JP2009160691A - Grinding control system, grinding control program, and manufacturing method of semiconductor device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To carry out grinding of a plurality of types of grinding objects in a simple manner and with high accuracy. <P>SOLUTION: According to a grinding control system, a grinding processing section 2 carries out grinding of the plurality of types of grinding objects, and a film thickness measuring section 3 measures the thickness of each object before and after the grinding. In a process control section 4, a grinding rate ratio calculating section 7 carries out multiple regression analyses, based on an actual grinding rate calculated by a grinding rate calculation section 5 by using the film thickness and a grinding time, the name of the type of the object, the grinding time, a grinding frequency after maintenance, and a maintenance frequency of each grinding object, to thereby calculate a ratio of the grinding rates of the respective types. Thereafter the grinding time calculated by a grinding time calculation section 6 is adjusted according to the type of the subsequent grinding object by using the ratio of the grinding rates, and the grinding processing section 2 carries out subsequent grinding based on an adjusted grinding time. Thus the grinding control system deals with the grinding process control of the plurality of types of grinding objects with high accuracy, based on actually measured data of the plurality of types of grinding objects without relying on complicated model equations or parameters. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a polishing control system, a polishing control program, and a method for manufacturing a semiconductor device, and more particularly to a polishing control system, a polishing control program for controlling a polishing process performed in the manufacturing process of a semiconductor device, and such control. The present invention relates to a method for manufacturing a semiconductor device.

  In recent years, in the field of semiconductor device manufacturing, establishment of a technology capable of dealing with high-mix low-volume production such as SoC (System On Chip) has been promoted. In addition, since the margin on the manufacturing apparatus side with respect to process variations is becoming increasingly narrow as semiconductor devices become finer, an APC (Advanced Process Control) technology that reduces such process variations is introduced into the production line. This is also being promoted. In the production line to which the APC technology is applied, the result of the next process is predicted from the measurement result of the previous process by Run-to-Run control, and the parameter adjustment of the recipe of the next process is automatically performed.

  Incidentally, in a semiconductor device manufacturing process, CMP (Chemical Mechanical Polishing) is widely performed for the purpose of performing planarization after film formation. The general CMP process is controlled by first predicting the next polishing rate using the polishing rate obtained from the polishing amount and polishing time of the polishing actually performed after film formation, and the predicted polishing rate at the next polishing. The polishing time is adjusted so that the desired film thickness can be obtained.

Conventionally, attempts have been made to control such a CMP process using various model equations and parameters (see, for example, Patent Documents 1 to 4).
JP 2005-252036 A JP 2000-252179 A JP 2005-328441 A JP-T-2001-523586

  However, in the CMP process, the polishing rate is likely to fluctuate between polishing processes due to the structure in which the polishing apparatus used for the CMP uses parts such as a polishing pad that is consumed during the process, and the parts can be replaced. Even when the polishing apparatus is maintained, the polishing rate largely fluctuates before and after the maintenance.

  In the control of the CMP process, the model formulas and parameters used for the control are complicated due to the nature of the process, the calculation required for the control is complicated, or the control with sufficient accuracy cannot be performed. There was a problem such as. Further, the existing control method has a problem that it may be difficult to directly apply to control of various types of CMP processes.

  The present invention has been made in view of these points, and an object of the present invention is to provide a polishing control system and a polishing control program capable of easily and accurately controlling a polishing process.

  It is another object of the present invention to provide a method for manufacturing a semiconductor device including a polishing process that is easily and highly accurately controlled.

  In order to solve the above problems, the following polishing control system is provided. That is, the polishing control system includes a polishing processing unit that polishes a film formed on a target to be polished in a predetermined polishing time, a film thickness measuring unit that measures the film thickness of the film before and after polishing, and the predetermined polishing time. And a polishing rate calculation unit that calculates an actual polishing rate using the film thickness, calculates a predicted polishing rate in the next polishing processing unit using the calculated actual polishing rate, and the calculated predicted polishing A polishing time calculation unit that calculates a polishing time in the polishing processing unit using a rate, and a multiple regression analysis using a predetermined number of the measured polishing rates acquired for a plurality of types of polishing objects, and polishing by the type And a polishing rate ratio calculation unit for calculating a polishing rate ratio for each type used for adjusting the time.

  According to such a polishing control system, the polishing rate ratio calculation unit calculates a polishing rate ratio for each type of polishing object using a predetermined number of actually measured polishing rates acquired for a plurality of types of polishing objects. When polishing a plurality of types of polishing objects, complicated model formulas and parameters are unnecessary.

  In order to solve the above problems, a disclosed polishing control program and a semiconductor device manufacturing method are provided.

  According to the disclosed polishing control system, polishing control program, and semiconductor device manufacturing method, polishing of a plurality of types of polishing objects can be controlled with high accuracy. As a result, a high-quality semiconductor device can be manufactured.

Hereinafter, it will be described in detail with reference to the drawings.
First, an outline of a CMP process performed in the manufacturing process of a semiconductor device will be described.

