KR101398567B1 - Apparatus and methods for spectrum based monitoring of chemical mechanical polishing - Google Patents

Abstract

The present invention relates to an apparatus and method for spectral-based monitoring of chemical mechanical polishing, including spectral-based endpoint determination, spectral-based polishing rate adjustment, flushing of a top surface of a pad or optical head with a window. The spectral-based endpoint determination uses a reference spectrum experimentally selected for a particular spectral-based endpoint decision logic so that the target thickness can be achieved when the endpoint is called by applying a specific spectral-based endpoint logic. The polishing endpoint may be determined using a sequence of index values or a deviation trace. The flushing system creates a flow of laminar gas across the upper surface of the optical head. Vacuum nozzles and vacuum sources are configured so that the flow of gas is laminar. The window includes a soft plastic portion and a crystalline or glass-like portion. Spectral-based polishing rate adjustment involves acquiring spectra for different regions on the substrate.

Description

[0001] APPARATUS AND METHODS FOR SPECTRUM BASED MONITORING OF CHEMICAL MECHANICAL POLISHING [0002]

The present invention generally relates to chemical mechanical polishing of a substrate.

Typically, an integrated circuit is formed on a substrate by sequentially depositing a conductive layer, a semiconductor layer, or insulating layers on a silicon wafer. One fabrication step involves depositing a filler layer on the non-planar surface and planarizing the filler layer. For certain applications, the fill layer is planarized until the top surface of the patterned layer is exposed. For example, a conductive fill layer may be deposited on the patterned insulating layer to fill the trenches or holes in the insulating layer. After planarization, portions of the conductive layer that remain between the raised patterns of the insulating layer form vias, plugs, and lines that provide a conductive path between the thin film circuits on the substrate. For other applications, such as oxide polishing, the fill layer is planarized until a predetermined thickness remains on the non-planar surface. Also, substrate surface planarization is usually required for photolithography.

Chemical mechanical polishing (CMP) is one acceptable planarization method. Typically, such a planarization method requires that the substrate be mounted on a carrier or polishing head. Typically, the exposed surface of the substrate is positioned opposite the rotating polishing disk pad or belt pad. The polishing pad may be one of a standard pad or a fixed-abrasive pad. While the standard pad has a durable roughened surface, the abrasive-fixed pad includes abrasive particles that are held in containment media. The carrier head provides the substrate with a controllable rod to be pressed against the polishing pad. The polishing slurry is typically supplied to the surface of the polishing pad. The polishing slurry comprises at least one reactive agent and, when used with a standard polishing pad, comprises abrasive particles.

One problem with CMP is to determine whether the polishing process has been completed, i. E. Whether the substrate layer has been flattened to the desired flatness or thickness, or when the desired amount of material has been removed. Excessive polishing (too much removal) of the conductive layer or film increases circuit resistance. On the other hand, under polishing (too little removal) of the conductive layer causes electrical shorting. The initial thickness of the substrate layer, the slurry composition, the polishing pad conditions, the relative speed between the polishing pad and the substrate, and variations in the load on the substrate can cause variations in the material removal rate. This variation can cause variations in the time required to reach the polishing endpoint. Thus, the polishing end point can not be determined solely as a function of polishing time.

SUMMARY OF THE INVENTION It is an object of the present invention to provide a method of chemical mechanical polishing of an improved substrate.

In one general aspect, the invention features a computer implemented method comprising selecting a reference spectrum. The reference spectrum is the spectrum of white light having a thickness greater than the target thickness and reflecting from an interest film located on the first substrate. By applying a specific spectrum-based endpoint logic, the reference spectrum can be empirically tested for a particular spectrum-based endpoint decision logic such that the target thickness can be achieved when the endpoint is called (determined as an endpoint) Is selected. The method includes obtaining a current spectrum. The current spectrum is a spectrum of white light reflected from a target film located on the second substrate and having a current thickness thicker than the target thickness. The target film on the second substrate undergoes a polishing step. The method includes determining, for a second substrate, when the end point of the polishing step has been achieved. The determination is based on a reference spectrum and a current spectrum.

In another general aspect, the invention features a computer implemented method comprising selecting two or more reference spectra. Each reference spectrum is a spectrum of white light reflected from a target film that is located on the first substrate and has a thickness greater than a target thickness. By applying a specific spectrum-based endpoint logic, the reference spectrum is empirically selected for a particular spectrum-based endpoint decision logic such that the target thickness can be achieved when the endpoint is called. The method includes obtaining two or more current spectra. Each current spectrum is a spectrum of white light reflected from a target film located on the second substrate and having a current thickness that is thicker than the target thickness. The target film on the second substrate undergoes a polishing step. The method includes determining, for a second substrate, whether an end point of the polishing step has been achieved, the determination being based on a reference spectrum and current spectra.

In another general aspect, the invention features a computer program product comprising instructions operable to cause a processor to select a reference spectrum. The reference spectrum is a spectrum of white light reflected from a target film located on the first substrate and having a thickness thicker than a target thickness. The reference spectrum is chosen empirically for a particular spectral-based endpoint decision logic so that the target thickness can be achieved when the endpoint is called by applying a specific spectrum-based endpoint logic. The product includes instructions that cause the processor to acquire the current spectrum. The current spectrum is a spectrum of white light reflected from a target film located on the second substrate and having a current thickness thicker than the target thickness. The target film on the second substrate undergoes a polishing step. The article of manufacture includes instructions that, for a second substrate, cause the processor to determine when an endpoint of the polishing step has been achieved. The determination is based on a reference spectrum and a current spectrum. The product is stored tangibly on a machine readable medium.

In another general aspect, the invention features a computer program product stored tangibly on a machine-readable medium. The product includes instructions operable to cause the processor to select two or more reference spectra. Each reference spectrum is a spectrum of white light reflected from a target film located on the first substrate and having a thickness greater than the target thickness. The reference spectrum is chosen empirically for a particular spectral-based endpoint decision logic so that the target thickness can be achieved when the endpoint is called by applying a specific spectrum-based endpoint logic. The product further comprises instructions for obtaining two or more current spectra. Each current spectrum is a spectrum of white light reflected from a target film located on the second substrate and having a current thickness that is thicker than the target thickness. The target film on the second substrate undergoes a polishing step. The product further comprises, for the second substrate, instructions for determining whether an end point of the polishing step has been achieved, the determination being based on a reference spectrum and a current spectrum.

In one general aspect, the invention features a flushing system for flushing a top surface of an optical head. The system includes a gas source configured to provide a flow of gas, a delivery nozzle, a delivery line connecting the gas source to the delivery nozzle, a vacuum source configured to provide a vacuum, a vacuum nozzle, Lt; / RTI > The gas source and delivery nozzle are configured to direct the flow of gas across the upper surface of the optical head. The vacuum nozzle and the vacuum source are configured such that the flow of gas is laminar flow.

In another general aspect, the invention features a flushing system for flushing the bottom surface of a polishing pad window. The system includes a gas source configured to provide a flow of gas, a delivery nozzle, a delivery line connecting the gas source to the delivery nozzle, a vacuum source configured to provide a vacuum, a vacuum nozzle, and a vacuum line connecting the vacuum source to the vacuum nozzle . The gas source and delivery nozzle are configured to direct the flow of gas to the bottom of the polishing pad window, preventing the formation of condensation at the bottom surface of the polishing pad window.

In one general aspect, the invention features an assembly for chemical mechanical polishing. The assembly includes a polishing pad having a polishing surface. The assembly includes a solid window positioned within the polishing pad to provide optical access through the polishing pad. The solid window comprises a first portion made of polyurethane and a second portion made of quartz. The first portion has a surface coplanar with the polishing surface of the polishing pad.

In another general aspect, the invention features a polishing pad comprising a polishing layer having a top surface and a bottom surface. The pad includes an aperture having a first opening in the top surface and a second opening in the bottom surface. The top surface is the polishing surface. The pad comprises a first portion made of soft plastic and a window comprising a crystalline or glass like second portion. The window is transparent to white light. The window is disposed in the opening such that the first portion is plugged into the opening and the second portion is located on the bottom side of the first portion, wherein the first portion is a slurry sealing barrier .

In another general aspect, the present invention features a method of manufacturing a polishing pad. The method includes disposing a mass of crystalline or free-form material within a mold of a polishing pad window, wherein the mass is transparent to white light. The method includes dispensing a liquid precursor of a soft plastic material into a mold, wherein the soft plastic material is transparent to white light. The method includes curing the liquid precursor to form a window comprising a first portion made of a soft plastic material and a crystalline or glass-like second portion. The method includes placing the window into a mold of a polishing pad. The method includes dispensing a liquid precursor of a polishing pad material into a mold of a polishing pad. The method includes curing a liquid precursor of a polishing pad material to produce a polishing pad, wherein when the polishing pad is produced, the window is positioned within the polishing pad such that the first portion acts as a slurry sealing barrier, Is positioned in the mold of the polishing pad.

