TWI467678B - Dynamically or adaptively tracking spectrum features for endpoint detection - Google Patents

Description

Spectral features for dynamic or adaptive tracking of endpoint detection

This disclosure relates to optical monitoring during chemical mechanical polishing of substrates.

The integrated circuit is usually formed on the substrate by sequentially depositing a conductive layer, a semiconductive layer or an insulating layer on the germanium wafer. One manufacturing step involves depositing a filler layer on a non-planar surface and planarizing the filler layer. For some applications, the filler layer is planarized until the top surface of the patterned layer is exposed. For example, a layer of conductive filler can be deposited over the patterned insulating layer to fill the trenches or holes in the insulating layer. After planarization, vias, plugs, and wires are formed in portions of the remaining conductive layer between the raised patterns of the insulating layer, and the vias, plugs, and wires provide a conductive path between the thin film circuits on the substrate. For other applications, such as oxide polishing, the filler layer is planarized until a predetermined thickness is left on the non-planar surface. In addition, photolithography generally requires planarizing the surface of the substrate.

Chemical mechanical polishing (CMP) is an acceptable planarization method. This planarization method typically requires mounting the substrate on a carrier head or a polishing head. The exposed surface of the substrate is typically placed against a rotating polishing pad. The carrier head provides a controllable load on the substrate to push the substrate against the polishing pad. The abrasive slurry is typically supplied to the surface of the polishing pad.

One problem with CMP is to determine if the polishing process is complete (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. Slurry distribution, polishing pad conditions, relative speed between the polishing pad and the substrate, and changes in load on the substrate can all cause changes in material removal rate. These changes and changes in the initial thickness of the substrate layer cause a change in the time required to reach the end of the polishing. Therefore, the end point of the grinding cannot be determined solely as a function of the grinding time.

In some systems, the substrate is optically monitored in situ during polishing, for example, via a window in the polishing pad. However, existing optical monitoring technologies may not meet the increased needs of semiconductor device manufacturers.

Some optical endpoint detection systems track selected spectral signature characteristics in spectral measurements to determine endpoints or to change a polishing rate. In a spectrum, spectral features similar to the selected spectral features can make tracking the selected spectral features difficult. Identifying one of the wavelength ranges of the optical endpoint detection system to search for the selected spectral signature allows the optical endpoint detection system to correctly identify the selected spectral signature and use reduced processing resources.

In some polishing processes, a second layer (eg, a barrier layer) of a second material (eg, a nitride such as tantalum nitride or titanium nitride) is removed from a substrate to expose a A first layer or layer structure of a different first material (eg, a dielectric material, a low dielectric value material, and/or a low dielectric value cap material). It is often desirable to remove the first material until a target thickness remains. Some optical endpoint detection techniques that track a selected spectral feature in a spectral measurement to determine an endpoint or change a polishing rate can be problematic in the polishing process because the initial thickness of the second material is not known. . However, if the spectral feature tracking is triggered by another monitoring technique (eg, motor torque, eddy current or optical intensity monitoring), another monitoring technique can reliably detect the removal of the second material and the exposure of the underlying layer or layer structure. , you can avoid these problems. In addition, there may be variations in the thickness of the layer or layer structure between the substrates. To increase the uniformity between the substrates of the final thickness of the layer or layer structure, the initial thickness of the layer or layer structure can be measured prior to grinding, and a target eigenvalue can be calculated from the initial thickness and the target thickness.

In one aspect, a method of controlling polishing includes the steps of: grinding a substrate; and receiving a selected one of the selected spectral features, having a wavelength range of one width, and one of the selected spectral features for performing during the grinding Monitoring. A series of spectra of light from the substrate is measured while the substrate is being ground. A series of values of the characteristic of the selected spectral feature is generated from the series of spectra. The generating step includes the steps of: generating, for the at least some of the spectra from the series of spectra, a modified wavelength range based on the spectral feature at a location within a previous wavelength range, searching for the selected spectral feature within the modified wavelength range And determining a value of one of the characteristics of the selected spectral feature, the previous wavelength range being used for one of the previous spectra in the series of spectra. At least one of a polishing endpoint or an adjustment for one of the polishing rates is determined based on the series of values.

Embodiments may include one or more of the following features. This wavelength range can have a fixed width. The step of generating the modified wavelength range can include the step of centering the fixed width at the location of the characteristic in the previous wavelength range. The step of generating the modified wavelength range may include the steps of determining a position of the characteristic in the previous wavelength range and adjusting the wavelength range such that in the modified wavelength range, the characteristic is positioned closer to the modified wavelength range One of the centers. The step of generating the modified wavelength range can include the steps of: determining a wavelength value of the selected spectral feature for at least some of the series of spectra to generate a series of wavelength values; fitting a function to the series of wavelength values; An expected wavelength value for the selected spectral signature for a subsequent spectral measurement is calculated based on the function. This function can be a linear function. The step of generating the modified wavelength range can include the step of centering the width of the wavelength range at the expected wavelength value. The method can include the steps of fitting a function to the series of values and determining at least one of a polishing endpoint or an adjustment for a polishing rate based on the function. The step of determining a polishing end point may include the steps of: calculating an initial value of the characteristic according to the function, calculating a current value of the characteristic according to the function, and calculating a difference between the initial value and the current value, and when When the difference reaches a target difference, the grinding is interrupted. This function can be a linear function. The selected spectral features can include: a spectral peak, a spectral trough, or a spectral zero crossing. The characteristic can include: a wavelength, a width, or an intensity. The selected spectral signature can include a spectral peak and the characteristic can include a peak width. The spectrum of visible light can be measured and the wavelength range can have a width between 50 and 200 nanometers.

In another aspect, a method of controlling polishing includes the steps of: receiving a user input selecting a fixed wavelength range that is a subset of wavelengths measured by an in situ monitoring system; receiving a selection One of the spectral features identifies and one of the selected spectral features to monitor during polishing; grinding a substrate; for each spectrum in the series of spectra, measuring a series of light from the substrate while grinding the substrate Generating a selected spectral characteristic of the fixed wavelength range of the respective spectra and determining a value of one of the characteristics of the selected spectral feature to generate a series of values; and determining a polishing endpoint or a polishing rate based on the series of values At least one of the adjustments.

Embodiments may include one or more of the following features. The in-situ monitoring system can measure the intensity of at least the wavelength of visible light, and the fixed wavelength range can have a width of between 50 and 200 nanometers. The selected spectral signature can be a spectral peak, a spectral valley, or a spectral zero crossing. The characteristic can be a wavelength, a width or an intensity.

In another aspect, a method of controlling polishing includes: polishing a substrate having a first layer; receiving a characteristic of a selected spectral feature and characteristic of the selected spectral feature for monitoring during grinding; Measuring a series of spectra of light from the substrate while grinding the substrate; determining a first value of the characteristic of the feature at a time of exposure of the first layer; adding an offset to the first value And generating a second value; and monitoring the characteristic of the feature, and suspending the grinding when determining that the characteristic of the feature reaches the second value.

Embodiments may include one or more of the following features. This characteristic can be a position, width or intensity. The selected feature may sustain an evolutionary site, width or intensity under the entire spectrum of the spectrum. This feature can be one of the peaks or troughs of the spectrum. The substrate can include a second layer covering the first layer, and the step of grinding can include the steps of: grinding the second layer and detecting the first layer of exposure with an in situ monitoring system. The first value can be determined when the first in-situ monitoring technique detects the exposure of the first layer. The step of detecting the exposure of the first layer can be a process separate from the step of monitoring the characteristic of the feature. The step of detecting the exposure of the first layer can include the step of monitoring the total reflected intensity from one of the substrates. The step of monitoring the total reflected intensity can include the step of integrating the spectrum over a range of wavelengths for each of the series of spectra to produce the total reflected intensity. The in-situ monitoring system can include a motor torque or friction monitoring system. The first value can be determined during the grinding of the first layer (e.g., immediately after initiation of the grinding of the first layer). The first layer can be exposed before the initiation of the polishing of the substrate. The step of monitoring the characteristic of the feature can include the step of determining a value of the characteristic for each of the spectra from the series of spectra to produce a series of values. The characteristic of the feature is determined to reach the second value by fitting a linear function to the series of values and determining that the linear function is equal to one of the endpoint times of the second value. A pre-polishing thickness of the first layer may be received, and the offset value may be calculated based on the pre-polishing thickness. The step of calculating the offset value ΔV may include the step of: calculating (D ₂ -d _T )/(dD/dV), where d _T is a target thickness, and D ₁ is from a first layer of a mounting substrate a thickness before polishing, D ₂ after grinding the first layer from a substrate of one of the mounted thickness and dD / dV is the change in thickness as one of the characteristic function of the rate. The step of calculating the offset value ΔV may include the step of calculating ΔV=ΔV _D +(d ₁ -D ₁ )/(dD/dV)+(D ₂ -d _T )/(dD/dV), where d ₁ For the pre-polishing thickness, D ₁ is a pre-polishing thickness from one of the first layers of a mounting substrate, and ΔV _D is between the pre-polishing thickness of the first layer of the mounting substrate and the post-polishing thickness. A difference in the value of the characteristic of the feature. The pre-grinding thickness d ₁ can be measured at a separate measuring station. The rate of change dD/dV of the thickness as a function of the characteristic may be a rate of change of thickness near one of the ends of the polishing. The first layer may comprise polycrystalline germanium and/or a dielectric material, for example consisting of substantially pure polycrystalline germanium, composed of a dielectric material or a combination of polycrystalline germanium and one of dielectric materials.

Embodiments may include one or more of the following advantages as desired. Identifying a range of wavelengths to search for selected spectral features allows for greater accuracy in detecting endpoints or determining a change in polishing rate. For example, the system is less likely to select an incorrect spectrum during subsequent spectral measurements. feature. Tracking spectral features over a range of wavelengths rather than over an entire spectrum allows for easier and faster identification of such spectral features. The processing resources needed to identify these selected spectral features can be reduced.

It can reduce the time for a semiconductor manufacturer to develop an algorithm that detects the end of a particular product substrate. Spectral feature tracking can be applied to a polishing operation that begins with a polishing of a reflective layer, and wafer-to-wafer thickness uniformity (WTWU) can be improved. The initial thickness of the layer can be measured prior to grinding, and a target feature value can be calculated based on the initial thickness and the target thickness to provide a more accurate endpoint decision.

