Dicing device and dicing depth setting method

Abstract

In the present disclosure, a first table map is generated by contact measurement of a table height. A master map is generated by non-contact measurement of a master height. A second table map indicating a table height distribution is generated on the basis of the master map and the known thickness of a master substrate. After the preparation is completed, a first height map is generated by non-contact measurement of a wafer height. A wafer thickness distribution is obtained by the first height map and the second table map, and a second height map indicating a wafer height distribution is generated on the basis of a first table part distribution and the wafer thickness distribution. A dicing depth is set on the basis of the second height map.

Description

Dicing apparatus and dicing depth setting method

[0001] The present invention relates to wafer dicing technology.

[0002] In the pre-process of semiconductor manufacturing, a large number of devices are formed on a wafer. In the post-process, a blade called a dicing saw (dicer) is rotated at high speed to cut a plurality of devices formed on the wafer into individual chips.

[0003] In the dicing step of cutting a wafer, the wafer is fixed to a frame by a sheet-shaped dicing tape. Hereinafter, the wafer, the dicing tape and the frame are collectively referred to as a "wafer set". A dicing apparatus cuts a wafer set fixed on a table with a blade.

[0004] The depth at which the blade cuts (hereinafter referred to as "dicing depth") is set to a depth that completely cuts the wafer and cuts into a part of the dicing tape. The thickness of the dicing tape is about 80 to 100 micrometers, and an ideal dicing depth is one that cuts about 15 micrometers into the dicing tape (hereinafter, this ideal dicing depth is referred to as "ideal depth").

[0005] If the dicing depth is too large, the contact area between the blade and the dicing tape increases. When the contact area increases, residue of the dicing tape melted by heat during cutting easily adheres to the blade, and the sharpness of the blade gradually deteriorates. On the other hand, if the dicing depth is too small, partial peeling easily occurs near the cut surface of the wafer. Such a peeling phenomenon is generally called "backside chipping". Therefore, precise setting of the dicing depth is required.

[0006] Japanese Unexamined Patent Publication No. 2021-125592

[0007] However, in reality, the table on which the wafer set is fixed is not perfectly horizontal, and even if the wafer itself is highly flat, the surface of the fixed wafer will not be perfectly horizontal. Furthermore, the table on which the wafer set is fixed cannot be moved perfectly horizontally. The inventors recognized that, taking into account the existence of these various "deviations," it is necessary to adjust the ideal depth depending on the wafer cutting location.

[0008] This invention was completed based on the inventor's recognition of the above-mentioned problems, and its main objective is to provide a technique for appropriately setting the dicing depth when dicing a wafer.

[0009] A dicing apparatus in one aspect of the present invention includes: a first table map generation unit that measures the height of a table using a first measuring instrument and generates a first table map showing the height distribution of the table; a master map generation unit that, with a master substrate of known thickness placed on the table, measures the height of the master substrate using a second measuring instrument different from the first measuring instrument and generates a master map showing the height distribution of the master substrate; a second table map generation unit that generates a second table map showing the height distribution of the table based on the master map and the thickness of the master substrate; and a second measuring instrument that, with a workpiece placed on the table... The system includes: a first height map generation unit that measures the height of a workpiece and generates a first height map showing the height distribution of the workpiece; a processing thickness map generation unit that generates a processing thickness map showing the thickness distribution of the workpiece using the first height map and a second table map; a second height map generation unit that generates a second height map showing the height distribution of the workpiece using the first table map and the processing thickness map; a depth map generation unit that generates a depth map showing the depth distribution of dicing on the workpiece based on the second height map; and a processing unit that dices the workpiece by adjusting the cutting depth of the die based on the depth map.

[0010] A dicing depth setting method in one aspect of the present invention includes the steps of: measuring the height of a table with a first measuring instrument and generating a first table map showing the height distribution of the table; measuring the height of a master substrate with a known thickness using a second measuring instrument different from the first measuring instrument, with the master substrate placed on the table, and generating a master map showing the height distribution of the master substrate; generating a second table map showing the height distribution of the table based on the master map and the thickness of the master substrate; measuring the height of a workpiece with a second measuring instrument, with the workpiece placed on the table, and generating a first height map showing the height distribution of the workpiece; generating a processing thickness map showing the thickness distribution of the workpiece using the first height map and the second table map; generating a second height map showing the height distribution of the workpiece using the first table map and the processing thickness map; and generating a depth map showing the dicing depth distribution for the workpiece based on the second height map.

[0011] According to the present invention, it becomes easier to appropriately set the dicing depth when dicing a wafer.

[0012] This is a perspective view showing the schematic configuration of the dicing apparatus. This is a perspective view showing the schematic configuration of the processing section. This is a cross-sectional view when a wafer set is cut. This is an external perspective view when a contact-type measuring instrument is attached to the spindle. This is an external perspective view when a non-contact-type measuring instrument is attached to the spindle. This is a functional block diagram of the dicing control unit. This is a flowchart showing the processing steps of the dicing process. This is a flowchart showing the initial measurement process in S10 of Figure 7. This is a top view and side view of the holding surface when the first table map is generated. This is a top view and side view of the holding surface when the master table is generated. This is a flowchart showing the wafer measurement process in S14 of Figure 7. This is a top view and side view of the holding surface when the second height map is generated. This is a schematic diagram of the cut surface of the wafer. This is a schematic diagram showing the position of the measurement points on the wafer.

[0013] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. In the following embodiment and its modified examples, substantially identical components are denoted by the same reference numerals, and their descriptions are omitted as appropriate.

