Devices and Methods for Optical Endpoint Detection During Semiconductor Wafer Polishing

FIELD OF THE INVENTIONS

The present invention is in the field of semiconductor wafer processing, and more specifically relates to a disposable polishing pad for use in chemical mechanical polishing. The polishing pad contains an optical sensor for monitoring the condition of the surface being polished while the polishing operation is taking place, thus permitting determination of the endpoint of the process.

BACKGROUND OF THE INVENTIONS

In U.S. Pat. No. 5,893,796 issued Apr. 13, 1999 and in continuation U.S. Pat. No. 6,045,439 issued Apr. 4, 2000, Birang et al. show a number of designs for a window installed in a polishing pad. The wafer to be polished is on top of the polishing pad, and the polishing pad rests upon a rigid platen so that the polishing occurs on the lower surface of the wafer. That surface is monitored during the polishing process by an interferometer that is located below the rigid platen. The interferometer directs a laser beam upward, and in order for it to reach the lower surface of the wafer, it must pass through an aperture in the platen and then continue upward through the polishing pad. To prevent the accumulation of slurry above the aperture in the platen, a window is provided in the polishing pad. Regardless of how the window is formed, it is clear that the interferometer sensor is always located below the platen and is never located in the polishing pad.

In U.S. Pat. No. 5,949,927 issued Sep. 7, 1999 to Tang, there are described a number of techniques for monitoring polished surfaces during the polishing process. In one embodiment Tang refers to a fiber-optic ribbon embedded in a polishing pad. This ribbon is merely a conductor of light. The light source and the detector that do the sensing are located outside of the pad. Nowhere does Tang suggest including a light source and a detector inside the polishing pad. In some of Tang's embodiments, fiber-optic decouplers are used to transfer the light in the optical fibers from a rotating component to a stationary component. In other embodiments, the optical signal is detected onboard a rotating component, and the resulting electrical signal is transferred to a stationary component through electrical slip rings. There is no suggestion in the Tang patent of transmitting the electrical signal to a stationary component by means of radio waves, acoustical waves, a modulated light beam, or by magnetic induction.

In another optical end-point sensing system, described in U.S. Pat. No. 5,081,796 issued Jan. 21, 1992 to Schultz there is described a method in which, after partial polishing, the wafer is moved to a position at which part of the wafer overhangs the edge of the platen. The wear on this overhanging part is measured by interferometry to determine whether the polishing process should be continued.

In earlier attempts to mount the sensor in the polishing pad, an aperture was formed in the polishing pad and the optical sensor was bonded into position within the aperture by means of an adhesive. However, subsequent tests revealed that the use of an adhesive could not be depended upon to prevent the polishing slurry, which may contain reactive chemicals, from entering the optical sensor and from penetrating through the polishing pad to the supporting table.

In conclusion, although several techniques are known in the art for monitoring the polished surface during the polishing process, none of these techniques is entirely satisfactory. The fiber optic bundles described by Tang are expensive and potentially fragile; and the use of an interferometer located below the platen, as used by Birang et al., requires making an aperture through the platen that supports the polishing pad. Accordingly, the present inventor set out to devise a monitoring system that would be economical and robust, taking advantage of recent advances in the miniaturization of certain components.

SUMMARY

The disposable polishing pad described below is composed of foamed urethane. It contains an optical sensor for monitoring, in situ, an optical characteristic of a wafer surface being polished. The real-time data derived from the optical sensor enables, among other things, the end-point of the process to be determined without disengaging the wafer for off-line testing. This greatly increases the efficiency of the polishing process.

The wafers to be polished are composite structures that include strata of different materials. Typically, the outermost stratum is polished away until its interface with an underlying stratum has been reached. At that point it is said that the end point of the polishing operation has been reached. The polishing pad and accompanying optics and electronics is able to detect transitions from an oxide layer to a silicon layer as well as transitions from a metal to an oxide, or other material.

The polishing pad described involves modifying a conventional polishing pad by embedding within it an optical sensor and other components. The unmodified polishing pads are widely available commercially, and the Model IC 1000 made by the Rodel Company of Newark, N.J., is a typical unmodified pad. Pads manufactured by the Thomas West Company may also be used.

The optical sensor senses an optical characteristic of the surface that is being polished. Typically, the optical characteristic of the surface is its reflectivity. However, other optical characteristics of the surface can also be sensed, including its polarization, its absorptivity, and its photoluminescence (if any). Techniques for sensing these various characteristics are well known in the optical arts, and typically they involve little more than adding a polarizer or a spectral filter to the optical system. For this reason, in the following discussion the more general term “optical characteristic” is used.