In another general aspect, the present invention features a method of manufacturing a polishing pad. The method includes disposing a mass of crystalline or glass-like material in a mold of a polishing pad window, wherein the mass is transparent to white light. The method includes dispensing a liquid precursor of a soft plastic material into a mold, wherein the soft plastic material is transparent to white light. The method includes curing the liquid precursor to form a window comprising a first portion made of a soft plastic material and a crystalline or glass-like second portion. The method includes forming a polishing layer comprising an opening, the polishing layer having a top surface and a bottom surface, the opening having a first opening in the top surface and a second opening in the bottom surface, The top surface is the polishing surface. The method includes inserting a window into an opening wherein the window is positioned within the opening such that a first portion is fitted to the opening and a second portion is on a bottom side of the first portion, Acts as a slurry sealing barrier.

In another general aspect, the present invention features a method of manufacturing a polishing pad. The method includes forming a first portion of a polishing pad window, the first portion having a recess and being transparent to white light. The method includes inserting a mass of crystalline or free-form material into the recess, wherein the mass is transparent to white light. The method includes forming a polishing layer comprising an opening, the polishing layer having a top surface and a bottom surface, the opening having a first opening in the top surface and a second opening in the bottom surface, The top surface is the polishing surface. The method includes inserting a window into an opening wherein the window is positioned within the opening such that a first portion is fitted into the opening and a second portion is on a bottom side of the first portion, Acts as a slurry sealing barrier.

In another general aspect, the invention features a computer implemented method. During the polishing sequence, a first spectrum of reflected light is obtained from a first region on the substrate and a second spectrum is obtained from a second region on the substrate. The first and second spectra are compared to a library of spectra to determine a first index for the first spectrum and a second index for the second spectrum. A third spectrum of light reflected at different times during the polishing sequence is obtained from the first region and a fourth spectrum is obtained from the second region. The third and fourth spectra are compared to the library to determine a third index for the first region and a fourth index for the second region. The polishing rate in the first area is determined from the first index and the third index, and the polishing rate in the second area is determined from the second index and the fourth index. Based on the first polishing rate, the second polishing rate, the first target relative thickness for the first region, and the second target relative thickness for the second region, the first region is polished to the first target relative thickness, And the polishing rate adjusted to simultaneously cause the second region to be polished to the second target relative thickness is determined for the second region.

Implementations of the present invention may include one or more of the following features. The first region may be an inner region, and the second region may be an outer annular region. Determining the adjusted polishing rate for the second region comprises determining when the first target relative thickness will be within a predetermined threshold from the second target relative thickness. Determining the adjusted polishing rate for the second region may include determining an estimated end point time for the polishing sequence. Acquiring the first and second spectra may comprise acquiring white light spectra. The method may further comprise adjusting a parameter of the polishing system such that the second area is polished at an adjusted polishing rate. The step of determining the adjusted polishing rate may be performed on a set-up substrate, and adjusting the parameters of the polishing system may be performed on the product substrate, or both of the steps may be performed on the product substrate . Adjusting the parameters of the polishing system may include adjusting the pressure. Determining the adjusted polishing rate may include determining a polishing rate such that when the polishing sequence is completed, the cross-section along the diameter of the substrate has a flat profile or a bowl-shaped profile. Obtaining the first and second spectra may include sampling the substrate at different rotational positions. Obtaining the first and second spectra may comprise measuring spectra reflected from the oxide layer. The method includes polishing the set-up substrate until the set-up substrate is overpolished, acquiring a plurality of spectra from a single area of the test substrate during polishing, and generating the spectra library And storing the plurality of spectra in combination with the time at which each spectrum is acquired. The method may further comprise generating indexes for a library of spectra, wherein the one index represents a spectrum obtained from the set-up substrate at a particular time.

In another general aspect, the invention features a method of monitoring a chemical mechanical polishing process. A multi-wavelength light beam is directed to the substrate being polished and the spectrum of the light reflected from the substrate is measured. The light beam travels along a path across the substrate surface. A sequence of spectral measurements is extracted from the signal and the radial position on the substrate is determined for each spectral measurement. The spectral measurements are classified into a plurality of radial ranges according to their radial positions. A polishing end point for the substrate is determined from spectral measurements in at least one of a plurality of radial ranges. The method may further comprise determining an adjusted polishing rate for one of the radial ranges and applying the adjusted polishing rate to one of the radial ranges.

As used herein, the term substrate includes, for example, an article substrate (e.g., comprising a plurality of memory or processor dies), a test substrate, a bare substrate, and a gating substrate can do. The substrate may be in various stages of integrated circuit manufacturing, for example the substrate may be a bare wafer, or the substrate may include one or more deposited and / or patterned layers. The term substrate may include circular disks and rectangular sheets.

Possible advantages of implementations of the present invention may include one or more of the following. The end point determination can be made virtually without considering the variation of the polishing rate. Factors affecting the polishing rate, e.g., consumables, generally need not be considered. The use of multiple criteria and / or target spectra (as opposed to a single reference spectrum and / or a single target spectrum) is a smoother or more smoother difference than a trace generated by using a single reference spectral technique, Improves the accuracy of end point determination by providing an endpoint trace. The flushing system can make the slurry on the surface of the substrate being polished less dry. The polishing pad window can improve the accuracy and / or accuracy of the end point determination.

The details of one or more embodiments of the invention are set forth in the accompanying drawings and the description below. Other features, objects, and advantages of the present invention will become apparent from the description of the invention, the drawings, and the claims.

Figure 1 shows a chemical mechanical polishing apparatus.
Figures 2a-2h illustrate implementations of a polishing pad window.
Figure 3 shows an implementation of a flushing system.
Figure 4 shows an alternative implementation of a flushing system.
FIG. 5 is a plan view of the polishing pad, showing locations where in-situ measurements are performed.
Figure 6A shows the spectrum obtained from in-situ measurements.
FIG. 6B shows the evolution of the spectra obtained from the in-situ measurements as the polishing progresses.
Figure 7A shows a method for obtaining a target spectrum.
Figure 7b shows a method for obtaining a reference spectrum.
Figures 8A and 8B show a method for end point determination.
Figures 9a and 9b illustrate an alternative method for end point determination.
Figures 10A and 10B illustrate another alternative method for end point determination.
Figure 11 shows an implementation for determining an end point.
Figure 12 shows the peak-to-trough normalization of the spectrum.
Figure 13 illustrates the smoothing effect using the provision of multiple reference spectra.
Figure 14 shows a method for obtaining spectra in regions during polishing.
Figure 15 shows a method for adjusting the polishing rate in regions to achieve a desired profile.
Figure 16 shows a graph of polishing progress versus time for a process in which the polishing rate is adjusted.
Figure 17 shows a graph of polishing progress versus time for a process in which the polishing rate is not adjusted.
Figure 18 shows a graph of polishing versus time for a process using a feed forward method of polishing control.
In various figures, similar reference symbols represent like elements.

Figure 1 shows a polishing apparatus 20 operable to polish a substrate 10. The polishing apparatus 20 includes a rotatable disc-shaped platen 24 on which the polishing pad 30 is placed. The platen is operable to rotate about an axis 25. For example, the motor may rotate the drive shaft 22 to rotate the platen 24. [ The polishing pad 30 may be releasably secured to the platen 24 by, for example, an adhesive layer. When worn, the polishing pad 30 can be removed and replaced. The polishing pad 30 may be a two-layer polishing pad having an outer polishing layer 32 and a softer backing layer 34. [

An optical access 36 through the polishing pad is provided by including an opening (i.e., a hole through which the pad travels) or a solid window. Although the solid window is supported on the platen 24 and may protrude into the opening in the polishing pad in some embodiments, the solid window may be secured to the polishing pad. It is common that the polishing pad 30 is placed on the platen 24 such that an opening or window is placed on the optical head 53 located in the recess 26 of the platen 24. [ As a result, the optical head 53 has optical access to the substrate being polished through the opening or window. The optical head will be further described below.

The window can be, for example, a rigid crystalline or glassy material, such as quartz or glass, or a more flexible plastic material, such as silicone, polyurethane or a halogenated polymer, For example, a fluoropolymer), or a combination of the above materials. The window may be transparent to white light. If the upper surface of the solid window is a hard crystalline or vitreous material, its upper surface must be sufficiently recessed from the polishing surface to prevent scratches. If the upper surface may be in close proximity to and in contact with the polishing surface, the upper surface of the window should be a more flexible plastic material. In some embodiments, the solid window is fixed to the polishing pad and is a polyurethane window, or a window having a combination of quartz and polyurethane. The window may have a high transmittance, for example, approximately 80% transmittance, for monochromatic light of a particular color, for example blue light or red light. The window may be sealed to the polishing pad 30 so that the liquid is not leaked through the interface of the window and the polishing pad 30.

