JP2011018738A - Production management system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a production management system capable of predicting the processing time required for the processing of a prescribed lot with high accuracy, in a semiconductor manufacturing apparatus.SOLUTION: The production management system includes: a data collecting part 3 for associating the actually measured processing time required for the processing of the prescribed lot with a prescribed data tag and storing it in a data storage part 4, when the semiconductor manufacturing apparatus completes the processing of the lot; and a processing time estimating part 8 for estimating the processing time of a scheduled lot, by using the actually measured processing time associated with the data tag in the data storage part 4, which matches with a prescribed scheduled data tag, when the processing time of the scheduled lot is estimated.

Description

  The present invention relates to a production management system, and more particularly to a production management system capable of accurately predicting a processing time required for processing a predetermined lot for a semiconductor manufacturing apparatus.

  In recent years, in the manufacturing process of semiconductor devices, where competition with other companies is intensifying, it is required to shorten the manufacturing lead time and improve the delivery time compliance rate, despite the complexity of the manufacturing process. Therefore, in a semiconductor device manufacturing process, it is common to manufacture a semiconductor device based on a lot progress plan calculated using a production simulator. Improving the accuracy of this lot progress plan has become an important factor in determining the success or failure of a semiconductor device business by shortening the manufacturing lead time and improving the delivery rate.

  The production simulator performs a simulation of assigning a lot to each semiconductor manufacturing apparatus (manufacturing apparatus) belonging to the production line at regular intervals according to the lot processing status and facility operation status in the production line. Creation of a lot progress plan requires the estimated processing time of each semiconductor manufacturing apparatus for the lot, and it is necessary to bring this predicted processing time closer to the actual measurement processing time when the lot is actually processed. This leads to an improvement in the accuracy. For this reason, it becomes increasingly important in the future to calculate a predicted processing time close to the actual measurement processing time.

  Conventionally, the predicted processing time is calculated by a model formula (equipment model formula) expressed by a linear function with the number of wafers belonging to a lot (also called the number of substrates or lot size) as a variable (Patent Document 1).

  For example, assuming that the predicted processing time Y (sec) is the processing time a (sec) per wafer, the processing time b (sec) for the first wafer in the lot, and the number of wafers X (sheets), It is expressed by equation (1).

Y = a * (X−1) + b (1)
The wafer processing time a (sec / sheet) and the processing time b (sec) for the first wafer in the lot are calculated based on the acquired data. Specifically, the user linearly approximates the variation of the measured data with a linear function based on experience, and sets the wafer processing time a (sec) and the processing time b (sec) for the first wafer in the lot. Has been decided.

  In Japanese Patent Application Laid-Open No. 2008-287442 (Patent Document 2), the equipment model formula (1) is expanded by paying attention to the equipment processing method.

  For example, a “single-wafer type” equipment model formula is used for a manufacturing apparatus that processes wafers one by one in one chamber. A “single-wafer step type” equipment model formula is used in a manufacturing apparatus that processes one wafer at a time in a plurality of chambers. A “batch injection type” equipment model formula is used in a manufacturing apparatus that simultaneously processes a plurality of wafers having a lot size or less. For a manufacturing apparatus capable of simultaneously processing a plurality of wafers exceeding the lot size, a “batch cleaning type” equipment model formula is used.

In each of the equipment model expressions shown above, the number of wafers X (sheets) and the processing time Y (sec) are expressed by the following expressions.
Single wafer type Y = a * (X-1) + b (2)
Single wafer step type Y = a * (X-1) + b + floor ((X-1) / c) * d (3)
Batch injection type Y = floor ((X-1) / c) * d (4)
Batch cleaning type Y = b (5)
In Expressions (3) and (4), the function floor (X) is a function that performs fraction processing to round off the decimal point of the argument X. In the equations (3) and (4), model parameters c (sheets) and d (sec) are parameters determined based on already acquired data.

Japanese Patent Application Laid-Open No. 2004-334643 JP 2008-287442 A

  However, in the processing time prediction method described in Patent Document 1 or 2, since the collected data (the number of wafers, the actual processing time) is approximated by a specific function, there is a variation in the actual processing time for a predetermined number of wafers. If it is large, there is a problem that an appropriate approximate expression cannot be obtained even if the equipment model formula described above is fitted to the data.

  FIG. 6 is a diagram showing the relationship between the conventional predicted processing time and the actual processing time.

  The data shown in FIG. 6 includes a plurality of lot sizes (number of wafers) (sheets) obtained when a lot is processed under the same semiconductor manufacturing apparatus and the same processing conditions (recipe), and a plurality of data. This corresponds to the processing time (actual measurement processing time) (sec). As shown in FIG. 6, the processing time 601 corresponding to the same lot size (for example, lot size “12”) is very varied with the difference between the maximum value and the minimum value being about 600 seconds. Is understood. Therefore, there is a difference between the equipment model equation 602 shown in FIG. 6 and the actual processing time 603, and it is impossible to calculate a processing time with high accuracy.

  Therefore, in the conventional processing time prediction method, no matter how high the accuracy of the equipment model formula is, the accuracy of the prediction processing time is not improved due to the influence of the processing time variation described above. There is a problem of not improving.

  Further, when attention is paid to the recipe used by the semiconductor manufacturing apparatus, the following problems also exist.

  In general, in a method of manufacturing a semiconductor device, in order to minimize a processing loss that occurs between lots, processing of a predetermined lot is often executed continuously without a space between lots. Such a lot processing mode is called continuous processing.

