JP2001100822A - Maintenance schedule preparing method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a means capable of preparing a maintenance schedule required and sufficient for achieving device characteristics requested for a product, and preparing the optimal maintenance schedule capable of guaranteeing the quality of the product. SOLUTION: A maintenance schedule preparing device 100 is provided with a device characteristic result data storage means 110, predicting means 120, device characteristic predictive data storage means 130, device characteristic managing value storage means 140, product characteristic specification value storage means 150, comparing means 160, maintenance reference value storage means 170, schedule-preparing means 180 and maintenance schedule storage means 190. Concerning this maintenance schedule preparing device 100 (maintenance schedule preparing method), the maintenance schedule which is necessary and sufficient for achieving the device characteristics requested for the product is prepared on the basis of device characteristic predictive data prepared from device characteristic result data, a device characteristic managing value and a product characteristic specification value.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for preparing a maintenance plan for preparing an appropriate maintenance plan for a manufacturing apparatus (and, consequently, a maintenance plan preparing apparatus). The present invention relates to a maintenance plan creation method (and, consequently, a maintenance plan creation device) capable of creating a necessary and sufficient maintenance plan in which unnecessary maintenance is not performed while sufficient maintenance is performed. The present invention is a technology effective for performing maintenance necessary for manufacturing a product according to a standard and preventing the occurrence of defective products, and for reducing a product of a manufacturing apparatus by reducing unnecessary maintenance. This is an effective technique to increase the time that can be manufactured and shorten the manufacturing lead time of the product.

[0002]

2. Description of the Related Art In general, in an apparatus for manufacturing industrial products, maintenance is appropriately performed according to a maintenance plan. FIG. 17 is a block diagram showing an example of a conventional maintenance plan creation device for creating such a maintenance plan. Hereinafter, a conventional maintenance plan creating apparatus or a conventional maintenance plan creating method will be described with reference to FIG.

As shown in FIG. 17, this conventional maintenance plan creating apparatus 1 is essentially composed of an apparatus characteristic management value storage means 10, an apparatus characteristic data storage means 20, a plan creation means 30, and a maintenance plan storage means. And a means 40 for creating a maintenance plan for the manufacturing apparatus.

Hereinafter, a method of creating a maintenance plan for a manufacturing apparatus by the maintenance plan creating apparatus 1 will be described. In this conventional maintenance plan creation method,
First, the maintenance plan creation device 1 is started. Subsequently, the device characteristic management value is obtained from the device characteristic management value storage unit 10. Next, device characteristic data is obtained from the device characteristic data storage unit 20. Thereafter, the plan creation means 30
As a result, the device characteristic prediction data is created from the device characteristic actual data, and the date and time at which the device characteristic prediction data is expected to deviate from the preset device characteristic management value is stored in the maintenance plan storage unit 40 as the maintenance start date and time. You.

[0005] As described above, in the conventional maintenance plan creating apparatus or the conventional maintenance plan creating method, when creating a maintenance plan for a manufacturing apparatus, only the apparatus characteristic prediction data and the preset apparatus characteristic management value are used, and the product is manufactured. The maintenance start date and time of the manufacturing apparatus is set without considering whether the product satisfies the standard.

[0006]

However, such a conventional maintenance plan creation device or maintenance plan creation method cannot create a necessary and sufficient maintenance plan according to the product characteristics, as described below. There is a problem. For example, if a product requires a manufacturing device to have a standard that is stricter than the preset device characteristic management value, the maintenance will be delayed, and the necessary maintenance to manufacture the product according to the standard will be performed reliably. There is a problem that you can not. As a result, the product is manufactured without performing necessary maintenance, and the manufactured product is out of specification, which may cause defective products.

On the other hand, if the preset device characteristic management value is excessively stricter than the product specification, there is a problem that a plan for performing unnecessary or excessive maintenance is created for manufacturing the product. . As a result, there is a problem in that the time during which the manufacturing apparatus can manufacture the product is reduced, and the manufacturing lead time of the product is lengthened.

The present invention has been made in view of the above situation and has been made in order to solve the above-mentioned conventional problems, and creates a maintenance plan necessary and sufficient for a manufacturing apparatus to manufacture a product according to a standard. The problem or object to be solved is to provide a maintenance plan creation method (and, consequently, a maintenance plan creation device) that can prevent the occurrence of defective products and reduce the production lead time. I do.

[0009]

According to the present invention, there is provided a method for preparing a maintenance plan for a manufacturing apparatus, which solves the above-described problems, generates prediction data from actual data of the apparatus characteristic management parameter, and generates a product of the management parameter. It is characterized in that a maintenance schedule of a manufacturing apparatus is created based on characteristic specification values and prediction data. In this maintenance plan creation method, when there is no product characteristic standard value, it is preferable to create a maintenance schedule for the manufacturing apparatus based on the apparatus characteristic standard value and the prediction data. Further, in this maintenance plan creation method, when the actual data of the device characteristic management parameter indicates an irregular data distribution, the prediction data is created using only the data indicating the maximum value or the minimum value in the actual data, When the actual data shows a regular data distribution, it is preferable to create the prediction data using all the actual data.
Further, in this maintenance plan creation method, prediction data is calculated by first-order or higher-order polynomial approximation using time as a variable based on the actual data of the device characteristic management parameters, and the prediction data and the product characteristic specification value or the device characteristic are calculated. It is preferable to set a time at which the standard value matches as the maintenance start time.

