Correction of Substrate Doping in Edge Reconfiguration for In-Situ Electromagnetic Induction Monitoring

Abstract

To provide a substrate polishing method that can improve the reliability of determination of control parameters and detection of an endpoint, can avoid insufficient polishing of a wafer, and can reduce non-uniformity within a wafer.SOLUTION: A substrate polishing method includes storing or generating a modified reference trace 450 for correcting a contribution of conductivity of a semiconductor wafer to a measured trace 220 by an in-situ electromagnetic induction monitoring system. The modified reference trace represents measurements of a bare doped reference semiconductor wafer by the in-situ electromagnetic induction monitoring system, which are modified by a neutral network. The substrate is monitored with the in-situ electromagnetic induction monitoring system, and at least a portion of the measured trace is applied to the neutral network to generate a modified measured trace 250. An adjusted trace 480 is generated, including subtracting the modified reference trace from the modified measured trace.SELECTED DRAWING: Figure 8

Description

Technical Field

[0001] The present disclosure relates to chemical mechanical polishing, and more specifically, to the monitoring of conductive layers during chemical mechanical polishing.

Background Art

[0002] Integrated circuits are typically formed on a substrate by sequentially depositing conductive layers, semiconductor layers, or insulating layers on a silicon wafer. In various manufacturing processes, it is necessary to planarize the layers on the substrate. For example, one manufacturing step includes depositing a fill layer on a non-planar surface and planarizing the fill layer. In certain applications, the fill layer is planarized until the top surface of the patterned layer is exposed. For example, a metal layer can be deposited on a patterned insulating layer to fill the trenches and holes in the insulating layer. After planarization, the remaining portions of the metal in the trenches and holes of the patterned layer form vias, plugs, and lines to provide conductive paths between the thin film circuits on the substrate.

[0003] Chemical mechanical polishing (CMP) is an accepted method of planarization. In this planarization method, it is usually necessary to attach the substrate to a carrier head. The exposed surface of the substrate is typically positioned against a rotating polishing pad. The carrier head applies a controllable load to the substrate to press the substrate against the polishing pad. A polishing slurry containing abrasive particles is typically supplied to the surface of the polishing pad.

[0004] One problem with CMP is determining whether the polishing process is complete, i.e., whether the substrate layer has been planarized to the desired flatness or thickness, or when the desired amount of material has been removed. Variations in the slurry composition, the condition of the polishing pad, the relative speed between the polishing pad and the substrate, the initial thickness of the substrate layer, and the load on the substrate can cause variations in the material removal rate. These variations cause the time required to reach the polishing endpoint to vary. Therefore, simply determining the polishing endpoint as a function of the polishing time can result in non-uniformity within or between wafers.

[0005] In some systems, for example, during polishing through a polishing pad, the substrate is monitored in-situ. One monitoring method is to induce eddy currents in the conductive layer and detect changes in the eddy currents when the conductive layer is removed. SUMMARY OF THE INVENTION

[0006] In one aspect, a method of correcting the contribution of the conductivity of a semiconductor wafer to a measurement trace by an in-situ electromagnetic induction monitoring system includes storing or creating a modified reference trace representing the measurement values of a bare doped reference semiconductor wafer by the in-situ electromagnetic induction monitoring system, modified by a neural network; monitoring a substrate using the in-situ electromagnetic induction monitoring system when a conductive layer is polished to create a measurement trace that depends on the thickness of the conductive layer; applying at least a portion of the measurement trace to a neural network to create a modified measurement trace; and creating an adjustment trace including subtracting the modified reference trace from the modified measurement trace.

[0007] In one aspect, a method of polishing a substrate includes storing or creating a modified reference trace representing the measurement values of a bare doped reference semiconductor wafer by the in-situ electromagnetic induction monitoring system, modified by a neural network; contacting a substrate having a conductive layer disposed thereon with a polishing pad; generating relative movement between the substrate and the polishing pad; monitoring the substrate using the in-situ electromagnetic induction monitoring system when the conductive layer is polished to create a measurement trace that depends on the thickness of the conductive layer; applying at least a portion of the measurement trace to a neural network to create a modified measurement trace; creating an adjustment trace including subtracting the modified reference trace from the modified measurement trace to at least partially correct the contribution of the conductivity of the semiconductor wafer to the measurement trace; and at least one of stopping the polishing or changing the polishing parameters based on the adjustment trace.

[0008] Each of these aspects is also applicable as a computer program product tangibly embodied on a computer-readable medium including instructions that cause a computer system to perform suitable operations (e.g., saving or creating a modified reference trace, applying a measurement trace, and creating an adjustment trace), or a polishing system including a controller configured to perform suitable operations.

[0009] Implementations of the method, computer program product, and / or system may include one or more of the following features.

[0010] The modified reference trace may include a series of equivalent thickness values, and the modified measurement trace may include a series of actual thickness values. To create the modified reference trace, at least a portion of the initial reference trace may be applied to a neural network. To create the initial reference trace, the raw signal values of the preliminary reference trace may be converted to thickness values. User input may be received to select a reference trace from a plurality of reference traces. Creating the modified reference trace may include scanning the sensors of the in-situ electromagnetic induction monitoring system over the entire bare doped reference semiconductor wafer.

[0011] Creating the adjustment trace may include scaling the difference between the modified reference trace and the modified measurement trace. The adjustment trace A(x) may be calculated as A(x) = (T(x) - S(x) - b) / k. In the above formula, T(x) is the modified measurement trace, S(x) is the modified reference trace, and b and k are constants. The constants b and k depend on the configuration of the sensors of the in-situ monitoring system.