The details of one or more embodiments are set forth in the accompanying drawings and description. Other aspects, features, and advantages will be apparent from the description and drawings.

One optical monitoring technique is to measure the spectrum of light reflected from the substrate during grinding and to identify matching reference spectra from the library. One potential problem with the spectral matching approach is that for some types of substrates, there are significant inter-substrate differences in the underlying grain features, resulting in a change in the spectrum of the substrate reflection from the surface having the same outer layer thickness. These changes increase the difficulty of proper spectrum matching and reduce the reliability of optical monitoring.

One technique to counteract this problem is to measure the spectrum of light reflected from the substrate being ground and to identify changes in spectral characteristic characteristics. Variations in the characteristics of the tracking spectrum (eg, the wavelength of the spectral peaks) allow for better uniformity of polishing between the substrates within the batch. By determining the target difference of the spectral characteristic characteristics, the end point can be called when the value of the characteristic has changed the target amount.

The substrate may be only a single dielectric layer disposed on the semiconductor layer, or have a significantly more complex layer stack. For example, the substrate can include a first layer and a second layer disposed on the first layer. The first dielectric layer may be, for example, oxides (such as silicon dioxide), or low-k dielectrics (low-k) materials, such as carbon-doped silicon dioxide, e.g., Black Diamond ^TM (from Applied Materials, Inc.) or Coral ^TM (from Novellus Systems, Inc.). The second layer may be a barrier layer, and the composition of the barrier layer is different from the first layer. For example, the barrier layer can be a metal or a metal nitride, such as tantalum nitride or titanium nitride. One or more additional layers are disposed between the first layer and the second layer, as desired, for example, a low dielectric value covering material, for example, a material formed of tetraethyl orthosilicate (TEOS). Both the first layer and the second layer are at least translucent. The first layer, together with one or more additional layers, if present, provides a layer stack below the second layer. However, in some embodiments, only a single layer containing, for example, polysilicon and/or dielectric is ground (although there may be additional layers below the layer being ground).

Chemical mechanical polishing can be used to planarize the substrate until the second layer is exposed. For example, if an opaque conductive material is present, the opaque conductive material can be ground until the second layer (eg, barrier layer) is exposed. Thereafter, portions of the second layer remaining on the first layer are removed and the substrate is polished until the first layer (eg, dielectric layer) is exposed. In addition, it is sometimes desirable to grind the first layer (eg, the dielectric layer) until the target thickness is left or the amount of target material has been removed.

One method of grinding is to polish the conductive layer on the first polishing pad at least until the second layer (eg, the barrier layer) is exposed. Additionally, a portion of the thickness of the second layer can be removed, for example, at the first polishing pad during the overgrinding step. Thereafter, the substrate is transferred to a second polishing pad wherein the second layer (eg, barrier layer) is completely removed and a portion of the thickness of the underlying first layer (eg, low dielectric value dielectric) is also removed. Additionally, if there is one or more additional layers between the first layer and the second layer, it can be removed in the same polishing operation at the second polishing pad.

However, when the substrate is transferred to the second polishing pad, the initial thickness of the second layer may not be known. As noted above, this situation can cause problems with optical endpoint detection techniques that track selected spectral feature characteristics in spectral measurements to determine the endpoint at the target thickness. However, this problem can be alleviated by triggering spectral feature tracking by another monitoring technique that can reliably detect the removal of the second layer and the exposure of the underlying first layer or layer structure. In addition, by measuring the initial thickness of the first layer and calculating the target feature value based on the initial thickness of the first layer and the target thickness, the inter-substrate uniformity of the thickness of the first layer can be improved.

Spectral features may include spectral peaks, spectral troughs, spectral inflection points, or spectral zero crossings. Characteristics of the features may include wavelength, width or intensity.

FIG. 1 illustrates a polishing apparatus 20 operable to polish a substrate 10. The grinding apparatus 20 includes a rotatable disc shaped platform 24 on which the polishing pad 30 is positioned. The platform is operable to rotate about the axis 25. For example, the motor can rotate the drive shaft 22 to rotate the platform 24. For example, the polishing pad 30 can be detachably secured to the platform 24 by an adhesive layer. The polishing pad 30 is detachable and replaceable when worn. The polishing pad 30 can be a two-layer polishing pad having an outer polishing layer 32 and a softer backing layer 34.

The optical access point 36 is provided through the polishing pad in a manner that includes an aperture (i.e., a hole through the pad) or a solid window. The solid window can be secured to the polishing pad, however in some embodiments the solid window can be supported on the platform 24 and protrude into the aperture in the polishing pad. The polishing pad 30 is typically placed on the platform 24 such that the aperture or window covers the optical head 53 positioned in the recess 26 of the platform 24. The optical head 53 can thus optically access the ground substrate via an aperture or window.

For example, the window can be a rigid crystalline or vitreous material (eg, quartz or glass), or a softer plastic material (eg, a silicone resin, a polyurethane, or a halogenated polymer (eg, a fluoropolymer) ), or a combination of materials mentioned. The window can be transparent to white light. If the top surface of the solid window is a rigid crystalline or vitreous material, the top surface should be sufficiently recessed from the abrasive surface to prevent scratching. If the top surface is close and accessible to the abrasive surface, the top surface of the window should be a softer plastic material. In some embodiments, the solid window is secured in 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 monochromatic light of a particular color (eg, blue or red light), for example, approximately 80% transmittance. The window may be sealed to the polishing pad 30 such that the liquid does not leak through the interface of the window and the polishing pad 30.

In one embodiment, the window comprises a rigid crystalline or vitreous material covered with an outer layer of a softer plastic material. The top surface of the softer material can be coplanar with the abrasive surface. The bottom surface of the rigid 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 integrated into the abrasive layer and the bottom layer can have an aperture aligned with the solid window.

The bottom surface of the window may include one or more grooves as desired. The recess can be shaped to accommodate, for example, the end of the fiber optic cable or the end of the eddy current sensor. The recess allows the end of the fiber optic cable or the end of the eddy current sensor to be positioned at a distance from the surface of the substrate being ground that is less than the thickness of the window. In the case where the window comprises a rigid crystalline portion or a glassy portion, and the recess is formed by machining into an embodiment in this portion, the recess is ground to remove scratches caused by machining. Alternatively, a solvent and/or liquid polymer can be applied to the surface of the recess to remove scratches caused by machining. The removal of scratches typically caused by machining reduces scattering and increases the transmission of light through the window.

The backing layer 34 of the polishing pad can be attached to the polishing layer 32 outside of the polishing pad, for example, by an adhesive. An aperture providing an optical access point 36 can be formed in the pad 30 (eg, by cutting or by modeling the pad 30 to include an aperture), and the window can be inserted into the aperture and secured to the pad 30, for example, by Adhesive. Alternatively, the liquid precursor of the window can be dispensed into the aperture in the pad 30 and allowed to cure to form a window. Alternatively, a solid transparent element (eg, the crystalline or glassy portion described above) can be positioned in the liquid pad material and the liquid pad material can be cured to form the pad 30 around the transparent element. In either of the latter two conditions, a mat of material can be formed and the layer of the polishing pad containing the modeling window can be cut from the block.

The grinding apparatus 20 includes a combined slurry/flushing arm 39. During milling, the arm 39 is operable to dispense a slurry 38 containing a liquid and a pH indicator. Alternatively, the grinding apparatus includes a slurry crucible that is operable to dispense the slurry onto the polishing pad 30.

The grinding 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 a support structure 72 (e.g., a rotating rack) and coupled to the carrier head rotation motor 76 by a carrier drive shaft 74 such that the carrier head is rotatable about the shaft 71. Additionally, the carrier head 70 can vibrate laterally in a radial slot formed in the support structure 72. In operation, the platform rotates about the platform center axis 25 and the carrier head rotates about the carrier head central axis 71 and translates laterally across the top surface of the polishing pad.

The grinding apparatus also includes an optical monitoring system that can be used to determine the grinding end point as discussed below. The optical monitoring system includes a light source 51 and a photodetector 52. Light is transmitted from the source 51, passes through the optical access point 36 in the polishing pad 30, collides and passes through the optical access point 36, and is reflected back from the substrate 10 and travels to the photodetector 52.

The bifurcated fiber optic cable 54 can be used to transmit light from the source 51 to the optical access point 36 and back to the photodetector 52 back from the optical access point 36. The split-fork fiber optic cable 54 can include a "trunk" 55 and two "spurs" 56 and 58.

As mentioned above, the platform 24 includes a recess 26 in which the optical head 53 is positioned. The optical head 53 holds one end of the main line 55 of the bifurcated fiber cable 54, and the bifurcated fiber cable 54 is configured to conduct light to and from the surface of the substrate being polished. The optical head 53 can include one or more lenses or windows that cover the ends of the bifurcated fiber optic cable 54. Alternatively, the optical head 53 can only hold the end of the main line 55 adjacent to the solid window in the polishing pad. The optical head 53 can be removed from the recess 26 as needed, for example, to achieve preventive maintenance or corrective maintenance.

The platform includes a removable in-situ monitoring module 50. The in-situ monitoring module 50 can include one or more of the following: a light source 51, a photodetector 52, and a circuit for transmitting and receiving signals to and from the light source 51 and the photodetector 52. For example, the output of the detector 52 can be a digital electronic signal that is transmitted to a controller of the optical monitoring system via a rotary coupler (eg, a slip ring) in the drive shaft 22. Similarly, the light source can be turned on or off in response to a control command in the digital electronic signal transmitted from the controller to the module 50 via the rotary coupler.

The in-situ monitoring module 50 can also hold the respective ends of the branch portions 56 and 58 of the bifurcated fiber 54. The light source is operable to transmit light that is conducted via the branch line 56 and is conducted from the end of the main line 55 located in the optical head 53 and impinges on 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 conducted to the photodetector 52 via the branch line 58.

In one embodiment, the bifurcated fiber cable 54 is a bundle of fibers. The bundle includes a first set of fibers and a second set of fibers. The fibers in the first set are connected to conduct light from the source 51 to the surface of the substrate being polished. The fibers in the second set are coupled to receive light reflected from the surface of the substrate being polished and to conduct the received light to the photodetector 52. The fibers can be arranged such that the fibers in the second set form an X-like shape centered on the longitudinal axis of the bifurcated fiber 54 (when viewed in cross section of the bifurcated fiber cable 54). Alternatively, other arrangements can be implemented. For example, the fibers in the second group can form a V-like shape that is mirror images of each other. Suitable bifurcated fibers are available from Verity Instruments, Inc., Carrollton, Texas.

