TW200305240A - Method and system for controlling the chemical mechanical polishing of substrates by calculating an overpolishing time and/or a polishing time of a final polishing step - Google Patents

Abstract

A method and a controller for the chemical mechanical polishing (CMP) of substrates and, in particular, for the chemical mechanical polishing of metallization layers is disclosed. In a linear model of the CMP process, the erosion of the metallization layer to be treated is determined by the overpolish time and possibly by an extra polish time on a separate polishing platen for polishing the dielectric layer, wherein the CMP inherent characteristics are represented by sensitivity parameters derived empirically. Moreover, the control operation is designed so that even with a certain inaccuracy of the sensitivity parameters due to subtle process variations, a reasonable controller response is obtained.

Description

200305240 V. Description of the invention (1) [Technical field to which the invention belongs] The present invention relates to the field of integrated circuit manufacturing, especially to the chemical mechanical mechanism of the metallization layer material layer in various manufacturing stages of integrated circuit Polishing (Chemical Mechanical Polishing, CMP) process. [Previous technology] In the manufacture of precision integrated circuits, a large number of semiconductor elements, such as field-effect transistors, capacitors, and the like, are manufactured on a plurality of wafer regions (die), which are distributed throughout The entire surface of the substrate extends. As the structural size of individual semiconductor devices continues to shrink, it is necessary to provide various material layers, which are arranged as uniformly as possible on all substrate surfaces to conform to the underlying surface shape to ensure a successful patterning process. For example, the quality required for 4 shadows, surname reduction and similar situations. Recently, chemical-mechanical polishing has become a widely used technique in the preparation process of planarizing an existing material layer for deposition as an adhering material layer. Chemical mechanical polishing specifically focuses on the envy of the so-called metallization layer, which includes layers such as recessed portions of channels and trenches. 7 ^ 'Also is filled with the appropriate metal to form a metal wire to connect the individual two components . In the prior art, aluminum has been used as the preferred bull carcass and in precision integrated circuits, it may be necessary to provide multiple metallization layers' metallization layers to obtain twelve layers between semiconductor elements. The connection. Due to the electrical migration of copper and its conductivity compared to aluminum, several companies are now beginning to replace aluminum with copper. In general, due to ·. The split garment has higher conductivity and the copper wire can be made into a smaller cross section. Second, compared to aluminum, it is! Made of steel

200305240 V. Description of the invention (2) It can reduce the number of metallization layers required for the required function. However, the planarization of individual metallization layers is still very important. The technique commonly used to form copper metallization lines is called the damascene process (d a m a s c e n e process), in which channels and trenches are formed in an insulating layer, and copper is filled into the channels and trenches. Thereafter, the excess metal is removed by chemical mechanical polishing after metal deposition, thereby obtaining a planarized metallization layer. Although CMP has been successfully used in the semiconductor industry, this method has proven to be both complex and difficult to handle, especially when handling most large substrates. φ In the CMP process, the substrate, such as a wafer supporting semiconductor elements, is mounted on a suitably formed carrier called a polishing head, and the carrier moves relative to the polishing pad, and the wafer surface is in contact with the polishing pad . During this process, an abrasive containing a chemical composition is supplied to the polishing pad, and the chemical composition reacts with the material or materials of the layer to planarize, for example, by converting a metal into an oxide The reaction product, such as copper oxide, is mechanically removed by the binder contained in the abrasive and polishing pad. The CMP process has a problem due to the fact that at some stage of the process, different materials may appear on the layer to be polished at the same time. For example, after removing most of the excess copper, it is necessary to apply a layer of insulating material such as silicon dioxide, as well as copper and copper oxide, through the use of abrasives, polishing pads, and adhesives in abrasives. Chemical and mechanical processing. Generally, the composition of the abrasive is selected to have the best polishing characteristics for a specific material. In general, different materials show different removal rates so that, for example,

92307.ptd Page 6 200305240 ---------___ 5. Description of the invention (3) A '-The surrounding insulation material is processed more quickly than the surrounding recessed part :: Effective at the top of the line Shape = during the removal of excess metal, it can be the same in the case of the same material: although the removal rate of the insulating material is compared to the removal of copper; = material shifts to, and subtracts " (er〇 Slon). The reduction of the thickness of the metal layer is generally called the "Melting" of the material of the dishing layer and the metal layer, regardless of the difference between the material and the thickness of the material layer including the insulating layer. May change along substrate surface. Even I can change the two in a single wafer area in response to a planarized pattern, that is, 'the removal rate of the metal and the insulating material is based on various factors', such as the type of abrasive, the structure of the polishing pad, Structure and type of polishing =, the relative movement between the polishing pad and the substrate, § the pressure applied to the substrate when moving relative to the polishing pad, the type of the feature pattern to be polished on the substrate, and the bottom layer Uniformity of insulation layer and metal layer, etc. From the above considerations, there are a number of related parameters that obviously affect the surface shape of the final obtained, chemically-modified layer. Therefore, a lot of efforts have been made to develop CMP tools and methods to improve the reliability and robustness of the CMP process. Example ϋ In a sophisticated CMP tool, the polishing head is mounted to provide the possible application of σ—adjustable pressure to two or more portions of the substrate to control friction and the area of the substrate corresponding to these different head portions On the mobile. Then ’in this way that the movement rate is as uniform as possible over the entire surface area’ the polishing platform and polishing head carrying the polishing pads are opposed to each other

200305240 i. Description of the invention (4) Move in order to maximize the service life of polishing pads that gradually wear out during operation. For this purpose, a so-called shim adjuster is additionally provided in the CMP tool, which moves on the polishing pad and revises the polishing surface so that a similar polishing condition can be maintained for best performance May be many substrates. The polishing pad is adjusted substantially uniformly, while the pad adjuster will not interfere with the movement of the polishing head and control the movement of the pad adjuster. Due to the complexity of the CMP process, it may be necessary to implement two or more process steps, preferably on different polishing platforms, in order to obtain a polishing result that meets the strict and necessary conditions for manufacturing sophisticated semiconductor devices. For example, when manufacturing a metallization layer, the minimum cross-section of individual metal wires must be established according to design rules to obtain the desired resistance. The resistance of individual metal wires depends on the material type, wire length, and cross section. Although the two previous factors have not changed substantially during the manufacturing process, the cross section of the metal wire may change significantly, and thus the resistance and quality of the metal wire will be affected due to erosion and polishing in the related CMP process. . Therefore, the semiconductor designer must take these changes into account and implement an additional “safety” thickness of the metal wires so that the cross section of each metal wire is reliably within a certain tolerance range after the polishing operation is completed. φ As can be understood from the above considerations, large-scale efforts are underway to improve the yield of the chemical mechanical polishing of the substrate while still maintaining high quality standards. Due to the characteristics of the CMP process, the in-situ measurement of the thickness of the layer to be removed and / or the removal rate are very difficult to predict. In fact, before or after a predetermined number of product substrates have been processed,