In one implementation, the window comprises a hard crystalline or vitreous material covered with an outer layer of a more ductile plastic material. The upper surface of the more malleable material may be coplanar with the polishing surface. The bottom surface of the hard material may be coplanar with the bottom surface of the polishing pad, or may be recessed relative to the bottom surface of the polishing pad. In particular, if the polishing pad comprises two layers, the solid window can be incorporated into the polishing layer, and the bottom layer can have openings aligned with the solid window.

Assuming that the window contains a combination of rigid crystalline or glassy material and a more ductile plastic material, the adhesive used for fixing the two parts is not required. For example, in one implementation, no adhesive is used to bond the polyurethane portion of the window to the quartz portion. Alternatively, an adhesive transparent to white light may be used, or an adhesive may be applied so that light passing through the window does not pass through the adhesive. By way of example, the adhesive may only be applied around the interface between the polyurethane portion and the quartz portion. A refractive index gel can be applied to the bottom surface of the window.

The bottom surface of the window may optionally include one or more recesses. The recess can be shaped, for example, to accommodate the end of an optical fiber cable or the end of an eddy current sensor. The recess allows the end of the fiber optic cable or the end of the eddy current sensor to be positioned a distance shorter than the thickness of the window from the substrate surface being polished. In an embodiment in which the window comprises a hard crystalline or glass-like portion and the recess is formed in such a portion by machining, the recess is polished to remove the scratches caused by machining. Alternatively, solvent and / or liquid polymer may be applied to the surfaces of the recess to remove scratches caused by machining. Removal of scratches, typically caused by processing, can reduce scattering and improve light transmission through the window.

Figures 2a-2h illustrate various implementations of a window. As shown in FIG. 2A, the window may have two portions, a polyurethane portion 202 and a quartz portion 204. The portions are those in which the polyurethane portion 202 is located at the top of the quartz portion 204. A window may be positioned within the polishing pad such that the top surface 206 of the polyurethane layer is flush with the polishing surface 208 of the polishing pad.

As shown in FIG. 2B, the polyurethane portion 202 may have a recess in which the quartz portion is located. The bottom surface 210 of the quartz portion is exposed.

2C, the polyurethane portion 202 may include protrusions 212, for example, protrusions 212 projecting into the quartz portion 204. The protrusions can act to reduce the likelihood that the polyurethane portion 202 will pull away from the quartz portion 204 due to friction from the substrate or retaining ring.

As shown in Figure 2D, the interface between the polyurethane portion 202 and the quartz portion 204 may be a rough surface. Such a surface can improve the bond strength of the two portions of the window and also reduces the likelihood that the polyurethane portion 202 will be displaced from the quartz portion 204 due to friction from the substrate or retaining ring.

As shown in Figure 2E, the polyurethane portion 202 may have a non-uniform thickness. The thickness at the location in the path 214 of the light beam is less than the thickness at the location that is not in the path 214 of the light beam. For example, the thickness t ₁ is thinner than the thickness t ₂ . Alternatively, the thickness at the edge of the window may be thinner.

2F, the polyurethane portion 202 may be attached to the quartz portion 204 by use of an adhesive 216. [ The adhesive can be applied so that it is not in the path 214 of the light beam.

As shown in FIG. 2G, the polishing pad may include a polishing layer and a backing layer. The polyurethane portion 202 extends through the polishing layer and at least partially into the backing layer. The holes in the backing layer may be larger in size than the holes in the polishing layer and the polyurethane sections in the backing layer may be wider than the polyurethane sections in the polishing layer. Thus, the polishing layer provides a lip 218 that can protrude above the window and act to prevent the polyurethane portion 202 from escaping from the quartz portion 204. The polyurethane portion 202 is conformed to the holes of the layers of the polishing pad.

The refractive index gel 220 can be applied to the bottom surface 210 of the quartz portion 204 to provide a medium through which light can travel from the fiber cable 222 to the window, have. The refractive index gel 220 can fill the volume between the fiber optic cable 222 and the quartz portion 204 and can match the refractive indices of the optical fiber cable 222 and the quartz portion 204 or have a refractive index therebetween have.

In embodiments where the window comprises both quartz and polyurethane portions, the polyurethane portion should have a thickness such that the polyurethane portion does not expose the quartz portion during the life of the polishing pad. Quartz can be recessed from the bottom surface of the polishing pad, and the fiber optic cable 222 can extend partially into the polishing pad.

The above-described window and polishing pad can be manufactured using various techniques. The backing layer 34 of the polishing pad may be attached to its outer polishing layer 32, for example, by an adhesive. An opening providing the optical access 36 may be formed in the pad 30 by cutting or molding the pad 30 to include, for example, an opening, and the window may be inserted into the opening, And can be fixed to the pad 30 by an adhesive. Alternatively, a liquid precursor of the window may be dispensed into the openings in the pad 30 and cured to form a window. Alternatively, a solid transparent element of the solid, such as, for example, the crystalline or glass-like portion described above, may be disposed within the liquid pad material, As shown in FIG. In either of the last two cases, a block of pad material may be formed and a layer of polishing pad having a molded window may be scratched from the block.

In an embodiment wherein the window comprises a crystalline or glass-like first portion and a second portion made of a soft plastic material, the second portion may be formed within the opening of the pad 30 by applying the liquid precursor technique described above. have. Thereafter, the first portion may be inserted. If the first portion is inserted before the liquid precursor of the second portion is cured, the cure may bond the first portion and the second portion. If the first portion is inserted after the liquid precursor is cured, the first portion and the second portion may be fixed using an adhesive.

The polishing apparatus 20 may include a flushing system to improve light transmission through the optical access 36. There are different implementations of the flushing system. In embodiments of the polishing apparatus 20 in which the polishing pad 30 includes an opening instead of a solid window, the flushing system provides a fluid, e.g., a gas or a laminar flow of liquid across the upper surface of the optical head 53 . (The upper surface may be the upper surface of the lens included in the optical head 53.) Laminar flow of the fluid across the upper surface of the optical head 53 may sweep / sweep the opaque slurry from the optical access Thereby preventing the slurry from drying on the upper surface, and consequently improving the permeability through optical access. In implementations in which the polishing pad 30 includes a solid window instead of an opening, the flushing system is implemented to direct the flow of gas to the bottom surface of the window. The flow of gas can prevent condensation from forming on the bottom surface of the solid window, which can interfere with optical access, if there is no flow of gas.

Figure 3 shows an implementation of a laminar flow flushing system. The flushing system includes a gas source 302, a delivery line 304, a delivery nozzle 306, a suction nozzle 308, a vacuum line 310 and a vacuum source 312. The gas source 302 and the vacuum source may be configured such that they can introduce and inhale gases of the same or similar volume. The transfer nozzle 306 is positioned such that the laminar flow of gas is directed across the transparent upper surface 314 of the in situ monitoring module and not toward the substrate surface being polished. As a result, the laminar flow of gas does not dry the slurry on the surface of the substrate being polished, which can have undesirable effects on the polishing.

Figure 4 shows an implementation of a flushing system to prevent condensate from forming on the bottom surface of a solid window. The system reduces or prevents the formation of condensate on the bottom surface of the polishing pad window. The system includes a gas source 402, a delivery line 404, a delivery nozzle 406, a suction nozzle 408, a vacuum line 410 and a vacuum source 412. The gas source 402 and the vacuum source may be configured such that they can introduce and inhale the same or similar volume of gas. The transfer nozzle 406 is positioned such that the flow of gas is directed to the bottom surface window in the polishing pad 30. [

In one alternative implementation to the implementation of FIG. 4, the flushing system does not include a vacuum source or a vacuum line. Instead of these components, the flushing system includes a vent formed in the platen so that the gas entering the space below the solid window can be vented to the side of the platen, or, alternatively, To any other location in the polishing apparatus which tolerates the pressure of the fluid.

The gas source and the vacuum source described above can be positioned apart from the platen so that the gas source and the vacuum source do not rotate with the platen. In this case, a rotatable coupler for transferring the gas is included in each of the supply line and the vacuum line.

Referring again to FIG. 1, the polishing apparatus 20 includes a combined slurry / rinse arm 39. During polishing, the arm 39 is operable to dispense a slurry 38 comprising a liquid and a pH adjuster. Alternatively, the polishing apparatus includes a slurry port operable to dispense the slurry to the polishing pad (30).

The polishing apparatus 20 includes a carrier head 70 that is operable to hold the substrate 10 against the polishing pad 30. The carrier head 70 is suspended from the support structure 72, e.g., a carousel and is suspended from the carrier drive shaft 74 to allow the carrier head to rotate about the shaft 71 And is connected to the carrier head rotation motor 76. [ In addition, the carrier head 70 may oscillate laterally within the radial slot formed within the support structure 72. During operation, the platen is rotated about its central axis 25, and the carrier head is rotated about its central axis 71 and translates laterally across the upper surface of the polishing pad. do.

