JP4892175B2 - Calibration method for semiconductor manufacturing equipment - Google Patents

Description

  The present invention relates to a calibration method for a semiconductor manufacturing apparatus.

  Usually, a semiconductor device such as a transistor is obtained by subjecting a wafer to a number of semiconductor manufacturing processes. In each semiconductor manufacturing process, a semiconductor manufacturing apparatus such as an etching apparatus, a CVD apparatus, a PVD apparatus, an ion implantation apparatus, and a heat treatment apparatus is used. For example, in an etching apparatus, when a predetermined recipe (processing condition) is executed, a wafer is etched corresponding to the recipe. In the recipe, for example, the type and flow rate of the process gas, the pressure and temperature in the chamber, and the like are set in advance.

  Many sensors are attached to the semiconductor manufacturing apparatus as described above. When these sensors are used, for example, the pressure in the chamber, the high-frequency power output, the gas flow rate, the heater temperature and the like during the operation of the semiconductor manufacturing apparatus can be monitored.

Also, a method is known in which part of data is extracted from various data collected from a semiconductor manufacturing factory that mass-produces semiconductor devices, and data mining is performed on the extracted data (for example, see Patent Document 1). ).
International Publication No. 03/012696 Pamphlet

  By the way, when a semiconductor manufacturing apparatus is newly introduced into a factory for manufacturing semiconductor devices, or when an existing semiconductor manufacturing apparatus in the factory is moved to another place, a so-called semiconductor manufacturing apparatus is started up. At the start-up, calibration is performed on the semiconductor manufacturing apparatus moved to the installation location. Accordingly, the semiconductor manufacturing apparatus can exhibit predetermined characteristics and can perform a desired semiconductor manufacturing process.

  In calibration, the operation and output of the control module (for example, pressure control valve) or sensor (for example, pressure gauge) corresponding to the inspection item (for example, pressure in the chamber) at the time of start-up is within a predetermined specification. Adjust manufacturing equipment and replace parts. The calibration of the semiconductor manufacturing apparatus is completed by sequentially performing this operation for all inspection items.

  However, since there are many cases where there are 100 or more inspection items at the time of start-up, the calibration method described above takes a considerable amount of time to calibrate the semiconductor manufacturing apparatus. For this reason, it usually takes about two months to start up the semiconductor manufacturing apparatus.

  Accordingly, an object of the present invention is to provide a semiconductor manufacturing apparatus calibration method capable of reducing the time required for calibration of a semiconductor manufacturing apparatus after being moved to an installation location.

  In order to solve the above-described problem, a calibration method for a semiconductor manufacturing apparatus according to the present invention is a calibration method for a semiconductor manufacturing apparatus that has been moved from a first location to a second location different from the first location, When a semiconductor manufacturing apparatus installed at a first location is operated under a first operating condition for inspecting a plurality of inspection items corresponding to a plurality of sensors attached to the semiconductor manufacturing apparatus, a plurality of A step of preparing a recording medium on which first data including a plurality of first output data collected by using the sensor is recorded, and a semiconductor manufacturing apparatus installed at the second location are set to a first operating condition. When operating below, the step of collecting second data including a plurality of second output data using a plurality of sensors is compared with the first data and the second data, The second output data indicating an abnormality with respect to the output data of There comprises generating a comparison result of the extraction, and performing the calibration of the semiconductor manufacturing apparatus based on the comparison result.

  In this method, the first data and the second data are compared, and the second output data indicating abnormality with respect to the first output data is extracted. Therefore, it is only necessary to calibrate the semiconductor manufacturing apparatus with respect to the inspection item corresponding to the second output data indicating abnormality, so that it is not necessary to calibrate all of the plurality of inspection items. Therefore, if this method is used, the time required for calibration of the semiconductor manufacturing apparatus can be shortened as compared with the case where the semiconductor manufacturing apparatus is calibrated for all of the plurality of inspection items.

  Furthermore, since the first data and the second data obtained from the same semiconductor manufacturing apparatus are compared, it is possible to eliminate the influence of individual differences in the semiconductor manufacturing apparatus.

  Further, it is preferable to use data mining in the step of generating the comparison result. In this case, a comparison result can be generated in a short time even if the amount of the first and second data is enormous. For this reason, the time required for calibration of the semiconductor manufacturing apparatus can be further shortened.

  The first operating condition is preferably an operating condition for operating the hardware of the semiconductor manufacturing apparatus. In this case, the hardware can be inspected without performing the semiconductor manufacturing process. Therefore, it is not necessary to consider the influence of the semiconductor manufacturing process on the hardware, so that the hardware inspection becomes easy.