1 92307. ptd page 8 200305240 V. Description of the invention (5), a plurality of imitation substrates are used to adjust and / or calibrate the CMP tool. Because the processing of the counterfeit wafer is extremely cost intensive and time consuming, it has recently attempted to significantly reduce the number of sample measurements by implementing appropriate control mechanisms to maintain the performance of the CMP process. In general, it would be highly desirable to have a control process in which specific CMP parameters are operated based on the measurement results of a substrate that has been processed so that the last layer in the specification can be accurately maintained Thickness and dishing and erosion. In order to accomplish this so-called '' batch-by-batch '' plutonium control on this production line, at least two conditions must be met. First, appropriate metrology tools (metr ο 1 〇gy) must be implemented into the production line, so that each substrate that has completed the CMP process can immediately receive a measurement result, and the result must be before the CMP process or at least immediately after it The substrate is provided to the CMP tool before the final stage of the CMP process. Second, a model of the CMP process must be established to expose appropriate operating variables to obtain the desired polishing results. The first case cannot be accomplished without significantly adversely affecting other manufacturing process parameters, such as productivity and therefore high cost effects. Therefore, in practice, a plurality of substrates are subjected to CMP processing until the first measurement results of the substrates that are initially processed are valid. That is, the control loop contains a specific amount of delay, which must be taken into account when adjusting process parameters based on the measurement results.

Regarding the second term, a number of CMP models have been established to take into account the fact that these operating variables are controlled based on the aging feedback results. For example, in the minutes of the eighth AEC / APC symposium in 2001, "CMP * 111 ill __i_ |

L

1 hidden JI1 slightly

i 92307.ptd Page 9 200305240 > V. Description of the invention (6) Comparison of R2R control algorithm and measurement delay 〃, Cham ness et al. revealed the three CMP models when operating under delayed measurement feedback. Comparing results. In this paper, the authors show that when this measurement is provided with some degree of delay to the CMP tool, the process_control of only a predictive model avoids any instability in the control function. Given this previous Technology, in general, a predictive model is needed, such as the model described in the cited paper and a set of experimental data, to obtain process variables that can be operated to obtain the desired output of the CMP process, such as applied to Substrate pressure, abrasive composition, etc. _ Although the CMP process control has been successfully applied to many semiconductor devices, from the considerations produced so far, the reliable and robust CMP process used in precision integrated circuits includes information about the process The large-scale efforts of tools and control operations, so it is highly desirable to have a simplified but still effective CMP control process and control system, but still ensure the same high quality standards required for processed substrates. The present invention proposes There are methods that may solve or at least reduce some or all of the above problems. SUMMARY OF THE INVENTION In the above, the present invention relates to a method and a controller that allow the control of the CMP process by operating a process number that can be easily used, whereby the clear process characteristics are explained through parameters obtained through experience, but its accuracy It is not important for proper control functions. Therefore, in the illustrative embodiment of the present invention, the method for controlling the chemical mechanical polishing of a substrate includes empirically obtaining a first sensitivity parameter

92307. ptd Page 10 200305240 V. Description of the invention (7) The number is a quantitative description of the relationship between the excessive polishing time of the first material layer and the control variable related to the first material layer, and the second is obtained empirically. The sensitivity parameter is a quantitative description of the relationship between the control variable related to the second material layer and the control variable related to the second material layer of the previous substrate. In addition, the method includes calculating an excessive polishing time of the first material layer from a linear model, the linear model including a control variable, a first sensitivity parameter, a second sensitivity parameter, and a command value for the control variable related to the second material layer The excessive polishing time of the second material layer, the control variable of the second material layer, and the control variable related to the second material layer of the previous substrate, wherein the excessive polishing time is obtained by a weighted moving average. In addition, the over-polishing time of the first material layer was adjusted to the calculated over-polishing time.

According to a further illustrative embodiment, a method for controlling chemical mechanical polishing of a first metallization layer in a substrate includes empirically determining a sensitivity parameter α, which quantitatively illustrates that the over-polishing time TQP is correlated to Effect of the control variable Efirst on the first metallization layer. Furthermore, the sensitivity parameter 7 is obtained empirically, which quantitatively describes the control variable E secc) nd of the second metallization layer of the substrate and the control variable Ep, seec) nd of the second metallization layer of the previous substrate. Effect on the control variable Efirst. Furthermore, the method includes calculating the over-polishing time T QP of the first metallization layer from a linear model. The linear model includes at least the following items: E, E p, first, J (T〇p Τ 〇ρ) Ν 7 (E second Ep, second) 'where T p, (is the over-polishing time of the previous substrate. In addition, the true over-polishing time of the chemical mechanical polishing process is adjusted to the calculated over-polishing time T QP. Follow the further instructions Specific embodiment, the chemical machinery of the substrate

92307. ptd Page 11 200305240 'Fifth, the description of the invention (8) The controller used for polishing, which includes an input section for inputting at least one of the measured values of the sensitivity parameter and the control variable, and an output Part for outputting at least one of an excessive polishing time and a 'last polishing time' as an operation variable. The controller further includes a calculation-calculation part, which is configured to calculate the over-polishing time of the first material layer from the linear model, wherein the linear model includes the second material layer in addition to the first material layer. The control variable, the first sensitivity parameter, the second sensitivity parameter, the command value of the control variable, the excessive polishing time of the second material layer, the control variable about the second material layer, and the control variable of the second material layer II of the previous substrate. Furthermore, a part of the installation will be calculated to find the operating variable by a weighted moving average. According to a further illustrative embodiment, the controller for chemical mechanical polishing of the first metallization layer of the substrate includes an input portion for inputting a sensitivity parameter α, a sensitivity parameter r, and at least one control variable. A measured value of Enm, where the control variable EfhSt represents one of erosion and dishing. Furthermore, the controller includes an output section for outputting at least an excessive polishing time T cp as an operation variable for controlling the chemical mechanical polishing. In addition, the controller includes a calculation section configured to calculate an over-polishing time T @ of at least the first metallization _ from a linear model of the CMP process. With this, the linear model includes at least the following:

Efirst, Ep, first, ^ (丁 op Τρ, ορ) '? (E second E Pi second), where E p, f represents the control variable of the first metallization layer related to the previous substrate, TPfC) P represents the previous substrate Over polishing time, E secc) nd represents the control variable of the second metallization layer of the substrate-plate, and Ep, secc) nd represents the