The polishing apparatus also includes an optical monitoring system, which may be used to determine a polishing endpoint, as described below. The optical monitoring system includes a light source 51 and a photodetector 52. Light travels from the light source 51 through the optical access 36 in the polishing pad 30 and impinges on the substrate 10 and again from the substrate 10 through the optical access 36, 52).

A bifurcated optical cable 54 may be used to transfer light from the light source 51 to the optical access 36 and again from the optical access 36 to the optical detector 52. The bifurcated optical cable 54 may include a "trunk" 55 and two "branches 56 and 58".

As mentioned above, the platen 24 includes a recess 26 in which the optical head 53 is located. The optical head 53 holds one end of the main line 55 of the bifurcated optical cable 54 configured to transmit light to the substrate surface being polished and from the substrate surface being polished. The optical head 53 may include one or more lenses or windows over the end of the bifurcated optical cable 54 (e.g., as shown in FIG. 3). Alternatively, the optical head 53 may hold the end of the main line 55 just adjacent to the solid window in the polishing pad. The optical head 53 can hold the nozzles of the flushing system described above. The optical head 53 may be detached from the recess 26 as required, for example, to perform maintenance or maintenance of the calibration.

The platen includes a detachable in-situ monitoring module 50. The in situ monitoring module 50 may include one or more of the following: a light source 51, a light detector 52, and a light source 51 and a light detector 52, A circuit for transmitting and receiving signals from a detector (52). For example, the output of the detector 52 may be a digital electronic signal within the drive shaft 22, e.g., via a rotational coupler, such as a slip ring, to a controller for the optical monitoring system. Similarly, the light source may be turned on or off in response to a control command in the digital electronic signal traveling from the controller to the module 50 via the rotary coupler.

In addition, the in situ monitoring module can hold the respective ends of the branch portions 56 and 58 of the bifurcated optical fiber 54. The light source is operable to transmit light, which is transmitted to the outside of the end of the main line 55 located in the optical head 53 via the branch line 56, and collides with the substrate being polished. Light reflected from the substrate is received at the end of the main line 55 located in the optical head 53 and is transmitted to the photodetector 52 via the branch line 58.

In one implementation, the bifurcated fiber optic cable 54 is a bundle of optical fibers. The bundle includes a first group of optical fibers and a second group of optical fibers. The optical fibers in the first group are connected to transmit light from the light source 51 to the surface of the substrate being polished. The optical fibers in the second group are coupled to receive light reflected from the substrate surface being polished and to transfer the received light to a photodetector. The optical fibers can be arranged so that the optical fibers in the second group form an X-shaped configuration centered in the longitudinal axis of the bifurcated optical fiber 54 (as viewed in cross section of the bifurcated optical fiber cable 54) . Alternatively, different arrangements may be implemented. For example, the optical fibers in the second group can form a V-shaped shape that is a mirror image of each other. Suitable two-prong optical fibers are available from Verity Instruments, Inc. of Carrollton, Texas, USA.

Generally, there is an optimum distance between the polishing pad window and the end of the main line 55 of the bifurcated fiber optic cable 54 adjacent to the polishing pad window. This distance can be determined experimentally and is influenced by, for example, the reflectivity of the window, the shape of the light beam emitted from the bifurcated fiber optic cable, and the distance to the substrate being monitored. In one implementation, a bifurcated fiber optic cable is positioned so as to be as close as possible to the bottom of the window, with the end adjacent the window not substantially contacting the window. In this implementation, the polishing apparatus 20 includes a mechanism operable to adjust the distance between the bottom surface of the polishing pad window and the end of the bifurcated fiber optic cable 54, for example, as part of the optical head 53 . Alternatively, the proximal end of the bifurcated fiber optic cable is embedded in the window.

The light source 51 is operable to emit white light. In one embodiment, the emitted white light comprises light having a wavelength of 200-800 nanometers. A suitable light source is a xenon lamp or a xenon mercury lamp.

The photodetector 52 may be a spectrometer. Basically, a spectrometer is an optical tool for measuring properties of light over a part of an electromagnetic spectrum, for example, intensity. A suitable spectrometer is a grating spectrometer. The typical output of a spectrometer is the intensity of light as a function of wavelength.

Optionally, the in-situ monitoring module 50 may include other sensor members. The in-situ monitoring module 50 may include, for example, an eddy current sensor, a laser, a light emitting diode, and photodetectors. In implementations in which the in-situ monitoring module 50 includes an eddy current sensor, it is common that the module 50 is located such that the substrate being polished is within the operating range of the eddy current sensors.

The light source 51 and the photodetector 52 are coupled to a computing device operable to control their operation and receive their signals. The computing device may include a microprocessor located adjacent to the polishing apparatus, such as, for example, a personal computer. In connection with the control, the computing device may, for example, synchronize the activation of the light source 51 with the rotation of the platen 24. As shown in FIG. 5, the computer causes the light source 51 to emit a series of flashes that begin and end immediately before the substrate 10 passes through the in-situ monitoring module (Each point 501-511 shown represents the location where light from the in-situ monitoring module is reflected and reflected). Alternatively, the computer may cause the light source 51 to initiate immediately before the substrate 10 passes through the in-situ monitoring module and successively emit light that terminates immediately after passing through.

In connection with receiving signals, the computing device may receive a signal that carries information describing, for example, the spectrum of light received by the optical detector 52. 6A shows examples of spectra measured from light emitted from a single flash of a light source and reflected from a substrate. The spectrum shows the raw spectrum, i.e., the spectrum before normalization. The spectrum 602 is measured from light reflected from the product substrate. The spectrum 604 is measured from light reflected from a base silicon substrate (a wafer having only a silicon layer). The spectrum 606 is obtained from the light received by the optical head 53 in the absence of a substrate located above the optical head 53. Under these conditions, referred to herein as dark conditions, the received light is typically scattered light from the pad window light.

The computing device may process the above-described signals to determine an end point of the polishing step. Without being limited to any particular theory, the spectra of light reflected from the substrate 10 evolve as the polishing progresses. FIG. 6B provides an example of the development as the polishing of the target film progresses. The spectra in FIG. 6B are normalized. The different lines of the spectrum represent different times in the polishing. As can be seen, the spectral characteristics of the reflected light change as the film thickness changes, and a particular spectrum is attributed to a specific thickness of the film. The computing device may execute logic to determine when an endpoint has been reached, based on one or more of the spectra. One or more spectra underlying the endpoint determination may include a target spectrum, a reference spectrum, or both.

As used herein, the target spectrum refers to the spectrum exhibited by the white light reflected from the target film when the target film has a target thickness. For example, the target thickness may be 1, 2, or 3 microns. Alternatively, for example, when the underlying film is exposed to clear the target film, the target thickness may be zero.

There can be many target spectra for a given target thickness, and are usually there. This is the case, as the polishing usually occurs at a finite speed, so that the target film maintains the target thickness for a duration that a plurality of spectra can be collected. Also, different regions of the patterned substrate usually produce different spectra (even when spectra are acquired at the same time during polishing). For example, the spectrum of light reflected from a scribe line in a substrate differs (i.e., has a different shape) from the spectrum of light reflected from an array of substrates. Such a phenomenon is referred to as a pattern effect in the present specification. Thus, there may be multiple spectra for a particular target thickness, and multiple spectra may contain different spectra due to pattern effects.

FIG. 7A illustrates a method 700 for obtaining one or more target spectra. The properties of the substrate having the same pattern as the product substrate are measured (step 702). In this specification, the substrate to be measured is referred to as a "set-up" substrate. The set-up substrate may simply be a substrate the same as or similar to the product substrate, or the set-up substrate may be a substrate from a batch. The properties may include the thickness prior to polishing of the object film at a particular object location on the substrate. Typically, thicknesses at multiple locations are measured. Typically, locations are selected such that die features of the same type are measured for each location. The measurement may be performed at a metrology station.

The set-up substrate is polished according to the object polishing step, and the spectra of the white light reflected from the substrate surface being polished are collected during polishing (step 704). Polishing and spectral collection can be performed in the polishing apparatus described above. The spectra are collected by the in-situ monitoring system during polishing. Multiple spectra can be collected for each platen revolution. The substrate is excessively polished so that the spectrum of the light reflected from the substrate can be obtained when the target thickness is achieved. (I.e., polished past the estimated end point).

The properties of the over-polished substrate are measured (step 706). The characteristics include the thicknesses of the target film after polishing at the specific positions or positions used for pre-polishing measurements.

The measured thicknesses and collected spectra are used to select one or more spectra that are determined to be exhibited by the substrate when the substrate has an interest thickness among the collected spectra (step 708). In particular, linear interpolation may be performed, using measured film thickness before polishing and substrate thickness after polishing, to determine which of the spectra appeared when the target film thickness was achieved. The spectrum determined to represent when the target thickness is achieved is specified to be the target spectrum for a batch of substrates. Typically, three of the collected spectra are designated to be target spectra. Alternatively, five, seven, and nine spectra are designated to be the target spectra.