  For example, there is a case where a predetermined lot is thrown into a semiconductor manufacturing apparatus that has not executed any processing on the lot. In this case, the semiconductor manufacturing apparatus starts processing only after the lot is input. The processing of the lot is executed with an interval between lots. Such a lot processing mode is called single processing.

  In such two types of lot processing modes, the processing time of the continuous processing semiconductor manufacturing apparatus corresponds to the recipe (processing conditions) used by the semiconductor manufacturing apparatus when compared with the processing time of the single-unit semiconductor manufacturing apparatus. There is a tendency to become longer.

  For example, the tendency to increase the time will be described based on the following example.

  The semiconductor manufacturing apparatus is provided with two processing units for processing a lot of wafers, the first processing unit performs a cleaning process, the second processing unit performs a drying process, and recipes A and B are Both assume that the wafer is configured to execute a drying process in the second processing unit after the wafer is cleaned in the first processing unit.

  In the single processing, it is assumed that the processing time per wafer of the first processing unit executing processing according to Recipe A is 1 minute / sheet, and the processing time per wafer of the second processing unit is 1 minute / sheet. To do. In the single processing, it is assumed that the processing time per wafer of the first processing unit that executes processing according to the recipe B is 2 minutes / sheet, and the processing time per wafer of the second processing unit is 1 minute / sheet. To do.

  Next, when the second processing unit performs a drying process on the wafer after the first processing unit performs the cleaning process on the wafer, the first processing unit is performing the drying process while the second processing unit is performing the drying process. Assume that the unit can perform the next wafer cleaning process. The wafer transfer process to the first processing unit and the wafer unloading process to the second processing unit are executed after both the first processing unit and the second processing unit have completed their respective processes. Assume.

  FIG. 7 shows a processing Gantt chart (FIG. 7A) when a predetermined semiconductor manufacturing apparatus executes a single process with a predetermined recipe (recipe A), and two different recipes (recipe A and recipe B). It is a figure which shows an example with the process Gantt chart (FIG.7 (B)) at the time of performing a continuous process by.

  As shown in FIG. 7A, a semiconductor manufacturing apparatus including a first processing unit and a second processing unit processes a lot (lot A) composed of 12 wafers using recipe A. When the first processing unit starts the cleaning process on the first wafer of the lot A, the processing time required for the second processing unit to complete the drying process on the last wafer of the lot A is 13 Calculated as minutes.

  On the other hand, as shown in FIG. 7B, when the semiconductor manufacturing apparatus processes a lot A using recipe A and then processes a new lot (lot B) using recipe B, that is, When the semiconductor manufacturing apparatus performs continuous processing, the first processing unit starts the cleaning process on the first wafer of the lot A until the second processing unit completes the drying process on the last wafer of the lot A. The processing time required for is calculated as 14 minutes.

  As described above, the processing time for lot A in the single processing (13 minutes) is delayed by one minute compared to the processing time for lot A in continuous processing (14 minutes).

  As shown in FIGS. 7 (A) and 7 (B), the second processing unit performs the drying process in the recipe A on the last wafer of the lot A, and the first processing unit has the lot B. When performing the cleaning process in Recipe B on the first wafer, the second processing unit can complete the drying process in one minute, but the first processing unit requires two minutes to complete the cleaning process. This is because the transfer processing of the last wafer of the lot A from the second processing unit is waited until the processing of the first processing unit is completed. As shown in FIG. 7B, a waiting time 801 of 1 minute occurs in the first processing unit.

  7A and 7B illustrate an example in which the recipe A affects the lot processing time in the recipe A by the recipe B immediately after the recipe A. The recipe C immediately before A may affect the lot processing time in the recipe A.

  For example, in the single processing, the processing time per wafer of the first processing unit that executed processing according to the recipe C is 2 minutes / sheet, the processing time per wafer of the second processing unit is 2 minutes / sheet, In the processing, it is assumed that the processing time per wafer of the first processing unit executing processing according to the recipe A is 1 minute / sheet, and the processing time per wafer of the second processing unit is 1 minute / sheet. is there.

  When the continuous processing semiconductor manufacturing apparatus processes the lot C using the recipe C and then processes the lot A using the recipe A, the second processing unit performs the drying process in the recipe C on the last wafer of the lot C. When the first processing unit executes the cleaning process in the recipe A on the first wafer of the lot A, the first processing unit completes the cleaning process in one minute, but the second processing unit has 2 Complete the drying process in minutes. Therefore, the transfer processing of the first wafer of the lot A from the first processing unit is waited until the processing of the second processing unit is completed. In this case, the recipe C immediately before the predetermined recipe A affects the processing time of the lot of the recipe A.

  As described above, when a semiconductor manufacturing apparatus that performs continuous processing executes a lot process, if the types of recipes before and after a predetermined recipe used by the semiconductor manufacturing apparatus are different, the processing time required for the process may be different. . Such a variation (variation) in processing time is not taken into account in the processing time prediction method described in Patent Document 1 or Patent Document 2 described above, and thus the problem is that the predicted processing time varies when compared with the actual processing time. There is.

  The present invention has been proposed in view of the above problems, and a production management system capable of predicting a lot processing time in a semiconductor manufacturing apparatus with high accuracy and creating a highly accurate lot progress plan. It is intended to provide.

  In order to solve the above-described problems and achieve the object, the production management system according to the present invention has a processing time required from the start of the processing of the scheduled lot scheduled for input by the semiconductor manufacturing apparatus to the completion of the processing. A production management system that creates a progress plan for a planned lot by estimation is assumed.