FIG. 1 is a schematic explanatory diagram of a CMP process.
First, in the manufacture of a semiconductor device, a predetermined number of circuit elements such as transistors are formed at predetermined positions on a semiconductor substrate such as silicon (Si), and an insulating film, a wiring, and a via that penetrates the insulating film are formed thereon. Etc. will be formed. The CMP process is performed, for example, when polishing and planarizing the film surface after depositing an insulating film by CVD or the like. Alternatively, in order to embed a conductive material in a wiring groove or via hole formed in the insulating film, the conductive material is first deposited on the entire surface and then polished to the insulating film to form a wiring or via.

  In such a CMP process, step S10 of measuring the film thickness before polishing (film thickness before polishing) of the film to be polished formed by the CVD method or the like, and polishing the film after measuring the film thickness before polishing. Step S11 has a step S12 of measuring the thickness of the film after polishing (film thickness after polishing). An object to be polished such as a wafer on which a film to be polished is formed is sent to these steps S10, S11, and S12 in order, and predetermined processing is performed in each of steps S10, S11, and S12. The object to be polished after the post-polishing film thickness measurement is sent to the next process. Alternatively, further polishing is performed as necessary.

  When polishing a plurality of types (variety) of polishing objects such as a plurality of types of wafers for forming different semiconductor devices, the pre-polishing film thickness obtained in step S10 and the pre-polishing film thickness were obtained. Using the target film thickness after polishing (target film thickness after polishing) specific to the type of polishing target, the predicted polishing rate (predicted polishing rate), and the polishing rate ratio specific to that type, the following equation (1) Is used to calculate the polishing time.

Polishing time = (film thickness before polishing−target film thickness after polishing) / (predicted polishing rate × polishing rate ratio) (1)
The polishing rate ratio used for the calculation of the polishing time includes a specific product (reference product) with respect to the polishing rate of one product (reference product) of a plurality of products in consideration of a polishing rate difference that may exist between different products. ) Polishing rate ratio. Further, the predicted polishing rate of the formula (1) is the predicted polishing rate of the standard product or the predicted polishing rate converted to the standard product converted into the predicted polishing rate of the standard product. According to the equation (1), the polishing time of the object to be polished is adjusted to a polishing time suitable for the type of the polishing time for the standard type.

The calculated polishing time is sent to Step S11, and in Step S11, polishing is performed with the sent polishing time.
After the measurement of the post-polishing film thickness in step S12, using the post-polishing film thickness acquired in the measurement, the pre-polishing film thickness acquired in step S10, and the polishing time, A polishing rate (measured polishing rate) of the polished object is calculated.

Actual polishing rate = (film thickness before polishing−film thickness after polishing) / polishing time (2)
Further, the calculated actual polishing rate is converted according to the following equation (3) using the polishing rate ratio of the polished product type.

Converted actual polishing rate = actual polishing rate of specific product / ratio of polishing rate of specific product (3)
By this formula (3), the actually measured polishing rate of the polishing target actually polished is converted into the polishing rate when the polishing target is the standard type. Then, as shown in the following equation (4), an index shift using the calculated conversion actual polishing rate and the standard product used to obtain the calculated polishing rate or the predicted polishing rate converted to the standard product (previous polishing rate) The predicted polishing rate for the next polishing is calculated based on the average (coefficient γ is a smoothing constant).

Predicted polishing rate = converted actual measured polishing rate × γ + previous predicted polishing rate × (1−γ) (4)
The coefficient γ is usually set to about 0.2 to 0.3.

  The next predicted polishing rate calculated is a predicted polishing rate when the next polishing is assumed to be a standard product, and is a predicted polishing rate for calculating the polishing time of the next polishing by the equation (1). In the next polishing, the pre-polishing film thickness to be polished next time, the post-polishing target film thickness and the polishing rate ratio specific to the product type, and the polishing time suitable for the product type are calculated by equation (1). Will be.

Here, the transition of the measured polishing rate obtained by a plurality of polishings and the polishing rate ratio used for the control of the CMP process will be described.
FIG. 2 is a conceptual diagram of the transition of the measured polishing rate.

  The measured polishing rate of polishing performed continuously a plurality of times gradually decreases as the number of polishing increases as shown by the trend line 100a in FIG. 2 due to pad wear due to the structure of the polishing apparatus. There is a tendency to change. The decreased polishing rate is recovered by performing maintenance such as pad replacement as shown by the trend line 100b in FIG. 2, and thereafter gradually decreases as the number of polishing increases again.

  In the CMP process in which the actually measured polishing rate tends to change as described above, during the plural times of polishing up to the maintenance, all of the same type may be polished, and different types may be polished. In that case, even if the polishing is performed under the same conditions, the polishing rate is not necessarily the same when the types are different.

  In the CMP process, it should be noted that, in addition to the state of the polishing apparatus to be used, a difference in polishing rate existing between different varieties may affect the control (adjustment of polishing time).

  FIG. 3 is an explanatory diagram of polishing of a plurality of types of polishing targets. FIG. 3A is a diagram illustrating an example of transition of an actual polishing rate when polishing is performed on a plurality of types of polishing targets, and FIG. ) Shows the measured polishing rate and its simple average for each type. However, FIG. 3 shows the measured polishing rate when polishing is performed without considering the polishing rate ratio (the polishing rate ratio is 1 for all types).

