JP2008108815A - System for specifying defectives-causing equipment - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To rapidly and accurately specify abnormal equipment (causing equipment candidate) as cause of defective products, or the like, in a production line containing a plurality of processes. <P>SOLUTION: Final defect distributions on each substrate are sorted into defect-distribution patterns P1 and P2 by using final defect-distribution information in a final inspection process 10g. The sorted final defect distributions and layer defect distributions obtained by each in-line inspection process 10a, 10e and 10f are compared, respectively, and the in-line inspection process 10e obtaining the layer defect distributions (such as 31b) similar to the final defect distributions (such as 26a) is specified. The substrate, similar to the layer defect distributions in the in-line inspection process 10e to the final defect distribution 26a, is extracted from the substrates through the in-line inspection process 10e. An equipment candidate causing the defect in a plurality of pieces of equipment is specified, on the basis of information displaying the substrate and production history information. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a failure cause facility identification system, and more particularly to a system for identifying an abnormal process or facility that causes a product failure or the like in a production line including a plurality of processes.

  In production lines for semiconductor devices and thin film devices that include many processes, such as hundreds of processes, in order to improve the quality of device products and stabilize equipment, various inspections such as foreign matter inspection and film thickness measurement inspection are performed. In-line inspection is performed in the production line. In addition, a final inspection process for finally judging the quality of the device product is also performed after the completion of all the processes. In this final inspection process, unlike foreign object inspection and film thickness measurement inspection, pass / fail judgment as a device product is performed. For example, in the final inspection process of a semiconductor device, pass / fail is determined for each chip on each wafer.

Conventionally, the defect distribution state on the wafer is classified from the failure information of the chips on each wafer obtained in these final inspection processes, and the product history information (which device was used to process each wafer in each process) A technique for identifying a facility that is a cause of failure using information) is well known.
JP 2005-197437 A

  However, in a production line for semiconductor devices and the like, since there are several hundreds of processes and each process has a plurality of production facilities, it takes a long time to compare with the production history individually, which is difficult in terms of time. is there.

  In addition, while many types of small-lot production are frequently performed, it is rare that the lots are processed in the order of lots from the initial process of wafer input to the final wafer test after several hundred processes. A lot is temporarily stored in a buffer in the layer, that is, an inventory stocker, and another lot proceeds to the next process while the lot is being stored. For this reason, in the wafer test of the final process, it is rare that the inspection is performed in the processing order of the manufacturing equipment. Therefore, at the final wafer test stage, inspection results of lots that are processed by the same abnormal equipment and have the same cause of failure are dispersed, making it difficult to detect the abnormal equipment. In addition, if inspection results are dispersed, it is difficult to determine a search period when collecting information from the final process.

  In Patent Document 1 (Japanese Patent Application Laid-Open No. 2005-197437), correspondence information between defect modes (defect classification, defect distribution, etc.) and apparatus types is stored in a database in advance using information obtained from each in-line inspection apparatus. And the method of specifying the equipment which is the cause of failure from the defect classification information of an in-line inspection result is disclosed using the database.

  However, in this method, it is necessary to define the correspondence between the defect mode and the device type in advance in order to create the database. In addition, since the correspondence between the actual defect mode and the device type changes every moment or every day, even if the correspondence is defined once, a lot having a certain defect distribution in the in-line inspection process becomes an actual product wafer. Sometimes it is not possible to accurately determine whether or not it will be defective.

  Further, in this document, defective defect modes are classified, and weighting is performed for each defect mode to specify the device type. However, in many cases, there are a plurality of defect modes on one wafer, and weighting them becomes complicated, and much time and labor are required only for maintenance. Even if weighting is performed, defects that occur from time to time vary from day to day, so it is not possible to accurately determine whether or not defects will occur when an actual product wafer is obtained.

  Accordingly, an object of the present invention is to provide a failure cause facility identification system capable of quickly and accurately identifying abnormal facilities (hereinafter referred to as “cause facility candidates”) that cause product defects in a production line including a plurality of processes. There is to do.

In order to solve the above problems, the failure cause equipment identification system of the present invention is
A failure cause facility identification system for identifying a facility that causes a failure in a production line that performs one or more steps on a substrate using one or more facilities capable of executing the steps. ,
The manufacturing line includes an in-line inspection process for acquiring layer defect distribution information indicating a position of a defect on each substrate for each layer indicating a stage of manufacturing, and a device manufactured on each substrate after the respective processes. A final inspection process for determining good and defective and obtaining final defect distribution information representing the position of the defective product,
Using the final defect distribution information for each substrate obtained in the final inspection step, a classification result acquisition unit that classifies the final defect distribution on each substrate for each defect distribution pattern;
For each final defect distribution classified into each defect distribution pattern, the final defect distribution classified for each defect distribution pattern by the classification result acquisition unit is compared with the layer defect distribution obtained in each inline inspection process. A layer specifying unit for specifying an inline inspection process in which a layer defect distribution similar to the final defect distribution is obtained;
A similar substrate extraction unit that extracts a substrate having a similar layer defect distribution in the specified inline inspection process with respect to the final defect distribution from the substrate that has undergone the specified inline inspection step;
Based on the information representing the substrate extracted by the similar substrate extraction unit and the manufacturing history information representing the facility that has performed processing in each step on each of the substrates, the occurrence of a failure among the plurality of facilities is determined. A causal equipment candidate specifying unit for specifying a causal equipment candidate that is a cause is provided.