[0010] The above-mentioned problems include (a) prediction means for acquiring apparatus characteristic actual data and creating apparatus characteristic prediction data from the apparatus characteristic actual data, and (b) apparatus characteristic management value and product characteristic standard. Comparing means for acquiring the device characteristic management value and the product characteristic standard value, and setting a maintenance reference value based on the comparison result; (c) device characteristic prediction data and maintenance reference value Thus, the present invention is also solved by a maintenance plan creating apparatus, which is provided with a plan creating means for creating a maintenance schedule.

In such a maintenance plan creation device, the device characteristic management value and the product characteristic specification value each include at least one of four statistics of the average value, variance, standard deviation, and total value of the device characteristic data. It is preferable that the setting be made using one of the following. Further, it is more preferable that the device characteristic management value and the product characteristic specification value are set using different statistics for each management item of the device characteristic.

[0012] Further, the above-mentioned problems include (a) a prediction step of creating device characteristic prediction data from device characteristic actual data;
(B) comparing a device characteristic management value with a product characteristic specification value and setting a maintenance reference value based on the comparison result; and (c) determining a maintenance schedule from the device characteristic prediction data and the maintenance reference value. The present invention is also solved by a maintenance plan creation method characterized by comprising a plan creation step of creating.

In the prediction step of such a maintenance plan creation method, (i) when the device characteristic actual data relates to a management item indicating an irregular data distribution, the maximum value in the device characteristic actual data is used. Alternatively, the device characteristic prediction data is created using only the data having the minimum value, and (i
i) When the device characteristic actual data relates to a management item indicating a regular data distribution, it is preferable to create the device characteristic prediction data using all of the device characteristic actual data. In the comparison step of the maintenance plan creation method, (iii) when the product characteristic standard value exists for the combination of each device (unit device) constituting the manufacturing apparatus and the management item, the product characteristic standard value is used. (Iv) When there is no product characteristic standard value for a combination of each device constituting the manufacturing apparatus and the management item, it is preferable to set the device characteristic management value as the maintenance reference value. Further, in the plan creation step of this maintenance plan creation method, (v) device property prediction data is calculated from the device characteristic actual data, and the device property prediction data is converted to a first or higher first-order data using time as a variable. (Vi) the maintenance reference value is converted into a second polynomial using time as a variable, and (vii)
It is preferable to set the solution of the first polynomial and the second polynomial as the maintenance start date and time.

According to the maintenance plan creation method (or maintenance plan creation device) according to the present invention, a necessary and sufficient maintenance plan, that is, a necessary maintenance is surely performed according to a product characteristic, while unnecessary or excessive maintenance is performed. Has a remarkable operation and effect that an appropriate maintenance plan that is not implemented can be created.

Specifically, for example, when a product requires a manufacturing device to have a standard that is stricter than a preset device characteristic management value, the maintenance necessary for manufacturing the product according to the standard is surely performed. A maintenance plan can be created that can be done. As a result, since the product is manufactured without performing necessary maintenance, the product is out of specification, and there is no problem such as the occurrence of poor quality products.

On the other hand, when the preset device characteristic management value is excessively stricter than the product standard, there is no risk of creating a plan for performing unnecessary or excessive maintenance. As a result, problems such as a reduction in the time during which the manufacturing apparatus can manufacture the product and an increase in the manufacturing lead time of the product do not occur.

[0017]

BRIEF DESCRIPTION OF THE DRAWINGS FIG.
A maintenance plan creation device and a maintenance plan creation method according to an embodiment of the present invention will be specifically described. FIG.
As shown in FIG. 1, a maintenance plan creating apparatus 10 for creating a maintenance plan for a semiconductor device manufacturing apparatus.
0 is substantially the device characteristic performance data storage means 110
, Prediction means 120, and device characteristic prediction data storage means 1
30, a device characteristic management value storage unit 140, a product characteristic specification value storage unit 150, a comparison unit 160, a maintenance reference value storage unit 170, a plan creation unit 180, and a maintenance plan storage unit 190. I have. It should be noted that management items for each device (hereinafter, referred to as “unit device”) constituting the semiconductor device manufacturing apparatus are stored in a device management item combination table 600 as shown in FIG. 6, for example. .