[0012] At least a portion of the measurement traces applied to the neural network may include a portion corresponding to the edge region of the substrate. At least a portion of the measurement traces applied to the neural network need not include a portion corresponding to the central region of the substrate. The neural network may be trained using a plurality of training traces representing measurements of one or more training substrates having a conductive layer on an undoped semiconductor wafer, the plurality of training traces including various training traces corresponding to various thicknesses and various edge profiles of the conductive layer.

[0013] The implementation aspects may include one or more of the following advantages. During the monitoring of substrate processing, such as polishing, inaccuracies that may occur in the correlation between the measured eddy current signal and the thickness of the conductive layer caused by the doping of the underlying semiconductor wafer can be reduced, particularly at the edges of the substrate. The eddy current signal adjusted using a correction process or the adjusted thickness of the conductive layer can be more accurate. The system can correct for distortion in a portion of the signal corresponding to the edge of the substrate. Using the adjusted eddy current signal and / or the adjusted conductive layer, control parameters during the polishing process can be determined and / or the end point of the polishing process can be determined. The reliability of determining the control parameters and detecting the end point can be improved, under-polishing of the wafer can be avoided, and non-uniformities within the wafer can be reduced.

[0014] Details of one or more implementation aspects are described in the accompanying drawings and the following description. Other aspects, features, and advantages will become apparent from the description, the drawings, and the claims.

Brief Description of the Drawings

[0015]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Mode for Carrying Out the Invention

[0016] One of the monitoring techniques for the polishing process is to induce a current in the conductive layer on the substrate. The induced current can be measured by an in-situ induction monitoring system during polishing to generate a signal. Assuming that the outermost layer during polishing is the conductive layer, the signal from the sensor should depend on the thickness of the conductive layer. Based on the monitoring, for example, the polishing control parameters can be adjusted so that the position of the layer has substantially the same thickness after polishing, or the polishing of the layer positions is completed almost simultaneously. The above profile control can be referred to as real-time profile control (RTPC). Furthermore, the polishing process can end based on an indication that the monitored thickness has reached the desired end-point thickness.

[0017] The in-situ monitoring system may be subject to signal distortion due to measurements at positions close to the edge of the substrate. For example, the induction monitoring system can generate a magnetic field. Near the edge of the substrate, since the magnetic field only partially overlaps with the conductive layer of the substrate, the signal can be artificially low. Various techniques can be used to correct the distortion. For example, the signal can be supplied to an artificial neural network to generate a modified signal.

[0018] In practice, the magnetic field generated by the eddy current sensor may extend down to the underlying substrate without stopping within the conductive layer. Without being limited to a particular theory, the skin depth of these ferromagnetic materials for the electromagnetic frequencies used in the eddy current sensor can be greater than the thickness of the conductive layer and the underlying semiconductor wafer. As a result, the signal generated by the eddy current sensor may depend on the conductivity of the semiconductor wafer.

[0019] If the semiconductor wafer is not doped, for example, when used as a "blank" wafer for system calibration or a basic substrate wafer commonly used, the electrical resistance of the wafer may be high enough that the presence of the wafer may not have a detectable effect on the eddy current signal. However, in actual device manufacturing, the wafer is typically doped for various purposes, for example, highly doped. In this situation, the signal generated by the eddy current sensor may contribute significantly from the semiconductor wafer depending on the conductivity of the wafer. For this reason, thickness measurements based on the signal captured by the eddy current sensor can be inaccurate. This inaccuracy can be corrected by using techniques, for example, by taking into account the contribution to the signal from the semiconductor wafer. However, the above correction may introduce additional errors at the edge of the substrate when edge reconstruction techniques are utilized.

[0020] However, the traces from the substrate and the doped wafer can be run separately through the edge reconstruction algorithm. The modified trace of the doped wafer can be subtracted from the modified measurement trace of the substrate. The resulting difference will be close to the actual thickness of the layer during polishing. Further, the difference can be scaled to correct the sensor configuration.

[0021] Figures 1 and 2 show an embodiment of the polishing station 20 of a chemical mechanical polishing system. The polishing station 20 includes a rotatable disk-shaped platen 24 on which a polishing pad 30 is located. The platen 24 is operable to rotate about an axis 25. For example, a motor 22 can rotate a drive shaft 28 to rotate the platen 24. The polishing pad 30 can be a two-layer polishing pad having an outer polishing layer 34 and a softer backing layer 32.

[0022] The polishing station 20 can include a supply port or a combined supply rinse arm 39 for dispensing a polishing liquid 38, such as a polishing slurry, onto the polishing pad 30. The polishing station 20 can include a pad conditioner device with a conditioning disk for maintaining the surface roughness of the polishing pad.

[0023] The carrier head 70 is operable to hold the substrate 10 against the polishing pad 30. The carrier head 70 is suspended from a support structure 72, such as a carousel or a track, and is connected to a carrier head rotation motor 76 by a drive shaft 74, whereby the carrier head can rotate about an axis 71. Optionally, the carrier head 70 can vibrate laterally, for example, by a slider on a carousel, by movement along a track, or by rotational vibration of the carousel itself.

[0024] The carrier head 70 can include a retaining ring 84 for holding the substrate. In some implementations, the retaining ring 84 can include a highly conductive portion. For example, the carrier ring can include a thin lower plastic portion 86 that contacts the polishing pad and a thick upper conductive portion 88. In some implementations, the highly conductive portion is a metal, such as the same metal as the layer being polished, for example, copper.

[0025] In the process, the platen is rotated about its central axis 25, the carrier head is rotated about its central axis 71, and is translated laterally across the upper surface of the polishing pad 30. When there are multiple carrier heads, each carrier head 70 can independently control its polishing parameters. For example, each carrier head can independently control the pressure applied to each substrate.

[0026] The carrier head 70 may include a flexible membrane 80 having a substrate mounting surface that contacts the back side of the substrate 10 and a plurality of pressure chambers 82 for applying different pressures to different zones on the substrate 10, such as different radial zones.