92307.ptd Page 12 200305240 V. Description of the invention (9) Control variable of the second metallization layer of the substrate 0 [Embodiment Mode The illustrative specific embodiment of the present invention is described as follows. 0 In order to make the description clear, the actual implementation is not implemented. All features are described in this specification. Of course, it will be understandable that in the development of any of the real embodiments, the determination of various explicit implementations must be carried out in order to obtain the clear goals of the developers, such as the compliance of system-related and enterprise-related restrictions. 5 The system will be implemented by a specific implementation. Example to another specific embodiment and it will be understood that this development effort may be complex and time consuming but will still be a routine that can be implemented for those skilled in the art to benefit from the disclosure of the present invention Work 〇— Generally speaking, the specific embodiments described before @ and the specific embodiments to be described below depend on the dishing and erosion of the material layer of the substrate. 5 For example, the metallization layer can be adjusted by the CMP process. The excessive polishing time is maintained at a tightly set tolerance. Generally, the excessive polishing time indicates that During the duration of the CMP process after the measurement, it has been shown that the material is removed from a predetermined area on the substrate. The process of detecting voids in a specific area is also referred to as endpoint detection, and it is used in the manufacture of metallization layers £ 5 or more in the CMP process. • As explained earlier, the CMP process used for the metallization of the metallization layer on the 1¾ end integrated circuit is often designed as a multi-step process, such as the last step of the process. 9 After the metal is removed, the polishing operation is performed on the dielectric layer. Therefore, by adjusting the process time of the final polishing step, the degree of reduction and dishing can be controlled. In order to reliably predict the CMP after old removal Steps

92307. ptd page 13 200305240 V. Description of the invention (ίο) Appropriate over polishing time and / or process time, the inventor proposes a linear model of the CMP process, the linear model depends on the same previous metallization layer and the previous substrate And / or dish-shaped depressions and / or layer thicknesses. In this model, the machinery inherent in the process is represented by two or more sensitivity parameters, which may be obtained by experiments and / or calculations and experiments. In some specific embodiments, the sensitivity parameters Accuracy is not essential to successful control operations because of the `` head-to-tail '' design of control functions. Therefore, in contrast to the conventional control strategy described, for example, in the background portion of this application, in the present invention, process parameters that can be simplified and precisely adjusted are selected as the operational variables of the control operation. Referring to Figure 1, a typical CMP tool and process description will be used in conjunction with the illustrative embodiments described herein. In FIG. 1, it is a schematic diagram illustrating a CMP system 100 including a CMP tool 110, a metrology tool 130, and a CMP controller 150. The CMP tool 110 includes an input section 1 1 1 for receiving a substrate for processing, and an output section 1 12 for receiving and storing the substrate after the CMP process is completed. The CMP tool 1 1 0 further includes a process box 1 1 3, which includes three polishing platforms 1 1 4, 1 15 and 1 1 6 which are referred to as a platform I, a platform Π, and a platform m, respectively. On each of the platforms 114, 11 5 and 11 6 are provided a shim adjuster 1 1 7, an abrasive supply 1 1 8, and a polishing head 1 1 9. On the platform Π, the measuring member 120 is configured and assembled to detect the end point of the CMP process. For the sake of simplicity, the substrate is transferred from the input part 1 1 1 to the platform -I, or from the platform I to any further component of the platform Π, etc.,

92307. ptd page 14 200305240 V. Description of the invention (11) and any components used to feed gas, liquid, such as water, abrasives and the like, are not described in the drawings. ~ In operation, the substrate 1 2 1 containing one or more metallization layers is attached to the polishing head of the platform 1. It should be noted that the substrate 1 21 represents the existing 〃 t I plate, and the operating parameters used to explain the control process will be established > = for the substrate, that is, the operation variable represents a process parameter, ^ The parameter value is changed to obtain a desired value of a control variable, such as the dish shape, surface reduction, and the final layer thickness. The metallization layer with a base = 121 immediately processed by the CMP tool 1 1 0 is also referred to as the first metallization layer, but any metal that is below the first "metallization layer and has been subjected to the CMP process on the substrate 1 2 1 Θ 'is also called the second metallization layer. Furthermore, the substrate that has been subjected to CMP ρ 4 is called the previous substrate, and the metallization layer corresponding to the existing substrate 1 2 1 is also called the metallization layer of the previous substrate. Are the first and second metallization layers, as in the existing substrate 121. The object is on the substrate 1 2 1 with a predetermined process parameter, such as a predetermined abrasive composition%, a predetermined relative movement between the optical head 11 9 and the platform 11 4, and a CMP process. < Duration and the like, and after the CMP process has been completed on the platform I, the substrate 1 2 1 may pass to the platform Π with different process parameters, ρ ^ in the second CMP step until the measurement device 1 2 0 indicates that the end point of the system has been reached. As explained previously, and the detailed description of the substrate u η f 'substrate 1 2 1 will be made with reference to FIG. 2 for an excessive polishing time T g on the platform Π The excessive polishing time T ^ is controlled by the controller 1 5 〇 Determined. Vivax excessive polishing _ Τ called temple elapsed after transmitting dish 121 to the platform will be a substrate, an insulating material in the polishing system of this fourth layer of metal with an appropriate process parameters Lysaght

200305240 V. Description of the invention (12) Application 5 For example, the composition of the abrasive, the relative movement between the platform 1 16 and the polishing head 1 1 9 to generate a pressure applied to the substrate 1 2 1 and the like. In the specific embodiment shown in FIG. 1, the process time at the platform m is also called τΒa, which is obtained by the controller 150. After the polishing step on the platform II is completed-the substrate 1 2 1 is transferred to the output part 1 1 2 and possibly to the metrology tool 130. The measurement results of the-metallization layer are obtained using this tool, for example Layer thickness, erosion, and dishing. In the various embodiments described, the layer thickness, erosion, and dish-shaped depressions are considered individually or in combination as control variables of the CMP process, but Ding and / or • T, a όν act as Operation • Variable. In general, the results of the control variable 1 are obtained by the well-known optical measurement technology, and its description will be omitted. 0 Refer to Figure 2 of BS, which will be used to obtain the operating variables T \ p and Tm. Illustrative specific embodiments. In the second figure, in the first step 2 10, the sensitivity parameter is obtained. In a specific embodiment, the sensitivity parameter is obtained by experiments based on the test substrate or product substrate processed previously. The first stem sensitivity parameter α is thus obtained, and explains the excessive polishing time Tc) P control: the influence of control variables, such as erosion reduction, dish-shaped depression, metallized layer thickness, and the like. The second sensitivity parameter / 3 may also be sought. • The effect of the polishing time 1 ΙΠ made by cmp on the platform m on the control variable will be specifically explained. In addition, the third sensitivity parameter 7 is obtained to quantitatively explain the control variables of the previous metallization layer, such as the dish-shaped depression of the previous layer and / or the reduction of the last name ', which affects the existing, that is, the control of the first metallization layer. Variable, the previous layer will also be referred to as the second gold as previously noted