  As shown in FIG. 3A, for example, when four different varieties A, B, C, and D are continuously polished in a predetermined order, as described above, overall polishing is performed. There is a tendency for the measured polishing rate to decrease as the number of times increases. Looking at each of the types A, B, C, and D, as shown in FIG. 3B, the simple averages A1, B1, C1, and D1 of the measured polishing rates are different for the types A, B, C, and D. .

  In the example of FIG. 3, based on the type A having the highest number of polishing times, the polishing rate ratio of the standard type A is 1, and the ratio of the simple average of each of the other types B, C, D to the simple average of the standard type A is shown. The polishing rate ratio of each product type B, C, D is used. The polishing time of step S10 shown in FIG. 1 is calculated using the polishing rate ratio of each of these varieties A, B, C, and D, and using the standard varieties (variety A) or the predicted polishing rate converted to the reference varieties. By doing so, it is possible to obtain a polishing time in consideration of the polishing rate difference among the varieties A, B, C, and D when viewed as a simple average.

  However, in the method using the polishing rate ratio peculiar to the product obtained by such a simple average, the polishing rate ratio is set without considering the transition of the actual polishing rate as shown in FIG. Seeking. For example, the type D having a small simple average of the measured polishing rate may have a small polishing rate in nature, and the measured polishing rate is reduced because of the large number of polishings performed before the polishing. There is a possibility.

Therefore, next, a method for obtaining the polishing rate ratio by multiple regression analysis using the polishing result will be described.
In the CMP process, first, the polishing rate ratio is set to 1 for all types of polishing objects without considering the polishing rate ratio, and the film thickness measurement before polishing in step S10 shown in FIG. And the post-polishing film thickness in step S12 are measured. As a result, the data of the polishing result when the polishing rate ratio is not taken into consideration (such as the name of the product to be polished, the polishing time, the number of polishings after maintenance, the number of maintenances, the measured polishing rate, etc.) is acquired. At that time, each data is acquired in association with each polishing object. Then, when a predetermined number of polishing result data is obtained in total for a plurality of types of polishing objects, a multiple regression analysis is performed as follows.

  In the multiple regression analysis, the product name, polishing time, the number of polishings after maintenance and the number of maintenances are used as explanatory variables (the product name and the number of maintenances are nominal scales, the polishing time and the number of polishings after maintenance are continuous scales), and the measured polishing rate. Is the objective variable. If such a multiple regression analysis is performed, the polishing rate is expressed by the following equation (5).

Actual polishing rate = a (product specific) + b × polishing time + c × number of polishing after maintenance + d (maintenance specific) + e (5)
In this equation, a is a value specific to the product type. On the other hand, b and c are coefficients of the polishing time and the number of polishing times after maintenance that are not unique to the product type. d is a value unique to maintenance. The coefficient e is an intercept.

  In the case of multiple regression analysis of polishing result data having the same maintenance frequency, the measured polishing rate of each product type is regarded as the value of e added to the value of a unique to each product type. Then, the polishing rate ratio of each type is obtained. At that time, one of the plurality of varieties is set as the reference varieties, and the ratio of the measured polishing rate of the specific varieties to the measured polishing rate of the reference varieties is obtained. The polishing rate ratio of each product is calculated by the following equation (6) using a and e.

Polishing rate ratio = (specific product specific a + e) / (reference product specific a + e) (6)
As a result, the polishing rate ratio of the other types is calculated with respect to the polishing rate ratio 1 of the standard type.

  Further, in order to evaluate the trend of the actual polishing rate in consideration of such a polishing rate ratio, the actual polishing rate of each product and the polishing rate ratio of each product are used, and the trend polishing rate is expressed by the following equation (7). Is calculated.

Trend polishing rate = Measured polishing rate for a specific product / Polishing rate ratio for a specific product (7)
After obtaining the polishing rate ratio of each product type, the CMP process is controlled by the flow shown in FIG. 1 using the polishing rate ratio of each product type.

The calculation of the polishing rate ratio and the control of the CMP process using the determined polishing rate ratio can be performed using, for example, a polishing control system as shown in FIG.
FIG. 4 is a diagram showing a configuration example of the polishing control system.

  A polishing control system 1 shown in FIG. 4 includes a polishing processing unit 2 that polishes a film to be polished formed on a polishing target, a film thickness measuring unit 3 that measures the film thickness of the film before and after polishing, and a polishing process. A process control unit 4 for controlling the polishing time in the unit 2 is provided.

  The object to be polished is measured for the film thickness before polishing by the film thickness measuring unit 3 before polishing, and then sent to the polishing processing unit 2. In the polishing processing unit 2, polishing is performed on the object to be polished in a predetermined polishing time adjusted by the process control unit 4. The polished object after polishing is returned to the film thickness measuring unit 3 again, and the film thickness after polishing of the film is measured.

  The polishing processing unit 2 includes a polishing unit 2a having a mechanism for performing polishing, and a control unit 2b for controlling the polishing operation of the polishing unit 2a. Further, the polishing processing unit 2 includes a memory unit 2c, and stores polishing result data (the name of the product, the polishing time, the number of polishings after maintenance, the number of maintenances, etc.) acquired by polishing. It has become.