There is typically an optimum distance between the polishing pad window and the end of the main line 55 of the bifurcated fiber cable 54 that is closest to the polishing pad window. This distance can be determined empirically and is affected by, for example, the reflective nature of the window, the shape of the beam emitted from the bifurcated fiber optic cable, and the distance from the substrate being monitored. In one embodiment, the bifurcated fiber optic cable is positioned such that the end closest to the window is as close as possible to the bottom of the window without actually touching the window. In the case of this embodiment, the polishing apparatus 20 can include a mechanism (e.g., as part of the optical head 53) that is operable to adjust between the end of the bifurcated fiber optic cable 54 and the bottom surface of the polishing pad window. distance. Alternatively, the closest end of the split fiber cable 54 is embedded in the window.

Light source 51 is operable to emit white light. In one embodiment, the emitted white light comprises light having a wavelength of from 200 to 800 nanometers. Suitable for the light source is a xenon lamp or a mercury lamp.

The photodetector 52 can be a spectrometer. A spectrometer is basically an optical instrument for measuring the properties (eg, intensity) of light over a portion of the electromagnetic spectrum. A suitable spectrometer is a grating spectrometer. The typical output of a spectrometer is the intensity of light, which is a function of wavelength.

The light source 51 and the photodetector 52 are connected to an operable computing device to control the operation of the light source 51 and the photodetector 52, and receive signals from the light source 51 and the photodetector 52. The computing device can include a microprocessor positioned adjacent to the grinding device, such as a personal computer. With regard to control, the computing device can, for example, synchronize the activation of the light source 51 with the rotation of the platform 24. As shown in FIG. 2, the computer can cause the light source 51 to emit a series of flashes that begin just before the substrate 10 passes over the home position monitoring module 50 and just after the substrate 10 has passed the home position monitoring module 50. Each of the dots 201-211 represents a spot on which the light from the in-situ monitoring module 50 impinges on the substrate 10 and is reflected from the substrate 10. Alternatively, the computer can cause the light source 51 to continuously emit light that begins just before the substrate 10 passes over the in-situ monitoring module 50 and ends just after the substrate 10 has passed the in-situ monitoring module 50.

A series of spectra is provided during the grinding process, for example, from the spectrum obtained by continuous pulsing of the inductors on the substrate from the platform. In some embodiments, light source 51 emits a series of light flashes onto portions of substrate 10. For example, the light source can emit a flash of light onto a central portion of the substrate 10 and a portion outside the substrate 10. Light reflected from substrate 10 can be received by photodetector 52 to determine a plurality of series of spectra from portions of substrate 10. The features are identifiable in a spectrum in which each feature is associated with a portion of the substrate 10. For example, features can be used to determine the endpoint conditions for the polishing of substrate 10. In some embodiments, monitoring of portions of substrate 10 allows for varying the polishing rate on one or more portions of substrate 10.

Regarding the received signal, the computing device can receive, for example, a signal carrying information describing the spectrum of light received by the optical side detector 52. Figure 3A illustrates an example of a spectrum measured from the amount of light reflected from a substrate from a single flash of light from a source. Spectrum 302 is measured based on light reflected from the product substrate. The spectrum 304 is measured from light reflected from a substrate 矽 substrate which is a wafer having only a ruthenium layer. The spectrum 306 is light received by the optical head 53 in the absence of a substrate positioned on the optical head 53. Under this condition (referred to as dark conditions in this specification), the received light is typically ambient light.

The computing device can process the signal or a portion of the signal to determine the end of the grinding step. Without being limited to any particular theory, the spectrum of light reflected from substrate 10 evolves as the grinding progresses. Figure 3B provides an example of the evolution of the spectrum as it progresses to the polishing of the film of interest. Different spectral lines represent different points in the polishing process. As can be seen, as the thickness of the film changes, the nature of the spectrum of the reflected light changes and the particular spectrum is exhibited by the particular thickness of the film. When the grinding of the film is performed, the peaks in the spectrum of the reflected light are observed (i.e., local maximum), the height of the peak typically changes, and as the material is removed, the peak tends to broaden. In addition to broadening, the wavelength at which a particular peak is located generally increases as the milling progresses. In some embodiments, the wavelength at which a particular peak is located generally decreases as the milling progresses. For example, peak 310(1) illustrates the peak in the spectrum at a particular time during the grinding, while peak 310(2) illustrates the same peak at a later time during the grinding. The peak 310(2) is located at a longer wavelength and is wider than the peak 310(1).

The relative variation in the wavelength and/or width of the peak can be used according to empirical formulas (eg, the width measured at a fixed distance below the peak, or the width measured at an intermediate height between the peak and the nearest trough), The absolute wavelength and/or width of the peak, or both, determine the end of the polishing. The optimum peak (or multiple peaks) used in determining the endpoint will vary depending on the material being ground and the pattern of the materials.

In some embodiments, a change in peak wavelength can be used to determine the endpoint. For example, the polishing apparatus 20 may stop grinding the substrate 10 when the difference between the initial wavelength of the peak and the current wavelength of the peak reaches a target difference. Alternatively, features other than the crests may be used to determine the difference in wavelength of the light reflected from the substrate 10. For example, the wavelength, inflection point, or x-axis or y-axis intercept of the trough can be monitored by photodetector 52, and polishing apparatus 20 can stop grinding substrate 10 when the wavelength has changed by a predetermined amount.

In some embodiments, in addition to the wavelength, the monitored characteristic can be the width or intensity of the feature or not. Features can be offset by a sequence of approximately 40 nm to 120 nm, although other offsets are also possible. For example, the upper limit can be much larger, especially in the case of dielectric grinding.

FIG. 4A provides an example of a spectrum 400a measured from the amount of light reflected from the substrate 10. The optical monitoring system can pass spectrum 400a through a high pass filter to reduce the overall slope of the spectrum, resulting in spectrum 400b as shown in Figure 4B. For example, there may be a large spectral difference between wafers during processing of multiple substrates in a batch. A high pass filter can be used to normalize the spectrum to reduce spectral variations on the substrate in the same batch. An exemplary high pass filter can have a cutoff frequency of 0.005 Hz and a filter order 4 . The high-pass filter is not only used to help filter out sensitivity to underlying changes, but is also used to "flatten" the legitimate signals to make feature tracking easier.

In order for the user to select which feature of the end point will be tracked to determine the end point, a contour map can be generated and displayed to the user. Figure 5B provides an example of a contour map 500b generated from a plurality of spectral measurements of light reflected from the substrate 10 during polishing, and Figure 5A provides an example of a measured spectrum 500a from a particular transient in the contour map 500b. . Contour map 500b includes features such as peak region 502 and trough region 504 generated by correlation peaks 502 and valleys 504 on spectrum 500a. Over time, the substrate 10 is ground and the light reflected from the substrate changes as illustrated by the change in spectral characteristics in the contour map 500b.

To generate the contour map 500b, a test substrate can be ground and the light reflected from the test substrate can be measured by the photodetector 52 during the polishing to produce a series of spectra of light reflected from the substrate 10. The series of spectrum can be stored, for example, in a computer system that can be part of an optical monitoring system as needed. Grinding of the mounting substrate can begin at time T1 and continue beyond the estimated endpoint time.

When the grinding of the test substrate is complete, the computer, for example on a computer screen, presents a contour map 500b to the operator of the polishing apparatus 20. In some embodiments, for example, by assigning red to a higher intensity value in the spectrum, blue is assigned to a lower intensity value in the spectrum, and an intermediate color (orange to green) is assigned to an intermediate intensity in the spectrum. Value, computer color mark contour map. In other embodiments, the darkest shades of gray are assigned to lower intensity values in the spectrum, and the brightest shades of gray are assigned to higher intensity values in the spectrum, and intermediate shadows are assigned to intermediate intensities in the spectrum. The value, and the computer produces a grayscale contour map. Alternatively, the computer can generate a three-dimensional contour map in which the maximum z value is used to represent the higher intensity value in the spectrum, and the minimum z value is used to represent the lower intensity value in the spectrum, and the intermediate z value is used to represent the intermediate value in the spectrum. For example, a three-dimensional contour map can be displayed in color, grayscale, or black and white. In some embodiments, an operator of the grinding apparatus 20 can interact with a three-dimensional contour map to view different features of the spectrum.

For example, a contour map 500b of reflected light generated from the monitoring of the test substrate during polishing may contain spectral features such as peaks, troughs, spectral zero crossings, and inflection points. Features may have characteristics such as wavelength, width, and/or intensity. As shown by the contour map 500b, when the polishing pad 30 removes material from the top surface of the mounting substrate, the light reflected from the mounting substrate can change over time, and thus the characteristic characteristics change over time.

Prior to the grinding of the device substrate, the operator of the polishing apparatus 20 can observe the contour map 500b and select characteristic characteristics for tracking during processing with a batch of substrates having similar grain characteristics as the mounting substrate. For example, an operator of the grinding apparatus 20 can select the wavelength of the peak 506 for tracking. A potential advantage of the contour map 500b (especially for color markers or three-dimensional contour maps) is that such graphical displays make it easier for the user to select the appropriate features, since the features (eg, features having characteristics that vary linearly over time) are visually It is easily distinguishable.

To select the endpoint criteria, the characteristics of the selected features can be calculated by linear interpolation based on the pre-polished thickness and the post-polishing thickness of the test substrate. For example, the thicknesses D1 and D2 of the layers on the test substrate can be tested before grinding (for example, the thickness of the substrate before the start of the polishing T1) and after the grinding (for example, after the end of the grinding time T2) The thickness of the substrate is measured and the value of the characteristic can be measured at time T' at which the target thickness D' is reached. T' can be calculated from T' = T1 + (T2-T1) * (D2-D') / (D2-D1), and the value of the characteristic V' can be determined based on the spectrum measured at time T'. The target difference δV of the characteristic of the selected feature, such as a particular change in the wavelength of the peak 506, can be determined from V'-V1, where V1 is the initial characteristic value (at time T1). Therefore, the target difference δV may be a change from the initial characteristic value V1 before the grinding at time T1 to the value V' of the characteristic at the time T' at which the grinding is expected to be completed. The operator of the grinding apparatus 20 can input a target difference 604 (e.g., δV) of the characteristic characteristics to be changed into the computer associated with the grinding apparatus 20.

