingly random failure distributions has increased, resulting in more detailed classification of failure distributions. Therefore, there is an increasing need to increase the accuracy associated with the cause of failure.
[0013]
For example, when the number of defects or the defect density is small, it may be difficult to judge whether the distribution is irregular or regular, or a defect due to manufacturing may look like a regular distribution to the naked eye. . For this reason, only applying the related technology proposed in Japanese Patent Application No. 11-1680 may fail to sufficiently associate the failure distribution with the failure cause.
[0014]
The present invention solves a problem that could not be solved in the related art, and accurately classifies a distribution of defective elements included in a semiconductor application device, and can associate the distribution with a cause of failure. It is an object of the present invention to provide a computer-readable recording medium on which an inspection analysis device and a method for an application device and a program for realizing the same are recorded.
[0015]
[Means for Solving the Problems]
In order to achieve the above object, an inspection and analysis apparatus for a semiconductor application device according to a first aspect of the present invention includes:
Interval calculating means for calculating the interval for all combinations of any two of the defective elements included in the semiconductor application device to be analyzed;
Combination number calculating means for calculating the number of combinations excluding the number of combinations having a value of 0 among combinations of intervals between defective elements calculated by the interval calculating means,
Divisor calculating means for respectively obtaining a divisor of the interval between the defective elements calculated by the interval calculating means,
Comparison means for comparing a value obtained by dividing the maximum value of the divisor calculated by the divisor calculation means by the number of combinations calculated by the combination number calculation means with a predetermined threshold value,
As a result of the comparison by the comparing means, when a value obtained by dividing the maximum value of the divisor by the number of combinations is larger than a predetermined threshold, the distribution of the defective elements included in the semiconductor application device is different from the regular distribution and the irregular distribution. First analysis means for determining that the distribution is a low-density distribution that cannot be distinguished from the regular distribution;
It is characterized by having.
[0016]
In the inspection and analysis apparatus for a semiconductor application device according to the first aspect, the predetermined threshold may be 1.
[0017]
The inspection / analysis device for a semiconductor application device according to the first aspect, wherein as a result of the comparison by the comparing means, a value obtained by dividing the maximum value of the divisor by the number of combinations is equal to or less than a predetermined threshold, Second analysis means for analyzing whether the distribution of defective elements included in the application device includes a regular distribution or an irregular distribution may be further provided.
[0018]
In the inspection / analysis device for a semiconductor application device according to the first aspect,
The second analysis means includes:
Frequency calculation means for each value to calculate the appearance frequency for each divisor found by the divisor calculation means,
Expected value function calculating means for calculating, for each divisor, a value of an expected value function corresponding to the appearance frequency of each divisor calculated by the value-based frequency calculating means,
Based on the expected value function calculated by the expected value function calculating means, a regularity determining means for determining whether the distribution of defective elements included in the semiconductor application device includes a regularity distribution or an irregularity distribution. May be provided.
[0019]
In this case, the inspection / analysis apparatus for a semiconductor application device according to the first aspect may include the expected value function calculated by the expected value function calculating means, a predetermined reference value, and a period of the predicted regularity distribution. And a third analysis means for analyzing the composition ratio of the distribution of the defective elements included in the semiconductor application device based on the irregular distribution and the number of types of regular distributions. It can be.
[0020]
The inspection and analysis device for a semiconductor application device according to the first aspect,
Frequency calculation means for each value to calculate the appearance frequency for each divisor found by the divisor calculation means,
Expected value function calculating means for calculating, for each divisor, a value of an expected value function corresponding to the appearance frequency of each divisor calculated by the value-based frequency calculating means,
When the first analysis means determines that the distribution is a low-density distribution, and the value obtained by dividing the maximum value of the divisor by the number of combinations falls within a predetermined range near the predetermined threshold, the expected value function calculation means And a fourth analyzing means for analyzing whether the distribution of the defective elements included in the calculation result includes a regular distribution or an irregular distribution.
[0021]
In this case, the fourth analyzing means may determine whether the distribution of the defective elements included in the semiconductor application device is regular according to whether the value of the expected value function for the prime divisor is close to 1 or close to the divisor. It can be determined whether the distribution includes a gender distribution or an irregular distribution.
[0022]
Further, when the value of the expected value function when the value of the divisor is 2 is close to 1, the fourth analysis means may determine the distribution of the defective elements included in the semiconductor application device as an irregular distribution. If it is close to 2, it can be determined that it includes a regular distribution with an even period.
[0023]
The inspection analysis apparatus for a semiconductor application device according to the first aspect may further include a defective element number investigation unit for examining the number of defective elements included in the semiconductor application device. in this case,
The interval calculating means may calculate an interval between defective elements when the number of defective elements checked by the defective element number investigating means is within a predetermined range.
[0024]
Here, the inspection analysis device for a semiconductor application device according to the first aspect, when the number of defective elements examined by the defective element number investigating means is not in the predetermined range, the semiconductor analyzer according to the number of defective elements. A fifth analysis means for classifying the distribution of defective elements included in the application device may be further provided.
[0025]
The fifth analysis means may include: a case where the semiconductor application device does not include a defective element; a case where the semiconductor application device includes a defective element but the number is equal to or less than a first predetermined number; The distribution of the defective elements included in the semiconductor application device can be classified according to whether the number of defective elements included in the semiconductor application device is equal to or more than the second predetermined number.
[0026]
The inspection and analysis device for a semiconductor application device according to the first aspect,
Analysis means for the distribution of defective elements included in the semiconductor application device, accumulation means for accumulating according to the type of the distribution,
When an analysis result on the distribution of defective elements included in the semiconductor application device is obtained, the obtained analysis result is analyzed by comparing the obtained analysis result with the analysis result previously stored in the storage unit, and May be further provided.
[0027]
In order to achieve the above object, a semiconductor application device inspection / analysis device according to a second aspect of the present invention includes:
Interval calculating means for calculating the interval for all combinations of any two of the defective elements included in the semiconductor application device to be analyzed;
Combination number calculating means for calculating the number of combinations excluding the number of combinations having a value of 0 among combinations of intervals between defective elements calculated by the interval calculating means,
Divisor calculating means for respectively obtaining a divisor of the interval between the defective elements calculated by the interval calculating means,
Frequency calculation means for each value to calculate the appearance frequency for each divisor found by the divisor calculation means,
Expected value function calculating means for calculating, for each divisor, a value of an expected value function corresponding to the appearance frequency of each divisor calculated by the value-based frequency calculating means,
Based on the expected value function calculated by the expected value function calculating means, a predetermined reference value, and a predicted period of the regularity distribution, the distribution of the defective elements included in the semiconductor application device is changed to an irregularity distribution. And analysis means for analyzing the composition ratio of what kind of regular distribution and what kind of regular distribution are included
It is characterized by having.
[0028]
The analysis means,
It may include means for determining whether the maximum value of the expected value function is 1 or less. in this case,
When it is determined that the maximum value of the expected value function is 1 or less, it can be determined that the distribution of the defective elements included in the semiconductor application element is an ideal irregular distribution.
[0029]
The analysis means,
When the maximum value of the expected value function is not equal to or less than 1, means for determining whether the maximum value of the expected value function is equal to a corresponding divisor may be included. in this case,
When it is determined that the maximum value of the expected value function is equal to a divisor corresponding thereto, the distribution of defective elements included in the semiconductor application element changes the cycle of the divisor corresponding to the maximum value of the expected value function. It can be determined that this is one type of regularity distribution.
[0030]
The analysis means,
If the maximum value of the expected value function is not equal to or less than 1, means for determining whether the maximum value of the expected value function is equal to one half of a corresponding divisor may be included. in this case,
If it is determined that the maximum value of the expected value function is equal to one half of the corresponding divisor, the distribution of the defective elements included in the semiconductor application element is calculated by summing the maximum value of the expected value function. Can be determined to be a regular distribution composed of two numbers of periods, which are divisors corresponding to.
[0031]
The analysis means,
When the divisor corresponding to the maximum value of the expected value function is not equal to one half thereof, it is determined whether the maximum value of the expected value function is larger than 1 and smaller than a predetermined reference value. Means,
When it is determined that the maximum value of the expected value function is greater than 1 and smaller than a predetermined reference value, a set of regular distribution periods whose divisor corresponding to the maximum value of the expected value function is predicted in advance And means for determining whether or not the data is included in the data. in this case,
When the divisor corresponding to the maximum value of the expected value function is included in the set of the periods, the distribution of the defective elements included in the semiconductor application element includes an irregularity distribution, and the maximum of the expected value function. It includes one type of regular distribution having a divisor period corresponding to the value, and can be determined to be a distribution mainly composed of irregular distribution.
[0032]
The analysis means,
If it is determined that the divisor corresponding to the maximum value of the expected value function is not included in the set of periods of the regular distribution that is predicted in advance, the sum of the divisor corresponding to the maximum value of the expected value function and Means for determining whether two such numbers are included in the set of periods. in this case,
When it is determined that two numbers whose sum is a divisor corresponding to the maximum value of the expected value function are included in the set of periods, the distribution of defective elements included in the semiconductor application element is determined by the two It is possible to determine that the distribution includes a regular distribution having a cycle of a number and is mainly a irregular distribution.
[0033]
The analysis means,
When it is determined that the maximum value of the expected value function is equal to or less than 1 or equal to or greater than a predetermined reference value, the maximum value of the expected value function is larger than half of the corresponding divisor, and Means may be included for determining if it is less than a divisor. in this case,
When it is determined that the maximum value of the expected value function is larger than half of the corresponding divisor and smaller than the divisor, the distribution of the defective elements included in the semiconductor application element is determined to be incorrect. It can be determined that the distribution includes a regular distribution and that is mainly a regular distribution whose cycle is a divisor corresponding to the maximum value of the expected value function.
[0034]
The analysis means,
A rule that the divisor corresponding to the maximum value of the expected value function is predicted in advance when it is determined that the maximum value of the expected value function is two minutes or more of the divisor corresponding to the expected function. It may include means for determining whether or not it is included in the set of cycles of the gender distribution. in this case,
When it is determined that the maximum value of the expected value function is included in the set of periods, the distribution of defective elements included in the semiconductor application element includes an irregularity distribution, and the maximum value of the expected value function It can be determined that the distribution is mainly a regular distribution having a period corresponding to the divisor.
[0035]
Further, the analysis means includes:
Either of the two expected value functions corresponding to the divisor larger by 1 and smaller by 1 than the divisor corresponding to the maximum value of the expected value function is equal to or larger than the predetermined reference value, and Means for determining whether it is smaller than a divisor corresponding to the maximum value. in this case,
Either of the two expected value functions corresponding to the divisor larger or smaller by 1 than the divisor corresponding to the maximum value of the expected value function is greater than or equal to the predetermined reference value, and When it is determined that the divisor is smaller than the divisor corresponding to the maximum value, the distribution of the defective elements included in the semiconductor application element includes an irregularity distribution and the divisor corresponding to the maximum value of the expected value function. Judging that the distribution is mainly a regular distribution due to a factor that affects adjacent coordinates as well as the period,
Which of the two expected value functions corresponding to the divisor larger or smaller by one than the divisor corresponding to the maximum value of the expected value function is equal to or smaller than the predetermined reference value or the maximum value of the expected value function If it is determined that the divisor is equal to or greater than the divisor, the distribution of the defective elements included in the semiconductor application element, including the irregularity distribution, and the cycle corresponding to the divisor corresponding to the maximum value of the expected value function. At the same time, it can be determined that the distribution is mainly a regular distribution due to a cause that does not affect adjacent coordinates.
[0036]
The analysis means,
When it is determined that the maximum value of the expected value function is not included in the set of the periods, two numbers whose sum is a divisor corresponding to the maximum value of the expected value function are included in the set of the periods. It may include means for determining whether a in this case,
If it is determined that two numbers whose sum is a divisor corresponding to the maximum value of the expected value function are included in the set of the periods, the distribution of the defective elements included in the semiconductor application element becomes irregular. The distribution can be determined to be a distribution mainly including a regular distribution having the two numbers as a cycle while including the distribution.
[0037]
Further, the analysis means includes:
When it is determined that two numbers whose sum is a divisor corresponding to the maximum value of the expected value function are not included in the set of the periods, the number is larger or smaller by one than the sum of the two numbers. Means for determining whether the expected value function corresponding to the divisor is equal to or larger than the predetermined reference value and smaller than the divisor corresponding to the maximum value of the expected value function. in this case,
Each of two expected value functions corresponding to a divisor larger or smaller by one than the sum of the two numbers is equal to or greater than the predetermined reference value and corresponds to a maximum value of the expected value function. When it is determined that the number is smaller than the number, the distribution of the defective elements included in the semiconductor application element includes an irregularity distribution, the two numbers are set as a cycle, and the adjacent coordinates are affected. Is determined to be mainly a regular distribution by
Which of the two expected value functions corresponding to a divisor larger or smaller by one than the sum of the two numbers is equal to or less than the predetermined reference value or equal to or greater than a divisor corresponding to the maximum value of the expected value function When it is determined that the distribution of defective elements included in the semiconductor application element includes an irregularity distribution, the two numbers are set as a period, and a rule due to a factor that does not affect adjacent coordinates is determined. It can be determined that the distribution is mainly a gender distribution.
