Substrate processing method and substrate processing system

Abstract

The method includes: acquiring multiple learning data containing radial etching amount distributions when the surface of a substrate is etched under multiple different etching conditions; selecting multiple effective learning data from the multiple learning data to determine the target thickness distribution; averaging the etching amount distributions of the multiple effective learning data at each radial position to obtain an averaged etching amount distribution; converting the averaged etching amount distribution to a reference etching amount to obtain a referenced etching amount distribution; converting the referenced etching amount distribution to converge to a desired range to obtain the target etching amount distribution; and adding the target etching amount distribution to the target thickness after etching the surface of the substrate to determine the target thickness distribution during grinding of the substrate surface.

Description

Technical Field

[0001] This disclosure relates to a substrate processing method and a substrate processing system. Background Technology

[0002] Patent Document 1 discloses a method for manufacturing a semiconductor wafer, which includes the following steps: planarizing at least the surface of a wafer obtained by slicing a semiconductor ingot; and etching the surface of the planarized wafer by spin etching.

[0003] Existing technical documents

[0004] Patent documents

[0005] Patent Document 1: Japanese Patent Application Publication No. 11-135464 Summary of the Invention

[0006] The problem the invention aims to solve

[0007] The technology disclosed herein appropriately controls the surface shape of the substrate after grinding and etching.

[0008] Solution for solving the problem

[0009] One aspect of this disclosure is a substrate processing method for processing a substrate, the substrate processing method comprising the following steps: determining a target radial thickness distribution when grinding the surface of the substrate; grinding the surface of the substrate based on the target thickness distribution; and etching the surface of the ground substrate, wherein the step of determining the target thickness distribution comprises: acquiring a plurality of learning data including radial etching amount distributions when the surface of the substrate is etched under a plurality of different etching conditions; selecting a plurality of valid learning data from the plurality of learning data for determining the target thickness distribution; averaging the etching amount distribution of the plurality of valid learning data at each radial position to obtain an averaged etching amount distribution; converting the averaged etching amount distribution to a reference etching amount adjustment to obtain a referenced etching amount distribution; converting the referenced etching amount distribution to converge to a desired range to obtain a target etching amount distribution; and adding the target etching amount distribution to a target thickness after etching the surface of the substrate to determine the target thickness distribution.

[0010] Invention Effects

[0011] According to this disclosure, the surface shape of the substrate after grinding and etching can be appropriately controlled. Attached Figure Description

[0012] Figure 1This is a top view showing an outline of the structure of the wafer processing system.

[0013] Figure 2 This is a side view showing an outline of the structure of the etching apparatus.

[0014] Figure 3 This is an explanatory diagram showing the situation where the nozzle moves radially.

[0015] Figure 4 This is a side view showing an example of the structure of the grinding unit and the retaining disc.

[0016] Figure 5 This is a flowchart illustrating the main processes involved in wafer processing.

[0017] Figure 6 This is a flowchart illustrating the main steps of the method for determining optimal etching conditions.

[0018] Figure 7 This is an explanatory diagram showing an example of a learning data acquisition screen displayed on a display panel.

[0019] Figure 8 This is an explanatory diagram showing an example of a learning data selection screen displayed on the display panel.

[0020] Figure 9 This is a flowchart illustrating the main steps of the method for determining the thickness distribution of the grinding target.

[0021] Figure 10 This is an illustrative diagram showing an example of multiple learning data.

[0022] Figure 11 This is an illustrative diagram showing an example of learning data excluded from the valid learning data in a set of multiple learning data sets.

[0023] Figure 12 This is an illustrative diagram showing an example of the averaged etching amount distribution.

[0024] Figure 13 This is an illustrative diagram showing an example of a baseline etch amount distribution.

[0025] Figure 14 This is an illustrative diagram showing an example of the target etching amount distribution.

[0026] Figure 15 This is an explanatory diagram showing an example of the target etching amount distribution, the first modified target etching amount distribution, and the second modified target etching amount distribution.

[0027] Figure 16 This is an explanatory diagram showing an example of the thickness distribution of a grinding target.

[0028] Figure 17This is an illustrative diagram showing the scenario of performing principal component analysis on multiple training datasets. Detailed Implementation

[0029] In the semiconductor device manufacturing process, the following process is performed: the cut surface of a disk-shaped silicon wafer (hereinafter referred to as a "wafer") cut from a single-crystal silicon ingot using a wire saw or similar tool is planarized and further smoothed to homogenize the wafer's thickness. Planarization of the cut surface is performed, for example, by surface grinding or polishing. Smoothing is also achieved, for example, by spin etching, in which etchant is supplied from above the cut surface of the wafer while it is being rotated.

[0030] Patent Document 1 disclosed the following: After planarizing at least one surface of a wafer obtained by slicing a semiconductor ingot through surface grinding, the surface is etched by spin etching. However, in the case of spin etching, the etching solution supplied from the nozzle to the center of the wafer diffuses to the outer periphery around the center due to centrifugal force, so the etching amount at the center of the wafer is smaller than the etching amount at the outer periphery. Therefore, the target thickness distribution in the radial direction of the wafer during grinding is preferably determined in consideration of the characteristics of spin etching, but such determination of the target thickness distribution during grinding has not been performed in the past.

[0031] Furthermore, in the aforementioned spin etching, the etchant supplied to the center of the wafer flows and etches through centrifugal force at the outer periphery of the wafer. On the other hand, directly below the etchant supply at the center, the supplied etchant is expelled by centrifugal force, and sometimes, during this expulsion, the flow required for etching to progress on the wafer surface fails to form. In this case, it is impossible to properly control the surface shape of the etched wafer.

[0032] Therefore, the following scheme has been proposed in the past: the wafer is rotated and etched by supplying etchant to the surface of the wafer through the nozzle while the nozzle reciprocates in a radial direction through the center of the wafer (scanning). In scanning etching, etchant is supplied to the center of the wafer while the surface of the wafer at the center is made to flow with etchant, thereby controlling the surface shape of the wafer.

[0033] However, even when scanning etching is performed, the characteristics of that etching process are not considered when determining the target thickness distribution of the wafer during grinding. Therefore, there is room for improvement in controlling the surface shape of the etched wafer.

[0034] The technology disclosed herein appropriately determines the target thickness distribution when grinding the surface of a substrate and appropriately controls the surface shape of the substrate after grinding and etching. Hereinafter, a wafer processing system as a substrate processing system according to this embodiment and a wafer processing method as a substrate processing method will be described with reference to the accompanying drawings. Furthermore, in this specification and the accompanying drawings, elements having substantially the same functional structure are omitted from repeated description by using the same reference numerals.

[0035] In the wafer processing system 1 according to this embodiment, a wafer W, which serves as a substrate and is cut from an ingot, is subjected to a process to improve the in-plane uniformity of its thickness. Hereinafter, the cut surfaces of the wafer W will be referred to as the first surface Wa and the second surface Wb. The first surface Wa is the surface opposite to the second surface Wb. Alternatively, the first surface Wa and the second surface Wb are sometimes collectively referred to as the surfaces of the wafer W.

[0036] like Figure 1 As shown, the wafer processing system 1 has a structure that connects the loading / unloading station 2 and the processing station 3 into one unit. The loading / unloading station 2 can accommodate multiple wafers W in a cassette C, for example, for loading and unloading to and from external locations. The processing station 3 is equipped with various processing devices for performing the desired processing on the wafers W.

