WO2006077994A1 - Substrate polishing method and apparatus - Google Patents

Abstract

A polishing apparatus is provided for optimizing a polishing profile in consideration of even such parameters as the temperature on the surface of an object to be polished, and the thickness of a polishing pad, in addition to a polished amount. The polishing apparatus for polishing the object to be polished under control of a control unit CU has at least two pressing sections (E1, E4), and comprises a top ring (T) which can apply an arbitrary pressure to the object to be polished from each of the pressing sections (E1, E4), a measuring device IM for measuring a polished amount of the object to be polished (W), and a monitoring device SM for monitoring the object to be polished for a polishing condition. The control unit CU forces the polishing apparatus to polish the object to be polished (W) in accordance with a simulation program for setting processing pressures required to optimize a polishing profile of the object to be polished to the top ring based on the output of the measuring device IM and the output of the monitoring device SM.

Description

DESCRIPTION

SUBSTRATE POLISHING METHOD AND APPARATUS

TECHNICAL FIELD

The present invention relates to a substrate polishing apparatus for polishing materials to be polished, represented by a semiconductor substrate , which can suppress a degradation in yield rate due to non-uniformity of residual films on the surface of substrates , mainly caused by aging changes of consumable materials , extend the life time of such consumable materials to reduce an operation cost , and a polishing apparatus which embodies the method.

BACKGROUND ART In recent years , as semiconductor devices are

(increasinglymore miniaturized and complicated in element structure, the semiconductor devices tend to have increased asperities and larger steps on the surface . As a result , thin films are formed in smaller thicknesses on such steps , and open circuits can be developed due to disconnections of wires , and short circuits attributable to defective insulation between wiring layers , leading to a lower yield rate . In a planarization technology for solving such problems , chemical mechanical polishing (CMP ) has been employed for planarizing asperities on the surface formed in course of deposition of insulating films and wiring metal films , for example , during a semiconductor device manufacturing process on semiconductor substrates .

In the CMP , a substrate, which is an object to be polished. is pressed against a polishing pad made of unwoven fabric or the like , and the substrate and polishing pad are slid relative to each other with polishing slurry supplied therebetween to polish the substrate . It has been found that concentric or lattice grooves formed on the surface of the pad are effective for supplying a sufficient amount of polishing slurry deep into a central region of the substrate during the CMP-based polishing . Also , the CMP entails so-called pad conditioning for trimming the surface of the padwith a diamond disk or the like for purposes of removing polishing debris which can stick to the surface of the polishing pad.

In a CMP process for polishing wiring metal and insulating films laminated on a substrate into a flat finish, polishing conditions applied to amanufacturing line have previously optimized, so that the polishing is performed under the same conditions until polishingmembers reach a limit consumption levelunder the optimized conditions . However , as the polishing members are consumed, the shape of the surface after polishing wiring metal and insulating films on a substrate (called the "polishing profile" ) is varying over time in step with the consumption level of thepolishingmembers . Generally, the polishing members are replaced at a timing which is set before the aging changes affect the performance of devices .

With the miniaturization of semiconductor device , an increased number of wiring layers , and a higher processing speed in recent years , a higher degree of flatness is required for the surface profile , i . e . , polishing profile of wiring metals and insulating films after polishing . Specifically, allowed aging changes in the polishing profile is increasingly narrowed in more miniaturized devices and devices having a larger number of layers , resulting in a higher frequency of replacement of consumed polishing members . However , the consumablemembers of the CMP are so expensive that ahigherfrequencyof replacement due to consumptionwill largely affect the cost of devices . Generally, it is widely known that a polished amount Q can be predicted with a certain accuracy from a relationship QockpvΔt (where Q represents a polished amount ; k a coefficient determined by materials of a polishing pad, a polishing liquid, and a substrate , and the like ; p a processing pressure , v a moving speed, and Δt a polishing time) which is known as the Preston ' s empirical formula in the field of polishing, and the Preston ' s empirical formula is generally established in the CMP as well . However, in the CMP , the speed of polishing based on chemical actions is largely affected by a processing temperature, thus making it difficult in some cases to predict the polished amount with high accuracy only from the Preston ' s empirical formula. Also , thebehavior of polishing slurry within grooves in the surface of a polishing pad is based on hydrodynamics and is therefore a factor which is not considered in the Preston ' s empirical formula . Further, the Preston ' s empirical formula fails to cover such factors as insufficient dressing associatedwith a reduced cutting rate of a pad conditioner, and a reduced amount of removed polishing debris .

DISCLOSURE OF THE INVENTION The present invention has been proposed in view of the foregoing problems , and it is an object of the invention to automatically optimize aprocessingpressure using a simulatorbased on the Preston ' s empirical formula within a polishing apparatus , sufficiently monitor even those parameters which cannot be covered by the Preston ' s empirical formula to improve a correction accuracy, and accomplish a uniform polishing profile associated with increasing miniaturization of integrated circuits . It is another object of the present invention to correctly manage the state of consumable materials which have been conventionally replaced after a certain number of substrates had been processed to extend the life time of the consumable materials and reduce the operational cost . To achieve the above objects , the polishing apparatus according to the present invention comprises a top ring for holding an object to be polished such as a wafer while applying a pressure to the object to be polished against a polishing member in order to polish the object to be polished. The top ring can arbitrarily set a pressure to the object to be polished in each of concentrically partitioned areas , and can therefore control a pressing force on the object to be polished. Therefore, if the object to be polished is not polished into a flat shape , a pressing, force for a required polishedamount canbe additionallyapplied, for example , to aportion whichis not sufficientlypolished, thusmaking it possibletoprovide high polishing performance with high accuracy flatness .

The pressure within the area of the top ring is generally set to provide a flat surface for a wiring metal or an inter-layer insulating film formed on a polished object to be polished. Generally, this pressure is often set in accordance with engineer ¹ s empirical rules , so that several objects to be polished must be polished for adjustment before conditions are defined for polishing the surface of the object to be polished to ' be flat . Accordingly, the present invention utilizes a first simulation program which receives a pressure setting condition for each area in the top ring structured as described above to estimate a polishing profile for an object to be polished. It has been found that the result of a simulation performed by the first simulation program has merely 1 to 5 % of errors as compared with the actual profile resulting from the polishing . The present invention can eliminate wasteful objects to be polished which have been used at a pressure setting stage , can instantaneously predict a polishing profile through the simulation, and can accordingly reduce a time required for setting the pressure as well .

Since the first simulation program can simply update a polishing coefficient ( coefficient including the influences by a pad and a slurry) which can be found from the result of measurements of the shape of a residual film (or polished shape) at a relatively small number of measuring points to predict the thickness of the residual film after polishing at a large number of points other than themeasuringpoints , the simulationprogram canreadily correct the influence caused by changes in polishing members such as the slurry, pad and the like , and can predict a polishing profile under a polishing condition which is set after the correction . When the polishing coefficient is updated using the result of polishing performed in the vicinity of a polishing condition set value used in the first simulation program, errors can be reduced even to 1 to 3 % . When objects to be polished are sequentially polished on an actual semiconductor production line , there is not a large difference in the polishing condition set value among the sequential objects to be polished, so that the simulation can be performed with a higher accuracy . When there are a relatively small number of points at which a polished shape is measured, the polishing coefficient may be calculated using a curve which is smoothly interpolated by the measuring points . The present invention also provides a desired polishing profile by creating a film shape on a wafer surface in a desired thickness . For this purpose, in the present invention , a desired polishing time , an average polished amount , and the shape of a residual film ( a polished shape may be used instead) are entered to calculate a set pressure for each area to satisfy these conditions by a second simulation program. The first simulation program is incorporated into the second simulation program in the form of a module . The first simulation program calculates a predicted value for a polishing profile at a certain set pressure , and the second simulation program compares this predicted value with a desired polishing profile to calculate amodified value for the set pressure . When the second simulation program is used to repeatedly calculate a predicted value for the polishing profile and calculate a modified value for the set pressure , it is possible to calculate a set pressure which is closer to the desired polishing profile .