[0038]
The analysis means,
When it is determined that two numbers whose sum is a divisor corresponding to the maximum value of the expected value function are not included in the set of the periods, the distribution of the defective elements included in the semiconductor application element is changed to a main component. Can be determined to be an unclear irregular distribution.
[0039]
In order to achieve the above object, a method for inspecting and analyzing a semiconductor application device according to a third aspect of the present invention includes:
An interval calculation step of calculating an interval for all combinations of any two of the defective elements included in the semiconductor application device to be analyzed;
A number-of-combinations calculating step of calculating the number of combinations excluding the number of combinations having a value of 0 among combinations of intervals between defective elements calculated in the interval calculating step;
A divisor calculation step for respectively obtaining a divisor of the interval between the defective elements calculated in the interval calculation step,
A comparison step of comparing a value obtained by dividing the maximum value of the divisor calculated in the divisor calculation step by the number of combinations calculated in the combination number calculation step with a predetermined threshold value,
As a result of the comparison in the comparison step, when a value obtained by dividing the maximum value of the divisor by the number of combinations is larger than a predetermined threshold, the distribution of the defective elements included in the semiconductor application device has a regular distribution and a regular distribution. A first analysis step of determining that the distribution is a low-density distribution that cannot be distinguished from the irregular distribution;
It is characterized by including.
[0040]
The inspection and analysis method for a semiconductor application device according to the third aspect,
As a result of the comparison in the comparison step, when a value obtained by dividing the maximum value of the divisor by the number of combinations is equal to or smaller than a predetermined threshold, the distribution of the defective elements included in the semiconductor application device has a regular distribution. The method may further include a second analysis step of analyzing whether the data includes the irregular distribution or the irregular distribution.
[0041]
In order to achieve the above object, a method for inspecting and analyzing a semiconductor application device according to a fourth aspect of the present invention includes:
An interval calculation step of calculating an interval for all combinations of any two of the defective elements included in the semiconductor application device to be analyzed;
A number-of-combinations calculating step of calculating the number of combinations excluding the number of combinations having a value of 0 among combinations of intervals between defective elements calculated in the interval calculating step;
A divisor calculation step for respectively obtaining a divisor of the interval between the defective elements calculated in the interval calculation step,
A frequency calculation step for each value for calculating the appearance frequency for each divisor obtained in the divisor calculation step,
An expected value function calculating step of calculating, for each divisor, a value of an expected value function corresponding to the appearance frequency of each divisor calculated in the frequency calculation step for each value,
Based on the expected value function calculated in the expected value function calculation step, a predetermined reference value, and a predicted period of the regularity distribution, the distribution of the defective elements included in the semiconductor application device is changed to the irregularity distribution. And an analysis step for analyzing the composition ratio of how many types of regular distributions are included and in what ratio
It is characterized by including.
[0042]
In order to achieve the above object, a computer-readable recording medium according to a fifth aspect of the present invention includes:
Interval calculation means for calculating the interval for all combinations of any two of the defective elements included in the semiconductor application device to be analyzed;
Combination number calculating means for calculating the number of combinations excluding the number of combinations having a value of 0 among combinations of intervals between defective elements calculated by the interval calculating means,
Divisor calculating means for respectively obtaining a divisor of the interval between the defective elements calculated by the interval calculating means,
Comparison means for comparing a value obtained by dividing the maximum value of the divisor calculated by the divisor calculation means by the number of combinations calculated by the combination number calculation means with a predetermined threshold value, and
As a result of the comparison by the comparing means, when a value obtained by dividing the maximum value of the divisor by the number of combinations is larger than a predetermined threshold, the distribution of the defective elements included in the semiconductor application device is different from the regular distribution and the irregular distribution. First analysis means for determining a low-density distribution that cannot be distinguished from a regular distribution
A program for causing a computer to function as a computer.
[0043]
The computer-readable recording medium according to the fifth aspect, wherein the value obtained by dividing the maximum value of the divisor by the number of combinations is equal to or less than a predetermined threshold value as a result of the comparison by the comparing means. A program for causing a computer to function as second analysis means for analyzing whether the distribution of defective elements included in the device includes a regular distribution or an irregular distribution may be further recorded.
[0044]
In order to achieve the above object, a computer-readable recording medium according to a sixth aspect of the present invention includes:
Interval calculation means for calculating the interval for all combinations of any two of the defective elements included in the semiconductor application device to be analyzed;
Combination number calculating means for calculating the number of combinations excluding the number of combinations having a value of 0 among combinations of intervals between defective elements calculated by the interval calculating means;
Divisor calculating means for respectively obtaining a divisor of the interval between the defective elements calculated by the interval calculating means,
Frequency calculation means for each value to calculate the appearance frequency for each divisor found by the divisor calculation means,
Expected value function calculating means for calculating, for each divisor, a value of an expected value function corresponding to the appearance frequency of each divisor calculated by the value-based frequency calculating means; and
Based on the expected value function calculated by the expected value function calculating means, a predetermined reference value, and a predicted period of the regularity distribution, the distribution of the defective elements included in the semiconductor application device is changed to an irregularity distribution. Analysis means for analyzing the composition ratio of what kind of regular distribution and what kind of regular distribution are included
A program for causing a computer to function as a computer.
[0045]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings.
[0046]
[First Embodiment]
FIG. 1 is a block diagram showing a configuration of an inspection and analysis device (hereinafter, simply referred to as “inspection and analysis device”) of a semiconductor application device according to this embodiment. As shown in the figure, the inspection and analysis apparatus includes an input device 1, an operation device 2, an analysis device 3, and an output device 4.
[0047]
The arithmetic device 2 includes an interval calculation unit 21, a combination number calculation unit 22, a divisor calculation unit 23, a frequency calculation unit 24, and an expected value function calculation unit 25. The analysis device 3 includes a low density distribution analysis unit 31 and a regular distribution analysis unit 32. Note that the arithmetic device 2 and the analysis device 3 are realized by a computer device such as an engineering workstation that operates by executing a program. These may be physically implemented on the same computer device.
[0048]
The input device 1 is configured by, for example, a database reading device that reads out test data of a semiconductor application device (for example, a memory LSI) to be analyzed from a database, and includes an analysis range, the number n of defective elements, and each defect in the semiconductor application device. The coordinate components of the element are input to the arithmetic unit 2. In addition, when the arithmetic device 2 performs the calculation by the method C described later, the input device 1 also inputs a reference value Kc described later.
[0049]
The interval calculation unit 21 calculates the coordinate interval | d | between any two defective elements in a coordinate component (vector component) of a specific dimension among the defective elements input from the input device 1 by the total number of combinations N ( = _n C ₂ = N (n-1) / 2), and the frequency distribution S (| d |) is obtained. When performing the calculation by the method B described later, the interval calculation unit 21 calculates the weighting W (| y2-y1 |), and obtains the frequency distribution S (d) therefrom.
[0050]
The number-of-combinations calculation unit 22 calculates the number of combinations (N−u) by subtracting the number u of combinations where the value of | d | is 0 from the number N of combinations calculated by the interval calculation unit 21. The divisor calculator 23 checks all the divisors f for all the intervals | d | between the defective elements, and obtains the frequency distribution S (f) and the maximum value fmax of the divisors f. The value-based frequency calculation unit 24 calculates the content rate (appearance probability) P (f) = S (f) / (N) of an arbitrary divisor f included in the number of combinations (N−u) obtained by the number-of-combinations calculation unit 22. -U).
[0051]
The expected value function calculator 25 calculates an expected value function T (f) = f × P (f) obtained by multiplying the content (probability of occurrence) P (f) of the divisor f by the corresponding divisor f. Calculate for divisor f.
[0052]
The low-density distribution analysis unit 31 determines whether the value Kabc = fmax / (Nu) obtained by dividing the maximum value fmax of the divisor f from the number of combinations (Nu) is greater than 1 or less than 1. . If this value is greater than 1, the low-density distribution analysis unit 31 determines that the distribution of defective elements included in the semiconductor application device being analyzed is the number of combinations required to determine the distribution of regularity or irregularity. It is determined that (N−u) is a low density distribution that is insufficient.
[0053]
When the low-density distribution analysis unit 31 determines that the distribution is not a low-density distribution, that is, when the Kabc value is 1 or less, the regular distribution analysis unit 32 includes a defective element included in the semiconductor application device to be analyzed. Is determined to include a regular distribution or an irregular distribution. The regularity distribution analyzing unit 32 determines that the distribution of the defective elements is irregular if the values of the expected value function T (f) are all 1 or less, and that the regularity distribution T It is determined that the distribution includes the distribution. The method of determining the distribution of defective elements by the analyzer 3 will be described later in detail.
[0054]
The output device 4 is configured by a display device such as a CRT (Cathode Ray tube) display device or a printing device, and outputs an analysis result in the analysis device 3, that is, the low-density distribution analysis unit 31 or the regular distribution analysis unit 32. The output device 4 may store the analysis result of the analysis device 3 in a medium such as a magnetic disk.
[0055]
Hereinafter, the processing in the inspection analysis apparatus according to this embodiment will be described in detail. In the following description, it is assumed that the analysis range of the failure distribution to be analyzed in the coordinate component Y is ymin ≦ y ≦ ymax. First, the calculation method up to obtaining the expected value function T (f) in the arithmetic unit 2 will be described in detail. The arithmetic unit 2 can perform the calculation according to any one of the methods A to C shown in FIGS.
[0056]
FIG. 2 is a flowchart illustrating an example A of a calculation method until the expected value function T (f) is obtained in the arithmetic device 2. When the analysis range, the number n of defective elements, and the coordinate components of each defective element are input from the input device 1 to the analysis target semiconductor device, the interval calculator 21 analyzes the coordinate range Ymin ≦ y ≦ ymax in the coordinate component Y. , The coordinate interval | d | between any two defective elements is calculated for all the combinations N, and the frequency distribution S (| d |) is obtained (step S101).
[0057]
Next, in the number-of-combinations calculation unit 22, the number u of combinations in which the value of | d | is 0 is calculated from the number of combinations N obtained as an intermediate result when the interval calculation unit 21 obtains the frequency distribution S (| d |). The number of combinations (Nu) removed is obtained (step S102).
[0058]
Next, the divisor calculation unit 23 checks all the divisors f for all the intervals | d | between the defective elements, and calculates the frequency distribution S (f) and the maximum value fmax of the divisor f. It is obtained (step S103). Further, the value-based frequency calculator 24 obtains the content rate (appearance probability) P (f) = S (f) / (Nu) of an arbitrary divisor f included in the number of combinations (Nu) ( Step S104).
[0059]
Then, the expected value function calculator 25 calculates an expected value function T (f) = f × P (f) by multiplying the content (probability of occurrence) P (f) of the divisor f by the corresponding divisor f. Calculation is performed for each divisor f (step S105).
[0060]
When the calculation of the expected value function T (f) is completed for all divisors f, the arithmetic unit 2 determines whether or not there are any other coordinate components to be analyzed (step S106). If there are other coordinate components to be analyzed, the process returns to step S101, and the same processing is performed on the other coordinate components. If there is no other coordinate component to be analyzed, the arithmetic unit 2 ends the processing of this flowchart, and shifts to the analysis of the distribution of defective elements by the analysis unit 3.
[0061]
FIG. 3 is a flowchart illustrating an example B of a calculation method until the expected value function T (f) is obtained in the arithmetic device 2. First, the interval calculation unit 21 obtains a frequency distribution S (y) of n defective elements with respect to the analysis range ymin ≦ y ≦ ymax in the coordinate component Y (step S201). Further, in the interval calculation unit 21, the interval | y2- between the defective elements when S (y1)> 0 and S (y2) (y1, y2 are arbitrary coordinate values in the analysis range: y1 ≠ y2). The weight W (| y2-y1 |) = S (y1) * S (y2) of y1 | is obtained (step S202).
[0062]
Further, the interval calculator 21 obtains the sum of weights W (| y2-y1 |) in which the combination of y1 and y2 is equal to the coordinate interval | d |, that is, the frequency distribution S (| d |) of the coordinate interval | d | (Step S203). The processing in the next steps S204 to S207 is substantially the same as the processing in steps S102 to S105 in the case of the method A, and thus the expected value function T (f) for the divisor f is calculated.