  Here, the “substrate” corresponds to, for example, a wafer in a semiconductor device and a glass substrate (also referred to as mother glass) in a thin film device.

  The “layer” corresponds to a stage of manufacturing performed on the substrate, and is normally determined for each of a plurality of processes.

  “Layer defect distribution” and “final defect distribution” mean defect distributions represented by “layer defect distribution information” and “final defect distribution information”, respectively.

  In the defect cause facility identifying system of the present invention, the classification result acquisition unit classifies the final defect distribution of each substrate for each defect distribution pattern using the final defect distribution information for each substrate obtained in the final inspection process. The layer specifying unit compares the final defect distribution classified for each defect distribution pattern by the classification result acquisition unit with the layer defect distribution obtained in each of the inline inspection processes, and is classified into a certain defect distribution pattern. For each final defect distribution, an in-line inspection process in which a layer defect distribution similar to the final defect distribution is obtained is specified. If the inline inspection process is specified, the layer where the inline inspection process is performed is also specified. The similar substrate extraction unit extracts, from the substrates that have undergone the specified inline inspection process, the substrates whose layer defect distribution in the specified inline inspection process is similar to the final defect distribution. . The cause facility candidate specifying unit is configured to generate a plurality of information based on information representing the substrate extracted by the similar substrate extraction unit and manufacturing history information representing the facility that has performed each of the substrates in each step. The cause equipment candidate that caused the occurrence of the failure is identified.

  As described above, in this failure cause equipment identification system, the cause equipment candidate is specified by skillfully using both the final defect distribution information obtained in the final inspection process and the layer defect distribution obtained in the inline inspection process. Therefore, the cause equipment candidate can be identified quickly and accurately.

In the defect cause equipment identification system of one embodiment,
The classification result acquisition unit comprises a defect category input unit for selecting and instructing the reason for the defect of the defective product forming the final defect distribution to be classified,
The classification result acquisition unit classifies a final defect distribution for defective products having the selected reason for failure in each of the substrates based on an instruction input by the failure category input unit. .

  In the defect cause facility identifying system of this embodiment, for example, the operator selects and instructs the reason for defective products having the final defect distribution to be classified by the classification result acquisition unit via the defect category input unit. In this way, when the reason for failure is selected, in this system, the classification result acquisition unit, based on the instruction input by the failure category input unit, the selected reason for failure in each substrate. Classify the final defect distribution for defective products with Therefore, the classification by the classification result acquisition unit is efficiently performed.

  In the defect cause facility identifying system according to an embodiment, the similar substrate extracting unit executes the inline inspection process on the substrate having the final defect distribution among the substrates that have undergone the identified inline inspection process. With the inspection date and time as a reference, a substrate on which the inline inspection process has been executed within a certain period before and after the inspection date and time is extracted.

  In the defect cause facility identifying system according to this embodiment, the similar substrate extracting unit executes the inline inspection process on the substrate having the final defect distribution among the substrates that have undergone the identified inline inspection process. Based on the inspection date and time, a substrate on which the inline inspection process has been executed within a certain period before and after the inspection date and time is extracted. Therefore, the similar substrate extraction unit can efficiently extract a substrate whose layer defect distribution obtained in the specified inline inspection process is similar to the final defect distribution. As a result, the cause facility candidate specifying unit can quickly and accurately specify the cause facility candidate.

In the defect cause equipment identification system of one embodiment,
A first defect distribution image creation unit that creates a first defect distribution superimposed image by superimposing defect distributions on each substrate processed by the cause facility candidate;
A second defect distribution that creates a second defect distribution superimposed image by superimposing defect distributions on each substrate processed by equipment other than the cause equipment candidate in the same process as that performed by the cause equipment candidate. An image creation unit;
A certain display screen includes a first display unit that displays the first defect distribution superimposed image and the second defect distribution superimposed image in a contrasting manner.

  In the defect cause equipment identifying system of this embodiment, a first defect distribution superimposed image obtained by superimposing the defect distribution on each substrate processed by the cause equipment candidate on a certain display screen, and the cause equipment A second defect distribution superimposed image formed by superimposing defect distributions on each substrate processed by equipment other than the cause equipment candidate in the same process as the process executed by the candidate is displayed in comparison. Therefore, users (including system operators; the same shall apply hereinafter) can intuitively grasp whether or not the cause / facility candidate specified by this system is a cause of abnormality, compared to the conventional case. Judge quickly and easily.

  In the defect cause equipment identifying system according to an embodiment, a second list of the defect distribution patterns and cause equipment candidate information representing cause equipment candidates corresponding to the defect distribution patterns are displayed on the display screen together. And a display unit.

  In the defect cause equipment identifying system according to this embodiment, each defect distribution pattern and cause equipment candidate information representing cause equipment candidates corresponding to the defect distribution pattern are displayed in a list on the display screen. . Therefore, the user can quickly and easily determine the correspondence between the defect distribution pattern (wafer in-plane pattern) of the final inspection and the cause facility candidate. In the defect cause equipment identification system according to one embodiment, the second display unit is limited to only the cause equipment candidates that satisfy a predetermined threshold for specifying the cause equipment among the cause equipment candidates corresponding to the defect distribution pattern. It is characterized by displaying.