To determine the value V' and determine the value of point 602 accordingly, a robust line fit can be used to fit the vector fit data line 508. The value of line 508 at time T' can be subtracted from the value of line 508 at T1 to determine point 602.

Features such as spectral peaks 506 can be selected based on the correlation between the target difference of the characteristic characteristics and the amount of material removed from the mounting substrate during grinding. The operator of the grinding apparatus 20 can select different features and/or characteristic characteristics to find characteristic characteristics of a good correlation between the target difference of the characteristics and the amount of material removed from the mounting substrate.

In other embodiments, the endpoint decision logic determines the spectral characteristics and endpoint criteria to be tracked.

Turning now to the grinding of the device substrate, Figure 6A is an exemplary graph 600a of the difference 602a-d of characteristic characteristics tracked during the grinding of the device substrate 10. The substrate 10 can be part of a batch of substrates that are being ground, wherein an operator of the polishing apparatus 20 selects characteristic characteristics (such as wavelengths of peaks or troughs) to track according to the contour map 500b of the mounting substrate.

When the substrate 10 is polished, the photodetector 52 measures the spectrum of the light reflected from the substrate 10. The endpoint decision logic uses the spectrum of light to determine the series of values of the characteristic characteristics. As the material is removed from the surface of the substrate 10, the value of the selected characteristic can be varied. The difference 602a-d is determined using the difference between the series of characteristic characteristics and the initial value V1 of the characteristic.

When the substrate 10 is being polished, the endpoint decision logic can determine the current value of the characterized characteristic being tracked. In some embodiments, the endpoint is invoked when the current value of the feature has changed the target difference 604 from the initial value. In some embodiments, line 606 is fitted to difference values 602a-d, for example, using a robust line fit. The function of line 606 can be determined based on the difference 602a-d to predict the grinding end time. In some embodiments, the function is a linear function of time versus characteristic difference. The function (eg, slope and intercept) of line 606 may change during grinding of substrate 10 when calculating new differences. In some embodiments, the time at which line 606 reaches target difference 604 provides an estimated end time 608. The estimated end time 608 may change as the function of line 606 changes to accommodate the new difference.

In some embodiments, the function of line 606 is used to determine the amount of material removed from substrate 10, and the change in the current value determined by the function is used to determine when the target difference is reached and when the end point needs to be called. Line 606 tracks the amount of material removed. Alternatively, when a particular thickness of material is removed from the substrate 10, a change in the current value determined by the function can be used to determine the amount of material removed from the top surface of the substrate 10 and when to call the endpoint. For example, the operator can set the target difference to a wavelength variation of 50 nm for the selected feature. For example, the change in wavelength of the selected peak can be used to determine how much material to remove from the top layer of substrate 10 and when to call the endpoint.

At time T1, the characteristic value difference of the selected features is zero before the polishing of the substrate 10. As the polishing pad 30 begins to polish the substrate 10, the characteristic values of the identified features may change as the material is abraded away from the top surface of the substrate 10. For example, during milling, the wavelength of the selected characteristic can be changed to a higher or lower wavelength. Excluding the noise effect, the wavelength of the feature (and therefore the difference in wavelength) tends to change monotonically and often changes linearly. At time T', the endpoint decision logic determines that the identified characteristic characteristic has changed the target difference δV and the endpoint is called. For example, when the wavelength of the feature has changed the target difference by 50 nm, the end point is called and the polishing pad 30 stops grinding the substrate 10.

When processing a batch of substrates, optical monitoring system 50 can, for example, track the same spectral features on all of the substrates. The spectral features can be associated with the same grain features on the substrate. Based on the underlying changes in the substrate, the starting wavelength of the spectral features can vary between substrates in the batch. In some embodiments, in order to minimize variability on a plurality of substrates, the endpoint decision logic is invoked when a function that selects a characteristic property value or a value that fits to the characteristic property changes the endpoint metric EM (rather than the target difference) Use the end point. The endpoint decision logic may use an expected initial value EIV determined according to the mounting substrate. At time T1 identifying the characteristic characteristic being tracked on the substrate 10, the endpoint decision logic determines the actual initial value AIV of the substrate being processed. The endpoint decision logic can use the initial value weight IVW to reduce the effect of the actual initial value on the endpoint decision, taking into account changes in the substrate on a batch. For example, the substrate variation can include the thickness of the substrate or the thickness of the underlying structure. Initial value weights can be correlated with substrate variations to increase uniformity between processes between substrates. For example, the endpoint metric can be determined by multiplying the initial value weight by the difference between the actual initial value and the expected initial value, plus the target difference, eg, EM=IVW*(AIV-EIV)+δV.

In some embodiments, a weighted combination is used to determine the endpoint. For example, the endpoint decision logic may calculate an initial value of the property from the function and calculate a current value of the property from the function and calculate a first difference between the initial value and the current value. The endpoint decision logic may calculate a second difference between the initial value and the target value and generate a weighted combination of the first difference and the second difference.

Figure 6B is an exemplary graph 600b of the characteristic magnitude difference versus time taken at two portions of the substrate 10. For example, optical monitoring system 50 can track one feature positioned toward an edge portion of substrate 10 and another feature positioned toward a central portion of substrate 10 to determine how much material has been removed from substrate 10. When testing the mounting substrate, the operator of the polishing apparatus 20 can, for example, identify two features corresponding to different portions of the mounting substrate for tracking. In some embodiments, the spectral features correspond to the same type of grain features on the mounting substrate. In other embodiments, the spectral features are associated with different types of grain features on the mounting substrate. When the substrate 10 is being ground, the photodetector 52 can measure a series of spectra of reflected light from two portions of the substrate 10 corresponding to selected features of the mounting substrate. The series of values associated with the characteristics of the two features can be determined by the endpoint decision logic. A series of first difference values 610a-b of one of the characteristic characteristics in the first portion of the substrate 10 can be calculated by subtracting the initial characteristic value from the current characteristic value as the polishing time progresses. A series of second differences 612a-b of one of the characteristic characteristics in the second portion of the substrate 10 can be similarly calculated.

The first line 614 can be fitted to the first difference 610a-b and the second line 616 can be fitted to the second difference 612a-b. The first line 614 and the second line 616 can be determined based on the first function and the second function, respectively, to determine an estimate of the polishing endpoint time 618 or the polishing rate 620 of the substrate 10.

During the grinding, an end point calculation based on the target difference 622 is performed at time TC using a first function of the first portion of the substrate 10 and using a second function of the second portion of the substrate. If the estimated end time of the first portion of the substrate and the second portion of the substrate is different (eg, the first portion will reach the target thickness before the second portion), the polishing rate 620 can be adjusted such that the first function and the second function will Has the same endpoint time 618. In some embodiments, the polishing rates of the first portion and the second portion of the substrate are adjusted such that the endpoints are simultaneously reached at both portions. Alternatively, the polishing rate of the first portion or the second portion can be adjusted.

For example, the polishing rate can be adjusted by increasing or decreasing the pressure in the corresponding region of the carrier head 70. The change in the polishing rate can be assumed to be proportional to the change in pressure, for example, a simple Prestonian model. For example, when the first region of the substrate 10 protrudes at the time TA to reach the target thickness, and the system has established the target time TT, the carrier head pressure in the corresponding region before the time T3 can be multiplied by TT/TA to The carrier head pressure is provided after time T3. Additionally, a control model for polishing the substrate can be developed that takes into account the effects of the speed of the platform or head, the second order effects of different head pressure combinations, the grinding temperature, the slurry flow rate, or other parameters that affect the polishing rate. At a subsequent time during the polishing process, the rate can be adjusted again if appropriate.

In some embodiments, the computing device uses a range of wavelengths to readily identify selected spectral features in the spectrum measured from light reflected by the device substrate 10. The computing device searches the selected spectral features in the wavelength range to distinguish between selected spectral features and other spectral features that are similar in intensity, width, or wavelength to selected spectral features in the measured spectrum.

FIG. 7A illustrates an example of a spectrum 700a based on the amount of light received by the photodetector 52. Spectrum 700a includes selected spectral features 702, such as spectral peaks. The selected spectral features 702 can be selected by endpoint decision logic to track during CMP of the substrate 10. The characteristic 704 (e.g., wavelength) of the selected spectral feature 702 can be identified by the endpoint decision logic. When the characteristic 704 has changed the target difference, the endpoint determines the logical recall endpoint.

In some embodiments, the endpoint decision logic determines a wavelength range 706 to search for selected spectral features 702 over the range of wavelengths. Wavelength range 706 can have a width between about 50 and about 200 nanometers. In some embodiments, the wavelength range 706 is predetermined, for example, by an operator (eg, by receiving a user input of a selected wavelength range), or as a process parameter for a batch of substrates (by making the wavelength range from The memory associated with the batch of substrates retrieves the wavelength range). In some embodiments, the wavelength range 706 is based on historical data, such as the average or maximum distance between sequential spectral measurements. In some embodiments, the wavelength range 706 is based on information about the test substrate, for example, twice the target difference δV.

FIG. 7B is an example of a spectrum 700b measured from the amount of light received by the photodetector 52. For example, after the spectrum 700a is taken following the rotation of the platform 24, the spectrum 700b is measured. In some embodiments, the endpoint decision logic determines the value of the characteristic 704 in the previous spectrum 700a (eg, 520 nm) and adjusts the wavelength range 706 such that the center of the wavelength range 708 is positioned closer to the characteristic 704.

In some embodiments, the endpoint decision logic uses a function of line 606 to determine the expected current value of characteristic 704. For example, the endpoint decision logic may use the current grind time to determine the expected difference and determine the expected current value of the characteristic 704 by adding the expected difference to the initial value V1 of the characteristic 704. The endpoint decision logic may center the wavelength range 708 on the expected current value of the characteristic 704.

FIG. 7C is another example of the spectrum 700c measured from the amount of light received by the photodetector 52. For example, after the spectrum 700a is taken immediately after the rotation of the platform 24, the spectrum 700c is measured. In some embodiments, the endpoint decision logic uses the previous value of the characteristic 704 for the center of the wavelength range 710.