Optionally, the collected spectra are processed to improve accuracy and / or accuracy. The spectra may be processed, for example, to normalize and / or averge them and / or filter noise from them, with a common reference. Specific implementations of such processing operations are described below.

As used herein, the reference spectrum refers to the spectrum associated with the target film thickness. By applying endpoint determination logic based on specific spectra, typically one, two, or three or more reference spectra for the endpoint determination logic based on specific spectra are experimentally determined so that the target thickness is achieved when the computer device calls the endpoint. . As described below with reference to FIG. 7B, the reference spectrum or spectra can be selected repetitively. The reference spectra are usually not the target spectra. Rather, the reference spectrum is usually the spectrum of light reflected from the substrate when the target film has a thickness that is thicker than the target thickness.

FIG. 7B shows a method 701 for selecting a particular spectrum-based endpoint determination logic and a reference spectrum for a particular target thickness. In some embodiments, two or more spectra may be selected instead of only one spectrum. The set-up substrate is measured and polished as described above in step 702 and step 706 (step 703). In particular, the collected spectra and the time at which each collected spectrum is measured are stored. Multiple spectra can be collected for each platen revolution during polishing.

The polishing rate of the polishing apparatus for a particular set-up substrate is calculated (step 705). By using the thicknesses (T ₁ , T ₂ ) before and after polishing and the actual polishing time (PT), the average polishing rate (PR), for example, PR = (T ₂ -T ₁ ) / PT, Can be calculated.

As discussed below, an endpoint time for a particular set-up substrate is calculated (step 707) to provide a calibration point to test the reference spectrum. The end point can be calculated based on the calculated polishing rate PR, the start thickness ST of the target film before polishing, and the target thickness TT of the target film. Assuming that the polishing rate is constant throughout the polishing process, the end point time ET can be computed as a simple linear interpolation, e.g., ET = (ST-TT) / PR.

Optionally, the calculated endpoint time may be evaluated by polishing another substrate in a batch of patterned substrates, stopping polishing at the calculated endpoint time, and measuring the thickness of the target film. If the thickness is within a satisfactory range of the target thickness, then the calculated endpoint time will be satisfactory. Otherwise, the computed endpoint time can be re-computed.

One of the collected spectra is selected and designated to be the reference spectrum (step 709). The selected spectrum is the spectrum of light reflected from the substrate when the target film is thicker than the target thickness and has a thickness approximately equal to the target thickness. Alternatively, two or more spectra are designated as reference spectra. Typically, three of the collected spectra are designated to be reference spectra. Alternatively, five, seven, or nine spectra are designated to be reference spectra. As with the target spectra, there can be multiple reference spectra because the polishing rate is finite.

In one implementation, a particular platen rotation corresponding to the end point time calculated in step 707 is identified, and the spectra collected during a particular platen rotation are selected to be assigned to the reference spectra. By way of example, the collected spectra may be from the central region of the substrate. The platen rotation corresponding to the calculated end point time is the platen rotation where the time corresponding to the calculated end point time occurs during the platen rotation. As an example, if the calculated endpoint time is 25.5 seconds, then the particular platen rotation corresponding to this calculated endpoint time is a platen rotation in which 25.5 seconds of polishing occurred in the polishing process during platen rotation.

Using the collected spectra for the set-up substrate and using the selected spectra or spectra designated to be reference spectra or spectra, the specific end point decision logic is executed in the simulation (step 711). Execution of the logic yields an empirically derived but simulated endpoint time that the logic has determined to be the endpoint.

The simulated endpoint time is experimentally induced but compared to the calculated endpoint time (step 713). If the experimentally derived endpoint time is within the threshold range of the calculated endpoint time, it is recognized that the currently selected reference spectrum or spectra produces a result that matches the calibration point. Thus, when the endpoint logic is executed using the reference spectrum or spectra in a run-time environment, the system must reliably detect the endpoint at the target thickness. Therefore, the reference spectrum or spectra may be maintained as a reference spectrum for the run-time polishing of the remaining substrates of the batch (step 718). Otherwise, steps 709 and 711 are repeated appropriately.

Alternatively, variables other than the selected spectrum or spectra may be varied for each iteration (i.e., performing each of steps 709 and 711). For example, the threshold range from the processing of the above-mentioned spectra (e.g., filter parameters) and / or the minimum value of the difference trace may be varied. The critical ranges of the minimum values of the deviation traces and deviation traces are described below.

8A illustrates a method 800 for using spectrum-based endpoint determination logic to determine an endpoint of a polishing step. Another batch of batches of patterned substrates is polished using the above-described polishing apparatus (step 802). In each rotation of the platen, the following steps are performed.

One or more spectra of the white light reflected from the substrate surface being polished are measured to obtain one or more current spectra for the current platen rotation (step 804). The spectral measurements at points 501-511 (see FIG. 5) are examples of spectra measured during the current platen revolution. As described above with reference to FIG. 7A and as described below with reference to FIG. 11, the spectra measured during the current platen revolution are selectively processed to improve accuracy and / or accuracy.

In some implementations, if only one spectrum is measured, that one spectrum is used as the current spectrum. If more than one current spectra are measured for platen rotation, the spectra are grouped and averaged within each group, and the averages are designated to be current spectra. The spectra can be grouped by radial distance from the center of the substrate. As an example, if a first current spectrum can be obtained from the spectra measured at point 502 and point 510 (see FIG. 5), and the second current spectrum is measured at point 503 and point 509 And a third current spectrum can be obtained from the spectra measured at point 504 and point 508 (and so on). The spectra measured at point 502 and point 510 are averaged to obtain a first current spectrum for the current platen rotation. The spectra measured at point 503 and point 509 are averaged to obtain a second current spectrum for the current platen rotation. The spectra measured at point 504 and point 508 are averaged to obtain a third current spectrum for the current platen rotation.

In some implementations, two or more of the spectra measured during the current platen revolution are selected to be the current spectra for the current platen revolution. In one implementation, the spectra that are selected to be the current spectra are the spectra measured at sample locations close to the center of the substrate (e.g., point 505, point 506, and point 507 shown in FIG. 5) admit. The selected spectra are not averaged and each selected spectrum is designated to be the current spectrum for the current platen revolution.

A difference between each of the one or more current spectra and each reference spectrum is calculated (step 806). The reference spectrum or spectra can be obtained as described above with reference to Figure 7b. In one implementation, the deviation is the sum of the deviations of the intensities over the wavelength range. In other words,

Where a and b are the lower and upper limits of the range of wavelengths of the spectrum, respectively, and I _current (lambda) and I _criteria (lambda) respectively represent the intensity of the current spectra and the intensity of the target spectra for a given wavelength.

One way to calculate the deviation between each of the current spectra and each of the reference spectra is to select each of the current spectra. For each selected current spectrum, the deviation for each reference spectrum is calculated. For example, given current spectra (e, f, and g) and reference spectra (E, F, and G), a deviation will be calculated for each of the following combinations of current and reference spectra: e And E, e and F, e and G, f and E, f and F, f and G, g and E, g and F, and g and G.

In some implementations, each calculated deviation is appended to the deviation trace (step 808). Generally, a deviation trace is a plot of the calculated deviation. The deviation traces are updated at least once per platen revolution. (If multiple current spectra are obtained for each platen revolution, the deviation trace may be updated more than once per platen revolution.)

Generally, a deviation trace is one of the calculated deviations (in this case, the smallest of the deviations calculated for the current platen revolution). As an alternative to the smallest deviation, other deviations of the deviations, e. G. The median of the deviations or the next deviation of the smallest deviations, can be added to the trace.

Taking the smallest deviation among the deviations can improve the accuracy of the end point determination process. The current spectrum may include spectra from light that is reflected at different locations of the substrate (e.g., scribe lines and arrays), and the pattern effects described above may cause these spectra to vary significantly. Similarly, the reference spectra may include spectra from light reflected at different locations on the substrate. The comparison of such different spectra is faulty and may cause errors in the end point determination. For example, comparing a current spectrum of light reflected from a scribed line of a patterned substrate with a reference spectrum of light reflected from an array of patterned substrates may introduce errors into the calculation of the end point determination . Ironically, such a comparison is a comparison between apples and oranges. Considering only the smallest deviation of deviations, this type of comparisons are factored out (even if comparisons have been made) from the calculations. Thus, by using multiple reference spectra and multiple current spectra, and by considering only the smallest one of the deviations between each of these spectra, errors that may be caused by the above-mentioned false comparison can be avoided.

Optionally, the deviation traces may be processed, for example, by filtering out the calculated deviations beyond the threshold from the preceding one or more calculated deviations to smooth out the deviation traces.

