JP2016096167A - Semiconductor chip manufacturing method, circuit board and electronic apparatus - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a semiconductor chip manufacturing method which can inhibit breakage of a stepped part in comparison with the case where a groove on a rear face side with a depth of reaching a groove on a surface side is formed by one-time cutting.SOLUTION: A semiconductor chip manufacturing method comprises: a process of forming from a surface of a substrate W, a surface side groove 140 along a cut region of the substrate; a process of forming from a rear face of the substrate and along the surface side groove 140, a first rear face side groove 170A which has a depth of not reaching the surface side groove 140 and a width Sb1 wider than a width Sa of the surface side groove 140; a process of forming a rear face side groove 170B having a width Sb2 wider than a width Sa of the surface side groove 140 from a bottom of the first rear face side groove 170A to reach the surface side groove 140 by a rotating cutting blade after forming the first rear face side groove 170A.SELECTED DRAWING: Figure 17

Description

  The present invention relates to a method for manufacturing a semiconductor piece, a circuit board, and an electronic device.

  A first groove is formed with a first blade from the front surface side of the sapphire substrate, and then a second groove wider and deeper than the first groove is formed with a second blade from the back surface side. There has been proposed a dicing method capable of improving the yield without reducing the number of chips that can be obtained from the substrate (Patent Document 1). Also, there is a method of increasing the number of semiconductor elements that can be formed on a wafer by forming a groove with a laser from the wafer surface to the middle of the wafer and then cutting with a blade from the back surface of the wafer to the position where the laser groove is reached. It has been proposed (Patent Document 2).

JP 2003-124151 A JP 2009-88252 A

  A depth that reaches the groove on the surface side by a rotating cutting member that forms a groove on the surface side from the surface of the substrate along the cutting region of the substrate and has a thickness larger than the groove width on the surface side from the back surface of the substrate. A method is known in which the substrate on the back surface side is formed along the groove on the front surface side and the substrate is separated into pieces.

In this method, when the cutting member is displaced in the groove width direction when forming the groove on the back surface side, the stress applied to the stepped portion formed by the difference in width between the groove on the front surface side and the groove on the back surface side is increased. In some cases, the stepped portion is increased and the stepped portion is damaged.
In particular, when the substrate is thick and the depth of cutting is deep, or when the thickness of the cutting member is thin and the cutting member is likely to warp, the groove on the back surface reaching the surface side groove is formed once. When trying to form by cutting, the cutting member warps in the middle of cutting, and there is a case where the stepped portion is likely to be damaged due to displacement in the groove width direction at the connection portion between the groove on the front surface side and the groove on the back surface side. there were.

  Therefore, in the present invention, compared to the case where the groove on the back surface having a depth reaching the groove on the front surface side is formed by one-time cutting, a method for manufacturing a semiconductor piece, a circuit board, and an electronic device capable of suppressing breakage of the stepped portion The purpose is to provide.

According to a first aspect of the present invention, there is provided a step of forming a groove on the surface side along the cutting region of the substrate from the surface of the substrate, and the groove on the surface side is not reached along the groove on the surface side from the back surface of the substrate. Forming a first backside groove having a depth that is wider than a width of the frontside groove; and after forming the first backside groove, the first backside Forming a second back-side groove having a width wider than the width of the front-side groove with a cutting member that rotates from the bottom of the side-side groove to the front-side groove. A manufacturing method of a piece.
According to a second aspect of the present invention, the groove on the first back surface side has a first depth from the substrate back surface to the bottom of the groove on the first back surface side, and the groove on the second back surface side is The first depth has a second depth from a bottom portion of the groove on the first back surface side to a bottom portion of the groove on the front surface side, and the first depth is larger than the second depth. The manufacturing method of the semiconductor piece of description.
According to a third aspect of the present invention, in the semiconductor piece manufacturing method according to the first or second aspect, the width of the groove on the first back surface side is narrower than the width of the groove on the second back surface side.
According to a fourth aspect of the present invention, the groove on the first back surface side is formed by a rotating cutting member, and the taper degree of the tip portion of the cutting member forming the groove on the second back surface side is the groove on the first back surface side. The manufacturing method of the semiconductor piece of any one of Claim 1 thru | or 3 which is larger than the taper degree of the front-end | tip part of the cutting member which forms.
According to a fifth aspect of the present invention, the groove on the first back surface side is formed by a rotating cutting member, and the thickness of the cutting member forming the groove on the second back surface side forms the groove on the first back surface side. The manufacturing method of the semiconductor piece of any one of Claim 1 thru | or 4 thinner than the thickness of a cutting member.
Claim 6 is the usage time of the cutting member that forms the groove on the second back surface side is longer than the usage time of the cutting member that forms the groove on the first back surface side. Manufacturing method of semiconductor piece.
Claim 7 includes a step of forming a third backside groove between the step of forming the first backside groove and the step of forming the second backside groove. The method for manufacturing a semiconductor piece according to claim 1.
According to claim 8, the first to third back side grooves are formed by a rotating cutting member, and the width of the third back side groove is narrower than the width of the first back side groove, The method of manufacturing a semiconductor piece according to claim 7, wherein a width of the groove on the second back surface side is narrower than a width of the groove on the third back surface side.
A circuit board on which at least one semiconductor piece manufactured by the manufacturing method according to any one of claims 1 to 8 is mounted.
An electronic device on which the circuit board according to claim 9 is mounted.

According to the first aspect, the stepped portion can be prevented from being damaged as compared with the case where the groove on the back surface side reaching the groove on the front surface side is formed by one cutting.
According to the second aspect, the contact resistance to the cutting member can be reduced as compared with the case where the depth of the groove on the second back surface side is made larger than the depth of the groove on the first back surface side.
According to the third aspect, the contact resistance to the cutting member can be reduced as compared with the case where the width of the groove on the second back surface side is made larger than the width of the groove on the first back surface side.
According to claim 4, the taper degree of the tip part of the cutting member that forms the groove on the second back surface side is made smaller than the taper degree of the tip part of the cutting member that forms the groove on the first back surface side. In comparison, damage to the stepped portion can be suppressed.
According to the fifth aspect, compared to the case where the thickness of the cutting member that forms the groove on the second back surface side is thicker than the thickness of the cutting member that forms the groove on the first back surface side, Damage can be suppressed.
According to claim 6, compared with the case where the usage time of the cutting member that forms the groove on the second back surface side is shorter than the usage time of the cutting member that forms the groove on the first back surface side, the step portion Can be prevented from being damaged.
According to the seventh aspect, the depth of the groove on the second back surface side can be reduced as compared with the case where the groove on the third back surface side is not formed.
According to the eighth aspect, damage to the step portion can be suppressed as compared with the case where the width of the groove on the second back surface side is not made smaller than the width of the groove on the third back surface side.

It is a flow which shows an example of the manufacturing process of the semiconductor piece which concerns on the Example of this invention. It is typical sectional drawing of the semiconductor substrate in the manufacturing process of the semiconductor piece which concerns on the Example of this invention. It is typical sectional drawing of the semiconductor substrate in the manufacturing process of the semiconductor piece which concerns on the Example of this invention. It is a schematic plan view of a semiconductor substrate (wafer) when circuit formation is completed. 5A is a cross-sectional view for explaining the cutting operation of the dicing blade, FIGS. 5B to 5F are enlarged cross-sectional views of the tip portion of the dicing blade of this embodiment, and FIG. It is an expanded sectional view of the tip part of a dicing blade used for general full dicing. 6A is an enlarged cross-sectional view of the tip of the dicing blade used in the simulation, and FIG. 6B is a groove formed on the semiconductor substrate when the dicing blade shown in FIG. 6A is used. 6C and 6D are enlarged cross-sectional views of the tip portion of the dicing blade having the curvature radii r = 0.5 and r = 12.5 used in the simulation. It is a graph when the relationship between the curvature radius of the front-end | tip part of a dicing blade and the stress value produced in the corner part of a level | step-difference part is simulated. It is a graph when the relationship between the curvature radius of the front-end | tip part of a dicing blade and the maximum stress value is simulated. 9A is a cross-sectional view illustrating the stress applied to the corner portion of the step portion, and FIG. 9B is a cross-section illustrating an example in which the step portion is damaged by the stress generated in the corner portion of the step portion. FIG. It is a figure explaining the stress of the level | step-difference part when the dicing blade of FIG. 5 (B) is used. FIG. 11A is a cross-sectional view of the stepped portion when the center of the groove 140 and the center of the groove 170 coincide with each other, and FIG. 11B shows the position when the center of the groove 140 and the center of the groove 170 are misaligned. It is sectional drawing of a level | step-difference part. FIGS. 12A to 12D are diagrams for explaining four types of dicing blades used in the simulation regarding the positional deviation. It is a graph which shows the result of having simulated the influence which position shift amount and a kerf width give to a level difference part. It is a figure which illustrates the position where the maximum stress occurs when the kerf width Sb is very narrow and the displacement amount Ds is large. It is a figure which shows the experimental result when an actual board | substrate is cut | disconnected by the various dicing blade from which the kerf width Sb and the curvature radius of a front-end | tip corner | angular part differ. It is a figure which shows the result of the experiment conducted in order to confirm the influence on the damage of the level | step-difference part by the difference in the groove width of the surface side, and the damage on the level | step-difference part by the difference in the thickness of the level | step-difference part. It is a figure explaining the half dicing which concerns on the other Example of this invention, FIG. 17 (A) is a schematic sectional drawing when performing the 1st step | paragraph, FIG. 17 (B) is the 2nd step | paragraph. It is a schematic sectional drawing when cutting is performed. It is a figure explaining curvature of a dicing blade when cutting a slot on the back side. It is a schematic sectional drawing explaining the half dicing of the 2nd aspect which concerns on the other Example of this invention. It is a figure explaining the half dicing of the 3rd mode concerning other examples of the present invention, and Drawing 20 (A) is an outline sectional view when performing the 1st stage cutting, and Drawing 20 (B) is It is a schematic sectional drawing when the 2nd-stage cutting is performed. It is a figure explaining the half dicing of the 3rd mode concerning other examples of the present invention. It is a figure explaining the half dicing of the 3rd mode concerning other examples of the present invention. It is a figure which shows the dicing blade applicable to the other Example of this invention, and is an example of the dicing blade with a large taper degree of the front-end | tip part which can be obtained from the outside. FIG. 24A shows the shape of the tip portion of the dicing blade before the start of use, and FIG. 24B shows the dicing blade with the tip portion worn after a certain period of use. It is sectional drawing which shows the typical structure of the fine groove | channel by the Example of this invention. It is a figure which shows the flow of the manufacturing method which forms the fine groove | channel by the Example of this invention. It is a schematic sectional drawing of the manufacturing process of a flask-shaped fine groove | channel by the manufacturing method by the Example of this invention.