  In the failure cause facility identification system according to this embodiment, the cause facility candidate information displayed by the second display unit satisfies a predetermined threshold regarding the cause facility identification among the cause facility candidates corresponding to the defect distribution pattern. Limited to those related to cause equipment candidates. That is, among the causal equipment candidates corresponding to the defect distribution pattern, the causal equipment candidates that do not satisfy the predetermined threshold for specifying the causal equipment are considered to be causal equipment having a low degree of association, and the display is omitted. Therefore, the user can narrow down the cause equipment that is really causing the abnormality from the cause equipment candidates specified by the system more quickly and easily.

In the defect cause equipment identification system of one embodiment,
A candidate instruction input unit for inputting an instruction to select one of the causal equipment candidates whose causal equipment candidate information is displayed on the display screen;
The first display unit is linked to the cause facility candidate selected via the candidate instruction input unit, and the first defect distribution superimposed image and the second defect for the selected cause facility candidate. The display contents are switched so as to display the distribution superimposed image.

  In the failure cause facility specifying system according to this embodiment, for example, the operator can select one of the cause facility candidates whose cause facility candidate information is displayed on the display screen via the candidate instruction input unit. In this way, when any of the causal equipment candidates is selected, in this system, the first display unit is linked to the causal equipment candidate selected via the candidate instruction input unit, and the The display contents are switched so as to display the first defect distribution superimposed image and the second defect distribution superimposed image for the selected causal facility candidate. Therefore, it is possible to superimpose the defect distribution on each substrate processed by the selected causal equipment candidate without re-searching the inspection information and manufacturing history information when the operator uses equipment other than the causal equipment candidate. The second defect distribution overlay image is superimposed on the defect distribution on each substrate processed by the equipment other than the cause equipment candidate in the same process as the process executed by the cause equipment candidate. The distribution superimposed image is displayed in contrast to the display screen. Therefore, the user can intuitively grasp whether or not the selected cause facility candidate is the cause of the abnormality through visual perception, and can quickly and easily determine as compared with the conventional case.

  In the defect cause equipment identifying system according to an embodiment, the first and second defect distribution image creation units are configured to perform the first and second based on the information for identifying each substrate and the manufacturing history information, respectively. A defect distribution superimposed image is created for each facility.

  In the defect cause equipment identifying system according to this embodiment, the first and second defect distribution image creation units create the first and second defect distribution superimposed images for each equipment, respectively. Thereby, the user can intuitively grasp for each facility whether or not the cause facility candidate specified by this system is a cause of abnormality, and can determine more quickly and easily.

  Hereinafter, the present invention will be described in detail with reference to the illustrated embodiments.

  FIG. 1 illustrates a block configuration of a failure cause facility identification system 1 according to an embodiment of the present invention and a thin film device production line 10 to which the system is applied.

  In general, a production line for semiconductor devices and thin film devices is composed of a number of processes that are sequentially executed in units of production lots from reception of a substrate (wafer or mother glass) to completion of the device.

  FIG. 1 shows a part of such a semiconductor device manufacturing line 10. In this example, the production line 10 includes an inline inspection process 10a using the inline inspection apparatus A, a process 10b using the processing apparatus B, a process 10c using the processing apparatus C, a process 10d using the processing apparatus D, and an inline inspection. Inline inspection process 10e using apparatus E, 10f using processing apparatus F, and final inspection process 10g using final inspection apparatus G are included. That is, each of the processing steps 10b, 10c, 10d, and 10f includes, for example, a cleaning step for cleaning a wafer, a film forming step for forming a thin film on the wafer (hereinafter referred to as “deposition step”), and a photoresist on the thin film. It refers to a photolithography process (including an exposure process, a development process, etc., hereinafter referred to as “photo process”) for forming a pattern, and an etching process for patterning the thin film using the photoresist as a mask. The processing device B, the processing device C, the processing device D, and the processing device F collectively represent a plurality of facilities. In other words, for example, in the XX process, a plurality of facilities called XXX-1 machine, XX-2 machine, XX-3 machine,... Are used in parallel. .

  In the inline inspection steps 10a, 10e,..., In this example, pattern defect inspection is performed for each layer, and the inspection result is acquired as inline inspection result information. This in-line inspection result information includes layer defect distribution information indicating the position of the defect on each wafer for each layer. As the defects on the wafer, various defect distributions occur as shown in FIGS. 8A, 8B, and 8C. 8A is a type in which defects appear on one surface 101a in a concentrated manner on the surface of the wafer 100 (single-site concentration type), and FIG. 8B is a type in which defects appear on the surface of the wafer 100 as cracked chips 101b (a crack shape type). FIG. 8C shows types (peripheral types) in which defects are concentrated on the surface of the wafer 100 along the periphery 101c of the wafer.

  In the final inspection process 10g, final inspection result information is obtained by performing an inspection for determining whether the chip generated on the wafer is a non-defective product / defective product, that is, pass / fail. This final inspection result information includes final defect distribution information indicating the positions of good products and defective products. That is, in this final inspection result information, as shown in FIG. 9A, each chip 102 in the wafer 10 is mapped (the position of each chip 102 is specified by a number such as first, second,... On this map. In addition, each chip 102 is shown as a non-defective product / defective product. In FIG. 9A (and FIG. 9B), the acceptable product chip is represented by a white background, and the unacceptable product chip is represented by diagonal lines or solid black. As shown in FIG. 9B, for defective products of each chip 102 in the wafer, failure reasons 1, 2, 3, 4 (respectively described determination codes 1001, 1002, 1003, respectively), which are reasons for failure. , 1004)).