92307.ptd page 16 200305240 V. Description of the invention (13) The chemical layer. In particular, the sensitivity parameters α and /? Include inherent CMP mechanisms, such as the rate of movement, which may change during the actual CMP process, for example, due to degradation of polishing pads, saturation of abrasives, and the like. In a particular embodiment, as will be explained in detail later, let α and shi represent separate numbers used for the benefits of the simple linear CMP model. It is considered that the change in the specific process of / 9 is substantially the same as that of the last. As a result, there will be no adverse effects, and the remaining control operations are designed accordingly without considering any changes in α and. In a further specific embodiment, the sensitivity parameters α and β are selected in accordance with time, that is, the number of substrates that have been processed or are being processed in view of slight changes in the process conditions. In step 2 2 0, the intermediate value for the operation variable (called T% ρ, T \) is calculated from the linear CMP model. In this aspect, the linear model is understood as a mathematical expression that illustrates the relationship between various variables, such as the operating variables T QP, I'm, and control variables, where the variable does not have any such as Τ 2ϋρ, Τ, etc. Higher-order terms appear as linear terms. Referring to Fig. 3, an illustrative specific embodiment for obtaining D and D will be described. In Figure 3, step 2 2 0 is further divided into the first sub-step 2 2 1 to describe the linear model of the CMP process. According to this method, the control variable of the first metallization layer is represented by E iirst, wherein it should be noted that the control variable may represent any of erosion, dishing, metallization layer thickness, and the like, and Efirst It is expressed by the following equation:

Efirst = Ep, first + ^ (Τ〇ρ— Τρ, ορ) + β (Tm — Τ p, ffl) + [r] (Ε second Ε ρ, second) (1)

92307.ptd Page 17 200305240 V. Description of the invention (14) ^ where the index p represents a variable, which refers to the previous substrate, and the indexes first and sec ο nd respectively refer to the first metallization layer being processed. And a second metallization layer that has been processed. Therefore, the sign of α is preferably selected as positive, while the sign of / 5 is preferably selected as negative. The size and sign of 7-was obtained experimentally. Furthermore, as previously explained, in a specific embodiment, if the final CMP step is not used on the platform II and there is only a single operation variable, TQP may be used to control the entire CMP process. As can be understood from Equation 1, for a particular Ep, fii-st 'for example, compared to the first metallization layer of the previous substrate Tρ, ορ, the excessive polishing time T Qp of a metallization layer The increased erosion of the first metallization layer will increase E fil »st by an amount, which is obtained by multiplying these over-polishing times (T 0p-T p, op) by the difference in sensitivity parameter α. It is therefore obvious that a change in the inherent mechanism of CMP made by a single number α or a certain inaccuracy in finding α may affect the result of Emst, and thus produce a T 叩 value, which is In some cases it may be considered inappropriate to obtain a desired E target, where E target is the target value of the control variable. The same is true for the sensitivity parameter cold. Therefore, in a specific embodiment, as mentioned earlier, in the sub-step 222, the parameters α and / 3 may be selected as time-depent parameters #, or more appropriately, determined as The number of substrates to be processed. In this way it is possible to take into account the general tendency of the degradation of polishing pads, abrasive compositions and the like, so that it is possible to compensate for system changes of α and / 3. That is, because the number of processed substrates increases, a systematic decrease in polishing rate over time may be increased by a corresponding increase.

1111 m fli 92307. ptd page 18 200305240 V. Description of the invention Add α and may be chosen among them (i) Some kind of device must be obviously extended and polished. It should be used for 'v elimination', soft evening display extreme only in each base from a previous change. Real-time operation on the board. Take into account the deputy (15) / or decrease / 3. Therefore, α and / or / 9 are equations (i) and / or / 5 = cold (i), and represent the number of processed substrates. This feature reveals the ability to predict the degree of CMP control, which may be beneficial when controlling the response to the measurement results and may have late correlations with existing processed substrates as described previously. In step 2 2 3, the intermediate value used for the excessive polishing time of the operating variable on the platform m is obtained from the model in step 2 2 1 and the reason for the intermediate variables T and T \ lies in the control operation except " Any short-term changes in the CMP process, and should correspond to the measurement results of previously processed substrates in a way that does not need to be insufficient and excessive. Such a control operation may be convenient only when a small number of measurement results of one plate is valid, so that the measurement result of the substrate to another previous substrate may show a clear, ie, measurement result of Ep, first, which is caused by Obtained from separate measurements of a single location determined by the previous basis. Therefore, before the variables Tϋρ and Tffl, find the intermediate operation variable D: and T%. Step 2 2 3 is used in the following example), first + r second

E p, secon d) target (2) This means that the common value E target does not change the over-polishing time compared to the previous polishing time of the previous substrate and does not change the polishing time compared to the polishing time on the platform II of the previous substrate. Get in case of time. As a result, T is equal to TP, QP, and T is equal.

I11P JI hidden

92307.ptd page 19 200305240 ‘fifth, the description of the invention (16) and Τ \ calculated for the following situations

In step 224, change the T shape:

Ep, first +? (E second E p, second ^ < target (3) which means that depends on the previous substrate, the metallization money reduction and / or dish depression, which E really represents, and The erosion effect of the second metallization layer of the existing substrate and the previous substrate results in less erosion than desired and or dish-shaped depressions and / or layer thicknesses. Obviously, the over-polishing time for the existing substrate must be equal to or greater than the previous The excessive polishing time of the substrate, and the polishing time on the platform m must be equal to or shorter than the polishing time of the front substrate j. Therefore,

Τ * Ω > T p, op (4) What's more, in general, the maximum and minimum over-polishing time K, τ QP, and the maximum and minimum polishing time on the platform in, 平台 ι, which may be based on process requirements It is set in advance. These limitations for the excessive polishing time and the polishing time of the platform m may be determined experimentally or empirically. For example, the maximum and minimum overpolishing times ϊ ":, τορ may be individually selected to be approximately 30 seconds and 5 seconds. The maximum and minimum polishing time 〒 I on the platform m, the individual may be selected to be about 120 seconds and 20 seconds, respectively. In a specific embodiment in which the excessive polishing time T # and the platform IΠ polishing time Tffl are used as operating variables, the intermediate value D: and T \ are desirable, so that these values are properly located at the smallest With the maximum over polishing time as well as the allowable range given by the platform m polishing time. In a specific embodiment, the intermediate excessive polishing time τ and the platform m polishing time τ \ are obtained to be placed in the middle around the center of the corresponding allowable range, where