[0063]
When the calculation of the expected value function T (f) is completed for all divisors f, the arithmetic unit 2 determines whether there are any other coordinate components to be analyzed (step S208). If there are other coordinate components to be analyzed, the process returns to step S201, and the same processing is performed on the other coordinate components. If there is no other coordinate component to be analyzed, the arithmetic unit 2 ends the processing of this flowchart, and shifts to the analysis of the distribution of defective elements by the analysis unit 3.
[0064]
FIG. 4 is a flowchart illustrating an example C of a calculation method until the calculation device 2 obtains the expected value function T (f). When performing the calculation by this method, first, a reference value Kc for selecting the method A or the method B is input from the input device 1 to the arithmetic device 2 (step S301).
[0065]
Next, the arithmetic unit 2 obtains a value of K = n / (ymax−ymin) for the analysis range ymin ≦ y ≦ ymax in the coordinate component Y (step S302). Then, the arithmetic unit 2 determines whether the obtained value K is smaller than the input reference value Kc (step S302).
[0066]
When it is determined that K is smaller than Kc, the arithmetic unit 2 performs calculation up to the expected value function T (f) using the method A, that is, according to the flowchart of FIG. 2 (step S304). On the other hand, if it is determined that K is not smaller than Kc, the arithmetic unit 2 calculates up to the expected value function T (f) using the method B, that is, according to the flowchart of FIG. 3 (step S305). However, when the calculation device performs the calculation in step S304 or S305, the determination regarding other coordinate components (steps S106 and S208) is not performed.
[0067]
When the calculation of the expected value function T (f) is completed for all the divisors f in step S304 or S305, the arithmetic unit 2 determines whether there are any other coordinate components to be analyzed (step S306). If there is another coordinate component to be analyzed, the process returns to step S302, and the same processing is performed on the other coordinate component. If there is no other coordinate component to be analyzed, the arithmetic unit 2 ends the processing of this flowchart, and shifts to the analysis of the distribution of defective elements by the analysis unit 3.
[0068]
Hereinafter, a method of analyzing the distribution of defective elements in the analyzer 3 will be described in detail. The analysis of any of the low density distribution, the distribution including the regular distribution, and the irregular distribution can be performed by any of the methods shown in FIG. 5 or FIG.
[0069]
FIG. 5 is a flowchart illustrating an example A of the analysis method of the distribution of defective elements in the analysis device 3. First, the low-density distribution analysis unit 31 obtains a value Kabc = fmax / (Nu) obtained by dividing the maximum value fmax of the divisor f from the number of combinations (Nu) (step S401). Next, the low-density distribution analysis unit 31 determines whether the value Kabc is greater than 1 (step S402).
[0070]
If it is determined that the value Kabc is greater than 1, the low-density distribution analysis unit 31 determines whether the distribution of the defective elements included in the semiconductor application device to be analyzed is a regular or irregular distribution. It is determined that the low-density distribution (B) lacks the required number of combinations (Nu) (step S403). When it is determined that the value Kabc is not larger than 1, the processing shifts to the processing by the regularity distribution analysis unit 32, and the expected value function T (f) is obtained by the expected value function calculation unit 25 and passed from the arithmetic unit 2. It is determined whether all the values are less than or equal to 1 (step S404).
[0071]
When it is determined that all of the expected value functions T (f) are 1 or less, the regularity distribution analysis unit 32 determines that the distribution of the defective elements included in the semiconductor application element to be analyzed is the irregularity distribution ( C) (step S404). On the other hand, when it is determined that the expected value function T (f) includes more than one, it is determined that the distribution (A) includes the regular distribution (step S405).
[0072]
When the state of any failure distribution is determined in step S403, 405, or S406, the analyzer 3 determines whether there is another coordinate component to be analyzed (step S406). If there are other coordinate components to be analyzed, the process returns to step S401, and the same processing is performed on the other coordinate components. If there is no other coordinate component to be analyzed, the analysis device 3 ends the processing of this flowchart, passes the analysis result to the output device 4, and outputs it.
[0073]
FIG. 6 is a flowchart illustrating an example B of a method of analyzing the distribution of defective elements in the analysis device 3. In this example, first, an arbitrary adjustment value Koff for adjusting the criterion of the low density distribution is input from the input device 1 to the analysis device 3 via the arithmetic device 2 (step S451).
[0074]
Subsequent processing is the same as that of the method A, except that step S402 is replaced with step S452. In step S452, the low-density distribution analysis unit 31 determines whether the value Kabc is larger than the adjustment value Koff. If it is determined that the value Kabc is larger than the adjustment value Koff, the process proceeds to step S403. If it is determined that the value Kabc is not larger than the adjustment value Kooff, the process proceeds to step S404. If it is determined in step S407 that there is another coordinate component to be analyzed, the process returns to step S401 as in method A, and does not return to the process in step S451.
[0075]
The fact that the classification is possible as described above will be described based on a graph showing the relationship between the expected value function T (f) and the divisor f shown in FIG. As shown in FIG. 7A, in the regularity distribution A, the value of the expected value function T (f) at a specific divisor f exceeds 1 and vibrates. As shown in FIG. 7B, in the low-density distribution C, the value of the expected value function T (f) at a specific divisor f exceeds 1, and the value of T (f) increases as the value of f increases. To increase. As shown in FIG. 7C, in the irregularity distribution, the value of the expected value function T (f) is 1 or less at an arbitrary divisor f, and converges to 1 as the number of defectives increases. It can be seen from the shape of each such graph that the above classification is possible.
[0076]
As described above, in the inspection and analysis apparatus according to the present embodiment, the analysis range, the number n of defective elements, and the coordinate components of each defective element are input from the input device 1 to the arithmetic unit. 2 and the units 21 to 25, 31, and 32 of the analysis device 3 sequentially perform processing, so that the distribution of defective elements included in the semiconductor application device to be analyzed is (A) a distribution including a regular distribution, It is classified into either B) irregular distribution or (C) low density distribution.
[0077]
That is, the distribution of defective elements included in the semiconductor application device to be analyzed can be classified into the low density distribution C that could not be classified by the prior application (Japanese Patent Application No. 11-1680). As a result, when the time-dependent change in the distribution of the defective elements is observed, when the low-density distribution C changes to the regular distribution A, the cause of the defect indicating the regular distribution increases. When the distribution changes to distribution B, it can be determined that the number of defective causes indicating the irregular distribution has increased. As described above, the trend analysis for each cause of failure becomes easy, and the yield of the semiconductor application device to be inspected can be improved.
[0078]
[Second embodiment]
FIG. 8 is a block diagram illustrating a configuration of the inspection analysis apparatus according to the present embodiment. The difference from the first embodiment (FIG. 1) is that the defect number analyzing device 5 is further provided, and the case where the calculation in the arithmetic unit 2 and the analysis in the analyzing device 3 are performed are performed in the defect number analyzing device 5. That is, it follows the analysis result of the number of defects.
[0079]
The failure number analysis device 5 changes the distribution of the failure elements included in the semiconductor application device to be analyzed in accordance with the following (D) to (F) according to the number n of the failure elements input from the input device 1. Classify. The details of the method of classifying the distribution of defective elements by the defect number analyzer 5 will be described later.
[0080]
(D) The number of defective elements is so large that the analysis of the distribution is not required (all elements are defective, or the density of defective elements is higher than a predetermined density)
(E) No defective element
(F) The number of defective elements is so small that the analysis of the distribution is not necessary (the number of defective elements is two or less, or the density of defective elements is a predetermined density or less)
[0081]
The arithmetic unit 2 and the analyzing unit 3 perform the processing when the defective element distribution is not classified into any of the above (D) to (F) in the defective number analyzing apparatus 5 (the number of defective elements is in a range other than the above range). ), Calculate and analyze, respectively. That is, when the number n of the defective elements is in a range other than the above (D) to (F), the calculation in the interval calculator 21 is performed. Then, as shown in the first embodiment, the data is classified into any one of the distribution including the regular distribution (A), the irregular distribution (B), and the low density distribution (C). In addition, the defect number analysis device 5 classifies the case where all the elements are not defective but the defective elements are equal to or higher than a predetermined density as (D), and further classifies the case where all the elements are defective as (D_all). Is also good.
[0082]
Hereinafter, a method of classifying the distribution of defective elements in the defect number analysis device 5 will be described in detail.
[0083]
FIG. 9 is a flowchart illustrating an example A of a method of classifying the distribution of defective elements in the defect number analysis device 5. The method shown in this flowchart corresponds to the case where the distribution of defective elements is classified into (D) to (F).
[0084]
First, the minimum value nmin and the maximum value nmax of the number of defective elements for causing the analysis device 3 to analyze the failure distribution are input from the input device 1 to the failure number analysis device 5 (step S501). Next, the analysis device compares the number n of defective elements included in the semiconductor application device to be analyzed with the input minimum value nmin and maximum value nmax (step S502).
[0085]
As a result of the comparison, first, it is determined whether or not the number n of defective elements is 0 (step S503). If it is determined that the number n of defective elements is 0, that is, if no defective element is included in the semiconductor application device to be analyzed, the number-of-defectives analyzer 5 determines the distribution (E) that does not include defective elements. It is determined that there is (Step S504).
[0086]
If it is determined that the number n of defective elements is not 0, it is further determined whether the number n of defective elements is smaller than the input minimum value nmin (step S505). If it is determined that the number n of defective elements is smaller than the minimum value nmin, the number-of-failures analyzer 5 determines that the defective elements have a distribution (F) that is so small that the analysis of the distribution is not required (step). S506).
[0087]
If it is determined that the number n of defective elements is not smaller than the minimum value nmin, it is further determined whether the number n of defective elements is smaller than the input maximum value nmax (step S507). If it is determined that the number n of defective elements is smaller than the maximum value nmax, the arithmetic unit 2 and the analyzing device 3 are made to analyze the failure distribution of the semiconductor application device to be analyzed (step S508). This method is shown in the first embodiment, and is thereby determined to be one of the distributions (A) to (C) described in the first embodiment.
[0088]
If it is determined that the number of defective elements is not smaller than the maximum value nmax, the defect number analyzer 5 determines that the defective elements have a distribution (D) that is so large that analysis of the distribution is not necessary (step). S509).
[0089]
When the state of any failure distribution is determined in steps S504, S506, S508, or S509, the failure number analysis device 5 determines whether there is another coordinate component to be analyzed (step S510). If there is another coordinate component to be analyzed, the process returns to step S502, and the same processing is performed on the other coordinate component. If there is no other coordinate component to be analyzed, the failure number analysis device 5 ends the processing of this flowchart, and passes the analysis result by the analysis device 3 or the failure number analysis device 5 to the output device 4 for output. .
[0090]
FIG. 10 is a flowchart illustrating an example B of a method of classifying the distribution of defective elements in the defect number analysis device 5. The method shown in this flowchart corresponds to a case where the distribution of defective elements is classified into (D), (D_all), (E), and (F).
[0091]
This classification method B is almost the same as the classification method A described above. However, if it is determined in step S503 that the number n of defective elements is not 0, the failure number analysis device 5 sends the failure analysis device 5 to the semiconductor application device to be analyzed. It is determined whether all the included semiconductor elements are defective (step S551). If it is determined that all of them are defective, the defect number analyzer 5 determines that the distribution is all (D_all) (step S552). On the other hand, if it is determined that there is any non-defective element, the process proceeds to step S505.
[0092]
When the state of any failure distribution is determined in steps S504, S506, S508, S509, or S552, the failure number analysis device 5 determines whether there is another coordinate component to be analyzed (step S510). . If there is another coordinate component to be analyzed, the process returns to step S502, and the same processing is performed on the other coordinate component. If there is no other coordinate component to be analyzed, the failure number analysis device 5 ends the processing of this flowchart, and passes the analysis result by the analysis device 3 or the failure number analysis device 5 to the output device 4 for output. .
[0093]
As described above, in the inspection and analysis apparatus according to this embodiment, the analysis range, the number n of defective elements, and the coordinate components of each defective element in the semiconductor application device to be analyzed are input from the input device 1 to determine the number of defective elements. By causing the analysis units 5, the arithmetic unit 2, and the units 21 to 25, 31, and 32 of the analysis unit 3 to sequentially perform processing, the distribution of defective elements included in the semiconductor application device to be analyzed can be calculated as described in (A) above. To (F). Alternatively, they are classified into seven categories: (A), (B), (C), (D), (D_all), (E), and (F). Thereby, the following effects can be obtained.