In the maintenance plan creating apparatus 100, each of the storage means 110, 130, 140, 150, 1
Reference numerals 70 and 190 denote storage devices (whether freely writeable and readable) of data or information (for example,
Magnetic recording disk device, etc.). Here, the device characteristic result data storage unit 110 stores, for example, device characteristic result data 500 as shown in FIG. The device characteristic prediction data storage means 130 stores, for example, device characteristic prediction data 700 having contents as shown in FIG. For example, FIG.
The device characteristic management value 200 having the contents shown in FIG. For example, FIG.
The product characteristic standard value 300 having the contents shown in FIG. The maintenance reference value storage means 170 stores, for example, a maintenance reference value 400 having the contents as shown in FIG.

In the maintenance plan creating apparatus 100, the predicting means 120, the comparing means 160, and the plan creating means 180 perform mathematical processing on data or information obtained (read) from the corresponding storage means, respectively. It is an electronic arithmetic unit (processor) that can perform various necessary arithmetic processing or calculation processing such as statistical processing.

Hereinafter, a method or procedure for preparing a maintenance plan by the maintenance plan preparing apparatus 100 will be described. As shown in FIG. 12, the maintenance plan creation method includes an apparatus characteristic actual data acquisition step (step S12).
10), a prediction step (step S1220), a device characteristic prediction data storage step (step S1230), a device characteristic management value acquisition step (step S1240), a product characteristic specification value acquisition step (step S1250), and a comparison step (Step S1260), a maintenance reference value storage step (Step S1270), and a plan creation step (Step S1280).

Then, the maintenance plan creation device 100
Roughly performs predetermined calculation processing (calculation processing) on the device characteristic actual data 500 (see FIG. 5) in accordance with the maintenance plan creating method to obtain the device characteristic prediction data 7.
00 (see FIG. 7), while the device characteristic management value 20
0 (see FIG. 2) or product characteristic specification value 300 (see FIG. 3)
Is used as a maintenance reference value 400 (see FIG. 4), and a maintenance start date and time is calculated (calculated) based on the device characteristic prediction data 700 and the maintenance execution reference value 400 to create a maintenance plan. I have.

In this maintenance plan creation method, a general management value for each management item, that is, a device characteristic management value is set for each unit device. When the characteristic value of the management item is worse than this when the characteristic value of each unit device is focused on only the characteristics of each unit device, while ignoring the individual circumstances of each product to be manufactured, This is a general reference value required to perform the maintenance of. In other words, the device characteristic management value is adopted when there is no special standard value relating to the characteristic value of the management item for a specific product to be manufactured, that is, when there is no product characteristic standard value. Here, when the product characteristic standard value is set for the product, the product characteristic standard value is stricter than the device characteristic management value (when the frequency of maintenance should be high), If it can be lower).

Basically, referring to FIGS. 1 and 12, when the product characteristic specification value is stricter than the device characteristic management value, that is, when the product is set to a value smaller than the preset device characteristic management value. In the case where a strict standard is required for a manufacturing apparatus, a maintenance plan creation method (each means and each step) will be described in detail. In this case, the unit device is a device for dry etching 1, the management item is an etching rate,
Maintenance item is maintenance A, product type is LO
The case of T-1 will be described as an example.

In this case, first, in the apparatus characteristic actual data acquisition step (step S1210), the apparatus characteristic actual data 500 (see FIG. 5) is acquired from the apparatus characteristic actual data storage means 110 for the etch rate of the apparatus for dry etching 1. (Read), and then the prediction process (step S1)
At 220), an arithmetic process (calculation process) is performed on the device characteristic actual data 500 to create device characteristic prediction data 700 (see FIG. 7). The device characteristic result data storage means 110 stores a large number of device characteristic result data 50 for each combination of a unit device and a management item in a form as shown in FIG.
0 is stored.

Here, the flow of processing in the apparatus characteristic actual data acquisition step (step S1210) will be described in detail with reference to FIG. As shown in FIG. 13, in this device characteristic actual data acquisition process (step S1210), first in step S1310, a device management item combination table 600 (see FIG. 6) containing possible combinations of unit devices and management items. Is obtained (input). Here, as described above, the case where the unit device is the device for dry etch 1 and the management item is the etch rate is taken as an example, so that the combination of the device for dry etch 1 and the etch rate is acquired. (Unit device = dry etch 1, control item = etch rate).

Next, in step S1320, it is determined whether the obtained management item is a particle. The management items used in this maintenance plan creation method include particles, dimensions, film thicknesses, etch rates, deposition rates, and the like. In these management items, the measurement data shows a regular distribution (there is a regular distribution). ) And those showing an irregular distribution (no regularity). In each of the above management items, a particle is a management item in which measurement data shows an irregular distribution, and all others are management items in which measurement data shows a regular distribution. The method of calculating device characteristic prediction data, which will be described later, differs between a management item indicating an irregular distribution and a management item indicating a regular distribution. Therefore, in step S1320, it is determined whether or not the management item is a particle in order to determine which of the management items is the acquired management item.

