JP2008166445A - Cutting method of semiconductor substrate - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a cutting method which does not has bad influence on a flat collet as a pickup apparatus even when debris is produced in the laser dicing of a semiconductor substrate. <P>SOLUTION: In the cutting method of the semiconductor substrate for cutting the semiconductor substrate into chips by irradiating it with a laser beam along the center line in the width direction of the cutting portion on the cutting schedule line, a recess is formed within the cutting schedule line, and cutting by the laser beam is carried out along the center line in the width direction of the recess. At that time, the depth of the recess is made so that the debris does not project above the substrate surface. Further, the depth of the recess from the substrate surface is made to be ≥3.5 μm. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a method of dividing and cutting a semiconductor substrate with a laser, and in particular, a processing by-product (hereinafter referred to as debris) generated during laser dicing adheres to a collet of a pickup device that sucks and conveys a semiconductor chip. The present invention relates to a method for cutting a semiconductor substrate to prevent.

In the manufacturing process of a semiconductor device, a semiconductor substrate (wafer) is divided into a large number of grid-like sections, and circuits (functional elements) such as ICs and LSIs are formed in each section, and then a planned cutting line (street) between the sections. Each semiconductor chip is manufactured by cutting (dicing) along the line.
Conventionally, blade dicing has been most commonly used for dicing semiconductor substrates. Blade dicing is a process in which a disk-shaped cutting blade rotating at high speed is pressed against a substrate held by a chuck for cutting. In many cases, a diamond blade is bonded to the outer periphery of the disk.

However, blade dicing has the following problems. First, since the substrate is a thin and brittle material, chipping or cracks occur on the cut surface, affecting the circuit area. In addition, waste generated thereby may affect the circuit area. Furthermore, since the cutting blade has a large width and chipping, it is necessary to increase the width of the line to be cut, and there is a problem that the ratio of the area that can be effectively used as a circuit region is reduced.
In general, when dicing, the substrate is affixed with dicing tape to prevent breakage and scattering of fragments, but in recent years, there has been a strong demand for thinner semiconductor devices, and the thickness is less than 100 μm. In the thin substrate as described above, the occurrence of chipping and cracking at the time of cutting becomes remarkable, and there is a problem that it is difficult to apply blade dicing.

For this reason, in recent years, a laser dicing technique has been frequently used. Laser dicing is performed by irradiating a laser beam having a wavelength that is well absorbed by the substrate and melting or evaporating the cut portion by the thermal energy of the laser. As processing lasers, CO ₂ laser, YAG laser, Ar laser, excimer laser, and the like are known.

Laser dicing has the advantage that there is no chipping and the width of the cutting region can be reduced, but the production of processing by-products called debris is a problem. This is a phenomenon in which the molten substrate material rises and adheres to the edge of the cut surface.
Such debris may be removed by providing an additional process for removing the debris after the cutting process. Examples of the additional process include a wet cleaning process and a dry plasma processing process.

An attempt to suppress the occurrence of debris in laser dicing is also disclosed. For example, Patent Document 1 below proposes a method of avoiding the influence of heat and debris on the circuit portion on the front surface side by irradiating a laser from the back surface of the substrate. This method is characterized in that it includes means for detecting a cut line (street) on the front surface side from the back surface of the substrate, and laser dicing is performed from the back surface side based on information of this detection means.
Patent Document 2 discloses laser dicing that suppresses the generation of debris by irradiating an ultrashort pulse laser of 1 picosecond or less. Patent Document 3 discloses two laser dicing devices having different oscillation wavelengths and focal diameters. A method is disclosed in which damage to the wafer is suppressed by performing laser dicing in conjunction with this laser.

Japanese Patent Laid-Open No. 2004-22936 JP 2002-324768 A JP-T-2004-526335

  The diced semiconductor chip is carried out of the dicing apparatus by the pickup means. As such pickup means, a flat collet is usually used. A flat collet is a cylindrical body with a bottom, and picks up by vacuum-sucking the upper surface of a semiconductor chip.

  The diced semiconductor chip is adsorbed and conveyed by the flat collet. At this time, debris adheres to the bottom surface of the cylindrical body of the flat collet. If the next chip is picked up in this state, the circuit portion of the semiconductor chip may be damaged by the irregularities on the bottom surface of the cylinder, which is not preferable.

Among the debris countermeasures described above, the method of removing debris by an additional process cannot be completely prevented from adhering to the pickup device, and the process is complicated because an additional process is required. There is.
Of the conventional debris suppression techniques described above, the method of Patent Document 1 is not preferable because it requires a special means of street detection from the back side of the substrate and the dicing process itself becomes complicated. Further, the methods of Patent Documents 2 and 3 are not preferable because a laser device used for dicing becomes complicated and expensive.