[0027] In some embodiments, the polishing station 20 includes a temperature sensor 64 for monitoring the temperature of the polishing station or the temperature of a component within the polishing station / components within the polishing station. Although shown in FIG. 1 as being arranged to monitor the temperature of the polishing pad 30 and / or the slurry 38 on the pad 30, the temperature sensor 64 may be disposed inside the carrier head 70 to measure the temperature of the substrate 10. The temperature sensor 64 can be in direct contact with (i.e., a contact sensor) the outermost layer of the polishing pad or the substrate 10 (which may be a conductive layer) to accurately monitor the temperature of the outermost layer of the polishing pad or the substrate. The temperature sensor may be a non-contact sensor (e.g., an infrared sensor). In some embodiments, for example, multiple temperature sensors are included in the polishing station 22 to measure the temperature of different components within the polishing station / different components within the polishing station. The temperature can be measured in real time, for example, periodically, and / or in relation to real-time measurements performed by an eddy current system. The monitored temperature can be used when adjusting the in-situ eddy current measurement.

[0028] Referring to FIG. 3A, a substrate 10 including conductive material on and / or embedded in a patterned dielectric layer can be polished using a polishing system. For example, substrate 10 can include a dielectric layer 14, such as silicon oxide or a high-k dielectric, covering and filling trenches with a conductive material 16, such as a metal, such as copper, aluminum, cobalt, or titanium. Optionally, the trenches can be lined with a barrier layer 18, such as tantalum or tantalum nitride, to separate the conductive material 16 from the dielectric layer 14. The conductive material 16 in the trenches can provide vias, pads, and / or interconnects in a completed integrated circuit. Although dielectric layer 14 is shown as being directly deposited on semiconductor wafer 12, one or more other layers can be inserted between dielectric layer 14 and wafer 12.

[0029] Semiconductor wafer 12 can be a silicon wafer, such as a single crystal silicon, although other semiconductor materials are possible. Further, semiconductor wafer 12 can be doped, for example, with p-type or n-type doping. The doping can be made uniform laterally across the entire wafer, or the wafer can be selectively doped, for example, as appropriate for the fabrication of transistors in an integrated circuit using the semiconductor wafer.

[0030] Initially, conductive material 16 covers the entire dielectric layer 14. As polishing progresses, most of the conductive material 16 is removed, exposing the barrier layer 18 (see FIG. 3B). Next, as polishing continues, the patterned top surface of the dielectric layer 14 is exposed (see FIG. 3C). Next, additional polishing can be used to control the depth of the trenches containing the conductive material 16.

[0031] In some implementations, the polishing system includes additional polishing stations. For example, the polishing system can include two or three polishing stations. For example, the polishing system can include a first polishing station with a first electromagnetic induction monitoring system and a second polishing station with a second electromagnetic induction current monitoring system.

[0032] For example, in a process, bulk polishing of a conductive layer on a substrate can be performed at a first polishing station, and polishing can be stopped when a target thickness of the conductive layer remains on the substrate. Next, the substrate is transferred to a second polishing station, and the substrate can be polished down to a layer below, such as a patterned dielectric layer.

[0033] Returning to FIG. 1, the polishing system includes an in-situ electromagnetic induction monitoring system 100 that can be coupled to or considered to include a controller 90. A rotary coupler 29 can be used to electrically connect components of the rotatable platen 24, such as sensors of the in-situ monitoring system, to components outside the platen, such as drive and sensing circuits or the controller 90.

[0034] The in-situ electromagnetic induction monitoring system 100 is configured to generate a signal that depends on the depth of a conductive material 16, such as a metal. The electromagnetic induction monitoring system can operate by generating eddy currents in a sheet of conductive material on top of the dielectric layer or by generating a current in a conductive loop formed in a trench of the dielectric layer on the substrate.

[0035] Using the electromagnetic induction monitoring system 100 as an eddy current monitoring system, the thickness of the conductive layer can be monitored by inducing eddy currents in the conductive sheet. Alternatively, as an induction monitoring system, the electromagnetic induction monitoring system can operate by inductively generating a current in a conductive loop formed in the dielectric layer 14 of the substrate 10 for monitoring purposes, as described, for example, in U.S. Patent Publication No. 2015-0371907.

[0036] In a process, a polishing system can use an in-situ monitoring system 100 to determine when a conductive layer has reached a target thickness, e.g., a target depth of metal in a trench or a target thickness of a metal layer on a dielectric layer, and then stop the polishing. Alternatively or additionally, the polishing system can use the in-situ monitoring system 100 to determine a difference in thickness of the conductive material 16 across the substrate 10 and use this information to adjust the pressure in one or more chambers 82 of the carrier head 80 during polishing to reduce polishing non-uniformity.

[0037] A recess 26 can be formed in the platen 24, and optionally, a thin section 36 can be formed in the polishing pad 30 that covers the recess 26. The recess 26 and the thin section 36 can be arranged to pass under the substrate 10 during a portion of the platen rotation, regardless of the translational position of the carrier head. Assuming the polishing pad 30 is a two-layer pad, the thin section 36 can be constructed by removing a portion of the backing layer 32 and optionally by forming a recess in the bottom of the polishing layer 34. The thin section can be, for example, optionally light transmissive if an in-situ optical monitoring system is integrated into the platen 24.

[0038] The in-situ monitoring system 100 can include a sensor 102 installed in the recess 26. The sensor 102 can include a magnetic core 104 at least partially disposed in the recess 26 and at least one coil 106 wound around a portion of the core 104. A drive and sense circuit 108 is electrically connected to the coil 106. The drive and sense circuit 108 generates a signal that can be transmitted to the controller 90. Although shown outside the platen 24, some or all of the drive and sense circuit 108 can be installed in the platen 24.