[0014] (Overall Configuration of the Dicing Apparatus) Figure 1 is a perspective view showing the schematic configuration of the dicing apparatus 100 according to this embodiment. For the sake of convenience in the following explanation, the front-to-back, left-to-right, and up-and-down directions of the dicing apparatus, viewed from the front, will be described as the X, Y, and Z directions, respectively.

[0015] The dicing apparatus 100 comprises a load port 102, a transport mechanism 104, a processing unit 106, and a cleaning unit 108. A cassette containing multiple wafers W is placed on the load port 102. The transport mechanism 104 transports the wafers W to each part of the apparatus. The processing unit 106 performs dicing on the wafers W. The cleaning unit 108 spin-cleans the diced wafers W. A dicing control unit 112 is provided inside the housing 110 of the dicing apparatus 100.

[0016] The dicing control unit 112 consists of a general-purpose computer and includes a CPU (Central Processing Unit) for executing various calculation processes, memory or storage for storing control programs, memory used as a work area for data storage and program execution, input / output interfaces, etc. The dicing control unit 112 controls each functional part (mechanism and device) of the dicing apparatus 1 according to the control program.

[0017] The wafer W is stored in a cassette while being held by the frame 114. The transport mechanism 104 takes the unprocessed wafer W stored in the cassette and transports it to the processing unit 106. The processing unit 106 performs dicing (cutting) on ​​the wafer W to divide the device. After that, the wafer W is transported by the transport mechanism 104 to the cleaning unit 108 for cleaning. Then, it is transported by the transport mechanism 104 to the load port 102 and stored in a cassette.

[0018] Figure 2 is a perspective view showing the schematic configuration of the processing unit 106. The processing unit 106 is a so-called twin-spindle dicer and comprises a pair of blades 116, a workpiece holder 118 for holding the wafer W, and a microscope 120 for alignment. The blades 116 are disc-shaped dicing saws.

[0019] A pair of blades 116 are arranged opposite each other in the Y direction and are rotatably supported by a spindle 124. Each blade 116 has a rotation axis extending in the Y direction. The spindle 124 incorporates a high-frequency motor and rotates the blades 116 at high speed around the rotation axis. Each spindle 124 is supported so as to be movable in the Y and Z directions by a moving mechanism described later.

[0020] The workpiece holding section 118 includes a chuck table 126, a rotary table 128, and an X table 130. The chuck table 126 has a holding surface 126a for adsorbing and holding a wafer W. The holding surface 126a is provided with numerous adsorption holes through which vacuum is applied to adsorb the back surface of the wafer W. By driving a vacuum adsorption source (not shown), the wafer W can be adsorbed and fixed to the holding surface 126a.

[0021] A base 140 is provided on the base (not shown) of the dicing device 100, and a pair of guide rails 142 extending in the X direction are provided on the upper surface of the base 140. The X table 130 is installed horizontally so that it can move in the X direction along the guide rails 142. The X table 130 is driven by an X movement mechanism 144. In this embodiment, the X movement mechanism 144 is implemented by a linear motor, but it may also be implemented by a screw feed mechanism and a servo motor that drives it.

[0022] The rotary table 128 is rotatably supported by the X table 130, and the chuck table 126 is fixed to the upper surface of the rotary table 128. The rotary table 128 can rotate around its own axis (in the θ direction around the axis L extending in the Z direction) by the rotation mechanism 146. The rotation mechanism 146 is implemented, for example, by a spindle motor. With this configuration, the chuck table 126 is movable in the X direction and the θ direction.

[0023] On the other hand, an arch-shaped column 150 is erected on the base, and a pair of blades 116 and a microscope 120 are supported by the column 150. Specifically, a pair of guide rails 152 extending in the Y direction are provided on the front of the column 150, and a pair of Y tables 154 are installed so that they can move in the Y direction along the guide rails 152. The Y tables 154 are driven by a Y movement mechanism 156. The Y movement mechanism 156 is realized by a screw feed mechanism and a servo motor that drives it, and the pair of Y tables 154 can be driven individually.

[0024] Each Y-table 154 is provided with a pair of guide rails 158 extending in the Z direction on its front surface, and a pair of Z-tables 160 are installed so that they can move in the Z direction along the guide rails 158. The Z-tables 160 are driven by a Z-movement mechanism 162. The Z-movement mechanism 162 is realized by a screw feed mechanism and a servo motor that drives it.

[0025] In the above configuration, the X-movement mechanism 144, the Y-movement mechanism 156, the Z-movement mechanism 162, and the rotation mechanism 146 function as "movement mechanisms" that move the microscope 120 and the chuck table 126 (i.e., the wafer W) relative to each other, and also move the pair of blades 116 and the chuck table 126 relative to each other.

[0026] The microscope 120 is mounted on the Z-table 160 integrally with the spindle 124 and is held to move freely in the Y and Z directions. The microscope 120 is an imaging device (camera) that incorporates an image sensor and optical system (not shown) and images the surface of the wafer W. The image captured by the microscope 120 is used for alignment between the wafer W and the blade 116.

[0027] Figure 3 is a cross-sectional view of the wafer set 164 when it is cut. A dicing tape T is attached to the bottom surface (negative Z-axis side) of the wafer W (main object). Furthermore, the dicing tape T is fixed to a frame F. The frame F is a ring-shaped member installed on the outer circumference of the wafer W. Since the dicing tape T is adhered to both the wafer W and the frame F, the wafer W, dicing tape T, and frame F are integrated. Hereinafter, the wafer W, tape dicing tape T, and frame F are collectively referred to as "wafer set 164". The dicing tape T is disposable for each wafer W, but the frame F is reused.

[0028] The wafer set 164 (workpiece) is placed on the holding surface 126a of the chuck table 126. As described above, the wafer set 164 is stably fixed to the chuck table 126 by vacuum adsorption of the numerous adsorption holes formed on the holding surface 126a.