  The method of manufacturing a semiconductor piece according to the present invention is, for example, a method of manufacturing individual semiconductor pieces (semiconductor chips) by dividing (dividing into pieces) a substrate-like member such as a semiconductor wafer on which a plurality of semiconductor elements are formed. Applies to The semiconductor element formed on the substrate is not particularly limited, and can include a light emitting element, an active element, a passive element, and the like. In a preferred embodiment, the manufacturing method of the present invention is applied to a method of taking out a semiconductor piece including a light emitting element from a substrate, and the light emitting element can be, for example, a surface emitting semiconductor laser, a light emitting diode, or a light emitting thyristor. One semiconductor piece may include a single light-emitting element, or may include a plurality of light-emitting elements arranged in an array. A driving circuit for driving one or a plurality of light emitting elements can also be included. Further, the substrate can be, for example, a substrate made of silicon, SiC, a compound semiconductor, sapphire, or the like, but is not limited thereto, and is a substrate including at least a semiconductor (hereinafter collectively referred to as a semiconductor substrate). Any other material substrate may be used. In a preferred embodiment, it is a III-V group compound semiconductor substrate such as GaAs on which a light emitting element such as a surface emitting semiconductor laser or a light emitting diode is formed.

  In the following description, a method for forming a plurality of light emitting elements on a semiconductor substrate and taking out individual semiconductor pieces (semiconductor chips) from the semiconductor substrate will be described with reference to the drawings. It should be noted that the scale and shape of the drawings are emphasized for easy understanding of the features of the invention, and are not necessarily the same as the scale and shape of an actual device.

  FIG. 1 is a flowchart showing an example of a manufacturing process of a semiconductor piece according to an embodiment of the present invention. As shown in the figure, the method for manufacturing a semiconductor piece of this example includes a step of forming a light emitting element (S100), a step of forming a resist pattern (S102), and a step of forming fine grooves on the surface of a semiconductor substrate ( S104), a step of peeling the resist pattern (S106), a step of attaching a dicing tape to the surface of the semiconductor substrate (S108), a step of half dicing from the back surface of the semiconductor substrate (S110), and an ultraviolet (UV) on the dicing tape. ), A step of attaching an expanding tape to the back surface of the semiconductor substrate (S112), a step of peeling the dicing tape and irradiating the expanding tape with ultraviolet rays (S114), and picking a semiconductor piece (semiconductor chip) And a step (S116) of die mounting on a circuit board or the like. The cross-sectional views of the semiconductor substrate shown in FIGS. 2A to 2D and FIGS. 3E to 3I correspond to steps S100 to S116, respectively.

  In the step of forming light emitting elements (S100), as shown in FIG. 2A, a plurality of light emitting elements 100 are formed on the surface of a semiconductor substrate W such as GaAs. The light emitting element 100 is, for example, a surface emitting semiconductor laser, a light emitting diode, a light emitting thyristor, or the like. Note that although one region is shown as the light emitting element 100 in the drawing, one light emitting element 100 illustrates an element included in one separated semiconductor piece, and one light emitting element 100 is illustrated. It should be noted that the region may include not only one light emitting element but also a plurality of light emitting elements and other circuit elements.

  FIG. 4 is a plan view illustrating an example of the semiconductor substrate W when the light emitting element formation process is completed. In the drawing, for the sake of convenience, only the light emitting element 100 at the center portion is illustrated. On the surface of the semiconductor substrate W, a plurality of light emitting elements 100 are formed in an array in the matrix direction. The planar area of one light emitting element 100 is generally rectangular, and each light emitting element 100 is separated in a grid by a cutting area 120 defined by a scribe line or the like having a constant interval S.

  When the formation of the light emitting element is completed, a resist pattern is then formed on the surface of the semiconductor substrate W (S102). As shown in FIG. 2B, the resist pattern 130 is processed so that the cutting region 120 defined by a scribe line or the like on the surface of the semiconductor substrate W is exposed. The resist pattern 130 is processed by a photolithography process.

  Next, a fine groove is formed on the surface of the semiconductor substrate W (S104). 2C, using the resist pattern 130 as a mask, a fine groove 140 (hereinafter referred to as a fine groove or a groove on the surface side for convenience) is formed on the surface of the semiconductor substrate W. The Such a groove can be formed by, for example, anisotropic etching, and is preferably formed by anisotropic plasma etching (reactive ion etching) which is anisotropic dry etching. It may be formed by a thin dicing blade or isotropic etching, but by using anisotropic dry etching, the width is narrower and deeper than forming grooves on the surface side by isotropic etching. It is preferable because a groove can be formed and the influence of vibration, stress, etc. on the light emitting element 100 around the fine groove can be suppressed more than when a dicing blade is used. The width Sa of the fine groove 140 is substantially equal to the width of the opening formed in the resist pattern 130, and the width Sa of the fine groove 140 is, for example, several μm to several tens of μm. Further, the depth is, for example, about 10 μm to 100 μm, and is formed deeper than at least a depth at which a functional element such as a light emitting element is formed. When the fine groove 140 is formed by a general dicing blade, the interval S between the cutting regions 120 becomes as large as about 40 to 60 μm as a total of the margin width considering the groove width and chipping amount of the dicing blade itself. On the other hand, when the fine groove 140 is formed by a semiconductor process, not only the groove width itself but also the margin width for cutting can be made narrower than the margin width when a dicing blade is used, in other words, The interval S between the cutting regions 120 can be reduced. For this reason, the number of obtained semiconductor pieces can be increased by arranging the light emitting elements at a high density on the wafer. In this embodiment, the “front side” means a surface side on which a functional element such as a light emitting element is formed, and the “back side” means a surface side opposite to the “front side”.

  Next, the resist pattern is peeled off (S106). As shown in FIG. 2D, when the resist pattern 130 is peeled from the surface of the semiconductor substrate, the fine groove 140 formed along the cutting region 120 is exposed on the surface. Details of the shape of the fine groove 140 will be described later.

  Next, an ultraviolet curable dicing tape is attached (S108). As shown in FIG. 3E, a dicing tape 160 having an adhesive layer on the light emitting element side is attached. Next, half dicing is performed along the fine groove 140 by a dicing blade from the back side of the substrate (S110). The positioning of the dicing blade can be done by placing an infrared camera on the back side of the substrate and indirectly detecting the fine groove 140 through the substrate, or by placing a camera on the surface side of the substrate and directly positioning the fine groove 140. The detection method and other known methods can be used. By such positioning, as shown in FIG. 3F, half dicing is performed by a dicing blade, and a groove 170 is formed on the back surface side of the semiconductor substrate. The groove 170 has a depth that reaches the fine groove 140 formed on the surface of the semiconductor substrate. Here, the fine groove 140 is formed with a width narrower than the groove 170 on the back surface side by the dicing blade. However, if the fine groove 140 is formed with a width narrower than the groove 170 on the back surface side, only the dicing blade is formed. This is because the number of semiconductor pieces that can be obtained from one wafer can be increased as compared with the case of cutting the semiconductor substrate. In addition, if it is possible to form fine grooves of about several μm to several tens of μm as shown in FIG. It is not easy to form a fine groove having such a depth. Therefore, as shown in FIG. 3F, the half dicing from the back surface by the dicing blade is combined.

  Next, the dicing tape is irradiated with ultraviolet rays (UV), and an expanding tape is attached (S112). As shown in FIG. 3G, the dicing tape 160 is irradiated with ultraviolet rays 180, and the adhesive layer is cured. Thereafter, an expanding tape 190 is attached to the back surface of the semiconductor substrate.