Here , the set polishing time may be treated as a reference value ( target value) , and the polishing may be terminated at the time the amount of residual film actually monitored by an end point system reaches a predetermined value . While an average polished amount has been simply stabilized in the past , the present invention can also control and stabilize the flatness after polishing or a desired shape of a residual film. Thus , in the present invention , after one test object to be polished is preferably processed to update the polishing coefficient , an optimized polishing condition is found by the second simulation program to provide a desired polishing time, average polished amount , and shape of residual film. While the object to be polished is polished under this optimized polishing condition , the polishing coefficient is updated as appropriate based on the degree of consumption of the polishing members to again optimize the polishing condition to stably provide the desired polishing time , average polished amount , and shape of residual film. Here , when a polishing condition under which an object to be polished was polished may be fed back for subsequent polishing, the quality of the polished object to be polished can be ensured with a high accuracy in consideration of the accuracy of the feedback control which is affected by the accuracy of the flatness of a residual film after polishing, and the polishing condition .

The preset invention can acquire data relating to the polished shape not only for a generated film which can be measured by an optical measuring device but also for, a metal film using a measuring device which can measure the metal film to conduct a feedback control , and is rich in general-purpose properties because it is not limited in applications of the CMP process . Also , data on thickness can be acquired by an arbitrarily selected means such as a measuring method using a measuring device which can make monitoring during polishing, a method of measuring a wafer which is transported to a measuring device after polishing, a method of transferring data measured by a measuring apparatus external to a CMP device to the CMP device and entering the data into the CMP device , and the like . Also , the foregoingmethods can be arbitrarily combined to use different methods for.acquiring thickness databefore polishing and after polishing , and the like , to facilitate the operation .

Further, in the present invention , the correction accuracy is improved by sufficiently monitoring those parameters which cannot be covered by the Preston ' s empirical formula, and the uniformization in the shape of polished wafers is realized, as required in step with increasing miniaturization of integrated circuits . For this purpose , the present invention controls the polishing operation in consideration of even the temperature on a polished surface of a wafer, the thickness of a pad, the depth of grooves in the pad, and the cutting rate value of a dresser .

Accordingly, the invention set forth in claim 1 of the present application provides a polishing apparatus for polishing an object to be polished under control of a control unit , which comprises : a top ringhavingat least two pressing sections andcapable of applying an arbitrary pressure to the object to be polished from each of the pressing sections ; a measuring device for measuring a polished amount of the object to be polished; and a monitoring device for monitoring the object to be polished for a polishing state , and is characterized in that : the control unit forces the polishing apparatus to polish the object to be polished in accordance with a simulation program for setting a processing pressure required to optimize a polishing profile of the object to be polished to the top ring based on the output of the measuring device and the output of the monitoring device .

The invention set forth in claim 2 is characterized in that the at least two pressing sections includes a plurality of concentric air bags , and a retainer ring surrounding the air bags , and the pressure of the retainer ring is kept at a value larger by 20 percent than an average value of the sum total of the pressures applied by the air bags .

The invention set forth in claim 3 is characterized in that when the output of the monitoring device indicates that an abrasion loss of the retainer ring falls below a threshold, the control unit instructs the polishing apparatus to stop polishing .

The invention set forth in claim 4 is characterized in that when the output of the monitoring device indicates that the temperature on the surface of the object to be polished exceeds

, a predetermined set temperature , the control unit stops using the simulation program or instructs the polishing apparatus to stop polishing, and when the output of the monitoring device indicates that the temperature on the surface falls below the set temperature , the control unit instructs the polishing apparatus to resume the polishing .

In the invention set forth in claim 5 , the polishing apparatus further comprises a polishing pad for polishing the object to be polished in a state in which the polishing pad is pressed by the top ring, and is characterized in that when the output of the monitoring device indicates that the thickness of the polishing pad falls below a threshold, the control unit stops using the simulation program or instructs the polishing apparatus to stop polishing .

The invention set forth in claim 6 is characterized in that the monitoring device comprises a laser displacement gage for measuring the thickness of the polishing pad. In the invention set forth in claim 7 , the polishing apparatus further comprises a polishing pad for polishing the object to be polished in a state in which the polishing pad is pressed by the top ring, and a dresser for conditioning the polishing pad, and is characterized in that when the output of the monitoring device indicates that a cutting rate of the dresser falls below a threshold, the control unit stops using the simulation program, or instructs the polishing apparatus to stop polishing.

The invention set forth in claim 8 is characterized in that the cutting rate is monitored using a torque of a motor for driving the dresser .

The invention set forth in claim 9 is characterized in that the control unit can adjust the amount of supplied slurry in accordance with the polishing state .

Generally, the polishing apparatus is provided with a touch panel for the operator to enter operational conditions , and when the control unit instructs the polishing apparatus to stop using the simulation program, this instruction is displayed on the touch panel . In response , the operator determines whether the polishing should be continued or stopped . Also , a setting can be previously made through manipulations on the touch panel to select a setting for stopping the polishing when the control unit generates an instruction of stopping the use of the simulation program. BRIEF DESCRIPTION OF THE DRAWINGS

Fig . 1 is a top plan view generally illustrating one embodiment of a polishing apparatus according to the present invention; Fig . 2 is a perspective view of the polishing apparatus in Fig . 1 ;

Fig . 3 is a diagram for describing some of components in the polishing apparatus of Fig . 1 ;

Fig . 4 is a diagram for describing some of components in the polishing apparatus of Fig . 1 ;

Fig . 5 is a cross-sectional view illustrating the structure of a top ring used in the polishing apparatus of Fig . 1 ;

Fig . 6 is a flow diagram for describing a procedure for collecting polishing rate distribution data in the polishing apparatus of Fig . 1 ;

Fig . 7 (A) is a diagram generally illustrating the configuration for detecting a change in the thickness of a polishing pad in the polishing apparatus of Fig . 1 using a laser displacement gage, and Fig . 7 (B) is a graph showing a change in the output of the laser displacement gage over time;

Fig .8 (A) is a table showing a comparison of measurements when the polishing method according to the present invention is used with those when not used, and Fig . 8 (B ) is a graph showing the result of the comparison;

Figs . 9 (A) to 9 ( C ) are diagrams showing the thickness of a film on a wafer before CMP (A) , the thickness of the film on the wafer after CMP (B) , and a polishing rate ( C) , respectively. when a polishing pad is new;

Figs . 10 (A) to 10 ( C) are diagrams showing the thickness of a film on a wafer before CMP (A) , the thickness of the film on the wafer after CMP (B) , and a polishing rate ( C ) , respectively, when the polishing pad has been consumed by 0.1 mm; and

Figs . 11 (A) to 11 (C) are diagrams showing the thickness of a film on a wafer before CMP (A) , the thickness of the film on the wafer after CMP (B) , and a polishing rate (C) , respectively, when the polishing pad has been consumed by 0.2 mm.