  The first output data and the second output data are preferably time series data. In this case, the detection sensitivity when detecting an abnormality of the second output data with respect to the first output data is improved.

  A calibration method for a semiconductor manufacturing apparatus according to the present invention is a calibration method for a first semiconductor manufacturing apparatus that has been moved from a first location to a second location different from the first location. The first semiconductor manufacturing apparatus installed at the place operates under a first operating condition for inspecting a plurality of inspection items corresponding to the plurality of first sensors attached to the first semiconductor manufacturing apparatus. And collecting a first data including a plurality of first output data using a plurality of first sensors, and a second reference that is used when the first semiconductor manufacturing apparatus is calibrated. When the semiconductor manufacturing apparatus is operated under the first operating condition, a plurality of second sensors collected using a plurality of second sensors attached to the second semiconductor manufacturing apparatus and corresponding to a plurality of inspection items are collected. Record in which second data including output data of 2 is recorded Preparing the body, comparing the first data and the second data, and generating a comparison result in which the first output data indicating abnormality with respect to the second output data is extracted; And a step of calibrating the first semiconductor manufacturing apparatus based on the comparison result.

  In this method, the first data is compared with the second data, and the first output data indicating abnormality with respect to the second output data is extracted. Therefore, it is only necessary to calibrate the semiconductor manufacturing apparatus with respect to the inspection item corresponding to the first output data indicating abnormality, so that it is not necessary to calibrate all of the plurality of inspection items. Therefore, if this method is used, the time required for calibration of the semiconductor manufacturing apparatus can be shortened as compared with the case where the semiconductor manufacturing apparatus is calibrated for all of the plurality of inspection items.

  Further, it is preferable to use data mining in the step of generating the comparison result. In this case, a comparison result can be generated in a short time even if the amount of the first and second data is enormous. For this reason, the time required for calibration of the semiconductor manufacturing apparatus can be further shortened.

  The first operating condition is preferably an operating condition for causing the first semiconductor manufacturing apparatus and the second semiconductor manufacturing apparatus to perform a semiconductor manufacturing process. Thereby, it becomes easy to match the semiconductor manufacturing process performed by the first semiconductor manufacturing apparatus with the desired semiconductor manufacturing process performed by the second semiconductor manufacturing apparatus serving as a reference.

  The first output data and the second output data are preferably time series data. In this case, the detection sensitivity when detecting an abnormality of the first output data with respect to the second output data is improved.

  The semiconductor manufacturing apparatus calibration method according to the present invention is a first semiconductor manufacturing apparatus calibration method that has been moved from a first location to a second location different from the first location. A first operating condition for inspecting a plurality of first inspection items corresponding to a plurality of first sensors attached to the first semiconductor manufacturing apparatus installed at the place Preparing a first recording medium having recorded thereon first data including a plurality of first output data collected using a plurality of first sensors when operated below; When the first semiconductor manufacturing apparatus installed at the location is operated under the first operating condition, the second data including the plurality of second output data is collected using the plurality of first sensors. Comparing the first data and the second data to the first output data. Generating a first comparison result from which second output data indicating abnormality is extracted, performing a first calibration of the first semiconductor manufacturing apparatus based on the first comparison result, A second semiconductor device for inspecting a plurality of second inspection items corresponding to a plurality of second sensors attached to the first semiconductor manufacturing device installed in the second location. Collecting the third data including the plurality of third output data using the plurality of second sensors when operated under the operating conditions of the second semiconductor device, and the second calibration of the first semiconductor manufacturing apparatus When the second semiconductor manufacturing apparatus serving as a reference for performing the operation is operated under the second operating condition, a plurality of second semiconductor manufacturing apparatuses attached to the second semiconductor manufacturing apparatus and corresponding to a plurality of second inspection items A first including a plurality of fourth output data collected using three sensors; Comparing the third data and the fourth data with the step of preparing the second recording medium having the data recorded thereon, and extracting the third output data indicating abnormality with respect to the fourth output data Generating a second comparison result, and performing a second calibration of the first semiconductor manufacturing apparatus based on the second comparison result.