[0029] With the wafer set 164 fixed to the chuck table 126, the blade 116 cuts the wafer set 164. By completely cutting the wafer W contained in the wafer set 164 (full cut), multiple devices (chips) that are pre-formed on the wafer W are separated into individual pieces. In Figure 3, the device layer is formed on the upper surface (positive Z-axis side) of the wafer W. The dicing depth of the blade 116 is set to a position that completely cuts the wafer W and cuts only a portion of the dicing tape T.

[0030] As mentioned above, increasing the dicing depth increases the contact area between the blade 116 and the dicing tape T, making it easier for melted dicing tape T residue to adhere to the blade 116 due to the heat generated during cutting, thus gradually reducing the cutting performance of the blade 116. On the other hand, if the dicing depth is too shallow, backside chipping will occur. Therefore, the ideal dicing depth is one that is neither too deep nor too shallow.

[0031] Figure 4 is an external perspective view of the spindle 124 with a contact-type measuring instrument 172 (probe) attached. The processing unit 106 is equipped with two spindles 124. The blades 116 on each of the two spindles 124 have different cutting widths. The spindle 124 rotates the blades 116 at high speed by rotating its built-in motor. As described above, the relative position of the spindle 124 and the wafer set 164 changes in the X direction by moving the X table 130. In the Y direction, the relative position of the spindle 124 and the wafer set 164 changes by moving the spindle 124 itself using the Y table 154, and in the Z direction, it changes using the Z table 160.

[0032] The distance from the spindle 124 (blade 116), which is set to a predetermined height, to the top surface of the wafer W or the holding surface 126a (table) is measured by a non-contact measuring instrument 170 (second measuring instrument) or a contact measuring instrument 172 (first measuring instrument / probe). In this embodiment, the measuring instrument 172 has a probe. Figure 4 shows the spindle 124 with the measuring instrument 172 (contact type) attached.

[0033] The measuring instrument 172 is attached to the spindle 124 after the blade 116 has been removed. This allows the measurement point of the measuring instrument 172 (the contact point of the probe tip) to perfectly coincide with the cutting point of the blade 116.

[0034] Figure 5 is an external perspective view of the spindle 124 with a non-contact measuring instrument 170 attached. As described above, the distance from the spindle 124 (blade 116), which is set to a predetermined height, to the top surface of the wafer W or the holding surface 126a (table) is also measured by the non-contact measuring instrument 170. Figure 5 shows the spindle 124 with the measuring instrument 170 (non-contact) attached.

[0035] The measuring instrument 170 is a known optical sensor or air sensor and is mounted on the side of the spindle 124. As shown in Figure 5, the point directly below the blade 116, i.e., the point where the blade 116 and the wafer set 164 come into contact (hereinafter referred to as the "cutting point"), and the point directly below the measuring instrument 170, i.e., the point measured by the measuring instrument 170 (hereinafter referred to as the "measurement point"), do not coincide. However, the distance from the cutting point to the measurement point is known.

[0036] On the other hand, the contact-type measuring instrument 172 (probe) is attached to the spindle 124 after removing the blade 116. Therefore, the measurement point of the measuring instrument 172 (the contact point of the probe tip) and the cutting point of the blade 116 can be perfectly aligned. At least, the measurement point of the measuring instrument 170 (non-contact type) is further from the cutting point than the measurement point of the measuring instrument 172 (contact type). On the other hand, in the case of contact measurement using the measuring instrument 172 (contact type), it is necessary to reattach the blade 116 to the spindle 124 after measurement. Therefore, from the viewpoint of work efficiency, the non-contact type measuring instrument 170 is preferable.

[0037] In summary, contact measurement has the following characteristics: - The measurement point and the cutting point are close together. - The measuring instrument 172 (contact) cannot be attached to the spindle 124 without first removing the blade 116 from the spindle 124. On the other hand, non-contact measurement has the following characteristics: - The measurement point and the cutting point are far apart. - The measuring instrument 170 (non-contact) can be attached to the spindle 124 with the blade 116 still attached.

[0038] Generally, the thickness of the dicing tape T is about 80 to 110 micrometers. The thinner the dicing tape T, the easier it is to peel it off the wafer W. Also, the thinner the dicing tape T, the less stress is placed on the wafer W during peeling, making it less likely to damage the individual devices. Therefore, a thinner dicing tape T is preferable. On the other hand, the thinner the dicing tape T, the more difficult it becomes to adjust the dicing depth. Furthermore, the thickness of the dicing tape T is not uniform, and there are variations in thickness depending on the location. This variation in thickness creates height differences on the surface of the wafer set 164 (hereinafter, the variation and non-uniformity of the thickness of the dicing tape T will be called "tape deviation"). The tape deviation is usually about 10 micrometers.

[0039] The holding surface 126a of the chuck table 126 is not a perfectly flat surface. There are slight differences in height depending on the location on the holding surface 126a (hereinafter, the non-uniformity of the holding surface 126a will be called "table deviation"). The table deviation is usually small, at most about 2 micrometers, compared to the tape deviation.

[0040] Because there is slight distortion in the guide rails 142, etc., the X table 130, Y table 154, and Z table 160 cannot move in a perfectly straight line. The same applies to the rotation of the chuck table 126. Such distortion in the direction of movement (hereinafter referred to as "mechanical deviation") causes variations in the relative distance between the blade 116 and the wafer set 164 depending on the location. The same applies to the relative distance with the holding surface 126a. The maximum mechanical deviation is expected to be around 10 micrometers.