BEST MODE FOR CARRYING OUT THE INVENTION In the following, several embodiments of a polishing method and apparatus according to the present invention will be described in detail with reference to the accompanying drawings . First , one embodiment of the polishing apparatus according to the present invention will be described with reference to Fig . 1 which is a top plan view illustrating the layout and configuration of respective components in the polishing apparatus , and Fig. 2 which illustrates a perspective view of the polishing apparatus . In Figs . 1 and 2 , a transportation mechanism common to two polishing stations installed in areas A, B comprises separately installed linear transporters , each of which includes two linearly reciprocating stages that are transportation mechanisms dedicated to the two polishing stations , respectively. Specifically, the polishing apparatus illustrated in Figs . 1 and 2 comprises four loading/unloading stages 2 for carrying wafer cassettes 1 which stockmultiplewafers . A carrier robot 4 having two hands is disposed on a running mechanism 3 such that the hands can reach each wafer cassette 1 on the loading/unloading stages 2. The runningmechanism 3 is based on linear motors . By employing the running mechanism based on linear motors , high-speed and stable transportation can be ensured even if wafers are increased in diameter and weight . In the polishing apparatus illustrated in Fig . 1 , the loading/unloading stage 2 for carrying thewafer cassette 1 comprises an SMIF ( StandardManufacturing Interface) pod or FOUP (Front Opening Unified Pod) with a loading/unloading stage being attached external thereto . The SMIF and FOUP are closed vessels which receive wafer cassettes therein and cover them with partitions to keep an environment independent of the external space . When SMIF or FOUP is installed as the loading/unloading stage 2 of the polishing apparatus , the polishing apparatus is integrated with the wafer cassette byopening a shutter S arrangedon ahousingHof thepolishing apparatus and a shutter of the SMIF or FOUP .

Upon termination of a polishing step, the SMIF or FOUP is separated from the polishing apparatus by closing the shutter, and is automatically or manually transported to another processing step , so that it must keep the internal atmosphere clean . For this purpose , a clean air down flow is formed through a chemical filter above an area C through which wafers pass immediately before they return to the cassette . Also, since a linear motor is used to move the carrier robot 4 , dust can be restrained to keep the atmosphere of the area C more normal . Additionally, in order to keep wafers clean within the wafer cassette 1 , a clean box which contains a chemical filter and a fan to maintain a certain degree of cleanness by itself may be employed for a closed vessel such as SMIF and FOUP .

Two washing machines 5 , 6 are disposed on the opposite sides to the wafer cassettes 1 in symmetric arrangement about the running mechanism 3 of the carrier robot 4. Each of the washing machines 5 , 6 is disposed at a position which can be accessed by the hands of the carrier robot 4. Awafer station 50 , which comprises four semiconductor wafer seats 7 , 8 , 9 , 10 , is disposed at a position between the two washing machines 5 , 6 , which can be accessed by the carrier robot 4.

A barrier 14 is disposed for ranking the cleanness in an area D in which disposed are the washing machines 5 , 6 and seats 7 , 8 , 9 , 10 and in an area C in which disposed are the wafer cassettes

I and carrier robot 4. The barrier 14 is provided with a shutter

II in an opening for carrying a semiconductor wafer from one area to another. A carrier robot 20 is disposed at a position from which the carrier robot 20 can access the washing machine 5 and three seats 7 , 9 , 10 , and a carrier robot 21 is disposed at a position from which the carrier robot 21 can access the washing machine 6 and three seats 8 , 9 , 10.

A washing machine 22 is disposed to be adjacent to the washing machine 5 and at a position which can be accessed by the hand of the carrier robot 20. Also , a washing machine 23 is disposed to be adjacent to the washing machine 6 and at a position which can be assessed by the hand of the carrier robot 21. The washing machines 22 , 23 can wash both sides of wafers . All of these washing machines 5 , 6 , 22 , 23 , seats 7 , 8 , 9 , 10 of the wafer station 50 , and carrier robots 20 , 21 are disposed in the area D in which the air pressure is adjusted to be lower than the air pressure in the area C .

The polishing apparatus illustrated in Figs . 1 and 2 has a housing H which surrounds the respective devices , and the housing H is partitioned into a plurality of chambers ( including the areas C , D) by partitions 14 , 24A, 24B . The partitions 24A, 24B define two areas A, B , separated from the area D, which form two polishing chambers . Each of the two areas A, B comprises two polishing tables , and a top ring for holding a semiconductor wafer and polishing the semiconductor wafer while pressing the same against the polishing table . Specifically, polishing tables 34 , 36 are disposed in the area A, while polishing tables 35 , 37 are disposed in the area B . Also , a top ring 32 is disposed in the area A, while a top ring 33 is disposed in the area B . Further disposed in the area A are an abrasive liquid nozzle 40 for supplying a polishing abrasive liquid to the polishing table 34 , and a mechanical dresser 38 for dressing the polishing table 34 , while disposed further in the area B are an abrasive liquid nozzle 41 for supplying a polishing abrasive liquid to the polishing table 35 , and a mechanical dresser 39 for dressing the polishing table 35. In addition , a dresser 48 is disposed for dressing the polishing table 36, in the area A, while a dresser 49 is disposed for dressing the polishing table 37 in the area B .

The polishing tables 34 , 35 comprise atomizers 44 , 45 , whichare dressers basedon liquidpressure, inaddition tomechanical dressers 38 , 39. The atomizer mixes a liquid ( for example , pure water) with a gas ( for example , nitrogen) into a sprayed fluidmixture which is blown from a plurality of nozzles onto a polishing surface to wash out polishing grounds and slurry grains deposited or clogged on the polishing surface . With the cleaning of the polishing surface by the fluid pressure of the atomizer, and the dressing of the polishing surface by the dressers 38 , 39 which involve mechanical contacts , more desirable dressing, i . e . , restoration of the polishing surface can be achieved.

Fig .3 is a diagram illustrating the relationship between the top ring 32 and the polishing tables 34 , 36. As appreciated, a similar relationship is established between the top ring 33 and the polishing tables 35 , 37. As illustrated in Fig . 3 , the top ring 32 is suspended from a top ring head 31 by a rotatable top ring driving shaft 91. The top ring head 31 is supported by a positionable rocking shaft 92 , and the top ring 32 is made accessible to the polishing tables 34 , 36. The dresser 38 is suspended from a dresser head 94 by a rotatable dresser driving shaft 93. The dresser head 94 is supported by a positionable rocking shaft 95 , and the dresser 38 can move between a standby position and a dresser position on the polishing table 34. The dresser head (rocking arm) ,97 is supported by a positionable rocking shaft 98 , and the dresser 48 can move between a standby position and a dresser position on the polishing table 36.

The dresser 38 has an elongated shape longer than the diameter of the table 36 , and the dresser head 97 rocks about the rocking shaft 98. The dresser 48 is suspended from the dresser head 97 by a dresser fixing mechanism 96 such that the dresser fixing mechanism 96 and the dresser 48 , opposite to the dresser head 97 fromthe rocking shaft 98 , make pivotalmovements , therebypermitting the dresser 48 to dress on the polishing table 36 through motions just like those of wipers of a car, without revolutions . Scroll type polishing tables can be used for the polishing tables 36 , 37.

Turning back to Fig . 1 , an inverter 28 is installed at a position to which the hands of the carrier robot 20 are accessible for inverting a semiconductor wafer within the area A separated from the area D by the partition 24A. Similarly, an inverter 28 ' is installed at a position to which the hands of the carrier robot 21 are accessible for inverting a semiconductor wafer within the area B separated from the area D by the partition 24B . The partitions 24A, 24B which separate the areas A, B from the area D, are formed with opening for transporting a semiconductor wafer therethrough, and shutters 25 , 26 dedicated to the inverters 28 , 28 ' are disposed on the respective openings .

Each of the inverters 28 , 28 ' comprises a chucking mechanism for chucking a semiconductorwafer, an invertingmechanism for inverting a semiconductorwafer upside down, and a wafer presence detecting sensor for confirming whether a semiconductor wafer is chucked by the chuckingmechanism. A semiconductorwafer is carried to the inverter 28 by the carrier robot 20 , while a semiconductor wafer is carried to the inverter 28 ' by the carrier robot 21.