  In this method, the first data and the second data are compared, and the second output data indicating abnormality with respect to the first output data is extracted. Therefore, since the first calibration of the semiconductor manufacturing apparatus may be performed for the first inspection item corresponding to the second output data indicating abnormality, it is necessary to perform the first calibration for all the plurality of first inspection items. Disappears. Therefore, when this method is used, the time required for the first calibration of the semiconductor manufacturing apparatus can be shortened compared to the case where the first calibration of the semiconductor manufacturing apparatus is performed for all of the plurality of first inspection items.

  Furthermore, since the first data and the second data obtained from the same first semiconductor manufacturing apparatus are compared, it is possible to eliminate the influence of individual differences in the semiconductor manufacturing apparatus.

  Further, the third data is compared with the fourth data, and third output data indicating abnormality with respect to the fourth output data is extracted. Therefore, since the second calibration of the semiconductor manufacturing apparatus has only to be performed for the second inspection item corresponding to the third output data indicating abnormality, it is necessary to perform the second calibration for all the plurality of second inspection items. Disappears. Therefore, if this method is used, the time required for the second calibration of the semiconductor manufacturing apparatus can be shortened compared to the case where the second calibration of the semiconductor manufacturing apparatus is performed for all of the plurality of second inspection items.

  Further, it is preferable to use data mining in the step of generating the first comparison result and the step of generating the second comparison result. In this case, the first and second comparison results can be generated in a short time even if the amount of the first to fourth data is enormous.

  The first operating condition is an operating condition for operating hardware of the first semiconductor manufacturing apparatus, and the second operating condition is a semiconductor manufacturing process in the first semiconductor manufacturing apparatus and the second semiconductor manufacturing apparatus. It is preferable that the operating conditions are such that

  In this case, when performing the first calibration, the hardware can be inspected without performing the semiconductor manufacturing process. Therefore, it is not necessary to consider the influence of the semiconductor manufacturing process on the hardware, so that the hardware inspection becomes easy. In addition, when performing the second calibration, it becomes easy to match the semiconductor manufacturing process performed by the first semiconductor manufacturing apparatus with the desired semiconductor manufacturing process performed by the second semiconductor manufacturing apparatus serving as a reference.

  The first to fourth output data are preferably time series data. In this case, the detection sensitivity when detecting the abnormality of the second output data with respect to the first output data and the abnormality of the third output data with respect to the fourth output data is improved.

  ADVANTAGE OF THE INVENTION According to this invention, the calibration method of the semiconductor manufacturing apparatus which can shorten the time required for calibration of the semiconductor manufacturing apparatus after moving to an installation place can be provided.

  Hereinafter, a calibration method for a semiconductor manufacturing apparatus according to a preferred embodiment of the present invention will be described with reference to the drawings. In the description of the drawings, the same reference numerals are used for the same or equivalent elements, and duplicate descriptions are omitted.

(Preparation process of the first recording medium)
1A to 1C show a process in which the semiconductor manufacturing apparatus 100 (first semiconductor manufacturing apparatus) is moved from a location A (first location) to a location B (second location). FIG. Location A and location B are different. The location A indicates, for example, a factory where the semiconductor manufacturing apparatus 100 is produced, and the location B indicates, for example, a factory where semiconductor devices are manufactured using the semiconductor manufacturing apparatus 100. Location A and location B may be in the same factory. Examples of the semiconductor manufacturing apparatus 100 include an etching apparatus, a CVD apparatus, a PVD apparatus, an ion implantation apparatus, and a heat treatment apparatus.

  First, as shown in FIG. 1A, the semiconductor manufacturing apparatus 100 installed at the location A is operated using a recipe R1 (first operating condition). The semiconductor manufacturing apparatus 100 includes, for example, chambers C1 to C4. In the semiconductor manufacturing apparatus 100, n sensors S1 to Sn (first sensors) are attached. Here, n is an integer of 2 or more, and the same applies to the following description. Examples of the sensors S1 to Sn include a pressure sensor, a temperature sensor, and a flow rate sensor. When the recipe R1 is used, as shown in FIG. 2A, data D1 (first data) including output data Ra1 to Ran (first output data) from the sensors S1 to Sn is collected.

  Using such output data Ra1 to Ran from the sensors S1 to Sn, the inspection items A1 to Am (first inspection items) can be inspected. Examples of the inspection items A1 to Am include the pressure and temperature in the chambers C1 to C4, the gas flow rate, and the like. FIG. 2A is a table showing output data Ra1 to Ran corresponding to the inspection items A1 to Am. Here, m is an integer of 2 or more, and the same applies to the following description. The value of m may be the same as or different from the value of n. For example, one inspection item Am may correspond to one sensor Sn, and one inspection item Am may correspond to a plurality of arbitrary sensors selected from the sensors S1 to Sn.