  Next, the dicing tape is peeled off, and the expanding tape is irradiated with ultraviolet rays (S114). As shown in FIG. 3H, the dicing tape 160 is peeled off from the surface of the semiconductor substrate. Further, the expanding tape 190 on the back surface of the substrate is irradiated with ultraviolet rays 200, and the adhesive layer is cured. The expanding tape 190 has a stretchable base material, stretches the tape so that it is easy to pick up semiconductor pieces separated after dicing, and extends the interval between the light emitting elements.

  Next, picking and die mounting of the separated semiconductor pieces are performed (S116). As shown in FIG. 3I, the semiconductor piece 210 picked from the expanding tape 190 is mounted on the circuit board 230 via a fixing member 220 such as a conductive paste such as an adhesive or solder.

  Next, details of half dicing by a dicing blade will be described. FIG. 5A is a cross-sectional view when half dicing is performed by the dicing blade shown in FIG.

  As described above, a plurality of light emitting elements 100 are formed on the surface of the semiconductor substrate W, and each light emitting element 100 is separated by a cutting region 120 defined by a scribe line having a spacing S or the like. A fine groove 140 having a width Sa is formed in the cut region 120 by anisotropic dry etching. On the other hand, as shown in FIG. 5A, the dicing blade 300 is a disc-shaped cutting member that rotates about the axis Q, and has a thickness corresponding to the groove 170 having a kerf width Sb. The dicing blade 300 is aligned outside the semiconductor substrate W in a direction parallel to the back surface of the semiconductor substrate W. Further, by moving by a predetermined amount in the direction Y perpendicular to the back surface of the semiconductor substrate W, alignment in the thickness direction with respect to the semiconductor substrate W is performed so that the stepped portion 400 has a desired thickness T. After the alignment, the dicing blade 300 is rotated, and at least one of the dicing blade 300 or the semiconductor substrate W is moved in a direction parallel to the back surface of the semiconductor substrate W, so that the semiconductor substrate W is moved. A groove 170 is formed. Since the kerf width Sb is larger than the width Sa of the fine groove 140, when the groove 170 reaches the fine groove 140, the cut region 120 has a cantilever-like shape having a thickness T due to the difference between the width Sb and the width Sa. A stepped portion 400 having a bowl shape is formed. If the center of the dicing blade 300 and the center of the fine groove 140 are completely coincident with each other, the length of the stepped portion 400 extending in the lateral direction is (Sb−Sa) / 2.

A) Explanation of the tip portion FIGS . 5B to 5F are enlarged sectional views of the tip portion A of the dicing blade 300 as an example in the embodiment of the present invention, and FIG. It is an expanded sectional view of the front-end | tip part A of the dicing blade used for full dicing. As shown in FIG. 5G, the tip of a dicing blade 300A used for general full dicing has one side surface 310, a side surface 320 facing the one side surface, and both side surfaces 310, 320. And a flat top surface 340 that intersects at substantially right angles. That is, the cross section viewed from the rotation direction has a rectangular tip. On the other hand, the dicing blade 300 of the present embodiment gradually moves toward the top of the dicing blade 300 as shown in FIGS. 5B to 5F, for example. The taper has a tapered shape in which the thickness of the taper is reduced.

  In this embodiment, the “top portion” is the most distal end portion of the dicing blade. If the shape is as shown in FIG. 5 (B), FIG. 5 (D) and FIG. One point at the tip. 5C and FIG. 5F, the top portion is a flat surface except for fine irregularities, and this flat surface is referred to as a “top surface”. In addition, “tapering” refers to a shape in which the tip of the dicing blade 300 has a portion that gradually decreases in thickness toward the top, and each of FIGS. 5B to 5F is tapered. This is an example of the shape.

  Here, each shape of FIG. 5B to FIG. 5G shows an initial shape when the semiconductor substrate is cut in the mass production process. That is, the dicing blade 300 of this embodiment shown in FIGS. 5B to 5F has such a shape in advance as an initial shape in the mass production process. 5 (G) used for general full dicing has a rectangular shape in the initial state, but as it continues to be used, FIG. 5 (B) through FIG. It wears into the taper shape which has the curved surface 330 as shown in FIG.5 (D).

  In the example illustrated in FIG. 5B, a pair of side surfaces 310 and 320 and a curved surface 330 is provided between the pair of side surfaces 310 and 320. Specifically, the distance between the pair of side surfaces 310 and 320 is a width corresponding to the kerf width Sb, and the tip portion includes a semicircular curved surface 330 between both side surfaces 310 and 320, 5 (C) and the top surface 340 as shown in FIG. 5 (F) are not included. The example illustrated in FIG. 5C is an intermediate shape between FIG. 5B and FIG. 5G, and has a curved surface 330 at the tip corner along with the top surface 340. The example illustrated in FIG. 5D does not have the top surface 340 but has a curved surface 330 having a radius of curvature larger than the curvature radius of the tip corner in FIGS. 5B and 5C, and the top portion. A curved surface 370 having a smaller radius of curvature than the curved surface 330 is formed at the position of. 5B to 5D, the rate at which the thickness of the dicing blade 300 decreases as the distance from the top of the dicing blade 300 increases.

  In the example illustrated in FIG. 5E, a curved surface 370 is formed between the two chamfers 350 and 360. Also in this case, the top surface 340 is not formed as in FIG. In the example illustrated in FIG. 5F, the side surfaces 310 and 320 that face each other and the top surface 340 between the side surfaces 310 and 320 are formed, and chamfers 350 and 360 are formed between the side surfaces 310 and 320 and the top surface 340. Has been. A curved surface 352 is formed at the corner between the chamfer 350 and the top surface 340, and a curved surface 362 is formed at the corner between the chamfer 360 and the top surface 340.

  The tip of the dicing blade according to the present embodiment may have a shape that is tapered from the rectangular tip as shown in FIG. 5 (G), as shown in FIGS. 5 (B) to 5 (F). Unless otherwise specified, it may or may not have a top surface. Further, the tip of the dicing blade 300 according to this embodiment shown in FIGS. 5B to 5F is based on the center K of the thickness of the dicing blade 300 as shown in FIG. 5D. It has a line-symmetric shape. However, unless otherwise specified, the shape is not necessarily line-symmetric, and the position of the top (top surface) may be shifted in the thickness direction of the dicing blade 300.

B) Explanation of simulation and experimental results Next, a simulation performed to confirm what kind of damage is caused for what cause in the case where fine groove widths of several μm to several tens of μm are communicated with each other. The experiment will be described.

B-1) Explanation of Simulation Regarding Tip Shape FIGS . 6 to 8 are diagrams for explaining the simulation performed to grasp the relationship between the radius of curvature of the tip corner portion of the dicing blade and the stress applied to the step portion and the result thereof. FIG. An example of the dicing blade 302 used for the simulation is shown in FIG. FIG. 6A shows a cross-sectional shape of the tip portion viewed from the rotating direction of the dicing blade 302. As shown in FIG. 6A, the tip of the dicing blade 302 has side surfaces 310 and 320, a top surface 340 having a certain length, and a curvature formed between the side surfaces 310 and 320 and the top surface 340. Including a curved surface 330 having a radius r, and the tip is configured symmetrically with respect to a line orthogonal to the rotation axis.

  FIG. 6B shows the shape of the groove formed in the semiconductor substrate when the tip-shaped dicing blade 302 shown in FIG. 6A is used. Here, due to the difference in the position of the side surface of the groove 140 on the front surface side of the substrate and the side surface of the groove 170 on the back surface side of the substrate, the width W is between the vertical side surface of the groove 140 on the front surface side and the groove 170 on the back surface side. A step is formed, and a step-shaped region 400 having a thickness T, that is, a step portion 400 is formed by the step. In other words, the stepped portion 400 is a portion between the step formed at the connecting portion of the front surface side groove 140 and the back surface side groove 170 and the surface of the semiconductor substrate.

  In this simulation, the curvature radius r (μm) of the curved surface 330 of the dicing blade 302 is set to r = 0.5, r = 2.5, r = 5.0, r = 7.5, r = 10.0. , R = 12.5, the stress value applied to the stepped portion 400 was calculated by simulation. The thickness of the dicing blade 302 is 25 μm, FIG. 6C shows the shape of the tip of r = 0.5, FIG. 6D shows the shape of the tip of r = 12.5, and FIG. The tip of D) has a semicircular shape in which the radius of curvature of the tip corner is ½ of the thickness of the dicing blade 302. The substrate to be processed is a GaAs substrate, the groove width of the surface-side groove 140 is 5 μm, the thickness T of the stepped portion 400 is 40 μm, and from the groove 170 on the back surface side to the surface side of the substrate with respect to the stepped portion 400. The load was set to 2 mN. The center of the width of the groove 140 on the surface side and the center of the thickness of the dicing blade 302 were matched.