  The production management system includes data storage means, data collection means, and processing time estimation means.

  The data storage means is an integrated data of the completed recipe used in the lot processing, the previous completed recipe used immediately before the completed recipe, and the later completed recipe scheduled to be used immediately after the completed recipe. The tag is stored in association with the actual measurement processing time required for processing the lot.

  When the semiconductor manufacturing apparatus completes the processing of a predetermined lot, the data collection means associates the actual measurement processing time required for the processing of the lot with the data tag and stores it in the data storage means.

  The processing time estimation means, when estimating the processing time of the scheduled lot, the scheduled recipe used in the processing of the scheduled lot, the previous scheduled recipe to be used immediately before the scheduled recipe, and immediately after the scheduled recipe The processing time of the scheduled lot is estimated by using the actually measured processing time associated with the data tag of the data storage means that matches the scheduled data tag integrated with the scheduled recipe used in the above.

  With this configuration, it is possible to estimate the processing time of the planned lot based on the actual measurement processing time considering the relationship between the recipes before and after the recipe of the planned lot. Therefore, the processing time of the scheduled lot can be set to a processing time with high reproducibility that does not greatly deviate from the processing time when the processing is actually executed. As a result, it is possible to accurately estimate the processing time of the scheduled lot and create a progress plan for the scheduled lot. Furthermore, by adopting an integrated data tag, it is possible to suppress the increase in the amount of data, eliminate the complexity of processing associated with the increase in the amount of data, and improve the ease and speed of data search. Become.

  Further, in the production management system, the data collection means stores the number of completed wafers constituting a lot for which processing has been completed, in association with the data tag, together with the actual measurement processing time, in the data storage means. Further, the processing time estimation means calculates an actual processing time associated with a set of the data tag of the data storage means and the number of completed wafers corresponding to a set of the scheduled data tag and the number of wafers constituting the scheduled lot. By using it, the processing time of the scheduled lot is estimated.

  Here, the set of the data tag and the number of completed wafers of the data storage means corresponding to the set of the scheduled data tag and the number of wafers includes a data tag that matches the scheduled data tag, and the wafer This means the correspondence (combination) with the number of completed wafers that matches the number of sheets.

  With this configuration, since the processing time of the scheduled lot is estimated using the data tag and the number of completed wafers, it is possible to estimate the processing time of the scheduled lot with higher accuracy. It is possible to create a progress plan.

  In the production management system, the processing time estimation means corresponds to the set of the scheduled data tag and the number of wafers, and the actual measurement process associated with the set of the data tag of the data storage means and the number of completed wafers. When there are a plurality of times, the processing time of the scheduled lot is estimated by using the statistical values of the plurality of actually measured processing times.

  With this configuration, since a predetermined statistical value is calculated from a plurality of actually measured processing times, the processing time of the estimated scheduled lot is a processing time with high reproducibility when compared with the processing time actually measured. Therefore, it is possible to create a progress plan for a scheduled lot with higher accuracy.

  In the production management system, when the processing time estimation unit does not have a set of the data tag of the data storage unit and the number of completed wafers corresponding to the set of the scheduled data tag and the number of wafers, By acquiring a relational expression from the relationship between the number of completed wafers and the actual measurement processing time associated with the data tag that matches the planned data tag, and using the relational expression and the number of wafers in the planned lot, Estimate processing time.

  With this configuration, it is possible to estimate the processing time of a scheduled lot from a predetermined relational expression even when there is no actual measurement processing time or the like that matches a predetermined condition in the data storage means.

  According to the present invention, it is possible to accurately estimate the processing time of a scheduled lot. As a result, it is possible to create a progress plan for the scheduled lot with high accuracy, and to efficiently operate the production line.

It is a schematic structure figure showing a production management system concerning an embodiment of the present invention. It is a flowchart for showing the procedure in which the production management system which concerns on embodiment of this invention produces the progress plan of a scheduled lot. It is a figure which shows an example of the data tag memorize | stored in the data storage part 4 which concerns on embodiment of this invention, the number of completed wafers, and measurement process time. It is a figure which shows an example of several measurement process time and average measurement process time which concern on embodiment of this invention. It is a figure which shows an example of the correspondence of the number of completed wafers and measurement process time, the approximate expression, and the process time of a scheduled lot which concern on embodiment of this invention. It is a figure which shows the relationship between the prediction processing time in a prior art, and actual processing time. A processing Gantt chart (FIG. 7A) when a predetermined semiconductor manufacturing apparatus executes a single process with a predetermined recipe, and a processing Gantt chart when a continuous process is executed with two different recipes (FIG. 7 (A)). It is a figure which shows an example with B)).

  Hereinafter, an embodiment of the present invention will be described with reference to the drawings. In the following embodiments, the present invention is embodied as a production management system and production management method for a semiconductor device production line.

  FIG. 1 is a schematic configuration diagram showing a production management system according to an embodiment of the present invention.

  The production management system includes a production line 1, a processing time storage unit 2, a production simulator unit 6, and a lot dispatch unit 9.

  The processing time storage unit 2, the production simulator unit 6, and the lot dispatch unit 9 may be collectively referred to as a manufacturing execution system (MES).

  The production line 1 has a predetermined number n of semiconductor manufacturing apparatuses i (where i = 1 to n: n is the number of semiconductor manufacturing apparatuses existing on the production line, which is indicated as equipment i in FIG. 1). The production line 1 executes lot processing in accordance with instructions from the lot dispatch unit 9. Here, the lot indicates a unit of a predetermined number of wafers, and the predetermined number is set, for example, with the number of wafers stored in the front open integrated pod as a maximum number.