  More specifically, for example, the output data Ran is preferably collected sequentially at predetermined time intervals. The output data Ran may be time series data composed of numerical values an1 to ank collected in time series at times t1 to tk. Here, k is an integer of 1 or more, and the same applies to the following description. Each interval between the times t1 to tk can be set to 1 second, for example.

  As shown in FIG. 1A, the data D1 is recorded from the data logger DL1 to the recording medium DB1 (first recording medium). Examples of the recording medium DB1 include a compact disk (CD), DVD, magneto-optical disk (MO), flexible disk (FD), and hard disk (HD).

  Next, as shown in FIG. 1B, at least a part of the semiconductor manufacturing apparatus 100 is disassembled as necessary, and the semiconductor manufacturing apparatus 100 is moved from the location A to the location B. Further, the data logger DL1 is moved from the location A to the location B as necessary. Furthermore, the recording medium DB1 is moved from the location A to the location B as necessary. The semiconductor manufacturing apparatus 100, the data logger DL1, and the recording medium DB1 may be moved separately or simultaneously. Further, the data D1 may be moved from the location A to the location B not through the recording medium but through a telecommunication line or the like. Further, at the location A, the data D1 may be copied to a recording medium different from the recording medium DB1, and the recording medium may be moved from the location A to the location B.

  As described above, the recording medium DB1 is prepared at the location B as shown in FIG.

(Second data collection process)
Next, as shown in FIG. 1C, the semiconductor manufacturing apparatus 100 is reassembled as necessary, and the semiconductor manufacturing apparatus 100 is installed in the place B. Subsequently, the semiconductor manufacturing apparatus 100 installed at the location B is operated using the recipe R1. When the recipe R1 is used, as shown in FIG. 2B, data D1a (second data) including output data Rb1 to Rbn (second output data) from the sensors S1 to Sn is collected. FIG. 2B is a table showing output data Rb1 to Rbn corresponding to the inspection items A1 to Am.

  More specifically, the output data Rbn is preferably collected sequentially at a predetermined time interval. The output data Rbn can be time series data composed of numerical values bn1 to bnk collected in time series at times t1a to tka. Each interval between the times t1a to tka can be assumed to be 1 second, for example.

  The data D1a is recorded from the data logger DL1 to the recording medium DB2 as shown in FIG. An example of the recording medium DB2 is the same as the recording medium DB1.

(First comparison result generation step)
Next, as shown in FIG. 3, the data D1a and the data D1 are compared to generate a comparison result CR1 (first comparison result). FIG. 3 is a block diagram illustrating a process of generating the comparison result CR1. In the comparison result CR1, for example, output data Rbn indicating an abnormality with respect to the output data Ran is extracted. More specifically, for example, a case where the difference between the output data Ran and the output data Rbn is equal to or greater than a predetermined threshold value can be a case where the output data Rbn indicates an abnormality. Further, as a method of displaying the comparison result CR1, the output data Ran, Rbn may be displayed on the display device.

  FIG. 4 is a graph showing an example of the comparison result CR1 displayed on the display device. In this graph, the vertical axis indicates the pressure P in the chamber, and the horizontal axis indicates time t. The chart (time series data) shown in the graph is obtained by collecting the pressure values in the chamber measured every second in time series. Here, the pressure P in the chamber is an example of the inspection item Am, and each chart corresponds to an example of the output data Ran, Rbn. In FIG. 4, output data collected after the first calibration described later is also displayed.

  In a region b in FIG. 4, it is confirmed that the difference Δ1 between the output data Ran and Rbn is equal to or greater than a predetermined threshold value. In such a case, it is confirmed that the output data Rbn is abnormal with respect to the reference output data Ran. More specifically, the zero point is shifted in the output data Rbn.

  Further, the comparison result CR1 as described above is preferably generated using data mining. When data mining is used, output data Rbn indicating an abnormality with respect to the reference output data Ran can be efficiently extracted. For this reason, even if the amount of the data D1a and the data D1 is enormous, the comparison result CR1 can be generated in a short time.