  As shown in FIG. 1, the failure cause facility identification system 1 includes a database 2, a classification result acquisition unit 15, a layer identification unit 16, a similar wafer extraction unit 17 as a similar substrate extraction unit, and a cause facility candidate identification. Unit 18, defect distribution image creation unit 19 serving as first and second defect distribution superimposed image creation unit, display unit 20 serving as first and second display units, defect category input unit and candidate instruction input And an instruction input unit 21 working as a unit. The database 2 includes a manufacturing history information storage unit 11 that stores manufacturing history information, an inline inspection result information storage unit 12 that stores inline inspection result information, a final inspection result information storage unit 13 that stores final inspection result information, And a master information storage unit 14 for storing master information.

  In the manufacturing history information storage unit 11, as illustrated in FIG. 4, manufacturing history information is stored in real time from the inline inspection apparatus A, the processing apparatus B, the final inspection apparatus G, and the like as needed. The form (format) of the manufacturing history information specifies a lot number (lot ID) for specifying a manufacturing lot, a wafer number (wafer ID) for specifying a wafer, and a process, as in the table illustrated in FIG. A process number, a machine number (processing equipment ID) that identifies equipment used in a certain process, and a processing date and time are associated with each other. From the manufacturing history information, for example, for the wafer AAA00001-1 included in the lot AAA00001, the process 10000 was executed at 10:10:00 on January 1, 2005 using the processing equipment AAA-1. I understand that.

  The inline inspection result information storage unit 12 stores inline inspection result information in real time from the inline inspection apparatus A and the like as illustrated in FIG. The output form (format) of the layer defect distribution information included in the in-line inspection result information represents the lot ID, the wafer ID, the process number, and the size of each defect as in the table illustrated in FIG. A defect size (which is classified into a small size S, a medium size M, and a large size L) and xy coordinates representing the position of the defect on the wafer surface are associated with each other. From this defect distribution information, for example, for the wafer AAA00001-1 included in the lot AAA00001, in step 30000, the size of the first defect is S, the x coordinate of the defect is 100, and the y coordinate of the defect is 200. Yes, the size of the second defect is M, the x coordinate of the defect is 100, the y coordinate of the defect is 200, the size of the nth defect is S, the x coordinate of the defect is 900, the defect It can be seen that the y coordinate is 800.

  In the final inspection result information accumulating unit 13, final inspection result information is accumulated in real time from the final inspection apparatus G and the like as illustrated in FIG. As for the output form (format) of the final inspection result information, the lot ID, wafer ID, process number, and determination code indicating pass / fail of each chip are associated with each other as in the table illustrated in FIG. It is what was done. From this final inspection result information, for example, for the wafer AAA00001-1 included in the lot AAA00001, in step 60000, the determination code of the first chip is 0000, which is a pass product, and the determination code of the second chip is 0001. It becomes a rejected product, and the determination code of the third chip is 0004 and the rejected product, and the determination code of the nth chip is 0001 and the rejected product.

  The master information storage unit 14 shown in FIG. 1 stores the determination code and determination reason for the above-mentioned acceptable product / non-acceptable product, the determination code 0000 is the acceptable product, the determination code 0001 is the unacceptable product, and the reason for the determination is poor. 1 and the determination code 0002 stores information such as a rejected product and a determination reason of failure reason 2. The reasons for failure 1 and failure 2 include, for example, a short circuit, large leakage current, Vth variation, etc. that occur when the thickness of the semiconductor device is not uniform. In addition, a threshold for specifying a similar wafer by the similar wafer extracting unit 17 and a threshold for specifying the cause facility by the cause facility candidate specifying unit 18 are stored.

  The classification result acquisition unit 15 uses the final defect distribution information for each wafer to classify the final defect distribution of each wafer for each defect distribution pattern. Here, the classification result acquisition unit 15 performs classification using a known method such as independent component analysis (ICA). As illustrated in FIG. 2, when pattern classification is performed from the result of the final inspection step 10g, a characteristic pattern is extracted from the defect distribution pattern of each wafer. Here, the top five defect distribution patterns P1, P2, and P3 are extracted. , P4 and P5 are selected. Each wafer 100 is classified into five groups 26a, 26b, 26c, 26d, and 26e that are similar to the defect distribution patterns P1, P2, P3, P4, and P5, respectively. When each wafer is classified, the definition of calculating the similarity with the defect distribution patterns P1 to P5 for each wafer and classifying the defect distribution pattern if the calculated similarity is equal to or greater than a certain threshold is as follows. The data stored in the master information storage unit 14 shown in FIG.

  As illustrated in FIG. 3, the layer specifying unit 16 includes the final defect distributions 26 a, 26 b,... Classified by the defect distribution patterns P 1 to P 5 by the classification result acquisition unit 15, and the inline inspection processes 10 a, 10 e, .. Are compared with each of the layer defect distributions 31a, 31b, 31c,... Obtained by 10f,..., And the final defect distributions 26a, 26b,. An in-line inspection process in which a layer defect distribution similar to the defect distribution is obtained is specified. If the inline inspection process is specified, the layer where the inline inspection process is performed is also specified.