[0037] A cassette loading stage 10 for holding multiple, for example, three cassettes C, is provided at the loading / unloading station 2. Furthermore, a wafer transport device 20 is provided adjacent to the cassette loading stage 10 on the negative X-axis side. The wafer transport device 20 is configured to move freely along a transport path 21 extending along the Y-axis. The wafer transport device 20 has, for example, two transport arms 22 for holding and transporting wafers W. Each transport arm 22 is configured to move freely in the horizontal and vertical directions and about a horizontal axis and about a vertical axis. Furthermore, the structure of the transport arms 22 is not limited to this embodiment and any structure can be adopted. Moreover, the wafer transport device 20 is configured to transport wafers W to the cassettes C of the cassette loading stage 10 and the transport device 30 described later.

[0038] On the negative X-axis side of the wafer transfer device 20 at the transfer station 2, a transfer device 30 for transferring wafers W between the wafer transfer device 20 and the processing station 3 is provided adjacent to the wafer transfer device 20.

[0039] For example, processing station 3 is provided with three processing blocks G1 to G3. The first processing block G1, the second processing block G2 and the third processing block G3 are arranged in the order described from the positive X-axis direction side (the side of the transfer station 2) to the negative direction side.

[0040] The first processing block G1 is equipped with an etching device 40, a thickness measuring device 50, a flipping device 51, and a wafer transport device 60. The etching device 40, the thickness measuring device 50, and the flipping device 51 are stacked. However, the number and arrangement of the etching device 40, the thickness measuring device 50, and the flipping device 51 are not limited thereto.

[0041] The etching apparatus 40 etches the silicon (Si) on the first surface Wa or the second surface Wb after grinding by the grinding apparatus 90 (described later). Multiple etching apparatuses 40 may be provided to improve wafer processing productivity.

[0042] like Figure 2 As shown, the etching apparatus 40 includes a wafer holding section 41, a rotating mechanism 42, a nozzle 43, and a moving mechanism 44.

[0043] The wafer holding portion 41 holds the outer edge of the wafer W at multiple points, specifically three points in this embodiment. Furthermore, the structure of the wafer holding portion 41 is not limited to the illustrated example; for instance, the wafer holding portion 41 may also include a holding disk (chuck) (not shown) that holds the wafer W from below. The wafer holding portion 41 is configured to rotate about a vertical rotation center line 41a via a rotation mechanism 42, thereby enabling the wafer W held on the wafer holding portion 41 to rotate.

[0044] The nozzle 43 supplies etching solution E to either the first surface Wa or the second surface Wb of the wafer W held in the wafer holding portion 41. The nozzle 43 is connected to an etching solution supply source (not shown) that supplies the etching solution E to the nozzle 43. The nozzle 43 is positioned above the wafer holding portion 41 and is configured to move horizontally and vertically via a moving mechanism 44. In one example, the nozzle 43 is configured to be able to pass through the rotation center line 41a of the wafer holding portion 41, i.e., ... Figure 3 As shown, it can move back and forth (scanning movement) or rotate above the center of wafer W.

[0045] The etching solution E contains hydrofluoric acid (HF), nitric acid (HNO3), and phosphoric acid (H3PO4) to appropriately etch the silicon of the wafer W, which can be the object of etching. In one example, the etching solution E is a mixture containing hydrofluoric acid, nitric acid, phosphoric acid, and water. Furthermore, the object to be etched can also be, for example, amorphous silicon.

[0046] In one case, Figure 1The thickness measuring device 50 shown includes a measuring unit (not shown) and a calculation unit (not shown). The measuring unit includes a sensor that measures the thickness of the etched wafer W at multiple points. The calculation unit obtains the thickness distribution of the wafer W based on the measurement results (thickness of the wafer W) from the measuring unit, and further calculates the flatness (TTV: Total Thickness Variation) of the wafer W. Alternatively, the calculation of the thickness distribution and flatness of the wafer W can be performed by the control device 130 described later, instead of the calculation unit. In other words, the calculation unit (not shown) can be provided within the control device 130 described later. Furthermore, the structure of the thickness measuring device 50 is not limited to this and can be configured arbitrarily.

[0047] The flipping device 51 flips the first surface Wa and the second surface Wb of the wafer W in the vertical direction. The structure of the flipping device 51 is arbitrary.

[0048] The wafer transport device 60 is disposed on the negative X-axis side of the transport device 30. The wafer transport device 60 has, for example, two transport arms 61 for holding and transporting the wafer W. Each transport arm 61 is configured to move freely in the horizontal and vertical directions and about the horizontal and vertical axes. Moreover, the wafer transport device 60 is configured to transport the wafer W to the transport device 30, the etching device 40, the thickness measuring device 50, the flipping device 51, the cleaning device 70 (described later), the thickness measuring device 71 (described later), the buffer device 72 (described later), and the flipping device 73 (described later).

[0049] The second processing block G2 is equipped with a cleaning device 70, a thickness measuring device 71, a buffer device 72, a flipping device 73, and a wafer transport device 80. The cleaning device 70, thickness measuring device 71, buffer device 72, and flipping device 73 are stacked. However, the number and arrangement of the cleaning device 70, thickness measuring device 71, buffer device 72, and flipping device 73 are not limited thereto.

[0050] The cleaning device 70 cleans at least the first surface Wa or the second surface Wb of the grinding device 90 after grinding.

[0051] In one example, the thickness measuring device 71 has the same structure as the thickness measuring device 50 described above. Furthermore, the structure of the thickness measuring device 71 is not limited to this and can be configured arbitrarily.

[0052] The buffer device 72 temporarily holds the unprocessed wafer W to be transferred from the first processing block G1 to the second processing block G2. The structure of the buffer device 72 is arbitrary. In addition, the buffer device 72 may also have an alignment mechanism (not shown) for adjusting the center position of the wafer W relative to the holding disks 93a, 93b described later and / or the orientation of the wafer W in the horizontal direction.

[0053] The flipping device 73 flips the first surface Wa and the second surface Wb of the wafer W in the vertical direction. The structure of the flipping device 73 is arbitrary.

[0054] The wafer transport device 80 is, for example, disposed on the positive Y-axis side of the cleaning device 70, the thickness measuring device 71, the buffer device 72, and the flipping device 73. The wafer transport device 80 has, for example, two transport arms 81 that use an adsorption and holding surface (not shown) to adsorb and hold the wafer W and transport it. Each transport arm 81 is supported by a multi-jointed arm member 82, configured to move freely in the horizontal and vertical directions and about the horizontal and vertical axes. Furthermore, the wafer transport device 80 is configured to transport the wafer W to the etching device 40, the thickness measuring device 50, the flipping device 51, the cleaning device 70, the thickness measuring device 71, the buffer device 72, the flipping device 73, and the grinding device 90 (described later).

[0055] A grinding device 90 is provided in the third processing block G3. The grinding device 90 grinds the first surface Wa or the second surface Wb of the wafer W to make it planar.

[0056] The grinding apparatus 90 includes a rotary table 91. The rotary table 91 is configured to rotate freely around a vertical rotation centerline 92 via a rotation mechanism (not shown). Four holding disks 93a and 93b are provided on the rotary table 91 for holding and holding the wafer W. For example, porous holding disks are used for the holding disks 93a and 93b. The surfaces of the holding disks 93a and 93b, i.e., the holding surfaces of the wafer W, have a convex shape where the central portion protrudes more than the outer periphery when viewed from the side. Furthermore, this central protrusion is minute, but in the following description, for clarity, the central protrusion of the holding disks 93a and 93b will sometimes be shown in an enlarged manner.

[0057] Two of the four retaining discs 93a and 93b are first retaining discs 93a used for grinding at the first machining position B1 (described later). These two first retaining discs 93a are arranged symmetrically across the rotation center line 92. The remaining two second retaining discs 93b are used for grinding at the second machining position B2 (described later). These two second retaining discs 93b are also arranged symmetrically across the rotation center line 92. That is, the first retaining discs 93a and the second retaining discs 93b are arranged alternately in the circumferential direction.