(First calibration process)
Next, the semiconductor manufacturing apparatus 100 is calibrated based on the comparison result CR1. In the comparison result CR1, for example, an abnormality of the output data Rbn (for example, pressure data) is confirmed as shown in a region b in FIG. In this case, it becomes necessary to inspect the inspection item Am (for example, the pressure in the chamber) corresponding to the output data Rbn. The semiconductor manufacturing apparatus 100 can be calibrated by repairing, adjusting or exchanging parts (for example, a pressure sensor) of the semiconductor manufacturing apparatus 100 related to the inspection item Am. As a result, the output data Ran and Rbn substantially overlap each other as shown in the region a of FIG. Here, the calibration of the semiconductor manufacturing apparatus 100 includes repair, adjustment or replacement of parts of the semiconductor manufacturing apparatus 100. When the semiconductor manufacturing apparatus 100 is calibrated, the semiconductor manufacturing apparatus 100 can exhibit predetermined characteristics. The difference between the chambers C1 to C4 can be adjusted based on the comparison result CR1.

  As described above, the first calibration of the semiconductor manufacturing apparatus 100 is performed. In the first calibration, the data D1a and the data D1 are compared, and output data Rbn indicating an abnormality with respect to the output data Ran is extracted. Therefore, since the semiconductor manufacturing apparatus 100 may be calibrated for the inspection item Am corresponding to the output data Rbn indicating abnormality, it is not necessary to calibrate all the inspection items A1 to Am. Therefore, if the calibration method of the semiconductor manufacturing apparatus according to the present embodiment is used, the first calibration of the semiconductor manufacturing apparatus 100 is performed compared to the case where the first calibration of the semiconductor manufacturing apparatus 100 is performed for all the inspection items A1 to Am. The time required can be shortened.

  Furthermore, since the data D1a obtained from the same semiconductor manufacturing apparatus 100 and the data D1 are compared, it is possible to eliminate the influence of individual differences in the semiconductor manufacturing apparatus. Since the semiconductor manufacturing apparatus is a very precise apparatus, it is an apparatus that is greatly influenced by individual differences. Therefore, if the influence due to individual differences can be eliminated, the first calibration of the semiconductor manufacturing apparatus 100 can be performed in a shorter time.

  In addition, the number of wafers required for the first calibration of the semiconductor manufacturing apparatus 100 can be greatly reduced. Further, by performing the first calibration, the reproducibility of the operation of the semiconductor manufacturing apparatus 100 moved from the location A to the location B can be confirmed.

  The recipe R1 is preferably a recipe for operating the hardware of the semiconductor manufacturing apparatus 100. Using this recipe R1, inspection items A1 to Am for hardware can be inspected. In this case, even if the semiconductor manufacturing apparatus 100 is operated using the recipe R1, the semiconductor manufacturing process is not performed on the wafer introduced into the semiconductor manufacturing apparatus 100. In the recipe R1, an inert gas such as nitrogen gas is supplied into the semiconductor manufacturing apparatus 100 instead of the process gas. Therefore, it is not necessary to consider the influence of the semiconductor manufacturing process on the hardware, so that the hardware inspection becomes easy. Examples of hardware include a transport system and a vacuum system.

  The output data Ra1 to Ran and the output data Rb1 to Rbn are preferably time series data. In this case, the detection sensitivity when detecting an abnormality of the output data Rbn with respect to the output data Ran is improved. For example, when the output data Ran and the output data Rbn are compared, the numerical values can be compared at a number of times. As a result, there is a high possibility that a difference is confirmed between the output data Ran and the output data Rbn.

(Third data collection process)
Next, as shown in FIG. 5, the semiconductor manufacturing apparatus 100 installed at the location B is operated using the recipe R2 (second operating condition). FIG. 5 is a diagram illustrating the semiconductor manufacturing apparatus 100 moved to the location B and the semiconductor manufacturing apparatus 200 (second semiconductor manufacturing apparatus) of the same model as the semiconductor manufacturing apparatus 100.

  In the semiconductor manufacturing apparatus 100, n sensors V1 to Vn (second sensors) are attached. Examples of the sensors V1 to Vn include the same sensors as the sensors S1 to Sn. An example of the semiconductor manufacturing apparatus 200 is the same as that of the semiconductor manufacturing apparatus 100. The semiconductor manufacturing apparatus 200 becomes a reference when performing second calibration of the semiconductor manufacturing apparatus 100 described later. Examples of the semiconductor manufacturing apparatus 200 include an existing semiconductor manufacturing apparatus that is actually operating as a production machine, a semiconductor manufacturing apparatus that is operating as a demo machine, or a laboratory machine.