In addition to the optics the disposable pad provides an apparatus for supplying electrical power to the optical sensor in the polishing pad.

The disposable polishing pad also provides an apparatus for supplying electrical power for use in transmitting an electrical signal representing the optical characteristic from the rotating polishing pad to an adjacent non-rotating receiver. The pad is removably connectable to a non-disposable hub that contains power and signal processing circuitry.

An optical sensor that includes a light source and a detector is disposed within a blind hole in the polishing pad so as to face the surface that is being polished. Light from the light source is reflected from the surface being polished and the detector detects the reflected light. The detector produces an electrical signal related to the intensity of the light reflected back onto the detector.

The electrical signal produced by the detector is conducted radially inward from the location of the detector to the central aperture of the polishing pad by a thin conductor concealed between the layers of the polishing pad.

The disposable polishing pad is removably connected, both mechanically and electrically, to a hub that rotates with the polishing pad. The hub contains electronic circuitry that is concerned with supplying power to the optical sensor and with transmitting the electrical signal produced by the detector to non-rotating parts of the system. Because of the expense of these electronic circuits, the hub is not considered to be disposable. After the polishing pad has been worn out from use, it is disposed of, along with the optical sensor and the thin conductor.

Electrical power for operating the electronic circuits within the hub and for powering the light source of the optical sensor may be provided by several techniques. In one embodiment, the secondary winding of a transformer is included within the rotating hub and a primary winding is located on an adjacent non-rotating part of the polishing machine. In another embodiment, a solar cell or photovoltaic array is mounted on the rotating hub and is illuminated by a light source mounted on a non-rotating portion of the machine. In another embodiment, electrical power is derived from a battery located within the hub. In yet another embodiment, electrical conductors in the rotating polishing pad or in the rotating hub pass through the magnetic fields of permanent magnets mounted on adjacent non-rotating portions of the polishing machine, to constitute a magneto.

The electrical signal representing an optical characteristic of the surface being polished is transmitted from the rotating hub to an adjacent stationary portion of the polishing machine by any of several techniques. In one embodiment, the electrical signal to be transmitted is used to frequency modulate a light beam that is received by a detector located on adjacent non-rotating structure. In other embodiments, the signal is transmitted by a radio link or an acoustical link. In yet another embodiment, the signal is applied to the primary winding of a transformer on the rotating hub and received by a secondary winding of the transformer located on an adjacent non-rotating portion of the polishing machine. This transformer may be the same transformer used for coupling electrical power into the hub, or it can be a different transformer.

There must be a viable optical path between the top of the sensor and the lower side of the wafer. However, a void would not be acceptable, because it would quickly become filled with polishing slurry, thereby rendering it incapable of serving as an optical medium. In addition, a void would present a large mechanical discontinuity in the otherwise homogenous and uniformly resilient polishing pad. Further, the components of the optical sensor must not come into direct mechanical contact with the wafer that is being polished, to avoid scratching the surface of the wafer.

To overcome this problem, the optical sensor is embedded into the polishing pad using techniques described in detail below. These techniques have been successful in overcoming the disadvantages described above.

In addition, the intensity of the detected light conveys information regarding the amount of material removed from a layer during the polishing process. The intensity of the detected light varies sinusoidally with time as the surface layer is removed. The distance between any two succeeding peaks on the sinusoidal curve represents a particular amount of material removed. Thus, the total amount of material removed during polishing can be measured in-situ by calibrating the sinusoidal curve and then counting the number of peaks measured or observed during the polishing process. Likewise, the amount of material removed can be measured by calibrating the sinusoidal curve, measuring the distance between the start point and the end point on the sinusoidal curve and then correlating that distance to the amount of material removed.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 shows a top view of a chemical mechanical planarization machine polishing wafers using a polishing pad embedded with optical sensors.

FIG. 2 is an exploded view in perspective showing the general arrangement of the elements of the hub and optical assembly as placed in a polishing pad.

FIG. 3 is a front top perspective view of the optical sensor.

FIG. 4 is a side elevational diagram showing an optical sensor without a prism.

FIG. 5 illustrates an electronics hub using an inductive coupler.

FIG. 6 is a diagram showing a cross sectional view of a hub using a light emitting means to transfer signals to a non-rotating hub.