[0039] Referring to FIGS. 1 and 4, drive and sense circuit 108 applies an AC current to coil 106, and coil 106 generates a magnetic field 150 between two poles 152a and 152b of core 104. Although a C-shaped core is shown in FIG. 4, other cores, such as E-shaped, I-shaped, etc. are possible. In the process, when substrate 10 intermittently overlays sensor 102, a part of magnetic field 150 extends into substrate 10.

[0040] Circuit 108 may include a capacitor connected in parallel with coil 106. Both coil 106 and the capacitor may form an LC resonant tank.

[0041] When monitoring the thickness of the conductive layer on the substrate is desired, when magnetic field 150 reaches the conductive layer, magnetic field 150 may pass through and generate a current (when the target is a loop), or generate eddy currents (when the target is a sheet). As a result, the effective impedance of the LC circuit is changed.

[0042] However, magnetic field 150 may also penetrate semiconductor substrate 12. For this reason, the effective impedance of the LC circuit, and thus the signal from drive and sense circuit 108, may depend on the doping of semiconductor substrate 12 and the resulting conductivity.

[0043] Drive and sense circuit 108 may include a marginal oscillator coupled to combined drive / sense coil 106, and the output signal may be the current required to maintain the peak-to-peak amplitude of the sine wave oscillation at a constant value, as described, for example, in U.S. Patent No. 7,112,960. Other configurations are possible for drive and sense circuit 108. For example, separate drive and sense coils may be wound around the core. For example, as described in U.S. Patent No. 6,975,107, drive and sense circuit 108 can apply a current at a fixed frequency, and the signal from drive and sense circuit 108 can be the phase shift of the current of the sense coil with respect to the drive coil, or the amplitude of the sensed current.

[0044] Referring to FIG. 2, as the platen 24 rotates, the sensor 102 sweeps under the substrate 10. By sampling the signals from the circuit 108 at a specific frequency, the circuit 108 generates measurement values at a series of sampling zones 94 across the entire substrate 10. For each sweep, the measurement values at one or more of the sampling zones 94 can be selected or combined. Thus, the selected or combined measurement values over a plurality of sweeps provide a series of values that change over time.

[0045] The polishing station 20 may also include a position sensor 96, such as an optical interrupter, to sense when the sensor 102 is under the substrate 10 and when the sensor 102 is away from the substrate. For example, the position sensor 96 can be attached at a fixed position on the opposite side of the carrier head 70. A flag 98 can be attached around the platen 24. The attachment point and length of the flag 98 are selected such that the position sensor 96 can send a signal when the sensor 102 sweeps under the substrate 10.

[0046] Alternatively or additionally, the polishing station 20 may include an encoder for determining the angular position of the platen 24.

[0047] Returning to FIG. 1, a controller 90, such as a general-purpose programmable digital computer, receives signals from the sensor 102 of the in-situ monitoring system 100. Since the sensor 102 sweeps under the substrate 10 each time the platen 24 rotates, information regarding the depth of a conductive layer, such as the bulk layer of a trench or a conductive material, is accumulated in-situ (once per rotation of the platen). The controller 90 can be programmed to sample the measurement values from the in-situ monitoring system 100 when the substrate 10 is entirely over the sensor 102.

[0048] Furthermore, the controller 90 can be programmed to calculate the radial position of each measurement value and classify the measurement values into radial ranges. By arranging the measurement values in radial ranges, data regarding the conductive film thickness of each radial range can be supplied to the controller (e.g., controller 90), and the polishing pressure profile applied by the carrier head can be adjusted. The controller 90 can also be programmed to apply endpoint detection logic to a series of measurement values generated by the signals of the in-situ monitoring system 100 to detect the polishing endpoint.

[0049] Each time the platen 24 rotates, the sensor 102 sweeps under the substrate 10, so information regarding the thickness of the conductive layer is continuously accumulated in-situ in real time. During polishing, the measurement values from the sensor 102 can be displayed on an output device so that the operator of the polishing station can visually monitor the progress of the polishing process.

[0050] Referring to FIGS. 2 and 5, a change in the position of the sensor head with respect to the substrate 10 can result in a change in the signal from the in-situ monitoring system 100. That is, when the sensor head scans the entire substrate 10, the in-situ monitoring system 100 makes measurements for a plurality of regions 94, e.g., measurement spots 211, at different positions on the substrate 10. The regions 94 may partially overlap.

[0051] FIG. 6 shows a graph of the signal 220 from the in-situ monitoring system 100 as the sensor 102 passes under the substrate 10 once. This signal 220 can be referred to as a "trace" of the entire substrate. The signal 220 is composed of a series of individual measurement values from the sensor head as the sensor head sweeps under the substrate. The graph can be a function of the measurement time or the measurement position on the substrate, e.g., the radial position. In either case, different portions of the signal 220 correspond to measurement spots 211 at different positions on the substrate 10 scanned by the sensor 102. Thus, the graph shows the corresponding measured signal values from the signal 220 for a given position on the substrate scanned by the sensor head.

[0052] Referring to FIGS. 5 and 6, signal 220 includes a first portion 222 corresponding to a position within edge region 203 of substrate 10 when sensor 102 crosses the front edge of substrate 10, a second portion 224 corresponding to a position within central region 201 of substrate 10, and a third portion 226 corresponding to a position within edge region 203 when sensor 102 crosses the rear edge of substrate 10. The signal may also include a portion 228 corresponding to an out-of-substrate measurement, i.e., a signal generated when the sensor head scans a region beyond edge 204 of substrate 10 in FIG. 5.

[0053] Edge region 203 may correspond to the portion of the substrate where measurement spot 211 of the sensor head overlaps with substrate edge 204. Central region 201 may include an annular anchor region 202 adjacent to edge region 203 and an inner region 205 surrounded by anchor region 202. The sensor head may scan these regions along its path 210 and generate a series of measurements corresponding to a series of positions along path 210.