  The graph shown in FIG. 7 is a result of simulation, and shows a change in the stress value applied to the stepped portion 400 when the radius of curvature of the tip corner is changed. Here, the vertical axis represents the stress value [Mpa], and the horizontal axis represents the X coordinate when the center of the surface-side groove 140 shown in FIG. From the graph, the stress increases as the X coordinate approaches 12.5 μm at any curvature radius r, that is, from the center side of the groove 170 on the back surface side toward the root side of the stepped portion 400. It can also be seen that as the value of the radius of curvature r increases, the stress applied to the base side of the stepped portion 400 decreases, and the way in which the stress rises becomes milder. In other words, in the tip shape range used in this simulation, that is, in the tip shape having a smaller degree of taper than the semicircular tip portion as shown in FIG. Maximum stress is occurring. Further, the stress applied to the base side of the stepped portion 400 is smaller in the semicircular tip shape as shown in FIG. 6D than in the shape close to a rectangle as shown in FIG. That is, the greater the degree of taper, the smaller the stress applied to the base side of the stepped portion 400. Further, in the case of a shape close to a rectangle as shown in FIG. 6C, for example, when r = 0.5, the stress is smaller in the range where the X coordinate is up to about 11 μm than in the case where the radius of curvature r is large. It can be seen that the stress suddenly increases in the range exceeding 1, that is, the portion closer to the root, and the stress is concentrated near 12.5 μm in the X coordinate.

  FIG. 8 shows the relationship between the radius of curvature on the horizontal axis and the maximum stress value on the vertical axis. In the graph, in addition to the value of the radius of curvature r shown in FIG. 7, a simulation is also performed for r = 25 μm and r = 50 μm, and the results are also shown. The tip shape when the curvature radius r exceeds a semicircular curvature radius of 12.5 μm, such as 25 μm or 50 μm, is a shape with a greater degree of taper, for example, as shown in FIG. From the graph, the smaller the radius of curvature r, that is, the closer the tip shape is to a rectangular shape, the higher the maximum stress value, and the greater the degree of change in the maximum stress with respect to the change in the radius of curvature r. Conversely, when the radius of curvature r increases, the maximum stress value decreases, and the degree of change of the maximum stress with respect to the change of the radius of curvature r is reduced from about 5 μm, and the radius of curvature is in the range of 12.5 to 50 μm. That is, it can be seen that the variation of the maximum stress value is almost constant in the range of the tapered shape having no top surface as shown in FIG. 6 (D) or FIG. 5 (D).

  Based on the above simulation results, the mechanism by which the semiconductor piece is broken will be described with reference to FIGS. As shown in FIG. 9A, in the case where the tip portion is rectangular like the dicing blade 300A (when the value of the radius of curvature r is very small), a groove 170 having a kerf width Sb is formed from the back surface of the semiconductor substrate. In doing so, the substrate is pressed by the top surface 340 of the dicing blade 300A. Although the force F by the dicing blade 300A is applied to the entire stepped portion 400, it is considered that the force F applied to the stepped portion 400 is concentrated on the base side region (the root region 410) of the stepped portion 400 by the lever principle. . Then, when the stress concentrated on the root region 410 exceeds the fracture stress of the wafer, as shown in FIG. 9B, the root region 410 of the step portion 400 is damaged (chip, crack, picking, etc.). If the stepped portion 400 is damaged, a margin M for cutting the stepped portion 400 must be secured. This is because the interval S between the cut regions 120 is equal to or larger than the margin M. It means you have to. From the result of the simulation in FIG. 8, when r = 0.5 and r = 12.5, the stress applied to the root region 410 of the stepped portion 400 is almost four times different. In the range where the value of the radius of curvature r is smaller than that of the semicircular tip as shown in FIG. 5B or FIG. 6D, that is, in the tip shape having the top surface, the tip It shows that the stress applied to the root region 410 of the stepped portion 400 varies greatly depending on the value of the radius of curvature r of the corner portion. Note that the “root region” in this embodiment is a stepped portion that is parallel to the substrate surface by using a tip shape having a top surface as shown in FIGS. 5C, 5F, and 5G. In this case, a region closer to the vertical side surface of the groove 170 on the back surface side than the position of ½ of the width Wh of the step portion parallel to the substrate surface formed on both sides of the groove on the front surface side is formed. Say. Note that the relationship between the width Wh and the width Wt of the stepped portion is shown in FIG. In addition, when a tapered step shape having no top surface as shown in FIGS. 5B, 5D, and 5E is used, dicing is not performed when a step portion parallel to the substrate surface is not formed. When the blade is assumed to be divided into three equal parts in the thickness direction, it refers to a stepped region corresponding to 1/3 of the thickness on each side of the central region.

  FIG. 10 is a cross-sectional view for explaining the application of stress to the stepped portion 400 when the groove 170 is formed by the dicing blade 300 of this embodiment shown in FIG. FIG. 10 shows an example in which the tip of the dicing blade 300 is semicircular, and in this case, the shape of the groove 170 is also semicircular so as to follow this. As a result, the force F applied to the stepped portion 400 by the tip of the dicing blade 300 is distributed in the direction along the semicircular shape of the groove. Therefore, as shown in FIG. 9A, stress concentration on the stepped portion 400 is suppressed from being concentrated on the root region 410 of the stepped portion 400, and thus the stepped portion 400 is damaged (chip, cracked, cracked, etc.). ) Is considered to be suppressed. In addition, the suppression of breakage in this specification is not limited to that which prevents cracks, cracks, cracks, etc. from being confirmed. The degree of suppression is not limited, including those that can be reduced.

B-2) Simulation for misregistration Next, the misregistration amount of the dicing blade in the groove width direction will be described. FIGS. 11A and 11B are views for explaining the positional relationship between the width Sa of the groove 140 on the surface side formed on the substrate surface and the kerf width Sb of the groove 170 formed by the dicing blade. Ideally, the center of the kerf width Sb coincides with the center of the width Sa of the groove 140 on the surface side, as shown in FIG. However, in practice, due to manufacturing variations, the center of the kerf width Sb is displaced from the center of the width Sa of the groove 140 on the surface side, as shown in FIG. As a result of the displacement, a difference also occurs in the width Wt of the left and right step portions 400. The difference between the center of the width Sa of the groove 140 on the surface side and the center of the kerf width Sb is defined as a positional deviation amount Ds. The manufacturing variation is mainly determined by the manufacturing conditions such as the positioning accuracy of the manufacturing apparatus to be used (including the detection accuracy of the alignment mark and the like) and the degree of deformation of the dicing blade (the amount of bending and warping).

  Next, a simulation performed for grasping the relationship between the positional deviation amount Ds in the groove width direction of the dicing blade and the stress applied to the step portion 400, and the kerf width Sb of the dicing blade and the stress applied to the step portion 400 are as follows. The simulation performed for grasping the relationship will be described. In this simulation, the kerf width Sb (μm) at the position of 12.5 μm from the top of the dicing blade is set to four types of Sb = 25, Sb = 20.4, Sb = 15.8, and Sb = 11.2. For each kerf width, the stress value when the positional deviation amount Ds (μm) from the surface side groove 140 was changed to Ds = 0, Ds = 2.5, and Ds = 7.5 was calculated by simulation. . Although the tip shape used in this simulation is different from the tip shape used in the simulation according to FIG. 6, it is common in that the tip shape is implemented using a plurality of tip shapes having different degrees of tapering. The substrate to be processed is a GaAs substrate, the thickness of the dicing blade is 25 μm, the radius of curvature of the tip corner is r = 5 μm, the width Sa of the groove 140 on the surface side of the semiconductor substrate is 5 μm, and the thickness of the step 400 T was set to 40 μm. Moreover, it set so that the load of a total of 10 mN might be applied to the normal line direction of the side surface of the level difference part 400 and the groove | channel 170 of the back surface side. The load on the side surface of the groove 170 on the back surface side takes into account the lateral vibration of the dicing blade during actual cutting.

  FIGS. 12A to 12D show the shapes of the four types of kerf widths (tip shape of the dicing blade) used in the simulation in a state where the positional deviation amount Ds is zero. FIG. 12A shows a shape with Sb = 25 μm, FIG. 12B shows a shape with Sb = 20.4 μm, FIG. 12C shows a shape with Sb = 15.8 μm, and FIG. ) Is the shape of Sb = 11.2 μm. In any of the shapes, the surface other than the curved surface at the corner of the tip is a linear shape. In the case of Sb = 11.2 μm in FIG. 12D, the curvature radius in the top region is 5 μm as shown in the figure. And a shape having no tip corner.

  FIG. 13 shows a result of simulating the influence of the positional deviation amount Ds and the kerf width Sb on the stepped portion. The vertical axis represents the maximum stress value applied to the stepped portion 400, and the horizontal axis represents the kerf width Sb. The kerf width Sb on the horizontal axis is a width at a position of 12.5 μm from the top of the dicing blade. The case results are plotted.

  As is apparent from the graph of FIG. 13, it can be understood that the maximum stress applied to the stepped portion 400 increases as the positional deviation amount Ds of the dicing blade in the groove width direction increases in any kerf width Sb. Although not expressed in FIG. 13, the maximum stress is generated in the root region 410 on the side where the width Wt of the stepped portion 400 is increased due to the displacement of the dicing blade. This is presumably because when the positional deviation amount Ds is increased, a greater stress is easily applied to the root region 410 of the stepped portion 400 on the side where the step is increased by the lever principle.