FIG. 7 is a diagram showing a cross sectional view of a hub utilizing radio emitting means to transfer signals to a non-rotating hub.

FIG. 8 is a diagram showing a cross sectional view of a hub utilizing sound waves to transfer signals to a non-rotating hub.

FIG. 9 shows a snap ring disposed in the polishing pad.

FIG. 10 is a top view of the snap ring, with a contact pad and conducting ribbon disposed on the bottom of the snap ring.

FIG. 11 shows a medial cross section of the optical sensor embedded into the polishing pad.

FIG. 12 shows a medial cross section of the injection molding process used to embed the optical sensor shown in FIG. 13.

FIG. 13 shows a medial cross section of the optical sensor and hub assembly embedded in a single injection molded pad.

FIG. 14 shows a medial cross section of the injection molding process used to embed both the optical sensor and the hub assembly.

FIG. 15 shows the polishing pad installed in a CMP system.

FIG. 16 illustrates the behavior of light of a selected wavelength when the light is incident on a thin layer of material disposed on the front side of a wafer.

FIG. 17 is a graph of the intensity of the detected light over time as the first layer of material is removed from a wafer.

DETAILED DESCRIPTION OF THE INVENTIONS

FIG. 1 is an overhead view of a chemical mechanical system 1 with the optical port 2 cut into the polishing pad 3. The wafer 4 (or other work piece requiring planarization or polishing) is held by the polishing head 5 and suspended over the polishing pad 3 from a translation arm 6. Other systems may use several polishing heads that hold several wafers, and separate translation arms on opposite sides (left and right) of the polishing pad.

The slurry used in the polishing process is injected onto the surface of the polishing pad through slurry injection tube 7. The suspension arm 8 connects to the non-rotating hub 9 that suspends over the electronic assembly hub 10. The electronics assembly hub 10 is removably attached to the polishing pad 3 by means of twist lock, detents, snap rings, screws, threaded segments, or any releasable mating mechanism. The hub 10 is attached to an electrical conducting assembly located within the pad where the hub attaches. The electrical conducting assembly can be either a single contact or a plurality of contacts attached to a thin, electrically conducting ribbon 11, also known as a flex circuit or ribbon cable. The ribbon 11 electrically connects an optical sensing mechanism, located within the optical port 2 and embedded in the pad 3, to the electronics in the electronics hub 10. The ribbon 11 may also comprise individual wires or a thin cable.

The window rotates with the polishing pad, which itself rotates on a process drive table, or platen 18, in the direction of arrow 12. The polishing heads rotate about their respective spindles 13 in the direction of arrows 14. The polishing heads themselves are translated back and forth over the surface of the polishing pad by the translating spindle 15, as indicated by arrow 16. Thus, the optical window 2 passes under the polishing heads while the polishing heads are both rotating and translating, swiping a complex path across the wafer surface on each rotation of the polishing pad/platen assembly.

The optical port 2 and the electrical conducting assembly (see FIG. 10) always remain on the same radial line 17 as the pad rotates. However, the radial line translates in a circular path as pad 3 rotates about the hub 9. Note that the conducting ribbon 11 lies along the radial line 17 and moves with it.

As shown in FIG. 2, the polishing pad 3 has a circular shape and a central circular aperture 23. A blind hole 24 is formed in the polishing pad, and the hole opens upwardly so as to face the surface that is being polished. An optical sensor 25 is placed in the blind hole 24 and a conductor ribbon 11, which extends from the optical sensor 25 to the central aperture 23, is embedded within the polishing pad 3.

When the polishing pad 3 is to be used, an electronics hub is inserted from above into the central aperture 23 and secured there by screwing a base 26, which lies below the polishing pad 3, onto a threaded portion of the hub 10. As seen in FIG. 5, the polishing pad 3 is thus clamped between portions of the hub and portions of the base 26. During the grinding process, the polishing pad 3, the hub 10 and the base 26 rotate together about a central vertical axis 28.

The non-rotating hub 9 of the polishing machine is located adjacent and above the hub 10. The non-rotating hub 9 is fixed during operation to the suspension arm 8.