  A predetermined recipe (processing condition) is associated with the lot, and when the semiconductor manufacturing apparatus i processes a predetermined lot, the process is executed according to the recipe associated with the lot.

  The processing time storage unit 2 includes a data collection unit 3, a data storage unit 4, and a data reading unit 5.

  When the production line 1 or a predetermined semiconductor manufacturing apparatus i completes processing of a predetermined lot, the data collection unit 3 detects the completion and acquires (collects) data relating to the processed lot from the production line 1 or the like. To do.

  The data collection unit 3 creates a data tag to be described later from the received data, and stores the data tag and the actual measurement processing time in association with each other in the data storage unit 4.

  When the data reading unit 5 receives a predetermined data read request from the production scheduling unit 7 described later, the data reading unit 5 acquires (extracts) the predetermined data from the data storage unit 4 and outputs it to the processing time estimation unit 8.

  Further, the production simulator unit 6 includes a production scheduling unit 7 and a processing time estimation unit 8.

  First, when the processing time estimation unit 8 receives predetermined data from the data reading unit 5, the processing time estimation unit 8 estimates the processing time of the scheduled lot using the received data. Since the method for estimating the processing time of the scheduled lot is appropriately changed according to the type of data received by the processing time estimation unit 8, the processing time of the planned lot corresponding to the type of data at the time of reception is estimated. .

  In addition, the production scheduling unit 7 acquires, from the production line 1 and the like, data related to scheduled lots (unprocessed lots) that are scheduled to be sequentially inserted into a predetermined semiconductor manufacturing apparatus i at predetermined intervals preset by the user. Then, a request for reading predetermined data is transmitted to the data reading unit 5.

  Further, as will be described later, when the production scheduling unit 7 receives the processing time of the planned lot from the processing time estimation unit 8, the production scheduling unit 7 creates (plans) the progress plan of the planned lot using the processing time of the planned lot. The production scheduling unit 7 transmits the created progress plan of the scheduled lot to the lot dispatch unit 9.

  When the lot dispatch unit 9 receives the progress plan of the scheduled lot from the production scheduling unit 7, the lot dispatch unit 9 receives the progress status of the in-process lot currently in progress from the production line 1, The production line 1 is instructed to process the scheduled lot and the in-process lot based on the progress status of the in-process lot.

  The processing time storage unit 2, the production simulator unit 6, and the lot dispatch unit 9 described above are, for example, dedicated arithmetic circuits or hardware including a processor and a memory such as a RAM (Random Access Memory) and a ROM (Read Only Memory). It can also be realized as software stored in the memory and operating on the processor.

  Next, with reference to FIG. 1 and FIG. 2, a description will be given of a procedure in which the production management system according to the embodiment of the present invention creates a progress plan for a scheduled lot with high accuracy.

  FIG. 2 is a flowchart for showing a procedure for the production management system to create a progress plan for a scheduled lot.

  First, when a predetermined semiconductor manufacturing apparatus i in the production line 1 completes processing of a predetermined lot, the data collection unit 3 of the processing time storage unit 2 detects the completion of processing of the lot, and the semiconductor manufacturing Data on the lot is acquired from the device i (FIG. 2A: S101).

  The lot-related data includes a completed recipe used in the processing of the lot by the semiconductor manufacturing apparatus i, a previous completed recipe used immediately before the completed recipe, and a post-completed recipe scheduled to be used immediately after the completed recipe. And the actual measurement processing time required for processing the lot and the number of completed wafers constituting the lot.

  The completed recipe, the pre-completed recipe, and the post-completed recipe are stored in advance in the production line 1 or the semiconductor manufacturing apparatus i, for example. The actual measurement processing time and the number of completed wafers are obtained, for example, by measuring (monitoring) the semiconductor manufacturing apparatus i that has executed the lot processing, and each time the lot processing is completed, a predetermined storage of the semiconductor manufacturing apparatus i is performed. Stored in the means.

  When the data collection unit 3 acquires data related to the lot, the completed recipe, the pre-completed recipe, and the post-completed recipe in the data are arranged in the order of the completed recipe, the pre-completed recipe, and the post-completed recipe, and these are integrated. Create information (data tag) to be used. Further, the data collection unit 3 associates the created data tag, the number of completed wafers, and the actual measurement processing time with each other and stores them in the data storage unit 4 (FIG. 2A: S102).

  When the data collection unit 3 creates a data tag and there is no pre-completed recipe or (and) post-completed recipe, in the present embodiment, the data collection unit 3 selects the pre-completed recipe or the post-completed recipe. The data tag is created by arranging information indicating that the data does not exist (for example, non-existence identification information such as “Z”) in correspondence with the non-existing pre-completed recipe or the post-completed recipe in the complete recipe. With this configuration, it is possible for the data collection unit 3 to create a data tag for all of the lot processing completed by the semiconductor manufacturing apparatus without determining whether the lot processing mode in the semiconductor manufacturing apparatus is continuous processing or single processing. It becomes.

  The storage processing of the data collection unit 3 is executed for each semiconductor manufacturing apparatus i in the production line 1, and the stored data tags and the like are organized in the data storage unit 4 for each semiconductor manufacturing apparatus i. In the organizing method, for example, a table is prepared in the data storage unit 4 for each device ID that identifies a predetermined semiconductor manufacturing apparatus, and the data tags and the like are stored and organized in the table.

  FIG. 3 is a diagram illustrating an example of the data tag, the number of completed wafers, and the actual measurement processing time stored in the table of the data storage unit 4 for a predetermined apparatus ID.