In the area Awhich defines one polishing chamber , a linear transporter 27A is installed for providing a transportation mechanism for transporting a semiconductor wafer between the inverter 28 and the top ring 32. In the area B which defines the other polishing chamber , a linear transporter 27B is installed for providing a transportation mechanism for transporting a semiconductor wafer between the inverter 28 ' and the top ring 33. The linear transporters 27A, 27B comprise two stages which linearly reciprocate , and a semiconductor wafer is passed between the linear transporter and the top ring or inverter through a wafer tray. The right-hand area of Fig . 3 illustrates the positional relationship among the linear transporter 27A, lifter 29 , and pusher 30. A similar positional relationship to that illustrated in Fig . 3 is established among the linear transporter 27B , lifter 29 ' , and pusher 30 " . Accordingly, the following description will be made only on the linear transporter 27A, lifter 29 , and pusher 30. As illustrated in Fig . 3 , the lifter 29 and pusher 30 are disposed below the linear transporter 27A. The inverter 28 is disposed above the linear transporter 27A. The top ring 32 , when rocking, can be positioned above the pusher 30 and linear transporter 27A. Fig . 4 is a diagram for describing how a semiconductor wafer is passed between the linear transporter and the inverter and between the linear transporter and the top ring . As illustrated in Fig . 4 , a semiconductor wafer 101 before polishing, carried to the inverter 28 by the carrier robot 20 , is inverted by the inverter 28. As the lifter 29 moves up, a wafer tray 925 on a loading stage ,901 is transferred onto the lifter 29. As the lifter 29 further moves up, the semiconductorwafer 101 is transferredfromthe inverter

28 to the wafer tray 925 on the lifter 29. Subsequently, the lifter

29 moves down, and the semiconductor wafer 101 is placed on the loading stage 901 together with the wafer tray 925. The wafer tray

925 and semiconductor wafer 101 are transported above the pusher

30 by a linear movement of the loading stage 901. In this event , an unloading stage 902 receives a polished semiconductor wafer 101 from the top ring 32 through the wafer tray 925 , and moves toward the lifter 29. The loading stage 901 and unloading stage 902 pass each other halfway in their movements . When the loading stage 901 reaches above the pusher 30 , the top ring 32 has previously rocked to the position indicated in Fig . 4. Next , the pusher 30 moves up , and further moves up , after it has received the wafer tray 925 and semiconductor wafer 101 from the loading stage 901 , to reach the top ring 32 to which the semiconductor wafer 101 alone is transferred. The wafer 101 , which has been transferred to the top ring

32 , is sucked by a vacuum sucking mechanism of the top ring 32 , transported to the polishing table 34 while it is still sucked. Next , the wafer 101 is polished by a polishing surface which has a polishing pad, a grindstone , or the like mounted on the polishing table 34. The second polishing table 36 is disposed at a location to which the top ring 32 can access . After the waferhas been polished on the first polishing table 34 in this way, the wafer is again polished on the second polishing table 36. However, depending on the type of a film formed on the semiconductor wafer, the semiconductor wafer may be first polished on the second polishing .table 36 and then polished on the first polishing table 34.

The polished wafer 101 is returned to the inverter 28 through the opposite path to the aforementioned. The wafer, which has returned to the inverter 28 , is rinsed with pure water or washing chemical liquid from arinse nozzle . Also , the wafer sucking surface of the top ring 32 , from which the wafer has been removed, is washed with pure water or chemical liquid from a top ring washing nozzle .

Now, a general description will be given of processing steps performed in the polishing apparatus illustrated in Figs . 1 to 4. In two-stage washing , two-cassette parallel processing , one wafer follows a path which passes wafer cassette ( CSl ) → carrier robot 4 → seat 7 of the wafer station 50 → carrier robot 20 → inverter 28 → loading stage 901 of the linear transporter 27A → top ring 32 → polishing table 34 → polishing table 36 ( as required) → unloading stage 902 of linear transporter 27A → inverter 28 —> carrier robot 20 → washing machine 22 → carrier robot 20 → washing machine 5 → carrier robot 4t → wafer cassette (CSl ) . The other wafer in turn follows a path which passes wafer cassette (CS2 ) -» carrier robot 4 → seat 8 of the wafer station 50 → carrier robot 21 → inverter 28 ' → loading stage 901 of linear transporter 27B → top ring 33 → polishing table 35 → top ring 33 → unloading stage 902 of linear transporter 27B → inverter 28 " → carrier robot 21 → washing machine 23 → carrier robot 21 → washing machine 6 → carrier robot 4 → and wafer cassette (CS2 ) .

In three-stagewashing, two-cassette parallel processing, one wafer follows a path which passes wafer cassette ( CSl ) → carrier robot 4 —* seat 7 of wafer station 50 → carrier robot 21 → washing machine 6 —> carrier robot 21 → seat 9 of wafer station 50 → carrier robot 20 → inverter 28 → loading stage 901 of linear transporter 27A —> top rig 32 → polishing table 34 → polishing table 36 ( as required) —> unloading stage 902 of the linear transporter 27A → inverter 28 → carrier robot 20 —* washing machine 22 → carrier robot 20 —* seat 10 of wafer station 50 → carrier robot 20 → washing machine 5 → carrier robot 4 → wafer cassette ( CSl ) . The other wafer in turn follows a path which passes wafer cassette (CS2 ) → carrier robot 4 → seat 8 of wafer station 50 → carrier robot 21 → inverter 28 ' → loading stage 901 of linear transporter 27B → top ring 33 → polishing table 35 → polishing table 37 ( as required) → unloading stage 902 of linear transporter 27B → inverter 28 ' → carrier robot 21 → washing machine 23 → carrier robot 21 → washing machine 6 → carrier robot 21 → seat 9' of the wafer station 50 → carrier robot 20 → washing machine 5 → carrier robot 4 → wafer cassette (CS2 ) .

Further, in the three-stage washing series processing, a wafer follows a path which passes wafer cassette (CSl ) → carrier robot 4 → seat 7 of wafer station 50 → carrier robot 20 → inverter 28 → loading stage 901 of linear transporter 27A → top ring 32 → polishing table 34 → polishing table 36 (as required) → unloading stage 902 of linear transporter 27A → inverter 28 → carrier robot 20 , washing machine 22 —> carrier robot 20 → seat 10 of wafer station 50 → inverter 28 " → loading stage 901 of linear transporter 27B → polishing table 35 → polishing table 37 ( as required) → unloading stage 902 of linear transporter 27B → top ring 33 → inverter 28 ' → carrier robot 21 —* washing machine 23 → carrier robot 21 → washing machine 6 → carrier robot 21 → seat 9 of wafer station 50 → carrier robot 20 → washing machine 5 → carrier robot 4 → .wafer cassette ( SCl ) .

According to the polishing apparatus illustrated in Figs . 1 to 4 , since thepolishingapparatus comprises the lineartransporter having at least two stages ( seats ) , which linearly reciprocate , as a transportation mechanism dedicated to each polishing station, the polishing apparatus can reduce a time required to transfer an object to be polished between the inverter and the top ring, and can increase the number of objects to be polished which can be processed per unit time . Also , when an object to be polished is transferredbetween a stage of the linear transporterandthe inverter, the object to be polished is transferred between the wafer tray and the inverter, and when the object to be polished is transferred between a stage of the linear transporter and the top ring, the object to be polished is transferred between the wafer tray and the top ring, so that the wafer tray can absorb impacts during the transfer, thus making it possible not only to increase the speed at which the object to be polished can be transferred, but also to improve the throughput, of the object to be polished. Also , since the transfer and placement of the wafer from the inverter to the top ring are performed through the tray which is removably held on each stage of the linear transporter, it is possible to eliminate a transfer of the wafer, for example , between the lifter and the linear transporter , andbetween the linear transporter andthe pusher, thus preventing damages possibly caused by produced dust and failure in holding the wafer .