  When calculating the similarity, if the calculated similarity is equal to or greater than a certain threshold, it is determined that the in-line inspection result and the final inspection result have a correlation. The threshold is used by referring to data stored in the master information storage unit 14 shown in FIG.

  Normally, a plurality of wafers (the final defect distribution) are classified into a certain defect distribution pattern. In that case, the in-line inspection process (and layer) may be specified, for example, by specifying the final defect distribution corresponding to a predetermined number of wafers or more.

  In this example, the final defect distribution 26a of the wafer classified into the defect distribution pattern P1 is similar to the layer defect distribution 31b obtained in the inline inspection process 10e and has a high similarity. Identified. Note that the inline inspection process is similarly specified for the final defect distributions 26b to 26e classified into the other defect distribution patterns P2 to P5, but the description thereof is omitted.

  The similar wafer extracting unit 17 shown in FIG. 1 performs the specified inline inspection process for the final defect distribution 26a from the wafers that have undergone the specified inline inspection process (in this example, the inline inspection process 10e). Wafers having similar layer defect distributions at 10e are extracted.

  Specifically, as shown in FIG. 14, the similar wafer extracting unit 17 selects the wafer 141 having the final defect distribution 26a from the wafers that have undergone the specified inline inspection process 10e (wafer ID = AAA0001-1). On the other hand, with reference to the inspection date and time when the inline inspection process 10e is executed as a reference, the wafer on which the inline inspection process 10e has been executed within a certain period (in this example, nn hours) before and after the inspection date and time is extracted. That is, it is assumed that the wafer 141 having the final defect distribution 26a is processed at 10:10:00 on January 1, 2005 by the inline inspection apparatus E in the inline inspection process 10e. Nn hours of wafers before and after this processing date, that is, a period from January 1, 2005 (10-nn) 10:00 to January 1, 2005 (10 + nn) 10:00 To do. Then, the degree of similarity is calculated by comparing the layer defect distribution obtained in the inline inspection process 10e for the wafer 141 with the layer defect distribution obtained in the inline inspection process 10e for those extracted wafers. If it is greater than a certain threshold, it is extracted as a wafer corresponding to the same defect distribution pattern. In FIG. 14, wafers having a high degree of similarity to the wafer 141 are represented by reference numerals 142a, 142b, and 142c. As the nn time and the threshold value, the data stored in the master information storage unit 14 shown in FIG.

  The cause facility candidate identification unit 18 targets a wafer extracted by the similar wafer extraction unit 17 and analyzes a common path based on the processing history information accumulated in the manufacturing history information accumulation unit 11 (a plurality of similar defect distributions having a plurality of defect distributions). An analysis in pursuit of which equipment the substrate is processed in common is performed), and a cause equipment candidate that causes a failure is identified from among a large number of equipment.

  The defect distribution image creation unit 19 creates a plurality of types of defect distribution superimposed images by superimposing defect distributions (bitmap images) on each wafer. This creation work will be described in detail later.

  The display unit 20 displays information on the cause facility candidate specified by the cause facility candidate specification unit 18 as a two-dimensional image on a certain display screen formed of a CRT (cathode ray tube) or LCD (liquid crystal display element).

  The instruction input unit 21 includes a mouse, a keyboard, and the like, and is used for an operator to input a desired instruction to the system 1.

  As will be appreciated by those skilled in the art, such a system 1 can be constituted by a computer, more specifically a personal computer. The operations of the units 15, 16, ..., 19 can be realized by a computer program (software).

  The system 1 operates as follows according to the processing flow shown in FIG.

  First, in step S <b> 101, the operator inputs search conditions for the final inspection result information stored in the final inspection result information storage unit 13 via the instruction input unit 21. For example, as shown in the right column (“Select” column) in FIG. 9B, this search condition is to select the reason for failure of the defective product that forms the final defect distribution to be classified by the classification result acquisition unit 15. In the example of FIG. 9B, the reasons for failure 1, 2, 4 and the accepted product are selected with a check.

  Next, in step S102 in FIG. 10, the classification result acquisition unit 15 reads the final inspection result information for each wafer from the final inspection result information storage unit 13 in accordance with the search condition.

  Subsequently, in step S103, the final defect distribution of each wafer is classified for each defect distribution pattern by using the final defect distribution information included in the final inspection result information (final inspection wafer in-plane pattern automatic classification). At this time, since the classification result acquisition unit 15 classifies the final defect distribution for the defective product having the reason for failure selected by the search condition, the classification is performed efficiently.

  Next, in step S104, the layer specifying unit 16 compares the final defect distribution classified for each defect distribution pattern by the classification result acquisition unit 15 with the layer defect distribution obtained by each in-line inspection process, respectively, and determines each defect. For each final defect distribution classified into the distribution pattern, an inline inspection process in which the layer defect distribution is similar to the final defect distribution and a layer on which the inline inspection process has been performed are identified (inline inspection layer identification) .

  Next, in step S105, the similar wafer extraction unit 15 uses the manufacturing history information 11 from the wafers that have undergone the inline inspection process specified in step S104 to inline the wafers having the final defect distribution. Based on the inspection date and time when the inspection process is executed, a wafer on which the inline inspection process is executed within a certain period before and after the inspection date and time is extracted. Then, a wafer whose layer defect distribution is similar in the inline inspection process specified in step S104 is extracted from the extracted wafer (similar wafer extraction).