[0041] In summary, the surfaces of the wafer set 164 and the holding surface 126a are not perfectly horizontal due to mechanical deviations resulting from the non-linearity when moving the chuck table 126 and spindle 124, table deviations resulting from the non-uniformity of the surface of the holding surface 126a, and tape deviations resulting from the non-uniformity of the thickness of the dicing tape T. Although there are deviations in the thickness of the wafer W, they are negligibly small compared to the three types of deviations described above.

[0042] In the present embodiment, the ideal depth is set as the dicing depth when the blade 116 completely cuts the wafer W and further cuts 15 micrometers into the dicing tape T. When debris (adhesive substance) from the dicing tape T adheres to the blade 116 and dulls its cutting performance, the blade 116 needs to be cleaned. Specifically, debris adhering to the blade 116 is removed by dicing a plate called a dressing plate (this operation is called "dressing"). An increase in the number of dressings is not preferable because it reduces the working efficiency of the dicing process.

[0043] On the other hand, if the dicing depth is too small, part of the kerf (groove) formed in the wafer W is likely to be partially peeled off on the lower surface side (the surface in contact with the dicing tape T), which is backside chipping. It cannot be determined whether backside chipping has occurred during dicing. Whether backside chipping has occurred can only be confirmed after the dicing process is completed and the dicing tape T is peeled off from the wafer W. If backside chipping occurs, the wafer W is determined as a "defective product". Backside chipping is also a factor that reduces the working efficiency of the dicing process.

[0044] For the above reasons, the ideal depth as the target value of the dicing depth needs to be set to a dicing depth that is neither too deep nor too shallow. As described above, due to various deviations such as mechanical deviations, the upper surface of the wafer set 164 is not perfectly flat. For this reason, it is necessary to set the ideal depth according to the cutting point (position coordinates) of the wafer set 164. In other words, it is necessary to finely adjust the ideal depth representing the depth from the surface of the wafer set 164 according to the cutting point (position coordinates) of the wafer set 164. To set the ideal depth corresponding to each cutting point, it is necessary to accurately measure the altitude distribution on the surface of the wafer set 164.

[0045] Although slight height differences occur on the surface of the wafer set 164 due to the aforementioned deviations, the wafer W itself has high flatness, so even if height differences occur, they do not change abruptly but gradually. Therefore, by measuring the surface height of the wafer set 164 (hereinafter referred to as "wafer height WH") at multiple measurement points, the wafer height WH at points other than the measurement points can also be calculated with high accuracy by interpolation calculation.

[0046] Figure 6 is a functional block diagram of the dicing control unit 112. Each component of the dicing control unit 112 is realized by hardware including arithmetic units such as a CPU and various coprocessors, memory and storage devices, and wired or wireless communication lines connecting them, and software stored in the storage devices that supplies processing instructions to the arithmetic units. The computer program may consist of device drivers, an operating system, various application programs located at a higher layer, and libraries that provide common functions to these programs. The blocks described below represent functional units, not hardware units.

[0047] The dicing control unit 112 includes a processing control unit 182, a data storage unit 184, and a map generation unit 180. The data storage unit 184 stores various types of data. The map generation unit 180 generates various maps, which will be described later, and sets the ideal depth at each point on the wafer set 164. The processing control unit 182 transmits various control signals to the processing unit 106, causing the processing unit 106 to dice the wafer set 164.

[0048] The map generation unit 180 includes a first table map generation unit 186, a second table map generation unit 188, a first height map generation unit 190, a second height map generation unit 192, a master map generation unit 194, a wafer thickness map generation unit 196 (processed thickness map generation unit), and a depth map generation unit 198. The first table map generation unit 186 measures the height distribution of the holding surface 126a (table surface) by the measuring instrument 172 (contact type), and stores the height distribution in the data storage unit 184 as the first table map. That is, the first table map indicates the height distribution of the holding surface 126a based on contact measurement. Hereinafter, the height of the surface of the holding surface 126a is referred to as "table height TH".

[0049] The master map generation unit 194 measures the height distribution on the surface of a master substrate M (described later), and stores the height distribution in the data storage unit 184 as a master map.

[0050] The second table map generation unit 188 generates a second table map based on the master map, and stores the second table map in the data storage unit 184. Although details will be described later, the second table map indicates the table height TH based on non-contact measurement.

[0051] The first height map generation unit 190 generates a first height map indicating the height distribution of the wafer set 164 placed on the holding surface 126a, and stores the first height map in the data storage unit 184. The wafer thickness map generation unit 196 generates a wafer thickness map (processed thickness map) indicating the thickness distribution of the wafer set 164 based on the first height map and the second table map, and stores the wafer thickness map in the data storage unit 184.

[0052] The second height map generation unit 192 generates a second height map based on the wafer thickness map and the first table map and stores it in the data storage unit 184. The depth map generation unit 198 generates a depth map showing the distribution of ideal depths based on the second height map and stores it in the data storage unit 184. The processing control unit 182 notifies the processing unit 106 of the dicing depth according to the depth map. More specifically, the processing control unit 182 appropriately notifies the processing unit 106 of the ideal depth corresponding to the cutting point, and the processing unit 106 dices the wafer set 164 while fine-tuning the height of the spindle 124 according to the notified ideal depth.

[0053] Figure 7 is a flowchart showing the processing steps of the dicing process. In the dicing process, "initial measurement" is performed using the master substrate M, and then "wafer measurement" is performed for each wafer set 164. S10 in Figure 6 corresponds to the initial measurement, and S14 corresponds to the wafer measurement.

[0054] The map generation unit 180 performs initial measurements (S10). The purpose of the initial measurements is to generate a first table map and a second table map. Both the first and second table maps are data showing the height distribution of the holding surface 126a. The first table map is generated based on contact measurements, and the second table map is generated based on non-contact measurements. Details of the initial measurements will be described later in relation to Figure 8.