[0054] In the first portion 222, the signal intensity rises from an initial intensity (usually the signal that occurs when the substrate and the carrier head are absent) to a higher intensity. This is caused by the monitoring position shifting from initially only slightly overlapping the substrate at the edge 204 of the substrate (generating a lower initial value) to almost completely overlapping the substrate (generating a higher value). Similarly, in the third portion 226, the signal intensity decreases as the monitoring position shifts to the edge 204 of the substrate.

[0055] The second portion 224 is shown as flat for simplicity, but this is for simplification, and the actual signal of the second portion 224 is likely to include variations due to both noise and variations in layer thickness. The second portion 234 corresponds to the monitoring position scanning the central region 201. The second portion 224 includes two sub-portions 230 and 232 caused by the monitoring position scanning the anchor region 202 of the central region 201, and a sub-portion 234 caused by the monitoring position scanning the inner region 205 of the central region 201.

[0056] As described above, the fluctuations in the signal intensities in regions 222 and 226 are not due to inherent fluctuations in the thickness or conductivity of the monitored layer, but are caused in part by the measurement regions of sensor 106 that overlap the substrate edge. As a result, this distortion of signal 220 can cause errors in the calculation of the characteristic values of the substrate, such as the thickness of the layer near the substrate edge. To address this issue, controller 90 may include a neural network, such as neural network 300 of FIG. 7, to generate a modified signal corresponding to those positions based on the measurement signals corresponding to one or more positions of substrate 10.

[0057] Referring now to FIG. 7, neural network 300 is configured to generate a modified signal that reduces and / or removes the distortion of the signal values calculated near the substrate edge when properly trained. Neural network 300 receives a set of inputs 304 and processes inputs 304 through one or more neural network layers to generate a set of outputs 350. The layers of neural network 300 include an input layer 310, an output layer 330, and one or more hidden layers 320.

[0058] Each layer of neural network 300 includes one or more neural network nodes. Each neural network node of a neural network layer receives one or more node input values (from inputs 304 into neural network 300 or from the outputs of one or more nodes of a preceding neural network layer), processes the node input values according to one or more parameter values, generates an activation value, and optionally applies a non-linear transformation function (e.g., a sigmoid function or a tanh function) to the activation value to generate the output of the neural network node.

[0059] Each node of input layer 310 receives one of inputs 304 into neural network 300 as a node input value.

[0060] The input 304 to the neural network includes measured signal values from the in-situ monitoring system 100 for a plurality of different spots 211 on the substrate 10, for example, from the first measured signal value 301, the second measured signal value 302 to the nth measured signal value 303. The measured signal values can be individual values of a series of values of the signal 220.

[0061] Generally, the plurality of different positions include the edge region 203 of the substrate 10 and, optionally, positions within the anchor region 202. In some implementations, the plurality of different positions exist only in the edge region 203 and the anchor region 202. In other implementations, the plurality of different positions span all regions of the substrate.

[0062] These measured signal values are received at the signal input node 344. Optionally, the input node 304 of the neural network 300 may also include one or more process state signals 304, for example, one or more state input nodes 316 that receive measurements of the wear of the pad 30 of the polishing apparatus 20.

[0063] The nodes of the hidden layer 320 and the output layer 330 are shown as receiving inputs from all nodes of the preceding layer. This is the case for a fully connected feedforward neural network. However, the neural network 300 can be a non-fully connected feedforward neural network or a non-feedforward neural network. Further, the neural network 300 can include at least one of one or more fully connected feedforward layers, one or more non-fully connected feedforward layers, and one or more non-feedforward layers.

[0064] The neural network generates a set of modified signal values 350 at the nodes of the output layer 330, i.e., the "output nodes" 350. In some implementations, there is an output node 350 for each measurement signal from the in-situ monitoring system supplied to the neural network 300. In this case, the number of output nodes 350 may correspond to the number of signal input nodes 304 of the input layer 310.

[0065] For example, the number of signal input nodes 344 may be equal to the number of measurement values in the edge region 203 and the anchor region 202, and there may be the same number of output nodes 350. Thus, each output node 350 generates a modified signal corresponding to each measurement signal supplied as an input to the signal input node 344, e.g., the first modified signal 351 of the first measurement signal 301, the second modified signal 352 of the second measurement signal 302, and the nth modified signal 353 of the nth measurement signal 303.

[0066] In some implementations, the number of output nodes 350 is less than the number of input nodes 304. In some implementations, the number of output nodes 350 is less than the number of signal input nodes 344. For example, the number of signal input nodes 344 may be equal to the number of measurement values in the edge region 203, or may be equal to the number of measurement values in the edge region 203 and the anchor region 202. Also in this case, each output node 350 of the output layer 330 generates a modified signal corresponding to each measurement signal supplied as a signal input node 304, e.g., the first modified signal 351 for the first measurement signal 301, but only for the signal input node 354 that receives a signal from the edge region 203.

[0067] The polishing apparatus 100 may use the neural network 300 to generate modified signals. Next, the modified signals may be used to determine the thickness of each position within the first group of positions of the substrate, e.g., positions within the edge region (and optionally the anchor region). For example, referring back to FIG. 6, the modified signal value of the edge region may provide the modified portion 230 of the signal 220.

[0068] In some implementations, for a changed signal value corresponding to a predetermined measurement position, the neural network 500 may be configured such that only input signal values from measurement positions within a predetermined distance from that predetermined position are used when determining the changed signal value.

[0069] To train the neural network, the sensor 102 of the in-situ monitoring system 100 can be used to generate a profile of a reference substrate. Further, a ground truth measurement of the thickness of the reference substrate can be obtained. These measurements can be performed on the positions to be processed by the neural network. The system can generate a ground truth measurement of the thickness using an electrical impedance measurement method such as the four-point probe method. The signal value from the reference substrate is applied to the input 304, the ground truth measurement value is applied to the output 350, and the system is run in a training mode (such as gradient descent with backpropagation).