Further, since the polishing apparatus has a plurality of trays classified into two groups , i . e . , a group dedicated to loading for holding objects to be polished before polishing, and a group dedicated to unloading for holding polished objects , a wafer before polishing is passed from the tray dedicated to loading, rather than from the pusher, to the top ring, while a polished wafer is passed from the top ring to the tray dedicated to unloading, rather than to the pusher . Thus , the loading of a wafer to the top ring is performed using a jig or member different from that which is used for the unloading of awafer from the top ring, making it possible to solve a problem that a polishing liquid or the like sticking to a polished wafer sticks to and solidifies on a wafer supporting member common to the loading and unloading, and scratches or sticks to a wafer before polishing.

An inline monitor IM is installed at an appropriate location in the area C of the polishing apparatus described above . such that a polished andwashed wafer is carried to the inline monitor IM by a carrier robot for measuring the thickness and profile of the wafer . The inline monitor IM can also measure a wafer before polishing, and the difference in thickness before polishing and after polishing can be regarded equal to a polished amount . Thus , the inline monitor IM acts as a thickness measuring device . Actually, the inline monitor IM is disposed above the robot 3. Further, the polishing apparatus comprises a state monitor SM for monitoring parameters representative of the operating state of the polishing apparatus such as the temperature on the polishing surface , the thickness of the polishing pad, the cutting rate of the dresser, and an abrasion loss of the retainer ring . The operation of the overall polishing apparatus is controlled by a control unit CU . The control unit CU stores a simulation program, later described, and a control flow program for measuring an arbitrary value among the temperature on the polishing surface , the thickness of the pad, the depth of the grooves in the pad, the cutting rate value of the dresser, and the abrasion loss of the retainer ring in the top ring to optimize the polishing . The control unit CU may be contained within the polishing apparatus as illustrated in Fig . 1 or may be separated from the polishing apparatus . The state monitor SM, inline monitor IM, and control unit CU are omitted in Fig. 2.

It is known from the Preston ' s formula that a pressing force for pressing the surface of a wafer against a polishing pad is generally proportional to a polished amount . However, an appropriate pressing force must be found by modeling a top ring which has a complicated structure , and taking into consideration the non-linearity of the polishing pad which is made of an elastic material , a large deformation of the wafer which is a thin plate , particularly, stress concentration which conspicuously appears on the end face of the wafer . Thus , difficulties would be encountered in analytically finding amathematical solution . On the other hand, the use of a finite element method or a boundary element method to find the pressing force involves dividing an object into a large number of elements , requiring an extremely large amount of calculations , a long calculation time , and high calculation capabilities . In addition , for achieving appropriate results , the operator is required to have expertism on numerical analysis , so that it is virtually infeasible , from viewpoints of the cost and practice to reference the mathematically derived pressing force in making a simple adjustment in the field, and incorporate the same in a CMP apparatus for utilization . Bearing the foregoing discussion in mind, the top ring in the polishing apparatus in the configuration described above is implemented by a profile control type one . The profile control type top ring, herein referred to , is a generic name for a top ring which has a plurality of pressing sections . Specifically, the profile control type top ringmaybe onehaving apluralityof pressing sections comprised of air bags or water bags concentrically partitioned by a plurality of membranes , one having a plurality of sections which directly press the back surface of a wafer with an air pressure by applying a pressure to partitioned air chambers , one having a section which generates a pressure with the aid of a spring, one having a local pressing section by disposing one or a plurality of piezo-electric elements , or a combination of those .

In the following, a pressing section will be described with reference to a top ring which has a plurality of concentrically partitioned air bags . As illustrated in Fig . 5 , the top ring comprises a plurality of concentric air bags , and adjusts a pressure applied from each air bag to an associated area of a wafer . In the following, the side of the wafer facing the air bags is called the "wafer back surface , " and the side facing the polishing pad the "wafer surface . " Fig . 5 illustrates a cross-sectional view taken along a plane including the axis of rotation of the top ring used in the polishing apparatus of the present invention , where the top ring T has a central discoidal air bag El , a troidal air bag E2 surrounding the air bag El , a troidal air bag E3 surrounding the air bag E2 , a troidal air bag E4 surrounding the air bag E3 , and a troidal retainer ring E5 surrounding the air bag E4. As illustrated, the retainer ring E5 is designed such that it can come into contact with the pad, and a wafer W carried on a polishing table is fitted in a space defined by the retainer ring E5 , and applied with pressures from the respective air bags El - E4.

As will be appreciated, the number of air bags which make up the top ring T is not limited to four, but may be increased or decreased in accordance with the size of a wafer . Also , though not shown in Fig . 5 , an air pressure feeder is disposed at an appropriate location of the top ring T for each air bag in order to adjust the pressure applied to the back surface of the wafer W by the associated air bag El - E4. Also , a pressure applied to the retainer ring E5 may be controlled by an air bag placed on the retainer ring E5 in a manner similar to the other air bags , or may be controlled by directly transmitting a pressure from a shaft which supports the top ring T . In the present invention , a combination of pressures applied by the respective air bags El - E4 and retainer ring E5 to the back surface of the wafer W and the polishing pad around the wafer W, and a resultant distribution of pressing forces on the surface of the wafer W have been previously stored in a memory of the control unit CU of the polishing apparatus . Preferably, the pressing force of the retainer ring E5 is set to 20 % or more of an average value of the sum total of pressing forces applied by the air bags El - E4 in order to prevent the wafer from slipping out . By using the structure as described above , assuming that a practical pressure setting range for the pressures applied from the air bags to the back surface of the wafer and the pressure applied from the retainer ring to the polishing pad (hereinafter called the "back surface pressure" ) is from 100 to 500 hPa, the air pressure is in a range of ±200 hPa, and the pressing force distribution on the surface of the wafer W can be regarded to be substantially linear ( i . e . , the principle of superposition is substantially established) , the pressing force distribution on the surface of the wafer resulting from a desired pressure applied by each air bag to an associated area on the back surface of the wafer can be found in a back surface pressure setting range of ±200 hPa by synthesizing pressing force distributions on the surface of the wafer by a combination of three different pressures of 100 hPa, 300 hPa, and 500 hPa applied to the back surface . In other words , by dividing set pressures on the back surface in a range in which a change in the surface pressing force can be regarded as substantially linear (the principle of superposition is established) , preparingpreviouslycalculateddata on a pressing force distribution on the surface of the wafer for a plurality of cases , and synthesizing data appropriately selected fromtheprepareddata, thepressing force distribution on the surface of the wafer corresponding to an arbitrary set pressure on the back surface of the wafer is found without complicated calculations based on the finite element method or the like . By storing a procedure for finding the pressing force distribution on the surface of the wafer in a computer, a simulation tool can be created for finding a pressing force distribution on the surface of a wafer for a set pressure on the back surface of the wafer .

Once the pressing force distribution on the surface of the wafer is found in this way, a predicted polishing profile can be foundfor thewaferbymultiplying this pressing force distribution by data on a distribution of a polishing coefficient on the wafer surface , previously found for the wafer to be polished. It is known from the aforementioned Preston ' s empirical formula that the amount Q of polished wafer is generally proportional to the product of the pressure applied to the wafer by each air bag, i . e . , the pressing force P , a moving speed v on the contact plane , and a polishing time Δt . While the moving speed ( i . e . , a relative speed of the wafer surface to the polishing pad) v of the contact plane on the wafer surface differs from one location to another on the wafer surface , and the polishing time Δt also differs depending on polishing conditions , the polishing coefficient corresponds to kv if the polishing rate per unit pressure is defined to be the polishing coefficient . When a distribution of a value corresponding to kv in the Preston ' s formula has been found for the wafer surface , the polished amount Q on the wafer surface , and a distribution of the polished amount Q per unit time , i . e . , the polishing rate Q/Δt can be found from the pressing force P . Since the amount of polished wafer (polishing rate ) can be found by such a simple calculation , the result of the calculation by the simulation tool can be referenced for a simple adjustment in the field, and incorporated in a CMP device for utilization .