[0055] After the initial measurement is complete, the transport mechanism 104 places the wafer set 164 to be diced onto the chuck table 126 (S12). The map generation unit 180 performs wafer measurement (S14). The purpose of wafer measurement is to generate a depth map. Details of wafer measurement will be described later in relation to Figure 11.

[0056] After wafer measurement is complete, the processing unit 106 dices the wafer set 164 according to the depth map (S16). If there are still unprocessed wafer sets 164 remaining (N in S18), the process returns to S12, and wafer measurement is performed for the next wafer set 164. When dicing is complete for all wafer sets 164 (Y in S18), the dicing process ends.

[0057] Figure 8 is a flowchart showing the initial measurement process in S10 of Figure 7. For the initial measurement, first, the measuring instrument 172 is attached to the spindle 124. The first table map generation unit 186 measures the table height TH by contacting the measuring instrument 172 with the holding surface 126a (S20). Hereafter, when it is clear that the table height TH was measured by the contact-type measuring instrument 172, it will be written as "table height THa". The measurement result is stored in the data storage unit 184 as the first table map. Details of the process in S20 will be described further later in relation to Figure 9. After the generation of the first table map (height distribution of table height THa), the measuring instrument 172 is removed and the blade 116 is attached to the spindle 124 instead.

[0058] Next, the master substrate M is placed on the holding surface 126a. The master substrate M is a disc of known thickness. The known thickness of the master substrate M is denoted as "master thickness MT". The master map generation unit 194 controls the spindle 124 to which the measuring instrument 170 (non-contact) is attached to measure the surface hardness of the master substrate M (hereinafter referred to as "master hardness MH") non-contact (S22). The measurement result is stored as a master map in the data storage unit 184. Details of the process in S22 will be described further later in relation to Figure 10.

[0059] Finally, the second table map generation unit 188 calculates the table height TH by subtracting the master thickness MT from the master height MH, and stores this as the second table map in the data storage unit 184. Hereafter, when it is explicitly stated that the table height TH is calculated using the master height MH and master thickness MT, it will be written as "table height THb". The second table map is data showing the table height distribution calculated based on the results of non-contact measurement of the master substrate M.

[0060] Figure 9 shows a top view and a side view of the holding surface 126a during the generation of the first table map. The ultimate goal of the various measurements performed before dicing is to generate a depth map. In order to generate a depth map, it is necessary to generate a second altitude map that shows the wafer altitude WH. In order to generate the second altitude map, it is necessary to measure the table altitude THa. The process described in Figure 9 corresponds to the process in S20 of Figure 8.

[0061] The first table map generation unit 186 sets a number of measurement points (hereinafter referred to as "n points") on the holding surface 126a. The first table map generation unit 186 associates the measurement point ID, which identifies the measurement point, with the XY coordinates (horizontal plane). Hereafter, when indicating that a measurement point is the k-th (k is a natural number) measurement point among multiple measurement points, its measurement point ID will be denoted as "Pk".

[0062] There are two methods for measuring height: contact measurement using the measuring instrument 172 and non-contact measurement using air or light. In non-contact measurement using air, compressed air is injected from the measuring instrument 170, and the distance to the object is measured by the pressure change at that time. Numerous suction holes are formed on the holding surface 126a, and these suction holes suck in some of the compressed air, so measuring the table height TH with air is not appropriate. The same applies to non-contact measurement using light such as a laser. For this reason, the first table map generation unit 186 measures the table height THa by contact measurement using the measuring instrument 172.

[0063] The first table map generation unit 186 stores data as a first table map in the data storage unit 184, associating the measurement point ID, XY coordinates, and table height THa for each of the n measurement points. In contact measurement, the cutting point, which is the lowest point of the blade 116, and the measurement point of the measuring instrument 172 coincide. Hereinafter, in the first table map, the table height THa at the measurement point with measurement point ID = Pk will be denoted as "THa(Pk)". Since contact measurement is performed while moving the X table 130, Y table 154, and Z table 160, the table height THa(Pk) ​​incorporates machine deviation and table deviation.

[0064] Figure 10 shows a top view and a side view of the holding surface 126a during master table generation. After the generation of the first table map, the master substrate M is placed on the holding surface 126a. The master substrate M has a degree of flatness that can be considered to be perfectly flat. Since the distortion of the surface of the master substrate M is negligibly small, the master thickness MT can be considered to be constant. The process described in Figure 9 corresponds to the processes S20 and S22 in Figure 7.

[0065] Since the master substrate M is a high-precision component, contact measurement that puts stress on the surface of the master substrate M is undesirable. Furthermore, since there are no structures such as suction holes on the surface of the master substrate M, non-contact measurement is possible. In this embodiment, the master height MH, i.e., the Z coordinate value of the surface of the master substrate M, is determined by non-contact measurement using air. The measurement of the master substrate M is performed by a non-contact measuring instrument 170 attached to the side of the spindle 124 (see Figure 5).

[0066] If the blade 116 is not properly attached to the spindle 124, eccentricity of the blade 116 will occur, which can lead to backside chipping. For this reason, when attaching the blade 116 to the spindle 124, it is necessary to check whether the blade 116 is properly attached to the spindle 124 by dicing an object other than the wafer W (generally called "pre-cutting"), and to make adjustments as needed.

[0067] Repeated adjustment work due to pre-cutting reduces overall work efficiency. For this reason, it is undesirable to remove the blade 116 from the spindle 124 and then attach the measuring instrument 170 to the spindle 124. In this embodiment, the measuring instrument 170 is additionally attached to the side of the spindle 124, where the blade 116 is already attached. As shown in Figure 5, when the measuring instrument 170 is attached to the side of the spindle 124, there is a discrepancy of about 5 centimeters between the measurement point and the cutting point.