The reason why the calculation method differs depending on whether the measured data indicates an irregular distribution or a regular distribution is as follows. That is, in general, in order to accurately generate the device characteristic prediction data, more noise-free device characteristic actual data is required.
When the measurement data has a regular distribution, such as an etch rate, the noise is small. Therefore, by using all the measurement data, extremely accurate device characteristic prediction data can be created. . In this case, since the noise is small, even when only the maximum value (or local maximum value) in the measurement data is used, relatively accurate device characteristic prediction data can be created, and therefore, the device characteristic prediction data can be generated in a short time. Accurate device characteristic prediction data can be obtained.

On the other hand, when the measurement data has an irregular distribution, such as particles, the noise is large, and it is difficult to accurately generate (calculate) the device characteristic prediction data using all the measurement data. It is. In other words, when the noise is large, a large amount of time and resources (computer resources such as a CPU and a memory) are required to generate device characteristic prediction data by performing arithmetic processing on a large number of measurement data. Therefore, in the device characteristic actual data in which the measurement data shows an irregular distribution like particles,
It is preferable to remove the influence of noise using only the maximum value (or local maximum value) in the measurement data. In this way, practical device characteristic prediction data can be obtained in a short time.

For this reason, in this embodiment, the method (creation method) of calculating the device characteristic prediction data from the device characteristic actual data differs depending on whether or not the management item is a particle. In this embodiment, since the management item indicating that the measurement data indicates an irregular distribution is only particles, the management item indicates an irregular distribution depending on whether or not the acquired management item is a particle. Is determined. However, when there is a management item having an irregular distribution other than the particles, it is necessary to add the management item to the determination condition. For example, if there is a management item “A” indicating that the measurement data indicates an irregular distribution other than the particle, in step S1320, the acquired management item is “whether the particle or A (YES)” or “ Whether it is neither particle nor A (N
O) "is determined.

If it is determined in step S1320 that the management item is a particle (YES), in step S1340, the maximum value data (or the maximum value data) is obtained from the device characteristic actual data corresponding to the obtained management item. ) Will be extracted. In this case, in the prediction process (Step S1220) to be executed next, only this maximum value data is used as the device characteristic actual data.
On the other hand, if it is determined in step S1320 that the management item is not a particle (NO), step S1330
Thus, all the device characteristic actual data corresponding to the acquired management items are acquired. In this case, in the prediction process to be executed next (step S1220), all the acquired device characteristic actual data are used. here,
As described above, since the acquired management item is the etch rate of the apparatus for the dry etch 1, step S1320
Is determined to be NO, and after step S1330 is executed, step S1220 is executed.

Next, the flow of processing in the prediction step (step S1220) will be described in detail with reference to FIGS. As shown in FIG. 14, in this prediction step (step S1220), first in step S1410,
The above-described device characteristic actual data acquisition step (step S121)
Based on the device characteristic actual data acquired in step (0), an approximate expression in the form of a polynomial that best fits these data (hereinafter referred to as an “approximate polynomial”) is created, and the coefficients of the approximate polynomial are calculated. You. For example, as shown in FIG. 8, when the approximate polynomial is a linear expression, a gradient A10 and an intercept B10 are calculated. Subsequently, step S1420
It is determined whether the order of the approximate polynomial is less than or equal to the first order. Here, when it is determined that the order of the approximate polynomial is less than or equal to the first order (YES), the coefficient A1 of the approximate polynomial is determined.
0 and B10 are provided to the device characteristic prediction data storage step (step S1230).

On the other hand, in step S1420, the order of the approximate polynomial is larger than the first order (ie, higher than the second order).
Is determined (NO), in step S1430,
The coefficients of the approximation polynomial of second order or higher are provided to the device characteristic prediction data storage step (step S1230), and the order of the approximation polynomial is reduced by one, and then the step S1
At 410, an approximate polynomial is again created. Thereafter, steps S1410 to S1430 are repeatedly executed while decreasing the order by one until the order of the approximate polynomial becomes 1st, and an approximate polynomial is created. In each case, the coefficients of the approximate polynomial are stored in the device characteristic prediction data storage. Process (Step S1)
230). Finally, a first-order approximate polynomial is created, and the coefficients of the first-order approximate polynomial are provided to the device characteristic prediction data storage step (step S1230). Note that not only the coefficients of the first-order approximation polynomial but also 2
The reason why the coefficients of the approximation polynomials of the second or higher order are also provided to the device characteristic prediction data storage step (step S1230) is that it is used when more accurate maintenance is required than ordinary maintenance using a first-order approximation polynomial. It is.

As shown in FIG. 12 again, after the prediction step (step S1220) is executed, a device characteristic prediction data storage step (step S1230) is executed. In the device characteristic prediction data storage step (step S1230), the device characteristic prediction data 700 (see FIG. 7) created in the prediction step (step S1220) is stored in the device characteristic prediction data storage unit 130.