  Next, in step S106, the causal facility candidate specifying unit 18 reads from the manufacturing history information storage unit 11 manufacturing history information representing the facilities that have been processed in each process for each wafer (reading manufacturing history information). Subsequently, in step S107, the cause facility candidate specifying unit 18 performs a common path analysis based on the information representing the wafer extracted by the similar wafer extracting unit 17 and the manufacturing history information read from the manufacturing history information accumulating unit 11. To identify the cause equipment candidate that caused the failure among the plurality of equipment (cause equipment identification).

  As described above, in the failure cause facility identification system 1, the cause facility candidate is identified by skillfully using both the final defect distribution information obtained in the final inspection process and the layer defect distribution obtained in the inline inspection process. Therefore, the causal equipment candidate can be identified quickly and accurately. In addition, since the wafer is extracted for a limited period in step S105, the cause facility candidate can be quickly and efficiently identified in step S107.

  Next, in step S108, the defect distribution image creation unit 19 creates first and second defect distribution overlay images (in-plane distribution overlay map creation). The first defect distribution superimposed image is obtained by superimposing the defect distribution on each wafer processed by the cause facility candidate in a certain process. The second defect distribution superimposed image is obtained by superimposing the defect distribution on each wafer processed by equipment other than the cause equipment candidate in the same process as the process executed by the cause equipment candidate. In the method of superimposing images, color settings including transparency are set on individual wafers so that defect information (defect size and position) can be distinguished even if the individual wafers overlap.

  Next, in step S109, the display unit 20 creates information about the causal facility candidate specified by the causal facility candidate specifying unit 18 on a certain display screen made up of a CRT or LCD by the defect distribution image creating unit 19. 1. Display as a two-dimensional image including a second defect distribution superimposed image (screen display).

  FIG. 7 illustrates what is displayed on the display screen 70 by the display unit 20. The display screen 70 of FIG. 7 is broadly divided into an upper table area 76 for displaying information on the cause equipment candidate for each defect distribution pattern (in-wafer surface pattern) of the final inspection, and a machine number for each equipment for the wafer defect distribution. It is composed of a lower table area 77 for displaying the difference.

  The upper table area 76 is divided into an item display area 72a and pattern classification areas 72b, 72c, 72d, 72e, and 72f representing classification results for each defect distribution pattern in the vertical direction. The upper table area 76 includes a check column 71a in which the operator checks with a mouse (not shown) in the horizontal direction, and the top five defect distribution patterns P1, P2, P3, P4, P5 (characterized by the final inspection result). The “feature map” column 71b for displaying the image of FIG. 2), the “classified wafer number” column 71c for displaying the number of wafers classified into the corresponding defect distribution pattern, and the wafer classified into the defect distribution pattern The “defect number wafer average” column 71d displaying the average number of defects per sheet, the “layer” column 71e representing the identified layer, the “process No.” column 71f representing the process number, and the process name The "process name" column 71g representing, the "equipment machine" column 71h representing the machine number of the cause candidate that executed the corresponding process, and the probability value of the corresponding cause facility candidate are scored. And the "score" column 71i which was, and is divided into a "rank" column 71j, which represents the score in the rank.

  In this display example, for example, in the pattern classification area 72b, the defect distribution pattern P1 (see FIG. 2) displayed on the feature map 71b of this pattern classification area is concentrated on the upper and right portions of the wafer. It can be seen that the number of wafers classified into the defect distribution pattern is 20 and the average number of defects per wafer is nine. Furthermore, it can be seen that the layer having the most defects is the second layer, and three candidates for the causal equipment are listed. The three candidates for the cause equipment are displayed in order from the highest degree of abnormality estimated as the cause candidates from the top. In the first candidate, the process No. Is 22000, the process name is the photo process, and the identified cause equipment candidate is BBB-4. It can be seen that the score (probability value) that is the cause of the abnormality is 95 and the rank is ●. The ranks are defined as ●, ◎ and ○ in descending order of degree of abnormality, but this definition may be changed by the master information storage unit 14 in FIG.

  In each of the pattern classification areas 72b, 72c, 72d, 72e, and 72f, the feature map images displayed in the feature map column 71b are arranged from the top in descending order of the number of classified wafers. Thereby, the operator can grasp | ascertain an important defect distribution pattern easily.

  In each pattern classification area 72b, 72c, 72d, 72e and 72f, since the image of the defect distribution pattern is displayed in the feature map column 71b, the operator intuitively visually understands the defect distribution pattern for each pattern classification area. Can grasp.

  In each pattern classification area 72b, 72c, 72d, 72e, and 72f, the “defect number wafer average” column 71d displays the average value of the number of defects per wafer. As a result, when an abnormal phenomenon related to a certain pattern classification area is classified into a characteristic defect distribution pattern, but the number of defects generated per wafer is small, the operator gives priority to measures against the abnormality detection. You can also make a decision to reduce the degree.

  In each of the pattern classification areas 72b, 72c, 72d, 72e, and 72f, the “facility number machine” column 71h displays the machine number of the cause equipment candidate identified by the cause equipment candidate identification unit 18. The causal facility candidate is displayed in association with the defect distribution pattern displayed in the feature map column 71b. That is, the defect distribution pattern of the feature map is displayed for each of the patterns 72b, 72c, 72d, 72e, and 72f, and the cause facility candidates corresponding to the cause facility candidates corresponding to the pattern are displayed in a list. That is, the operator can easily determine whether or not the cause is a cause of abnormality because the cause facility candidates can be listed while viewing the defect distribution pattern.