For example, the endpoint decision logic determines the average variance between the values of the characteristics 704 determined during the two sequential transfers of the optical head 53 below the substrate 10. The endpoint decision logic can set the width of the wavelength range 710 to twice the mean variance. In some embodiments, the endpoint decision logic uses the standard deviation of the variance between the values of the characteristic 704 when determining the width of the wavelength range 710.

In some embodiments, the width of the wavelength range 706 is the same for all spectral measurements. For example, the wavelength range 706, the wavelength range 708, and the wavelength range 710 have the same width. In some embodiments, the width of the wavelength range is different. For example, when the estimated characteristic 704 changes by 2 nanometers from the previous measurement of the characteristic, the width of the wavelength range 708 is 60 nanometers. When the estimated characteristic 704 changes by 5 nanometers from the previous measurement of the characteristic, the width of the wavelength range 708 is 80 nm, and the wavelength range of 80 nm is larger than the wavelength range having a smaller characteristic change.

In some embodiments, the wavelength range 706 is the same for all spectral measurements during the grinding of the substrate 10. For example, the wavelength range 706 is 475 nm to 555 nm, and for all spectral measurements made during the polishing of the substrate 10, the endpoint decision logic searches for selected spectral features in wavelengths between 475 nm and 555 nm. 702, however other wavelength ranges are also possible. The wavelength range 706 can be selected by the user input as a subset of the full spectrum range measured by the in situ monitoring system.

In some embodiments, the endpoint decision logic searches for selected spectral features 702 in the modified wavelength range of some spectral measurements and in the wavelength range of the previous spectrum used in the remainder of the spectrum. For example, the endpoint decision logic searches for a selected spectral feature 702 in a wavelength range 706 of the spectrum measured during the first rotation of the platform 24 and a wavelength range 708 of the measured spectrum during sequential rotation of the platform 24, wherein Both measurements are made in the first region of the substrate 10. Continuing with the example, the endpoint decision logic searches for another selected spectral feature in the wavelength range 710 of the two spectra measured during the same platform rotation, two of which are measured in a second region of the substrate 10 that is different from the first region. In progress.

In some embodiments, the selected spectral feature 702 is a spectral valley or a spectral zero crossing point. In some embodiments, characteristic 704 is the intensity or width of a peak or trough (eg, measured at a fixed distance below the peak, or measured at an intermediate height between the peak and the nearest trough).

Figure 8 illustrates a method 800 for selecting a target difference δV for use in determining the end of the polishing process. The properties of the substrate having the same pattern as the product substrate are measured (step 802). The substrate to be measured is referred to as a "mounting" substrate in this specification. The mounting substrate may simply be a substrate similar or identical to the product substrate, or the mounting substrate may be one substrate from a batch of product substrates. The properties of the measurement can include the pre-grind thickness of the film of interest at a particular site of interest on the substrate. Typically, the thickness at multiple sites is measured. Sites are typically selected to measure the same type of grain characteristics for each site. Measurements can be performed at the measurement station. Prior to grinding, the in-situ optical monitoring system can measure the spectrum of light reflected from the substrate.

The mounting substrate is ground according to the grinding step of interest, and the spectrum obtained during the grinding is collected (step 804). Grinding and spectral collection can be performed at the above grinding apparatus. The spectrum is collected by the in situ monitoring system during grinding. Excessive grinding of the substrate (i.e., grinding beyond the estimated endpoint) allows for a spectrum of light that is reflected from the substrate when the target thickness is achieved.

The properties of the overgrinded substrate are measured (step 806). Properties include the post-grinding thickness of the film of interest at a particular site, or at a site for measurement prior to milling.

The measured thickness and the collected spectrum are used to select (by examining the collected spectrum) specific features (such as peaks or troughs) for monitoring during grinding (step 808). The features may be selected by an operator of the grinding apparatus, or the selection of features may be automatic (e.g., based on a conventional peak-finding algorithm and an empirical peak-selection formula). For example, as described above with reference to FIG. 5B, a contour map 500b can be presented to an operator of the grinding apparatus 20, and an operator can select features from the contour map 500b for tracking. If a particular spectral region is expected to contain features that are desired to be monitored during grinding (eg, from past experience, or calculations based on theoretical characteristic behavior), then only features in that region need to be considered. A feature is typically selected that exhibits a correlation between the amount of material removed from the top of the mounting substrate as the substrate is being polished.

Linear interpolation can be performed using the measured pre-polished film thickness and post-grinding substrate thickness to determine the approximate time to achieve the target film thickness. The approximate time can be compared to the spectral contour map to determine the endpoint value of the selected characteristic. The difference between the end point value of the characteristic characteristic and the initial value can be used as the target difference. In some embodiments, a function is fitted to the value of the characteristic characteristic to normalize the value of the characteristic characteristic. The difference between the endpoint value of the function and the initial value of the function can be used as the target difference. The same features are monitored during the grinding of the remaining substrates of the batch of substrates.

Process the spectrum as needed to improve accuracy and/or accuracy. For example, the spectrum can be processed to normalize the spectrum to a common reference to average the spectrum and/or filter the noise from the spectrum. In one embodiment, a low pass filter is applied to the spectrum to reduce or eliminate burst spikes.

It is often empirically chosen to determine the logic characteristics for a particular endpoint, the spectral features to be monitored, such that the target thickness is achieved when the computer device invokes the endpoint by applying the endpoint-based logic based on the particular feature. The endpoint decision logic uses the target difference in the feature characteristics to determine when the endpoint should be called. When the grinding begins, the change in characteristics can be measured relative to the initial characteristic value of the feature. Alternatively, in addition to the target difference δV, the end point may be called with respect to the expected initial value EIV and the actual initial value AIV. The endpoint logic may multiply the difference between the actual initial value and the expected initial value by the starting value weight SVW to compensate for the underlying variation between the substrates. For example, when the endpoint metric EM=SVW*(AIV-EIV)+δV, the endpoint decision logic can end the grinding.

In some embodiments, a weighted combination is used to determine the endpoint. For example, the endpoint decision logic may calculate an initial value of the property from the function and calculate a current value of the property from the function and calculate a first difference between the initial value and the current value. The endpoint decision logic may calculate a second difference between the initial value and the target value and generate a weighted combination of the first difference and the second difference. The endpoint can be called if the weighted value reaches the target value. The endpoint decision logic can determine when the endpoint should be called by comparing the difference (or differences) of the monitored characteristics to the target. If the monitored difference matches the target difference or exceeds the target difference, the end point is called. In one embodiment, the monitored difference must match the target difference or exceed the target difference for a certain period of time (eg, the platform is rotated twice) to call the end point.

Figure 9 illustrates a method 901 for selecting a target value for a characteristic associated with a particular target thickness and a selected spectral feature of a particular endpoint decision logic. As described above in steps 802-806, the mounting substrate is measured and ground (step 903). In detail, the spectrum is collected and the time of each collected spectrum is stored.

The polishing rate of the polishing apparatus for a particular mounting substrate is calculated (step 905). The average polishing rate PR can be calculated by using the pre-polished thickness D1 and the post-grinding thickness D2 and the actual grinding time PT, for example, PR = (D2-D1) / PT.

The endpoint time of the particular mounting substrate is calculated (step 907) to provide a calibration point to determine the target value of the characteristic of the selected feature, as discussed below. The endpoint time can be calculated based on the calculated polishing rate PR, the pre-grinding initial thickness ST of the film of interest, and the target thickness TT of the film of interest of interest. Assuming that the polishing rate is constant throughout the polishing process, the endpoint time can be calculated as a simple linear interpolation, for example, ET = (ST - TT) / PR.

If desired, the calculated endpoint time can be evaluated by grinding another of the batch of patterned substrates, stopping the grinding at the calculated endpoint time, and measuring the thickness of the film of interest. If the thickness is within the satisfactory range of the target thickness, the calculated end time is satisfactory. Otherwise, the calculated end time can be recalculated.

At the calculated end time, the target characteristic value of the selected feature is recorded from the spectrum collected from the self-installed substrate (step 909). If the parameter of interest involves a change in the location or width of the selected feature, then the information can be determined by examining the spectrum collected during the time period prior to the calculated endpoint time. The difference between the initial value of the characteristic and the target value is recorded as the target difference of the feature. In some embodiments, a single target difference is recorded.

Figure 10 illustrates a method 1000 for determining the end of a grinding step using peak-based endpoint decision logic. The other of the batch of patterned substrates is ground using the polishing apparatus described above (step 1002).

The identification of the selected spectral features, the range of wavelengths, and the characteristics of the selected spectral features are received (step 1004). For example, the endpoint decision logic receives recognition from a computer that has processing parameters for the substrate. In some embodiments, the processing parameters are based on information determined during processing of the mounting substrate.

The substrate is initially ground, the light reflected from the substrate is measured to produce a spectrum, and the characteristic values of the selected spectral features are determined in the wavelength range of the measured spectrum. During each rotation of the platform, the following steps are performed.

One or more spectra of light reflected from the surface of the substrate being ground are measured to obtain one or more current spectra of the current platform revolution (step 1006). The one or more spectra measured by the current platform are processed as needed to improve accuracy and/or precision as described above with reference to FIG. If only one spectrum is measured, the one spectrum is used as the current spectrum. If more than one current spectrum is measured for platform rotation, the current spectrum is grouped, averaged within each group, and the average represents the current spectrum. The spectrum can be grouped by a radial distance from the center of the substrate.

For example, the first current spectrum can be obtained from the spectrum measured from points 202 and 210 (Fig. 2), and the second current spectrum can be obtained from the spectrum measured from points 203 and 209, which can be obtained from The measured spectrum at points 204 and 208 obtains a third current spectrum, and so on. The characteristic values of selected spectral peaks for each current spectrum can be determined, and the grinding can be monitored separately in each region of the substrate. Alternatively, the worst case value for the characteristics of the selected spectral peak can be determined based on the current spectrum and can be used by the endpoint decision logic.

Additional one or more spectra may be added to the series of spectra of the current substrate during each rotation of the platform. At least some of the spectra in the sequence are different as the milling proceeds, as the material is removed from the substrate during milling.

As described above with reference to Figures 7A-C, a modified wavelength range for the current spectrum is generated (step 1008). For example, the endpoint logic determines the modified wavelength range of the current spectrum based on previous characteristic values. The modified wavelength range can be centered on previous characteristic values. In some embodiments, the modified wavelength range is determined based on the expected characteristic value, for example, the center of the wavelength range coincides with the expected characteristic value.