Fig . 6 illustrates an exemplary procedure for finding data on a distribution of the polishing coefficient on the surface of a wafer . First , at step Sl , the shape of a film deposited on a certain wafer is previously measured . Next , at step S2 , the measured wafer is actually polished under a particular set pressure condition for a particular polishing time . In this event , at step S3 , a distribution of a pressing force on the surface of the wafer under this pressure condition is calculated using the simulation tool . The shape of the film on the surface of the wafer thus polished is again measured, and a distribution of the amount of polishing on the surface of the wafer is calculated from the difference in the shape before polishing and after polishing ( step S4 ) . Next , at step S5 , the calculated distribution of the polished amount is divided by the polishing time and the calculated distribution of the pressing force to find a distribution of the polishing rate per unit pressure at each point on the surface of the wafer, i . e . , a distribution of the polishing coefficient on the surface of the wafer. Here , instead of dividing by the polishing time , a distribution of the polished amount may be found per unit pressure . Alternatively, an initial conditionof the polishingpad, a situation after it has been used for a certain time , and a distribution of the polishing coefficient near a use limit may be previously calculated and stored within the control unit CU as data on aging changes of the polishing coefficient .

As has been so far described, the present invention is not limited to the profile control type top ring using air bags , but it is apparent that only if a force acting from the back surface of the wafer is found, the profile can be predicted by calculating a distribution of a pressing force on the surface of the wafer based on the acting force . Therefore, a top ring to which the present invention can be applied may be made up of respective pressing sections which comprise liquid bags that can accept a pressurized liquid therein , partitioned air chambers that directly press a wafer with a pressurized gas , resilient bodies that generate pressures using, for example , springs , piezo-electric elements that press a wafer, or a combination of those options . In the present invention, the simulation tool as described

, above is used to configure the top ring such that a polishing pressure can be set for each area, estimate a pressure which must be set for each area to accomplish a target polishing profile, and feed a calculatedpressurevalue back to awaferwhich is to be subsequently polished. In this way, even if the polishing profile varies over time as the polishing member consumes more and more , the variations can be corrected as appropriate to stably ensure a desired polishing profile .

To realize the foregoing , the present invention executes the following control flow:

1. A wafer is polished under an arbitrary polishing condition .

2. A distribution of the thickness of a wiring metal or an insulating film is measured on the polished wafer . This measurement can be made with a thickness measuring device contained in the polishing apparatus or a measuring apparatus external to the polishing apparatus , and measured data may be fetched online , or measured data recorded on another storage mediummay be retrieved. The measurement should be made at at least one location within each area.

3. Based on the result of the measurements , a polishing pressure condition is calculatedin order to create a target polishing profile . This step is performed in the following procedure :

3-1 ) A target polishing profile is set . For example , a plurality of arbitrary points at which a polished amount should be managed are specified on the surface of the wafer, and the polished amount QT is set at each of the specified points , or the polishing rate QτΔt⁼Qτ/Δt is set at each point . The processing can be carried out by any method. Here , a description will be given of a method of setting the polished amount .

3-2 ) The polished amount Qpoli is calculated for each of the areas of the actually polished wafer . The calculation of the polished amount requires data on an initial thickness of the wafer before polishing , and the initial thickness is measured using one of the measuring device contained in the polishing apparatus and the measuring device installed external to the polishing apparatus . The initial thickness data may be fetched by any of the methods described in Step 2.

3-3 ) The polished amount calculated at each point is divided by a pressure P applied to an area which includes the point to calculate the polished amount per unit contact pressure QpoliΔp=^Qpoli/P -

3-4 ) The target polished amount QT at the point closest to the point at which the distribution was measured at step 2 is extracted. Alternatively, the target polished amount QT may be approximated from two locations near the measuring location in a linear fashion .

3-5 ) At each point , a difference Qτ~Qpoli ^is calculated between the target polished amount QT set at 3-1 and the polished amount Qpoli calculatedat 3-2 , and the polished amount corresponding to the difference is divided by the polished amount per unit contact pressure calculated at 3-3 to calculate a correction polishing pressure (Qτ-Qpoli) /QpoliΔp -

3-6 ) The correction polishing pressure calculated at

3-5 is added to the pressure which was set upon polishing to find a pressure value Pinput - When a plurality of measuring points are included in an area, pressure values calculated at the plurality of points are averaged, and the average is set to the pressure value l?input of the area.

3-7 ) The pressure value Pinput calculated at 3-6 is entered into the simulation tool according to the present invention to estimate the polished amount at each of the points specified at 3-1 to find a estimate of the polished amount Qest -

3-8 ) The difference Qτ~Qest ^is calculated between the estimate of the polished amount Qest and the target polished amount QT .

3-9 ) The polished amount Qest calculated at 3-7 is divided by the pressure value Pinput to calculate a polished amount QestΔp ( =Qest/Pinput ) per unit contact pressure . 3-10 ) The difference Qτ~Qest calculatedat 3-8 is divided by the polished amount QestΔp P^er unit contact pressure to find a correction pressure value (Qτ-Qest ) /QestΔp which is then added to the pressure value Pinput - The calculated pressure values at points within the area are averaged, and the resultant average is defined to be a recommended pressure value Poutput iⁿ each area .

3-11 ) The recommendedpressure value Poutput calculated at 3-10 is again entered into the simulation tool . If the difference between the estimate of the polished amount at each point and the target polished amount falls within a previously set arbitrary allowable range , this recommended pressure value Poutout is applied

( fed back) to wafers which are to be actually polished from then on . If the difference falls out of the allowable range , steps 3- 7 to 3-10 are repeated until the difference falls within the allowable range to find the recommended pressure value .

The period of the feedback may be freely set , and an exemplary method of setting the period may involve measuring all wafers and feeding the recommended pressure, value back to wafers which are to be subsequently polished, or not conducting the feedback when the polishing member is not so consumed because of small variations in the polishing profile , and conducting the feedback when the polishing member has been much consumed. Further, the period set by the latter method may also be measured every arbitrary number of wafers , and a polishing condition fed back immediately before the measurement is continuously applied from the time the measurement is once made to the time the wafer is next measured. As the polishing member is more consumed, the period can be set shorter . Alternatively, for setting the polishing rate , each polished amount may be divided, by the polishing time at the aforementioned step 3.

Further, instead of correcting the polishing coefficient due to the influence of the edge shape , which has been made for predicting the polishing profile, the pressure on the back surface resulting from the measurement of the edge shape can be corrected after the calculation of the recommended pressure value to correct the edge polishing profile , restraining variations in polishing of an outer peripheral region of a wafer due to the edge shape . For example, for an oxide film on a wafer, a recommended pressure value for the retainer ring (E5 ) may be multiplied by a pressure correction coefficient in accordance with the magnitude of roll-off (Corrective Pressure Value for Retainer Ring = Pressure Correction Coefficient x Recommended Pressure Value for Retainer Ring) . Here , the pressure correction coefficient is createdby actuallypolishing awaferhaving, for example , apreviouslyknown roll-offwhilevarying the retainer ring pressure . Alternatively, the finite element method may be relied on to calculate the relationship between the pressingpressure andthemagnitude of roll-off to create thepressure correction coefficient .

Since the magnitude of roll-off varies from minute to another as the polishing is advanced, the magnitude of roll-off can be measured during polishing by a measuring device associated with the polishing apparatus to correct the pressure duringpolishing . Alternatively, the pressure can also be corrected by creating the pressure correction coefficient in consideration of the polishing time without measuring the magnitude of roll-off during polishing .