  The film thickness measurement unit 3 includes a measurement unit 3a provided with a mechanism for measuring the film thickness, and a control unit 3b that performs measurement operation control of the measurement unit 3a. Furthermore, the film thickness measuring unit 3 includes a memory unit 3c, and stores data on the film thickness measured before and after polishing.

  For example, the polishing processing unit 2 is configured by a CMP apparatus and a computer that cooperates with the CMP apparatus, and the film thickness measuring unit 3 is configured by a film thickness measuring apparatus that can be extended by using a laser or the like and a computer that cooperates with the film thickness measuring apparatus. The

The process control unit 4 of the polishing control system 1 includes a polishing rate calculation unit 5, a polishing time calculation unit 6, a polishing rate ratio calculation unit 7, and a data storage unit 8.
The polishing rate calculation unit 5 calculates the actual polishing rate of the polishing target polished by the polishing processing unit 2, the converted actual polishing rate obtained by converting the polishing rate into the polishing rate of the standard product, and the predicted polishing rate of the next polishing. . The measured polishing rate, the converted measured polishing rate, and the predicted polishing rate are calculated using the equations (2), (3), and (4), respectively. That is, the polishing rate calculation unit 5 performs actual polishing using the polishing result data acquired by the polishing processing unit 2 and the pre-polishing film thickness and post-polishing film thickness acquired by the film thickness measurement unit 3 according to the equation (2). Calculate the rate. Further, the polishing rate calculation unit 5 calculates the converted actual polishing rate by the equation (3), and acquires the converted actual polishing rate and the previous predicted polishing rate (the converted actual polishing rate by the equation (4)). The predicted polishing rate is calculated using the predicted polishing rate used. Further, the polishing rate calculation unit 5 calculates a trend polishing rate by the equation (7) using the calculated actual polishing rate and the polishing rate ratios of a plurality of types.

  The polishing time calculation unit 6 calculates the polishing time of the next polishing using the predicted polishing rate calculated by the polishing rate calculation unit 5 and the polishing rate ratios of a plurality of types. The polishing time of the next polishing is set for each predicted polishing rate of the next polishing, the polishing rate ratio of the type that performs the next polishing, the pre-polishing film thickness acquired by the film thickness measuring unit 3, and each type. Using the target film thickness after polishing, it is calculated by equation (1). The data of the polishing time calculated by the polishing time calculation unit 6 is transmitted to the polishing processing unit 2, and the polishing processing unit 2 performs polishing on the next polishing target using the received data of the polishing time. Become.

  The polishing rate ratio calculation unit 7 calculates the polishing rate ratio of each type using a predetermined number of polishing result data continuously acquired by the polishing processing unit 2 for a plurality of types of polishing targets. The polishing rate ratio calculation unit 7 executes a multiple regression analysis process using polishing result data of a plurality of types of polishing objects acquired without considering the polishing rate ratio, and each of the equations (5) and (6) Calculate the polishing rate ratio of the product type. The number of polishing result data used for calculating the polishing rate ratio can be set in advance from outside the polishing control system 1.

  The data storage unit 8 includes first and second memory units 8a and 8b. The first memory unit 8a stores the measured polishing rate and the predicted polishing rate calculated by the polishing rate calculation unit 5. When calculating the predicted polishing rate of the next polishing, the polishing rate calculation unit 5 uses the predicted polishing rate of the previous time (one time before the next polishing) stored in the first memory unit 8a to predict the next polishing. The polishing rate is calculated. Further, the polishing rate calculation unit 5 calculates a trend polishing rate using the actually measured polishing rate stored in the first memory unit 8a and the polishing rate ratio of each product stored in the second memory unit 8b. . The second memory unit 8b stores the polishing rate ratio of each product type calculated by the polishing rate ratio calculation unit 7. The polishing time calculation unit 6 calculates the polishing time using the previous predicted polishing rate stored in the first memory unit 8a and the polishing rate ratio of each product stored in the second memory unit 8b. .

The processing function of the process control unit 4 can be realized by a computer.
The control of the CMP process using the polishing control system 1 having the configuration as shown in FIG. 4 is performed in the following flow, for example.

  First, in order to predict an appropriate polishing time, a polishing rate ratio is obtained. At that time, with respect to a plurality of types of polishing objects, the polishing rate ratio is set to 1 for all types and the data is acquired according to the flow shown in FIG. Perform multiple regression analysis to determine the polishing rate ratio for each product type.

  In the first polishing, the film thickness measurement unit 3 measures the film thickness before polishing of a specific type of polishing target, and the film thickness before polishing is stored in the memory unit 3c. Further, the polishing time calculation unit 6 sets the polishing rate ratio of the formula (1) to 1, and appropriately sets the predicted polishing rate (for example, the polishing rate predicted from previous results or other types), and further. The polishing time is calculated using the target film thickness after polishing of the object to be polished. The calculated polishing time is transmitted to the polishing processing unit 2 (control unit 2b).