FIG. 3 shows the optical sensor 25 in greater detail. The optical sensor 25 includes a light source 35, a detector 36, a reflective surface 37 (which could be a prism, mirror, or other reflective optical component), and the conductor ribbon 11. The conductor ribbon 11 includes a number of generally parallel conductors laminated together for the purpose of supplying electrical power to the light source 35 and for conducting the electrical output signal of the detector 36 to the central aperture 23. Preferably, the light source 35 and the detector 36 are a matched pair. In general, the light source 35 is a light emitting diode and the detector 36 is a photodiode. The central axis of the beam of light emitted by the light source 35 is directed horizontally initially, but upon reaching the reflective surface 37 the light is redirected upward so as to strike and reflect from the surface that is being polished. The reflected light also is redirected by the reflective surface 37 so that the reflected light falls on the detector 36, which produces an electrical signal in relation to the intensity of the light falling on it. The arrangement shown in FIG. 3 was chosen to minimize the height of the sensor. The reflective surface 37 may be omitted and instead the arrangement shown in side view in FIG. 4 may be used.

The optical components and the end of the conductor ribbon 11 are encapsulated in the form of a thin disk 38 that is sized to fit snugly within the blind hole 24 of FIG. 2. Note that in the arrangements of FIGS. 3 and 4 baffles may be used to reduce the amount of non-reflective light reaching the detector 36. Included within the conductor ribbon 11 are three conductors: a power conductor 39, a signal conductor 40, and one or more return or ground conductors 41.

FIG. 5 illustrates an electronics hub using an inductive coupler. The power conductor 39 terminates adjacent the central aperture 23 of the polishing pad 3 at a power plug 46, and the signal conductor 40 likewise terminates at a signal plug 49. When the hub 10 is inserted into the central aperture 23, the power plug 46 makes electrical contact with the power jack 50, and the signal plug 49 makes electrical contact with the signal jack 51. An O-ring seal 52 prevents the liquids used in the polishing process from reaching the plugs and jacks. A ring seal 53 is provided in the base 26 to further insure that the electronic circuits within the hub remain uncontaminated.

An electrical signal produced by the detector and related to the optical characteristic is carried by the conductor 54 from the signal jack 51 to a signal processing circuit 55, that produces in response to the electrical signal a processed signal on the conductor 56 representing the optical characteristic. The processed signal on the conductor 56 is then applied to a transmitter 57.

The process by which the signal is passed from the rotating hub 10 to the non-rotating hub 9 is referred to as inductive coupling, or RF coupling. The overall assembly may be referred to as an inductive coupler or an RF coupler.

The transmitter 57 applies a time-varying electrical current to the primary winding 58 of a transformer that produces a varying magnetic field 59 representative of the processed signal. The magnetic field 59 extends upward through the top of the hub 10 and is intercepted by a secondary winding 60 of the transformer which is located on an adjacent non-rotating portion 9 of the polishing machine, or on some other non-rotating object. The varying magnetic field 59 induces a current in the secondary winding 60 that is applied to a receiver 61 that produces on the terminal 62 a signal representative of the optical characteristic. This signal is then available for use by external circuitry for such purposes as monitoring the progress of the polishing operation or determining whether the end point of the polishing process has been reached.

A similar technique may be used to transfer electrical power from the adjacent non-rotating portion 9 of the polishing machine to the rotating hub 10. A prime power source 63 on the non-rotating portion 9 applies an electrical current to the primary winding 64 of a transformer that produces a magnetic field 65 that extends downward through the top of the hub 10 and is intercepted by a secondary winding 66 in which the varying magnetic field induces an electrical current that is applied to a power receiver circuitry 67. The power receiver 67 applies electrical power on the conductor 68 to the power jack 50, from which it is conducted through the power plug 46 and the power conductor 46 to the light source. The power receiver 67 also supplies electrical power to the signal processing circuit 55 through the conductor 69, and to the transmitter 57 through the conductor 70. Thus, power for operation of the LED may also be provided by inductive coupling.

The winding 58 is the same winding as winding 66, and winding 60 is the same winding as winding 64. Alternatively, the windings may be different. The superimposed power and signal components are at different frequency ranges and are separated by filtering.

FIGS. 6 through 8 show other techniques used to transfer signals from the rotating hub 10 to a non-rotating hub 9 of the polishing machine, and to transfer electrical power from the non-rotating portion 9 into the rotating hub 10.

FIG. 6 shows the transmitter 57 further includes a modulator 75 that applies to a light emitting diode or laser diode 76 a frequency modulated current representative of the processed signal that represents the optical characteristic. The light-emitting diode 76 emits light waves 77 that are focused by a lens 78 onto a photodiode detector 79. The detector 79 converts the light waves 77 into an electrical signal that is demodulated in the receiver 80 to produce on the terminal 62 an electrical signal representative of the optical characteristic.