[0094]
For example, in a memory LSI which is an example of a semiconductor application device, the inspection according to this embodiment is performed by using the entire memory element group included in the LSI or the memory element group divided into several ranges as an analysis range. Analyze with an analyzer. Thereby, how the memory LSIs having the above distributions (A) to (F) are distributed on the memory LSI wafer, or having the above distributions (A) to (F) on the memory LSIs The distribution of the memory element blocks can be mapped and output from the output device 4.
[0095]
This output data output converts the defective state of the element group in the analysis range (for example, coordinate data of tens of thousands of defective elements) into a simple character string from “A” to “F” or a color or pattern corresponding to the character string. Can be converted, and the effect of reducing the data size can be increased. For this reason, it is possible to reduce the load when the output result is distributed to a related person via a communication means such as the Internet or is stored in a storage medium. In addition, the effect of reducing the data size facilitates comparative analysis between a large number of memory LSIs and memory LSI wafers, so that a trend analysis on a longer manufacturing process can be performed as compared with other methods.
[0096]
For example, in a conventional inspection analysis apparatus (method), one to several tens of memory LSIs on a wafer are sampled and analyzed to represent a defective element distribution of the whole wafer, or a magazine in which several tens of wafers are stored. Sampling analysis of one to several wafers in a specific slot to represent the defective element distribution of the entire magazine, or sampling analysis of about 0.01% to 0.1% of all lots manufactured annually. This was regarded as the tendency of defective element distribution of the lot.
[0097]
On the other hand, in the inspection analyzer according to the present embodiment, the data size can be reduced to tens of thousands by setting the range including tens of thousands of memory elements as the analysis range, and the sampling density can be reduced. It can be increased to 10,000 times or more. In addition, features of the distribution maps (A) to (F) are extracted using image recognition technology, and the distribution features are compared with design conditions, manufacturing conditions, appearance inspection results, and characteristic test results such as electrical tests. By doing so, the cause of the failure can be estimated in a short time.
[0098]
As an example, if the range having the distribution (E) not including the defective element is large on the memory element block and the memory LSI wafer, it can be said that the yield of the product is high. On the other hand, due to an adjustment error or a design error in the manufacturing apparatus, a defect having a large number of defective elements (D) due to a partial defect of the element structure being detected in the appearance inspection or not operating electrically increases. For example, the condition change, the inspection result, and the distribution change are stored as one case, and then, when the distribution changes from (E) to (D), whether or not the distribution is in the same place as in the previous case, and the change has occurred. The number of ranges can be compared.
[0099]
In addition, if the range having the distribution (F) in which the number of defective elements is small increases, it is possible to expect the drop of fine dust having a size of about 1 to 2 elements. As for the distributions (A), (B), and (C), as in the first embodiment, the distribution of in-wafer distributions of defective causes indicating a regular distribution or defective causes indicating an irregular distribution and between manufacturing lots It is possible to analyze changes over time and trends between different varieties.
[0100]
In addition, regarding the distributions (D) and (F), it is more effective to provide a standard value for the defective element density in order to reduce variations in analysis time and analysis accuracy.
[0101]
[Third Embodiment]
FIG. 11 is a block diagram illustrating a configuration of the inspection analysis apparatus according to the present embodiment. The difference from the second embodiment (FIG. 8) is that a mixture distribution analyzer 6 is further provided.
[0102]
The mixture distribution analysis device 6 is configured such that when the low-density distribution analysis unit 31 does not determine that the distribution of defective elements of the semiconductor application device to be analyzed is the low-density distribution, When it is determined that the data includes an irregular distribution or an irregular distribution, this is analyzed in further detail.
[0103]
More specifically, the mixture distribution analyzer 6 includes an expectation function T (f) calculated by the expectation function calculator 25, a reference value Kca described later, and a regular distribution predicted as described later. Based on the set of periods, the composition ratio of the distribution of the defective elements included in the semiconductor application device to be analyzed includes the irregularity distribution and the type of regularity distribution in what ratio.
[0104]
Hereinafter, the method of analyzing the mixture distribution in the mixture distribution analyzer 6 will be described in detail. FIG. 12 and FIG. 13 are flowcharts illustrating an example of a method of analyzing a mixture distribution in the mixture distribution analysis device 6.
[0105]
First, a reference value Kca (where Kca ≦ 1.5 or 1) for analyzing the mixture distribution is input from the input device 1 to the mixture distribution analysis device 6 via the defect number analysis device 5, the arithmetic device 2 and the analysis device 3. <Kca <L / 2 (L is a divisor indicating the maximum value T (f) max of the expected value function T (f)) is input (step S601). Similarly, a set Lmd (i) (i: natural number) of regularity distribution periods expected to be included in the semiconductor application element to be analyzed is input from the input device 1 to the mixture distribution analysis device 6 (step S602). ).
[0106]
Next, the mixture distribution analyzer 6 compares the values of the maximum value T (f) max of the expected value function and the expected value function T (f) with values such as the divisor L and the reference value Kca. The relationship is examined (step S603).
[0107]
Based on the result of this comparison, the mixture distribution analyzer 6 first determines whether or not the maximum value T (f) max of the expected value function is 1 or less (step S604). When it is determined that the maximum value T (f) max of the expected value function is 1 or less, the mixture distribution analyzer 6 determines that the distribution is an ideal irregularity distribution (C) (step S605).
[0108]
When it is determined that the maximum value T (f) max of the expected value function is greater than 1, the mixture distribution analyzer 6 further determines whether the maximum value T (f) max of the expected value function is equal to the divisor L. A determination is made (step S606). When it is determined that the maximum value T (f) max of the expected value function is equal to the divisor L, the mixture distribution analyzer 6 determines that the distribution is a regular distribution (A) having only one type of period L. (Step S607).
[0109]
When it is determined that the maximum value T (f) max of the expected value function is not equal to the divisor L, the mixture distribution analyzer 6 determines that the mixture is a mixture of a regular distribution and an irregular distribution ( Step S608). Then, the mixture distribution analyzer 6 further determines whether or not the maximum value T (f) max of the expected value function is equal to a half of the divisor L (step S609). When it is determined that the maximum value T (f) max of the expected value function is equal to one half of the divisor L, the mixture distribution analyzer 6 determines two types of periods L1 and L2 (where L = L1 + L2). (AW) is determined (step S610).
[0110]
When it is determined that the maximum value T (f) max of the expected value function is not equal to one half of the divisor L, the mixture distribution analyzer 6 sets the maximum value T (f) max of the expected value function to 1 It is determined whether the value is larger than the reference value Kca (step S611).
[0111]
When it is determined that the maximum value T (f) max of the expected value function is greater than 1 and smaller than the reference value Kca, the mixture distribution analyzer 6 sets the divisor L to the set Lmd ( It is determined whether it is included in i) (step S612). When it is determined that the divisor L is included in the set of periods Lmd (i), the mixture distribution analyzer 6 includes the mixture distribution (Ca) including the regular distribution of the period L and mainly having the irregular distribution. ) Is determined (step S613).
[0112]
When it is determined that the divisor L is not included in the cycle set Lmd (i), the mixture distribution analyzer 6 determines whether the cycles L1 and L2 are included in the cycle set Lmd (i) ( Step S614). When it is determined that the periods L1 and L2 are included in the set of periods Lmd (i), the mixture distribution analyzer 6 includes two types of regular distributions of the period L1 and the period L2 and the irregular distribution Is the main mixture distribution (Caw) (step S615).
[0113]
On the other hand, when it is determined in step S611 that the maximum value T (f) max of the expected value function is equal to or greater than 1 or equal to or greater than the reference value Kca, the mixture distribution analyzer 6 determines the maximum value T (f ) It is determined whether max is greater than half the divisor L and smaller than the divisor L (step S616). When the maximum value T (f) max of the expected value function is equal to or smaller than half of the divisor L, the mixture distribution analyzer 6 determines that the maximum value T (f) of the expected value function is equal to or larger than the reference value Kca. , Is smaller than half the divisor L (step S617). Then, the mixture distribution analyzer 6 determines whether or not the divisor L is included in the cycle set Lmd (i) (Step S618).
[0114]
When it is determined in step S616 that the maximum value T (f) max of the expected value function is larger than half of the divisor L and smaller than the divisor L, or in step S618, the divisor L is a set Lmd of periods. If it is determined that the expected value function is included in (i), the mixture distribution analyzer 6 determines that both the expected value functions T (L-1) and T (L + 1) are larger than the reference value Kca, and It is determined whether is also smaller (step S619).
[0115]
When it is determined that both the expected value functions T (L-1) and T (L + 1) are larger than the reference value Kca and smaller than the divisor L, the mixture distribution analyzer 6 determines the irregularity distribution. It is determined that the regular distribution of the period L due to the cause p that also includes and affects adjacent coordinates is the main distribution (Ap) (step S620).
[0116]
If any one of the expected value functions T (L-1) and T (L + 1) is determined to be equal to or less than the reference value Kca or equal to or more than a few L, the mixture distribution analyzer 6 determines that It is determined that the regular distribution having the period L due to the cause q not including the regular distribution and affecting the adjacent coordinates is the main distribution (Ac) (step S621).
[0117]
If it is determined in step S618 that the divisor L is not included in the cycle set Lmd (i), the mixture distribution analyzer 6 determines whether the cycles L1 and L2 are included in the cycle set Lmd (i). It is determined whether or not it is (step S622). When it is determined that the periods L1 and L2 are included in the set of periods Lmd (i), the mixture distribution analyzer 6 determines that both the expected value functions T (L1 + L2-1) and T (L1 + L2 + 1) are equal to the reference value Kca. It is further determined whether it is greater than and less than half the divisor L (step S623).
[0118]
When it is determined that both the expected value functions T (L1 + L2-1) and T (L1 + L2 + 1) are larger than the reference value Kca and smaller than a half of the divisor L, the mixture distribution analyzer 6 determines It is determined that two types of regular distributions (AWp) including the irregular distribution and the period L1 and the period L2 due to the cause p affecting adjacent coordinates are the main distributions (AWp) (step S624).
[0119]
If it is determined that either one of the expected value functions T (L1 + L2-1) and T (L1 + L2 + 1) is equal to or less than the reference value Kca or equal to or more than half the divisor L, the mixture distribution analysis is performed. The device 6 determines that the distribution (AWc) is mainly two types of regular distributions including a period L1 and a period L2 due to a cause q that includes an irregular distribution and does not affect adjacent coordinates (step S1). S625).
[0120]
If it is determined in steps S614 and S622 that the periods L1 and L2 are not included in the set of periods Lmd (i), the mixture distribution analyzer 6 has a mixture distribution (AX) whose principal component is not clear. Is determined (step S626).
[0121]
When any of the failure distribution states is determined in steps S605, S607, S610, S613, S615, S620, S621, S624, S625, and S626, the mixture distribution analyzer 6 determines the other coordinate components to be analyzed. It is determined whether or not there is (step S627). If there is another coordinate component to be analyzed, the process returns to step S603, and the same processing is performed on the other coordinate component. If there is no other coordinate component to be analyzed, the failure number analysis device 5 ends the processing of this flowchart, and passes the analysis result by the analysis device 3 or the failure number analysis device 5 to the output device 4 for output. .
[0122]
The fact that the classification can be performed as described above will be described based on graphs showing the relationship between the expected value function T (f) and the divisor f shown in FIGS. As shown in FIG. 14A, in the mixture distribution Ac in which the regular distribution of the period L is main, the value of the expected value function T (f) is larger than 1 for any divisor f, and the period L Less than. As shown in FIG. 14B, in the mixed distribution Ca including the regular distribution with the period L and the irregular distribution being the main, the expected value function T (f) when the divisor f is equal to the period L is shown. Greater value.
[0123]
As shown in FIG. 15A, in the regularity distribution AW having two types of periods L1 and L2, the value of the expected value function T (f) when the divisor f is equal to the sum L of the periods L1 and L2 is It is equal to the average L / 2 of the two periods. As shown in FIG. 15B, in the mixture distribution AWc mainly composed of two types of periods L1 and L2, the value of the expected value function T (f) at any divisor f exceeds 1. Does not exceed the value of L / 2. As shown in FIG. 15C, in the case of a regular distribution Caw that includes a regular distribution composed of two types of periods L1 and L2 and has a main irregular distribution, the tendency is close to the irregular distribution C. When the number f is equal to the sum of the periods L1 and L2, the value of the expected value function T (f) is larger.
[0124]
As shown in FIG. 16A, in the regular distribution Ap of the period L due to the cause p which also affects the adjacent coordinates, the expected value function T (f) when the divisor f is equal to the period L is It is equal and the maximum value T (f) max of the expected value function. As shown in FIG. 16B, in a distribution AWp including two types of regular distributions of a period L1 and L2 due to a cause p which also causes adjacent coordinates, an expected value function T when the divisor f is equal to L1 + L2. (F) is equal to the average of the periods L1 and L2, and becomes the maximum value T (f) max of the expected value function.