  Thereafter, the object to be polished is sent to the polishing processing unit 2 and polished at the polishing time calculated by the polishing time calculating unit 6, and the polishing result data (product name, polishing time, number of polishing after maintenance, maintenance) The number of times is acquired and stored in the memory unit 2c. The polished object after polishing is sent to the film thickness measuring unit 3 to measure the film thickness after polishing of the film, and the film thickness after polishing is stored in the memory unit 3c.

  The polishing result data of the memory unit 2c acquired by the polishing processing unit 2 and the data of the film thickness before polishing and the film thickness after polishing of the memory unit 3c acquired by the film thickness measuring unit 3 are transmitted to the polishing rate calculation unit 5. The polishing rate calculation unit 5 calculates the actual polishing rate of the object to be polished by equation (2). The calculated actual polishing rate is stored in the first memory unit 8a of the data storage unit 8. For example, this measured polishing rate is set as the predicted polishing rate of the second polishing. In addition, using the predicted polishing rate set when calculating the polishing time, the predicted polishing rate of the second polishing is calculated according to formulas (3) and (4) (polishing rate ratio 1, converted actual polishing rate = actual polishing rate). You can also

  In the subsequent second polishing, the film thickness measurement unit 3 measures the film thickness before polishing of a specific type (the same or different type as the first polishing), and the film thickness before polishing is stored in the memory. Stored in the unit 3c. Further, the polishing time calculation unit 6 sets the polishing rate ratio of the formula (1) to 1, and calculates the predicted polishing rate of the first memory unit 8a calculated by the first polishing and the target thickness after polishing of the polishing target. Using this, the polishing time is calculated. The calculated polishing time is transmitted to the polishing processing unit 2.

  Thereafter, like the first polishing, the polishing target is polished at the polishing time in the polishing processing unit 2, and the acquired polishing result data is stored in the memory unit 2c. The polished object after polishing is sent to the film thickness measuring unit 3, the film thickness after polishing of the film is measured, and the film thickness after polishing is stored in the memory unit 3c.

  The polishing result data acquired by the polishing processing unit 2, the pre-polishing film thickness data and the post-polishing film thickness data acquired by the film thickness measurement unit 3 are transmitted to the polishing rate calculation unit 5, and the polishing rate calculation unit 5 The actual polishing rate is calculated, and the predicted polishing rate for the third polishing is calculated using the predicted polishing rate for the second polishing in the first memory unit 8a under the condition of the polishing rate ratio of 1. The actually measured polishing rate after calculation and the predicted polishing rate of the third polishing are stored in the first memory unit 8a.

The third and subsequent polishing is performed in the same manner.
Then, when a predetermined number of polishing result data (product name, polishing time, measured polishing rate, number of polishings after maintenance, number of maintenances, etc.) are obtained for a plurality of types of polishing objects, the polishing results are obtained. Using the data, the polishing rate ratio calculation unit 7 calculates the polishing rate ratio of each product type. The polishing rate ratio calculation unit 7 executes a multiple regression analysis process using a predetermined number of polishing result data acquired without considering the polishing rate ratio of each product type, and uses Equations (5) and (6) for each product type. Determine the polishing rate ratio. The obtained polishing rate ratio of each product is stored in the second memory unit 8b of the data storage unit 8. Further, the polishing rate calculation unit 5 calculates the trend polishing rate using the measured polishing rate stored in the first memory unit 8a and the polishing rate ratio of each type stored in the second memory unit 8b. .

  After determining the polishing rate ratio of each product in this way, the subsequent CMP process is controlled using the polishing rate ratio. In that case, it is only necessary to perform data acquisition and process control according to the flow shown in FIG. 1 using the polishing rate ratio of each product type.

  That is, in the first polishing after acquiring the polishing rate ratio, first, the film thickness measuring unit 3 measures the film thickness before polishing of the film and stores it in the memory unit 3c for a specific type of polishing target. Next, the polishing time is calculated by the polishing time calculation unit 6 using the polishing rate ratio of the product to be polished, using the predetermined predicted polishing rate and the target film thickness after polishing of the product. The calculated polishing time is transmitted to the polishing processing unit 2.

  Thereafter, the object to be polished is sent to the polishing processing unit 2 and polished at the polishing time calculated by the polishing time calculating unit 6, and the polishing result data obtained by the polishing (product name, polishing time, polishing after maintenance) The number of times, the number of maintenance times, etc.) are stored in the memory unit 2c. The object to be polished after polishing is sent to the film thickness measuring unit 3, and the film thickness after polishing of the film is measured and stored in the memory unit 3c.

  The polishing result data in the memory unit 2 c and the data on the film thickness before polishing and the film thickness after polishing in the memory unit 3 c are transmitted to the polishing rate calculation unit 5. The polishing rate calculation unit 5 calculates the actual polishing rate of the object to be polished according to the equation (2). For example, this measured polishing rate is stored in the first memory unit 8a as the predicted polishing rate for the second polishing.

  In the subsequent second polishing, the film thickness measurement unit 3 measures the film thickness before polishing of the specific type (same or different type as the first polishing) and stores it in the memory unit 3c. . Next, the polishing time calculation unit 6 stores the pre-polishing film thickness stored in the memory unit 3c, the predicted polishing rate of the second polishing stored in the first memory unit 8a, and the second memory unit 8b. The polishing time is calculated by using the polishing rate ratio of the type to be polished and the target film thickness after polishing of the type. The calculated polishing time is transmitted to the polishing processing unit 2.