Incidentally, the above-described apparatus characteristic actual data acquisition step (step S1210) and the prediction step (step S1)
220) and a device characteristic prediction data storage step (step S1).
230), the device characteristic management value acquisition step (step S1240), the product characteristic specification value acquisition step (step S1250), and the comparison step (step S1).
260) and a maintenance reference value storing step (step S1).
270) is executed. That is, first, in the device characteristic management value acquisition step (step S1240), the device characteristic management value 200 (see FIG. 2) is read from the device characteristic management value storage unit 140.
Is obtained. The device characteristic management value storage unit 140 stores the device characteristic management value and the maintenance item for each combination of the unit device and the management item in a form as shown in FIG. Here, as described above, since the unit device is the device for dry etching 1 and the management item is the etching rate, the acquired device characteristic management value is 100 n.
m / min.

In the product characteristic specification value storage step (step S1250), the product characteristic specification value 300 (see FIG. 3) is obtained from the product characteristic specification value storage means 150. In addition,
The product characteristic standard value storage means 150 stores the product characteristic standard value required by the product for each combination of the unit device and the management item in the form as shown in FIG. Here, as described above, the product type is LOT-
1, the unit device is a device for dry etching 1,
Since the management item is the etch rate, the acquired product characteristic standard value (required standard value) is 80 nm / min.

Next, a comparison step (step S1260)
Then, the device characteristic management value (= 100 nm / min) acquired in step S1240 is compared with the product characteristic standard value (= 80 nm / min) acquired in step S1250, and then maintenance is performed using the comparison result. A reference value storing step (Step S1270) is performed.

Here, the flow of processing in the comparison step (step S1260) will be described in detail with reference to FIG. As shown in FIG. 15, this comparison step (step S1)
In step 260), first, in step S1510, it is determined whether or not a product characteristic standard value exists for the combination of the product, the unit device, and the management item. here,
If it is determined that a product characteristic standard value exists for this combination (YES), step S1520 is executed, and the product characteristic standard value is provided to the next maintenance reference value storage step (step S1270).
On the other hand, if it is determined in step S1510 that the product characteristic specification value does not exist for this combination, (N
O), Step S1530 is executed, and the device characteristic management value is stored in the maintenance reference value storage step (Step S127).
0).

Here, as described above, since the product characteristic specification value (= 80 nm / min) exists, step S1 is performed.
YES is determined in 510, step S1520 is executed, and the product characteristic specification value (= 80 nm / min) is provided to the maintenance reference value storing step (step S1270) as the maintenance reference value.

As shown in FIG. 12 again, after the comparison step (step S1260) is executed, the maintenance reference value (= 80 nm / min) is stored in the maintenance reference value storage means 170 in the maintenance reference value storage step (step S1270). Is stored. As a result, the maintenance reference value storage unit 170 stores the maintenance reference value for each combination of the unit device and the management item in the form as shown in FIG. Here, the unit device is a device for dry etching 1, and 80 nm / min is stored as a maintenance reference value in a column where the management item is an etching rate.

Thus, steps S1210, S1220 and S1230 are executed, and
And step S1240 and step S1 in parallel with this.
After steps 250, S1260, and S1270 have been executed, a plan creation step (step S1280) is executed. This plan creation step (Step S1280)
Now, the device characteristic prediction data 700 stored in the device characteristic prediction data storage unit 130 and the maintenance reference value 4 stored in the maintenance reference value storage unit 170 will be described.
00 is obtained, and these device characteristic prediction data 700
The maintenance start date and time are calculated based on the maintenance reference value 400 and the maintenance reference value 400.

Here, the flow of processing in the plan creation step (step S1280) will be described in detail with reference to FIG. 10 and FIG. As shown in FIG. 16, in this plan creation step (step S1280), first, in step S1280,
In step 1610, the device characteristic prediction data storage unit 130 is used in the device characteristic prediction data acquisition step (step S1230).
The coefficients A10 and B10 of the approximation polynomial of the device characteristic prediction data corresponding to the management item (etch rate) of the corresponding unit device (dry etch 1) are acquired from the device characteristic prediction data 700 stored in. Next, step S16
At 20, an approximate polynomial of the device characteristic prediction data is created from the coefficients A10 and B10 obtained in step S1610. Here, the case of approximation by a first-order polynomial is described. In FIG. 10, the approximate polynomial of the device characteristic prediction data is represented by a straight line S1.

On the other hand, step S1610 and step S1620 are executed in parallel with step S1620.
At 1630, the management item (etch rate) of the corresponding unit device (dry etch 1) is obtained from the maintenance reference value 400 stored in the maintenance reference value storage unit 170.
Maintenance reference value (= 80 nm / min) corresponding to
Is obtained. Next, in step S1640, the maintenance reference value (= 80n) acquired in step S1630
m / min) is created. In FIG. 10, the approximate polynomial of the maintenance reference value is a straight line L
It is represented by 1.

As described above, after Steps S1610 and S1620 are executed, and Steps S1630 and S1640 are executed in parallel with this,
In step S1650, the solution of the approximate polynomial S1 of the device characteristic prediction data and the polynomial L1 of the maintenance reference value (= 80 nm / min) is calculated. This solution corresponds to the intersection P1 between the straight line S1 and the straight line L1 in FIG. Next, in step S1660, a maintenance start date and time corresponding to the calculated solution is set. The maintenance start date and time is represented by a time T1 corresponding to the intersection P1 in FIG.