11

92307. ptd page 20 200305240 V. Description of the invention (17) ΊΛΡ and ΊΛ must be selected at the same time, so that the CMP model provides the instruction value Etarget, so D: and T \ are then obtained by: E p, first (T % p-T p > op) + / 3 (-T Ρι) Π) +7 (Ε second Ε ρ, second)-Ε target (5) T, and T ′ placed in the middle of each allowable range May be obtained by calculating the minimum of the following expression:

IK -Tορ! + W K -TUI i Minimum (6) T7P-TOP 2 \ — / Jni-TjiL where Equations 4 and 5 are therefore secondary conditions for finding the minimum T and T T. In a similar manner, in the sub-step 2 2 5, Ding, and TΛ are calculated for the following situations:

Ep, first +7 (ESecond — E p > second) ^^ Target (7) This situation means that the erosion of the first metallization layer of the previous substrate and the incorporated second metallization layer exceeds the desired erosion value. Therefore, the intermediate over-polishing time must be selected to be equal to or shorter than the previous substrate's over-polishing time, and the intermediate platform m-polishing time must be selected to be equal to or longer than the previous substrate polishing time. As a result, τρ, ορ; T \ ^ Tp, ffl (8) is similar to the calculation performed in the sub-step 2 2 4 and also in this case

92307.ptd Page 21 200305240 i. In the description of the invention (18), the minimum value of the expression (6) is obtained by the secondary conditions (5) and (8). In order to qualitatively summarize the above sub-steps used to obtain the intermediate over-polishing time T * QP and the medium 涧 platform Π polishing time τ \, it should be noted that the measurement results of the previous substrate on the second metallization layer, or individually, So the calculated value? The table shows that the expected loss is equal to the desired loss, and then the intermediate over-polishing time t * qp and the plateau m polishing time τ \ correspond to the previous substrate's over-polishing time tp, qpw and platen-in polishing time Tp, ffl. In the case where the "subtraction" # of the previous substrate and the existing substrate 2 2 1 and the second metallization layer of the previous substrate did not yield to the desired erosion reduction value E target, the intermediate polishing time was calculated so that these values were It is centered around the middle of the allowable range, but satisfies the secondary conditions (5) and (6) at the same time, that is, the intermediate polishing time must yield to the desired loss E target and must also comply with the situation (4 ) And (8). In particular, the secondary conditions (4) and (8) ensure that any change in T is not compensated by the corresponding change in the polishing time of the platform 1Π. The corresponding behavior may lead to a simpler solution when the minimum value is obtained according to (6), but it does cause a control operation in the imprecise parameter α and the wrong direction used, and thus makes the control function Unstable. • It should be understood that, in practice, the calculation may be performed with a predetermined accuracy, so any description of the solution equation will of course be subject to some degree of `` mutation '' and tolerable depending on the algorithm Degree of 'inaccuracy'. Therefore, the calculation results described here are generally obtained in rough numbers, and the degree of roughness is determined by, for example, effective calculation functions.

92307.ptd page 22 200305240 V. Description of the invention (19) The rate, the required accuracy, and the factors of the analog can be calculated. For example, in many applications, an accuracy of about 1 second between the overpolishing time and the platform m time is sufficient because polishing activity within one second results in changes in erosion that may be properly within the range of measurement variation. The weighting factor to find the minimum value in Expression (6) may be selected w \ β \ U Η \ — / The weighting factor may also be obtained based on experience. Furthermore, it should be noted that after using only one operation variable, such as the time of excessive polishing time τ @, it is not necessary to find the intermediate value by calculating the minimum value. Referring to FIG. 2 again, in step 230, the actual output values used for the excessive polishing time and the polishing time of the platform m are derived from the intermediate excessive polishing time of the previous substrate and the intermediate platform m polishing time and the excessive polishing time and the platform. Dish polishing time is calculated. Depending on the algorithm used, this situation ensures that the over-polishing time and the stage m polishing time are reasonably stable to the `` progression '' of the previous substrate's over-polishing time and the stage m polishing time. Referring to FIG. 4, an illustrative embodiment is shown to obtain an over-polishing time and a plateau ffl polishing time in step 23. In the first sub-step 231, it may be possible to detect whether T% ρ and / or T is within a predetermined range.

92307. ptd page 23 200305240 ^ V. In the description of the invention (20), the predetermined range may be different from the range defined by the minimum and maximum excessive polishing time and the platform Π polishing time. With these predetermined ranges, it may be possible to detect whether the control operation tends to systematically move out of a properly defined range, indicating that the parameters α and / ?, and therefore the CMP situation, have changed significantly. In this case, in the sub-step 2 3 2, it may indicate that, in the CMP process that is considered “close to the future”, the linear model of the CMP process is no longer valid or invalid. This representation may take the form of any sign that unexpected changes have taken place as a mechanism inherent in CMP. It is noteworthy that the sub-step 2 31 is optional and can be omitted. In the sub-steps 2 3 3, the over-polishing time and the plateau polishing time are calculated by using a weighted moving average of the over-polishing time from the previous substrate and the intermediate over-polishing time T% ρ, and the table-m polishing time is from the previous The polishing time of the substrate platform m and the polishing time τ of the intermediate platform π are calculated as a weighted moving average. As described in 2 3 3, the polishing time T is excessive.卩 is expressed by the following formula: Τ〇ρ = λ T * 〇p + (1— λ) τρ, ορ where I is a parameter in the range of 0 to 1. With the parameter λ, it is possible to adjust the adaptive speed of the previously developed control swing in relation to the excessive polishing time. Similarly, the polishing time of the platform m may be obtained by:

Tin = // + (1-//) T p, ffl where the parameter // adjusts the adaptive speed related to the polishing time of the platform of the previous substrate. Obviously, when, for example, previous substrate measurements show

92307. ptd page 24 200305240 V. Description of the invention (21) When the target value E target has a relatively large deviation value, the value of λ and / or close to 1 results in an immediate response to the excessive polishing time and the polishing time of the platform dish. On the other hand, finding λ and // as fairly low values will only result in a very slow response to any changes in the CMP process. In a particular embodiment, an algorithm called an exponentially weighted moving average (EWMA) is used, in which the same lambda value is used for the over polishing time and the platform Π polishing time. With this EWMA model, it is possible to take into account the effects of the recent progress of the CMP process more effectively than any of the process conditions of aging and aging. Corresponding specific embodiments including EWMA are particularly suitable when there is no insignificant delay from the measurement results from the previous substrate, that is, only some or no substrate is processed between the existing substrate 1 2 1 and the previous substrate deal with. Now referring to FIG. 2 again, in step 24 0, the over-polishing time and the stage calculated in step 2 30 will be used! Π time is transferred to the CMP tool 1 1 0 ′ in FIG. 1 to adjust the corresponding processing time of the substrate 1 2 1 currently being processed. In step 250, the substrate is transferred to a metrology tool 130 to obtain a measurement value for controlling a variable. These measurements may then be used for E, £ p, second, Ep, for calculations on the following substrates. As previously explained, 'this may be delayed to some extent until the measurement result is valid for the controller 150, and in this case, a specific implementation may be described with respect to the sub-step 2 2 2 The example is advantageously used, in which the sensitivity parameters α and / 5 are generated based on parameters depending on the number of substrates that have been processed and are being processed, 'because the controller 15 then displays 1', predictive " behavior, and