  Therefore, it is most convenient if there is a means that does not adversely affect the pickup device even if debris is generated by performing normal laser dicing without using a special laser device or the like.

  Accordingly, an object of the present invention is to provide a method for dividing and cutting a semiconductor substrate without fear of adversely affecting a flat collet as a pickup device even if debris occurs in laser dicing of the semiconductor substrate.

  A semiconductor substrate cutting method according to the present invention for solving the above-described problem is a semiconductor substrate cutting method in which a laser beam is irradiated along a center line in the width direction of a cutting portion of a line to be cut to cut the semiconductor substrate into a chip. In the method of cutting a semiconductor substrate, a recess is formed in the planned cutting line, and the semiconductor substrate is cut by a laser beam along a center line in the width direction of the recess.

  In the above method, the depth of the concave portion is preferably a depth at which debris does not come out above the substrate surface.

  Moreover, it is preferable that the value which pulled the width | variety of the recessed part from the width | variety between the center lines of the said cutting planned line is a larger value than the internal diameter of the collet which adsorb | sucks and conveys a chip | tip.

  In the method of the present invention, the width of the recess is preferably 12 μm or more from the center of the width of the cutting portion of the cutting line, and the depth of the recess from the substrate surface is 3.5 μm or more. Is preferred.

  In the method of the present invention, the recess can be formed by cutting with a blade or dry or wet etching.

  The semiconductor chip of the present invention is a semiconductor chip cut by a laser beam, and the semiconductor chip is provided with a cutting remaining area having a step downward from the chip surface around the periphery, and is generated by laser beam irradiation. The semiconductor chip is characterized in that a by-product (hereinafter referred to as debris) adheres to the uncut region without protruding upward from the chip surface. Here, the remaining cutting region is a stepped region around the chip after the semiconductor substrate is cut by a laser beam along the center line in the width direction of the recess formed in the planned cutting line.

  In this semiconductor chip, the step is preferably 3.5 μm or more, and the width of the remaining cutting region is preferably 2 μm or more.

  According to the present invention, it is possible to provide a method for dividing and cutting a semiconductor substrate without fear of adversely affecting a flat collet as a pickup device even if debris occurs in laser dicing of the semiconductor substrate. Therefore, it is not necessary to suppress the occurrence of debris itself, and it is not necessary to use a special laser device or perform laser dicing by a complicated process.

  FIG. 1 is a diagram for explaining the procedure and effects of the cutting method of the present invention. First, as shown in FIGS. 1A and 1B, a recess 3 having a substantially flat bottom is formed in a planned cutting line of a substrate 1 having a dicing tape 2 attached to the back surface. The width W of the concave portion is 4 μm or more larger than the maximum value of the width w of the cut portion 4 at the time of laser dicing, and the average depth d of the concave portion 3 is equal to or larger than the maximum height of the debris 5 generated at the time of laser dicing. And The formation of the recess is preferably performed (in a scribing process) prior to circuit formation on the substrate.

  According to the knowledge of the present inventors, for example, the maximum value of the cut width w at the time of laser dicing by the third harmonic of the YAG laser is about 20 μm. Therefore, the width W of the recess 3 may be 24 μm or more. Further, the maximum height of debris generated during the laser dicing is about 3.5 μm. Therefore, the depth d of the recess 3 may be 3.5 μm or more.

  As a method of forming the recess 3, it is preferable to use a dry or wet etching method. Further, cutting with a blade may be performed except when the substrate is extremely thin. However, laser processing is not preferable because debris may be formed at this stage.

  In this way, after the device is formed on the substrate on which the concave portion having a flat bottom is formed in advance, the laser beam 6 is formed along the center line in the width direction of the concave portion 3 as shown in FIG. Dicing with

  As a method for preventing the meandering of the laser beam, for example, a CCD camera installed at a dicer processing point may be used to confirm the pattern of the recess 3, adjust θ, and perform alignment such as cut line confirmation.

  In the present invention, it is not necessary to specifically limit the processing conditions of the laser used for dicing. For example, when a commonly used YAG third harmonic laser is used, the debris 5 is 3. The height is not more than 5 μm, and the debris 5 does not come out above the surface of the substrate 1. Further, if the width of the concave portion 3 is 24 μm or more, there is a margin of about 2 μm or more on one side from the edge of the cutting portion 4, and even if debris protrudes laterally along the laser cutting line, the width is 2 μm or less. Therefore, there is no possibility that the debris 4 reaches the edge of the recess 3.