[0058] like Figure 4As shown, four retaining discs 93a and 93b are respectively held in four retaining disc bases 94. A tilt adjustment mechanism 95 is provided in the retaining disc base 94 for adjusting the relative tilt of the grinding sections 101 and 111 (described later) with the retaining discs 93a and 93b. The tilt adjustment mechanism 95 has a fixed shaft 96 provided on the lower surface of the retaining disc base 94, and multiple lifting shafts 97, such as two. Each lifting shaft 97 is configured to extend and retract freely, allowing the retaining disc base 94 to rise and fall. Through this tilt adjustment mechanism 95, the other end of the outer periphery of the retaining disc base 94 is raised and lowered in the vertical direction with one end (the position corresponding to the fixed shaft 96) as a base point, thereby tilting the retaining discs 93a and 93b and the retaining disc base 94. Furthermore, this allows adjustment of the relative tilt of the grinding surfaces of the grinding sections 101 and 111 at the machining positions B1 to B2 (described later) with the upper surfaces of the retaining discs 93a and 93b. Furthermore, the structure of the tilt adjustment mechanism 95 is not limited to this, as long as it can adjust the relative angle (parallelism) of the surfaces (holding surfaces) of the holding discs 93a and 93b relative to the grinding surfaces of each grinding section 101 and 111.

[0059] like Figure 1 As shown, by rotating the rotary table 91, the four holding disks 93a and 93b can move to the junction positions A1~A2 and the processing positions B1~B2. In addition, the four holding disks 93a and 93b are each configured to rotate about a vertical axis via a rotating mechanism (not shown).

[0060] The first handover position A1 is located on the positive X-axis and positive Y-axis side relative to the rotation center line 92 of the rotary table 91, and is used for handing over the wafer W to the first holding disk 93a during grinding of the first surface Wa. The second handover position A2 is located on the positive X-axis and negative Y-axis side relative to the rotation center line 92 of the rotary table 91, and is used for handing over the wafer W to the second holding disk 93b during grinding of the second surface Wb.

[0061] The first processing position B1 is located on the negative X-axis and negative Y-axis side relative to the rotation center line 92 of the rotary table 91, and a first grinding unit 100 is arranged at the first processing position B1. As an example, the first grinding unit 100 grinds the first surface Wa or the second surface Wb of the wafer W held on the first holding disk 93a.

[0062] like Figure 4As shown, the first grinding unit 100 has a grinding section 101. The grinding section 101 includes a grinding stone 102, a grinding wheel 103, a mounting base 104, a spindle 105, and a drive unit 106. The grinding wheel 103 has an annular shape, and the grinding stone 102 is supported on its lower surface. The mounting base 104 supports the grinding wheel 103. The spindle 105 rotates the grinding wheel 103 and the grinding stone 102 via the mounting base 104. The drive unit 106 is mounted on the spindle 105 and, for example, incorporates a motor (not shown) to rotate the spindle 105. Additionally, as... Figure 1 As shown, the grinding unit 101 is configured to move in the vertical direction along the support column 107 via the drive unit (not shown).

[0063] The second processing position B2 is located on the negative X-axis and positive Y-axis side relative to the rotation center line 92 of the rotary table 91. A second grinding unit 110 is disposed at the second processing position B2. As an example, the second grinding unit 110 grinds the second surface Wb or the first surface Wa of the wafer W held on the second holding disk 93b. The second grinding unit 110 has the same structure as the first grinding unit 100. That is, as... Figure 1 and Figure 4 As shown, the second grinding unit 110 has a grinding section 111 (grinding stone 112, grinding wheel 113, mounting base 114, spindle 115 and drive section 116) and a support column 117.

[0064] Here, as described above, the holding surfaces of the holding disks 93a and 93b have a convex shape. Therefore, during the grinding process of the wafer W using the grinding units 100 and 110, the wafer W contacts the annularly arranged grinding wheels 102 and 112 in an arc shape from the center to the outer periphery. By rotating the holding disks 93a and 93b and the grinding wheels 103 and 113 respectively in this state, the entire surface of the wafer W is ground.

[0065] Furthermore, in the grinding units 100 and 110, by adjusting the relative angle (tilt) between the holding surfaces of the holding disks 93a and 93b and the grinding surfaces of the grinding tools 102 and 112 using the aforementioned tilt adjustment mechanism 95, the shape of the ground wafer W can be controlled in any shape selected from the following types: planar, convex, concave (V-shaped), W-shaped, and M-shaped, or any combination of two of these shapes. A planar shape is one where the entire surface of the wafer W is adjusted to a desired flatness (TTV) or less, and preferably, the thickness of the entire surface of the wafer W is controlled to be uniform. A convex shape is one where the thickness at the center of the wafer W is greater than the thickness at the outer periphery. A concave shape is one where the thickness at the concave portion of the wafer W is less than the thickness at the outer periphery. A W-shaped shape is one where the thickness at the radius center point is less than the thickness at the center and outer periphery of the wafer W. An M-shaped shape is one where the thickness at the radius center point is greater than the thickness at the center and outer periphery of the wafer W.

[0066] In addition, a thickness measuring device (not shown) for measuring the thickness of the ground wafer W can also be installed at the handover positions A1, A2 or the processing positions B1, B2.

[0067] A display panel 120 is provided for the wafer processing system 1 described above. The display panel 120 may be, for example, a monitor or a touch panel, and may be directly mounted on the wafer processing system 1, or it may be a display panel that can be viewed remotely. The display panel 120 displays screens showing the operations performed by the wafer processing system 1 for each process. The operation results displayed on the display panel 120 are output to the control device 130, which will be described later.

[0068] At least one control device 130 is provided for the wafer processing system 1 described above. The control device 130 processes computer-executable instructions that cause the wafer processing system 1 to execute the various processes described herein. The control device 130 can be configured to control the various elements of the wafer processing system 1 to execute the various processes described herein. In one embodiment, part or all of the control device 130 may be included in the wafer processing system 1. The control device 130 may include a processing unit, a storage unit, and a communication interface. The control device 130 is implemented, for example, by a computer. The processing unit can be configured to perform various control actions by reading a program from the storage unit that provides logic or routines enabling various control actions and executing the read program. The program may be stored in the storage unit in advance or retrieved via a medium when needed. The retrieved program is stored in the storage unit and read from and executed by the processing unit. The medium may be various computer-readable storage media or a communication line connected to the communication interface. The storage medium may be transient or non-transient. The processing unit can be a CPU (Central Processing Unit) or one or more circuits. The storage unit can include RAM (Random Access Memory), ROM (Read Only Memory), HDD (Hard Disk Drive), SSD (Solid State Drive), or a combination thereof. The communication interface can communicate with the wafer processing system 1 via a communication line such as a LAN (Local Area Network).

[0069] Next, the wafer processing performed using the wafer processing system 1 configured as described above will be explained. In this embodiment, the wafer W cut from an ingot by a wire saw or the ground wafer W is processed to improve the in-plane uniformity of thickness.

[0070] First, before processing wafer W in wafer processing system 1, the control device 130 determines the target thickness distribution (thickness profile) in the radial direction (within the wafer surface) when grinding the first surface Wa of wafer W (described later). Hereinafter, this target thickness distribution during grinding will be referred to as the "grinding target thickness distribution". Furthermore, the control device 130 also determines the target thickness distribution in the radial direction when grinding the second surface Wb of wafer W (described later). Figure 5 (St1). Furthermore, the method for determining the grinding target thickness distribution of the first surface Wa and the second surface Wb in the control device 130 will be described later.

