JP2007250790A - Manufacturing method of semiconductor chip - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a manufacturing method of semiconductor chips to which a normal carrying device or the like can be applied even if a divided semiconductor chip is being connected by a pressure-sensitive adhesive sheet for protecting the surface. <P>SOLUTION: A wafer-dividing method comprises processes of: forming a groove having a cutting depth that is shallower than the thickness of a wafer along a circuit at the circuit surface side of the wafer in which a plurality of circuits are formed; laminating a fixing jig 3 on the circuit surface; and a grinding the back of the wafer for dividing into a group of chips until the groove is reached. In the wafer-dividing method, the fixing jig comprises a jig substrate 30 that has a plurality of projections 36 on one surface, and a sidewall having nearly the same height as that of the projections at the outer-periphery section on one surface, and an adhesive layer 31 that is laminated on a surface having the projections of the jig substrate and is adhered to the upper surface of the sidewall. A divided space is formed on a surface having the projections of the jig substrate by the adhesive layer, the projections, and the sidewall, and at least one through hole 38 penetrating the outside and the divided section is provided in the jig substrate. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a method for manufacturing a semiconductor chip in which the wafer is thinned by grinding the back surface and finally divided into individual chips, and particularly, the semiconductor chip is excellent in handling property after back grinding. The present invention relates to a method for manufacturing a semiconductor chip.

  In recent years, IC cards have been widely used, and further reduction in thickness has been desired. For this reason, it is necessary to reduce the thickness of a semiconductor chip, which has conventionally been about 350 μm, to a thickness of 50 to 100 μm or less.

  As a method of achieving such chip thinning, there is known a method of manufacturing a semiconductor chip in which a groove having a predetermined depth is formed from the front surface side of the wafer (half-cut dicing) and then ground from the back surface side of the wafer. . Such a process is also called a “first dicing method” (Patent Document 1).

  In the process by the first dicing method, when the wafer is in an extremely thin state, it is divided into chips at the same time, and the area becomes smaller compared to the thickness. Moreover, chipping (chipping) of the chip cross section, which is likely to occur when full cut dicing is performed, is unlikely to occur with the prior dicing method.

By the way, the chips divided by the tip dicing method are connected with the adhesive sheet for surface protection, and the adhesive sheet does not have rigidity. There was a problem that the process could not be performed. In addition, when the adhesive sheet is rigid enough to support the chip, there is a problem that it is difficult to peel the adhesive sheet from the chip.
JP-A-5-335411

  An object of the present invention is to provide a method for manufacturing a semiconductor chip that can solve the problems associated with the prior art as described above, and that can be used for an ordinary transfer device or the like in the prior dicing method. Accordingly, it is also an object to realize a very thin and highly reliable chip at a relatively low cost.

The method for manufacturing a semiconductor chip according to the present invention includes (I) a step of forming a groove having a depth of cut shallower than the thickness of the wafer along the circuit on the circuit surface side of the wafer on which a plurality of circuits are formed. II) including a step of laminating a fixed jig on the circuit surface, and (III) a step of grinding the back surface of the wafer until it reaches the groove and dividing the wafer into chip groups, wherein the fixed jig has a plurality of protrusions on one side. A jig base having a side wall substantially the same height as the protrusion on the outer peripheral portion of the one surface, and a jig base layer laminated on the surface having the protrusion of the jig base, and bonding on the upper surface of the side wall A partition space is formed by the contact layer, the protrusion, and the side wall on the surface of the jig base having the protrusions, and the jig base includes an outside and the partition space. At least one through-hole penetrating through the adhesive layer Characterized in that it is laminated on the circuit surface.

In the manufacturing method, after the wafer is divided into chip groups in the step (III), a transfer tape fixed to the frame is attached to the back surface of the chip group, and then the inside of the partition space is formed from the through hole. It is preferable to remove the chip group from the adhesion layer and transfer it to the transfer tape by sucking gas to deform the adhesion layer into an uneven shape.

The adhesion layer is preferably laminated on the circuit surface of the wafer via an adhesive sheet.
In this way, by using the fixed jig having rigidity that allows the divided chips to be easily removed, it is possible to proceed to the next process without using a special transfer device after performing the first dicing. it can.

Hereinafter, the present invention will be described more specifically with reference to the drawings.
<Fixing jig>
First, the fixed jig used in the present invention will be described. The fixed jig 3 used in the present invention includes a jig base 30 and an adhesion layer 31 as shown in FIG. Examples of the shape of the jig base 30 include a substantially circular shape, a substantially elliptical shape, a substantially rectangular shape, and a substantially polygonal shape, and a substantially circular shape is preferable. As shown in FIGS. 2 and 3, a plurality of protrusions 36 are formed on one surface of the jig base 30 so as to protrude upward at intervals. Although the shape of the protrusion 36 is not particularly limited, a cylindrical shape or a truncated cone shape is preferable. Further, a side wall 35 having substantially the same height as the protrusion 36 is formed on the outer peripheral portion of the surface having the protrusion. An adhesion layer 31 is laminated on the surface having the protrusions. The adhesion layer 31 is bonded on the upper surface of the side wall 35, and the upper surface of the protrusion 36 and the adhesion layer 31 may or may not be bonded. A partition space 37 is formed by the projections 36, the side walls 35, and the adhesion layer 31 between the surface of the jig base 30 having the projections, that is, between the jig base 30 and the adhesion layer 31 (FIG. 1). On the other hand, on the surface of the jig base 30 that does not have a protrusion, a through hole 38 that penetrates the outside of the surface side and the partition space 37 is provided in the thickness direction of the jig base 30. As long as at least one through hole 38 is provided on the jig base 30, a plurality of through holes 38 may be provided. Further, instead of the through hole 38 on the surface of the jig base 30 having no protrusions, the through hole 38 may be provided in the horizontal direction of the jig base 30, and an opening may be provided in the side wall 35. By connecting a detachable vacuum device to the opening of the through hole 38, the gas in the partition space is exhausted, and the adhesion layer 31 can be deformed into an uneven shape.