[0068] When measuring the master substrate M, a measuring instrument 170 (non-contact type) is attached to one spindle 124, and a microscope 120 is attached to the other spindle 124. Since the positional relationship between the point directly below the measuring instrument 170 (measurement point) and the point directly below the microscope 120 (imaging point) is known, the position of the measuring instrument 170 is adjusted so that the measurement point of the measuring instrument 170 and the measurement point (cutting point) of the measuring instrument 172 (probe) perfectly coincide.

[0069] The master map generation unit 194 generates data associating each of the multiple measurement points with the measurement point ID, XY coordinates, and master altitude MH, and stores it as a master map in the data storage unit 184. In the following, the master altitude MH at the measurement point with measurement point ID = Pk will be denoted as "MH(Pk)" in the master map. The master altitude MH(Pk) incorporates machine deviation and table deviation. Because the position adjustment described above has been performed, the position coordinates of the measurement points in the first table map and the measurement points in the master map are in perfect agreement.

[0070] Since the master thickness MT is known, the table height can be obtained by subtracting the master thickness MT from the master height MH. The second table map generation unit 188 calculates the table height TH at each measurement point based on the master map and the master thickness MT. The set of table heights obtained in this way becomes the second table map. The first table map shows the table height THa measured by contact measurement using the measuring instrument 172, while the second table map shows the table height TH calculated from non-contact measurement of the master substrate M. Hereafter, the table height of each measurement point (Pk) calculated as the second table map will be denoted as "THb(Pk)". The table height THb(Pk) incorporates both mechanical deviation and table deviation.

[0071] Figure 11 is a flowchart showing the wafer measurement process in S14 of Figure 7. The purpose of wafer measurement is to determine the actual wafer height WH when the wafer set 164 is placed on the chuck table 126. Here, "actual" means the wafer height WH after taking into account machine deviation, table deviation, and tape deviation. The depth map generation unit 198 generates a depth map showing the distribution of ideal depths based on a second height map showing the height distribution of the actual wafer height WH. Figure 11 shows the overall flow, and further details will be described later in relation to Figure 12.

[0072] After the master measurement is completed, the wafer set 164 to be diced is placed on the holding surface 126a. The first altitude map generation unit 190 performs non-contact measurement of the wafer altitude WH (S30). Since the last measurement in the initial measurement (S22) is a non-contact measurement, the wafer set 164 can continue to be measured non-contact in S30 without removing the blade 116. Also, since the wafer set 164 is a delicate object, though not as delicate as the master substrate M, non-contact measurement is preferable to contact measurement. In S30, the measurement result of the wafer altitude WH becomes the first altitude map. Hereafter, when it is necessary to specify that the wafer altitude WH was measured by the non-contact measuring instrument 170, it will be written as "wafer altitude WHa".

[0073] Next, the wafer thickness map generation unit 196 calculates the thickness of the wafer set 164 (hereinafter referred to as "processing target thickness OT") based on the first height map (non-contact) and the second table map (non-contact) (S32). The distribution of the processing target thickness OT becomes the wafer thickness map, details of which will be described later. Subsequently, the second height map generation unit 192 generates a second height map by calculating the wafer height WH based on the wafer thickness map and the first table map (contact) that was initially generated (S34). Finally, the depth map generation unit 198 generates a depth map based on the second height map (S36). Hereinafter, when it is explicitly stated that the wafer height WH is calculated from the wafer thickness map and the first table map, it will be referred to as "wafer height WHb".

[0074] Figure 12 shows a top view and a side view of the holding surface 126a during the generation of the second height map. As described above, wafer measurement is performed after the first table map (contact type) and the second table map (calculated from the non-contact measurement results of the master substrate M) are obtained as a result of the initial measurement.

[0075] After an initial measurement is performed, multiple wafer measurements are carried out (see Figure 7). To improve work efficiency, the number of measurement points set in wafer measurement (hereinafter referred to as "m points") is less than the number of measurement points set in the initial measurement (n points) (m < n). In other words, in the initial measurement, which is performed only once, a large number of measurement points are targeted, and a precise high-level measurement is performed over a long period of time. On the other hand, in wafer measurement, which is performed multiple times for each wafer set 164, a relatively small number of measurement points are set to take work efficiency into consideration.

[0076] Since the wafer set 164 includes tape deviation, the processing target thickness OT, which is the thickness of the wafer set 164, is not constant. First, the first height map generation unit 190 non-contact measures the wafer height WHa (S30 in Figure 10). For each of the multiple measurement points, the first height map generation unit 190 stores the data as a first height map in the data storage unit 184, as data that associates the measurement point ID, XY coordinates, and wafer height WHa. In the following, in the first height map, the wafer height WHa at the measurement point with measurement point ID = Pk will be expressed as "WHa(Pk)". The wafer height WHa(Pk) ​​incorporates machine deviation, table deviation, and tape deviation. The set of non-contact measured wafer heights WHa(Pk) ​​becomes the first height map.

[0077] The measurement points on the first altitude map (non-contact) are set to coincide with any of the measurement points on the second table map (non-contact).

[0078] The wafer thickness map generation unit 196 calculates the processing target thickness OT by subtracting the table height THb shown in the second table map (non-contact) from the wafer height WHa shown in the first height map (non-contact). For example, the processing target thickness OT at measurement point (Pk) can be obtained by subtracting the table height THb(Pk) at the same measurement point (Pk) from the wafer height WHa(Pk) ​​at the same measurement point (Pk). The wafer thickness map generation unit 196 generates a wafer thickness map as a set of processing target thicknesses OT at each measurement point. In the following, the processing target thickness OT at the measurement point ID = Pk will be denoted as "OT(Pk)" in the wafer thickness map.