The prime source of electrical power is a battery 81 that supplies power to a power distribution circuit 82 that, in turn, distributes electrical power to the power jack 50, to the signal processing circuit 55, and to the transmitter circuit 57. In FIG. 7 the transmitter 57 is a radio transmitter having an antenna 87 that transmits radio waves 88 through the top of the hub 9. The radio waves 88 are intercepted by the antenna 89 and demodulated by the receiver 90 to produce an electrical signal on the terminal 62 that is representative of the optical characteristic.

Electrical power is generated by a magneto consisting of a permanent magnet 91 located in the non-rotating portion 29 and an inductor 92 in which the magnetic field of the permanent magnet 91 induces a current as the inductor 92 rotates past the permanent magnet 91. The induced current is rectified and filtered by the power circuit 93 and then distributed by a power distribution circuit 94.

In FIG. 8, the transmitter 57 further includes a power amplifier 100 that drives a loudspeaker 101 that produces sound waves 102. The sound waves 102 are picked up by a microphone 103 located in the non-rotating portion 29 of the polishing machine. The microphone 103 produces an electrical signal that is applied to the receiver 104 which, in turn, produces an electrical signal on the terminal 62 that is representative of the optical characteristic.

Electrical power is generated in the rotating hub 9 by a solar cell or solar panel 105 in response to light 106 applied to the solar panel 105 by a light source 107 located in the non-rotating portion 29. The electrical output of the solar panel 105 is converted to an appropriate voltage by the converter 108, if necessary, and applied to the power distribution circuit 94.

FIGS. 9 through 16 show the hub insertion assembly and the optical-electrical insertion assembly 25. They also disclose methods of sealing a snap ring (to releasably attach the electronics hub) and optical-electrical assemblies into the polishing pad. The polishing pads 3 shown in these Figures are typical polishing pads available in the industry, such as the model IC 1000 produced by Rodel Co. The model comprises two 0.045-inch thick layers of foamed urethane bonded face to face by a 0.007-inch thick layer of adhesive. However, each has been modified to allow for a conducting ribbon 11, a snap ring 114, and an optical assembly 25 to be placed into the pad.

FIG. 9 shows a cross section of a molded insert, comprising a snap ring, 114 used to fix the electronics hub 10 into the center aperture of the polishing pad 3. The snap ring 114 is placed inside the center aperture 23 of the polishing pad 3. An inwardly extending flange 115, or collar, is cut out of the snap ring 114 so that the electronics hub 10 will snap securely into place. A guide pin hole 116 receives an electronics hub guide pin 117 to help assure proper alignment of the electronics hub 10. The snap ring is sealed inside of the polishing pad 3 by means of an adhesive or by a liquid urethane which subsequently dries and solidifies. The electronics hub 10 has a flange or ridge 118 disposed around its bottom section 119. This flange 118 is sized to provide a releasable fit with the molded insert snap ring 114.

The electrically conducting ribbon 11 conveys electrical signals and power between the optical assembly 25 and the electronics hub 10. The terminus of ribbon 11 is disposed on a contact pad 126 in the bottom of the hub-receiving aperture 120. The contact pad is provided with contacts for establishing electrical contact with matching contacts 122 disposed on the hub 10. The contacts 122 are preferably spring loaded or biased contacts (such as pogo pins). The contacts may be provided in redundant groups. As shown, three contacts are provided in the group visible in this view.

The snap ring assembly 114 is preferably isoplanar with the polishing pad 3 such that multiple pads may be easily stacked on top of each other.

FIG. 10 shows a top view of the snap ring 114. The circular lip of the snap ring 115, the guide pin hole 116, and the electrically conducting ribbon 11 are the same as shown in FIG. 9. Also shown in this Figure are three electrical contacts disposed on the contact pad 126. Specifically, the three contacts are used for power conduction (contact 123), signal conduction (contact 124), and common ground (contact 125), all of which lie on the contact pad 126. The contact pad 127 is disposed on the bottom inside surface of the snap ring assembly.

The electronics hub will snap into place inside the lip 115 of the snap ring 114. Proper alignment of the contacts of the hub with the contacts of the contact pad 127 is assured by the guide pin 116. Thus, the contacts of the hub establish electrical contact with contacts 123, 124, and 125 of the contact pad 126 when the hub is secured in the snap ring.