[0071] Once the target grinding thickness distribution is determined as a pretreatment step, wafer processing in wafer processing system 1 is then performed. In this wafer processing, firstly, a cassette C containing multiple wafers W is placed on the cassette stage 10 of the transfer station 2. In cassette C, the wafers W are stored with their first surface Wa facing upwards and their second surface Wb facing downwards. Next, the wafers W in cassette C are removed using wafer transport device 20 and transferred to transport device 30. The wafers W transferred to transport device 30 are then transferred to buffer device 72 using wafer transport device 60.

[0072] Next, the wafer W is transferred to the grinding apparatus 90 using the wafer transfer device 80 and handed over to the first holding disk 93a at the first handover position A1. The second surface Wb of the wafer W is held and held by the first holding disk 93a.

[0073] Next, the rotary table 91 is rotated to move the wafer W to the first processing position B1. Then, the first surface Wa is ground by the first grinding unit 100 based on the grinding target thickness distribution of the first surface Wa determined in St1. Figure 5 (St2).

[0074] Next, the rotary table 91 is rotated to move the wafer W to the first junction position A1.

[0075] Next, the wafer W is transferred to the cleaning apparatus 70 using the wafer transfer device 80. In the cleaning apparatus 70, the first surface Wa of the wafer W is cleaned. Figure 5 St3). In St3, the second surface Wb of wafer W can also be cleaned.

[0076] Next, the wafer W is transferred to the flipping device 73 using the wafer transfer device 80. In the flipping device 73, the first surface Wa and the second surface Wb of the wafer W are flipped in the vertical direction. Figure 5 (St4). That is, the wafer W is flipped so that the first surface Wa faces downward and the second surface Wb faces upward.

[0077] Next, the wafer W is transferred to the grinding apparatus 90 using the wafer transfer device 80 and handed over to the second holding disk 93b at the second handover position A2. The first surface Wa of the wafer W is held by adsorption using the second holding disk 93b.

[0078] Next, the rotary table 91 is rotated to move the wafer W to the second processing position B2. Then, the second surface Wb of the wafer W is ground by the second grinding unit 110 based on the grinding target thickness distribution of the second surface Wb determined in St1. Figure 5 St5).

[0079] Next, the rotary table 91 is rotated to move the wafer W to the second junction position A2.

[0080] Next, the wafer W is transferred to the cleaning apparatus 70 using the wafer transfer device 80. In the cleaning apparatus 70, the second surface Wb of the wafer W is cleaned. Figure 5 St6). In St6, the first surface Wa of wafer W can also be cleaned.

[0081] Next, the wafer W is transported to the thickness measuring device 71 using either the wafer transport device 80 or the wafer transport device 60. In the thickness measuring device 71, the thickness distribution of the wafer W is obtained by measuring the thickness of the wafer W after grinding the second surface Wb at multiple points, and the flatness of the wafer W is calculated. Figure 5 (St7). The calculated thickness distribution and flatness of the wafer W are output to the control device 130, for example. In addition, if the grinding device 90 is equipped with a thickness measuring device, the thickness of the ground wafer W can also be measured by the thickness measuring device of the grinding device 90.

[0082] In the control device 130, based on the thickness distribution and flatness of the wafer W calculated in St7 and output to the control device 130, the optimal etching conditions for the second surface Wb are determined to optimize the etching amount distribution (etching profile) in the etching process of the second surface Wb. Figure 5 (St8). Furthermore, the etching amount is the amount removed from wafer W by etching, and the etching amount distribution is the distribution of the etching amount in the radial direction (within the wafer plane) of wafer W. The method for determining the optimal etching conditions for the second surface Wb in the control device 130 is described later.

[0083] Next, the wafer W is transported to the etching apparatus 40 using the wafer transport device 60. In the etching apparatus 40, under the optimal etching conditions determined by St8, the second surface Wb of the wafer W is etched using the etching solution E. Figure 5 (St9). In St9, the second surface Wb is etched into the target shape by etching under optimal etching conditions.

[0084] Next, the wafer W is transferred to the flipping device 51 using the wafer transfer device 60. In the flipping device 51, the first surface Wa and the second surface Wb of the wafer W are flipped in the vertical direction. Figure 5 (St10). That is, the wafer W is flipped so that the first surface Wa faces upward and the second surface Wb faces downward.

[0085] Next, the wafer W is transported to the thickness measuring device 50 using the wafer transport device 60. In the thickness measuring device 50, the thickness distribution of the wafer W is obtained by measuring the thickness of the wafer W after etching the second surface Wb at multiple points, and the flatness of the wafer W is calculated. Figure 5 (St11). The calculated thickness distribution and flatness of the wafer W are output to the control device 130, for example.

[0086] In the control device 130, the optimal etching conditions for the first surface Wa are determined based on the thickness distribution and flatness of the wafer W calculated in St11 and output to the control device 130, which optimizes the etching amount distribution in the etching process of the first surface Wa. Figure 5 (St12). The method for determining the optimal etching conditions for the first surface Wa in the control device 130 is described later.

[0087] Next, the wafer W is transported to the etching apparatus 40 using the wafer transport device 60. In the etching apparatus 40, the first surface Wa of the wafer W is etched using the etching solution E under the optimal etching conditions determined by St12. Figure 5 (St13). In St13, the etching amount distribution is optimized by etching the first surface Wa under optimal etching conditions, and the first surface Wa is processed into the target shape.

[0088] Next, the wafer W is transported to the thickness measuring device 50 using the wafer transport device 60. In the thickness measuring device 50, the thickness distribution of the wafer W is obtained by measuring the thickness of the wafer W after etching the first surface Wa and the second surface Wb at multiple points. Figure 5 (St14). In the thickness measuring device 50, the flatness of the wafer W can also be calculated. The obtained thickness distribution of the wafer W is output to the control device 130, for example, and used in the subsequent processing of other wafers W by the wafer processing system 1.

[0089] Afterwards, the wafer W, which has undergone all processing, is transferred via the transfer device 30 to the cassette C of the cassette stage 1. In this way, a series of wafer processing steps in the wafer processing system 1 are completed.

[0090] Furthermore, in the above embodiments, after grinding the first surface Wa by St2, grinding the second surface Wb by St5 is performed, but the grinding order of these surfaces can also be reversed. Additionally, after etching the second surface Wb by St9, etching the first surface Wa by St13 is performed, but the etching order of these surfaces can also be reversed.

[0091] Furthermore, in the above embodiments, the etching of the second surface Wb in St9 is performed under the optimal etching conditions determined by St8, but alternatively, it can be performed under predetermined etching conditions. In this case, St7 and St8 of this embodiment are omitted. Similarly, the etching of the first surface Wa in St13 is performed under the optimal etching conditions determined by St12, but alternatively, it can be performed under predetermined etching conditions. In this case, St11 and St12 of this embodiment are omitted.

[0092] Next, the method for determining the above-mentioned optimal etching conditions ( Figure 5 The following description will explain the optimal etching conditions for the first surface Wa in St12, but the same method will be used to determine the optimal etching conditions for the second surface Wb in St8.

[0093] First, before processing wafer W in wafer processing system 1, multiple training data are acquired ( Figure 6 (St100). Furthermore, the learning data is the etching amount distribution of wafer W for a given etching condition. Additionally, multiple learning data points are used, as described later, to determine the optimal etching conditions, and, as described later, to determine the grinding target thickness distribution.

[0094] In St100, for example, wafer dummy etches are performed under multiple different etching conditions. Specifically, for example, the dummy wafer rotation speed R (also known as rotational speed) during etching, the scanning speed V of nozzle 43 (also known as oscillation speed), and the scanning width L of nozzle 43 (refer to...) are varied. Figure 3 The scanning width L, also known as the oscillation radius, or the number of cycles N of nozzle 43, is used to etch the dummy wafers. In this case, the etching time for each dummy wafer is the same. Similar to the etching in St13, the dummy wafer is rotated and etchant E is supplied to the dummy wafer from nozzle 43 while the nozzle 43 reciprocates. In the following description, the reciprocating movement of nozzle 43 between the two ends of the dummy wafer is considered as one cycle.