In some embodiments, different methods are used to determine some of the wavelength ranges of the current spectrum. For example, the wavelength range of the spectrum measured from the amount of light reflected in the edge region of the substrate is determined by centering the wavelength range from the characteristic values of the previous spectrum measured in the same edge region of the substrate. Continuing with the example, the wavelength range of the spectrum measured from the amount of light reflected in the central region of the substrate is determined by centering the wavelength range over the expected characteristic value of the central region.

In some embodiments, the width of the wavelength range of the current spectrum is the same. In some embodiments, the widths of the wavelength ranges of some of the current spectra are different.

Identifying the wavelength range to search for selected spectral feature characteristics may allow for greater accuracy in the determination of endpoint detection or polishing rate changes, for example, the system may not select incorrect spectral features during subsequent spectral measurements. Tracking spectral features in the wavelength range rather than the entire spectrum allows for easier and faster identification of spectral features. The processing resources needed to identify selected spectral features can be reduced.

The current characteristic value of the selected peak is extracted from the modified wavelength range (step 1010), and the current characteristic value and the target characteristic value are compared using the endpoint decision logic discussed above in the context of FIG. 8 (step 1012). For example, a series of values of the current characteristic characteristics are determined based on the series of spectra, and a function is fitted to the series of values. For example, the function can be a linear function that can approximate the amount of material removed from the substrate during grinding based on the difference between the current characteristic value and the initial characteristic value.

As long as the endpoint decision logic determines that the endpoint condition has not been met ("NO" branch of step 1014), the grinding is allowed to continue, and steps 1006, 1008, 1010, 1012, and 1014 are repeated as appropriate. For example, the endpoint decision logic is based on a function that determines the target difference for which the feature has not yet been reached.

In some embodiments, when the equivalent is measured from the spectrum of light reflected by portions of the substrate, the endpoint decision logic may determine that the polishing rate of one or more portions of the substrate needs to be adjusted such that the same time or near the same time is completed. Grinding of multiple parts.

When the end point decision logic determines that the end condition has been met ("Yes" in step 1014), the end point is called and the grinding is stopped (step 1016).

The spectrum can be normalized to remove or reduce the effects of unwanted light reflections. Light reflection by a medium other than one or more films of interest, including light reflection from the polishing pad window and the substrate layer from the substrate. Light reflection from the window can be estimated by measuring the spectrum of light received by the in situ monitoring system under dark conditions (i.e., when the substrate is not placed on the in situ monitoring system). The light reflection from the germanium layer can be estimated by measuring the spectrum of the light reflected from the bare germanium substrate. These light reflections are typically obtained prior to the beginning of the grinding step. The original spectrum is normalized as follows:

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

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

In the described embodiment, endpoint detection is performed using changes in wavelength peaks in the spectrum. It is also possible to use variations in wavelength troughs (ie, local minima) in the spectrum instead of peaks or combined peaks. It is also possible to use multiple peaks (or troughs) to change at the end of the detection. For example, individual peaks can be individually monitored and the endpoint can be called when most of the peak changes meet the endpoint conditions. In other embodiments, the inflection point or spectral zero crossing change can be used to determine endpoint detection.

In some embodiments, the algorithmic mounting process 1100 (Fig. 11) is followed by the grinding of one or more substrates using the triggered feature tracking technique 1200 (Fig. 12).

Initially, the characteristics of the feature of interest in the spectrum are selected, for example, using one of the techniques described above for use in tracking the grinding of the first layer (step 1102). For example, the feature can be a crest or a trough, and the characteristic can be the position or width in the wavelength or frequency of the crest or trough, or the intensity of the crest or trough. Features and characteristics can be pre-selected by the equipment manufacturer if the characteristics of the feature of interest are applicable to a plurality of product substrates having different patterns.

In addition, the polishing rate dD/dt near the end of the polishing is determined (step 1104). For example, a plurality of mounting substrates can be ground depending on the polishing process to be used for the polishing of the product substrate, but at different polishing times near the expected end grinding time. The mounting substrate can have the same pattern as the product substrate. For each mounting substrate, the thickness of the layer before and after the polishing can be measured, and the amount of removal can be calculated according to the difference, and the removal amount of the mounted substrate and the related grinding time can be stored to provide a data set. A linear function of the removal amount can be fitted to the data set, the linear function of the removal amount being a function of time; the slope of the linear function provides the polishing rate.

Algorithm mounted installation process comprising measuring the initial thickness D of the first layer of the substrate ₁ (step 1106). The mounting substrate can have the same pattern as the product substrate. The first dielectric layer may be, e.g., a low dielectric value of the material, e.g., carbon-doped silicon dioxide, e.g., Black Diamond ^TM (from Applied Materials, Inc.) Or Coral ^TM (from Novellus Systems, Inc. ).

Depending on the composition of the first material, depending on the composition of the first material, a different layer than the first and second materials (eg, a low dielectric value covering material such as tetraethoxy decane (TEOS)) may be deposited on the first layer. One or more additional layers of a material (eg, a dielectric material) (step 1107). The first layer provides a layer stack with the one or more additional layers.

Next, a second layer (eg, a barrier layer) of a different second material (eg, a nitride, such as tantalum nitride or titanium nitride) is deposited on the first layer or layer stack (step 1108). Additionally, a conductive layer, such as a metal layer, such as copper, may be deposited on the second layer (and in the trench provided by the pattern of the first layer) (step 1109).

The measurement can be performed at a measurement system other than the optical monitoring system used during grinding, for example, by embedding or separating a measurement station, such as a profiler using an ellipsometer or an optical measurement station. For some measurement techniques (eg, profilers), the initial thickness of the first layer is measured before the second layer is deposited, but for other measurement techniques (eg, ellipsometers), before the second layer is deposited Or after the measurement is performed.

Thereafter, the substrate is mounted in accordance with the polishing process of interest (step 1110). For example, a first polishing pad can be used at the first polishing station to grind and remove the conductive layer and a portion of the second layer (step 1110a). Thereafter, a second polishing pad can be used at the second polishing station to grind and remove the second layer and a portion of the first layer (step 1110b). However, it should be noted that for some embodiments, there is no conductive layer, for example, the second layer is the outermost layer at the beginning of the grinding.

The spectrum is collected using the techniques described above during at least the removal of the second layer, and possibly during the entire polishing operation at the second polishing station (step 1112). Additionally, a separate detection technique is used to detect the removal of the second layer and the exposure of the first layer (step 1114). For example, exposure of the first layer can be detected by a sudden change in the total torque of the motor torque or light reflected from the substrate. At time T ₁ at which the second layer is cleared, the value V ₁ of the characteristic of the feature of interest of the spectrum at time T ₁ is stored. The time T ₁ at which the clearing is detected can also be stored.

After the detection of the clearing, the grinding may be paused at a preset time (step 1118). The preset time is large enough that the grinding is paused after exposure to the first layer. The preset time is selected such that the thickness after grinding is sufficiently close to the target thickness that the polishing rate can be assumed to be linear between the thickness after grinding and the target thickness. At the time the polishing is paused, the value V ₂ of the characteristic of the feature of interest that can detect and store the spectrum can also store the time T _{2 of the} grinding pause.

For example, the post-grinding thickness D _{2 of the} first layer is measured using a measurement system that is the same as the initial thickness used to measure (step 1120).

The preset target change ΔV _D of the value of the characteristic is calculated (step 1122). The preset target change for this value will be used in the endpoint detection algorithm for the product substrate. The preset target change may be calculated based on the difference between the value at the time of cleaning of the second layer and the value at the time of the grinding pause, that is, ΔV _D = V ₁ - V ₂ .

The rate of change dD/dV of the thickness as a function of the monitored characteristic is calculated near the end of the grinding operation (step 1124). For example, assuming that the wavelength position of the peak is being monitored, the rate of change can be expressed as the number of angstroms of the removed material for the wavelength position offset for each eibo peak. As another example, assuming that the frequency width of the peak is being monitored, the rate of change can be expressed as the offset of the frequency for the width of the Hertz peak, the number of grams of material removed.

In one embodiment, the rate of change dV/dt of the value as a function of time can be simply calculated based on the value at the exposure time of the second layer and at the end of the grinding, for example, dV/dt = (D ₂ -D ₁ )/(T ₂ -T ₁ ). In another embodiment, the end use of the substrate from the polishing mounted close (e.g., the last 25% or smaller between time T ₁ and T ₂₎ at the data, to be a function of the amount of time as measured The value fits the line; the slope of the line provides the rate of change dV/dt of the value as a function of time. In either case, thereafter, the rate of change dD/dV of the thickness as a function of the monitored characteristic is calculated by dividing the rate of change by the rate of change of the value, i.e., dD/dV = (dD/dt) / ( dV/dt). Once the rate of change dD/dV is calculated, the rate of change should remain constant for the product; it is not necessary to recalculate dD/dV for the same product in different batches.

Once the assembly process has been completed, the product substrate can be ground.

The initial thickness d ₁ of the first layer from at least one of the plurality of substrate substrates is measured as needed (step 1202). The product substrate has at least the same layer structure as the mounting substrate, and has the same pattern as the mounting substrate as needed. In some embodiments, each product substrate is not measured. For example, one substrate from a batch can be measured and the initial thickness can be used for all other substrates from the batch. As another example, one substrate from the cartridge can be measured and the initial thickness can be used for all other substrates from the cartridge. In other embodiments, each product substrate is measured. The measurement of the thickness of the first layer of the product substrate can be performed before or after the installation process is completed.

As described above, the first layer may be a dielectric, e.g., low dielectric value of the material, e.g., carbon doped silicon dioxide, e.g., Black Diamond ^TM (from Applied Materials, Inc.) Or Coral ^TM (from Novellus Systems, Inc.). The measurement can be performed at a measurement system other than the optical monitoring system used during grinding, for example, by embedding or separating a measurement station, such as a profiler using an ellipsometer or an optical measurement station.

Depending on the composition of the first material, different materials may be deposited on the first layer on the product substrate than the first and second materials (eg, a low dielectric value covering material, such as tetraethoxy decane (TEOS). One or more additional layers of another material (step 1203). The first layer provides a layer stack with the one or more additional layers.