  FIG. 2 is an explanatory view of a situation where a semiconductor substrate (chip) cut by the method of the present invention is sucked and conveyed by a flat collet. Since the depth of the concave portion 3 formed around the chip 7 is larger than the height of the debris 4, there is no possibility that the debris adheres to the lower surface when picking up by the flat collet 8.

However, in the present invention, the value of W must be determined so that the convex portion of the square chip does not enter the hollow portion of the cylindrical flat collet 8. For this purpose, when the collet 8 is circular, the diagonal length 2 ^1/2 (L−W + w) of the convex portion of the chip 7 must be larger than the inner diameter R of the collet 8. Further, when the collet 8 is rectangular, L−W + w must be larger than the side length R of the inner cylinder (hollow part) of the collet 8. If this condition is not satisfied, the top surface of the chip 7 enters the hollow portion of the flat collet 8, and the debris 4 may come into contact with the bottom surface of the collet 8.

  FIG. 3 is a flow chart of a semiconductor chip manufacturing process when the recess 3 is formed by a dry etching method, and shows a case where a chip having a PN junction is manufactured. First, an oxide film is formed on the wafer surface (S1). Next, the PN junction is opened by photolithography (also called photolithography) (S2) and oxide film etching (S3), and a PN junction is formed by diffusion / implantation (also called implantation) (S4).

  Next, the concave portion 3 which is a process unique to the present invention is formed. A portion other than the recess (scribe region) is masked by photolithography (S5), and a recess having a predetermined shape is formed in the scribe region by silicon etching (S6).

S5 and S6 will be described in more detail with reference to the examples. In the embodiment, a novolac positive resist agent is applied to a Si substrate to a thickness of about 2 μm by a spin coating method, a mask pattern is baked with a reduction projection exposure apparatus using a mercury lamp as a light source, and developed with a TMAH developer. Then, a dry etch mask pattern was formed. By etching the entire surface of the substrate on which this pattern was formed using, for example, CF ₄ gas, it was possible to form a flat recess at the bottom.

  Next, processing for attaching electrodes to the circuit portion of the chip is performed (S7 to S10). First, an oxide film for masking the electrode part is formed (S7), the electrode part is opened by photolithography (S8) and oxide film etching (S9), and a surface electrode is formed by vapor deposition or sputtering (S10).

  Next, the substrate is thinned (S11 to S12). The back surface of the substrate is ground by back grinding (S11), and the processing strain due to BG is removed by back surface etching (S12). Further, a back electrode is then formed by vapor deposition or sputtering (S13). Now that the processing of the circuit portion of the chip is completed, dicing is performed.

First, a dicing tape, for example, a UV curable tape is attached to the entire back surface of the substrate (S14). It is not limited to this as long as the chip cut by the laser is not scattered and can be easily peeled off after cutting.
Thereafter, as the surface protective film for applying the surface protective film to the entire surface of the substrate, for example, PVA (polyvinyl alcohol) is used. Thereafter, laser dicing is performed (S15).

  In the examples, laser dicing was performed using a YAG third harmonic laser. Note that it is desirable to scan the laser beam along the center line of the recess 3. In this embodiment, the chip pattern is stored in the laser dicing apparatus so that it is automatically aligned with the center line of the cutting area, thereby being aligned with the center line of the cutting area, and there is no horizontal inclination. After confirming this, cutting was started.

  Thereafter, in the post-cleaning step, the surface protective film and the debris attached to the surface are cleaned and removed (S16). This cleaning was performed, for example, by spraying cleaning water (pure water) onto the substrate placed on the rotary stage.

When manufacturing the semiconductor chip in the process shown in FIG. 3, after forming the concave portion by dry etching, laser dicing was performed, and the cross-sectional shape of the cut portion before and after dicing was investigated.
First, a resist mask was formed on the wafer surface by photolithography to form a recess in the scribe area (scheduled cutting area). A novolac positive resist agent (for example, a novolak positive resist agent manufactured by Tokyo Ohka Kogyo Co., Ltd.) is dropped onto the wafer by static dropping on the Si substrate, and is applied to a thickness of about 2 μm by a spin coating method in which the wafer is rotated at a rotational speed of 3000 rpm. Bake processing was performed at about 100 ° C.