  Thereafter, as in the first polishing, the object to be polished is polished at the polishing time in the polishing unit 2, and acquired polishing result data (product name, polishing time, number of polishing after maintenance, number of maintenance, etc.). ) Is stored in the memory unit 2c. The polished object after polishing is sent to the film thickness measuring unit 3, the film thickness after polishing of the film is measured, and the film thickness after polishing is stored in the memory unit 3c.

  The polishing result data in the memory unit 2 c and the data on the film thickness before polishing and the film thickness after polishing in the memory unit 3 c are transmitted to the polishing rate calculation unit 5. The polishing rate calculation unit 5 calculates the actual polishing rate of the object to be polished according to the equation (2). The calculated actual polishing rate is stored in the first memory unit 8a. Further, in the polishing rate calculation unit 5, the measured polishing rate stored in the first memory unit 8a, the predicted polishing rate of the second polishing, and the polishing of the type to be polished stored in the second memory unit 8b. Using the rate ratio, the predicted polishing rate of the third polishing is calculated by the equations (3) and (4).

The third and subsequent polishing is performed in the same manner.
Hereinafter, a specific example will be described.
FIG. 5 is a diagram showing an example of the transition of the measured polishing rate.

  FIG. 5 shows the above-described figure without considering the polishing rate ratio (wafer rate ratio 1) for each of the polishing objects (wafers) of varieties A, B, C, and D on which four types of semiconductor devices are respectively formed. 1 shows the actual polishing rate of each time when polishing was continuously performed according to the flow as shown in FIG.

  Multiple regression analysis was performed using the polishing result data (variety name, polishing time, number of polishings after maintenance, number of maintenances, etc.) acquired together with these 15 actual polishing rates. The polishing rate can be expressed as follows according to the equation (5).

Polishing rate = a + b × polishing time + c × number of polishing after maintenance + d + e
The values (estimated values) of the coefficients a, b, c, and d obtained by executing the multiple regression analysis are as shown in Table 1 below.

  Here, the cultivar A having the largest number of times of polishing (the number of n) was set as the reference cultivar, and the polishing rate ratio of each cultivar was calculated according to the equation (6). The polishing rate ratio of the standard product A is 1. The polishing rate ratio of product type B is as follows.

(Variety B a + e) / (reference variety A a + e) = (0.1569 + 10.0649) / (− 0.0448 + 10.0649) = 1.020
Similar calculations were made for other varieties C and D. The polishing rate ratios for the varieties A, B, C, and D are as shown in Table 2 below. Table 2 also shows a simple average of the measured polishing rates of the various types.

  According to Table 2, the polishing rate ratios for each of the varieties B, C, and D are different between the cases obtained by the multiple regression analysis and the case obtained by the simple average. It was very different.

The trend polishing rate was calculated according to the equation (7) using the polishing rate ratio thus determined.
FIG. 6 is a diagram showing the transition of the trend polishing rate.

  In FIG. 6, the measured polishing rate obtained as shown in FIG. 5 is obtained by dividing by the polishing rate ratio of the corresponding product obtained by multiple regression analysis, and a simple average is obtained. The trend polishing rate calculated by dividing by the polishing rate ratio of the corresponding product is shown.

  From FIG. 6, it can be seen that the trend polishing rate changes more smoothly when the polishing rate ratio obtained by multiple regression analysis is used than when the polishing rate ratio obtained by simple average is used. . In particular, for the type D, the difference in trend polishing rate is large in both cases.

  In this way, by using multiple regression analysis to calculate the polishing rate ratio, it is possible to properly grasp the transition of the polishing rate, and it is possible to more accurately predict the polishing time, and the film thickness after polishing It becomes possible to reduce the variation of. After calculating the polishing rate ratio using the above method, the CMP process was carried out according to the flow shown in FIG. 1 above. As a result, the process capability was about 1.3 times the deviation from the target film thickness after polishing. It was confirmed to improve. As a result, manufacturing variations in the semiconductor device manufacturing process after polishing could be reduced.

  Note that the number of n to be used for calculating the polishing rate ratios of a plurality of types is not limited to the above example (15 times). As the number n increases, the calculation load increases, but a more appropriate polishing rate ratio can be calculated. This n number is preset. Moreover, the polishing rate ratio calculation of a plurality of types may be performed every time the polishing using the polishing rate ratio is performed a predetermined number of times.

  As described above, the CMP process can be controlled with high accuracy by using the above-described method of calculating the polishing rate ratio of each product type using a measured number of polishing rates acquired for a plurality of product types. High-quality semiconductor devices can be manufactured. The above method does not require parameters indicating the unevenness of the film to be polished and the state of the polishing apparatus, model formulas using such parameters, parameter optimization based on experiments and simulations, etc. By using the necessary and sufficient data, it is possible to cope with high-precision CMP process control of a plurality of types.

  In the above description, the method of calculating the polishing rate ratio by multiple regression analysis has been described. However, the same effect can be obtained by calculating the polishing rate ratio using the following method. .