92307.ptd Page 25 200305240 V. Description of the invention (22) It is possible to output reliable values for over-polishing time and platform m polishing time, and even for a considerable delay in the control cycle. Furthermore, when this predictive model is applied, the number of measurement operations may be significantly reduced. -In the present embodiment, the currently processed substrate and the previous substrate are referred to as a single substrate, but in an illustrative embodiment, the existing substrate and the previous substrate may represent multiple substrates, such as many substrates , Where the control variables E iirst, E p, iirst, E second, E p, second and the operation variables T ^ and Tffl represent the corresponding intermediate values of the plurality of substrates. The relative # configuration has proven to be particularly useful in production lines, where the well-established CMP process settings are set and the deviations from substrate to substrate within a defined plurality are properly placed within acceptable process parameters. As a result, process control can be implemented on a volume-by-volume basis in a simple, yet effective manner, for most substrates. In a specific embodiment, as shown in FIG. 1, a controller 1 5 0 for controlling operations according to one of the illustrative embodiments described with reference to FIGS. 2 to 4 includes an input portion 1 5 1 A calculation part 1 5 2 and an output part 1 5 3, where the input part 1 5 1 is operatively connected to the metrology tool 1 3 0 and the output part 1 3 3 is operatively connected to the CMP Tools 1 1 0. When the CMP process is controlled based on the board-to-substrate repair, the metrology tool 130 and the controller 150 are implemented as in-line devices, so as to minimize the transport of the substrate and accelerate the input of measurement results. In the input section 1 5 1. In a further specific embodiment, preferably, when a plurality of substrates are used by the plurality of substrates, the intermediate polishing time and / or the intermediate value of the polishing time of the platform m

92307. ptd page 26 200305240 V. Description of the invention (23) For control, the weighing and weighing tool 130 and the controller 150 may be placed outside the production line. The controller 150 can be implemented as a single-chip microprocessor, because the microcontroller has analog or digital signals that may be supplied directly from the metrology tool 130, or may be part of an external computer, such as a PC or workstation Or it may be a management system commonly used in semiconductor manufacturing plants. In particular, the calculation steps 2 2 0 and 2 3 0 may be performed by any numerical algorithm, which includes the methods used to solve related equations, fuzzy logic, desktop parameters, and especially the analysis method used for EWMA, and The corresponding operation code may be set in the controller 150. Furthermore, because it is only necessary to obtain the sensitivity parameter α and / or it is necessary, the above specific embodiment may be easily adapted to any known CMP tool, which explains the corresponding CMP tool and the operations performed on this tool. Inherent in the basic CMP process. The specific embodiments disclosed above are merely illustrative. Although the present invention may be modified and implemented in different but equivalent ways that those skilled in the art know, it still has the benefit of its teachings. For example, the process steps described above may be performed in a different order. Furthermore, with the exception of the scope of patent applications described below, there is no intention to limit the details of the structure or design shown here. It is therefore clear that the particular embodiments disclosed above may be altered or modified, and all such alterations are considered to be within the scope and spirit of the present invention. Therefore, the protection sought here is set out in the scope of the patent application below. Although the present invention allows various modifications and alternative forms,

92307.ptd Page 27 200305240 ^ V. Description of the Invention (24) The examples have been shown by examples in the drawings and explained in detail here. In any case, it should be understood that the description of the specific embodiments herein is not intended to limit the present invention to the particular type disclosed, but instead, the present invention covers the essence of the present invention as defined by the scope of the additional patent application • All modifications, equivalents, substitutes within God and scope.

92307.ptd Page 28 200305240 Brief description of the drawings [Simplified description of the drawings] The present invention can be understood by referring to the following descriptions related to the drawings, wherein the same reference numerals represent the same elements, and among them: The figure shows a schematic diagram of an exemplary CMP tool, showing an illustrative specific embodiment for implementing the present invention; FIG. 2 depicts a flowchart representing a specific embodiment of a method for controlling CMP; and FIG. 3 represents a second embodiment FIG. 4 is a flowchart showing details of the specific embodiment shown in FIG. 4; and FIG. 4 is a flowchart showing further details of calculating a manipulation parameter according to the specific embodiment shown in FIG. 100 CMP system 110 CMP tool 111 input part 112 output part 1 14 polishing platform 115 polishing platform 116 polishing platform 117 cymbal adjuster 118 abrasive supply 119 polishing head 120 measuring member 121 substrate 130 weighing tool 150 CMP controller 151 Rarely entered part 152 Calculation part 153 ¥ m Out part

92307.ptd 29th 1

Claims (1)