  When the recipe R2 is used, as shown in FIG. 6B, data D2a (third data) including output data Rd1 to Rdn (third output data) from the sensors V1 to Vn is collected. Using output data Rd1 to Rdn from such sensors V1 to Vn, inspection items B1 to Bm (second inspection items) can be inspected. FIG. 6B is a table showing output data Rd1 to Rdn corresponding to the inspection items B1 to Bm. As inspection items B1-Bm, the thing similar to inspection items A1-Am is mentioned. For example, the output data Rdn includes numerical values dn1 to dnk collected in time series at times t1b to tkb.

  As shown in FIG. 5, the data D2a is recorded from the data logger DL1 to the recording medium DB3. An example of the recording medium DB3 is the same as the recording medium DB1.

(Preparation process of second recording medium)
On the other hand, as shown in FIG. 5, for example, the semiconductor manufacturing apparatus 200 installed at the location B is operated using the recipe R <b> 2. The semiconductor manufacturing apparatus 200 includes, for example, chambers C1a to C4a. In the semiconductor manufacturing apparatus 200, n sensors U1 to Un (third sensors) are attached. Examples of the sensors U1 to Un include the same as the sensors S1 to Sn.

  When the recipe R2 is used, as shown in FIG. 6A, data D2 (fourth data) including output data Rc1 to Rcn (fourth output data) from the sensors U1 to Un is collected. The data D2 may be an average of data collected from a plurality of semiconductor manufacturing apparatuses of the same model as the semiconductor manufacturing apparatus 200. When the sensors U1 to Un are used, the inspection items B1 to Bm can be inspected. FIG. 6A is a table showing output data Rc1 to Rcn corresponding to the inspection items B1 to Bm. For example, the output data Rcn includes numerical values cn1 to cnk collected in time series at times t1c to tkc.

  As shown in FIG. 5, the data D2 is recorded from the data logger DL2 to the recording medium DB4 (second recording medium). An example of the recording medium DB4 is the same as the recording medium DB1.

(Second comparison result generation step)
Next, as shown in FIG. 7, the data D2a and the data D2 are compared to generate a comparison result CR2 (second comparison result). FIG. 7 is a block diagram illustrating a process of generating the comparison result CR2. In the comparison result CR2, for example, output data Rdn indicating an abnormality with respect to the output data Rcn is extracted. More specifically, for example, a case where the difference between the output data Rcn and the output data Rdn is equal to or greater than a predetermined threshold value can be a case where the output data Rdn indicates an abnormality. Further, as a method of displaying the comparison result CR2, the output data Rcn and Rdn may be displayed on the display device.

  FIG. 8 is a graph showing an example of the comparison result CR2 displayed on the display device. In this graph, the vertical axis indicates the flow rate V of oxygen gas, and the horizontal axis indicates time t. The chart (time series data) shown in the graph is obtained by collecting oxygen gas flow rate values measured every second in a time series. Here, the flow rate V of the oxygen gas is an example of the inspection item Bm, and the chart corresponds to an example of the output data Rcn and Rdn.

  In a region c in FIG. 8, a difference Δ2 between the output data Rcn and Rdn is confirmed. In such a case, it is confirmed that the output data Rdn is abnormal with respect to the reference output data Rcn. More specifically, the flow rate of oxygen gas overshoots in the output data Rdn.

  The comparison result CR2 as described above is preferably generated using data mining. When data mining is used, output data Rdn that shows an abnormality with respect to the reference output data Rcn can be efficiently extracted. For this reason, even if the amount of the data D2a and the data D2 is enormous, the comparison result CR2 can be generated in a short time.

(Second calibration process)
Next, the semiconductor manufacturing apparatus 100 is calibrated based on the comparison result CR2. In the comparison result CR2, for example, an abnormality in the output data Rdn (for example, oxygen flow rate data) is confirmed as shown in a region c in FIG. In this case, it is necessary to inspect the inspection item Am (for example, oxygen flow rate) corresponding to the output data Rdn. The semiconductor manufacturing apparatus 100 can be calibrated by repairing, adjusting, or exchanging parts (for example, a flow rate sensor) of the semiconductor manufacturing apparatus 100 related to the inspection item Am. Thereby, the semiconductor manufacturing apparatus 100 is adjusted so that the output data Rdn overlaps the output data Rcn. Here, the calibration of the semiconductor manufacturing apparatus 100 includes repair, adjustment or replacement of parts of the semiconductor manufacturing apparatus 100. When the semiconductor manufacturing apparatus 100 is calibrated, the semiconductor manufacturing apparatus 100 can exhibit predetermined characteristics. The difference between the chambers C1 to C4 can be adjusted based on the comparison result CR2.