[0070] The reference substrate can include a blank undoped wafer on which a conductive material of uniform thickness is deposited. The amount of the conductive material can be selected to simulate the presence of a doped wafer.

[0071] The reference substrate can also include a sample device substrate that is at the same processing stage as the device substrate on which the in-situ monitoring system is used for polishing control, for example, a substrate having layers with different edge profiles.

[0072] As described above, the signals generated by the in-situ monitoring system also include contributions from doped wafers. If not properly processed, attempts to correct the contribution of the signal from the doped wafer, for example, when edge reconstruction techniques are utilized, additional errors may occur at the substrate edge.

[0073] Referring to FIG. 8, a reference trace 420 of the entire blank doped wafer is created. This reference trace 420 is created before polishing the substrate. The blank doped wafer has the same doping profile as the wafer used for the device substrate to be polished. In some implementations, the reference trace is created by scanning a sample blank doped wafer, e.g., a sacrificial wafer, using the sensor 102 of the in-situ monitoring system 100. For example, the reference trace can be created by a manufacturing operator. Alternatively, the system manufacturer can create reference traces of wafers with various different dopings (e.g., concentration and / or doping material), and these traces can be stored in a library. Next, the operator can select one of the reference traces from the library, e.g., from a drop-down menu or a similar user interface, that most closely corresponds to the doping of the wafer of the device substrate to be polished.

[0074] The raw signal values of the reference trace 420 from the sensor 102 can be converted to thickness values (represented by the reference trace 420) using a correlation curve.

[0075] FIG. 9 shows a correlation curve 510 between the thickness of a conductive layer of a predetermined resistivity and the signal from the electromagnetic induction monitoring system 100 for the predetermined resistivity. D START represents the initial thickness of the conductive layer, and S START is the desired signal value corresponding to the initial thickness D START . D FINAL represents the final thickness of the conductive layer, and S FINAL is the desired signal value corresponding to the final thickness. K is a constant representing the value of the signal when the thickness of the conductive layer is zero.

[0076] The relationship curve 510 can be represented in the controller 90 by a function, e.g., a polynomial function, e.g., a quadratic function, a cubic function, or a higher-order function. The correlation between the signal X(x) and the thickness D(x) can be represented by the following equation. X(x)=W1·D(x) 2+W2·D(x)+W3 (Equation 1) In the above equation, W1, W2, and W3 are real coefficients. Therefore, the controller may store the values of the coefficients of the function, for example, W1, W2, and W3, and the resistivity ρ0 to which the relationship curve 510 is applied. Further, this relationship may be represented by a linear function, a Bézier curve, or a non-polynomial function such as an exponential function or a logarithm.

[0077] Using the relationship curve 510, the signal value of the unprocessed signal 420 can be converted from the reference wafer to an "equivalent" thickness measurement value. That is, although there is no conductive layer on the doped reference wafer, the measurement value can be expressed as a thickness value. These are "equivalent" thickness values. This is because each is the thickness of an equivalent conductive layer on an undoped wafer and generates the same signal as the doped reference wafer.

[0078] Next, referring back to FIG. 8, the reference trace 420' is processed by the neural network as if it were a normal signal in order to execute an edge reconstruction algorithm on the reference trace. As a result, a modified reference trace 450 having a portion with a modified signal value 430 is created.

[0079] In some implementations, the conversion to thickness is performed beforehand, and a reference trace 420' containing the thickness value is stored in the library (and selected by the operator). In some implementations, the conversion to thickness and the edge reconstruction are performed beforehand, and the modified reference trace 450 is stored in the library (and selected by the operator).

[0080] During the polishing process, the substrate 10 is monitored by an in-situ monitoring system, and a measurement trace 220 of the substrate 10 is created in each sweep of the sensors 102 across the entire substrate 10. Since this measurement trace 220 includes contributions from both the conductive layer being polished and the doped wafer beneath it, it may also be referred to as a "total" trace or signal.

[0081] Using the relationship curve 510 (see FIG. 9), the signal value of the signal 220 from the substrate to be polished can be converted into a thickness measurement value (represented by the measurement trace 220´).

[0082] Each measurement trace 220´ is processed by a neural network as described above, and a modified measurement trace 250 having a portion with a modified value 230 is created.

[0083] In some implementations, the conversion from the unprocessed signal to thickness can be performed on both the reference wafer and the substrate to be polished after edge reconstruction has been performed.

[0084] The controller 190 can create an adjustment trace 480 to correct the wafer doping. Creating the adjustment trace includes subtracting the modified reference trace 450 from the modified measurement trace 250. Assuming that the modified reference trace 450 is represented by S(x) and the modified measurement trace 250 is represented by T(x), where x is the radial position, T(s) - S(x) becomes the apparent thickness trace.

[0085] In some configurations of the sensor 102, the contribution to the trace from the doped wafer and substrate is not a simple superposition. Rather, the apparent thickness of the conductive layer may be somewhat thinner than the actual thickness. This problem can become more prominent at higher drive frequencies.

[0086] However, for certain sensor configurations (e.g., drive frequency, core shape and dimensions, coil position and number of turns, etc.), there generally appears to be a linear relationship between the actual thickness and the apparent thickness. This relationship is shown in FIG. 10. The function 520 that associates the apparent thickness with the actual thickness can be represented as a linear function with a slope of k and a y-intercept (where the thickness must be non-zero) of b. These values of k and b are determined empirically by testing and can vary between sensor configurations. The value of k tends to be 1 or less, for example, a value from 0.7 to 1.

[0087] Therefore, the adjusted film thickness profile A(x) of the conductive layer on the substrate can be calculated by A(x) = (T(x) - S(x) - b) / k.