For the shape at an end of a metal film on a wafer, the correction can be made in a similar method to the oxide film roll-off correcting method . The method of correcting the edge shape using the pressure correction coefficient can also be applied when the recommended pressure value is not calculated. The polishing apparatus illustrated in Fig . 1 can be applied to a variety of objects to be polished by exchanging the top ring . When the top ring is exchanged in order to change an object to be polished, it is necessary to change a set of pressing force distributions on the surface of an object to be polished which have been previously calculated in conformity to the shape of the top ring . In this regard, the result of calculations of separately and previously calculated pressing force distributions may be set , or parameters such as the number of air bags of the top ring, an available pressure range and the like may be entered when the polishing apparatus is initially activated, and a plurality of pressing force distributions on the surface of an object to be polished, corresponding to the enteredparameters , maybe calculated within the polishing apparatus and stored in the control unit . In this way, in the polishing apparatus of Fig .1 , a recipe can be created not only to polish a wafer to be flat but also to polish a wafer into a particular shape . Even when a film surface shape of a wafer before polishing is not flat , a recipe can be created to make the shape of the residual film flat after polishing in consideration of the original shape . Also , the polishing condition can be optimized without relying on empirical rules of engineers as before , but an optimal condition can be calculated to polish into a desired polishing profile . As compared with the prior art which sets a polishing condition after a plurality of test wafers have been polished, efforts , time, and cost can be reduced.

In the foregoing description , the simulation program has used two variables which are the thicknesses of an initial wafer andthepolishedwaferandthepressingforce of the topring . Further , in the present invention , the accuracy for the correction is improved by sufficiently monitoring those parameters which cannot be covered by the Preston ' s empirical formula, and the temperature on the polishing surface , the thickness of the pad, the depth of the grooves in the pad, the cutting rate value of the dresser, and the amount of worn retainer ring in the top ring are also reflected to the polishing in order to accomplish uniform shapes , resulting from the polishing , in association with further miniaturization of integrated circuits .

To implement the foregoing, the state monitor SM (Fig . 1 ) in the polishing apparatus according to the present invention performs the following operations , and supplies resultant outputs to the control unit CU to further optimize the polishing using parameters which are not considered by the , simulation program. ( 1 ) In regard to the temperature on the polishing surface , a temperature range in which the polishing may be continued is set , and the temperature on the polishing surface is monitored by the state monitor SM. This can be implemented by providing the state monitor SM,. for example , with a radiation temperature . As a result of the monitoring, when the state monitor SM detects that the temperature on the polishing surface exceeds an upper limit or a lower limit of the set temperature range , the control unit CU stops the polishing and cools down the polishing surface . The polishing surface is cooled down in the following manner . A flow path is provided within the polishing table for communicating a cooling medium such as water therethrough. As a polishing stop signal is output from the control unit , the flow rate of the cooling medium is increased or the temperature of the cooling medium itself is reduced. Here , while the flow rate or temperature of the cooling medium is manipulated on the basis of the stop signal from the control unit , the flow rate and temperature of the cooling medium may be manipulated in accordance with the output of the state monitor SM, i . e . , a change in the temperature on the polishing surface . Subsequently, when the state monitor SM detects that the temperature on the polishing surface falls within the temperature range , the control unit CU resumes the polishing . In this event , the simulation program may be paused in a period in which the polishing is stopped.

( 2 ) The state monitor SM also monitors the thickness of the polishing pad or the depth of the grooves in the polishing pad on the polishing table ( described in greater detail in connection with Fig . 7 ) . Each time the state monitor SM detects that the thickness of the polishing pad or the depth of the grooves in the polishing pad is reduced by 0.1 mm, a monitoring wafer is polished instead of the wafer which has been so far polished, and the state monitor SM modifies a default value of the simulation application from the result of the polishing to optimize a pressure balance of the retainer ring and air bags in the top ring for a wafer which is to be next polished. When the state monitor SM detects that the thickness of the polishing pad or the depth of the grooves falls belowapredetermined thresholdwhile thewaferis thus beingpolished, the control unit CU stops the polishing . In response , the operator replaces the polishing pad . While the state monitor SM comprises a laser displacement gage so that the thickness of the polishing pad can be monitored by directly monitoring the surface of the polishing pad by the laser displacement gage or by measuring the distance to a member which comes into contact with the polishing pad by the laser displacement gage , the present invention is not so limited.

( 3 ) In order to prevent insufficient dressing of the dresser and a reduction in the amount of removed polishing debris , the state monitor SM monitors the dresser for the cutting rate when the pad is conditioned. When the state monitor SM detects that the cutting rate falls below a predetermined threshold, the control unit CU stops the polishing, or extends a conditioning time for the dresser, i . e . , a time for which the polishing pad is cut . In this way, since the polishing pad is uniformly cut away at all times , the polishing can be performed with a high accuracy. Variations in the cutting rate can be detected by monitoring the torque of a motor used by the dresser for the conditioning .

( 4 ) Further, the state monitor SM can monitor the retainer ring in the top ring for an abrasion loss . Then, the control unit CU instructs the polishing apparatus to stop the polishing when the state monitor SM detects the abrasion loss of the retainer ring falls below a certain threshold.

When a desired result cannot be achieved even if the polishing is performed in consideration of those parameters which cannot be covered by the Preston ' s basic formula, the amount of supplied slurry is preferably adjusted. The control flow from the foregoing ( 1 ) to ( 4 ) is stored in the control unit CU as a program. Fig . 7 (A) generally illustrates an exemplary configuration for measuring a relative change in the positions of the mechanical dressers 38 , 39 (Fig . 1 ) by the laser displacement gage associatedwith the state monitor SM for detecting the thickness of the polishing pad. As illustrated, a bar member 1001 is attached to an appropriate location of the driving shaft 93 of each dresser . The bar member 1001 is formed of a material which can reflect laser light , or has a film formed on the surface thereof and made of a material which can reflect laser light . A laser displacement gage 1002 is attached by an appropriate attaching means at a position at which the laser displacement gage 1002 can receive laser light which is irradiated to the bar member 1001 and reflected from the bar member 1001. In this way, as the thickness of the polishing pad is reduced with the advancement of the conditioning, the laser displacement gage 1002 outputs a signal corresponding to a change in the distance between the barmember 1001 and the laser displacement

,gage 1002 , i . e . , a reduction in the thickness of the polishing pad.

Fig . 7 (B) shows the relationship between a conditioning time and a reduction in the thickness of the polishing pad, derived by making use of the output from the laser displacement gage 1003. It can be understood from this graph that the thickness of the polishing pad substantially linearly reduces as the conditioning advances . By utilizing this relationship, a temporal changing rate of the thickness of the polishing pad, i . e . , the cutting rate of the dresser can be found. When the polishing apparatus as described above was used to actually polish wafers , the following results were obtained. For reference, the polishing was performed using IC1000/Suba400 (K-gr) for the polishing pad, and SS-25 for slurry with the rotational speed of the polishing table set at 70/71 rpm, the rotational speed of the top ring set at 71 rmp , and a default value for the pressure of the air bags set at 250 hPa, and a pressing force of the dresser set at 200 N. Under the foregoing conditions , the polishing was performed in the following procedure . First , after a polishing pad is replaced, a monitoring wafer is polished. The pressure balance of the air bags within the top ring is optimized from the result of the polishing to polish a wafer . Next , after the polishing pad is cut away by 0.1 mm, the monitoring wafer is polished . The pressure balance of the air bags within the top ring is optimized from the result to polish awafer . After the polishing pad is further cut away by 0.1 mm, the monitoring wafer is polished. The pressure balance of the air bags is optimized from the result to polish a wafer . In the following, this procedure is repeated a required number of times .