Next, a second layer of a different second material (eg, a nitride, such as tantalum nitride or titanium nitride), such as a barrier layer, is deposited on the first layer or layer stack of the product substrate (step 1204). Alternatively, a conductive layer, such as a metal layer, such as copper, may be deposited on the second layer of the product substrate (and in the trench provided by the pattern of the first layer) (step 1205). However, it should be noted that for some embodiments, there is no conductive layer, for example, the second layer is the outermost layer at the beginning of the grinding.

For some measurement techniques (eg, profilers), the initial thickness of the first layer is measured before the second layer is deposited, but for other measurement techniques (eg, ellipsometers), before the second layer is deposited Or after the measurement is performed. The deposition of the second layer and the conductive layer can be performed before or after the mounting process is completed.

For each product substrate to be ground, the target characteristic difference ΔV is calculated based on the initial thickness of the first layer (step 1206). Typically, this occurs before the start of the grind, but the calculation may occur after the start of the grind but before the spectral feature tracking is initiated (in step 1210). In detail, for example, the autonomous computer receives the initial thickness d _{1 of the} stored product substrate and the target thickness d _T . In addition, the initial thickness D ₁ and the end thickness D ₂ , the rate of change dD/dV of the thickness as a function of the monitored characteristic, and the predetermined target change ΔV _D of the determined value of the mounted substrate can be received.

In one embodiment, the target characteristic difference ΔV is calculated as follows:

ΔV=ΔV _D +(d ₁ -D ₁ )/(dD/dV)+(D ₂ -d _T )/(dD/dV)

In some embodiments, the front thickness will not be available. In this case, "(d ₁ -D ₁ )/(dD/dV)" will be omitted from the above equation, that is,

ΔV=ΔV _D +(D ₂ -d _T )/(dD/dV)

The product substrate is ground (step 1208). For example, a first polishing pad can be used at the first polishing station to grind and remove the conductive layer and a portion of the second layer (step 1208a). Thereafter, a second polishing pad can be used at the second polishing station to grind and remove the second layer and a portion of the first layer (step 1208b). However, it should be noted that for some embodiments, there is no conductive layer, for example, the second layer is the outermost layer at the beginning of the grinding.

In-situ monitoring techniques are used to detect the removal of the second layer and the exposure of the first layer (step 1210). For example, exposure of the first layer at time t1 can be detected by a sudden change in motor torque or total intensity of light reflected from the substrate. For example, Figure 13 illustrates a graph of the total intensity of light received from the substrate as a function of time during the grinding of the metal layer to expose the underlying barrier layer. This total intensity can be generated based on the spectral signal acquired by the spectrum monitoring system by integrating the spectral intensities over all wavelengths measured or over a preset wavelength range. Alternatively, the intensity at a particular monochromatic wavelength can be used instead of the total intensity. As shown in Fig. 13, when the copper layer is removed, the total strength decreases, and when the barrier layer is completely exposed, the total strength is stable. A steady state of intensity can be detected, and a steady state of intensity can be used as a trigger to initiate spectral feature tracking.

The detection of at least the cleaning of the second layer begins (and possibly earlier, for example, from the beginning of polishing the product substrate using the second polishing pad), and the spectrum is obtained during the grinding using the in-situ monitoring technique described above (step 1212). The above techniques are used to analyze the spectrum to determine the value of the characteristics of the features being tracked. For example, Figure 14 illustrates a graph of the wavelength position of the spectral peak as a function of time during grinding. The value v ₁ of the characteristic of the feature being tracked in the spectrum at time t ₁ at which the second layer is cleared is determined.

The target value v _{T of the} characteristic can now be calculated (step 1214). The target value v _T can be calculated by adding the target characteristic difference ΔV to the value v ₁ of the characteristic at the time t ₁ of the second layer erasing, that is, v _T = v ₁ + ΔV.

When the characteristic of the tracked feature reaches the target value, the grinding is paused (step 1216). In particular, for each measurement spectrum (eg, in each platform rotation), the value of the characteristic of the tracked feature is determined to produce a series of values. As described above with reference to Figure 6A, a function can be fitted to a series of values (e.g., a linear function of time). In some embodiments, the function can be fitted to values within the time window. The end time of the pause grinding is provided if the function satisfies the target value. T also by part of _a series of values of the fit function (e.g., a linear function) to the closest time to determine values t ₁ V characteristic of _a place at the time of detecting remove the second layer.

Although the method illustrated by Figures 12 and 13 includes the deposition and removal of the second layer, for some embodiments, there is no second layer, for example, the first layer is the outermost layer at the beginning of the grinding . For example, the initial thickness of the first layer is measured prior to grinding, and the processing of calculating the target eigenvalue based on the initial thickness and the target thickness may be applicable with or without covering the second layer; The layers are optional. In detail, the step of depositing the second layer and the step of detecting the exposure of the first layer may be omitted. The first layer may comprise a polycrystalline germanium and/or a dielectric material, for example, consisting of substantially pure polycrystalline germanium, composed of a dielectric material or a combination of polycrystalline germanium and a dielectric material. The dielectric material can be an oxide (eg, hafnium oxide), or a nitride (eg, tantalum nitride) or a combination of dielectric materials.

For example, an initial thickness d ₁ of a first layer from at least one substrate of a batch of product substrates is measured (eg, as discussed for step 1202). The target characteristic difference ΔV is calculated based on the initial thickness of the first layer (eg, as discussed for step 1206). The grinding of the first layer of the product substrate is initiated and the spectrum is obtained using the in situ monitoring technique described above during the grinding of the first layer. The value of the characteristic v ₁ can be measured during the grinding of the first layer (for example, immediately after the initiation of the grinding of the first layer, or shortly after the grinding of the first layer is initiated (for example, after a few seconds)). Waiting for a few seconds allows the signal from the monitoring system to stabilize, making the measurement of the value v ₁ more accurate. The target value v _{T of the} characteristic can be calculated (e.g., as discussed for step 1214). For example, the target characteristic difference ΔV can be added to the value of the characteristic v ₁ , that is, v _T = v ₁ + ΔV. When the characteristic of the tracked feature reaches the target value, the grinding is paused (e.g., as discussed for step 1216). This method allows for removal to the target thickness while compensating for inter-substrate variations in absolute crest sites caused by differences between substrates in the underlying structure.

There are many techniques for removing noise from series values. Although the above describes the fitting of the line to the sequence, it is also possible to fit the nonlinear function to the sequence, or a low-pass median filter can be used to smooth the sequence (in this case, the filtered value can be directly compared to the target value) Than to determine the end point).

As used herein, the term substrate can include, for example, a product substrate (eg, a product substrate that includes a plurality of memory or processor dies), a test substrate, a bare substrate, and a gate substrate. The substrate can be at various stages of integrated circuit fabrication, for example, the substrate can be a bare wafer, or the substrate can include one or more deposited and/or patterned layers. The term substrate can include circular disks and rectangular sheets.

The embodiments of the invention and all of the functional operations described in this specification can be implemented in digital electronic circuitry, or in computer software, firmware or hardware (including structural components and structural equivalents thereof disclosed in this specification). Medium, or implemented in a combination thereof. Embodiments of the invention may be implemented as one or more computer program products (ie, one or more computer programs tangibly embodied in an information carrier (eg, in a machine readable storage device or in a propagated signal) ), or by a data processing device (eg, a programmable processor, a computer or multiple processors or computers), or a data processing device (eg, a programmable processor, a computer, or multiple processors or computers) Operation. Any form of programming language (including compiled or interpreted languages) may be written to a computer program (also known as a program, software, software application or code) and may be in any form (including as a stand-alone program or as a module, A component, subroutine, or other unit suitable for use in a computing environment to deploy the computer program. The computer program does not have to correspond to the file. The program can be stored in a file that stores other programs or data, stored in a single file dedicated to the program or stored in multiple coordination files (for example, one or more modules, subprograms or partial code files). In the file). Computer programs can be deployed to execute on one computer or multiple computers at one location, or in multiple locations and interconnected by a communication network.

The processes and logic flows described in this specification can be performed by executing one or more computer programs or a plurality of programmable processors to perform functions by operating on an input data and generating an output. The processing and logic flow may also be performed by a dedicated logic circuit (eg, a field programmable gate array (FPGA) or an application-specific integrated circuit (ASIC)), and the device may also be implemented as This dedicated logic circuit.

The above grinding apparatus and method can be applied to various grinding systems. The polishing pad or carrier head or both can be moved to provide relative motion between the abrasive surface and the substrate. For example, the platform can orbit rather than rotate. The polishing pad can be a circular (or some other shaped) pad that is fixed to the platform. Some aspects of the endpoint detection system are applicable to linear abrasive systems, for example, where the polishing pad is a linearly moving continuous or reel-to-reel belt system. The abrasive layer can be a standard (eg, polyurethane with or without filler) abrasive material, soft material, or fixed abrasive material. The term relative positioning is used; it should be understood that the abrasive surface and substrate can be held in a vertical orientation or other orientation.

Specific embodiments of the invention have been described. Other embodiments are within the scope of the following patent claims. For example, the acts recited in the scope of the claims can be performed in a different order and still achieve the desired result.

10. . . Substrate

20. . . Grinding equipment

twenty two. . . Drive shaft

twenty four. . . platform

25. . . Axis/central axis

26. . . Groove

30. . . Abrasive pad/mat

32. . . Outer abrasive layer

34. . . Back layer

36. . . Optical access point

38. . . Slurry

39. . . Slurry arm / flushing arm / arm

50. . . In-situ monitoring module

51. . . light source

52. . . Light detector

53. . . Optical head

54. . . Bifurcated fiber optic cable / bifurcated fiber cable / bifurcated fiber

55. . . Trunk

58. . . Branch line

56. . . Branch line

71. . . Axis/central axis

70. . . Carrier head

74. . . Drive shaft

72. . . supporting structure

201. . . point

76. . . Carrier head rotation motor

203. . . point

202. . . point

205. . . point

204. . . point

207. . . point

206. . . point

209. . . point

208. . . point

211. . . point

210. . . point

304. . . Spectrum

302. . . Spectrum

310(1). . . crest

306. . . Spectrum

400a. . . Spectrum

310(2). . . crest

500a. . . Spectrum

400b. . . Spectrum

502. . . Crest area/peak

500b. . . Contour map

506. . . crest

504. . . Valley area/trough

602b. . . Difference

508. . . line

602d. . . Difference

602c. . . Difference

700b. . . Spectrum

700a. . . Spectrum

702. . . Selected spectral characteristics

700c. . . Spectrum

706. . . Wavelength range

704. . . characteristic

710. . . Wavelength range

708. . . Wavelength range

802. . . step

800. . . method

806. . . step

804. . . step

901. . . method

808. . . step

905. . . step

903. . . step

909. . . step

907. . . step

1002. . . step

1000. . . method

1006. . . step

1004. . . step

1010. . . step

1008. . . step

1014. . . step

1012. . . step

1100. . . Algorithm installation process

1016. . . step

1104. . . step

1102. . . step

1107. . . step

1106. . . step

1109. . . step

1108. . . step

1110a. . . step

1110. . . step

1112. . . step

1110b. . . step

1118. . . step

1114. . . step

1122. . . step

1120. . . step

1200. . . Triggered feature tracking

1124. . . step

1202. . . step

1203. . . step

1204. . step

1205. . . step

1206. . . step

1208. . . step

1208a. . . step

1208b. . . step

1210. . . step

1212. . . step

1214. . . step

1216. . . step

T1. . . time

T2. . . time

T'. . . time

δV. . . Target difference

T1. . . time

Figure 1 illustrates a chemical mechanical polishing apparatus.