[0095] The wafer dummy is etched under each etch condition for a predetermined desired time (desired number of cycles). Then, the etch amount distribution of the wafer dummy is acquired and output to the control device 130. Furthermore, in the control device 130, the output etch amount distribution under each etch condition is compressed into an etch amount distribution per unit time (unit number of cycles), and the compressed etch amount distributions are stored as the aforementioned learning data.

[0096] When acquiring learning data in St100, such as Figure 7 As shown, a learning data acquisition screen 200 is displayed on the display panel 120. The learning data acquisition screen 200 displays a list 201 of etching conditions used when acquiring learning data. The list 201 of etching conditions displays, for example, the ID of the learning data, rotation speed R, scan speed V, scan width L, and number of cycles N.

[0097] Additionally, an add button 202 for adding learning data is displayed on the learning data acquisition screen 200. When the user presses the add button 202, an etching condition input screen (not shown) is displayed on the display panel 120. In the etching condition input screen, multiple preset values ​​are displayed for conditions such as rotational speed R, scan speed V, scan width L, and number of cycles N, allowing the user to select the value for each condition. Alternatively, the etching condition input screen can be configured to allow the user to directly input, for example, rotational speed R, scan speed V, scan width L, and number of cycles N. The etching conditions input to the etching condition input screen are displayed in a list of etching conditions 201 on the learning data acquisition screen 200.

[0098] Additionally, a delete button 203 for deleting learning data is displayed on the learning data acquisition screen 200. When the user presses the delete button 203, a learning data deletion screen (not shown) for the deleted object is displayed, for example. The learning data deletion screen displays, for example, multiple learning data IDs, and the user can select the learning data ID to delete. Furthermore, when the user selects a learning data ID, the etching condition for that learning data ID is deleted from the etching condition overview table 201 on the learning data acquisition screen 200.

[0099] also, Figure 7 The learning data acquisition screen shown in Figure 200 is an example; other items can also be displayed.

[0100] Furthermore, while the example described above illustrates obtaining the learning data by etching a dummy wafer, the etching target for obtaining the learning data is not limited to a dummy wafer. Specifically, for example, the etching result of a product processed by the wafer processing system 1 using wafer W can also be stored as the learning data. Additionally, if a film is formed on the first surface Wa of wafer W, the etching target can be the film, and the etching result of that film can be stored as the learning data.

[0101] Furthermore, while the acquisition of the aforementioned learning data is performed within the wafer processing system 1, it can also be performed outside of the wafer processing system 1. In this case, the control device 130 determines the optimal etching conditions based on multiple learning data acquired outside the wafer processing system 1.

[0102] Next, based on the thickness distribution in the target shape of the etched wafer W and the thickness distribution in the surface shape of the etched wafer W obtained in St11 (hereinafter referred to as the "measured shape"), the target etching amount distribution in the etching process of St13 is obtained. Figure 6 (St110). As an example, the target etching amount distribution of the etching process can be obtained by calculating the difference between the thickness distribution of the target shape of the wafer W and the thickness distribution of the measured shape.

[0103] Next, an optimization method is used to optimize the learning data used for superposition and the number of times the learning data is superimposed to form the target etching distribution obtained in St110, so that multiple learning data (etching amount distributions) are superimposed. Figure 6 (St111).

[0104] In St111, for example, the control of the etching amount distribution is applied to the knapsack problem, optimizing the number of times the learning data and the learning data are superimposed. For example, the etching amount distribution is the knapsack of the knapsack problem, and the learning data is the items of the knapsack problem. Moreover, the number of times the learning data and the learning data are superimposed is optimized in a way that minimizes the difference between the superimposed etching amount distribution and the target etching amount distribution in St110. In other words, the etching amount distribution during the etching of the first surface Wa in St13 is optimized.

[0105] Next, the etching conditions corresponding to the most optimized learning data in St111 are integrated to determine the optimal etching conditions. Figure 6 (St112). Specifically, the optimal etching condition is determined by integrating multiple etching conditions in a way that optimizes the number of superpositions of the selected etching conditions. In other words, the optimal etching condition that optimizes the etching amount distribution is determined.

[0106] The optimal etching conditions for the first surface Wa in St12 are determined as described above. In this case, by etching the first surface Wa of wafer W under the optimal etching conditions in St13, the etching amount distribution can be optimized, and the first surface Wa can be processed into the target shape.

[0107] Furthermore, the optimal etching conditions for the first surface Wa, determined in St12, are stored in the control device 130, which retains a historical record of etching conditions. Learning data used to determine the optimal etching conditions is also stored at this time. Similarly, the optimal etching conditions (and learning data) for the second surface Wb, determined in St8, are also stored in the control device 130, which retains a historical record of etching conditions. Additionally, the optimal etching conditions for these surfaces Wa and Wb can also be displayed on the etching condition history screen (not shown) of the display panel 120.

[0108] Furthermore, as mentioned above, the etching of the second surface Wb in St9 and the etching of the first surface Wa in St13 are sometimes performed under predetermined etching conditions, rather than under optimal etching conditions. Therefore, when setting the etching conditions, it is also possible to allow the user to choose whether to use learning data.

[0109] In this case, for example, Figure 8 As shown, a learning data selection screen 210 is displayed on the display panel 120. The learning data selection screen 210 displays a learning button 211 and a calibration button 212. When the user presses the learning button 211, the user selects to use the learning data when etching the surfaces Wa and Wb of wafer W. Conversely, when the user presses the calibration button 212, the etching conditions are calibrated to determine the optimal etching conditions when etching the surfaces Wa and Wb of wafer W. Furthermore, it is determined whether the etching of the surfaces Wa and Wb of wafer W is performed under the optimal etching conditions or under predetermined etching conditions. Additionally, Figure 8 The learning data selection screen 210 shown is one example; other items can also be displayed.

[0110] Next, the method for determining the thickness distribution of the grinding target mentioned above ( Figure 5 The following description will explain the method for determining the grinding target thickness distribution of the first surface Wa, but the method for determining the grinding target thickness distribution of the second surface Wb is the same.

[0111] First, before processing wafer W in wafer processing system 1, multiple training data are acquired ( Figure 9 (St100). These multiple learning data are compared with... Figure 6 The multiple learning data obtained in St100 shown are the same.

[0112] Figure 10 This example illustrates multiple sets of learning data. Figure 10 The horizontal axis represents the radial position from the center of the wafer (0 (zero)) to the outer periphery, and the vertical axis represents the amount of etching. Figure 10 The examples of multiple learning data shown represent a case where a total of 30 types of learning data are stored in the control device 130; in other words, it is an example of obtaining an etching amount distribution under a total of 30 different etching conditions. Furthermore, in this example, the wafer W is, for example, a wafer with a diameter of 300 mm, but it is not limited to this. Additionally, in this example, the etching amount values ​​are illustrative and not limited to this.

[0113] Next, select several effective learning data from multiple learning datasets to determine the thickness distribution of the grinding target. Figure 9(St120). In St120, the method for selecting valid learning data is arbitrary; for example, statistical methods can be used to select multiple valid learning data.

[0114] In St120, for example, when using standard deviation as a statistical method, it is possible to exclude learning data with etching amounts that deviate from the average ±2σ of multiple learning data (etching amount distribution). Figure 11 This illustrates an example of learning data that was excluded from the valid learning data in a dataset. Figure 11 The learning data shown contains deviations from the mean by ±2σ. Figure 11 The etch amount (the dashed line) is excluded from the learning data. Furthermore, the range of valid learning data is not limited to the average value ±2σ of the etch amount distribution; for example, it could be the average value ±σ, the average value ±3σ, etc.