  As shown in FIG. 3, the data collection unit 3 performs “A” as the completion recipe, “B” as the previous completion recipe, “C” as the subsequent completion recipe, “5” as the number of completed wafers, and “5” as the actual measurement processing time. When “5.0 minutes” is acquired from the predetermined semiconductor manufacturing apparatus i, the data collection unit 3 arranges the data tag as a data tag in the order of the completed recipe, the pre-completed recipe, and the post-completed recipe, and “A, B, C”. The data tag 301 is created and stored in the data storage unit 4. Further, the data collection unit 3 stores the number of completed wafers “5” 302 and the actually measured processing time “5.0” 303 in the data tag 301 in association with each other.

  Now, the production simulator section 6 creates a progress plan for a predetermined scheduled lot when a predetermined period elapses. For example, the scheduled lot for which the progress plan is to be created is transmitted in advance to the production simulator 6 by the production line 1 or the like.

  When the production scheduling unit 7 of the production simulator unit 6 creates a progress plan for a planned lot, it communicates with the production line 1, the semiconductor manufacturing apparatus i, etc., and the semiconductor manufacturing apparatus into which the planned lot is input, and data related to the planned lot Is acquired (FIG. 2B: S201).

  The data related to the planned lot includes a planned recipe used for processing the planned lot, a pre-planned recipe to be used immediately before the planned recipe, a post-planned recipe to be used immediately after the planned recipe, and the planned lot. The number of wafers. These data are stored in advance in the production line 1, the semiconductor manufacturing apparatus i, and the like.

  Next, the production scheduling unit 7 arranges the planned recipe, the pre-planned recipe, and the post-planned recipe in the data related to the planned lot in the order of the planned recipe, the pre-planned recipe, and the post-planned recipe, and integrates them. (Schedule data tag) is created.

  In addition, when the production scheduling unit 7 creates the planned data tag, if the previous planned recipe or (and) the subsequent planned recipe does not exist, as described above, the data collection unit 3 performs the pre-completed recipe or ( And) Similar to the configuration in which the data tag is created using the non-existence identification information corresponding to the post-completed recipe, the production scheduling unit 7 has the pre-existing recipe that does not exist or (and) the non-existence identification information corresponding to the post-planning recipe Create a scheduled data tag using. With this configuration, it is possible to associate the scheduled data tag created by the production scheduling unit 7 with the data tag created by the data collection unit 3.

  When the production scheduling unit 7 creates the scheduled data tag, a request for reading the measured processing time corresponding to the semiconductor manufacturing apparatus in which the scheduled lot processing is executed is transmitted to the data reading unit 5 of the processing time storage unit 2 to create the scheduled data tag. The data tag and the number of wafers in the planned lot are transmitted to the data reading unit 5 (FIG. 2B: S202).

  When the data reading unit 5 receives from the production scheduling unit 7 a request for reading an actual processing time corresponding to a predetermined semiconductor manufacturing apparatus, the scheduled data tag, and the number of wafers, the data reading unit 5 refers to the data storage unit 4, A table corresponding to the semiconductor manufacturing apparatus is specified, and it is determined whether or not a data tag that matches the scheduled data tag exists in the table (FIG. 2B: S203).

  If the data reading unit 5 determines that there is a data tag that matches the scheduled data tag as a result of the determination, the data reading unit 5 next matches the number of wafers and the number of completed wafers associated with the data tag. Is determined (FIG. 2B: S203 YES → S204).

  If the data reading unit 5 determines that there is a completed wafer number that matches the number of wafers, the data reading unit 5 acquires an actual measurement processing time associated with the completed wafer number (FIG. 2B ): S204 YES → S205).

  As shown in FIG. 3, when the data reading unit 5 receives the scheduled data tag “A, B, C” and the number of wafers “5”, the data tag “A, B, C” that matches the scheduled data tag. 301 and the number of completed wafers corresponding to the number of wafers “5” 302 exist in a predetermined table (data storage unit 4), the data reading unit 5 includes data tags “A, B, C” 301, wafers The actual measurement processing time “5.0” 303 associated with the number “5” 302 is acquired.

  Although FIG. 3 shows one actual measurement processing time “5.0”, the actual measurement processing time associated with the data tag that matches the scheduled data tag and the number of completed wafers that matches the number of wafers. When there are a plurality of tables (data storage unit 4), the data reading unit 5 acquires all the actual measurement processing times.

  When the data reading unit 5 acquires the actual measurement processing time, the acquired actual measurement processing time, the scheduled data tag, and the number of wafers are transmitted to the processing time estimation unit 8.

  When the processing time estimation unit 8 receives the actual measurement processing time, the scheduled data tag, and the number of wafers, when the actual processing time is received, the data tag corresponding to the planned data tag and the number of wafers; It is determined that the number of completed wafers exists in the table, and predetermined statistical values (for example, average value, median value, maximum value, minimum value, etc.) are obtained from the actually measured processing time received according to a statistical method preset by the user. Is calculated (FIG. 2B: S206).

  The predetermined statistical value is arbitrarily designed by the user in accordance with the configuration of the production line, the configuration of each semiconductor manufacturing apparatus, the processing applied to the lot, the recipe used for the processing of the lot, and the like. In the embodiment of the present invention, the average value is adopted as the statistical value.

  FIG. 4 is a diagram illustrating an example of a plurality of actual measurement processing times and an average actual measurement processing time according to the embodiment of the present invention.