  The “score” column 71i displays the score of the corresponding causal facility candidate. This value represents the degree of cause of abnormality in the cause facility candidate specified by the cause facility candidate specifying unit 18. That is, if the probability value (for example, 2.66e-14) when the causal equipment candidate is specified is displayed as it is, the operator is confused. For example, the numerical value is displayed as 100 points above a certain upper limit. By doing so, it is possible to easily determine the degree of abnormality.

  The “rank” column 71j displays the rank of the corresponding causal facility candidate. This value represents the degree of cause of abnormality in the cause facility candidate specified by the cause facility candidate specifying unit 18, and the score calculated for display in the “score” column 71i is also included. And rank. In other words, the operator can make a quick decision by visually expressing equipment with high relevance or low equipment by displaying ● above a certain score, ◎ at the next rank, and ○ at the lower rank. can do.

  In each pattern classification area 72d, 72f, no causal facility candidate is displayed. This reason is classified as a feature map, but when identifying the cause equipment candidate, the cause equipment candidate that does not satisfy a certain threshold is determined to be a cause equipment with a low degree of association, and the cause equipment candidate is displayed. This is because is omitted.

  As a result, only the information narrowed down from a large number of information can be displayed, so that the operator can quickly and easily determine the correspondence between the defect distribution pattern (wafer in-plane pattern) of the final inspection and the cause equipment candidate. .

  Next, each page (switched by tabs 73a, 73b, 73c) in the lower table area 77 is set to the process No. in the vertical direction. "Wafer" which displays a heading area 74a for displaying process name, equipment number, score and rank, "equipment number" display area 74b for specifying equipment, and first and second defect distribution superimposed images. A “map overlay” display area 74c, a “number of wafers” display area 74d for displaying the number of wafers in which the classified wafer has been processed by the corresponding equipment, a “number of processed wafers” display area 74e for indicating the number of processes, and a specification It is divided into a “%” display area 74 f representing the ratio of the corresponding wafer. Moreover, regarding the horizontal direction, the item display column 75a, the first information display column 75b for displaying information on the identified causal equipment candidate, and the information on equipment other than the causal equipment candidate in the same process as the process executed by the causal equipment candidate Is displayed in a second information display field 75c.

  The process and equipment display area 74a includes a process number, a process name, and a cause equipment candidate machine number (equipment) according to the pattern classification area (that is, the defect distribution pattern) checked by the operator in the check column 71a of the upper table area 76. Unit), score, and rank are displayed. In the example of FIG. 7, the process and equipment display area 74 a includes a process number “22000”, a process name “photo process”, a cause equipment candidate machine number (equipment machine number) “BBB-4 machine”, and a score “95”. , Rank “●” is displayed. This is information corresponding to the first candidate BBB-4 among the three causal equipment candidates BBB-4, CCC-4 and AAA-5 listed in the pattern classification area 71h. is there.

  In the “equipment number machine” display area 74b, the equipment number machine identified by the cause equipment candidate is displayed in the first information display field 75b. In contrast, in the second information display field 75c, equipment numbers processed by equipment other than the cause equipment candidate in the same process as the cause equipment candidate is executed are displayed.

  In the “wafer map overlay” display area 74c, a first defect distribution overlay image is displayed in the first information display column 75b by superimposing the defect distributions on each wafer processed by the cause candidate facility. In contrast, the second defect is formed by superimposing the defect distribution on each wafer processed by the equipment other than the cause equipment candidate in the same process as the process executed by the cause equipment candidate in the second information display column 75c. A distribution overlay image is displayed. Thus, the operator can intuitively grasp whether or not the causal equipment candidate specified by the system 1 is really the cause of the abnormality through the visual sense, and can quickly and easily determine compared with the conventional case. That is, unlike the conventional case, the operator does not have to search for inline inspection information, manufacturing history information, and the like when using equipment other than the causal equipment candidate.

  The plurality of pages in the lower table area 77 can be switched by the operator selecting the tabs 73a, 73b, 73c with a mouse or the like. For example, when the operator selects the pattern classification area 72e from the display in which the pattern classification area 72b is selected, the content displayed in the lower area 77 is the process number. “33000”, process name “etching process”, equipment number “CCC-03”, score “75”, rank “◯”. Therefore, the wafer in-plane distribution information and the information on the difference between facilities can be displayed sufficiently effectively even in a small screen display area.

  In this embodiment, an example in which the present invention is applied to a semiconductor device production line has been described. However, the present invention is used to identify equipment that causes a defect in a production line that executes one or more processes on a substrate using one or more equipment capable of executing the processes. Can be widely applied. For example, the present invention can be applied to a production line for thin film devices. The present invention is effective for in-plane distribution based on in-line inspection in the device manufacturing process, and if this in-line inspection and final inspection can be linked with history information, Applicable.