When the mechanical dressers 38 , 39 are fed by such a mechanism as ball screws , a number of pulses .required for driving a motor for feeding can be measured to calculate the amount by which the mechanical dresser is fed.

Fig . 8 (A) is a diagram for describing the state of a residual film before and after CMP when the present invention is applied and not applied. The surface of the wafer is not flat but partially has asperities and steps . The difference between a maximum value and a minimum value of the thickness of a film to be polished in the wafer is called the "thickness difference . " When a polished surface of the wafer is flat , the thickness difference is zero . Also , the difference between the "thickness difference" after the polishing and the "thickness difference" before the polishing is called the "residual film difference . "

Fig. 8 (A) shows the residual film difference Δ when the present invention is applied and when not applied, when the grooves in the polishing pad have the depth of 0.4 mm, 0.3 mm, and 0.2 mm, respectively, under conditions in which the pressures of the air bags El - E5 in the top ring are set as illustrated. Specifically, the residual film difference Δ was :

3.3 nm when the grooves had a depth of 0.4 mm and the present invention was not applied;

-43.5 nm when the grooves had a depth of 0.4 mm and the present invention was applied;

7.2 nm when the grooves had a depth of 0.3 mm and the present invention was not applied; -29.4 nm when the grooves had a depth of 0.3 mm and the present invention was applied;

68.6 nm when the grooves had a depth of 0.2 mm and the present invention was not applied; and

-65.3 nm when the grooves had a depth of 0.2 mm and he present invention was applied.

Fig. 8 (B) is a graphic representation of the foregoing result . A minus residual film difference means that the "thickness difference" after polishing is smaller than the "thickness difference" before polishing, so that the difference in thickness was improved as compared with before polishing, i . e . , the flatness was improved. It can therefore be understood that the difference in thickness after CMP was largely reduced by applying the present invention . Next , Fig . 9 shows the thickness and polishing rate when the polishing pad has not at all consumed, where Φ indicates a value when the present invention was applied, and ♦ indicates a value when the present invention was not applied. Fig . 9 (A) is a graph showing the relationship between a radial distance from the center of a 300-mm wafer and the thickness before CMP; and Fig . 9 (B) is a graph showing the relationship between a radial distance from the center of the wafer and the thickness after CMP in Fig . 9 (A) . Then , when the polishing rate was derived from the thickness before and after CMP when the present invention was applied and when not applied, a graph shown in Fig. 9 (C) was obtained. When the result of a simulation for the polishing rate ( indicated by O ) was plotted on this graph, it was understood that the polishing rate when the present invention was applied was fairly consistent with the result of the simulation .

Fig . 10 shows the thickness and polishing rate when the polishing pad has been consumed by 0.1 mm, where Φ indicates a value when the present invention was applied, and, ♦ indicates a value when the present invention was not applied. Fig . 10 (A) is a graph showing the relationship between a radial distance from the center of a 300-mm wafer and the thickness before CMP; and Fig . 10 (B) is a graph showing the relationship between a radial distance from the center of the wafer and the thickness after CMP in Fig. 10 (A) . Then , when the polishing rate was derived from the thickness before and after CMP when the present invention was applied and when not applied, a graph shown in Fig . 10 (C) was obtained. When the result of a simulation for the polishing rate ( indicated by O) was plotted on this graph, it was recognized that the polishing rate was reduced. though slightly, at the center as the polishing pad was consumed more, i . e . , thepolishingpadhas shallower grooves , but thepolishing rate was fairly consistent with the result of the simulation at the center, while in an outer peripheral region , actual values were slightly different from _. the result of the simulation .

Fig . 11 shows the thickness and polishing rate when the polishing pad has been consumed by 0.2 mm, where Φ indicates a value when the present invention was applied, and ♦ indicates a value when the present invention was not applied. Like Figs . 9 and 10 , Fig. 11 (A) is a graph showing the relationship between a radial distance from the center of a 300-mm wafer and the thickness before CMP; and Fig . 11 (B) is a graph showing the relationship between a radial distance from the center of the wafer and the thickness after CMP in Fig . 11 (A) . Then, when the polishing rate was derived from the thickness before and after CMP when the present invention was applied and when not applied, a graph shown in Fig . 11 (C) was obtained. When the result of a simulation for the polishing rate indicated by O was plotted on this graph, it was recognized that the polishing rate was largely reduced at the center, and largely differed from the result of the simulation in an outer peripheral region . A default value of the simulation application should be modified.

INDUSTRIAL AVAILABILITY As will be understood from the foregoing description , since the present invention optimizes a processing pressure using a simulation program based on the Preston ' s basic formula, and performs the polishing in consideration of even those parameters which cannot be coveredby the Preston ¹ s basic formula, it is possible to realize the uniformization in the shape of polished wafers , as required in step with increasing miniaturization of integrated circuits . It is further possible to extend the life time of a consumablematerialby correctlymanaging the state of the consumable material to reduce the operation cost .

Claims

1. A polishing apparatus for polishing an object to be polished under control of a control unit , comprising : atopringhaving at least two pressing sections andcapable of applying an arbitrary pressure to the object to be polished from each of said pressing sections ; a measuring device for measuring a polished amount of the object to be polished; and a monitoring device for monitoring the object to be polished for a polishing state , said polishing apparatus characterized in that : said control unit forces said polishing apparatus to polish the object to be polished in accordance with a simulation program for setting a processing pressure required to optimize a polishing profile of the object to be polished to said top ring based on the output of said measuring device and the output of said monitoring device .

2. A polishing apparatus according to claim 1 , characterized in that : said at least two pressing sections include a plurality of concentric air bags , and a retainer ring surrounding said air bags , and the pressure of said retainer ring is kept at a value larger by 20 percent than an average value of the sum total of the pressures applied by said air bags .

3. A polishing apparatus according to claim 1 or 2 , further comprising a polishing pad for polishing the object to be polished such that said pad is depressed by said top ring , wherein said control unit instructs said polishing apparatus to polish a monitor wafer instead of the object when said monitoring device detects that said polishing pad is cut away by a predetermined depth.

4. A polishing apparatus according to claim 2 , characterized in that when the output of said monitoring device indicates that an abrasion loss of said retainer ring falls below a threshold, said control unit instructs said polishing apparatus to stop polishing.

5. A polishing apparatus according to claim 1 or 2 , characterized in that : when the output of said monitoring device indicates that the temperature on the surface of the object to be polished exceeds a predetermined set temperature , said control unit stops using the simulation program or instructs said polishing apparatus to stop polishing, and when the output of said monitoring device indicates that the temperature on the surface falls below the set temperature , said control unit instructs said polishing apparatus to resume the polishing.

6. A polishing apparatus according to claim 1 or 2 , further comprising a polishing pad for polishing the object to be polished in a state in which said polishing pad is pressed against the object to be polished by said, top ring, said polishing apparatus characterized in that : when the output of said monitoring device indicates that the thickness of said polishing pad falls below a threshold, said control unit stops using the simulation program or instructs said polishing apparatus to stop polishing.

7. A polishing apparatus according to claim 6 , characterized in that said monitoring device comprises a laser displacement gage for measuring the thickness of said polishing pad.

8. A polishing apparatus according to claim 1 or 2 , further comprising a polishing pad for polishing the object to be polished in a state in which said polishing pad is pressed against the object to be polished by said top ring, and a dresser for conditioning said polishing pad, said polishing apparatus characterized in that : when the output of said monitoring device indicates that a cutting rate of said dresser falls below a threshold, said control unit stops using the simulation program, or instructs said polishing apparatus to stop polishing .

9. A polishing apparatus according to claim 7 , characterized in that the cutting rate is monitored using a torque of a motor for driving said dresser .

10. A polishing apparatus according to any of claims 1 to

9 , characterized in that said control unit can adjust the amount of supplied slurry in accordance with the polishing state .