  As shown in FIG. 4, when the processing time estimation unit 8 receives a plurality of actual measurement processing times 401 (seven actual measurement processing times), an average actual measurement processing time “5.0” 402 is calculated by averaging the actual measurement processing times. To do. The processing time estimation unit 8 estimates the average actually measured processing time “5.0” 402 as the processing time of the scheduled lot.

  Next, a case will be described in which there is a data tag that matches the scheduled data tag, but there is no completed wafer number in the data storage unit 4 that matches the number of wafers.

  If the data reading unit 5 determines that there is no completed wafer number that matches the number of wafers, the data reading unit 5 determines the number of completed wafers associated with the data tag that matches the scheduled data tag and the actual measurement process. The number of images is acquired (FIG. 2: S204 NO → S207).

  As shown in FIG. 3, when the data reading unit 5 receives the scheduled data tag “A, B, C” and the number of wafers “8”, the data tag “A, B, C” that matches the scheduled data tag. "301" exists in the table, but the number of completed wafers corresponding to the number of wafers does not exist in the table. Therefore, the data reading unit 5 includes the number of completed wafers (“5” 302, “12” 304) associated with two existing data tags “A, B, C” 301 and the actual measurement processing time (“5.0 "303," 10.0 "305).

  When the data reading unit 5 acquires the number of completed wafers and the actual measurement processing time, it transmits the acquired number of completed wafers, actual measurement processing time, scheduled data tag, and number of wafers to the processing time estimation unit 8.

  When the processing time estimation unit 8 receives the number of completed wafers, the actually measured processing time, the scheduled data tag, and the number of wafers, when the received number of completed wafers and the actually measured processing time are received, the number of completed wafers and the wafer Since the number is different, it is determined that the data tag that matches the scheduled data tag exists in the table, but the completed wafer number that matches the number of wafers does not exist in the table. Then, the processing time estimation unit 8 calculates a processing time corresponding to the number of wafers by a method (hereinafter referred to as a first method) different from the statistical method described above, and calculates the processing time of the scheduled lot. Estimation is performed (FIG. 2B: S208).

  Any method may be used as the first method for the processing time estimation unit 8 to estimate the processing time of the scheduled lot. For example, the following estimation method is adopted.

  First, the processing time estimation unit 8 applies a predetermined function (for example, a linear function) to the received number of completed wafers and the actually measured processing time by a least square method or the like to obtain an approximate expression. Next, the processing time estimation unit 8 substitutes the number of wafers in the planned lot for the parameter corresponding to the number of completed wafers in the approximate expression, and calculates the calculated actual processing time as the processing time corresponding to the number of wafers. . Then, the processing time estimation unit 8 estimates the calculated processing time as the processing time of the scheduled lot.

  FIG. 5 is a diagram illustrating an example of a correspondence relationship between the number of completed wafers and an actual measurement processing time, an approximate expression, and a processing time of a scheduled lot according to the embodiment of the present invention.

  As shown in FIG. 5, when the processing time estimation unit 8 receives the completed wafer number “5”, the actually measured processing time “5.0”, the completed wafer number “12”, and the actually measured processing time “10.0”. Then, a linear function is obtained by fitting a linear function to both. In FIG. 5, when the number of completed wafers is X (number) and the actual measurement processing time is Y (minutes), the linear approximation formula 501 is Y = 0.71X + 1.43. Therefore, when the number of wafers in the planned lot is “8”, the processing time estimation unit 8 substitutes the number of wafers “8” for X in the linear approximation formula 501, and actually measures the number of wafers “8”. The processing time is calculated as “7.1” 502. Thereby, the processing time estimation unit 8 estimates the processing time of the scheduled lot as “7.1” 502.

  Note that the function adopted by the processing time estimation unit 8 in the linear approximation need not be a linear function. For example, a straight line corresponding to the equipment model, a step straight line, or the like may be adopted. Further, the function adopted by the processing time estimation unit 8 in the linear approximation need not use a predetermined function. For example, the processing time estimation unit 8 may be configured to estimate the processing time of the scheduled lot using a simple interpolation formula. I do not care.

  By the way, the above-described first method can be adopted when the number of completed wafers is two or more and the actual measurement processing time is present, but is adopted when there is one completed wafer number and the actual measurement processing time. I can't. Therefore, when the processing time estimation unit 8 acquires the number of completed wafers of 1 and the actual measurement processing time, for example, the following second method is adopted.

  First, a function or the like corresponding to a conventionally used equipment model is previously stored by a user in a predetermined storage unit. Next, when the processing time estimation unit 8 acquires the number of completed wafers of 1 and the actual processing time, the function is read from the storage unit, and the function is based on the function, the number of completed wafers, and the actual processing time. Determine the parameters. Then, the processing time estimation unit 8 estimates the processing time of the planned lot based on the function that has determined the parameter and the number of wafers of the planned lot. In this way, the processing time of the scheduled lot can be estimated without adopting the first method described above.

  In the above, after determining the parameters of the function corresponding to the equipment model, the processing time of the planned lot is estimated from the function and the number of wafers of the planned lot. For example, the parameter is fixed. In this case, the processing time for the planned lot may be estimated from the function with the fixed parameters and the number of wafers of the planned lot without executing the process for determining the parameter.

  Finally, a case will be described in which neither the data tag that matches the scheduled data tag nor the number of completed wafers that matches the number of wafers exists in the data storage unit 4.

  If the data reading unit 5 determines that there is no data tag that matches the scheduled data tag, the data reading unit 5 transmits the scheduled data tag and the number of wafers to the processing time estimation unit 8 (FIG. 2). (B): S203 NO).