[0115] Additionally, in St120, for example, when using a 95% confidence ellipse as a statistical method, training data that deviates from the 95% confidence ellipse can be excluded from multiple training datasets.

[0116] In St120, multiple valid learning data are selected as described above. In actual etching processes, it is difficult to estimate the amount of etching that varies drastically in the radial direction of wafer W, so learning data that contains etching amounts that deviate from the average ±2σ or deviate from the 95% confidence ellipse are excluded.

[0117] Furthermore, if all the learning data acquired in St100 are effective in determining the thickness distribution of the grinding target, then all of these learning data should be used as effective learning data. In this case, St120 can be omitted.

[0118] Next, the selected valid learning data (etching amount distribution) in St120 will be averaged to obtain the averaged etching amount distribution. Figure 9 (St121). In St121, the etch amount from multiple valid learning data is averaged over each radial position of the wafer W to calculate the average etch amount. Figure 12 An example of averaged etching amount distribution is shown.

[0119] By obtaining an averaged etching amount distribution in St121, it is possible to increase the control margin of the etching amount distribution when etching the first surface Wa in the subsequent St13. The reasons for this ability to increase the control margin of the etching amount distribution are explained later.

[0120] Next, the averaged etch amount distribution obtained in St121 is converted to a reference etch amount distribution by adjusting for a reference etch amount. Figure 9(St122). In St122, the reference etch amount B is calculated based on the average etch amount A using the following formula (1). Furthermore, the reference etch amount α can be arbitrarily set according to the specifications.

[0121]

[0122] Where B is the baseline etching amount, and A is the average etching amount. AVG α is the radial average of the averaged etching amount, and α is the baseline etching amount.

[0123] Figure 13 An example of a benchmark etching amount distribution is shown. Figure 13 In the example, the reference etching amount α is 1.5 μm. In St122, the average etching amount distribution is made consistent with the reference etching amount α, thus enabling the acquisition of a reference etching amount distribution that corresponds to the specifications.

[0124] Next, the target etching amount distribution is obtained by converting the baseline etching amount distribution obtained in St122 to converge to the desired interval (range). Figure 9 (St123). In St123, the target etching amount C is calculated based on the reference etching amount B using the following formula (2). Furthermore, the desired range Z is the (maximum value - minimum value) of the reference etching amount B, which can be arbitrarily set according to the specifications. For example, when grinding the first surface Wa by St2, there is a target flatness of the ground first surface Wa, and the desired range Z is set to this target flatness.

[0125]

[0126] Where C is the target etching amount, B is the baseline etching amount, and B0 is the standard etching amount. AVG B is the radial average value of the baseline etching amount. RGE For (BB) AVG The scope of (BB) AVG (Maximum value - Minimum value) where Z is the desired range and α is the baseline etching amount.

[0127] Figure 14 This illustrates an example of the target etching amount distribution. Figure 14 In the example, the desired interval Z is 0.5 μm, and the baseline etching amount α is 1.5 μm as described above. In St123, the baseline etching amount distribution is converged to the desired interval Z, thus enabling the acquisition of an appropriate target etching amount distribution. For example, when the desired interval Z is the target flatness during grinding, an appropriate etching amount distribution can be obtained, resulting in the ability to determine an appropriate grinding target thickness distribution.

[0128] Here, the target etching amount distribution obtained in St123 is an ideal etching amount distribution, and when etching is performed with this target etching amount distribution, the surface shape of the wafer W can be appropriately controlled. Moreover, if the target etching amount distribution obtained in St123 can be achieved, St124 to St126 described later are omitted, and we proceed to St127 described later, where the grinding target thickness distribution is determined based on the target etching amount distribution.

[0129] On the other hand, the target etching amount distribution obtained in St123 becomes an etching amount distribution that uses all effective learning data. In St120, for example, some learning data are excluded from 30 learning data points, such as selecting more than 20 effective learning data points, and in St123, more than 20 effective learning data points are used to obtain the target etching amount distribution. However, the number of learning data points used in St12 to determine the optimal etching conditions is mostly less than 15, sometimes less than the number of effective learning data points used in St123. In this case, it is sometimes difficult to achieve the target etching amount distribution obtained in St123.

[0130] Therefore, the number of combinations of effective learning data can be reduced to obtain an etching amount distribution that is close to the target etching amount distribution obtained in St123. For example, an optimization method can be used to superimpose multiple effective learning data (etching amount distributions) to form the target etching amount distribution obtained in St123. The combination of the effective learning data used for superposition and the number of times the effective learning data is superimposed is optimized to obtain a modified target etching amount distribution. In this embodiment, the constraints used in the optimization calculation are changed to perform the first optimization calculation ( Figure 9 St124) and the second optimization calculation ( Figure 9 St125).

[0131] In both the first optimization calculation in St124 and the second optimization calculation in St125, control over, for example, the etching amount distribution is applied to the knapsack problem to optimize the number of times the effective learning data and the effective learning data are superimposed. For example, the etching amount distribution is the knapsack in the knapsack problem, and the effective learning data is the items in the knapsack problem. Moreover, the number of times the effective learning data and the effective learning data are superimposed is optimized in a way that minimizes the difference between the superimposed etching amount distribution and the target etching amount distribution in St123.

[0132] Here, if the number of valid learning data selected in St120 is, for example, 20, then when averaging the valid learning data in St121, the 20 valid learning data are added together equally. That is, 1 / 20 (=0.5) of the etching amount distribution of each valid learning data is added together. Thus, the average etching amount is a decimal, which is difficult to achieve.

[0133] Regarding this, both the first optimization calculation in St124 and the second optimization calculation in St125 optimize the number of stacks of the effective learning data. In this case, the optimized number of stacks is an integer, thus enabling the optimization of the achievable etching amount distribution. In other words, by solving the knapsack problem, the weights of the averaged etching amount, which are continuous values, can be transformed into the number of stacks, which are discrete values.

[0134] The constraint used in the first optimization calculation in St124 is the amount of effective learning data. This amount of effective learning data is an upper limit on the amount of effective learning data used in the first optimization calculation, determined, as described above, in a manner consistent with the amount of learning data used to determine the optimal etching conditions in St12. Furthermore, the first optimization calculation is performed to obtain the first corrected target etching amount distribution.

[0135] Figure 15 The first corrected target etching amount distribution is shown. Figure 15 An example of a dashed line (in Chinese). Figure 15 In the example, the number of valid learning data points is 5, which serves as a constraint. In St124, the number of valid learning data points is used as a constraint to perform the first optimization calculation, thus enabling the acquisition of an achievable first corrected target etching amount distribution.

[0136] The constraints used in the second optimization calculation in St125 are the number of valid data points and the grinding conditions when grinding the first surface Wa in St2. The number of valid data points is the same as the constraint used in the first optimization calculation described above. That is, the grinding conditions were not used as constraints in the first optimization calculation, but they are added to the constraints in the second optimization calculation. Furthermore, the second optimization calculation is performed to obtain the second corrected target etching amount distribution.

[0137] Figure 15 The second corrected target etching amount distribution is shown. Figure 15 An example of a solid line (in the text). Figure 15 In the example, the number of valid learning data points used as constraints is 5. Furthermore, the grinding conditions used as constraints are, for example, (1) the shape of the first surface Wa of the ground wafer W and (2) the target flatness of the first surface Wa of the ground wafer W. In this embodiment, both (1) and (2) are used as grinding conditions, but either one may be used. Alternatively, other grinding conditions besides (1) and (2), such as a desired grinding model for the first surface Wa, may also be used.