  Further, the narrower the kerf width Sb (the greater the degree of taper), the smaller the maximum stress value tends to be. However, this is a stress that presses the stepped portion 400 toward the substrate surface side due to the greater degree of taper. This is considered to be because the stress is less likely to concentrate on the root region 410 of the stepped portion 400 because of weakening. In addition, when the kerf width Sb is very narrow (Sb = 11.2) and the positional deviation amount Ds is large (Ds = 7.5 μm), the location where the maximum stress value is generated changes abruptly (approximately 7. It can be seen that 2) increases. This is because a dicing blade having a wide kerf width Sb (a dicing blade having a small taper degree) gives stress to the stepped portion 400 on a wide surface, but a dicing blade having a very narrow kerf width Sb (a degree of taper). In the case of a very large dicing blade), it is considered that when the top (vertex) deviates from the range of the groove 140 on the surface side of the semiconductor substrate, stress concentrates on the tapered top (vertex) region. Although not expressed in FIG. 13, according to the simulation results, the maximum stress when the kerf width Sb is very narrow (Sb = 11.2) and the positional deviation amount Ds is large (Ds = 7.5 μm) is This position is shown as P in FIG. Note that the “top region” in the present embodiment is a region including the top, and is a region closer to the center of the groove on the back surface side than the root region 410 of the stepped portion 400.

B-3) Description of First Experimental Results Next, FIG. 15 shows experimental results when a plurality of dicing blades having different degrees of taper are prepared and an actual substrate is cut. In this experiment, the tip of a dicing blade having a thickness of 25 μm was processed, and a plurality of dicing blades having a radius of curvature r of the tip corner of 1 μm to 23 μm and a kerf width of 5 μm to 25 μm at a position 5 μm from the top. Got ready. A specific combination of the radius of curvature and the kerf width is as shown in FIG. 15, and the dicing blades were prepared so that the degree of tapering was substantially uniform. Further, a GaAs substrate is used, the width of the groove 140 on the surface side is set to about 5 μm, the thickness T of the stepped portion 400 is set to about 40 μm, and the positional deviation amount Ds of the dicing blade in the groove width direction is less than ± 7.5 μm. It was. Since the thickness of the dicing blade is 25 μm, when the radius of curvature r at the tip corner is 12.5 μm or more, the tip has a tapered shape with no top surface, while the radius of curvature is less than 12.5 μm. In the smaller range, the smaller the degree, the smaller the degree of taper. When the radius of curvature is 1 μm, the tip has a substantially rectangular shape.

  “◯” in FIG. 15 indicates that the stepped portion 400 is sufficiently prevented from being damaged, and that the degree of taper can be used in the mass production process. “X” indicates that the stepped portion 400 is sufficiently prevented from being damaged. This indicates that the degree of taper is not usable in the mass production process. In FIG. 15, there are unusable ranges both in the range where the degree of taper is small (the radius of curvature r is 8 μm or less) and in the large range (the radius of curvature r is 22 μm or more). Is present. As shown in the previous simulation results, stress concentrates on the root region 410 of the stepped portion 400 in the range where the taper degree is small, and the stepped portion 400 is damaged. In the range where the taper degree is large, the top (vertex) of the dicing blade. This is because stress concentrates at the position of, and the stepped portion 400 is damaged. Note that when the radius of curvature r is 8 μm or less, the stepped portion is damaged because the degree of taper is small, and when the radius of curvature r is 22 μm or more, the stepped portion is damaged because the degree of taper is large.

  As shown in the simulation of FIG. 8, the maximum stress received by the stepped portion 400 varies greatly depending on the degree of tapering of the tip portion. Therefore, even if a rectangular tip shape or any other tip shape is used, even if it is damaged, as shown in the experiment in FIG. If the tip shape is controlled so as to fit within the range, the manufacturing conditions are changed, such as increasing the thickness T of the stepped portion 400 so that the strength of the stepped portion is increased (the width of the groove 140 on the surface side is made wider and deeper). Even without this, it can be seen that damage to the stepped portion is suppressed to a level that does not cause a problem in the mass production process.

B-4) Explanation of Second Experiment Result Next, in order to confirm the influence on the stepped portion due to the difference in the groove width on the surface side and the influence on the stepped portion due to the difference in the thickness of the stepped portion. The results of the experiment conducted are shown in FIG. In this experiment, a GaAs substrate was used, and a dicing blade having a thickness T of 25 μm and 40 μm and a kerf width of 16.7 μm at a position of 5 μm from the tip was used. Then, for each width Sa of the groove 140 on the front surface side and for each thickness T of the stepped portion 400, damage to the stepped portion 400 is suppressed to what extent the positional displacement is in the groove width direction of the dicing blade. It was confirmed whether it can be used in the mass production process. “A” to “D” in FIG. 16 indicate the range of the positional deviation amount Ds in which the result that the damage of the stepped portion 400 is sufficiently suppressed is obtained.

  For example, when the thickness T of the step portion is 25 μm and the groove width Sa on the surface side is 7.5 μm, it is “B”, which means that the dicing blade is in the range of ± 5 μm to less than ± 7.5 μm in the groove width direction. Even if there is a variation, this indicates that the stepped portion 400 is sufficiently prevented from being damaged and can be used in a mass production process, and the stepped portion 400 for a positional deviation of ± 7.5 μm or more. This indicates that the breakage of is not sufficiently suppressed. Further, when the thickness T of the stepped portion 400 is 45 μm and the groove width Sa on the front side is 5 μm, it is “A”. This is even when the dicing blade is displaced ± 7.5 μm or more in the groove width direction. It shows that 400 is sufficiently damaged and can be used in a mass production process. Further, when the thickness T of the stepped portion 400 is 25 μm and the groove width Sa on the front surface side is 5 μm, it is “D”. This is only when the deviation in the groove width direction of the dicing blade is less than ± 3 μm. The damage was sufficiently suppressed, and when it was shifted by ± 3 μm or more, it was indicated that the damage of the stepped portion 400 was not sufficiently suppressed.

  From the experimental results of FIG. 16, it is shown that the stepped portion 400 is more resistant to displacement in the groove width direction of the dicing blade as the width Sa of the groove 140 on the front side is wider. That is, the stepped portion 400 is less likely to be damaged by the stress from the dicing blade as the width Sa of the groove 140 on the surface side is wider. This is presumably because the principle W of the lever becomes difficult to work because the width W of the stepped portion 400 becomes narrower as the width Sa of the groove 140 on the front surface side becomes wider. Further, it is shown that the thickness T of the stepped portion 400 is stronger against the positional deviation in the groove width direction of the dicing blade. That is, the stepped portion 400 is less likely to be damaged by the stress from the dicing blade when the thickness T of the stepped portion 400 is thicker. This is because the strength against stress increases as the thickness T of the stepped portion 400 increases.

C Half Dicing Next, another embodiment of half dicing (S110) from the back surface shown in FIG. 1 will be described. In the above-described embodiment, the back surface side groove 170 communicating with the fine groove 140 on the front surface side is formed by a single cutting process using a dicing blade. However, in this embodiment, the back surface side by half dicing is used. The example which forms this groove | channel by the cutting process of multiple times is shown. In the following description, an example in which half dicing is performed by a two-stage cutting process will be described unless otherwise specified. If half-dicing is performed in multiple stages of three or more stages, the first-stage cutting process in the two-stage cutting process is applied to the multi-stage first-stage cutting process, and the two-stage cutting process is performed. The second-stage cutting step can be applied to a multi-stage final-stage cutting step, that is, a step of forming a back-side groove communicating with the front-side fine groove.

  FIG. 17 is a schematic cross-sectional view illustrating an example of half dicing according to the present embodiment. As shown in the figure, the substrate W has a thickness Wz, and on the surface thereof, light emitting elements 100 as device regions are spaced apart by a spacing S, and fine grooves 140 having a width Sa smaller than the spacing S are formed. It is formed by a dry process. The width Sa of the fine groove 140 may be 5 μm, for example. As shown in FIG. 17A, a first back side groove 170A is formed from the back side of the substrate, and then the surface along the first back side groove 170A as shown in FIG. 17B. By forming the second back side groove 170B communicating with the fine groove 140 on the side, half dicing from the back side is completed.

  Although the cutting method of the groove 170A on the first back surface side is not particularly limited, for example, the groove 170A on the first back surface side is formed by a dicing blade. In addition to using a dicing blade, the first backside groove 170A may be formed by thermally melting or evaporating the substrate by laser dicing. When forming the groove 170A on the first back surface side, the dicing blade is positioned on the alignment mark on the front surface or the back surface of the substrate, and the rotated dicing blade is moved in a direction parallel to the back surface of the substrate. A kerf width Sb1 equal to the thickness of the dicing blade and a first back side groove 170A having a bottom portion with a depth Da from the back side of the substrate are formed. The kerf width Sb1 is larger than the width Sa of the fine groove 140 on the surface side.

  After forming the first back side groove 170A, the second back side groove 170B is formed using a dicing blade. For example, the dicing blade uses the first back surface side groove 170A as an alignment mark, and is positioned there, and the rotated dicing blade is moved in a direction parallel to the back surface of the substrate to thereby form the first back surface side groove. A second back surface side groove 170B that connects 170A and the front surface side fine groove 140 is formed. The groove 170B on the second back surface side has a kerf width Sb2 equal to the thickness of the dicing blade and a bottom portion having a depth Db, and the curve width Sb2 is larger than the width Sa of the fine groove 140 on the front surface side. The length Db has a distance (Wz−D1−Da) from at least the bottom of the groove 170A on the first back surface side to the bottom of the groove 140 on the front surface side.