  When the processing time estimation unit 8 receives the scheduled data tag and the number of wafers, when the scheduled data tag and the number of wafers are received, the data tag corresponding to the scheduled data tag and the number of wafers, the completed wafer It is determined that the number of sheets does not exist in the table. Then, the processing time estimation unit 8 estimates the processing time of the scheduled lot by another method (FIG. 2B: S209).

  The method for estimating the processing time of the scheduled lot by the processing time estimation unit 8 may be any method. For example, the following method is adopted.

  First, a function or the like corresponding to a conventionally used equipment model is previously stored by a user in a predetermined storage unit. When the processing time estimation unit 8 determines that the scheduled data tag, the data tag corresponding to the number of wafers, and the number of completed wafers do not exist in the table, the processing time estimation unit 8 matches the scheduled recipe of the scheduled data tag via the data reading unit 5. It is determined whether or not a data tag completion recipe exists in the data storage unit 4.

  When it is determined that a completed recipe that matches the scheduled recipe exists in the table, the processing time estimation unit 8 acquires the number of completed wafers associated with the completed recipe and the actually measured processing time via the data reading unit 5. Next, the processing time estimation unit 8 reads a function corresponding to the equipment model from the storage unit, and determines a parameter related to the function based on the function, the number of completed wafers, and the actually measured processing time. Then, the processing time estimation unit 8 estimates the processing time of the planned lot based on the function that has determined the parameter and the number of wafers of the planned lot. Such a method is the same as the second method employed in the case where the number of completed wafers of 1 and the actual measurement processing time exist as described above.

  In this way, even if there is no data tag that matches the above-mentioned scheduled data tag in the table, it is possible to acquire the measured processing time from the table and estimate the processing time of the scheduled lot.

  When the processing time estimation unit 8 estimates the processing time of the scheduled lot, the processing time estimation unit 8 transmits the estimated processing time of the planned lot to the production scheduling unit 7. The production scheduling unit 7 creates a scheduled lot progress plan based on the received processing time of the planned lot (FIG. 2B: S210).

  In addition, when the planned lot is processed via a plurality of semiconductor manufacturing apparatuses among the progress plans of the planned lot, the production scheduling unit 7 communicates with the production line 1 and the like, and a plurality of planned lots are input. The semiconductor manufacturing apparatus is specified, and the processing time of the scheduled lot is acquired for each semiconductor manufacturing apparatus.

  Specifically, the production scheduling unit 7 transmits a read request to the data reading unit 5 so as to read the actual measurement processing time corresponding to the scheduled lot for each semiconductor manufacturing apparatus, and estimates the processing time in response to the read request. The processing time of the scheduled lot for each semiconductor manufacturing apparatus estimated from the unit 8 is received. The production scheduling unit 7 creates a scheduled lot progress plan based on the received processing time of the scheduled lot for each semiconductor manufacturing apparatus.

  When the production scheduling unit 7 creates a progress plan for the scheduled lot, the production scheduling unit 7 transmits the created progress plan for the scheduled lot to the lot dispatch unit 9. When the lot dispatch unit 9 receives the progress plan of the scheduled lot, the lot dispatch unit 9 obtains the progress status of the in-process lot currently in progress from the production line 1, and the progress plan of the scheduled lot and the progress of the in-process lot. The order of the scheduled lot and the in-process lot for the predetermined semiconductor manufacturing apparatus i is determined using the situation. For example, the determination method is the order of insertion that minimizes the processing time.

  When the order of input is determined, the lot dispatch unit 9 instructs the production line 1 so that a predetermined semiconductor manufacturing apparatus processes the scheduled lot and the in-process lot in the order of input. That is, the lot dispatch unit 9 assigns a scheduled lot to a predetermined semiconductor manufacturing apparatus i and causes the semiconductor manufacturing apparatus i to execute the processing of the scheduled lot.

  In the semiconductor manufacturing apparatus i, processing of the assigned scheduled lot is executed. When the semiconductor manufacturing apparatus i processes a planned lot, it processes according to the schedule plan of the planned lot described above. For this reason, as described in the related art, the planned lot is generated without causing a large variation between the actual processing time for processing the planned lot in the semiconductor manufacturing apparatus i and the processing time used for creating the progress plan. It becomes possible to improve the delivery rate compliance.

  The embodiment of the present invention is not limited to the embodiment described above, and various modifications and applications can be made without departing from the technical idea of the present invention.

  For example, the production simulator unit 6 is configured to create a progress plan for a scheduled lot every predetermined period. However, the production simulator unit 6 is not limited to every predetermined period, and each time a predetermined semiconductor manufacturing apparatus i completes processing of a predetermined lot, You may comprise so that the said production simulator part 6 produces the progress plan of a scheduled lot.

  With this configuration, when the processing time required for processing a lot by the semiconductor manufacturing apparatus i is shorter than the period, data relating to the lot is stored in the data storage unit 4 and the data reading unit 5 is related to the lot. Since the data is also a search target, the production scheduling unit 7 can create a progress plan for the scheduled lot corresponding to the data relating to the newly stored lot. As a result, since the progress plan of the planned lot to be created reflects new data, it is possible to further improve the accuracy of the progress plan of the planned lot.

  In the embodiment of the present invention, the data collection unit 3 stores the data tag, the number of completed wafers, and the actual measurement processing time in association with each other in the data storage unit 4, and the data reading unit 5 stores the scheduled data tag and the wafer. The data tag corresponding to the number of sheets and the actual measurement processing time associated with the number of completed wafers are acquired, but other structures may be used.