FIGS. 11 and 12 show cross sections of the optical sensor 25 and a method of securing the optical sensor 25 in the optical port 2 into the polishing pad 3. An aperture, or hole, 143 is produced in the polishing pad. The aperture 143 must be large enough to accommodate the optical sensor 25. The optical assembly 25 is placed into an optical assembly puck so that it may be easily disposed into the aperture. The puck is sized and dimensioned such that the surface of the puck facing the wafer is substantially flush with the surface of the polishing pad (the surface of the puck is within about 0.015 inches or less of the surface of the pad). Portions of the aperture adjacent to the upper surface 144 and lower surface 145 of the polishing pad 3 extend a short distance radially outwardly from the aperture. This creates a spool-shaped void with the boundaries of the pad.

A channel is produced in the underside of the upper layer 147 to accommodate the conducting ribbon 11 used to convey electrical power and signals from the electronics hub 10 to the optical sensor 25. The conducting ribbon 11 may intrude into the space generally occupied by the layer of adhesive 148, which secures the upper layer 147 of the polishing pad to the lower layer 149 of the polishing pad. Alternatively the conducting ribbon 11 may lie above or beneath the adhesive layer 148.

After the aperture 143 has been formed in the polishing pad 3, the optical sensor 25 and its conductor ribbon 11 are inserted into their respective places, where they are supported and held in place by spacers composed of urethane or by portions of the upper layer 147 and lower layer 149.

Thereafter, the assembly is placed into a fixture that includes flat, non-stick surfaces 155 and 156. The non-stick surfaces 155 and 156 are brought into contact with the upper pad surface 144 and lower pad surface 145 and pressed together.

Next, a liquid urethane is injected by syringe 157 through a passage 158 in the lower mold plate 159 and into the void immediately surrounding the optical sensor 25 until the injected urethane begins to emerge through the vent passage 160 of upper mold plate 161. During the injection, it is helpful to tilt the assembly slightly in the clockwise direction so that the liquid is injected at the lowest point of the void and the vent passage 160 is at the highest point. Tilting the assembly in this manner prevents air from becoming trapped in the void.

The injected urethane 162 directly above the optical sensor 25 serves as a window through which the optical sensor 25 can view the underside of the wafer, which is placed on top of the upper layer 147. The liquid urethane is a type of urethane that is optically transparent when it has cured. Because it is chemically similar to the urethane of the polishing pad 3, it forms a durable, liquid-proof bond with the material of the polishing pad 3.

The snap-ring assembly can be inserted into the pad, as shown in FIG. 9, or formed or integrally with the pad with injection molding processes. As shown in FIGS. 13 and 14, the polishing pad 3, including the upper pad layer 147, lower pad layer 149 and adhesive layer 148, has been punched and cut to provide voids 168 for the optical sensor, ribbon cable and the electrode pad. The ribbon cable 11, contact pad, and optical sensor 25 are placed in the corresponding voids in the pad, and a snap ring hub mold is inserted into the hub aperture. The electrode pad may be glued with a weak pressure sensitive adhesive (sticky glue) to the snap ring mold 169.

As shown in FIG. 13, an upper mold base 172 and a lower mold base 173 are pressed against the polishing pad's upper layer 147 and lower 149 layer, respectively. Urethane or other injectable plastic is then injected through the injection port 174, and the urethane fills the voids. When the void between the plates is filled, the liquid urethane 162 will exit through the exit vent 175, signaling that the injection process is complete. As shown in FIG. 14, the injected urethane 176 forms the snap ring assembly and fills the ribbon cable channel and the optical sensor assembly aperture. The injected urethane seals and connects the entire length of void between the snap ring 114 and the optics insert 25, and it locks the ribbon cable and the sensor assembly into place within the pad.

This process can be accomplished using a snap ring insert as shown in FIGS. 9 and 10 by sizing the hub aperture in the pad slightly larger than the snap ring insert, and using the injected urethane to fix the snap ring insert to the pad.

FIG. 15 shows a detailed view of the overall polishing pad 3 installed in a CMP system, using the pad design shown in FIGS. 13 and 14. The pad comprises the upper pad layer 147, lower pad layer 149, adhesive layer 148, injected urethane 176, electrically conductive ribbon 11, optical sensor 25, described in the previous Figures. The pad is placed on the platen 18. The electronics hub 10 is inserted in to the snap ring, so that the pogo pin electrical contacts 137 are in contact with the electrodes of the electrode pad. The non-rotating receiving hub 9 is suspended from the suspension arm 8 over the rotating electronics hub 10. The electronics in the rotating electronics hub may be the electronics shown in FIGS. 5 through 8, inside the box numbered as item 10 in those drawings, and the non-rotating receiving hub 9 will house the corresponding electronics in the boxes marked as items 9. After extended use, the pad will be exhausted and may be removed and discarded. A new pad may be placed on the platen, and the rotating hub may be inserted into the snap ring of the new pad.