It is determined whether the deviation trace is within a minimum threshold (step 810). After the minimum value is detected, the endpoint is called when the deviation trace begins to rise past a certain threshold of minimum value. Alternatively, the endpoint may be invoked based on the slope of the deviation trace. In particular, the slope of the deviation trace approximates zeor at the minimum value of the deviation trace and becomes zero. The end point can be called when the slope of the deviation trace is within the critical range of the slope close to zero.

Optionally, window logic may be applied to facilitate determination of step 808. [ Windows logic suitable for use is described in U.S. Patent Nos. 5,893,796 and 6,296,548, herein incorporated by reference and commonly assigned to the assignee of the present application.

If it is not determined that the deviation trace has reached the minimum critical range, polishing is allowed to continue and steps 804, 806, 808, and 810 are repeated appropriately. Otherwise, the endpoint is called and the polishing is stopped (step 812).

Figure 8B shows the method described above for determining the end point. Trace 801 is a raw deviation trace. The trace 803 is a smoothed deviation trace. The endpoint is called when the smoothed deviation trace 803 reaches a threshold value 805 that is greater than the minimum value 807. [

As an alternative to using the reference spectrum or spectra, a target spectrum or spectra may be used in method 800. The deviation calculation will be between the current spectrum or spectra and the target spectrum or spectra and the end point will be determined when the deviation trace reaches the minimum value.

FIG. 9A illustrates an alternative method 900 for using spectrum-based endpoint determination logic to determine the endpoint of the polishing step. The set-up substrate is polished and a target spectrum or spectra and a reference spectrum or spectra are obtained (step 902). These spectra can be obtained as described above with reference to Figs. 7A and 7B.

The target deviation is calculated (step 904). If one reference spectrum is used, the target deviation is a deviation between the reference spectrum and the target spectrum and can be calculated using the deviation equation described above. If two or more reference spectra are used, the target deviation is the smallest deviation of the deviations between the reference spectra and the target spectra, which is calculated using the method for calculating the deviation and the deviation equation described above Step 808).

Polishing of another substrate in a batch of substrates is started (step 906). The following steps are performed for each platen revolution during polishing. One or more spectra of white light reflected from the substrate surface being polished to obtain one or more current spectra for the current platen rotation is measured (step 908). A deviation between one or more current spectra and the reference spectrum is calculated (step 910). Calculated deviations or deviations (if one or more current spectra are present) are added to the deviation traces (step 912). (Steps 908, 910, and 912 are respectively similar to steps 804, 806, and 808.) It is determined whether the deviation traces are within the threshold range of the target deviation (step 914). If it is not determined that the deviation trace has reached the critical range of the target deviation, then polishing is allowed to continue and steps 908,910, 912, and 914 are repeated appropriately. Otherwise, the endpoint is called and the polishing is stopped (step 916).

FIG. 9B shows the above-described method for determining an end point. Trace 901 is a raw deviation trace. Trace 903 is a smoothed deviation trace. The end point is called when the smoothed deviation trace 903 is within the critical range 905 of the target deviation 907. [

10A shows another method 1000 for determining the end point of the polishing step. Reference or reference spectra are obtained (step 1002). The reference spectrum or spectra are obtained as described above with reference to FIG. 7B.

The spectra collected from the process of obtaining the reference spectrum are stored in a library (step 1004). Alternatively, the library may contain the theoretically generated spectra without being collected. Spectra containing the reference spectrum are indexed such that each spectrum has a unique index value. Indexing is implemented such that the index values are ordered in the order in which the spectra are measured. Thus, the index can be correlated with time and / or platen rotation. In one implementation, the first spectrum collected at the first time point will have an index value less than the second spectrum collected at a later time point. The library may be implemented in the memory of the computing device of the polishing apparatus.

A substrate from a batch of substrates is polished, and the following steps are performed for each platen revolution. One or more spectra are measured to obtain current spectra for the current platen rotation (step 1006). The spectrum is obtained as described above. Each current spectrum is compared to the spectra stored in the library, and a library spectrum that best fits any of the current spectra is determined (step 1008). An index of the library spectrum that is determined to best fit any of the current spectra is added to the endpoint index trace (step 1010). When the endpoint trace reaches the reference spectrum or the index of any of the spectra, an endpoint is called (step 1012).

Figure 10B shows the above-described method for end point determination. Trace 1004 is a raw index trace. Trace 1003 is a smoothed deviation trace. Line 1005 represents the index value of the reference spectrum. In each sweep of the optical head below the substrate, a plurality of current spectra can be obtained, e.g., spectra for each radial region on the substrate being tracked, and index traces generated for each radial region .

Figure 11 illustrates an implementation for determining an end point during the polishing step. For each platen revolution, the following steps are performed. A plurality of raw spectra of the white light reflected from the substrate surface being polished is measured (step 1102).

Each measured source spectrum is normalized to remove light reflections contributed by media other than the target film or films (step 1104). Normalization of the spectra facilitates comparison of the spectra between each other. Light reflections contributed by a medium other than the target film or films include light reflections from the polishing pad window and the base silicon layer of the substrate. The contribution from the window can be estimated by measuring the spectrum of light received by the in situ monitoring system under dark conditions (i.e., no substrates are placed on the in situ monitoring system). The contribution from the silicon layer can be estimated by measuring the spectrum of light reflected from the bare silicon substrate. Contributions from the bare silicon substrate can be obtained prior to initiation of the polishing step. However, the contribution from the window, the so-called dark contribution, is obtained dynamically, i.e., for each platen revolution, as at point 511 in FIG.

The measured raw spectrum is normalized as follows.

Normalized spectrum = (A-Dark) / (Si-Dark)

Where A is the source spectrum, Dark is the spectrum obtained under dark conditions, and Si is the spectrum obtained from the bare silicon substrate.

Alternatively, the collected spectra may be classified based on the region of the pattern that generated the spectrum, and the spectra from some regions may be excluded from the endpoint calculation. In particular, spectra from light reflected from the scribe line can be removed from consideration (step 1106). Different regions of the patterned substrate usually yield different spectra (even when spectra are acquired at the same time during polishing). For example, the spectrum of light reflected from a scribe line in a substrate differs from the spectrum of light reflected from an array of substrates. Because of their different shapes, the use of spectra from both regions of the pattern usually causes an error in the end point determination. However, the spectra can be grouped into groups for scribe lines and groups for scribes based on their shapes. Since there is often a greater variation in the spectra for the scribe lines, these spectra can usually be excluded from consideration for precision improvement.

Step 1106 may be an alternative to techniques (as described above in step 808 of method 800) that use multiple reference spectra to compensate for the errors described above caused by erroneous comparisons . Step 1106 may be performed instead of or in addition to step 808. [

So far, a subset of the processed spectra is selected and averaged in some cases (step 1108). The subset consists of spectra obtained from light reflected from the substrate, at points of the area on the substrate. For example, the region may be an area 503 or an area 507 (FIG. 5).

Optionally, a high pass filter is applied to the measured raw spectrum (step 1110). Typically, the application of a high pass filter eliminates the low frequency distortion of the average of a subset of the spectra. The high pass filter can be applied to both the raw spectra, their average, or the raw spectrum and their average.

The average is normalized such that the amplitude of the mean is equal to or similar to the amplitude of the reference spectrum (step 1112). The amplitude of the spectrum is the peak-to-low point of the spectrum. Alternatively, the reference spectrum is also normalized such that the reference spectrum of the average is equal to or similar to the normalized reference amplitude. In some implementations, each of the spectra of a subset of the spectrum, such that the amplitude of each spectrum of a subset of the spectrum is the same or similar to the amplitude of the reference spectrum, or the reference spectrum is also equal or similar to the normalized reference amplitude Is normalized.

The deviation between the normalized mean or spectrum and the reference spectrum is calculated (step 1114). As described with reference to Fig. 7B, a reference spectrum or spectra are obtained. The deviations are calculated using the above-described equation for calculating deviations between spectra.

The deviation trace is updated with the smallest deviation of the current deviation or calculated deviations (step 1116). The deviation traces represent the calculated deviations between the normalized averages or spectra and the reference spectra or spectra as a function of time (or platen rotation).

A median filter and a lowpass filter are applied to the updated deviation traces (step 1118). Typically, the application of these filters smooths the trace (by reducing or eliminating spikes in the trace).

The end point determination is performed based on the updated and filtered deviation traces (step 1120). A determination is made based on when the deviation trace has reached the minimum value. The window logic described above is used to make decisions.

More generally, the signal processing steps in steps 1104-1112 can be used to improve the end point determination procedure. For example, instead of generating the deviation traces as described above with reference to FIG. 10A, normalized average spectra to generate index traces may be used to select the spectrum from the library.