[0125]
As described above, in the inspection and analysis apparatus according to this embodiment, the analysis range, the number n of defective elements, and the coordinate components of each defective element in the semiconductor application device to be analyzed are input from the input device 2 to determine the number of defective elements. The analysis unit 5, the arithmetic unit 2, and the units 21 to 25, 31, and 32 of the analysis device 3 and the mixture distribution analysis device 6 sequentially perform processing, so that the distribution of defective elements included in the semiconductor application device to be analyzed ( A) or (B) is further classified as follows.
[0126]
(C) Ideal irregularity distribution
(A) Regular distribution consisting of only one type of period L
(AW) Mixed distribution composed of two types of periods L1 and L2
(Ca) A mixed distribution that includes a regular distribution with a period L and is mainly irregular.
(Caw) A mixed distribution including two types of regular distributions, that is, a period L1 and a period L2, and in which an irregular distribution is mainly used.
(Ap) Distribution mainly including regular distribution with period L due to cause p that includes irregular distribution and also affects adjacent coordinates
(Ac) The mixture distribution analyzer 6 is a distribution that includes an irregular distribution and is mainly a regular distribution with a period L due to a cause q that does not affect adjacent coordinates.
(AWp) A distribution mainly including two types of regular distributions of a period L1 and a period L2 due to a cause p including an irregular distribution and affecting adjacent coordinates.
(AWc) A distribution mainly including two types of regular distributions of a period L1 and a period L2 due to a cause q that includes an irregular distribution and does not affect adjacent coordinates.
(AX) Mixture distribution whose principal component is not clear
[0127]
Since such classification is possible, the following effects can be obtained according to the inspection and analysis apparatus for a semiconductor application device according to this embodiment.
[0128]
First, the inspection analysis apparatus according to the present embodiment extracts a regularity hidden in a seemingly irregular distribution or extracts a small regularity hidden in a clear regularity distribution with high sensitivity. can do. That is, a typical failure distribution as shown in FIGS. 14 to 16 can be identified.
[0129]
In particular, regarding the causes of failures p and q, in the design of a semiconductor application device, for example, when dimensions or positions are not correctly drawn with respect to a wiring, a circuit, or an electrode for selecting a specific element in a memory element group of a memory LSI This may cause the related elements to become electrically defective. If an adjacent element has no relation to the cause, the element does not lead to an electrical failure.
[0130]
On the other hand, in the manufacture of a semiconductor application device, for example, the element at the center of the memory element group of the memory LSI is evenly and multiplely surrounded on all sides by other elements, but the element at the periphery is one side or There are elements whose two sides are not surrounded by other elements. In the memory element group, depending on whether or not the element itself is surrounded by other elements, and how much is surrounded by other elements, the ease of heat transmission, the contact area with reactive gas and liquid, the polishing time, Since the abrasion resistance and the reflection characteristics and scattering characteristics during light irradiation are different, the shape and electrical characteristics of the manufactured memory element tend to be different depending on the central part or the peripheral part of the memory element group. This tendency changes continuously along the line connecting the central portion and the peripheral portion. Therefore, when the defective element of interest is an electrical failure, the adjacent element is also often an electrical failure.
[0131]
For this reason, the inspection and analysis apparatus according to the present embodiment makes it possible to identify whether or not the cause of failure indicating the regular distribution of the same period L is the cause of failure p which also affects the adjacent element. This is important in increasing the accuracy of determining whether the distribution is due to design or manufacturing.
[0132]
If the specific distribution is biased to a specific location (right side or left side) on the memory LSI, the manufacturing conditions may not be uniform with respect to the wafer, or the specific distribution may be in a specific direction on the LSI wafer. If there are several places where the design rules are not adhered to, such as an LSI mask, a reticle, or an internal circuit, which is repeatedly arranged on the wafer, there are several causes of failure. Candidates can be estimated. Therefore, by tracing the change in the number of such distributions, it is possible to know whether the change in the distribution is increasing, decreasing, or suddenly occurring over time.
[0133]
[Fourth Embodiment]
FIG. 17 is a block diagram illustrating a configuration of the inspection analysis apparatus according to the present embodiment. The difference from the third embodiment (FIG. 11) is that a critical distribution analyzer 7 is further provided.
[0134]
The critical distribution analysis device 7 determines that the value of fmax / (N−u) is 1 when the low-density distribution analysis unit 31 determines that the distribution of defective elements of the semiconductor application device to be analyzed is the low-density distribution. If it is larger than 1, it is further determined whether or not this value is within a predetermined range near 1 (such a distribution is called a critical distribution). If the critical distribution analysis device 7 determines that the distribution is a critical distribution, the distribution of defective elements of the semiconductor application device to be analyzed is determined according to the expected value function T (2) when the value of the divisor f is 2. Analyze whether a regular distribution is included or irregular.
[0135]
More specifically, it is determined whether the value of T (2) when the value of the divisor f is 2 is close to 1 or close to 2. When it is close to 1, the distribution of defective elements is determined to be an irregular distribution. On the other hand, when it is close to 2, it is determined that the regular distribution including the even period is included. This is because, in the irregular distribution, the probability of occurrence P (f) of the divisor f is inversely proportional to the value of the divisor f, so that f = 2 is most likely to be included in the distribution, and (N−u) T (2) converges to 1 as fast as T increases, while the regular distribution has many periods (even periods) where f = 2 is a divisor, and unless the period is odd, T (2) increases as (N−u) increases. (2) is based on converging to 2 fastest.
[0136]
The critical distribution analyzer 7 also analyzes in detail the expected value function T (f) for the divisor f in relatively small prime numbers from one to two digits. That is, when the value of the expected value function T (f) is close to 1, the distribution of defective elements is determined to be an irregular distribution, and when the value of the expected value function T (f) is close to f, It is determined that a regular distribution is included. This is because the expected value function T (f) converges relatively quickly even if the cycle is an odd number, if the divisor f includes relatively small prime numbers such as 3, 5, and 7.
[0137]
Hereinafter, a method of analyzing the distribution of defective elements in the critical distribution analyzer 7 will be described in detail. FIG. 18 is a flowchart illustrating an example of a method of analyzing the distribution of defective elements in the critical distribution analysis device 7.
[0138]
First, in the third embodiment, the critical distribution analysis device 7 is input from the input device 1 via the defect number analysis device 5, the arithmetic device 2, and the analysis device 3 to the critical distribution analysis device 7 in the same manner as that used for the analysis of the mixture distribution. A reference value Kca (where Kca ≦ 1.5 or 1 <Kca <f / 2) is input (step S701). Next, the critical distribution analyzer 7 compares the expected value function T (f) relating to the divisor f composed of relatively small prime numbers up to one or two digits including 2, with the reference value Kca and the divisor f, The magnitude relationship is checked (step S702).
[0139]
The critical distribution analyzer 7 determines whether the expected value function T (2) for the divisor 2 is larger than Kca and smaller than 2 which is the value of the divisor itself from the result of the comparison (step S703). . When it is determined that the expected value function T (2) is larger than Kca and smaller than 2, the critical distribution analysis device 7 determines that the low density distribution (Ab) having a high possibility of including a regular distribution having an even period. Is determined (step S704).
[0140]
When it is determined that the expected value function T (2) is equal to or smaller than Kca or equal to or larger than 2, the critical distribution analysis device 7 further determines one of the divisors to be compared from the result of the comparison in step S702. It is determined whether the expected value function T (f) for f is smaller than Kca and smaller than the value itself of the divisor f (step S704). If any one of the expected value functions T (f) is determined to be larger than Kca and smaller than the divisor f, the critical distribution analysis device 7 determines that the probability that the periodicity distribution includes an odd-numbered regular distribution is high. It is determined that the distribution is the density distribution (Ab) (step S705).
[0141]
If it is determined that any of the expected value functions T (f) is equal to or smaller than Kca or equal to or larger than 2, the critical distribution analysis device 7 does not include the regular distribution and may have the irregular distribution. Is determined to be a high low-density distribution (Cb) (step S705).
[0142]
When the state of any failure distribution is determined in step S704, S706, or S707, the critical distribution analysis device 7 determines whether there is another coordinate component to be analyzed (step S708). If there is another coordinate component to be analyzed, the process returns to step S702, and the same processing is performed on the other coordinate component. If there is no other coordinate component to be analyzed, the critical distribution analysis device 7 ends the processing of this flowchart, passes the analysis result by the critical distribution analysis device 7 to the output device 4, and outputs it.
[0143]
The fact that the classification is possible as described above will be described based on a graph showing the relationship between the expected value function T (f) and the divisor f shown in FIG. As shown in FIG. 19A, in the low-density distribution Ab estimated to include a regular distribution having an even period, the value of the expected value function T (2) when the divisor f is 2 is larger than Kca. , Smaller than 2. As shown in FIG. 19B, in the low-density distribution Cb estimated to be an irregular distribution, the value of the expected value function T (f) is equal to or smaller than Kca at any divisor f.
[0144]
As described above, in the inspection and analysis apparatus according to this embodiment, the analysis range, the number n of defective elements, and the coordinate components of each defective element in the semiconductor application device to be analyzed are input from the input device 1 to determine the number of defective elements. The analysis unit 5, the arithmetic unit 2, and the respective units 21 to 25, 31, 32 of the analysis unit 3, the mixture distribution analysis unit 6, and the critical distribution analysis unit 7 are sequentially processed to be included in the semiconductor application device to be analyzed. The distribution of defective elements (C) is further classified as follows.
[0145]
(Ab (even period)) Low density distribution that is likely to include a regular distribution with an even period
(Ab (odd period)) A low-density distribution that is likely to include an odd-numbered regular distribution
(Cb) Low density distribution likely to be irregular distribution
[0146]
Since such classification is possible, the following effects can be obtained according to the inspection and analysis apparatus for a semiconductor application device according to this embodiment.
[0147]
That is, in the actual defective element distribution, it is not unusual for irregularities and regularities to be mixed, and for the distribution of defective elements to be difficult to analyze due to low density. In order to enhance the practicality of the distribution analysis technique, it is important that the intermediate distributions (Cb) and (Ab) can be analyzed as described above. In particular, the inspection and analysis apparatus of this embodiment can improve the accuracy of identifying irregularities and regularities in a low-density defective element distribution. Can be identified, and a time margin until an appropriate countermeasure is planned can be secured.
[0148]
[Fifth Embodiment]
FIGS. 20 to 23 are block diagrams showing the configuration of the inspection analysis apparatus according to this embodiment. These correspond to those of the first to fourth embodiments, respectively, but differ from the respective embodiments in that a history analysis device 8 and a history database 81 are further provided.
[0149]
The history analysis device 8 stores the analysis results of the analysis device 3 (and the number of defective devices 5, the mixture distribution analysis device 6, and the critical distribution analysis device 7; hereinafter, these are referred to as the analysis device 3) in the history database 81. Compare with the stored history. The history analysis device 8 estimates the change state of the distribution of defective elements included in the semiconductor application device to be analyzed based on the comparison result, and outputs the estimated change state to the output device 4. The history analysis device 8 stores the analysis result of the analysis device 3 in the history database 81 in association with the already stored history.
[0150]
Hereinafter, a method in which the history analysis device 8 compares and associates the analysis result of the analysis device 3 with the history stored in the history database 81 will be described. FIG. 24 is a flowchart illustrating a method of associating with a history in the history analysis device 8.
[0151]
According to the methods described in the first to fourth embodiments, the history analysis device 8 analyzes the distribution of defective elements of the semiconductor application device to be analyzed by the analysis device 3 and determines the type thereof. Then, the analysis result is compared with the history stored in the history database 81 (step S801). The history analyzer 8 determines whether the newly obtained analysis result has changed from the past analysis results accumulated in the history database as a result of the comparison (whether the number of defects has increased or whether the type of distribution has changed). Is determined (step S802).
[0152]
If it is determined that the analysis result has changed, the history analysis device 8 estimates that the yield has decreased if the number of defects has increased, and that the cause of the defect has changed if the type of distribution has decreased. Is the analysis result with the history (step S803). Further, as a result of the comparison, the history analysis device 8 determines whether the operator, the manufacturing device, the operation time, and the location where the defect has occurred have regularity (step S804).
[0153]
If it is determined that the failure occurrence part has regularity, the history analysis device 8 stores the analysis result of the analysis device 3 and the regularity in the history database 81 in association with the history so far (step S1). S805). Then, the process proceeds to step S808. If it is determined that there is no regularity in the location where the failure has occurred, the history analysis device 8 stores the analysis result of the analysis device 3 in the history database 81 in association with the history so far (step S806). Then, the process proceeds to step S808.