FIG. 16 illustrates the behavior of light 190 of a selected wavelength when the light is incident on a thin layer of material disposed on the front side of a wafer. The wafer 4 is greatly magnified to show the two outermost layers built up on the front side 191 of the wafer. The first, outermost, layer 192 covers the second layer 193. Each layer may have a thickness of about 30 micrometers or less, usually between about 10 micrometers and about 1,000 Angstroms (about 1/10 of a micrometer), and a plurality of additional layers may be disposed beneath the first and second layers. During the polishing process the first layer is polished to remove the layer either partially or completely. To determine how much of the first layer has been removed, light 190 of a selected wavelength is emitted from the light source 35 and directed at the front side of the wafer at a fixed angle relative to the axis of the optical puck. The reflected light is detected by the detector 36. Both the light source and light detector are disposed within the optical sensor puck and the optical sensor puck may be disposed completely within the polishing pad. The intensity of the light reflected from the wafer conveys information regarding the amount of material removed during polishing. (The wavelength of the light is selected so that a portion of the light will transmit through the thin layer of material. For many layer materials, such as silicon, silicon dioxide, copper and other materials, the wavelength selected is in the range of about 300 nanometers (blue light) or less to about 1500 nanometers or more (infrared light). The angle of incidence and reflection is fixed between about 0 degrees and 70 degrees, preferably about 5 degrees, as measured between the axis of the puck and the light source.)

When light 190 is directed onto the front side of the wafer, a portion 194 of the light reflects from the surface of the wafer and a portion 195 of the light passes through the surface and through the first layer 192 of material. Portion 195 of the light reflects from the surface of the second layer 193 and escapes through the first layer 192. Portion 194 and portion 195 combine together before reaching the detector. Because portion 195 travels a greater distance than portion 194, the light reflected from the surface of the first layer 192 (portion 194) and the light reflected from the surface of the second layer 193 (portion 195) may be out of phase. Depending on the relative phase of portions 194 and 195, the two portions either constructively or destructively interfere with each other, thereby causing the detected light to become either more or less intense, respectively.

As the first layer 192 is removed, the distance traveled by portion 195 relative to portion 194 changes, thereby changing their phase relationship. As a result, the intensity of the detected light changes as the first layer is removed. As the phase shift between the two light rays repeatedly varies between 0 and 90 degrees as the layer is removed, the intensity of the detected light varies approximately sinusoidally. 196197198199

FIG. 17 is a graph of the intensity of the detected light over time as the first layer of material is removed from a wafer. (The intensity of the reflected light is a function of layer thickness and sinusoidally varies with layer thickness. Layer thickness varies over the time of polishing.) When light portion 194 and light portion 195 completely constructively interfere with each other, the intensity of the detected light is at a peak 200. When light portion 194 and light portion 195 completely destructively interfere with each other, the intensity of the detected light is at a trough 201.

To measure the amount of material removed during polishing, the curve must be calibrated. To calibrate the sinusoidal curve, the absolute thickness of the outer layer is first measured by spectral reflectance, ellipsometry or other technique for measuring absolute thickness. (These techniques may be performed using equipment provided by a variety of vendors. The equipment is relatively bulky, expensive or delicate and slurry and other aspects of the polishing process interfere with precise measurements of the index of refraction and of layer thickness. Thus, these other techniques for measuring layer thickness are not practical for use within a polishing pad during polishing or for use during mass production.) Next, the intensity of the reflected light signal is measured with the optical sensor 25. The outer layer of a test wafer is then polished until one or more wavelengths of the sinusoidal curve is measured or observed. Thus, if the initial intensity of the reflected light was at a peak or trough, then the wafer is polished until a second or subsequent peak or trough is measured. If the initial intensity of the reflected light signal was at some other point on the sinusoidal curve, then the wafer is polished until the same intensity is measured two or more times. The polishing process is then stopped and the absolute thickness of the outer layer is measured again.

The difference between the two measurements of layer thickness is the initial change in layer thickness. The initial change in layer thickness is also represented by one wavelength along the sinusoidal curve, but only if using the same polishing process on the same kind of wafer (or outer wafer layers) and if using the same wavelength of incident light. Multiple wavelengths along the curve may be counted, in which case the total change in layer thickness is the number of wavelengths measured times the initial change in layer thickness.