  FIGS. 9A and 9B are graphs showing the in-plane distribution of the etching rate over the entire wafer surface. In this graph, the vertical axis represents the etching rate, and the horizontal axis represents the distance from the center of the wafer. FIG. 9A is a graph showing the results when the flow rate of oxygen gas overshoots as indicated by the region c in FIG. FIG. 9B is a graph showing the results when the flow rate of oxygen gas does not overshoot as shown in the region d in FIG. From these graphs, it is confirmed that the in-plane uniformity of the etching rate is improved by performing the second calibration.

  As described above, the second calibration of the semiconductor manufacturing apparatus 100 is performed. In the second calibration, the data D2 and the data D2a are compared, and output data Rdn indicating an abnormality with respect to the output data Rcn is extracted. Therefore, since the semiconductor manufacturing apparatus 100 may be calibrated for the inspection item Bm corresponding to the output data Rdn indicating abnormality, it is not necessary to calibrate all the inspection items B1 to Bm. Therefore, if the semiconductor manufacturing apparatus calibration method according to the present embodiment is used, the second calibration of the semiconductor manufacturing apparatus 100 is performed compared to the case where the second calibration of the semiconductor manufacturing apparatus 100 is performed for all the inspection items B1 to Bm. The time required can be shortened. In addition, the number of wafers required for the second calibration of the semiconductor manufacturing apparatus 100 can be greatly reduced.

  The recipe R2 is preferably a recipe that causes the semiconductor manufacturing apparatuses 100 and 200 to perform a semiconductor manufacturing process. Using this recipe R2, inspection items B1 to Bm for the semiconductor manufacturing process can be inspected. When the semiconductor manufacturing apparatus 100 is operated using the recipe R2, a semiconductor manufacturing process such as etching is performed on the wafer introduced into the semiconductor manufacturing apparatus 100. Examples of the semiconductor manufacturing process include etching, film formation, ion implantation, heat treatment, and the like. By using such a recipe R2, it becomes easy to match the semiconductor manufacturing process performed by the semiconductor manufacturing apparatus 100 with the desired semiconductor manufacturing process performed by the semiconductor manufacturing apparatus 200. For example, the second calibration can set the etching rate, the ashing rate, and the like to desired values in a short time.

  The output data Rc1 to Rcn and the output data Rd1 to Rdn are preferably time series data. In this case, the detection sensitivity when detecting abnormality of the output data Rdn with respect to the output data Rcn is improved. For example, when comparing the output data Rcn and the output data Rdn, the numerical values can be compared at a number of times. As a result, there is a high possibility that a difference is confirmed between the output data Rcn and the output data Rdn.

  When the first calibration and the second calibration are sequentially performed, the reproducibility of the semiconductor manufacturing apparatus can be confirmed in a short time by the first calibration. Next, it is possible to approach the desired semiconductor manufacturing process in a short time by the second calibration. In addition, when the inspection items A1 to Am and the inspection items B1 to Bm are the same, the reproducibility of the collected data can be confirmed.

  Further, when the semiconductor manufacturing apparatus 100 is started up using a conventional method, it takes about two months. On the other hand, if the first calibration and the second calibration are sequentially performed, the time required for startup can be reduced by 60% or more.

  Further, when the semiconductor manufacturing apparatus 100 is started up using a conventional method, the start-up man-hours exceed 500 hours. On the other hand, if the first calibration and the second calibration are performed in order, the startup man-hour can be reduced by 50% or more.

  Further, when the semiconductor manufacturing apparatus 100 is started up using a conventional method, 200 to 300 wafers are required. On the other hand, if the first calibration and the second calibration are sequentially performed, the number of wafers required for start-up can be reduced by about 50%.

  As mentioned above, although preferred embodiment of this invention was described in detail, this invention is not limited to the said embodiment.

  For example, the second recording medium preparation step may be performed before the second comparison result generation step. Therefore, the second recording medium preparation step may be performed before or simultaneously with the third data collection step.

  One of the first and second calibrations may be performed. In either case, the time required for calibration of the semiconductor manufacturing apparatus can be shortened.

  Further, instead of the data D1, data including output data collected using the sensors U1 to Un when the semiconductor manufacturing apparatus 200 is operated using the recipe R1 may be used. This data may be obtained by averaging the data collected from a plurality of semiconductor manufacturing apparatuses of the same model as the semiconductor manufacturing apparatus 100.