[0154]
On the other hand, if it is determined in step S802 that there is no change in the analysis result, the history analysis device 8 estimates that the distribution of the defective elements is in a stable state, and uses this as the analysis result with the history (step S807). . Then, the process proceeds to step S808.
[0155]
In step S808, it is determined whether there are any other coordinate components to be analyzed. If there is another coordinate component to be analyzed, the process returns to step S801, and the same processing is performed on the other coordinate component. If there is no other coordinate component to be analyzed, the history analysis device 8 ends the processing of this flowchart, and passes the analysis result to the output device 4 for output.
[0156]
As described above, in the inspection analysis apparatus according to this embodiment, a change in the analysis result by the analysis apparatus 3 is recorded as a history in the history database 81. In particular, if there is a regularity in the location where the failure occurs, the regularity is also recorded in association with the analysis result. Therefore, by examining the history of the analysis results accumulated in the history database 81, it is possible to accurately associate the cause with the cause of the failure.
[0157]
In the inspection analyzers according to the above-described first to fourth embodiments, the examiner takes out the analysis results at each time point and associates them with each other in order to check the temporal change of the analysis results. Had to be weighed. In contrast, in the inspection analysis apparatus according to the present embodiment, a new analysis result is recorded in association with the history of past analysis results, and the regularity of change is also recorded in association with the analysis result. Therefore, the labor of the inspector can be reduced.
[0158]
[Result output example]
Hereinafter, an example of an output from the output device 4 will be described with respect to a result of analyzing the distribution of defective elements included in the semiconductor application device by the inspection and analysis device described in the first to fifth embodiments.
[0159]
FIGS. 30 to 32 are diagrams illustrating examples of output from the output device 4 when the inspection and analysis device of the semiconductor application device described in the first to fifth embodiments is applied.
[0160]
The output example shown in FIG. 30 will be described. As an example of analyzing a failure distribution, in the case of a failure distribution having a density gradient in which the failure density increases from left to right, the failure distribution is divided into eight parts on the left and right, and the failure distribution regularity of each divided region is examined.
[0161]
In the case where the irregularity distribution indicates a density gradient, a region determined to be E at the left end indicates that no defect is included in the region. The area determined to be F on the right side of the E area includes a defect in the area, but has a very small number of several to several tens, so the analysis of the regularity / irregularity distribution is performed. The low density is not suitable for
[0162]
The area determined to be B on the right of the F area has a higher defect density than the E area, but the defect density is high even if the regularity / irregularity distribution is analyzed to detect the presence or absence of the regularity distribution. Indicates too low.
[0163]
Although the value of T (f) converges to 1 in the region determined to be Cb on the right of the B region as compared with the B region, the value of fmax / (N−u), that is, the Kabc, is threshold. It is larger than the value of 1 or Koff, which indicates that defects are not concentrated.
[0164]
The region determined to be C on the right of Cb has a value of fmax / (N−u), that is, Kabc is smaller than 1 or Koff, which is a threshold value, and the value of T (f) is an analysis target f Converges to approximately 1 or less in the range of, indicating that the distribution is dense enough to be judged as an irregular distribution.
[0165]
The D region on the right side of the C region is a region having a defect density exceeding the upper limit of the defect density mainly for the purpose of shortening the analysis time, and indicates that the defect density is higher than the C region. Although the setting of the upper limit is not always necessary, the analysis time of the regularity / irregularity distribution is almost proportional to the square of the defect density, so that the effect of reducing the defect analysis time is great.
[0166]
The DALL regions on the right and right of the D region indicate that all the elements in the corresponding region are defective.
[0167]
In the above eight areas, the area suitable for analysis of the regularity / irregularity distribution is C and the density gradient is continuous. It can be estimated that the defect distribution has spread.
[0168]
On the other hand, when the regularity distribution (one cycle) shows a density gradient, the region from E to B at the left end is the same as the above example, and the region does not include a defect corresponding to the defect. Or it indicates that the defect density is too low to judge the regularity / irregularity.
[0169]
In the Ab area on the right side of the B area, the value of T (f) in the specific period L is more than 1 in the specific period L, but the value of fmax / (N−u), that is, Kabc, is greater than the B area. Indicates that the value is greater than the value 1 or Koff.
[0170]
The Ca region to the right of the Ab region is smaller than the value of fmax / (Nu), that is, Kabc is 1 or Koff, which is the threshold value, indicating that defects are not concentrated. It also indicates that the maximum value of T (f), T (f) max = T (L), is greater than 1.
[0171]
In the region determined to be Ca on the right of the Ab region, the value of Kabc is larger than the threshold value of 1 or Koff. I This indicates that defects are more concentrated. Also, it indicates that the maximum value of T (f), T (f) max = T (L), is larger than 1, but not prominent (T (f) max is close to 1).
[0172]
In the region determined to be Ac on the right of the Ca region, the maximum value of T (f) T (f) max = T (L) is L / 2 <T (L) <L, and Indicates that the density of the regular distribution including the period L is higher than that of the Ca region, but the mixture distribution includes a small amount of the irregular distribution.
[0173]
The area determined to be A on the right of the Ac area is the value of fmax / (Nu), that is, Kabc is smaller than 1 or Koff, which is the threshold value, and the maximum value T (f) of T (f) ) Max = the value of T (L) protrudes from 1 and the value of T (L) is close to L, and the distribution is a high-density distribution sufficient to determine that the distribution is a regular distribution having a single period L. It indicates that.
[0174]
The D region on the right end is a region with a defect density exceeding the upper limit of the defect density mainly for the purpose of shortening the analysis time, and indicates that the defect density is higher than the A region.
[0175]
In the above eight regions, the region suitable for analysis of the regularity / irregularity distribution is A and the density gradient is continuous. It can be estimated that the failure distribution has spread.
[0176]
If the mixture distribution (two types of periods) shows a density gradient, a region suitable for analysis of the regularity / irregularity distribution is the mixture distribution AW having two types of periods L1 and L2, and Since the density gradient is continuous as in the previous two examples, it can be estimated that the regularity distribution of the same cause of failure as in the AW area is a failure distribution spread with a density gradient.
[0177]
When the mixture distribution (having one type of cycle and affecting adjacent coordinates) shows a density gradient, a region suitable for analyzing the regularity / irregularity distribution has a single period L, and The defect distribution Ap is characterized in that the values of T (L-1) and T (L + 1) before and after are close to the maximum value of T (f) T (f) max = T (L), and Since the density gradient is continuous as in the previous three examples, it can be estimated that the regularity distribution of the same cause of failure as the Ap region affecting the adjacent coordinates has spread with the density gradient.
[0178]
When the mixture distribution (two types of periods and influence on adjacent coordinates) indicates a density gradient, a region suitable for analysis of regularity / irregularity has two types of periods L1 and L2, and A defect distribution AWp characterized in that the value of T (f) before and after these cycles is close to the maximum value of T (f) T (f) max = T (L) ≒ (L1 + L1) / 2. Since the density gradient is continuous as in the four examples, it can be estimated that the regular distribution of the same cause of failure as the AWp region is a failure distribution spread with a density gradient.
[0179]
As described above, when a continuous density gradient spreads around a region where regularity / irregularity distribution can be classified, these regions can be grouped as a region group sharing the same cause of failure.
[0180]
Next, an output example shown in FIG. 31 will be described. When managing the distribution characteristics of the cause of the failure at a specific failure phenomenon to be analyzed and the specific coordinates in the divided analysis region in a database, for example, even if the failure distribution is roughly classified as a regular distribution, a detailed description is given depending on the type of the period. Classification is possible, it is possible to classify according to whether or not it includes a plurality of cycles, or it is also possible to classify according to whether or not the cause of failure affects adjacent coordinates. Can also be classified.
[0181]
When these are classified, for example, they are classified according to the defect density as a primary identification method, and a secondary identification method is applied to each area roughly classified into a regular distribution, an irregularity, or a low density distribution. Are classified according to the type of mixture distribution, and furthermore, each region including the regular distribution is classified as a tertiary identification method depending on whether or not there is an influence on adjacent coordinates. Are classified according to the period of the regular distribution included in.
[0182]
As a result of these hierarchical classifications, database management is facilitated for each area by hierarchical digital identifiers (such as alphabets, numbers, or a combination of numbers and alphabets, or RGB color arrangements). With respect to distribution features that cannot be represented by these identifiers, the T (f) vs. f graph is filed in a tabular form, or the failure distribution is converted into an image file, and is associated with the identifier of each region, thereby obtaining the target on the database. If the identifier is searched, the data of the corresponding area can be easily extracted.
[0183]
That is, a region roughly classified into a regular distribution is searched by the identifier A, a region including the irregular distribution slightly in the extracted A region group is searched by the identifier Ac, and the period included in the region is identified by the identifier L. = 2, an Ac area group with a single period L = 2 is output, and when the number of defects included in the area is searched for with an identifier n = 100, an area with the number of defects is 100 is output. A T (f) vs. f graph file or an image file of a defective distribution can be output to check the distribution shape.
[0184]
In addition, a mixture distribution in which a regular distribution having two types of periods and a small amount of irregularity distribution are slightly searched from the A region group previously searched for by the identifier AWc. L1, L2, the number of defects n and T (f) vs. f graph file and an image file of defect distribution can be output and confirmed.
[0185]
Similarly, an area having a defect cause that affects adjacent coordinates can be searched using the identifier Ap, and an area including a plurality of regular distributions that affect adjacent coordinates can be searched using the identifier AWp. Similarly, an irregularity distribution that is not classified as A can be searched using the identifier C, and the number of defects n or T (f) vs. f graph or a defect distribution image file can be searched.
[0186]
In other words, with respect to a specific failure phenomenon, for example, a single-bit failure, the failure distribution data in the divided area at the specific coordinates X = 10 and Y = 12 is recorded and managed for the above-mentioned purpose, so that the conventional database management is required. It became easy to search for features of difficult distribution of defects.
[0187]
A feature search for a defective distribution is similar to searching for a specific feature (a photograph of a flower and a child near a lake) using a computer, for example, from a collection of photographs, and a method for accurately defining the feature has been established. I didn't. As described above, each of the above-described embodiments provides information on regularity / irregularity and information on defect density and defect cycle, which are important in defining the characteristics of the defect distribution.
[0188]
Next, an output example of FIG. 32 will be described. It is important to analyze the inside of the divided target area, and it is also important to compare features between the areas. In particular, if the characteristics of the defect distribution are continuous between adjacent regions, it can be estimated that these regions belong to the same group.
[0189]
It is also important to compare the same type of apparatus having different manufacturing dates and times to check the temporal change of the target area having the same coordinates. For example, when the distribution of the area of a device manufactured later changes to an irregularity while the failure distribution of the device manufactured earlier is a regular distribution, it can be estimated that the manufacturing conditions have changed. Conversely, the same applies to a case where a device having an irregular distribution first changes to a regular distribution in a device manufactured later, which is a trigger for checking manufacturing conditions.
[0190]
Not only the change in the regularity / irregularity of the defect distribution, but also the rate of change in the number of defects is important in observing a change in manufacturing conditions. The number of defects included in the analysis target region of the manufactured device is 800, and the change rate of the number of defects is + 10% or + 200% or + 400% as compared with the same region of the previously manufactured device. In such a case, the production conditions may be reviewed.
[0191]
When the manufacturing conditions of the device to be analyzed are changed, it is also important to confirm whether or not the characteristic parameter group of the device has changed before and after the change. For example, when the characteristic value of the parameter 1 before the change is plotted on the horizontal axis, and the characteristic value of the parameter 1 after the change is plotted on the vertical axis, the plotted value is obliquely elongated (the horizontal axis and The distribution becomes stronger (correlation on the vertical axis is stronger), and the less insensitive, the rounder the distribution is (the correlation between the horizontal axis and the vertical axis is weaker).
[0192]
The upper right diagonal distribution indicates that the larger the characteristic value before the change, the larger the characteristic value after the change. This is called a strong positive correlation. The lower-right diagonal distribution indicates that the larger the characteristic value before the change, the smaller the characteristic value after the change. This is called a strong negative correlation. By comparing some characteristic parameters and confirming whether or not deterioration has occurred after the change of the manufacturing conditions, the quality of the change of the manufacturing conditions can be evaluated.
[0193]
In this example, by setting the variation of the manufacturing condition parameters on the horizontal axis and the variation of the defect density on the vertical axis, when the magnitude of each parameter is changed, whether the defect density increases or decreases is evaluated by the strength of the correlation. It is possible to determine whether the manufacturing conditions are changed.
[0194]
As described above, in the above-described first to fifth embodiments, the distribution of each analysis range on the wafer of the memory LSI can be mapped and output. In any of the embodiments, the types of distribution are limited, so that the visibility can be enhanced by colors and patterns. Further, since the memory LSI has XY addresses, it is possible to obtain output data for each of the X and Y components, and it is possible to associate a failure cause with a combination of the classification results of the defective element distribution in both components.