In such a conventional maintenance plan creation method, in such a case, the device characteristic management value (= 100 nm / min) for the etch rate is used as the maintenance reference value. In this case, the approximate polynomial of the maintenance reference value (= 100 nm / min) is
In FIG. 10, it is represented by a straight line L2. Therefore, the maintenance start date and time is set to the time T2 corresponding to the intersection Q1 in FIG. 10, and the maintenance is performed considerably later than the time when the maintenance should be performed.

As described above, in the maintenance plan creating method according to this embodiment, when the product characteristic standard value is stricter than the device characteristic management value, that is, when the product is set to a value smaller than the preset device characteristic management value. When a strict standard is required for a manufacturing apparatus, a maintenance execution plan necessary for manufacturing a product according to the standard can be created. As a result, it is possible to reliably perform the required maintenance to manufacture the product, to surely manufacture the product that meets the standard, and to prevent the occurrence of a defective product.

Contrary to the above-mentioned case, when the preset device characteristic management value is excessively (unnecessarily) stricter than the product characteristic standard value, the maintenance plan creation method (each means, each process) ) Will be described in detail. In order to avoid duplication of description, description of items common to the above case will be omitted. Here, the unit device is a device for dry etching 1, the management item is a particle,
Maintenance item is maintenance A, product type is LO
The case of T-2 will be described as an example.

In this case, first, in the device characteristic result data acquisition step (step S1210), the device characteristic result data 500 (see FIG. 5) for the particles of the device for dry etching 1 is acquired from the device characteristic result data storage means 110. (Read), and then the prediction process (step S1
At 220), the device characteristic actual data 500 is subjected to arithmetic processing to create device characteristic prediction data 700 (see FIG. 7).

Here, the flow of processing in the device characteristic actual data acquisition step (step S1210) will be described in detail with reference to FIG. As shown in FIG. 13, in this device characteristic actual data acquisition step (step S1210), first in step S1310, the device management item table 600
One specified combination selected from is obtained (input)
Is done. Here, as described above, the case where the unit device is the device for dry etch 1 and the management item is a particle is taken as an example, so a combination of the device for dry etch 1 and the particles is obtained (unit device). =
Dry etch 1, control item = particle).

Next, in step S1320, it is determined whether the acquired management item is a particle. Here, as described above, since the acquired management item is a particle of the device for the dry etch 1, step S
In step 1320, YES is determined, and step S1340 is performed.
Then, the maximum value data (or the maximum value data) is extracted from the device characteristic actual data corresponding to the acquired management item (particle), and only the maximum value data is extracted in the prediction process (step S1220) to be executed next. Are used as the device characteristic result data.

Next, the flow of processing in the prediction step (step S1220) will be described in detail with reference to FIGS. As shown in FIG. 14, in this prediction step (step S1220), first in step S1410,
The above-described device characteristic actual data acquisition step (step S121)
An approximate polynomial (approximate expression in the form of a polynomial) that best fits these data is created based on the device characteristic result data acquired in step 0), and the coefficients of the approximate polynomial are calculated. As shown in FIG. 9, when the approximate polynomial is a linear expression,
The gradient A20 and the intercept B20 are calculated. Subsequently, in step S1420, it is determined whether or not the degree of the approximate polynomial is less than or equal to the first degree. If it is determined that the degree is less than or equal to the first degree (YES), the coefficients (A20, B20) of the approximate polynomial are determined.
Is the device characteristic prediction data storage step (step S1230)
Provided to

On the other hand, if it is determined in step S1420 that the degree of the approximate polynomial is second or higher (NO),
In step S1430, the coefficients of the second or higher order approximation polynomial are stored in the device characteristic prediction data storing step (step S123).
0), the order of the approximate polynomial is reduced by one, and an approximate polynomial is created again in step S1410. Thereafter, steps S1410 to S14 are performed while decreasing the order by one until the order of the approximate polynomial becomes the first order.
30 are repeatedly executed, and each time the coefficients of the approximate polynomial are provided to the device characteristic prediction data storing step (step S1230). Finally, a first-order approximate polynomial is created, and the coefficients of the first-order approximate polynomial are provided to the device characteristic prediction data storage step (step S1230).

As shown in FIG. 12 again, after the prediction step (step S1220) is executed, the prediction step (122) is performed in the apparatus characteristic prediction data storage step (step S1230).
The device characteristic prediction data 700 (see FIG. 7) created in step (0) is stored in the device characteristic prediction data storage unit 130. Then, steps S1210 and S1 are performed.
In parallel with the execution of 220 and step S1230, the apparatus characteristic management value acquisition step (step S1240), the product characteristic specification value acquisition step (step S1250), the comparison step (step S1260), and the maintenance reference value storage step (step S1260). Step S1270) is performed.