  Next, using a reduction projection exposure apparatus (for example, Canon projection exposure apparatus FPA5500iX) using an ultrahigh pressure mercury lamp (wavelength 365 nm) as a light source, a mask pattern of a design circuit drawn in chrome on quartz glass was baked. The exposed resist agent was developed with a tetramethyl ammonium hydroxide (TMAH) developer to form a dry etch mask pattern.

Thereafter, a reactive ion etching type dry etching apparatus (for example, a dry etcher E620 manufactured by Matsushita Electric Industrial Co., Ltd.) is used to form a recess. The etching section of this apparatus is composed of a substrate holder disposed in the etching chamber, a high-frequency coil wound around the etching chamber, a gas introduction pipe, and the like. A mixed gas of chlorofluorocarbon (C _n F _{n + 2} ) and oxygen gas is introduced from a gas introduction pipe, and a high frequency of, for example, 13.56 MHz is applied to the high frequency coil to excite the mixed gas. The substrate is placed on a substrate holder at some distance from the plasma. In this example, CF ₄ gas was used, and the etching process was performed for 12 minutes with the target of an etching amount of about 5.2 μm.

  Thereafter, dicing tape (UV curable tape) was applied to the entire back surface of the substrate, and laser dicing was performed. Laser dicing was performed using a YAG third harmonic laser. Note that the chip pattern is stored in advance in the laser dicing apparatus so as to scan the laser beam along the center line of the recess, so that it is automatically aligned with the center line of the cutting region.

  In FIG. 4, the example of the microscope picture of the board | substrate cutting part cross-sectional shape after a recessed part formation (before dicing) and a dicing is shown. FIG. 5 shows a sketch for explaining the micrograph. In either case, the left side is a cross section before dicing, and the right side is a cross section after dicing. As can be seen from these drawings, it is known that a recess having a uniform depth and a flat bottom is formed by the dry etching. In addition, debris is generated on both edges of the cut portion after dicing, but the height of the debris is much lower than the depth of the recess 3, and there is no concern that the debris will adhere to the bottom surface of the collet when picked up by the collet. I know that.

  When laser dicing was actually performed by the method of the present invention, after the collet was continuously used for a long period of time, the adhesion state of debris on the bottom surface was investigated, but almost no deposits were observed. In addition, there was no case where the semiconductor chip became defective due to the deposit on the collet.

It is a figure for demonstrating the procedure and effect of the cutting method of this invention. It is explanatory drawing of the condition where the semiconductor chip cut | disconnected by the method of this invention is adsorbed and conveyed with a flat collet. It is a flowchart which shows the example of the semiconductor chip manufacturing process in the case of forming a recessed part by the dry etching method. It is a microscope picture which shows the example of the cross-sectional shape of the board | substrate cutting part before dicing and after dicing. It is a sketch figure of the photograph of FIG.

Explanation of symbols

1 Substrate 2 Dicing tape 3 Recess 4 Cutting section 5 Debris 6 Laser beam 7 Chip 8 Flat collet

Claims (9)

In the method for cutting a semiconductor substrate, a chip is formed by irradiating a laser beam along the center line in the width direction of the cut part of the line to be cut to cut the semiconductor substrate.
Forming a recess in the line to be cut;
A method for cutting a semiconductor substrate, comprising cutting with a laser beam along a center line in a width direction of the recess.   2. The semiconductor substrate according to claim 1, wherein a depth of the concave portion is a depth at which a processing by-product (hereinafter referred to as debris) generated by laser beam irradiation does not come out from the substrate surface. Cutting method.   3. The width of the concave portion is a value obtained by subtracting the width of the concave portion from the width between the center lines of the scheduled cutting lines, and a value larger than the inner diameter of the collet that sucks and conveys the chip. A method for cutting a semiconductor substrate as described in 1.   4. The method for cutting a semiconductor substrate according to claim 1, wherein the width of the concave portion is 12 μm or more from the center of the width of the cutting portion of the cutting line. 5.   The semiconductor substrate cutting method according to claim 1, wherein a depth of the recess from the substrate surface is 3.5 μm or more.   6. The method of cutting a semiconductor substrate according to claim 1, wherein the recess is formed by cutting with a blade or dry or wet etching. A semiconductor chip cut by a laser beam,
The semiconductor chip comprises a cutting remaining region with a step downward from the chip surface around the periphery,
A semiconductor chip characterized in that a processing by-product (hereinafter referred to as debris) generated by laser beam irradiation adheres to the uncut region without protruding above the chip surface.   The semiconductor chip according to claim 7, wherein the step is 3.5 μm or more.   9. The semiconductor chip according to claim 7, wherein a width of the uncut region is 2 μm or more.

Images