FIG. 12 shows the normalization of step 1112. FIG. As shown, only a portion of the spectrum (mean or of the spectra) is considered for normalization. In this specification, the portion considered is referred to as the normalization range, which may also be user selectable. Normalization is performed so that the highest and lowest points in the normalization range are normalized to 1 and 0, respectively. The normalization is computed as:

Here, g is the gain (gain), and, h is the offset (offset) and, r _max is the maximum value in the normalized range, r _min is the minimum value in the normalized range, N is the normalized spectra, R is normalized before (pre normalized spectrum.

Figure 13 shows the smoothing effect using the provision of multiple reference spectra. Trace 1302 was generated using a single reference spectrum (which is averaged). Trace 1304 (as described above with reference to Figure 8) was generated using three reference spectra. Trace 1306 was generated using nine reference spectra. As can be seen, the trace 1304 includes fewer spikes than the trace 1302 includes. That is, the trace 1304 is smoother than the trace 1302. In addition, the trace 1304 has a more defined dip than the trace 1306 has, which means that a more well defined dip is required to allow the endpoint decision logic to call the endpoint It is important because it is. Thus, a more well defined dip facilitates end point determination.

Figure 14 illustrates a method 1200 for using spectra to achieve a desired substrate profile. An expected end time to polish the product substrate is determined (step 1210). In some implementations, the set-up substrate is polished using predetermined process parameters (e.g., by conventional off-line metrology measurements) and the set-up substrate is determined to reach the desired thickness, The expected end point time is determined by using the polishing time at which the polishing-up substrate reaches the desired thickness as the estimated end point time.

Product substrate polishing begins (step 1218). Spectra are acquired at more than one radial position of the substrate (step 1226). For each spectral measurement, the radial position on the substrate can be determined and the spectral measurements binned into the zones based on their radial positions. The substrate may have multiple regions such as a center region, an intermediate region, and an edge region. In a 300 mm wafer, the central region can extend from the center to a radius of 50 mm, the middle region can extend from a radius of 50 mm to a radius of about 100 mm, and the edge can extend from about 100 mm to about 150 mm. In some implementations, the substrate may have more or less than three regions as described above. The location at which the spectra are obtained can be found in, for example, U.S. Patent Application No. 10 (1998), entitled " Determination of Position of Sensor Measurements During Polishing " filed on August 18, 2004, incorporated herein by reference for all purposes. / 922,110, or may be determined as described in U.S. Patent No. 7,018,271.

10A, the spectra from each region (or, for each region, the average of the spectra from within a region obtained from a single sweep of the sensor across the substrate) Are compared to the spectra (step 1234). From the comparison with the spectra library, the corresponding index index for each area is determined (step 1238).

Polishing is aborted if the indices for the regions meet one or more endpoint criteria. For example, when a desired index is reached for a preselected area, or when any of the areas first reaches a desired index, or when desired indices are achieved for all of the areas, (Step 1244). The desired index for each area is determined by the final desired profile for the substrate. If the substrate should have a flat profile or have a uniform oxide layer when polishing is completed, the spectra obtained in each area should be the same or nearly the same and each area will have the same or similar desired index index.

The polishing rate in the areas may be adjusted using a feedback loop such that the final index index in each area is equal to the desired final index index. FIG. 15 illustrates one method 1400 for adjusting the polishing process to achieve the desired substrate profile at the expected endpoint time. The desired index index at the expected endpoint time is determined for each area on the substrate (step 1402). Polishing may begin (step 1404) and the substrate may be optically monitored (step 1406) as described above so that index traces are determined for each area on the substrate. After the initial delay time, which allows the polishing process to stabilize, the rate of change of the index over time is calculated (the number of rotations of the platen can be used to represent time) (step 1408). The index variation rate may simply be calculated as a deviation from the indexes at two different times divided by the elapsed platen rotation rate between the spectral measurements that generated the indexes at different times. The rate of change of the index index represents the polishing rate. Typically, if none of the polishing parameters have been changed, it may be assumed that the polishing rate is steady.

Extrapolate (step 1412) the index traces extrapolated using the rate of change of the index for each region to determine the index exponent to be achieved at the expected endpoint time for the associated region. If, at the expected end point time, the desired index index has passed or has not yet reached, the polishing rate may be adjusted high or low as desired (step 1420). If the desired index index is reached at the expected endpoint time, no adjustment is required. A determination as to whether more than one extrapolated estimation and adjustment should be made may occur across the polishing sequence. Determining whether or not an adjustment of the polishing rate needs to be performed may include determining whether a desired index index is to be achieved when a polishing endpoint occurs or determining whether the last index falls within an acceptable range from a desired final index index ≪ / RTI >

In some implementations, the expected endpoint time is determined for one region, such as the central region. If necessary, the polishing rate in the remaining areas can then be adjusted such that the desired end point of the remaining areas is achieved, for example at the same time as the expected end point time for the selected area, such as the center area. The polishing rate can be adjusted, for example, by increasing or decreasing the pressure in the corresponding area in the carrier head. In some carrier heads, such as the carrier head described in US-A-2005-0211377, the carrier head has an adjustable pressure area. The variation in the polishing rate can be assumed to be directly proportional to the pressure change. (E.g., a simple Prestonian model). Additionally, the influence of other parameters affecting the platen or head rotational speed, the secondary effect of different head pressure combinations, the polishing temperature, the slurry flow, A control model for polishing the substrates may be developed.

The spectral-based endpoint determination logic described above in method 800 can also be used to determine a polishing endpoint and can be used in conjunction with adjusting the polishing process to achieve a desired substrate profile. From the provided equations in relation to step 806, the relative thickness for each area is determined using the deviation between the areas. As the substrate is polished, the spectra are acquired and binned into regions. Optionally, a single processing and filtering is applied to the spectra. A sum-of-squared-difference calculation is applied to the predetermined reference spectrum and the collected spectra for each region. The predetermined reference spectrum is the spectrum obtained when the polishing end point is reached.

When the square sum of the deviations for the reference spectrum approaches a minimum value in one area, the polishing pressure for the remaining areas is checked to determine whether the polishing rate should be changed in any one of the areas. The polishing rate in the region where the square sum of the deviations approaches the minimum value can be reduced and the polishing rate in the remaining regions can be increased. Further, the square sum of the deviations can be analyzed throughout the polishing, so that the adjustment to the polishing rate can be changed earlier in the polishing sequence. Unlike the method described in method 1400, the method does not require correlation between index exponents from the spectral libraries and polishing spectra.

16, if a specific profile, such as a uniform thickness, is required across the surface of the substrate, the slope of the polishing rate, as indicated by the change in index index over time, can be monitored and the polishing rate adjusted . After the polishing stabilization period 1505, a spectrum is obtained in the central region 1510, in the edge region 1515, and in the middle region 1520 therebetween. Here, the regions are circular or annular regions. Each spectrum is correlated to its respective index. This process is repeated over the number of platen revolutions, or over time, and the polishing rate is determined in each of the central region 1510, the middle region 1520, and the edge region 1515. [ The polishing rate is indicated by the slope of the line obtained by plotting the index 1530 (y-axis) according to the number of revolutions 1535 (x-axis). If any of the velocities appear to be faster or slower than the rest velocities, the velocity in that region can be adjusted. Here, the adjustment is based on the end point (C _E ) of the central region 1510. An approximate polish end point ("EDP") or an estimated end point time ("EET") is calculated for the central region after sufficient platen rotations or data points have been collected for the current substrate. ) Is determined. The EET is recalculated after each platen revolution. At the first polishing time (T ₁ ) during the polishing process, the polishing rate in the middle region 1520 is reduced and the polishing rate in the edge region is increased. Without adjusting the polishing rate in the middle area 1520, the middle area will be polished faster than the rest of the substrate and polished with excess polishing rate M _A. The edge region 1515 will be underpolished at the speed E _{u with} no adjustment of the polishing rate at the first polishing time T ₁ to the edge region 1515. [

At a subsequent time (T ₂ ) during the polishing process, if necessary, the speed can be adjusted again. The purpose in such a polishing process is to terminate polishing when the substrate has a relatively flat oxide layer over a flat surface or surface. One way to determine the amount to adjust the speed of polishing is to adjust the speed so that the index of each of the center, middle, and edge areas is equal at the approximate polishing end point (EDP). Thus, the polishing rate in the edge area requires adjustment, while the center area and middle areas are polished at the same rate as before T ₂ . The EDP is determined when the fitted line to the center region hits the desired ASL level. All other areas are controlled so that the fitted lines of all other areas are as close as possible and cross the ASL level at the same time.