  When half dicing is performed in a single cutting process, cutting with a dicing blade with a thick substrate, or with a thin dicing blade even if the substrate is not so thick, Due to the stress of time, as shown in FIG. 18, the dicing blade 300 warps or bends in the direction perpendicular to the rotation axis thereof. The warping of the dicing blade 300 increases the stress in the side surface direction of the dicing blade, and easily causes misalignment at the connecting portion between the fine groove 140 and the back surface side groove 170, and as a result, the back surface side groove 170. A crack 420 or the like from the bottom to the substrate surface occurs, and the stepped portion 400 is easily damaged. For this reason, it is necessary to select a dicing blade having a large thickness in order to increase the rigidity of the dicing blade. On the other hand, using such a thick dicing blade requires a large cutting margin for cutting the semiconductor piece, that is, the interval S between the light emitting elements 110 in FIG. 17 must be set large. In other words, the number of semiconductor pieces that can be obtained from one substrate is reduced.

  In this embodiment, compared with the case where the groove on the back surface side is formed by a single cutting step, the amount of cutting when forming the second groove on the back surface side leading to the fine groove on the front surface side is reduced, and dicing is performed. The cutting resistance to the blade is reduced, and the warping or deflection amount of the dicing blade is reduced. As a result, the stress in the side surface direction of the dicing blade is reduced, and the displacement at the connection portion between the fine groove and the groove on the second back side is suppressed. Therefore, a substrate having a large thickness can be cut while suppressing breakage of the stepped portion. In addition, since the thickness of the dicing blade when forming the second back surface side groove can be reduced compared to the case where the back surface side groove is formed by a single cutting process, the cutting margin is reduced. It is possible to increase the number of semiconductor pieces that can be obtained from one substrate.

  In the present embodiment, as described above, the formation of the last back side groove communicating with the fine groove on the front side is always performed using a dicing blade, and the formation of the back side groove in other cutting processes is performed. The dicing blade is not limited. Further, when a dicing blade is used in each cutting process, the dicing blade in each cutting process may be the same or different. When different dicing blades are used, the thickness and taper degree of each dicing blade may be the same or different. When different dicing blades are used in each cutting process, the order in which the dicing blades are used is not limited. For example, the first backside groove may be formed by a thick dicing blade, and then the second backside groove may be formed by a thin dicing blade, and vice versa. A groove on the first back surface side may be formed by a dicing blade, and then a groove on the second back surface side may be formed by a thick dicing blade. Alternatively, the groove on the first back surface side may be formed by a dicing blade having a small taper degree, and then the groove on the second back surface side may be formed by a dicing blade having a large taper degree. A groove on the first back surface side may be formed by a dicing blade having a large degree, and then a groove on the second back surface side may be formed by a dicing blade having a small taper degree.

  In each cutting process, when the first backside groove and the second backside groove are formed by a dicing blade having the same thickness, that is, when the kerf width Sb1 = Sb2, the cut surface of the semiconductor piece is flat. It becomes a serious aspect. On the other hand, when the first back surface side groove and the second back surface side groove are formed by dicing blades having different thicknesses, that is, when the kerf width Sb1 ≠ Sb2, A step corresponding to the thickness of the film can be formed.

  In addition, the dicing apparatus includes a one-axis type that mounts one dicing blade and a two-axis type that simultaneously mounts two dicing blades. When using different dicing blades in each cutting process, if it is a single axis type, replace the dicing blade for each cutting process, and if it is a two axis type, install two different dicing braces, 1. Select one of the dicing blades. If it is a biaxial type, a different dicing blade can be used with the substrate attached.

Next, some aspects of the half dicing of the present embodiment will be described.
In the first aspect, cutting is performed such that the depth Da of the groove 170A on the first back surface side is larger than the depth Db of the groove 170B on the second back surface side. As shown in FIG. 17A, the first back surface side groove 170A has a bottom portion having a depth Da from the back surface of the substrate, and the second back surface side groove 170B is formed as shown in FIG. 17B. Each of the cutting steps is performed such that Da> Db, with a depth from the bottom of the first back-side groove 170A to the bottom of the front-side groove 140. By making the depth Da of the groove 170A on the first back surface side larger than the depth Db of the groove 170B on the second back surface side, the fine groove 140 on the front surface side is compared with the case of Da <Db. By reducing the cutting amount of the groove 170B on the second back surface side that communicates with the cutting force, cutting resistance is alleviated, warping or bending of the dicing blade is suppressed, and damage to the step portion is suppressed.

  In the second aspect, half dicing is performed by a three-stage cutting process. FIG. 19 is a schematic cross-sectional view illustrating a three-stage cutting process. As shown in the figure, in the first stage cutting process, a first back surface side groove 170A is formed from the back surface of the substrate, and in the second stage cutting process, the first back surface side groove 170A is formed along the first back surface side groove 170A. The third back surface side groove 170C is formed, and in the third cutting step, the second back surface side groove 170B is formed along the third back surface side groove 170B. The second back-side groove 170B allows the third back-side groove 170C and the front-side fine groove 140 to communicate with each other. According to the second aspect, the cutting amount of the groove 170B on the second back surface side is further reduced and the cutting resistance of the dicing blade is reduced as compared with the case where the groove on the back surface side is formed by a two-stage cutting process. Thus, warping of the dicing blade and breakage of the stepped portion are suppressed.

  In the third aspect, when all the grooves on the back surface side are formed by using a dicing blade, the thickness of the second dicing blade that cuts the second stage is larger than the thickness of the first dicing blade that performs the first stage cutting. Reduce the thickness. FIG. 20 is a schematic cross-sectional view illustrating the cutting process of the third aspect. As shown in FIG. 20A, in the first cutting process, a first backside groove 170A is formed by the first dicing blade. At this time, the kerf width Sb1 of the groove 170A on the first back surface side is substantially equal to the thickness of the first dicing blade. Next, as shown in FIG. 20B, in the second-stage cutting process, a second backside groove 170B is formed by the second dicing blade. The second back-side groove 170B allows the first back-side groove 170A and the front-side fine groove 140 to communicate with each other. At this time, the kerf width Sb2 of the groove 170B on the second back surface side is substantially equal to the thickness of the second dicing blade. By such a cutting process, the relationship between the groove widths of the first back surface side groove 170A, the second back surface side groove 170B, and the front surface side fine groove 140 is Sb1> Sb2> Sa.

  As in the third aspect, the thickness of the second dicing blade that forms the second back surface side groove 170B is smaller than the thickness of the first dicing blade that forms the first back surface side groove 170A. By doing so, the width Wt (see FIG. 6B) of the stepped portion is reduced, and breakage of the stepped portion is suppressed. In addition, by reducing the thickness of the second dicing blade, the rigidity of the second dicing blade is reduced. However, since the first dicing blade 170A is formed and then the second dicing blade is used for cutting. The cutting amount that the second dicing blade itself cuts is small, and therefore the cutting resistance of the second dicing blade is reduced, and the warping or bending of the second dicing blade is suppressed. In the third aspect, Da> Db can be satisfied as in the first aspect.

  Next, FIG. 21 shows an example in which the third aspect is applied to the second aspect, that is, an example in which half dicing is performed in a three-stage cutting process. As shown in the figure, a first back side groove 170A having a width Sb1 and a depth Da is formed, and then a third back side groove 170C having a width Sb3 and a depth Dc is formed. When the groove 170B on the second back surface side of Sb2 and depth Db is formed, cutting is performed so that Sb1 = Sb3> Sb2. Further, as in the first embodiment, the depth Db <Dc or Db <Da can be satisfied. By performing such cutting, the same effect as described above can be obtained.

  Further, FIG. 22 shows that the width Sb3 of the groove 170C on the third back surface side is smaller than the width Sb3 of the groove 170C on the third back surface side so that the width Sb1 of the groove 170A on the first back surface side is smaller. 2 shows an example in which the thickness of the dicing blade in each cutting process is selected to be gradually reduced so that the width Sb2 of the groove 170B on the back surface side 2 is reduced (Sb1> Sb3> Sb2). Also in this case, the depth Db <Dc or Db <Da can be satisfied.

  In the fourth aspect, when all of the grooves on the back surface side are formed by the dicing blade, the second back surface side is less than the taper degree of the tip of the first dicing blade that forms the groove on the first back surface side. The taper degree of the tip of the second dicing blade forming the groove is increased. That is, the curvature radius (R shape) of the tip portion of the second dicing blade is increased. As shown in the simulation result of FIG. 7, the greater the degree of tapering of the tip of the dicing blade, the smaller the stress applied to the base side of the stepped portion. Also, as shown in the simulation result of FIG. The greater the degree, the smaller the maximum stress. In the cutting with the first dicing blade, since the groove on the first back surface side does not communicate with the fine groove on the front surface side, the stress due to the tip shape of the first dicing blade is not directly applied to the step portion. For this reason, the tip of the first dicing blade may have a small degree of taper, for example, a rectangular shape as shown in FIGS. 5 (G) and 6 (C). Such a shape with a small degree of taper can be said to be a state in which the tip is not worn compared to a case where the degree of taper is large, and the dicing blade can be used for a long time. On the other hand, since the cutting with the second dicing blade communicates with the fine groove on the surface side, the stress due to the tip shape is directly applied to the step portion. For this reason, as the second dicing blade, a tip having a tapered tip is used, so that stress on the root region of the stepped portion is reduced, and a shape having a small taper degree as shown in FIG. 5G is used. As compared with the case, the breakage of the stepped portion is suppressed. For example, a tapered shape having no top surface as shown in FIG. 5B or 5D is used as the second dicing blade. In FIG. 8, the range in which the radius of curvature is 12.5 to 50 μm is the range in which the maximum stress value is saturated at a low level, and does not have a top surface as shown in FIG. 5 (B) or FIG. 5 (D). Since the taper shape is included in a range where the maximum stress value is saturated at a low level, the stepped portion is effectively prevented from being damaged. In the present embodiment, the degree of taper of dicing blades having different thicknesses refers to the size of the dicing blade when compared by enlarging or reducing the dicing blade with the same thickness without changing the shape. .