[0195]
[Modification of Embodiment]
The present invention is not limited to the above-described first to fifth embodiments, and various modifications and applications are possible. Hereinafter, modifications of the above-described embodiment applicable to the present invention will be described.
[0196]
In the above-described third embodiment, the inspection analysis device in which the mixture distribution analysis device 6 is added to the configuration of the inspection analysis device shown in the second embodiment has been described. On the other hand, the mixture distribution analyzer 6 may be added to the configuration in the first embodiment as shown in FIG. Further, as shown in FIG. 26, the mixture distribution analyzer 6 is added to an inspection analyzer having a configuration similar to that of the related art, that is, a configuration obtained by removing the low-density distribution analyzer 31 from the configuration of FIG. It may be.
[0197]
In the above-described fourth embodiment, the inspection analyzer in which the critical distribution analyzer 7 is added to the configuration of the inspection analyzer shown in the third embodiment has been described. On the other hand, the critical distribution analyzer 7 may be added to the configuration in the first embodiment or to the configuration in the second embodiment, as shown in FIGS. 27 and 28, respectively. . As shown in FIG. 29, a configuration in which a mixture distribution analyzer 6 and a critical distribution analyzer 7 are added to the configuration in the first embodiment may be adopted. Further, the history analysis device 8 and the history database 81 described in the fifth embodiment are not particularly shown, but may be added to the configurations of FIGS.
[0198]
In the above-described first to fourth embodiments, when it is determined in steps S407, S510, S627, and S708 that there are other coordinate components to be analyzed, the process returns to steps S401, S502, S603, and S702, respectively. I was However, the process returns to steps S451, S501, S601, and S701, and the input of the reference value or the like may be performed again for each coordinate component.
[0199]
In the first to fifth embodiments, the regularity distribution analysis unit 32 determines whether the distribution of defective elements included in the semiconductor application device to be analyzed is the distribution (A) including the regularity distribution. The analysis of the regular distribution (B) was performed by the same method as that of the related art described above. However, the method of analyzing whether the distribution (A) includes a regular distribution or the irregular distribution (B) is not limited to this, and an arbitrary method can be applied.
[0200]
A program for constructing the inspection analyzer (excluding the input device 1 and the output device 4) shown in the first to fifth embodiments is stored in a computer-readable recording medium such as a CD-ROM, for example. May be distributed. That is, the program stored in such a recording medium may be installed and executed on a fixed disk device included in a general-purpose computer. As a result, the above-described inspection analysis device can be constructed on the general-purpose computer.
[0201]
Such a program is stored in a fixed disk device of a server device on a network such as a WAN (Wide Area Network) or a LAN (Local Area Network), and downloaded to a client device via the network as necessary. You may. As a result, the above-described inspection analysis device can be constructed on the client device.
[0202]
【The invention's effect】
As described above, according to the present invention, the distribution of defective elements included in a semiconductor application device can be accurately and finely classified. By recording the history and comparing the new analysis result with the history, it becomes possible to accurately associate the cause with the cause of the failure.
[Brief description of the drawings]
FIG. 1 is a block diagram illustrating a configuration of an inspection and analysis apparatus for a semiconductor application device according to a first embodiment of the present invention.
FIG. 2 is a flowchart illustrating an example (A) of a calculation method up to obtaining an expected value function in the arithmetic device of FIG. 1;
FIG. 3 is a flowchart illustrating another example (B) of a calculation method up to obtaining an expected value function in the arithmetic device of FIG. 1;
FIG. 4 is a flowchart showing another example (C) of a calculation method up to obtaining an expected value function in the arithmetic device of FIG. 1;
FIG. 5 is a flowchart showing an example (A) of an analysis method of the distribution of defective elements in the analyzer of FIG. 1;
FIG. 6 is a flowchart showing another example (B) of an analysis method of the distribution of defective elements in the analyzer of FIG. 1;
FIG. 7 is a graph showing that the method of classifying the distribution of defective elements according to the first embodiment of the present invention is applicable.
FIG. 8 is a block diagram illustrating a configuration of an inspection / analysis device for a semiconductor application device according to a second embodiment of the present invention.
9 is a flowchart illustrating an example (A) of a method of classifying the distribution of defective elements in the defect number analysis apparatus of FIG. 8;
10 is a flowchart illustrating an example (B) of a method of classifying the distribution of defective elements in the defect number analysis apparatus of FIG. 8;
FIG. 11 is a block diagram illustrating a configuration of an inspection and analysis apparatus for a semiconductor application device according to a third embodiment of the present invention.
FIG. 12 is a flowchart showing a method of analyzing the distribution of defective elements in the mixture distribution analyzer of FIG.
FIG. 13 is a flowchart showing a method of analyzing the distribution of defective elements in the mixture distribution analyzer of FIG.
FIG. 14 is a graph showing that a method for classifying the distribution of defective elements according to the third embodiment of the present invention is applicable.
FIG. 15 is a graph showing that a method for classifying the distribution of defective elements according to the third embodiment of the present invention is applicable.
FIG. 16 is a graph showing that a method for classifying the distribution of defective elements according to the third embodiment of the present invention is applicable.
FIG. 17 is a block diagram illustrating a configuration of an inspection / analysis device for a semiconductor application device according to a fourth embodiment of the present invention.
18 is a flowchart showing a method for analyzing the distribution of defective elements in the critical distribution analyzer of FIG.
FIG. 19 is a graph showing that a method for classifying the distribution of defective elements according to the fourth embodiment of the present invention is applicable.
FIG. 20 is a block diagram illustrating a configuration of an inspection / analysis device for a semiconductor application device according to a fifth embodiment of the present invention.
FIG. 21 is a block diagram showing another configuration of the inspection / analysis device for a semiconductor application device according to the fifth embodiment of the present invention.
FIG. 22 is a block diagram showing another configuration of the inspection / analysis device for a semiconductor application device according to the fifth embodiment of the present invention.
FIG. 23 is a block diagram showing another configuration of the inspection and analysis apparatus for a semiconductor application device according to the fifth embodiment of the present invention.
FIG. 24 is a flowchart illustrating a method of comparing and associating with a history in the history analysis apparatus of FIGS. 20 to 23;
FIG. 25 is a block diagram showing a configuration of an inspection / analysis device for a semiconductor application device according to a modification of the embodiment of the present invention.
FIG. 26 is a block diagram illustrating a configuration of a test analysis device of a semiconductor application device according to a modification of the embodiment of the present invention.
FIG. 27 is a block diagram showing a configuration of an inspection and analysis apparatus of a semiconductor application device according to a modification of the embodiment of the present invention.
FIG. 28 is a block diagram illustrating a configuration of an inspection / analysis device for a semiconductor application device according to a modification of the embodiment of the present invention.
FIG. 29 is a block diagram showing a configuration of an inspection / analysis device of a semiconductor application device according to a modification of the embodiment of the present invention.
FIG. 30 is a diagram showing an output example from the output device of the inspection analysis device of the semiconductor application device according to the embodiment of the present invention.
FIG. 31 is a diagram showing an output example from an output device of the inspection analysis device of the semiconductor application device according to the embodiment of the present invention.
FIG. 32 is a diagram showing an output example from the output device of the inspection analysis device of the semiconductor application device according to the embodiment of the present invention.
[Explanation of symbols]
1 Input device
2 arithmetic unit
3 analyzer
4 Output device
5 Defect count analyzer
6 Mixture distribution analyzer
7 Critical distribution analyzer
8 History analyzer
21 Interval calculator
22 Combination number calculator
23 Factor calculation unit
24 Individual frequency calculator
25 Expected Value Function Calculator
31 Low Density Distribution Analysis Section
32 Regular distribution analyzer
81 History Database

Claims (30)

Interval calculating means for calculating the interval for all combinations of any two of the defective elements included in the semiconductor application device to be analyzed;
A combination number calculating means for calculating the number of combinations excluding the number of combinations having a value of 0 among combinations of intervals between defective elements calculated by the interval calculating means; and a defective element calculated by the interval calculating means. A divisor calculating means for determining a divisor of each interval;
Comparison means for comparing a value obtained by dividing the maximum value of the divisor calculated by the divisor calculation means by the number of combinations calculated by the combination number calculation means with a predetermined threshold value,
As a result of the comparison by the comparing means, when a value obtained by dividing the maximum value of the divisor by the number of combinations is larger than a predetermined threshold, the distribution of the defective elements included in the semiconductor application device has a regular distribution and an irregular distribution. A first analysis means for determining that the distribution is a low-density distribution that cannot be distinguished from a regular distribution. The inspection and analysis apparatus according to claim 1, wherein the predetermined threshold is 1. As a result of the comparison by the comparing means, when a value obtained by dividing the maximum value of the divisor by the number of combinations is equal to or less than a predetermined threshold, the distribution of the defective elements included in the semiconductor application device includes a regular distribution. The inspection analysis apparatus for a semiconductor application device according to claim 1, further comprising a second analysis unit configured to analyze whether the distribution is an irregular distribution. The second analysis means includes:
Frequency calculation means for each value to calculate the appearance frequency for each divisor found by the divisor calculation means,
Expected value function calculating means for calculating, for each divisor, a value of an expected value function corresponding to the appearance frequency of each divisor calculated by the value-based frequency calculating means,
Based on the expected value function calculated by the expected value function calculating means, a regularity determining means for determining whether the distribution of defective elements included in the semiconductor application device includes a regularity distribution or an irregularity distribution. The inspection and analysis apparatus for a semiconductor application device according to claim 3, further comprising: Based on the expected value function calculated by the expected value function calculating means, a predetermined reference value, and a predicted period of the regularity distribution, the distribution of the defective elements included in the semiconductor application device is changed to the irregularity distribution. The inspection analysis apparatus for a semiconductor application device according to claim 4, further comprising third analysis means for analyzing a composition ratio of what kind of regular distribution and what kind of regular distribution are included. Frequency calculation means for each value to calculate the appearance frequency for each divisor found by the divisor calculation means,
Expected value function calculating means for calculating, for each divisor, a value of an expected value function corresponding to the appearance frequency of each divisor calculated by the value-based frequency calculating means,
When the first analysis means determines that the distribution is a low density distribution, and the value obtained by dividing the maximum value of the divisor by the number of combinations is within a predetermined range near the predetermined threshold, the expected value function calculation means 6. The apparatus according to claim 1, further comprising: fourth analysis means for analyzing whether the distribution of the defective elements included in the calculation result includes a regular distribution or an irregular distribution. Item 2. An inspection / analysis device for a semiconductor application device according to item 1. The fourth analysis means determines that the distribution of defective elements included in the semiconductor application device includes a regular distribution according to whether the value of the expected value function for the prime divisor is close to 1 or close to the divisor. The inspection and analysis apparatus for a semiconductor application device according to claim 6, wherein it is determined whether the distribution is an irregular distribution. When the value of the expected value function when the value of the divisor is 2 is close to 1, the fourth analysis unit determines that the distribution of the defective elements included in the semiconductor application device is an irregular distribution. 8. The inspection and analysis apparatus according to claim 6, wherein when the number is close to 2, it is determined that a regular distribution having an even period is included. Further comprising a defective element number investigation means for examining the number of defective elements included in the semiconductor application device,
9. The apparatus according to claim 1, wherein said interval calculation means calculates an interval between defective elements when the number of defective elements checked by said defective element number inspection means is within a predetermined range. Inspection and analysis equipment for semiconductor application equipment according to the paragraph. In a case where the number of defective elements checked by the defective element number checking means is not in the predetermined range, a fifth analyzing means for classifying a distribution of defective elements included in the semiconductor application device according to the number of the defective elements is further provided. The inspection / analysis device for a semiconductor application device according to claim 9, wherein the inspection / analysis device is provided. The fifth analysis means may be configured to determine whether the semiconductor application device does not include a defective element, determine whether the semiconductor application device includes a defective element but the number is equal to or less than a first predetermined number, 11. The semiconductor application device according to claim 10, wherein a distribution of the defective elements included in the semiconductor application device is classified into a case where the number of defective elements included in the device is equal to or more than a second predetermined number. Inspection analyzer. Analysis means for the distribution of defective elements included in the semiconductor application device, accumulation means for accumulating according to the type of the distribution,
When an analysis result on the distribution of defective elements included in the semiconductor application device is obtained, the obtained analysis result is analyzed by comparing the obtained analysis result with the analysis result previously stored in the storage unit, and The inspection and analysis apparatus for a semiconductor application device according to any one of claims 1 to 11, further comprising: Interval calculating means for calculating the interval for all combinations of any two of the defective elements included in the semiconductor application device to be analyzed;
Combination number calculating means for calculating the number of combinations excluding the number of combinations having a value of 0 among combinations of intervals between defective elements calculated by the interval calculating means,
Divisor calculating means for respectively obtaining a divisor of the interval between the defective elements calculated by the interval calculating means,
Frequency calculation means for each value to calculate the appearance frequency for each divisor found by the divisor calculation means,
Expected value function calculating means for calculating, for each divisor, a value of an expected value function corresponding to the appearance frequency of each divisor calculated by the value-based frequency calculating means,