[0088] The point at which the adjusted thickness value A(x) reaches the target thickness value D TARGET can be called the end point. Similarly, the adjusted thickness value A´(x) can be used for controlling polishing parameters, for example, calculating the polishing pressure to reduce non-uniformity.

[0089] In some cases, the relationship between the apparent thickness and the actual thickness of a specific sensor configuration may not be linear. In such cases, more complex equations, such as polynomials, can be used to calculate the actual thickness.

[0090] In some implementations, the raw signal is normalized before being converted to a thickness value. This technique is applicable to both the reference trace 420 and the substrate trace 220. For example, a calibrated signal X´(x) can be generated as follows. X´(x) = G * X(x) - ΔK (Equation 2) In the above equation, G is the gain and ΔK is the offset, which are determined experimentally in the in-situ monitoring system using a blank wafer having a conductive layer of known thickness and conductivity. X(x) represents the raw signal value from either the reference trace 420 or the substrate trace 220, which is appropriate for the processing of each trace. Next, the calibrated signal X´(x) is used in the correlation curve, for example, instead of X(x) in Equation 1 above, to determine the thickness value.

[0091] Furthermore, when the raw signal value is converted to a thickness value, the resistivity of the layer can be considered. For example, the thickness value calculated using a correlation curve, such as Equation 1 above, can be adjusted based on the resistivity of the layer to provide a corrected thickness value. This technique can be used for both the reference trace 420 and the substrate trace 220.

[0092] The corrected thickness value D´(x) can be calculated as follows. D´(x) = D(x) * (ρ X / ρ0) (Equation 3) In the above equation, ρ X is the resistivity of the conductive layer, ρ0 is the resistivity to which the relationship curve 410 (and the values W1, W2, W3) applies, and D(x) represents the initial thickness value calculated using the correlation curve (from either the reference trace 420 or the substrate trace 220 as required). The edge reconstruction algorithm can be applied to the modified thickness value D´(x) instead of the initial thickness value D(x).

[0093] In addition to the inter-substrate variation in resistivity, a change in the temperature of the layer can cause a change in the resistance of the conductive layer. For example, the conductive layer can become hotter as the polishing progresses and thus become more conductive (lower resistivity). In particular, the controller executing the process can calculate the resistivity ρ T of the conductive layer at the real-time temperature T(t). The real-time temperature T(t) can be determined from the temperature sensor 64. In some implementations, the adjusted resistivity ρ T is calculated based on the following equation. ρ T = ρ X [1 + α(T(t) - T ini )] (Equation 4) In the above equation, T ini is the initial temperature of the conductive layer when the polishing process is started. Next, for example, in the above Equation 3 (or the calculation of the gain and offset of Equation 2), the adjusted resistivity ρ X is used instead of the resistivity ρ T .

[0094] In a situation where the polishing process is performed at room temperature, T ini can take on a value of approximately 20°C. ρ X is the resistivity of the conductive layer at T ini and can be at room temperature. Usually, α is a known value, which may be described in the literature or obtainable from experiments. The unprocessed signal 220 includes contributions from the underlying doped wafers, but the value of α for the conductive layer can be used as a first approximation when calculating the thickness value of the trace 220´.

[0095] In some implementations, the temperatures T and T used when adjusting the measured eddy current signal ini are, for example, the temperature of the conductive layer measured by a temperature sensor of the carrier head. In some implementations, the temperatures T and T ini may be the temperature of the polishing pad or the temperature of the slurry instead of the temperature of the conductive layer.

[0096] The polishing apparatus and method described above can be applied to various polishing systems. Relative movement can be provided between the polishing surface and the substrate by moving the polishing pad, or the carrier head, or both. For example, the platen may orbit rather than rotate. The polishing pad may be a circular (or any other arbitrary shape) pad fixed to the platen. Some aspects of the endpoint detection system may be applicable to a linear polishing system where, for example, the polishing pad moves linearly on a continuous belt or a reel-to-reel belt. The polishing layer can be a standard (e.g., polyurethane with or without fillers) polishing material, a soft material, or a fixed abrasive material. The term relative positioning is used to refer to the relative positioning within the system or the substrate. It should be understood that during the polishing process, the polishing surface and the substrate can be held in a vertical direction or some other direction.

[0097] The functional operations of the controller 90 can be implemented by digital electronic circuitry, or by computer software, firmware, or hardware, and structural equivalents thereof, or combinations thereof. The computer software can be implemented as one or more computer program products, i.e., one or more computer programs tangibly embodied in a non-transitory computer-readable storage medium for execution by, or to control the operation of, a data processing apparatus, such as a programmable processor, a computer, or multiple processors or computers. The computer program (also referred to as a program, software, software application, or code) can be written in any form of programming language, including a compiled or interpreted language, and it can be deployed as a stand-alone program or as a module, component, subroutine, or other unit suitable for use in a computing environment. The computer program does not necessarily correspond to a file. The program can be stored in a portion of a file that holds other programs or data, in a single file dedicated to the program, or in multiple coordinated files (e.g., files that store one or more modules, subprograms, or portions of code). The computer program can be executed on one computer or on multiple computers interconnected by a communication network and distributed across one site or multiple sites.

[0098] The processes and logical flows described in this specification can be performed by one or more programmable processors executing one or more computer programs to perform functions by operating on input data and generating output. The processes and logical flows can also be performed by, and the apparatus can be implemented as, special purpose logic circuitry, e.g., an FPGA (field programmable gate array) or an ASIC (application specific integrated circuit).

[0099] Some embodiments of the present invention have been described. However, it will be understood that various modifications can be made without departing from the spirit and scope of the present invention. For example, while the above description has focused on chemical mechanical polishing, the control system can be adapted to other semiconductor processing techniques, such as etching or deposition, such as chemical vapor deposition. Further, the technique can be applied to in-line or stand-alone metrology systems rather than in-situ monitoring. Accordingly, other embodiments are within the scope of the following claims.