  Next, a method for preparing a dicing blade having a high degree of taper will be described. In the first preparation method, a dicing blade having a large degree of taper at the tip is obtained from another main body (other person). In the second preparation method, a dicing blade with a small taper degree is processed to increase the taper degree. In the third preparation method, a dicing blade having a small taper degree is used for a dicing blade having a large taper degree when the use period of the dicing blade exceeds a certain time.

  As a first preparation method, FIGS. 23A to 23E show an example of a dicing blade having a high degree of taper that can be obtained from another main body (other person). A dicing blade 500 illustrated in FIG. 23A includes a pair of side surfaces 510 and 520 and a pair of inclined surfaces 512 and 522 extending obliquely and linearly from the pair of side surfaces 510 and 520. A sharp apex 530 is formed at a portion where the pair of inclined surfaces 512 and 522 intersect. The inclination angle θ of the sharp apex 530 is the angle formed between the surface H orthogonal to the side surfaces 510 and 520 and the inclined surfaces 512 and 522, or the surface H parallel to the rotation axis of the dicing blade and the inclined surfaces 512 and 522. It is specified in angle. The distance between the pair of side surfaces 510 and 520 corresponds to the kerf width Sb. A dicing blade 502 shown in FIG. 23B is obtained by forming a flat surface (top surface) 532 on the pointed top portion 530 of FIG. In this case, the angle formed by the plane H parallel to the flat surface 523 and the inclined surfaces 512 and 522 is the inclination angle θ of the top (top surface). In the dicing blade 504 shown in FIG. 23C, inclined surfaces 514 and 524 extending from the pair of side surfaces 510 and 520 are curved, and a sharp top portion 534 is formed at the intersecting portion. In the dicing blade 506 shown in FIG. 23D, a linear side surface 510 and an inclined surface 522 extending linearly in an oblique direction from the other side surface 520 intersect with each other, and a sharp top portion 536 is formed there. The A dicing blade 508 shown in FIG. 23E is obtained by forming a flat surface (top surface) 532 on a sharp top 536 of the dicing blade shown in FIG.

  As a second preparation method, for example, a dicing blade having a rectangular tip suitable for full dicing as shown in FIG. 5G is obtained from another main body, and this is shown in FIGS. F is processed into a tip portion having a large taper degree as shown in F). The dicing blade is, for example, an electroformed blade in which abrasive grains such as diamond are bonded by metal plating on a side surface of a base such as aluminum, a resinoid blade in which abrasive grains such as diamond are bonded by resin bond, and abrasive grains such as diamond. It consists of metal blades baked and hardened with metal bonds. The tip shape of such a dicing blade can generally be processed by a processing apparatus using a dressing board made of abrasive grains that are harder than the dicing blade and larger than the abrasive grains of the dicing blade.

  As a third preparation method, when the dicing blade is used repeatedly, the radius of curvature of the tip shape increases, and the taper degree gradually increases. For example, FIG. 24A shows the tip shape of a dicing blade before the start of mass production. When such a dicing blade is used for a long period of time, the tip portion is worn, and the tip portion is curved as shown in FIG. Therefore, a dicing blade that is less than a certain usage time is used as the first dicing blade, and a dicing blade that is longer than the certain usage time is used as the second dicing blade.

D Modified Example of Fine Groove on Surface Side Next, a modified example of the fine groove formed on the substrate surface side will be described. The fine groove 140 shown in FIG. 2D is processed into a straight groove having a side surface extending substantially perpendicularly from the substrate surface by anisotropic dry etching, but the fine groove has a shape other than this. It may be.

  25A, 25B, 25C, and 25D show another configuration example of the fine groove of this embodiment. These groove shapes have a shape in which the lower part side of the groove is widened, and even when the top of the dicing blade varies in the groove width direction, the stepped portion has a shape that is difficult to receive stress. A fine groove 800 shown in FIG. 25A is connected to a first groove part 810 including a straight side surface forming a substantially uniform width Sa1 having a depth D1, and a lower part of the first groove part 810. And a second groove portion 820 having a spherical side surface and a bottom surface having a depth D2. The width Sa2 of the second groove portion 820 is an inner diameter between opposing side walls in a direction parallel to the substrate surface, and has a relationship of Sa2> Sa1. In the illustrated example, the width Sa2 is maximized near the center of the second groove portion 820.

  A fine groove 800A shown in FIG. 25B is connected to a first groove part 810 including a straight side surface forming a substantially uniform width Sa1 having a depth D1, and a lower part of the first groove part 810. And a rectangular second groove portion 830 having a substantially straight side surface with a depth D2. The second groove portion 830 is obtained by linearly changing the spherical side surface and the bottom surface of the second groove portion 820 shown in FIG. 25A, and the width Sa2 of the second groove portion 830 is: This is the distance between opposing side walls in a direction parallel to the substrate surface, and this distance is substantially constant (Sa2> Sa1). The shape of the second groove portion shown here is an exemplification, and the shape of the second groove portion may be a shape having a width larger than the width Sa1 of the first groove portion. An intermediate shape between the second groove portion 820 shown in FIG. 25 (A) and the second groove portion 830 shown in FIG. 25 (B), that is, the second groove portion may be elliptical. In other words, the second groove portion may have a shape having a space wider than the width of the groove at the boundary with the first groove portion (the width of the groove at the depth of D1).

  A fine groove 800B shown in FIG. 25C is connected to a first groove portion 810 having a side surface forming a substantially uniform width Sa1 having a depth D1, and a depth D2 below the first groove portion 810. And a second groove portion 840 having a reverse taper shape. The side surface of the second groove portion 840 is inclined so that the width gradually increases toward the bottom. The width Sa2 of the second groove portion 840 is a distance between the side surfaces facing each other in the horizontal direction with respect to the substrate surface, and the distance is maximum in the vicinity of the lowermost portion (near the lower end) of the second groove portion 840.

  The fine groove 800C shown in FIG. 25D has a shape in which the width gradually increases from the opening width Sa1 on the substrate surface to the width Sa2 near the bottom. That is, it is composed of an inversely tapered groove having a depth D2. The fine groove 800C is obtained by making the depth D1 of the first groove portion 810 shown in FIG. Note that the shape in FIG. 25D is not a shape in which the angle of the side surface changes at the boundary between the first groove portion and the second groove portion, as in the shapes in FIGS. 25A to 25C. However, when the upper part and the lower part of the entire groove are compared, the lower part has a wider groove width, and the first groove part (upper part) and the second groove wider than the first groove part. It has a part (lower part).

  Here, in order to suppress the remaining of the adhesive layer of the dicing tape that occurs when the dicing tape 160 is peeled off, the shape of the first groove portion is from the substrate surface to the back surface as shown in FIG. A vertical groove as shown in FIGS. 25 (A) to 25 (C) is preferable to a shape in which the width gradually increases toward () (reverse taper shape). This is because the reverse taper shape makes it difficult for UV light to hit the adhesive layer that has entered the groove, and the adhesive layer is difficult to cure. This is because stress is easily applied to the base portion of the layer, and it is easy to tear off.

  Furthermore, from the viewpoint of suppressing the remaining of the adhesive layer, the shape of the side surface of the first groove portion is wider from the substrate surface to the back surface than the vertical shape of FIGS. 25 (A) to 25 (C). It is more preferable that the shape becomes gradually narrower (forward tapered shape). That is, the shape of the first groove portion is preferably a shape that does not have a portion (reverse taper shape) whose width increases from the front surface to the back surface of the substrate.

  The fine grooves 800, 800A, 800B, and 800C shown in FIGS. 25A to 25D are preferably axisymmetric with respect to a center line orthogonal to the substrate. FIGS. 25A to 25D are drawn by straight lines or curved surfaces for easy understanding of the characteristics of the fine grooves. Steps or irregularities are formed on the side surfaces of the actually formed fine grooves. It should be noted that the corners are not necessarily strictly angular and can be formed with curved surfaces. Also, FIGS. 25A to 25D show the shape as an example of a fine groove to the bottom, and a width larger than the first width is provided below the first groove portion. Any other shape may be used as long as the second groove portion is formed. For example, the shape which combined each shape shown to FIG.25 (A)-(D), and the shape further deform | transformed after combining may be sufficient. In addition, the forward and reverse mesa shape angles shown in FIGS. 25C and 25D are merely examples, and may be inclined with respect to a plane perpendicular to the substrate surface, and the degree of inclination is not limited.