  That is, a ratio (actual polishing rate / predicted polishing rate) between each measured polishing rate as shown in FIG. 5 and each predicted polishing rate calculated immediately before each acquisition is calculated, and these are calculated as varieties A, B, By averaging for each C and D, the polishing rate ratio of each of the varieties A, B, C and D is calculated. Also by such a method, it is possible to appropriately calculate the polishing rate ratio of each of the varieties A, B, C, and D, thereby enabling accurate prediction of the polishing time.

  Even when such a method is used, a system similar to the polishing control system shown in FIG. 4 can be used. In the polishing rate ratio calculation unit 7, necessary measured polishing rate and predicted polishing rate are set. It is sufficient to calculate the ratio between them, and to perform an averaging process for each of the varieties A, B, C, and D.

  The processing functions of the polishing control system 1 as described above can be realized using a computer. In that case, a program describing the processing contents of the functions that the polishing control system 1 should have is provided. By executing the program on the computer, a predetermined processing function is realized on the computer. The program describing the processing contents can be recorded in a computer-readable recording medium such as a magnetic recording device, an optical disk, a magneto-optical recording medium, a semiconductor memory, or the like.

Regarding the embodiment described above, the following additional notes are further disclosed.
(Supplementary note 1) In a polishing control system for controlling a polishing process,
A polishing processing unit for polishing a film formed on the object to be polished in a predetermined polishing time;
A film thickness measuring unit for measuring the film thickness of the film before and after polishing;
A polishing rate calculation unit that calculates an actual polishing rate using the predetermined polishing time and the film thickness, and calculates a predicted polishing rate in the next polishing processing unit using the calculated actual polishing rate;
A polishing time calculation unit that calculates a polishing time in the polishing processing unit using the calculated predicted polishing rate;
A polishing rate ratio calculation unit for calculating a polishing rate ratio for each type used for adjusting a polishing time according to the type by multiple regression analysis using a predetermined number of the measured polishing rates acquired for a plurality of types of polishing targets. When,
A polishing control system comprising:

(Supplementary Note 2) After calculation of the polishing rate ratio by the polishing rate ratio calculation unit,
The polishing rate calculation unit calculates the predicted polishing rate using the polishing rate ratio corresponding to the type of the polishing target for which the measured polishing rate is calculated,
The polishing time calculation unit adjusts the calculated polishing time using the calculated predicted polishing rate and the polishing rate ratio corresponding to the type of the next polishing target,
The polishing control system according to supplementary note 1, wherein:

  (Supplementary Note 3) The polishing rate ratio calculation unit may perform the multiple regression analysis using the predetermined polishing time and the type each time the predetermined number of actual polishing rates are acquired together with the predetermined number of actual polishing rates. The polishing control system according to appendix 1 or 2, wherein:

(Supplementary Note 4) In a polishing control system for controlling a polishing process,
A polishing processing unit for polishing a film formed on the object to be polished in a predetermined polishing time;
A film thickness measuring unit for measuring the film thickness of the film before and after polishing;
A polishing rate calculation unit that calculates an actual polishing rate using the predetermined polishing time and the film thickness, and calculates a predicted polishing rate in the next polishing processing unit using the calculated actual polishing rate;
A polishing time calculation unit that calculates a polishing time in the polishing processing unit using the calculated predicted polishing rate;
By averaging each ratio of the predetermined number of the measured polishing rates acquired for a plurality of types of polishing targets with respect to the predicted polishing rate calculated immediately before each acquisition, the polishing time according to the types A polishing rate ratio calculating unit for calculating a polishing rate ratio for each type used for the adjustment of
A polishing control system comprising:

(Supplementary Note 5) In the polishing control program used for controlling the polishing process,
Computer
A polishing rate calculation unit that calculates a measured polishing rate using a film thickness before and after polishing of the film and a predetermined polishing time of the film, and calculates a next predicted polishing rate using the calculated measured polishing rate,
A polishing time calculation unit that calculates a polishing time using the calculated predicted polishing rate,
A polishing rate ratio calculation unit for calculating a polishing rate ratio for each type used for adjusting a polishing time according to the type by multiple regression analysis using a predetermined number of the measured polishing rates acquired for a plurality of types of polishing targets. ,
A polishing control program characterized by functioning as:

(Supplementary Note 6) After calculation of the polishing rate ratio by the polishing rate ratio calculation unit,
The polishing rate calculation unit calculates the predicted polishing rate using the polishing rate ratio corresponding to the type of the polishing target for which the measured polishing rate is calculated,
The polishing time calculation unit adjusts the calculated polishing time using the calculated predicted polishing rate and the polishing rate ratio corresponding to the type of the next polishing target,
The polishing control program according to supplementary note 5, wherein

  (Supplementary Note 7) The polishing rate ratio calculation unit may perform the multiple regression analysis using the predetermined polishing time and the type each time the predetermined number of actual polishing rates are acquired together with the predetermined number of actual polishing rates. The polishing control program according to appendix 5 or 6, characterized in that:

(Additional remark 8) In the manufacturing method of the semiconductor device which has a polishing process,
Forming a transistor on a semiconductor substrate;
Forming a film on an upper layer of the semiconductor substrate on which the transistor is formed;
Measuring the film thickness of the formed film;
Polishing the film at a predetermined polishing time;
Measuring the film thickness of the film after polishing;
Calculating an actual polishing rate using the film thickness before and after polishing and the predetermined polishing time, and calculating a next predicted polishing rate using the calculated actual polishing rate;
Calculating a polishing time using the calculated predicted polishing rate;
A step of calculating a polishing rate ratio for each type used for adjustment of a polishing time according to the type by multiple regression analysis using a predetermined number of the measured polishing rates acquired for a plurality of types of polishing targets;
A method for manufacturing a semiconductor device, comprising:

(Supplementary Note 9) After calculation of the polishing rate ratio by the polishing rate ratio calculation unit,
Calculate the predicted polishing rate using the polishing rate ratio corresponding to the type of polishing object for which the measured polishing rate was calculated,
Adjusting the calculated polishing time using the calculated predicted polishing rate and the polishing rate ratio corresponding to the type of the next polishing target;
The method for manufacturing a semiconductor device according to appendix 8, wherein:

  (Supplementary Note 10) The multiple regression analysis is performed using the predetermined polishing time and the type each time the predetermined number of actual polishing rates are acquired together with the predetermined number of actual polishing rates. The method for manufacturing a semiconductor device according to appendix 8 or 9.

It is a schematic explanatory drawing of CMP process. It is a conceptual diagram of transition of an actually measured polishing rate. It is explanatory drawing of the grinding | polishing of multiple types of grinding | polishing object, Comprising: (A) is a figure which shows an example of transition of the measurement polishing rate at the time of grinding | polishing to multiple types of grinding | polishing object, (B) is a kind of (A) It is a figure which shows the measured polishing rate for every and its simple average. It is a figure which shows the structural example of a grinding | polishing control system. It is a figure which shows an example of transition of a measurement grinding | polishing rate. It is a figure which shows transition of a trend polishing rate.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Polishing control system 2 Polishing process part 2a Polishing part 2b Control part 2c Memory part 3 Film thickness measurement part 3a Measurement part 3b Control part 3c Memory part 4 Process control part 5 Polishing rate calculation part 6 Polishing time calculation part 7 Polishing rate ratio calculation Section 8 Data storage section 8a First memory section 8b Second memory section

Claims (5)

In the polishing control system that controls the polishing process,
A polishing processing unit for polishing a film formed on the object to be polished in a predetermined polishing time;
A film thickness measuring unit for measuring the film thickness of the film before and after polishing;
A polishing rate calculation unit that calculates an actual polishing rate using the predetermined polishing time and the film thickness, and calculates a predicted polishing rate in the next polishing processing unit using the calculated actual polishing rate;
A polishing time calculation unit that calculates a polishing time in the polishing processing unit using the calculated predicted polishing rate;
A polishing rate ratio calculation unit for calculating a polishing rate ratio for each type used for adjusting a polishing time according to the type by multiple regression analysis using a predetermined number of the measured polishing rates acquired for a plurality of types of polishing targets. When,
A polishing control system comprising: After calculating the polishing rate ratio by the polishing rate ratio calculation unit,
The polishing rate calculation unit calculates the predicted polishing rate using the polishing rate ratio corresponding to the type of the polishing target for which the measured polishing rate is calculated,
The polishing time calculation unit adjusts the calculated polishing time using the calculated predicted polishing rate and the polishing rate ratio corresponding to the type of the next polishing target,
The polishing control system according to claim 1.   The polishing rate ratio calculation unit performs the multiple regression analysis using the predetermined polishing time and the type each time the predetermined number of actual polishing rates are acquired together with the predetermined number of actual polishing rates. The polishing control system according to claim 1 or 2, characterized by the above. In the polishing control program used to control the polishing process,
Computer
A polishing rate calculation unit that calculates a measured polishing rate using a film thickness before and after polishing of the film and a predetermined polishing time of the film, and calculates a next predicted polishing rate using the calculated measured polishing rate,
A polishing time calculation unit that calculates a polishing time using the calculated predicted polishing rate,
A polishing rate ratio calculation unit for calculating a polishing rate ratio for each type used for adjusting a polishing time according to the type by multiple regression analysis using a predetermined number of the measured polishing rates acquired for a plurality of types of polishing targets. ,
A polishing control program characterized by functioning as: In a method for manufacturing a semiconductor device having a polishing process,
Forming a transistor on a semiconductor substrate;
Forming a film on an upper layer of the semiconductor substrate on which the transistor is formed;
Measuring the film thickness of the formed film;
Polishing the film at a predetermined polishing time;
Measuring the film thickness of the film after polishing;
Calculating an actual polishing rate using the film thickness before and after polishing and the predetermined polishing time, and calculating a next predicted polishing rate using the calculated actual polishing rate;
Calculating a polishing time using the calculated predicted polishing rate;
A step of calculating a polishing rate ratio for each type used for adjustment of a polishing time according to the type by multiple regression analysis using a predetermined number of the measured polishing rates acquired for a plurality of types of polishing targets;
A method for manufacturing a semiconductor device, comprising:

Images