[0079] When data for the wafer height WHa(Pk) ​​corresponding to a measurement point (Pk) does not exist, the wafer thickness map generation unit 196 can calculate the wafer height WHa(Pk) ​​by interpolation calculation based on the measured values ​​of multiple wafer heights WHa around the measurement point (Pk), and then calculate the processing target thickness OT(Pk) using this data. Since the wafer height WHa changes smoothly, such interpolation calculation is possible (see Patent Document 1).

[0080] Wafer measurement includes mechanical and table deviations, but initial measurement also includes mechanical and table deviations. Therefore, mechanical and table deviations can be removed by subtracting the table height THb(Pk) from the wafer height WHa(Pk). In other words, the processing target thickness OT(Pk) at each measurement point shown in the wafer thickness map includes only tape deviation and is therefore close to the actual measured thickness of the wafer set 164. In other words, the actual thickness of the wafer set 164 can be accurately calculated from the wafer height WHa(Pk) ​​and table height THb(Pk) without directly measuring the thickness of the wafer set 164. For each of the n measurement points included in the second table map, the processing target thickness OT is calculated.

[0081] The first table map is based on contact measurements by a probe, so the measurement points and cutting points coincide. The second height map generation unit 192 calculates the wafer height WHb (hereinafter referred to as "WHb(Pk)") at the measurement point (Pk) by adding the processing target thickness OT(Pk) at the measurement point (Pk) to the table height THa(Pk) ​​in the first table map. The second height map is generated as a set of wafer heights WHb(Pk) obtained in this way. The second height map is data that shows the wafer height at the cutting point by the blade 116.

[0082] Next, the depth map generation unit 198 generates a depth map showing the distribution of ideal depths based on the second elevation map. For example, if the wafer elevation at the measurement point (Pk) is WHb(Pk), then the ideal depth at the cutting point (Pk) will be a position (negative Z-axis direction) deeper than WHb(Pk) by "WT+d". WHb(Pk) is the wafer elevation at the cutting point (Pk), i.e., the Z-coordinate value of the surface of the wafer set 164. "WT" is the thickness of the wafer W. The thickness of the wafer W is known. The deviation of the wafer W thickness is significantly smaller than the tape deviation, table deviation, and machine deviation, so it can be considered a constant value. Also, "d" is the depth of the cut in the dicing tape T, and the set value in this embodiment is 15 micrometers.

[0083] Figure 13 is a schematic diagram of the cross-section of the wafer W. In Figure 12, the blade 116 cuts into the wafer set 164 at the cutting point (Pk). According to the second height map, the wafer height WH (Z coordinate value) at the cutting point (Pk) is WHb(Pk). The ideal depth is a position deeper than the wafer height WHb(Pk) by "WT+d". The processing control unit 182 notifies the processing unit 106 of the ideal depth for each cutting point set in the depth map, and the processing unit 106 dices the wafer set 164 while finely adjusting the height of the spindle 124 according to the cutting point.

[0084] [Summary] The dicing apparatus 100 has been described above based on the embodiment. According to this embodiment, the ideal depth can be determined based on the advantages and disadvantages of both non-contact measurement and contact measurement. Contact measurement is preferable for table height, while non-contact measurement is preferable for master height and wafer height. With contact measurement, the measurement point and the cutting point can be made to coincide, but with non-contact measurement, the measurement point and the cutting point do not coincide, or it is difficult to make them coincide.

[0085] The dicing apparatus 100 measures the table height THa of the holding surface 126a by generating a first table map through contact measurement. Next, the master map generation unit 194 generates a master map by non-contact measurement of the master substrate M, and generates a second table map by calculating the table height THb based on the master map. In other words, two types of table maps are generated: a first table map based on contact measurement and a second table map based on non-contact measurement.

[0086] After the initial measurements described above are completed, the dicing apparatus 100 performs wafer measurements for each wafer set 164. The first height map generation unit 190 non-contactly measures the wafer height WHa and generates a first height map. The wafer thickness map generation unit 196 calculates a wafer thickness map (processing target thickness OT) based on the first height map (wafer height WHa: non-contact) and the second table map (table height THb: non-contact).

[0087] The second altitude map generation unit 192 calculates the processing target thickness OT for each of the numerous measurement points included in the first table map by interpolation calculation based on this wafer thickness map. Then, the second altitude map generation unit 192 generates a second altitude map showing the wafer altitude WHb at the cutting point based on the wafer thickness map and the first table map. After the above processing steps, the depth map generation unit 198 generates a depth map, which is the distribution of ideal depths corresponding to each cutting point of the wafer set 164, while utilizing the characteristics of both contact measurement and non-contact measurement.

[0088] In the depth map, the ideal depth for each cutting point is set, taking into account tape deviation, table deviation, and machine deviation. This prevents the blade 116 from cutting deeper than necessary, thereby reducing the number of times the blade 116 needs to be replaced due to the adhesion of the dicing tape T. It also suppresses backside chipping caused by the blade 116 cutting too shallowly.

[0089] In this embodiment, the efficiency of wafer measurement can be improved by reducing the number of measurement points set on the wafer W to fewer than the number of measurement points set in the master measurement.

[0090] It should be noted that the present invention is not limited to the embodiments and modifications described above, and the components can be modified and implemented without departing from the spirit of the invention. Various inventions may be formed by appropriately combining the multiple components disclosed in the embodiments and modifications described above. In addition, some components may be deleted from all the components shown in the embodiments and modifications described above.