Another way to use the spectral-based observation of the polishing rate to achieve a particular profile is to polish the first substrate, monitor the polishing rate, and subsequently feed-forward the polishing rate information to the substrate being polished. Referring to FIG. 17, the first set-up substrate is polished and the spectrum is acquired to determine the oxide thickness relative to the polishing rate in the central region 1610, the middle region 1620 and the edge region 1630. The starting indices for the middle area 1620, the center area 1610 and the edge area 1630 are M _O , C _O, and E _O, respectively. The center region 1610 has an endpoint (C _E ) spectrum that is selected to be the target spectrum. If at the end of the current wafer polishing, the other two regions have index indices within the threshold distance 1640 from the index of the center end CE, the edge region 1630 or the middle region 1620 of the subsequent wafer, No adjustment is made to the polishing rate of the polishing pad. Similarly, if polishing rates and index indices during polishing are within an allowable margin 1650, adjustments to edge area 1630 or intermediate area 1620 may not need to be made. At the end of polishing, the end point M _E for the middle area shows that the middle area is over-polished, and the end point E _E for the edge area indicates that the edge area is under-polished . As a result, for subsequent wafers, the polishing rate parameters will be adjusted such that the thickness at the end point for all the areas E, M, and C falls within an acceptable range 1640. Figure 18 shows the evolution of the expected signals for the E, M and C regions for the next wafer.

Referring to FIG. 18, it is desirable to change the polishing rate only a few times, for example, 4, 3, 2 or 1 times, when in situ adjustment is used during the polishing process. The adjustment may be made in the middle or at the end of the polishing process, adjacent to the beginning of the polishing process. Associating the spectra with the index index can create a linear comparison for polishing in each area, and can simplify the calculations required to control the polishing process and determine whether complex software or processing steps can be ruled out.

Spectra obtained from various regions of the substrate may represent the profile of the substrate, but do not necessarily provide an absolute thickness of the oxide layer. Thus, some of the spectral-based polishing rate adjustment methods described herein can be used to monitor the relative thickness of the oxide over the substrate. Because the spectral-based methods can be used to determine and adjust the polishing rate in regions of the substrate, the spectrum-based methods can also compensate for wafer induced thickness variation as well as substrate in- have.

As described herein, the relative thickness can be used to achieve the desired substrate profile. In some of the above examples, the desired substrate profile after polishing is a flat profile. However, a non-flat profile can also be achieved. Often, the substrate is polished on more than one platen. Some polishing processes are known to essentially polish one region faster than other regions. To compensate for this non-uniform polishing, polishing in the first platen can be controlled to leave a thicker region than other regions, such as, for example, a region to be polished faster on a subsequent platen. This thickness difference can be achieved by selecting a ratio between the ending index indexes for one area versus another, or a deviation in the target index indices.

All of the functional operations described herein as well as embodiments of the present invention may be implemented in digital electronic circuitry, or computer software, firmware, or hardware, including structural means disclosed herein and structural equivalents thereof Or a combination thereof. Embodiments of the present invention may be practiced with other computer system configurations, for example, for controlling the operation of a data processing apparatus, such as a programmable processor, a computer or a plurality of processors or computers, Such as one or more computer program products tangentially embodied in an information carrier, such as a readable storage device or a signal to be propagated. A computer program (also known as a program, software, software application, or code) may be written in any form of programming language, including a compiled or interpreted language, Which may be deployed in any form, including as a component, subroutine, or unit suitable for use in a computing environment. A computer program does not necessarily correspond to a file. A program may be stored in a portion of a file that has other programs or data, may be stored in a single file dedicated to the program in question, or may be stored in a plurality of coordinated files (e.g., one Or files that store portions of two or more modules, subprograms, or portions of code). A computer program may be deployed to run on one computer or a plurality of computers at a site, or may be distributed across a plurality of sites and interconnected by a communications network.

The processes and logic flows described herein may be performed by one or more programmable processors executing one or more computer programs to perform functions by generating operations and outputs to input data. have. Process and logic flows may also be performed by special purpose logic circuitry, such as, for example, a field programmable gate array (FPGA) or an application specific integrated circuit (ASIC) Purpose logic circuit such as a field programmable gate array (FPGA) or an application specific integrated circuit (ASIC).

The above-described polishing apparatuses and methods can be applied to various polishing systems. One or both of the polishing pad and the carrier head may move to provide relative motion between the polishing surface and the substrate. For example, instead of rotating the platen, it may orbit. The polishing pad may be a circular (or any other shape) pad fixed to the platen. Some aspects of the endpoint detection system can be applied to linear polishing systems (e.g., in the case where the polishing pad is a continuous or linearly moving reel-to-reel belt). The polishing layer may be a standard (e.g., polyurethane with or without fillers) polishing material, a soft material, or a fixed-abrasive material. Terms of relative positioning are used; It should be understood that the polishing surface and substrate can be maintained in a vertically oriented state or in any other oriented state.

Specific embodiments of the invention have been described. Other embodiments are within the scope of the following claims. For example, the operations described in the claims may be performed in a different order and still achieve desirable results.

Claims (15)

A method of spectral-based monitoring of chemical mechanical polishing,
Obtaining a sequence of current spectra from a substrate being polished;
Determining, for each current spectrum in the sequence, whether the current spectrum corresponds to a first region of the substrate based on the current spectrum; And
Determining a polishing end point using current spectra that are not determined to correspond to the first region
Containing
Spectral - based monitoring method of chemical mechanical polishing. The method according to claim 1,
The first area is a scribe line
Spectral - based monitoring method of chemical mechanical polishing. 3. The method according to claim 1 or 2,
Wherein determining based on the current spectrum comprises classifying the current spectrum into one of a plurality of groups based on the shape of the current spectrum
Spectral - based monitoring method of chemical mechanical polishing. The method of claim 3,
Wherein a first one of the plurality of groups comprises current spectra from a scribe line of the substrate and a second one of the plurality of groups comprises a current spectrum from an array of substrates
Spectral - based monitoring method of chemical mechanical polishing. The method according to claim 1,
Further comprising averaging current spectra that are not determined to correspond to the first region
Spectral - based monitoring method of chemical mechanical polishing. As a spectral-based monitoring method of chemical mechanical polishing,
Storing one or more normalized predetermined spectra;
Obtaining a sequence of current spectra from a substrate being polished;
Normalizing each current spectrum in the sequence to produce a sequence of normalized current spectra; And
Determining a policing endpoint using the at least one normalized predetermined spectrum and the normalized current spectra, and determining a difference between the normalized current spectra and the at least one normalized predetermined spectrum, Determining an end point;
Containing
Spectral - based monitoring method of chemical mechanical polishing. The method according to claim 6,
The outermost layer of the substrate is polished,
The normalizing step includes dividing the spectrum of the lower layer below the outermost layer
Containing
Spectral - based monitoring method of chemical mechanical polishing. 8. The method of claim 7,
The lower layer is a silicon layer
Spectral - based monitoring method of chemical mechanical polishing. 9. The method according to any one of claims 6 to 8,
The normalizing step includes subtracting a spectrum indicating that the substrate is not positioned on a sensor of an in-situ optical monitoring system
Containing
Spectral - based monitoring method of chemical mechanical polishing. 10. The method of claim 9,
The normalizing step is a spectrum obtained when A is the current spectrum and Dark is no substrate located on top of the in-situ optical monitoring system, and Si is the spectrum obtained from the bare silicon substrate, -Dark) / (Si-Dark)
Containing
Spectral - based monitoring method of chemical mechanical polishing. A method of implementing spectral-based monitoring of chemical mechanical polishing,
Storing one or more reference spectra;
Obtaining a sequence of current spectra from a substrate being polished;
Normalizing each current spectrum in the sequence to produce a sequence of the same normalized current spectra such that the amplitude of each current spectrum is equal to or similar to the amplitude of the reference spectrum; And
Determining a policing endpoint using the one or more reference spectra and the normalized current spectra, the method comprising: determining a polishing endpoint comprising determining a difference between the normalized current spectra and the one or more reference spectra;
Containing
Method of implementing spectral based monitoring of chemical mechanical polishing. 12. The method of claim 11,
The amplitude is the peak-to-low point intensity deviation
Containing
Method of implementing spectral based monitoring of chemical mechanical polishing. 12. The method of claim 11,
Further comprising receiving a range indicative of a portion of the spectrum,
The sequence of normalized current spectra is such that the amplitude of each current spectrum within the range is equal to or similar to the amplitude of the reference spectrum within the range
Method of implementing spectral based monitoring of chemical mechanical polishing. 14. The method of claim 13,
The normalizing step allows the high and low points in the range from each normalized spectrum to be normalized to 1 and 0
Method of implementing spectral based monitoring of chemical mechanical polishing. 15. The method of claim 14,
The normalizing step comprises the steps of: g is a gain, h is an offset, r _max is a highest value within the range of the current spectrum, r _min is a minimum value of the range of the current spectrum, R is a current spectrum and N is a normalized spectrum time,
calculating g = (1-0) / (r _max -r _min );
calculating h = 1 - r _max g; And
Calculating N = Rg + h
Containing
Method of implementing spectral based monitoring of chemical mechanical polishing.

Images