That is, first, in the device characteristic management value acquisition step (step S1240), the device characteristic management value storage means 1
An apparatus characteristic management value 200 (see FIG. 2) is obtained from 40. Here, as described above, since the unit device is the device for dry etching 1 and the management item is a particle, the acquired device characteristic management value is 20 pieces.

In the product characteristic specification value storing step (step S1250), the product characteristic specification value 300 (see FIG. 3) is acquired from the product characteristic specification value storage means 150. Here, as described above, since the product type is LOT-2, the unit device is a device for dry etching 1, and the management item is a particle, the acquired product characteristic standard value (required standard value) ) Is 30 pieces.

Next, a comparison step (step S1260)
Then, the device characteristic management value (= 20) acquired in step S1240 is compared with the product characteristic standard value (= 30) acquired in step S1250, and then the maintenance reference value is obtained using the comparison result. Storage step (Step S
1270) is executed.

Here, the flow of processing in the comparison step (step S1260) will be described in detail with reference to FIG. As shown in FIG. 15, this comparison step (step S1)
In step 260), first, in step S1510, it is determined whether or not a product characteristic standard value exists for the combination of the product, the unit device, and the management item. Here, as described above, since there is a standard value (= 30) of the product characteristics, YES is determined in step S1510,
Step S1520 is executed, and the product characteristic specification value (=
30) are provided to the maintenance reference value storing step (step S1270) as the maintenance reference value.

As shown in FIG. 12 again, after the comparison step (step S1260) is executed, the maintenance reference values (= 30) are stored in the maintenance reference value storage means 170 in the maintenance reference value storage step (step S1270). It is memorized. As a result, the maintenance reference value storage means 1
In the column 70, as shown in FIG. 4, 30 units are stored as the maintenance reference values in the column where the unit device is the device for dry etching 1 and the management item is a particle.

Thus, steps S1210, S1220 and S1230 are executed, and
And step S1240 and step S1 in parallel with this.
After the execution of step 250, step S1260, and step S1270, the plan creation step (step S1280)
The device characteristic prediction data 700 stored in the device characteristic prediction data storage means 130 and the maintenance reference value 40 stored in the maintenance reference value storage means 170
0 is obtained, and the maintenance start date and time are calculated based on the device characteristic prediction data 700 and the maintenance reference value 400.

Here, the flow of processing in the plan creation step (step S1280) will be described in detail with reference to FIG. 11 and FIG. As shown in FIG. 16, in this plan creation step (step S1280), first, in step S1280,
In step 1610, an apparatus characteristic prediction data acquisition step (step S
In step 1230), from the device characteristic prediction data 700 stored in the device characteristic prediction data storage unit 130, the coefficient A20 of the approximate polynomial of the device characteristic prediction data corresponding to the management item (particle) of the manufacturing apparatus (dry etch 1) and B20 is obtained. Next, in step S1620,
Coefficients A20 and B20 obtained in step S1610
, An approximate polynomial of the device characteristic prediction data is created.
Here, the case of approximation by a first-order polynomial is described. In FIG. 11, the approximate polynomial of the device characteristic prediction data is represented by a straight line S2.

On the other hand, step S1610 and step S1
620 is executed in parallel with step S163
At 0, a maintenance reference value (= 30) corresponding to the management item (particle) of the corresponding manufacturing apparatus (dry etch 1) is acquired from the maintenance reference value 400 stored in the maintenance reference value storage unit 170. . Next, in step S1640, an approximate polynomial of the maintenance reference value (= 30) acquired in step S1630 is created. In FIG. 11, the approximate polynomial of the maintenance reference value is represented by a straight line L4.

After steps S1610 and S1620 have been executed in this way and steps S1630 and S1640 have been executed in parallel with this, in step S1650, the approximate polynomial S2 of the device characteristic prediction data and the maintenance The solution of the reference value (= 30) with the polynomial L4 is calculated. This solution corresponds to the intersection Q2 between the straight line S2 and the straight line L4 in FIG. Next, in step S1660, a maintenance start date and time corresponding to the calculated solution is set. This maintenance start date and time is represented by a time T4 corresponding to the intersection Q2 in FIG.

In such a conventional maintenance plan creation method, in such a case, the device characteristic management value (= 20) for particles is used as a maintenance reference value. In this case, the approximate polynomial of the maintenance reference value (= 20) is represented by a straight line L3 in FIG. Therefore, the maintenance start date and time
It will be set at time T3 corresponding to intersection P2,
Unnecessary or excessive maintenance will be performed.

Thus, in the maintenance plan creating method according to the present embodiment, the maintenance that has conventionally been started at time T3 can be extended to time T4.
That is, the time interval (maintenance cycle) from the last (latest) maintenance to the next maintenance can be extended from T3 to T4, and the time during which the manufacturing apparatus can manufacture a product can be increased.

As described above, in the maintenance plan creating method according to the present embodiment, when the preset device characteristic management value is excessively (unnecessarily) stricter than the product characteristic standard value, it is difficult to manufacture a product. A maintenance plan in which unnecessary or useless maintenance is not performed can be created. As a result, the time during which a product can be manufactured by the manufacturing apparatus can be increased, thereby shortening the manufacturing lead time of the product.