  For example, the data collection unit 3 stores the data tag and the actual measurement processing time in association with each other except the number of completed wafers in the data storage unit 4, and the data reading unit 5 matches (corresponds to) the scheduled data tag. You may comprise so that the measurement processing time linked | related with the tag may be acquired.

  Even in this configuration, it is possible to estimate the processing time of the scheduled lot in consideration of the waiting time resulting from the mutual relationship between the recipes, and it is possible to improve the accuracy of the processing time of the estimated scheduled lot.

  In the embodiment of the present invention, the data collection unit 3 arranges the completed recipe, the pre-completed recipe, and the post-completed recipe in this order, creates a data tag that integrates them, and corresponds to the creation method. In addition, the production scheduling unit 7 is configured to arrange the order recipe, the pre-scheduled recipe, and the post-scheduled recipe in this order to create a schedule data tag that integrates them, but other configurations may be used. .

  For example, without considering the arrangement mode, the data collection unit 3 creates a data tag in which a completed recipe, a pre-completed recipe, and a post-completed recipe are integrated, and corresponds to the data tag, The production scheduling unit 7 may be configured to create a schedule data tag that integrates the scheduled recipe, the previous scheduled recipe, and the later scheduled recipe.

  In the embodiment of the present invention, the data collection unit 3 is configured to create the data tag by arranging the completed recipe, the pre-completed recipe, and the post-completed recipe in this order. You may comprise so that it may produce.

  Examples of the arrangement form include, but are not limited to, an aspect of arranging in the order of pre-completed recipe, completed recipe, and post-completed recipe, an aspect of arranging in the order of post-completed recipe, completed recipe, and pre-completed recipe. It is not a thing.

  In the embodiment of the present invention, information indicating that a predetermined recipe does not exist (non-existence identification) regardless of whether the data collection unit 3 creates a data tag or the production scheduling unit 7 creates a scheduled data tag. The data collection unit 3 uses the information) and the production scheduling unit 7 creates the scheduled data tag. However, other configurations may be used.

  For example, when the data collection unit 3 completes the processing of a predetermined lot by a predetermined semiconductor manufacturing apparatus i, it is determined whether the semiconductor manufacturing apparatus has completed the lot processing by a single process or the lot processing by a continuous process. It is configured so as to determine, and together with the data relating to the lot, information indicating whether it is a single process or continuous process (processing mode information) is acquired and stored in the data storage unit 4 including the processing mode information. It doesn't matter. Then, the production scheduling unit 7 adds the processing mode information to the data related to the planned lot, and acquires data (for example, actual measurement processing time) related to the lot corresponding to the data related to the planned lot from the data storage unit 4. You may comprise.

  With this configuration, the production scheduling unit 7 can acquire the actual measurement processing time that is the actual value in consideration of the variation in the processing time of the lot according to the processing mode. It is possible to improve the accuracy of the progress plan. In addition, a large amount of accumulated data can be efficiently organized.

  The present invention can predict the processing time required for processing a predetermined lot for a semiconductor manufacturing apparatus with high accuracy, and as a result, can create a highly accurate lot progress plan and It has the effect of being able to operate with high efficiency and is useful as a production management system.

DESCRIPTION OF SYMBOLS 1 Production line 2 Processing time storage part 3 Data collection part 4 Data storage part 5 Data reading part 6 Production simulator part 7 Production scheduling part 8 Processing time estimation part 9 Lot dispatch part

Claims (4)

In the production management system that creates the progress plan of the planned lot by estimating the processing time required from the start of the processing of the planned lot scheduled for input by the semiconductor manufacturing equipment to the completion of the processing,
When the semiconductor manufacturing apparatus completes the processing of the predetermined lot, the completed recipe used in the processing of the lot, the previous completed recipe used immediately before the completed recipe, and the schedule to be used immediately after the completed recipe A data collection means for associating an actual processing time required for processing of the lot with a data tag integrated with a post-completion recipe, and storing it in a data storage means;
When estimating the processing time of a scheduled lot, the scheduled recipe used in the processing of the scheduled lot, the previous scheduled recipe used immediately before the scheduled recipe, and the scheduled schedule used immediately after the scheduled recipe Processing time estimation means for estimating the processing time of the scheduled lot by using the actual measurement processing time associated with the data tag of the data storage means, which matches the recipe data tag integrated with the recipe;
A production management system comprising: The data collection means associates with the data tag the number of completed wafers constituting the processed lot together with the actual measurement processing time, and stores it in the data storage means.
The processing time estimation means uses an actual processing time associated with a set of the data tag of the data storage means and the number of completed wafers corresponding to a set of the scheduled data tag and the number of wafers constituting the scheduled lot. The production management system according to claim 1, wherein the processing time of the scheduled lot is estimated by:   When the processing time estimation means has a plurality of measured processing times associated with a set of the data tag of the data storage means and the number of completed wafers corresponding to the set of the scheduled data tag and the number of wafers, a plurality of The production management system according to claim 2, wherein the processing time of the scheduled lot is estimated by using the statistical value of the actual measurement processing time.   When the processing time estimation unit does not have a set of the data tag and the number of completed wafers corresponding to the set of the scheduled data tag and the number of wafers, the data that matches the scheduled data tag A relational expression is acquired from the relationship between the number of completed wafers and the actual processing time associated with the tag, and the processing time of the planned lot is estimated by using the relational expression and the number of wafers of the planned lot. The production management system according to claim 2 or 3.

Images