  Instead of the data D2, data including output data collected using the sensors V1 to Vn when the semiconductor manufacturing apparatus 100 is operated at the location A using the recipe R2 may be used. In this case, it is possible to eliminate the influence due to individual differences of semiconductor manufacturing apparatuses.

  Further, instead of the data D1, data including output data collected using the sensors U1 to Un when the semiconductor manufacturing apparatus 200 is operated using the recipe R1, and the data D2 is used instead. Thus, when the semiconductor manufacturing apparatus 100 is operated at the location A using the recipe R2, data including output data collected using the sensors V1 to Vn may be used.

FIG. 1A to FIG. 1C are diagrams illustrating a process in which the first semiconductor manufacturing apparatus is moved from the first place to the second place. FIGS. 2A and 2B are tables showing first and second output data corresponding to the first inspection item. It is a block diagram which shows the process of producing | generating a 1st comparison result. It is a graph which shows an example of the 1st comparison result displayed on a display. It is a figure which shows the 1st semiconductor manufacturing apparatus and the 2nd semiconductor manufacturing apparatus which were moved to the 2nd place. FIG. 6A and FIG. 6B are tables showing third and fourth output data corresponding to the second inspection item. It is a block diagram which shows the process of producing | generating a 2nd comparison result. It is a graph which shows an example of the 2nd comparison result displayed on a display. FIGS. 9A and 9B are graphs showing the in-plane distribution of the etching rate over the entire wafer surface.

Explanation of symbols

  A ... 1st place, B ... 2nd place, 100 ... 1st semiconductor manufacturing apparatus, S1-Sn ... 1st sensor, A1-Am ... 1st test | inspection item, R1 ... Recipe (1st operation | movement) Condition), Ra1 to Ran ... first output data, D1 ... first data, DB1 ... first recording medium, Rb1 to Rbn ... second output data, D1a ... second data, CR1 ... first As a result of comparison, V1 to Vn ... second sensor, B1 to Bm ... second inspection item, R2 ... recipe (second operating condition), Rd1 to Rdn ... third output data, D2a ... third data, 200: second semiconductor manufacturing apparatus, U1-Un: third sensor, Rc1-Rcn: fourth output data, D2: fourth data, DB4: second recording medium, CR2: second comparison result .

Claims (4)

A calibration method for a first semiconductor manufacturing apparatus moved from a first location to a second location different from the first location,
The first semiconductor manufacturing apparatus installed at the first location is used for inspecting a plurality of first inspection items corresponding to a plurality of first sensors attached to the first semiconductor manufacturing apparatus. A first recording medium on which first data including a plurality of first output data collected by using the plurality of first sensors when operated under a first operating condition is prepared And a process of
When the first semiconductor manufacturing apparatus installed at the second location is operated under the first operating condition, a plurality of second output data is included using the plurality of first sensors. Collecting second data; and
Comparing the first data with the second data to generate a first comparison result in which the second output data indicating an abnormality with respect to the first output data is extracted;
Performing a first calibration of the first semiconductor manufacturing apparatus based on the first comparison result; and converting the first semiconductor manufacturing apparatus installed at the second location into the first semiconductor manufacturing process. When operating under a second operating condition for inspecting a plurality of second inspection items corresponding to a plurality of second sensors attached to the apparatus, a plurality of the plurality of second sensors are used. Collecting third data including the third output data of:
When the second semiconductor manufacturing apparatus serving as a reference in performing the second calibration of the first semiconductor manufacturing apparatus is operated under the second operating condition, it is attached to the second semiconductor manufacturing apparatus. Preparing a second recording medium on which fourth data including a plurality of fourth output data collected using a plurality of third sensors corresponding to the plurality of second inspection items is recorded; ,
Comparing the third data with the fourth data to generate a second comparison result in which the third output data indicating abnormality with respect to the fourth output data is extracted;
Performing the second calibration of the first semiconductor manufacturing apparatus based on the second comparison result;
Calibration method for semiconductor manufacturing equipment including In the first comparison result to produce a step and the second comparison result generating a calibration method of a semiconductor manufacturing apparatus according to claim 1 using data mining. The first operating condition is an operating condition for operating hardware of the first semiconductor manufacturing apparatus,
Said second operating condition, the calibration method of a semiconductor manufacturing apparatus according to claim 1 or 2 of a semiconductor manufacturing process is an operation condition for implementation in the first semiconductor manufacturing device and the second semiconductor manufacturing device. The semiconductor manufacturing apparatus calibration method according to any one of claims 1 to 3 , wherein the first to fourth output data are time-series data.

Images