As described above, according to the maintenance plan creating apparatus or the maintenance plan creating method according to this embodiment, when creating a maintenance plan for a manufacturing apparatus, a maintenance plan necessary and sufficient for manufacturing a product in accordance with a standard is prepared. It can be created and does not suffer from defects such as the occurrence of defective products or an increase in manufacturing lead time. That is, all the problems to be solved by the present invention are solved.

In this embodiment, the device characteristic management value, the product characteristic standard value, and the standard value for maintenance are assumed to be average values in lot units. A statistic such as a deviation or a total amount may be employed. Further, statistics such as an average value, a variance, a standard deviation, and a total amount for each wafer may be employed. In this embodiment, the approximate polynomial of the device characteristic prediction data is
Although a first-order polynomial is used, a second-order or higher-order polynomial may be used.

Thus, if a maintenance plan creation program for realizing the maintenance plan creation method according to this embodiment is created, and this maintenance plan creation program is recorded on a computer-readable recording medium, this By mounting the recording medium in the auxiliary storage device of the computer, the maintenance plan creation program can be loaded into the main storage device of the computer. Then, when a predetermined event occurs, the function of each procedure is executed by the CPU of the computer.

[Brief description of the drawings]

FIG. 1 is a block diagram of a maintenance plan creation device according to an embodiment of the present invention.

FIG. 2 is a diagram showing storage contents (device characteristic management values and the like) of a device characteristic management value storage means.

FIG. 3 is a diagram showing storage contents (product characteristic standard values and the like) of a product characteristic standard value storage means.

FIG. 4 is a diagram showing storage contents (maintenance reference values and the like) of a maintenance reference value storage means.

FIG. 5 is a diagram showing storage contents (device characteristic result data and the like) of a device characteristic result data storage means.

FIG. 6 is a diagram showing possible combinations of individual devices constituting a manufacturing apparatus and management items.

FIG. 7 is a diagram illustrating storage contents (such as coefficients of an approximate polynomial) of device characteristic prediction data.

FIG. 8 is a diagram showing a polynomial that approximates device characteristic prediction data when the management item is an etch rate.

FIG. 9 is a diagram illustrating a polynomial that approximates device characteristic prediction data when a management item is a particle.

FIG. 10 is a diagram showing a time-series transition of device characteristic prediction data when a management item is an etch rate.

FIG. 11 is a diagram illustrating a time-series transition of device characteristic prediction data when a management item is a particle.

FIG. 12 is a flowchart illustrating a maintenance plan creation method according to the embodiment of the present invention.

FIG. 13 is a flowchart showing specific contents of an apparatus characteristic actual data acquisition step in the flowchart shown in FIG.

FIG. 14 is a flowchart showing specific contents of a prediction step in the flowchart shown in FIG. 12;

FIG. 15 is a flowchart showing specific contents of a comparison step in the flowchart shown in FIG. 12;

FIG. 16 is a flowchart showing specific contents of a plan creation step in the flowchart shown in FIG. 12;

FIG. 17 is a block diagram of a conventional maintenance plan creation device.

[Explanation of symbols]

Reference numeral 100: maintenance plan creation device, 110: device characteristic actual data storage means, 120: prediction device, 130: device characteristic prediction data storage device, 140: device characteristic management value storage device, 150: product characteristic specification value storage device, 160: Comparison means, 170 ... maintenance reference value storage means, 180 ...
Plan creation means, 190: maintenance plan storage means.

 ──────────────────────────────────────────────────続 き Continued on the front page F term (reference) 5B049 AA06 BB07 CC32 CC34 EE01 EE03 5H215 AA06 BB14 BB20 CC07 CC09 CX01 GG09 GG17 5H269 AB26 BB11 EE11 EE25

Claims (4)

[Claims] 1. A method for creating a maintenance plan for a manufacturing apparatus, comprising: creating prediction data from actual data of an apparatus characteristic management parameter; and generating the forecast data based on a product characteristic standard value of the management parameter and the prediction data. A maintenance plan creation method characterized by creating a maintenance schedule. 2. If no product characteristic specification value exists,
2. A maintenance schedule of the manufacturing apparatus is created based on the apparatus characteristic specification value and the prediction data.
The maintenance plan creation method described in. 3. When the performance data of the device characteristic management parameter indicates an irregular data distribution, predictive data is created using only data indicating a maximum value or a minimum value in the performance data, and the performance data is generated. 3. The maintenance plan creation method according to claim 1 or 2, wherein, when indicates a regular data distribution, forecast data is created using all of the actual data. 4. Based on the actual data of the device characteristic management parameter, predictive data is calculated by a first-order or higher-order polynomial approximation using time as a variable, and the predicted data is compared with the product characteristic standard value or the device characteristic standard value. 3. The maintenance plan creation method according to claim 1, wherein a coincident time is set as a maintenance start time.