200305240 VI. Scope of patent application 1. A method for controlling chemical mechanical polishing of a substrate, the method includes: obtaining a first sensitivity parameter, which quantitatively explains the excessive polishing time used for the first material layer and the time period associated with the first material layer; The relationship between the control variables * to obtain the second sensitivity parameter is a quantitative description of the relationship between the control variable related to the second material layer and the control variable related to the second material layer of the previous substrate; from the chemical mechanical formula The linear model of the polishing process calculates the over-polishing time of the first φ material layer, wherein the model includes at least the control variables related to the second material layer, the first sensitivity parameter, the second sensitivity parameter, and the The command value, the excessive polishing time of the second material layer, the control variable related to the second material layer, and the control variable related to the second material layer of the previous substrate; calculating the weight of the excessive polishing time of the first material layer A moving average; and the chemistry of the substrate corresponding to the calculated over-polishing time During the mechanical polishing period, the excessive polishing time of the first material layer is adjusted. # The method according to item 1 of the patent application scope, wherein the control variable represents at least one of erosion reduction, dish-shaped depression, and material layer thickness. 3. The method according to item 1 of the patent application scope, further comprising: measuring at least one of the first and second material layers of the previous substrate to obtain at least one of erosion reduction, dish-shaped depression, and material layer thickness. one of them 92307.ptd page 30 200305240 6. Scope of patent application. 4. For the method in the first item of the patent application scope, wherein each control variable represents an intermediate value for a plurality of substrates. 5. The method according to item 1 of the patent application scope, wherein the first sensitivity parameter depends on one of the number of substrates processed and the number of substrates to be processed. 6. The method according to item 1 of the patent application scope, wherein the chemical mechanical polishing process includes: a final polishing step, which can be performed on a separate polishing platform with adjustable additional polishing time. 7. The method according to item 6 of the patent application scope, further comprising: obtaining a third sensitivity parameter, which quantitatively explains the relationship between the control variables and the additional polishing time. 8. The method according to item 7 of the patent application scope, further comprising: calculating the additional polishing time from the linear model. 9. The method of claim 8 in the scope of patent application, wherein the calculation of the excessive polishing time and the additional polishing time includes calculating the intermediate excessive polishing time and the intermediate additional polishing time, such that the distance between the intermediate excessive polishing time and the intermediate additional polishing time The compound deviation value of the middle point of the corresponding allowable range is approximate to the minimum value. 1 〇. The method according to item 9 of the patent application range, wherein the minimum value is obtained under the following conditions: when compared to the previous substrate values, the intermediate over-polishing time and the intermediate additional polishing time are changed in a different direction , And the intermediate excessive polishing time and the intermediate additional polishing time produce a control variable value related to the first material layer that is substantially equal to the command value. 92307.ptd Page 31 200305240 4 Scope of patent application • 1 1. A chemical mechanical polishing method for controlling a first metallization layer in a substrate, the method includes: obtaining a sensitivity parameter α, the sensitivity parameter is quantitatively Explain the effect of the over polishing time T% used on C MP (C hemica 1 M echanica 1 P ο 1 ishing) after detecting an endpoint on the control variable E first related to the first metallization layer ; Find the sensitivity parameter T, which quantitatively describes the control variable E secQnd related to the second metallization layer of the substrate and the control variable Ep, seeQnd to the control variable Efirst | Influence; and calculate the over-polishing time TQP of the first metallization layer from a linear model, the linear model includes at least the following items: E ~ st, Ep, first ^ a T〇p- T p, op r second E p, second where TD ,. ^ Over-polishing time of the previous substrate; and during the chemical mechanical polishing of the first metallization layer of the substrate, the calculated over-polishing time T π is selected as the actual over-polishing time. 1 2. The method according to item 11 of the scope of patent application, wherein calculating T QP includes: calculating an intermediate over-polishing time T% p, which is equal to the desired value Etarget of the control variable Eiirst obtained by 0. Necessary; and Calculate T QP from the excessive polishing time of the previous substrate T p,% and the intermediate excessive polishing time T as the weighted moving average. 13. The method according to item 12 of the scope of patent application, wherein the weighted moving average is an exponentially weighted moving average. mm 92307.ptd page 32 200305240 VI. Application for patent scope 1 4. As for the method of patent application scope item 11, each of the control variables represents the middle value of a plurality of substrates. 15. The method according to item 11 of the scope of patent application, wherein each of the control variables represents one of erosion of the first and second metallization layers, dish-shaped depressions, and layer thickness. 16. The method according to item 11 of the scope of patent application, further comprising: measuring the control variable of the previous substrate, and using the measured value of the control variable to calculate the over-polishing time T ορ. 17. The method according to item 12 of the scope of patent application, wherein when the intermediate excessive polishing time exceeds the predetermined value range, the loss of validity of the linear model will be expressed. 18. The method according to item 11 of the scope of patent application, wherein the chemical mechanical polishing process includes: a final polishing step, which is performed on a separate polishing platform, thereby using the process time of the final polishing step as a manipulation variable , Tffl. 19. The method according to item 18 of the scope of patent application, further comprising: finding a sensitivity parameter / 5, which quantitatively explains the effect of the final polishing time TE on the control variable Efirst. 20. The method according to item 19 of the patent application scope, wherein the linear model further includes the item: cold (Tm — T p, ffl), where T p, m represents the last polishing time of the previous substrate, and wherein the The excessive polishing time T QP and the final polishing time Tffl are calculated from a model including this term. 2 1. The method according to item 20 of the scope of patent application, wherein the model is given by: 92307. ptd page 33 200305240 6. Scope of patent application first d, first \ a (T op Τ ρ, op) + / 5 τ Τ Ρ, ΠΙ Ύ second Ε ρ, second 2 2 The method further includes: before calculating the over-polishing time T QP and the final polishing time Tm, calculating-calculating the intermediate over-polishing time T, and the intermediate final polishing time T \ 23 · 如 时 此 别 2 4 · 如The method of item 22 of the second patent application range, wherein the intermediate over-polishing room and the intermediate final polishing time are calculated under the secondary conditions. Under the conditions, Ding: and T * m are selected to substantially generate the desired value Etarget At the same time, Ding ^ and T \ distance Ding: the sum of the deviations from the center points in the predetermined value range of T \ is minimized. The method in the 23rd scope of the patent application, in which T% ρ and is calculated under the necessary conditions, under which, when Ep, first + r (Ese_d Ep, sec: Qnd) is greater than the desired value Etarget 'T * Qp is equal to or less than the previous polishing time of the previous substrate, and T is equal to or greater than the last polishing time of the previous substrate. 2 5. The method according to item 23 of the scope of patent application, in which Ding, and T \ are calculated under the secondary conditions. Under this condition, when E p, first + 7 " (E Sec〇nd —Ep When secc) nd) is less than the desired value Etarget, 'T * Qp is equal to or less than the excessive polishing time of the front substrate, and T \ is equal to or greater than the additional polishing time of the previous substrate. 2 6. The method according to item 21 of the scope of patent application, wherein the excessive polishing time T @ and the final polishing time Tffl are respectively calculated as weighted moving averages. Iff 92307. ptd page 34 200305240 6. Scope of patent application 27. The method of the 20th scope of patent application further includes: measuring the control variable of the front substrate. 28. The method according to item 11 of the scope of patent application, wherein the sensitivity parameter α depends on one of the number of substrates to be processed and the number of substrates processed. 2 9. The method according to item 21 of the scope of patent application, wherein the sensitivity parameter / 3 depends on one of the number of substrates to be processed and the number of substrates already processed. 