BRIEF DESCRIPTION OF THE DRAWINGS It is a figure which shows the general | schematic block structure of the manufacturing line to which the defect cause equipment identification system of one Embodiment of this invention and the defect cause equipment identification system are applied. It is a figure which illustrates the state which classified the final defect distribution information obtained at the final inspection process for every defect distribution pattern. It is a figure which compares and exemplifies the final defect distribution classified for every defect distribution pattern, and the layer defect distribution obtained by each in-line inspection process, respectively. It is a figure explaining how a manufacturing history information is stored in a manufacturing history information storage part. It is a figure explaining how final inspection result information is stored in a final inspection result information storage part. It is a figure explaining how inline inspection result information is stored in an inline inspection result information storage part. It is a figure which illustrates the display screen displayed by the said failure cause equipment specific system. It is a figure which illustrates the one-point concentration type defect distribution which generate | occur | produces on the wafer surface. It is a figure which illustrates the crack shape type defect distribution which generate | occur | produces on the wafer surface. It is a figure which illustrates the periphery type defect distribution which generate | occur | produces on the wafer surface. It is a figure which illustrates final defect distribution. It is a figure which illustrates the state where the judgment code and the reason for failure are given to the rejected product in FIG. 9A. It is a figure which shows the operation | movement flow of the said failure cause equipment identification system. It is a figure which shows the format of the database of the said inline test result information. It is a figure which shows the format of the database of the said last test result information. It is a figure which shows the format of the said manufacture log | history information. It is a figure explaining how to extract the wafer which has similar layer defect distribution from the wafer which passed through the in-line inspection process.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Defect cause identification system 2 Database 10 Manufacturing line 11 Manufacturing history information storage part 12 Inline inspection result information storage part 13 Final inspection result information storage part 14 Master information storage part 15 Classification result acquisition part 16 Layer specification part 17 Similar wafer extraction part 18 Cause equipment candidate specifying unit 19 Defect distribution image creating unit 20 Display unit 21 Instruction input unit 70 Display screen 100, 141, 141a, 141b, 141c Wafer P1, P2, P3, P4, P5 Defect distribution pattern

Claims (8)

A failure cause facility identification system for identifying a facility that causes a failure in a production line that performs one or more steps on a substrate using one or more facilities capable of executing the steps. ,
The manufacturing line includes an in-line inspection process for acquiring layer defect distribution information indicating a position of a defect on each substrate for each layer indicating a stage of manufacturing, and a device manufactured on each substrate after the respective processes. A final inspection process for determining good and defective and obtaining final defect distribution information representing the position of the defective product,
Using the final defect distribution information for each substrate obtained in the final inspection step, a classification result acquisition unit that classifies the final defect distribution on each substrate for each defect distribution pattern;
For each final defect distribution classified into each defect distribution pattern, the final defect distribution classified for each defect distribution pattern by the classification result acquisition unit is compared with the layer defect distribution obtained in each inline inspection process. A layer specifying unit for specifying an inline inspection process in which a layer defect distribution similar to the final defect distribution is obtained;
A similar substrate extraction unit that extracts a substrate having a similar layer defect distribution in the specified inline inspection process with respect to the final defect distribution from the substrate that has undergone the specified inline inspection step;
Based on the information representing the substrate extracted by the similar substrate extraction unit and the manufacturing history information representing the facility that has performed processing in each step on each of the substrates, the occurrence of a failure among the plurality of facilities is determined. A failure cause facility identification system comprising a cause facility candidate identification unit for identifying a cause facility candidate that is a cause. In the defect cause equipment identification system according to claim 1,
The classification result acquisition unit comprises a defect category input unit for selecting and instructing the reason for the defect of the defective product forming the final defect distribution to be classified,
The classification result acquisition unit classifies a final defect distribution for defective products having the selected reason for failure in each of the substrates based on an instruction input by the failure category input unit. Defect cause equipment identification system. In the defect cause equipment identification system according to claim 1,
The similar substrate extraction unit is configured to determine whether the inline inspection process is performed on a substrate having the final defect distribution from among the specified inline inspection processes before and after the inspection date. A failure cause facility identifying system characterized in that a substrate on which the in-line inspection process has been executed within a certain period is extracted. In the defect cause equipment identification system according to claim 1,
A first defect distribution image creation unit that creates a first defect distribution superimposed image by superimposing defect distributions on each substrate processed by the cause facility candidate;
A second defect distribution that creates a second defect distribution superimposed image by superimposing defect distributions on each substrate processed by equipment other than the cause equipment candidate in the same process as that performed by the cause equipment candidate. An image creation unit;
A failure cause facility identifying system comprising: a first display unit that displays the first defect distribution superimposed image and the second defect distribution superimposed image on a display screen. . In the defect cause equipment identification system according to claim 1,
The display screen includes a second display unit that displays a list of each defect distribution pattern and cause equipment candidate information representing cause equipment candidates corresponding to the defect distribution pattern. Defect cause equipment identification system. In the defect cause equipment identification system according to claim 5,
The second display unit displays only the cause of the cause of failure corresponding to the defect distribution pattern, with respect to the cause of the cause of the candidate of the cause satisfying a predetermined threshold related to the specification of the cause of the failure. Equipment identification system. In the defect cause equipment identification system according to claim 5,
A candidate instruction input unit for inputting an instruction to select one of the causal equipment candidates whose causal equipment candidate information is displayed on the display screen;
The first display unit is linked to the cause facility candidate selected via the candidate instruction input unit, and the first defect distribution superimposed image and the second defect for the selected cause facility candidate. A failure cause facility identification system characterized by switching display contents so as to display a distribution superimposed image. In the defect cause equipment identification system according to claim 4,
The first and second defect distribution image creation units create the first and second defect distribution superimposed images for each facility based on information for specifying each substrate and the manufacturing history information, respectively. Defect cause equipment identification system characterized by that.

Images