  Next, the manufacturing method of the fine groove | channel of a present Example is demonstrated. FIG. 26 is a flow showing a manufacturing method for manufacturing the fine groove of the present embodiment. In the fine groove as shown in FIGS. 25A to 25D, the first groove portion having the width Sa1 is formed by the first etching (S800), and then below the first groove portion. Forming a second groove portion having a width Sa2 wider than the width Sa1 by second etching (S810). Here, the second etching uses etching whose etching strength in the side wall direction is stronger than that of the first etching. As an example, an example in which anisotropic etching is used as the first etching and isotropic etching is used as the second etching will be described.

  FIG. 27 is a schematic cross-sectional view for explaining a manufacturing process of the fine groove 800 shown in FIG. A photoresist 900 is formed on the surface of the GaAs substrate W. The photoresist is, for example, an i-line resist having a viscosity of 100 cpi, and is applied to a thickness of about 8 μm. An opening 910 is formed in the photoresist 900 using a known photolithography process, for example, an i-line stepper and 2.38% TMAH developer. The width of the opening 910 defines the width Sa1 of the first groove portion.

A first groove portion 810 is formed on the substrate surface by anisotropic dry etching using the photoresist 900 as an etching mask. In a preferred embodiment, inductively coupled plasma (ICP) is used as a reactive ion etching (RIE) apparatus. Etching conditions are, for example, inductively coupled plasma (ICP) power of 500 W, bias power of 50 W, pressure of 3 Pa, etching gas of Cl ₂ = 150 sccm, BCl ₃ = 50 sccm, C ₄ F ₈ = 20 sccm, and etching time of 20 minutes. . As is well known, a protective film 920 is formed on the side wall simultaneously with etching by adding a CF-based gas. Radicals and ions are generated by the reactive gas plasma. The side wall of the groove is attacked only by radicals, but is not etched because of the protective film 920. On the other hand, the protective film is removed from the bottom portion by vertically incident ions, and the removed portion is etched by radicals. For this reason, anisotropic etching is achieved.

Next, isotropic etching is performed by changing the etching conditions. As an example, here, the supply of C ₄ F ₈ for forming the sidewall protective film 920 is stopped. The power of inductively coupled plasma (ICP) is 500 W, the bias power is 50 W, the pressure is 3 Pa, the etching gas is Cl ₂ = 150 sccm, BCl ₃ = 50 sccm, and the etching time is 10 minutes. Since the supply of C ₄ F ₈ is stopped, the sidewall protective film 920 is not formed, so that isotropic etching is achieved at the bottom of the first groove portion 810. As a result, a second groove portion 820 is formed below the first groove portion 810. The second groove portion 820 has spherical side surfaces and a bottom surface that further extend laterally and downward from the width Sa1 of the first groove portion 810. In addition, said etching conditions are an example and an etching condition can be suitably changed according to the width | variety, depth, shape, etc. of a fine groove | channel.

Note that the shape as shown in FIG. 25C has a weaker etching strength in the side wall direction than when the second groove portion in FIG. 25A is formed when the second groove portion is formed. That's fine. The etching strength in the direction of the side wall can be changed by changing the etching conditions such as the output of the etching apparatus and the etching gas. Specifically, for example, the supply of C ₄ F ₈ which is a gas for protecting the side wall is completely performed. Without stopping, the flow rate for forming the first groove portion may be decreased, the flow rate of Cl ₂ or the like as an etching gas may be increased, or these may be combined. In other words, in both the formation of the first groove portion and the formation of the second groove portion, both the side wall protection gas and the etching gas contained in the etching gas are supplied, but the respective flow rates are set. What is necessary is just to form by changing. Then, by setting such a flow rate in advance before forming the first groove portion, the first groove portion and the second groove portion can be formed in a series of continuous etching processes. . In addition, in order to suppress the remaining of the adhesive layer, such a shape is obtained when the first groove portion is formed in a shape (a forward taper shape) in which the width gradually decreases from the front surface to the back surface of the substrate. Thus, the flow rate of C ₄ F ₈ or Cl ₂ and the output of the etching apparatus may be optimized or the flow rate may be switched. 25D can be formed by omitting the formation of the first groove portion in FIG. 25C. Such etching is generally achieved as anisotropic etching.

  Although the manufacturing method for manufacturing the fine groove of the present embodiment has been described, the first groove portion and the second groove portion having a width wider than the first groove portion can be formed. It may be formed by a method. For example, it may be formed by a combination of dry etching and wet etching. Further, the first groove portion does not need to be formed only by the first etching, and the second groove portion does not need to be formed only by the second etching. That is, for the first groove portion, if the first etching is the main etching, etching other than the first etching may be included, and for the second groove portion, the second etching is performed. If this etching is the main etching, etching other than the second etching may be included. Further, since at least the first groove portion and the second groove portion only need to be formed, for example, between the first groove portion and the second groove portion or the back surface side of the substrate with respect to the second groove portion. The third and fourth groove portions may be present at positions close to, and they may be formed by the third etching or the fourth etching. In this specification, the width of the fine groove (groove on the front surface side) is compared with the width of the groove on the back surface side, such as “the groove on the back surface side having a width wider than the width of the groove on the front surface side”. In the expression, the width of the groove on the surface side refers to the width of the inlet portion of the groove when the width of the fine groove (groove on the surface side) is not constant as in the shapes of FIGS. .

  The preferred embodiment of the present invention has been described in detail above, but the present invention is not limited to the specific embodiment, and various modifications can be made within the scope of the present invention described in the claims. Deformation / change is possible. For example, the present invention may be applied to the case where individual elements are separated from a substrate that does not include a semiconductor such as glass or polymer. For example, you may apply to the board | substrate for MEMS which does not contain a semiconductor. In addition, as long as there is no order contradiction, each step in the embodiment of the present invention may be performed at least partially in the design stage before the mass production process, or may be performed entirely as part of the mass production process. Good. In addition, each step in the embodiment of the present invention may be performed by a plurality of subjects. For example, the first main body performs the formation of the groove on the front surface side, and the second main body supplies the substrate on which the groove on the front surface side is formed by the first main body. The second main body may form a groove on the back side to separate (divide) the substrate. That is, the first main body may prepare the substrate on which the groove on the front side is formed, or the second main body may prepare itself.

100: Light emitting element 120: Cutting region (scribe line)
130: resist pattern 140: groove on the surface side (fine groove)
160: Dicing tape 170: Back side groove 170A: First back side groove 170B: Second back side groove 170C: Third back side groove 180, 200: Ultraviolet ray 190: Expanding tape 210: Semiconductor chips 300, 300A, 302: Dicing blade 310, 320: Side surfaces 330, 332, 352, 362: Curved surface 340: Top surface 350, 360: Chamfer 400: Stepped portion 410: Root region 420: Crack

Claims (10)

Forming a groove on the surface side along the cutting region of the substrate from the surface of the substrate;
A groove on the first back surface side having a width that does not reach the groove on the front surface side along the groove on the front surface side from the back surface of the substrate and has a width wider than the width of the groove on the front surface side. Forming, and
After forming the groove on the first back surface side, the cutting member that rotates from the bottom of the groove on the first back surface side to the groove on the front surface side has a width wider than the width of the groove on the front surface side. Forming a groove on the second back surface side,
A method of manufacturing a semiconductor piece. The groove on the first back surface side has a first depth from the substrate back surface to the bottom of the groove on the first back surface side, and the groove on the second back surface side is the first back surface. 2. The semiconductor piece according to claim 1, having a second depth from a bottom of the groove on the side to a bottom of the groove on the surface side, wherein the first depth is greater than the second depth. Production method. 3. The method of manufacturing a semiconductor piece according to claim 1, wherein a width of the groove on the first back surface side is narrower than a width of the groove on the second back surface side. The groove on the first back surface side is formed by a rotating cutting member, and the taper degree of the tip portion of the cutting member forming the groove on the second back surface side forms the groove on the first back surface side. 4. The method of manufacturing a semiconductor piece according to claim 1, wherein the taper degree is larger than a taper degree of the tip of the semiconductor piece. The groove on the first back surface side is formed by a rotating cutting member, and the thickness of the cutting member forming the groove on the second back surface side is larger than the thickness of the cutting member forming the groove on the first back surface side. The method of manufacturing a semiconductor piece according to claim 1, wherein the semiconductor piece is thin. The semiconductor piece manufacturing method according to claim 4 or 5, wherein a usage time of the cutting member for forming the groove on the second back side is longer than a usage time of the cutting member for forming the groove on the first back side. Method. 7. A step of forming a third back surface side groove between the step of forming the first back surface side groove and the step of forming the second back surface side groove. The manufacturing method of the semiconductor piece as described in any one. The grooves on the first back surface side are formed by a rotating cutting member, and the width of the groove on the third back surface side is narrower than the width of the groove on the first back surface side. The width | variety of this groove | channel is a manufacturing method of the semiconductor piece of Claim 7 narrower than the width | variety of the groove | channel on the said 3rd back surface side. A circuit board on which at least one semiconductor piece manufactured by the manufacturing method according to claim 1 is mounted. An electronic device for mounting the circuit board according to claim 9.

Images