3 0. — A controller for controlling chemical mechanical polishing of a substrate, including: an input section for inputting at least one of a measurement value of a sensitivity parameter and a control variable; an output section for outputting At least one of an over-polishing time and an additional polishing time as an operational variable; and a calculation part for performing the following steps: obtaining a first sensitivity parameter that quantitatively describes the over-polishing time of the first material layer and The relationship between the control variable related to the first material layer; obtaining the second sensitivity parameter quantitatively illustrates the relationship between the control variable related to the second material layer and the control variable related to the second material layer of a previous substrate Relationship; calculating the over-polishing time of the first material layer from the linear model of the chemical mechanical polishing process, wherein the model includes at least a control variable, a first sensitivity parameter, and a second spirit related to the second material layer 92307.ptd Page 35 200305240 Sixth, the scope of patent application ϋ Sensitivity parameter The command value for the first material layer Λ '-the excessive polishing time of the material layer, the control variables related to the second material layer, and the previous substrate The control variable of the second material layer and the weighted average of the weighted movement of the excessive polishing time of the first material layer are 0. 31. For example, the controller of the 30th item in the patent scope is requested, wherein the control variable represents 1 insect reduction. Dish-shaped depressions, and at least one of the thickness of the material layer. 032. For example, the controller of the 30th patent scope, where each control variable f number represents a median value for multiple substrates of 0. 33. Such as The controller 5 of the 30th patent scope of the patent application, among which the sensitivity parameter depends on the number of processed substrates g and the One of the number of processed substrates g is 34, such as the controller y of the 30th scope of the patent, which is further used to obtain a third sensitivity parameter, which quantitatively explains the control variable and the additional of the final polishing step The relationship between the polishing time is 0 35. The controller y of the patent scope item 34 is further used to calculate the additional polishing time from the linear model. 36 The controller y calculates the over-polishing parameter time and the extra polishing time including: • Find the intermediate over-polishing time and the intermediate extra-polishing time, so that the intermediate point of the intermediate over-polishing time and the intermediate extra-polishing time corresponds to the compound of the middle point of the allowable Capricorn The deviation value is approximately the minimum value of 0-37. Take the smaller value under 92307. ptd page 36 200305240 VI. Calculated under the conditions of the patent application range: when compared to the previous values of the substrate, the intermediate excessive polishing time and the intermediate additional polishing time are changed in a different direction, and the intermediate excessive The polishing time and the intermediate additional polishing time produce a control variable value substantially equal to the command value. 3 8. —A chemical mechanical polishing controller for controlling a first metallization layer in a substrate, comprising: an input section for inputting at least one sensitivity parameter α, sensitivity parameter / ?, sensitivity parameter λ And the measured values of the control variables; the output section is used to output at least one of the excessive polishing time T π and the additional polishing time I'm as the operating variables; and the calculation section is used to implement: from the linear model The over-polishing time T @ for the first metallization layer is calculated. The linear model includes at least the following terms: E fir st Ep, first, ^ (T op Τρ, 〇ρ) Λ Ί '(E second-Ep, se_d) , Where TP, QP are the excessive polishing time of the previous substrate. 39. The controller according to item 38 of the scope of patent application, which further comprises at least one of a microprocessor, a microcontroller, a personal computer, and a communication line used to communicate with the facility management system. 40. The controller according to item 38 of the scope of patent application, wherein calculating T QP includes: calculating an intermediate over-polishing time rQP, which is necessary to obtain a desired value E target of the control variable, and Calculate T from the over-polishing time of the previous substrate T p, @ and the intermediate over-polishing time T% ρ as a weighted moving average. 41. The controller according to item 40 of the scope of patent application, wherein the weighted moving flat 92307.ptd Page 37 200305240 'VI. Patent Application Scope • The mean is a moving average weighted exponentially. 4 2. The controller according to item 38 of the scope of patent application, wherein the control variable represents an intermediate value of a plurality of substrates. 3 4 3. The controller according to item 38 of the scope of patent application, wherein the control variable represents one of the erosion of the first and second metallization layers, the dish-shaped depression, and the layer thickness. 4 4. The controller according to item 38 of the scope of patent application, which is further used to receive the measurement value of the control variable of the previously processed substrate, and use the measurement value of the control variable to calculate the over-polishing time T @. f 5. According to the controller of the 38th patent application range, when the intermediate over-polishing time exceeds the predetermined value range, it is used to indicate the loss of effectiveness of the linear model. 46. The controller according to item 46 of the scope of patent application, which is used to receive an empirical sensitivity parameter, which quantitatively describes the effect of the additional polishing time Tm on the control variable Ef] 1.st &. 47. The controller according to item 46 of the patent application scope, wherein the linear model further includes an item: / 5 (Tm—Tp, ffl), where Tp, ffl represents additional polishing time of the previous substrate, and wherein the excessive The polishing time T QP and the additional polishing time Tffl are calculated from a model including this term. |. If the controller of the scope of patent application item 47 is used, it is used to calculate the intermediate excessive polishing time T% p and the intermediate additional polishing time T * m before calculating the excessive polishing time T @ and the additional polishing time Tffl. 4 9. The method according to item 48 of the scope of patent application, wherein the intermediate over-polishing time and the intermediate additional polishing time are calculated under secondary conditions. 92307. ptd page 38 200305240 6. Under the scope of the patent application, select ΊΛΡ and ΊΛ to get the desired value Etarget, and at the same time make D, and T \ distance D: the deviation from the individual center points in the predetermined value range of T \ The sum of the values is minimized. 50. As the controller of the 48th item in the scope of patent application, ΊΛΡ and Τ \ are calculated under the secondary condition, under which, when Ep, first + r (EsecQnd — EP, se_d) is greater than the desired value Etarget T% p is equal to or less than the excessive polishing time of the previous substrate, and rm is equal to or greater than the additional polishing time of the previous substrate. 5 1. The controller according to item 48 of the scope of patent application, in which D ^ and T \ are calculated under the secondary conditions. Under this condition, when Ep, iirst + r (Esecond — EP, sec〇nd) is less than At the desired value Etarget, T * QP is equal to or less than the excessive polishing time of the previous substrate, and T \ is equal to or greater than the additional polishing time of the previous substrate. 5 2. The controller according to item 48 of the scope of patent application, wherein the excessive polishing time T ^ and the additional polishing time are respectively calculated as weighted moving averages. 5 3. An apparatus for chemical mechanical polishing of a substrate, the apparatus comprising: a polishing tool having at least one polishing platform; an endpoint detector that supplies an endpoint signal indicating an endpoint of the polishing; and a control After the endpoint signal has been supplied, the over-polishing time of the existing substrate having the first material layer to be processed is obtained in advance; 92307.ptd Page 39 200305240 6. Scope of patent application • The controller calculates the excessive polishing time based on the following: • At least one of the erosion of the first material layer of the previous substrate, dish-shaped depression, and layer thickness Or _ at least one of the erosion of the second material layer of the previous substrate, the dish-shaped depression, and the layer thickness, the erosion of the second material layer of the existing substrate, the dish-shaped depression, and at least one of the layer thickness One of them, and the sensitivity parameter obtained empirically, represent the inherent mechanism of the polishing process. @ 4. The device according to item 53 of the patent application scope, wherein the controller is operatively coupled to a facility management system. 55. The device according to item 53 of the patent application scope, further comprising a final polishing platform disposed downstream of the at least one polishing platform, wherein the controller is used to obtain a polishing time on the final polishing platform. 92307.ptd p. 40