For convenience, wavelengths along the sinusoidal curve may be easily counted by counting the number of peaks or the number of troughs measured during a polishing process. Since the peaks or troughs may be thought of as nodes on the sinusoidal curve, this process of measuring layer thickness may be referred to as node counting. (The term node counting refers to the process of counting wavelengths along a sinusoidal reflectance curve and is not limited to counting only peaks and troughs.)

For example, the outer layer of a wafer is 10,000 Angstroms (1 micrometer) thick, as measured using ellipsometry. The layer is polished using a particular process until one wavelength on the sinusoidal curve is measured. After polishing the layer thickness is 8,000 Angstroms thick, as measured using ellipsometry. Thus, the distance between peaks on the sinusoidal curve (one wavelength) corresponds to a change in layer thickness equal to 2,000 Angstroms. If the final desired thickness of the layer is 4,000 Angstroms, the layer is polished until a total of 3 wavelengths are counted (representing 6,000 Angstroms of removed material), at which point the polishing process reaches its endpoint.

This process may also be used to continuously measure smaller changes in layer thickness. A fraction of a wavelength along the sinusoidal curve equals a corresponding fractional change in the thickness of the polished layer. Continuing the above example, ½ of the wavelength (the peak-to-peak distance shown by arrows “X”) represents a change in layer thickness equal to 1,000 Angstroms. Thus, if the wafer is polished again and another half wavelength along the sinusoidal curve is measured, then the final layer thickness will be 3,000 Angstroms. Since fractions of a wavelength can be counted, node counting may make in-situ measurements of very small changes in layer thickness.

Calibrating the sinusoidal curve at many points along the curve or over multiple wavelengths may be necessary where the wavelength of the curve varies over the time of polishing and where the different wavelengths represent different amounts of material removed. Thus, as shown in FIG. 17, when the distance along arrows “X” does not equal the distance along arrows “Y”, then more of the sinusoidal curve may have to be calibrated. In addition, the absolute thickness of the layer may be measured at any number of points along the sinusoidal curve to increase the precision of the calibration curve. This may be necessary if the sinusoidal curve is subject to noise, represented by the variations in the sinusoidal curve shown in FIG. 17.

A processor and software are provided to correlate the change in intensity of reflected light to the change in layer thickness according to the above methods. A display may be provided to display the progress of the polishing process. A control system, such as computer hardware and software, may be provided to modify the polishing process or to slow, stop or otherwise change the rate of polishing in response to a change in the layer thickness. Thus, the control system may cause polishing to slow as the endpoint of a process is neared and stop when the endpoint is reached. (The control system can control any aspect of the polishing process in response to the change in layer thickness over time.)

It should be noted that the various inventions may be employed in various combinations. For example, the releasable hub embodiments, described in connection with inductive couplers and other non-contacting couplers, can also be employed with slip rings and other contacting couplers. While urethane has been discussed as the material to be used as for injection and use as the injected sealant, other materials may be used, so long as they provide substantial adhesion and sealing between the several inserts and the pad. Additionally, while the pad construction has been discussed in relation to optical sensors, electrical sensors, heat sensors, impedance sensors and other sensors may be used instead, and the benefits of the molding and releasable hub still achieved.

In addition, libraries of sinusoidal reflectance curves may be generated to save time during production. Each curve will be the same for a particular process on a particular wafer. Thus, when polishing a known type of wafer with a known process for which a calibration curve has already been established, the calibration step may be skipped. In addition, each reflectance curve may be further refined by measuring the absolute thickness of each layer removed for each wavelength counted over the entire polishing process. Thus, the calibration curve will be precise over the entire duration of a polishing process (regardless of changes in index of refraction, layer materials or in processing parameters). Thus, while the preferred embodiments of the devices and methods have been described in reference to the environment in which they were developed, they are merely illustrative of the principles of the inventions. Other embodiments and configurations may be devised without departing from the spirit of the inventions and the scope of the appended claims.

	Number	Date	Country
Parent	10754360	Jan 2004	US
Child	11768873	Jun 2007	US

Devices and Methods for Optical Endpoint Detection During Semiconductor Wafer Polishing

Information

Publication Number

Date Filed

Date Published

Inventors

Original Assignees

CPC

US Classifications

International Classifications

Abstract

Description

Claims

Parent Case Info

Continuations (1)