[0138] In the grinding conditions described in (1) above, the shape of the first surface Wa of the ground wafer W is as shown in the side view. Figure 15As shown, for example, at a position 75mm from the center of the wafer ( Figure 15 The broken line with two straight lines (with the black dot in the image) as vertices was approximated in shape. That is, the broken line was made to have... Figure 15 The range shown is 55mm to 95mm from the center of wafer W. Figure 15 The rounded corners are defined within the area indicated by the dashed lines in the diagram. Specifically, the difference in slope between the two straight lines is divided by the number of measurement points to gradually decrease the slope from one slope to the other at the same rate. In this case, the curvature of the broken line at a position 75 mm from the center of the wafer can also be set below a predetermined threshold.

[0139] Here, when grinding the first surface Wa in St2, as described above, the shape of the ground wafer W can be controlled in the first grinding unit 100 with any shape among flat, convex, concave, W-shaped, and M-shaped, or any combination of two of these shapes. The inflection point of the convex, concave, W-shaped, and M-shaped shapes is 75 mm from the center of the wafer W, which is easy to control.

[0140] In addition, the range of 55mm to 95mm from the center of wafer W where the bend is made is one example, and it can be set arbitrarily.

[0141] Under the grinding conditions described in (2) above, the target flatness of the first surface Wa of the ground wafer W is the same as the desired range Z used in St123. In this embodiment, the target flatness is 0.5 μm.

[0142] In St125, grinding conditions are used as constraints for the second optimization calculation in addition to the amount of effective learning data, so that the second corrected target etching amount distribution achievable by the first grinding unit 100 can be obtained.

[0143] Here, the first corrected target etching amount distribution obtained in St124 is sometimes not as... Figure 15 As shown, it varies smoothly in the radial direction. In this case, as described later, when grinding the first surface Wa in St2, it is sometimes impossible to achieve the grinding target thickness distribution determined by the first modified target etching amount distribution. Therefore, the first modified target etching amount distribution can also be smoothed to obtain a smoothed target etching amount distribution. Figure 9 (St126). The smoothing method is arbitrary; for example, spline smoothing can be used.

[0144] Furthermore, if the first corrected target etching amount distribution obtained in St124 is smoothed to a degree that can be achieved by grinding in St2, the smoothing of the first corrected target etching amount distribution in St126 can be omitted. Additionally, in St125, the grinding conditions are used as constraints for the second optimization calculation, therefore the smoothing of the first corrected target etching amount distribution in St126 is not required.

[0145] Next, the target etching amount distribution is added to the target thickness after etching the first surface Wa in St13 to determine the grinding target thickness distribution. Figure 9 (St127). The target etch amount distribution is any one of the target etch amount distribution obtained in St123, the first modified target etch amount distribution obtained in St124, the second modified target etch amount distribution obtained in St125, and the smoothed target etch amount distribution obtained in St126.

[0146] Figure 16 An example of the thickness distribution of a grinding target is shown. Figure 16 In the example, the target thickness after etching the first surface Wa using St13 is 775 μm. Furthermore, in... Figure 16 The image shows the grinding target thickness distribution based on the target etching amount distribution of St123, the first modified target etching amount distribution of St124, and the second modified target etching amount distribution of St125.

[0147] According to the above embodiment, the grinding target thickness distribution determined in St1 (St120~St127) is the distribution obtained by adding the optimal etching amount distribution achievable when etching the first surface Wa to the target thickness. Furthermore, after grinding the first surface Wa based on the grinding target thickness distribution in St2, the first surface Wa is etched in St13 according to the optimal etching conditions determined in St12. Therefore, the flatness of the etched first surface Wa can be minimized, and the shape of the first surface Wa can be appropriately controlled. The reasons for achieving this effect will be explained below.

[0148] Here, as indicators of whether the grinding target thickness distribution is appropriate, the following are used: (1) the similarity between the grinding target thickness distribution and the target etching amount distribution, and (2) the control margin of the etching amount distribution during the etching of the first surface Wa.

[0149] Regarding the similarity in (1) above, when the grinding target thickness distribution and the target etching amount distribution are similar, the grinding target thickness distribution becomes a distribution that can be reproduced by optimizing the etching amount distribution of St12. Therefore, it is easy to control the etching amount distribution when etching the first surface Wa. As a result, it is easy to improve the flatness of the first surface Wa after etching to control the shape of the first surface Wa.

[0150] Regarding the control margin mentioned above (2), the control margin is widened when the etching amount distribution during the etching of the first surface Wa is easily controlled. As a result, even if the thickness distribution of the wafer W after grinding deviates, the deviation can be absorbed during the etching of the first surface Wa, thereby making it easier to control the shape of the first surface Wa.

[0151] Therefore, the state of high similarity in (1) and wide control margin in (2) is ideal, in which the shape of the first surface Wa can be appropriately controlled.

[0152] Regarding this, previous methods have included making the thickness distribution of the grinding target flat. However, due to the etching characteristics of the first surface Wa, the etching amount distribution tends to form a V-shape, making it difficult to make the etching amount distribution flat. Therefore, the similarity of (1) above is low, making it difficult to improve the flatness of the first surface Wa after etching. In addition, the etching amount distribution of the first surface Wa tends to form a V-shape, and conversely, when making the thickness distribution of the grinding target flat, it is difficult to control the etching amount distribution. Therefore, the control margin of (2) above is narrow, making it difficult to control the shape of the first surface Wa.

[0153] Furthermore, conventionally, the grinding target thickness distribution is set to a V-shape. In this case, the etching amount distribution of the first surface Wa is easily V-shaped, so the similarity of (1) above is higher than that when the grinding target thickness distribution is flat. In addition, when the grinding target thickness distribution is set to a V-shape, it is easier to perform the etching amount distribution of the first surface Wa, and the control margin of (2) above is also wider than that when the grinding target thickness distribution is flat. However, even when the grinding target thickness distribution is set to a V-shape, it is not optimized for the learning data that optimizes the etching amount distribution of the first surface Wa. Therefore, there is room for improvement in the conventional grinding target thickness distribution.

[0154] In contrast, in St1 of this embodiment, the learning data obtained in St100 is used, specifically the effective learning data selected in St120, so the grinding target thickness distribution determined in St127 can be optimized. Therefore, the flatness of the first surface Wa after etching can be improved, and the similarity between the grinding target thickness distribution and the target etching amount distribution in (1) above can be increased.

[0155] Here, when determining the optimal etching conditions in St12, the target etching amount distribution is optimized by centering on the average value of the learning data (etching amount distribution). In this respect, in St1 of this embodiment, the averaged etching amount distribution obtained by averaging the effective learning data in St127 is used to determine the grinding target thickness distribution. Moreover, the first surface Wa is etched when the thickness distribution of the ground wafer W is the grinding target thickness distribution, so the control margin of the etching amount distribution of the first surface Wa in (2) above can be widened to maximize it.

[0156] Figure 17 This illustrates the scenario of performing principal component analysis on multiple learning datasets obtained in St100. Figure 17 (a) is Figure 10 The example shows multiple training data points. When these multiple training data points are compressed to four dimensions, as shown... Figure 17 As shown in (b), the components are divided into first principal components to fourth principal components. Furthermore, when determining the optimal etching conditions in St12, the variations of these first to fourth principal components are summed to optimize the target etching amount distribution. At this time, when the first to fourth principal components are each 0 (zero), that is, the control margin of the average value of each of the first to fourth principal components is the widest. Regarding this, in this embodiment, the effective learning data is averaged in St121 to obtain an averaged etching amount distribution, thus maximizing the control margin described above (2).