[0091] [Modifications] In this embodiment, the workpiece was described assuming a wafer set including a wafer W, a dicing tape T, and a frame F. The workpiece may consist only of a wafer W, without the dicing tape T, frame F, etc. The workpiece may also include other members such as a base plate that supports the wafer W.

[0092] In this embodiment, the explanation was given assuming a full cut of the wafer W, but the dicing method shown in this embodiment can also be applied to a half cut. For example, when making a cut 150 micrometers from the surface of a wafer with a thickness of 775 micrometers, a constant dicing depth of 150 micrometers can be maintained by taking into account machine deviation, tape deviation, table deviation, etc.

[0093] Figure 14 is a schematic diagram showing the location of measurement points on wafer W. It is desirable that measurement points be set on all of the numerous devices formed on wafer W. While devices appear flat or smooth macroscopically, they have fine irregularities microscopically. For example, the areas where electrode pads are formed within a device are slightly raised, and the areas where streets are formed are slightly recessed. In non-contact measurement, especially optical measurement, the detection accuracy is high, so slight differences in the wafer height measurement may occur depending on where on the device the measurement point is set.

[0094] Taking into account the minute altitude differences specific to such devices, the first altitude map generation unit 190 may set measurement points at positions moved by a predetermined distance in both the X-axis and Y-axis directions, relative to a predetermined position on the device, for example, the intersection of the processing line (a corner of the device).

[0095] In the example shown in Figure 14, processing lines CLX1, CLX2, and CLX3 are set in the X-axis direction on the wafer W. Processing lines CLY1, CLY2, and CLY3 are also set in the Y-axis direction. Devices D1 to D4 are formed between these six processing lines. For device D1, the first height map generation unit 190 sets a measurement point P1 at a point moved by a distance L1 in the Y-axis direction and a distance L2 in the X-axis direction from the intersection of processing line CLX1 and processing line CLY1. Measurement points P2 to P4 are set in the same manner for the other devices D2 to D4.

[0096] This method of setting measurement points allows for the measurement of wafer height WH at the same location on all wafers. In other words, it eliminates the influence of subtle height differences due to the device's specific structure, allowing each device to be measured under the same conditions.

Claims

1. A first table map generation unit that measures the height of a table using a first measuring instrument and generates a first table map showing the height distribution of the table; a master map generation unit that, with a master substrate of known thickness placed on the table, measures the height of the master substrate using a second measuring instrument different from the first measuring instrument and generates a master map showing the height distribution of the master substrate; a second table map generation unit that generates a second table map showing the height distribution of the table based on the master map and the thickness of the master substrate; a first height map generation unit that, with a workpiece placed on the table, measures the height of the workpiece using a second measuring instrument and generates a first height map showing the height distribution of the workpiece; a machining thickness map generation unit that generates a machining thickness map showing the thickness distribution of the workpiece using the first height map and the second table map; a second height map generation unit that generates a second height map showing the height distribution of the workpiece using the first table map and the machining thickness map. A dicing apparatus comprising: a depth map generation unit that generates a depth map showing the depth distribution of dicing for the workpiece based on the second height map; and a processing unit that dices the workpiece by adjusting the cutting depth of the dicer based on the depth map.

2. The dicing apparatus according to claim 1, wherein a dicing tape is attached to the lower surface of the main body of the workpiece, and the depth map generation unit sets the depth at which the main body and a portion of the dicing tape are cut as the dicing depth in the depth map.

3. The dicing apparatus according to claim 1, wherein the first measuring instrument is a contact-type measuring instrument and the second measuring instrument is a non-contact-type measuring instrument.

4. The dicing apparatus according to claim 3, wherein the workpiece includes a wafer on which a plurality of devices are formed, the second measuring instrument is an optical measuring instrument, and the first height map generation unit measures the height of the workpiece with respect to predetermined positions for each of the plurality of devices using the second measuring instrument while the workpiece is placed on the table.

5. The dicing apparatus according to claim 1, wherein the second measuring instrument is attached to a movable spindle to which the dicer is fixed, and the master map generation unit adjusts the position of the spindle so that the coordinates of the multiple measurement points in the first table map and the multiple measurement points in the master map are the same.

6. The dicing apparatus according to claim 1, wherein the first measuring instrument and the second measuring instrument are attached to a movable spindle on which the dicer is fixed, and the distance from the cutting point of the dicer to the measuring point of the first measuring instrument is set to be shorter than the distance from the cutting point of the dicer to the measuring point of the second measuring instrument.

7. The dicing apparatus according to claim 1, wherein the number of measurement points in the first altitude map is less than the number of measurement points in the master map.

8. The dicing apparatus according to claim 7, wherein the first height map generation unit expands the first height map by calculating the expected height of the workpiece at multiple measurement points in the second table map based on the measurement values ​​of each of the multiple measurement points in the first height map, and the processing thickness map generation unit generates the processing thickness map based on the expanded first height map.

9. A dicing depth setting method comprising: measuring the height of a table with a first measuring instrument and generating a first table map showing the height distribution of the table; measuring the height of a master substrate with a known thickness using a second measuring instrument different from the first measuring instrument, with the master substrate placed on the table, and generating a master map showing the height distribution of the master substrate; generating a second table map showing the height distribution of the table based on the master map and the thickness of the master substrate; measuring the height of a workpiece with a second measuring instrument, with the workpiece placed on the table, and generating a first height map showing the height distribution of the workpiece; generating a processing thickness map showing the thickness distribution of the workpiece based on the first height map and the second table map; generating a second height map showing the height distribution of the workpiece based on the first table map and the processing thickness map; and generating a depth map showing the dicing depth distribution for the workpiece based on the second height map.

Images