Based on the expected value function calculated by the expected value function calculating means, a predetermined reference value, and a predicted period of the regularity distribution, the distribution of the defective elements included in the semiconductor application device is changed to the irregularity distribution. inspection analyzing apparatus for a semiconductor application device, characterized in that it comprises analyzing means for analyzing the composition ratio of whether to include in any ratio what kind of regularity distribution and. The analysis means,
Means for determining whether the maximum value of the expected value function is 1 or less,
The method according to claim 1, wherein when determining that the maximum value of the expected value function is 1 or less, the distribution of the defective elements included in the semiconductor application element is determined to be an ideal irregularity distribution. An inspection / analysis apparatus for a semiconductor application apparatus according to claim 13. The analysis means,
Means for determining whether the maximum value of the expectation function is equal to a corresponding divisor when the maximum value of the expectation function is not less than 1;
When it is determined that the maximum value of the expected value function is equal to a divisor corresponding thereto, the distribution of the defective elements included in the semiconductor application element changes the cycle of the divisor corresponding to the maximum value of the expected value function. The inspection and analysis apparatus for a semiconductor application device according to claim 14, wherein the inspection and analysis apparatus determines that the distribution has one type of regularity. The analysis means,
Means for determining whether the maximum value of the expectation function is equal to one half of the corresponding divisor if the maximum value of the expectation function is not less than 1;
When it is determined that the maximum value of the expected value function is equal to one half of the corresponding divisor, the distribution of defective elements included in the semiconductor application element is determined by adding the maximum value of the expected value function. The inspection and analysis apparatus for a semiconductor application device according to claim 14 or 15, wherein the inspection and analysis apparatus determines that the distribution is a regular distribution including two periods, which is a divisor corresponding to (i). The analysis means,
When the divisor corresponding to the maximum value of the expected value function is not equal to one half thereof, it is determined whether the maximum value of the expected value function is larger than 1 and smaller than a predetermined reference value. Means,
When it is determined that the maximum value of the expected value function is greater than 1 and smaller than a predetermined reference value, a set of periods of the regular distribution in which a divisor corresponding to the maximum value of the expected value function is predicted in advance Means for determining whether or not
When the divisor corresponding to the maximum value of the expected value function is included in the set of the periods, the distribution of the defective elements included in the semiconductor application element includes an irregularity distribution, and the maximum of the expected value function is included. The method according to any one of claims 13 to 16, further comprising one type of regular distribution having a divisor period corresponding to the value, and determining that the distribution is mainly an irregular distribution. Inspection and analysis equipment for semiconductor application equipment according to the description. The analysis means,
If it is determined that the divisor corresponding to the maximum value of the expected value function is not included in the set of periods of the regular distribution that is predicted in advance, the sum of the divisor and the divisor corresponding to the maximum value of the expected value function is determined. Means for determining whether two numbers are included in the set of periods,
When it is determined that two numbers whose sum is a divisor corresponding to the maximum value of the expected value function are included in the set of the periods, the distribution of the defective elements included in the semiconductor application element is determined by the two 18. The inspection and analysis apparatus for a semiconductor application device according to claim 17, wherein the inspection and analysis apparatus includes a regular distribution having a cycle of a number and judges that the distribution is mainly a random distribution. The analysis means,
When it is determined that the maximum value of the expected value function is equal to or less than 1 or equal to or greater than a predetermined reference value, the maximum value of the expected value function is larger than half of the corresponding divisor, and Including means for determining whether it is less than a divisor,
If it is determined that the maximum value of the expected value function is larger than half of the corresponding divisor and smaller than the divisor, the distribution of the defective elements included in the semiconductor application element is determined to be unacceptable. 19. The semiconductor according to claim 17 or 18, wherein the semiconductor is determined to be a distribution mainly including a regular distribution having a periodicity corresponding to a maximum value of the expected value function and including a regular distribution. Inspection and analysis equipment for applied equipment The analysis means,
A rule that the divisor corresponding to the maximum value of the expected value function is predicted in advance when it is determined that the maximum value of the expected value function is two minutes or more of the divisor corresponding to the expected value function; Means for determining whether or not they are included in a set of gender distribution cycles,
When it is determined that the maximum value of the expected value function is included in the set of periods, the distribution of the defective elements included in the semiconductor application element includes an irregularity distribution, and the maximum value of the expected value function 20. The inspection and analysis apparatus according to claim 19, wherein the inspection and analysis apparatus determines that the distribution is mainly a regular distribution having a cycle corresponding to a divisor. The analysis means,
Either of the two expected value functions corresponding to the divisor larger and smaller by 1 than the divisor corresponding to the maximum value of the expected value function is greater than or equal to the predetermined reference value, and Means for determining whether it is less than a divisor corresponding to the maximum value,
Either of the two expected value functions corresponding to the divisor larger or smaller by 1 than the divisor corresponding to the maximum value of the expected value function is greater than or equal to the predetermined reference value, and When it is determined that the divisor is smaller than the divisor corresponding to the maximum value, the distribution of the defective elements included in the semiconductor application element includes an irregularity distribution and the divisor corresponding to the maximum value of the expected value function. Judging that the distribution is mainly a regular distribution due to a factor that affects adjacent coordinates as well as the period,
Which of the two expected value functions corresponding to the divisor larger or smaller by 1 than the divisor corresponding to the maximum value of the expected value function is equal to or less than the predetermined reference value, or the maximum value of the expected value function If it is determined that the divisor is equal to or greater than the divisor, the distribution of the defective elements included in the semiconductor application element, including the irregularity distribution, and the cycle corresponding to the divisor corresponding to the maximum value of the expected value function. 21. The inspection / analysis apparatus for a semiconductor application device according to claim 19, wherein the distribution is determined to be mainly a regular distribution due to a factor that does not affect adjacent coordinates. The analysis means,
When it is determined that the maximum value of the expected value function is not included in the set of the periods, two numbers whose sum is a divisor corresponding to the maximum value of the expected value function are included in the set of the periods. Means for determining whether
When it is determined that two numbers whose sum is a divisor corresponding to the maximum value of the expected value function are included in the set of the periods, the distribution of the defective elements included in the semiconductor application element becomes irregular. 21. The inspection analysis apparatus for a semiconductor application device according to claim 20, wherein the inspection analysis apparatus includes a distribution and determines that the distribution is mainly a regular distribution having a cycle of the two numbers. The analysis means,
If it is determined that two numbers whose sum is a divisor corresponding to the maximum value of the expected value function are not included in the set of the periods, the number is larger or smaller by one than the sum of the two numbers. Means for determining whether the expected value function corresponding to the divisor is greater than or equal to the predetermined reference value and smaller than the divisor corresponding to the maximum value of the expected value function,
Each of two expected value functions corresponding to a divisor larger or smaller by one than the sum of the two numbers is equal to or greater than the predetermined reference value and corresponds to a maximum value of the expected value function. When it is determined that the number is smaller than the number, the distribution of the defective elements included in the semiconductor application element includes an irregularity distribution, the two numbers are set as a cycle, and the adjacent coordinates are affected. Is determined to be mainly a regular distribution by
Which of two expected value functions corresponding to a divisor larger or smaller by one than the sum of the two numbers is equal to or less than the predetermined reference value or greater than or equal to a divisor corresponding to a maximum value of the expected value function When it is determined that the distribution of defective elements included in the semiconductor applied element includes an irregularity distribution, the two numbers are set as a period, and a rule due to a factor that does not affect adjacent coordinates. 23. The inspection / analysis apparatus for a semiconductor application device according to claim 22, wherein the distribution is determined to be mainly a gender distribution. The analysis means,
If it is determined that two numbers whose sum is a divisor corresponding to the maximum value of the expected value function are not included in the set of the periods, the distribution of the defective elements included in the semiconductor application element is changed to a main component. 24. The inspection / analysis apparatus for a semiconductor application device according to claim 17, wherein it is determined that the distribution is an unclear irregular distribution. An interval calculation step of calculating an interval for all combinations of any two of the defective elements included in the semiconductor application device to be analyzed;
A number-of-combinations calculating step of calculating the number of combinations excluding the number of combinations having a value of 0 among combinations of intervals between defective elements calculated in the interval calculating step;
A divisor calculation step for respectively obtaining a divisor of the interval between the defective elements calculated in the interval calculation step,
A comparison step of comparing a value obtained by dividing the maximum value of the divisor calculated in the divisor calculation step by the number of combinations calculated in the combination number calculation step with a predetermined threshold value,
As a result of the comparison in the comparison step, when a value obtained by dividing the maximum value of the divisor by the number of combinations is larger than a predetermined threshold, the distribution of the defective elements included in the semiconductor application device is a regular distribution and A first analysis step of determining that the distribution is a low-density distribution that cannot be distinguished from an irregular distribution. As a result of the comparison in the comparison step, when a value obtained by dividing the maximum value of the divisor by the number of combinations is equal to or less than a predetermined threshold, the distribution of the defective elements included in the semiconductor application device has a regular distribution. 26. The inspection analysis method for a semiconductor application device according to claim 25, further comprising a second analysis step of analyzing whether the distribution is an irregular distribution or an irregular distribution. An interval calculation step of calculating an interval for all combinations of any two of the defective elements included in the semiconductor application device to be analyzed;
A number-of-combinations calculating step of calculating the number of combinations excluding the number of combinations having a value of 0 among combinations of intervals between defective elements calculated in the interval calculating step;
A divisor calculation step for respectively obtaining a divisor of the interval between the defective elements calculated in the interval calculation step,
A frequency calculation step for each value for calculating the appearance frequency for each divisor obtained in the divisor calculation step,
An expected value function calculating step of calculating, for each divisor, a value of an expected value function corresponding to the appearance frequency of each divisor calculated in the frequency calculation step for each value,
Based on the expected value function calculated in the expected value function calculating step, a predetermined reference value, and a predicted period of the regularity distribution, the distribution of the defective elements included in the semiconductor application device is changed to the irregularity distribution. And an analysis step of analyzing a composition ratio of what kind of regularity distribution is included in what ratio. Interval calculation means for calculating the interval for all combinations of any two of the defective elements included in the semiconductor application device to be analyzed;
Combination number calculating means for calculating the number of combinations excluding the number of combinations having a value of 0 among combinations of intervals between defective elements calculated by the interval calculating means,
Divisor calculating means for respectively obtaining a divisor of the interval between the defective elements calculated by the interval calculating means,
Comparison means for comparing a value obtained by dividing the maximum value of the divisor calculated by the divisor calculation means by the number of combinations calculated by the combination number calculation means with a predetermined threshold value, and
As a result of the comparison by the comparing means, when a value obtained by dividing the maximum value of the divisor by the number of combinations is larger than a predetermined threshold, the distribution of the defective elements included in the semiconductor application device has a regular distribution and an irregular distribution. A computer-readable recording medium on which a program for causing a computer to function as first analysis means for determining that the distribution is a low-density distribution that cannot be distinguished from a regular distribution is recorded. As a result of the comparison by the comparing means, when a value obtained by dividing the maximum value of the divisor by the number of combinations is equal to or less than a predetermined threshold, the distribution of defective elements included in the semiconductor application device includes a regular distribution. 29. The computer-readable recording medium according to claim 28, further comprising a program for causing a computer to function as second analysis means for analyzing whether the distribution is irregular distribution. Interval calculation means for calculating the interval for all combinations of any two of the defective elements included in the semiconductor application device to be analyzed;
Combination number calculating means for calculating the number of combinations excluding the number of combinations having a value of 0 among combinations of intervals between defective elements calculated by the interval calculating means;
Divisor calculating means for respectively obtaining a divisor of the interval between the defective elements calculated by the interval calculating means,
Frequency calculation means for each value to calculate the appearance frequency for each divisor found by the divisor calculation means,
Expected value function calculating means for calculating, for each divisor, a value of an expected value function corresponding to the appearance frequency of each divisor calculated by the value-based frequency calculating means; and
Based on the expected value function calculated by the expected value function calculating means, a predetermined reference value, and a predicted period of the regularity distribution, the distribution of the defective elements included in the semiconductor application device is changed to the irregularity distribution. A computer-readable recording medium having recorded thereon a program for causing a computer to function as analysis means for analyzing the composition ratio of what kind of regular distribution and what kind of regular distribution are included.

Images