Figure 2 is a top view of the polishing pad and shows the location for in situ measurement.

Figure 3A illustrates the spectrum obtained from in situ measurements.

Figure 3B illustrates the evolution of the spectrum obtained by in situ measurement as the grinding progresses.

Figure 4A illustrates an exemplary graph of the spectrum of light reflected from the substrate.

Figure 4B illustrates a graph through Figure 4A of the high pass filter.

Figure 5A illustrates the spectrum of light reflected from the substrate.

Figure 5B illustrates a contour plot of the spectrum obtained from in situ measurements of light reflected from the substrate.

Figure 6A illustrates an exemplary graph of the progress of the grinding, which is measured in terms of the difference in characteristics versus time.

Figure 6B illustrates an exemplary graph of the progress of the grinding, measured in terms of characteristic difference versus time, wherein the characteristics of the two different features are measured to adjust the polishing rate of the substrate.

Figure 7A illustrates another spectrum of light obtained by in situ measurement.

Figure 7B illustrates the spectrum of light obtained after the spectrum of Figure 7A.

Figure 7C illustrates another spectrum of light obtained after the spectrum of Figure 7A.

Figure 8 illustrates a method of selecting a peak for monitoring.

Figure 9 illustrates the method of obtaining the target parameters for the selected peak.

Figure 10 illustrates the method used for endpoint determination.

Figure 11 shows the method of setting the endpoint detection.

Figure 12 illustrates another method for endpoint determination.

Figure 13 illustrates a graph of total reflection intensity as a function of time during grinding.

Figure 14 illustrates a graph of the wavelength position of the spectral peak as a function of time during grinding.

The same component symbols and representations in the various figures represent the same elements.

700b. . . Spectrum

702. . . Selected spectral characteristics

704. . . characteristic

708. . . Wavelength range

Claims (42)

A method of controlling polishing comprising the steps of: grinding a substrate; receiving a selected one of a selected spectral feature, having a wavelength range of one width and one of the selected spectral features for monitoring during polishing; grinding the substrate Simultaneously measuring a series of spectra of light from the substrate; generating a series of values of the characteristic of the selected spectral feature from the series of spectra, the generating step comprising the step of: based on at least some of the spectra from the series of spectra, based on The spectral feature produces a modified wavelength range at a location within a prior wavelength range, searching for the selected spectral feature within the modified wavelength range, and determining a value of one of the selected spectral features, the prior wavelength range being used One of the previous spectra in the series of spectra; and at least one of determining a polishing endpoint or adjusting for one of the polishing rates based on the series of values. The method of claim 1, wherein the wavelength range has a fixed width. The method of claim 2, wherein the step of generating the modified wavelength range comprises the step of centering the fixed width at the location of the characteristic in the previous wavelength range. The method of claim 1, wherein the step of generating the modified wavelength range comprises the steps of: determining a position of the characteristic in the previous wavelength range, and adjusting the wavelength range such that in the modified wavelength range, the characteristic The system is positioned closer to the center of one of the modified wavelength ranges. The method of claim 1, wherein the step of generating the modified wavelength range comprises the step of determining a wavelength value of the selected spectral feature for at least some of the series of spectra to generate a series of wavelength values; The series of wavelength values is fitted to a function; and an expected wavelength value for the selected spectral signature for a subsequent spectral measurement is calculated based on the function. The method of claim 5, wherein the function is a linear function. The method of claim 5, wherein the step of generating the modified wavelength range comprises the step of centering the width of the wavelength range to the expected wavelength value. The method of claim 1, further comprising the step of fitting a function to the series of values and determining at least one of a polishing endpoint or an adjustment for a polishing rate based on the function. The method of claim 8, wherein the step of determining a polishing end point comprises the steps of: calculating an initial value of the characteristic according to the function, calculating a current value of the characteristic according to the function, and calculating the initial value and the current value A difference between the values, and the grinding is interrupted when the difference reaches a target difference. The method of claim 8, wherein the function is a linear function. The method of claim 1, wherein the selected spectral feature comprises: a spectral peak, a spectral trough, or a spectral zero crossing. The method of claim 11, wherein the characteristic comprises: a wavelength, a width, or an intensity. The method of claim 12, wherein the selected spectral feature comprises a spectral peak and the characteristic comprises a peak width. The method of claim 1, wherein the spectrum of visible light is measured and the wavelength range has a width between 50 and 200 nm. A method of controlling polishing comprising the steps of: receiving a user input selecting a fixed wavelength range, the fixed wavelength range being a subset of wavelengths measured by an in situ monitoring system; receiving a selected one of the selected spectral features and Selecting one of the spectral features to monitor during polishing; grinding a substrate; measuring a series of spectra of light from the substrate while grinding the substrate; searching for each of the spectra in the series of spectra The selected spectral characteristic of the fixed wavelength range and a value determining one of the characteristics of the selected spectral characteristic to produce a series of values; and determining at least one of a polishing endpoint or an adjustment for a polishing rate based on the series of values . The method of claim 15, wherein the in-situ monitoring system measures an intensity comprising at least a wavelength of visible light, and the fixed wavelength range has a width between 50 and 200 nm. The method of claim 15, wherein the selected spectral feature comprises: a spectral peak, a spectral trough, or a spectral zero crossing. The method of claim 15, wherein the characteristic comprises: a wavelength, a width, or an intensity. A method of controlling polishing comprising the steps of: grinding a substrate having a first layer; receiving a characteristic of a selected spectral feature and characteristic of the selected spectral feature for monitoring during polishing; grinding the substrate Measuring a series of spectra of light from the substrate; determining a first value of the characteristic of the feature at a time of exposure of the first layer; adding an offset to the first value to generate a first Binary; and monitoring the characteristic of the feature and pausing the grinding when determining that the characteristic of the feature reaches the second value. The method of claim 19, wherein the characteristic comprises: a position, a width or an intensity. The method of claim 20, wherein the selected feature continues for an evolutionary site, width or intensity in the entirety of the series of spectra. The method of claim 21, wherein the feature comprises a peak or trough of the spectrum. The method of claim 19, wherein the substrate comprises a second layer covering the first layer, wherein the step of grinding comprises the steps of: grinding the second layer, and further comprising the step of: using an in situ monitoring system Detecting the exposure of the first layer. The method of claim 23, wherein the first value is determined by a time during which the first in-situ monitoring technique detects exposure of the first layer. The method of claim 23, wherein the step of detecting the exposure of the first layer is one of a process separate from the step of monitoring the characteristic of the feature. The method of claim 25, wherein the step of detecting exposure of the first layer comprises the step of monitoring a total reflected intensity from one of the substrates. The method of claim 25, wherein the step of monitoring the total reflected intensity comprises the step of integrating the spectrum over a range of wavelengths for each of the series of spectra to produce the total reflected intensity. The method of claim 25, wherein the in-situ monitoring system comprises a motor torque or friction monitoring system. The method of claim 19, wherein the first value is determined during the grinding of the first layer. The method of claim 29, wherein the first value is determined immediately after initiation of the grinding of the first layer. The method of claim 29, wherein the first layer is exposed prior to the initiation of the grinding of the substrate. The method of claim 19, wherein the step of monitoring the characteristic of the feature comprises the step of determining a value of the characteristic for each of the spectra from the series of spectra to produce a series of values. The method of claim 32, wherein the characteristic of the feature reaches the second value by fitting a linear function to the series of values and determining that the linear function is equal to an end time of the second value. . The method of claim 19, further comprising the steps of: receiving a pre-polished thickness of the first layer, and calculating the offset value from the pre-polishing thickness. The method of claim 34, wherein the step of calculating the offset value ΔV comprises the step of: calculating (D ₂ -d _T )/(dD/dV), wherein d _T is a target thickness and D ₁ is from a mounting one of the substrates before the first one of the thickness of the polishing layer, D ₂ is mounted from a first layer of the thickness of the substrate after the polishing of one, and dD / dV is the rate of change in thickness as one of the functions of the characteristic. The method of claim 34, wherein the step of calculating the offset value ΔV comprises the steps of: ΔV = ΔV _D + (d _{1 -} D ₁ ) / (dD / dV) + (D _{2 -} d _T ) / ( dD / dV) wherein d _{1 is} the thickness before the polishing, d _T is a target thickness, D ₁ is a thickness from the first layer of one of the mounting substrates, and D ₂ is the first from a mounting substrate a thickness of one of the layers after polishing, ΔV _D is a difference between the pre-polished thickness of the first layer of the mounting substrate and the thickness of the feature between the thicknesses of the features, and dD/dV is taken as The rate of change in the thickness of one of the properties of the function. The method of claim 36, further comprising the step of measuring the pre-polation thickness d ₁ at a separate measurement station. The method of claim 35, wherein dD/dV is a rate of change of thickness near the end of the grinding. The method of claim 19, wherein the first layer comprises polysilicon and/or a dielectric material. The method of claim 39, wherein the first layer consists of substantially pure polycrystalline germanium. The method of claim 39, wherein the first layer is comprised of a dielectric material. The method of claim 39, wherein the first layer is a combination of polycrystalline germanium and one of dielectric materials.