Claims

1. A method for polishing a substrate, comprising: storing or creating a first reference trace representing a measurement value of a bare doped reference semiconductor wafer by an in-situ electromagnetic induction monitoring system; bringing a substrate having a conductive layer disposed thereon into contact with a polishing pad; generating a relative motion between the substrate and the polishing pad; monitoring the substrate using the in-situ electromagnetic induction monitoring system when the conductive layer is polished to create a measurement trace that depends on the thickness of the conductive layer; using the measurement trace, the first reference trace, and a neural network configured to reduce distortion of a signal value calculated near an edge of the substrate of the measurement trace to create an adjustment trace that at least partially corrects a contribution of the conductivity of the semiconductor wafer to the measurement trace and distortion of the signal value calculated near the edge of the substrate of the measurement trace; based on the adjustment trace, at least one of stopping polishing or changing polishing parameters; and a method comprising the steps of:

2. The method according to claim 1, wherein creating comprises applying at least a portion of the measurement trace to the neural network to create a modified measurement trace and subtracting the first reference trace from the modified measurement trace.

3. The method according to claim 2, wherein the first reference trace represents a measurement value of a bare doped reference semiconductor wafer modified by the neural network to reduce distortion of a signal value calculated near an edge of the substrate when applied to the neural network.

4. The method according to claim 2 or 3, wherein at least a portion of the measurement trace applied to the neural network includes a portion corresponding to an edge region of the substrate.

5. Creating the first reference trace includes scanning the sensor of the in-situ electromagnetic induction monitoring system across the entire bare doped reference semiconductor wafer, according to the method of claim 1.

6. Scanning the bare doped reference semiconductor wafer using the sensor of the in-situ electromagnetic induction monitoring system creates a preliminary reference trace having unprocessed signal values, and the method includes converting the unprocessed signal values of the preliminary reference trace to thickness values to create an initial reference trace, and applying at least a portion of the initial reference trace to the neural network to create the first reference trace, according to the method of claim 5.

7. A computer program product tangibly embodied on a non-transitory computer-readable medium, causing one or more computers to save or create a first reference trace representing measurement values of a bare doped reference semiconductor wafer by an in-situ electromagnetic induction monitoring system, receive a measurement trace dependent on the thickness of the conductive layer from the in-situ electromagnetic induction monitoring system when the conductive layer on a substrate disposed on the semiconductor wafer is polished, create an adjustment trace that at least partially corrects the contribution of the conductivity of the semiconductor wafer to the measurement trace and the distortion of the signal values calculated near the edge of the substrate of the measurement trace, using the measurement trace, the first reference trace, and a neural network configured to reduce the distortion of the signal values calculated near the edge of the substrate of the measurement trace, and cause at least one of stopping the polishing or changing the polishing parameters based on the adjustment trace A computer program product including instructions for execution.

8. The computer program product according to claim 7, wherein the instructions for creating the adjustment trace include instructions for applying at least a part of the measurement trace to the neural network to create a modified measurement trace, and for subtracting the first reference trace from the modified measurement trace.

9. The computer program product according to claim 8, wherein the first reference trace represents measured values of a bare doped reference semiconductor wafer that has been modified by the neural network so as to reduce distortion of signal values calculated near the edge of the substrate when applied to the neural network.

10. The computer program product according to claim 8, wherein the instructions for creating the adjustment trace include instructions for dividing the difference between the first reference trace and the modified measurement trace by a constant.

11. In the instructions for creating the adjustment trace, the adjustment trace A(x) is calculated as follows: A(x) = (T(x) - S(x) - b) / k In the above formula, T(x) is the modified measurement trace, S(x) is the first reference trace, and b and k are constants. The computer program product according to claim 8.

12. The computer program product according to claim 8, wherein the at least a part of the measurement trace applied to the neural network includes a part corresponding to the edge region of the substrate.

13. A polishing system, a support for holding a polishing pad, a carrier head for holding a substrate disposed on a semiconductor wafer in contact with the polishing pad, an in-situ electromagnetic induction monitoring system for monitoring the substrate when the conductive layer on the substrate is polished to create a measurement trace that depends on the thickness of the conductive layer. A controller configured to save or create a first reference trace representing a measurement value of a bare doped reference semiconductor wafer by the in-situ electromagnetic induction monitoring system, receive the measurement trace from the in-situ electromagnetic induction monitoring system, using the measurement trace, the first reference trace, and a neural network configured to reduce distortion of a signal value calculated near an edge of the substrate of the measurement trace, create an adjustment trace that at least partially corrects a contribution of a conductivity of the semiconductor wafer to the measurement trace and a distortion of the signal value calculated near the edge of the substrate of the measurement trace, and based on the adjustment trace, perform at least one of stopping polishing or changing polishing parameters and a polishing system comprising the controller

14. The polishing system according to claim 13, wherein the controller is configured to create the adjustment trace by applying at least a part of the measurement trace to a neural network to create an altered measurement trace and subtracting the first reference trace from the altered measurement trace.

15. The polishing system according to claim 14, wherein the first reference trace represents a measurement value of a bare doped reference semiconductor wafer altered by the neural network to reduce distortion of a signal value calculated near an edge of the substrate when applied to the neural network.

16. The polishing system according to claim 14, wherein the controller is configured to create the adjustment trace by dividing a difference between the first reference trace and the altered measurement trace by a constant.

17. The adjustment trace A(x) is calculated as follows: A(x) = (T(x) - S(x) - b) / k In the above formula, T(x) is the changed measurement trace, S(x) is the first reference trace, and b and k are constants. The polishing system according to claim 14.

Images