[0157] As described above, according to this embodiment, the grinding target thickness distribution is determined by St120 to St127 in St1, thus improving the similarity of (1) and expanding the control margin of (2). Furthermore, after grinding the first surface Wa based on the grinding target thickness distribution in St2, when etching the first surface Wa in St13 according to the optimal etching conditions determined in St12, the flatness of the first surface Wa can be minimized. As a result, the shape of the first surface Wa can be appropriately controlled.

[0158] Alternatively, one could consider determining the shape of the etched wafer W and then determining the target grinding thickness distribution based on that shape. However, in this case, determining the target grinding thickness distribution requires etching the wafer W each time, which is time-consuming. In this embodiment, once the learning data (etching amount distribution) is obtained, the target grinding thickness distribution can be determined arbitrarily, and the shape of the first surface Wa can be appropriately controlled.

[0159] Furthermore, the above explanation illustrates the effect when the target grinding thickness distribution of the first surface Wa is determined in St1, but the same effect can be achieved when the target grinding thickness distribution of the second surface Wb is determined. That is, after grinding the second surface Wb based on the target grinding thickness distribution in St5, the second surface Wb is etched in St9 according to the optimal etching conditions determined in St8. Thus, the flatness of the etched second surface Wb can be minimized, and the shape of the second surface Wb can be appropriately controlled.

[0160] In the above embodiments, an etching amount distribution was used as the etching index distribution (learning data) for controlling the etching process of wafer W, but an etching amount deviation distribution can also be used. The etching amount deviation distribution is a distribution of values ​​obtained by subtracting the average value of the etching amount within the wafer surface from the etching amount. In this case, the etching amount deviation distribution is used instead of the etching amount distribution when determining the optimal etching conditions in St8 and St12. Furthermore, the etching amount deviation distribution is also used when determining the grinding target thickness distribution in St1.

[0161] In the above embodiments, the example described is the case where various treatments are performed on both sides (first surface Wa and second surface Wb) of a wafer W obtained by cutting or grinding from an ingot using a wire saw or the like. However, various treatments may also be performed on only one side of the wafer W.

[0162] Furthermore, in the above embodiments, examples were given of various processing of wafers W obtained by cutting or grinding from an ingot using a wire saw or the like. However, the techniques disclosed herein can also be applied, for example, in post-processing steps in the manufacturing process of semiconductor devices. Specifically, the techniques disclosed herein can also be applied, for example, when etching the surface of the ground first wafer after grinding the first wafer in an overlapping wafer formed by bonding a first wafer and a second wafer.

[0163] The embodiments disclosed herein should be considered illustrative in all respects and not restrictive. The above embodiments can also be omitted, substituted, or modified in various ways without departing from the appended claims and their spirit. For example, the constituent elements of the above embodiments can be arbitrarily combined. Based on such arbitrary combinations, the functions and effects of each constituent element related to the combination can be obtained, and other functions and effects can be clearly understood by those skilled in the art based on the description herein.

[0164] Furthermore, the effects described in this specification are merely illustrative or exemplary and not limiting. That is to say, the technology disclosed herein can achieve the aforementioned effects and other effects that are clearly known to those skilled in the art based on the description in this specification, or can replace the aforementioned effects to achieve other effects that are clearly known to those skilled in the art based on the description in this specification.

[0165] Explanation of reference numerals in the attached figures

[0166] 1: Wafer processing system; 40: Etching device; 90: Grinding device; 130: Control device; W: Wafer; Wa: First surface; Wb: Second surface.

Claims

1. A substrate processing method for processing a substrate, the substrate processing method comprising the following processes: Determine the target radial thickness distribution when grinding the surface of the substrate; The surface of the substrate is ground based on the target thickness distribution; as well as The surface of the ground substrate is then etched. in, The process for determining the target thickness distribution includes: Acquire multiple training data sets containing the radial etching amount distribution when the surface of a substrate is etched under multiple different etching conditions; Select multiple effective learning data from the plurality of learning data to determine the target thickness distribution; The etching amount distribution of the multiple effective learning data is averaged over each radial position to obtain an averaged etching amount distribution. The average etching amount distribution is converted to a reference etching amount distribution by adjusting it towards a reference etching amount. The target etching amount distribution is obtained by converting the benchmark etching amount distribution in a manner that makes it converge to the desired range; and The target etching amount distribution is determined by adding the target thickness after etching the surface of the substrate.

2. The substrate processing method according to claim 1, wherein, The process of determining the target thickness distribution includes: Using an optimization method, the method of superimposing the etching amount distributions of the multiple effective learning data into the target etching amount distribution is optimized. This involves optimizing the combination of the etching amount distributions used for superposition and the number of times these distributions are superimposed, and obtaining the corrected target etching amount distribution. The target thickness distribution is determined by adding the corrected target etching amount distribution to the target thickness after etching the surface of the substrate. In the optimization method for obtaining the corrected target etching amount distribution, the amount of effective learning data is added to the constraints.

3. The substrate processing method according to claim 2, wherein, In the optimization method for obtaining the corrected target etching amount distribution, the grinding conditions during grinding of the substrate surface are added to the constraint conditions.

4. The substrate processing method according to claim 3, wherein, The grinding conditions include at least one of the surface shape of the ground substrate and the target flatness of the ground substrate surface.

5. The substrate processing method according to claim 2, wherein, The process for determining the target thickness distribution includes: The modified target etching amount distribution is smoothed to obtain a smoothed target etching amount distribution; and The target thickness distribution is determined by adding the smoothed target etching amount distribution to the target thickness after etching the surface of the substrate.

6. The substrate processing method according to claim 1, wherein, Statistical methods were used to select the multiple valid learning data sets.

7. A substrate processing system for processing a substrate. The substrate processing system includes: A grinding device that grinds the surface of a substrate; An etching apparatus that etches the surface of a ground substrate; as well as Control device, The control device performs the following controls: Acquire multiple training data sets containing the radial etching amount distribution when the surface of a substrate is etched under multiple different etching conditions; Select multiple effective learning data from the plurality of learning data to determine the radial target thickness distribution when grinding the surface of the substrate; The etching amount distribution of the multiple effective learning data is averaged over each radial position to obtain an averaged etching amount distribution. The average etching amount distribution is converted to a reference etching amount distribution by adjusting it towards a reference etching amount. The target etching amount distribution is obtained by converting the benchmark etching amount distribution in a manner that makes it converge to the desired range; and The target etching amount distribution is determined by adding the target thickness after etching the surface of the substrate.

8. The substrate processing system according to claim 7, wherein, The control device performs the following controls: An optimization method is used to make the etching amount distributions of the multiple effective learning data superimposed to become the target etching amount distribution. The combination of the etching amount distribution used for superposition and the number of times the etching amount distribution is superimposed is optimized, and the corrected target etching amount distribution is obtained. The target thickness distribution is determined by adding the corrected target etching amount distribution to the target thickness after etching the surface of the substrate; as well as In the optimization method for obtaining the corrected target etching amount distribution, the amount of effective learning data is added to the constraints.

9. The substrate processing system according to claim 8, wherein, The control device performs the following control: in the optimization method for obtaining the corrected target etching amount distribution, the grinding conditions during the grinding of the substrate surface are added to the constraint conditions.

10. The substrate processing system according to claim 9, wherein, The grinding conditions include at least one of the surface shape of the ground substrate and the target flatness of the ground substrate surface.

11. The substrate processing system according to claim 8, wherein, The control device performs the following controls: The modified target etching amount distribution is smoothed to obtain a smoothed target etching amount distribution; and The target thickness distribution is determined by adding the smoothed target etching amount distribution to the target thickness after etching the surface of the substrate.

12. The substrate processing system according to claim 7, wherein, The control device uses statistical methods to select the multiple valid learning data.

Images