The material of the jig base 30 is not particularly limited as long as it has excellent mechanical strength. For example, a thermoplastic resin such as polycarbonate, polypropylene, polyethylene, polyethylene terephthalate resin, acrylic resin, polyvinyl chloride; aluminum alloy, magnesium Examples thereof include metal materials such as alloys and stainless steel; inorganic materials such as glass; and organic-inorganic composite materials such as glass fiber reinforced epoxy resins. The bending elastic modulus of the jig base 30 is preferably 1 GPa or more. If it has such a bending elastic modulus, rigidity can be given, without making the thickness of a jig base unnecessarily thick. By using such a material, the wafer can be sufficiently supported without being bent after the backside grinding of the wafer. Further, it has a strength that opposes the shear stress caused by the rotation of the grinding wheel of the semiconductor wafer, and does not disturb the chip arrangement after the first dicing.

  The outer diameter of the jig base 30 is preferably substantially the same as the outer diameter of the semiconductor wafer or larger than the outer diameter of the semiconductor wafer. If the jig base 30 has an outer diameter that can accommodate the maximum diameter (for example, 300 mm diameter) of the standard size of the semiconductor wafer, it can be applied to all smaller semiconductor wafers. Moreover, 0.5-2.0 mm is preferable and, as for the thickness of the jig base 30, 0.5-0.8 mm is more preferable. When the thickness of the jig base is within the above range, the wafer can be sufficiently supported without being bent after the backside grinding of the wafer.

As for the height of the protrusion 36 and the side wall 35, 0.05-0.5 mm is more preferable. The diameter of the upper surface of the protrusion 36 is preferably 0.05 to 1.0 mm. Furthermore, the interval between the protrusions (distance between the centers of the protrusions) is preferably 0.2 to 2.0 mm. When the size of the protrusion 36 and the interval between the protrusions are within the above ranges, the adhesion layer 31 can be sufficiently deformed by deaeration in the partition space 37, and the semiconductor chip can be easily removed from the adhesion layer 31. Can be removed. Furthermore, even after the deformation of the unevenness of the adhesion layer 31 is repeated many times, the original flat state can be restored.

Although the diameter of the through-hole 38 is not specifically limited, 2 mm or less is preferable.
Such a jig base may be manufactured by integrally molding the bottom of the jig base, the side wall 35 and the projection 36 by, for example, thermoforming a thermoplastic resin material using a mold. The side wall 35 and the protrusion 36 may be formed on the plate, or may be manufactured, or the protrusion 36 may be formed on the inner surface of the concave portion of the concave disk. As a method for forming the protrusions 36, a method of depositing metal into a predetermined shape by electroforming, a method of forming protrusions by screen printing, a photoresist layered on a flat circular plate, exposure, development, and protrusion And a method of forming an object. Also, the jig base 30 is manufactured by a method in which the surface of a metal flat circular plate is eroded and removed by leaving a projection-forming portion by etching, or a method of removing the surface of a flat circular plate by leaving a projection-forming portion by sandblasting. You can also Note that the through hole 38 may be formed in advance before the projection is formed, or may be formed later. Moreover, you may form simultaneously with shaping | molding of a jig base.

  An adhesion layer 31 is formed on the surface of the jig base 30 having the protrusions. Examples of the material of the adhesion layer 31 include urethane-based, acrylic-based, fluorine-based, or silicone-based elastomers that are excellent in flexibility, flexibility, heat resistance, elasticity, adhesiveness, and the like. You may add various additives, such as a reinforcing filler and hydrophobic silica, to this elastomer as needed.

  The adhesion layer 31 is preferably a flat plate having substantially the same shape as the jig base 30, the outer diameter of the adhesion layer 31 is preferably substantially the same as the outer diameter of the jig base 30, and the thickness is 20 to 200 μm. Is preferred. If the thickness of the adhesion layer 31 is less than 20 μm, the mechanical durability against repeated suction may be poor. On the other hand, if the thickness of the adhesion layer 31 exceeds 200 μm, it may take a considerable time to peel off from the pressure-sensitive adhesive sheet 2 by suction, which is not preferable.

  Further, the tensile breaking strength of the adhesion layer 31 is preferably 5 MPa or more, and the tensile breaking elongation is preferably 500% or more. When the tensile rupture strength and the tensile rupture elongation are within the above ranges, the adhesive layer 31 is not broken or loosened even when the adhesive layer 31 is repeatedly deformed, and the original flat state is restored. Can do.

  Moreover, the bending elastic modulus of the adhesion layer 31 is preferably in the range of 10 to 100 MPa. When the bending elastic modulus of the adhesion layer 31 is less than 10 MPa, the adhesion layer 31 may be deflected by gravity at portions other than the contact points with the protrusions 36, and may not be able to adhere to the wafer. On the other hand, when the pressure exceeds 100 MPa, deformation due to suction becomes difficult to occur, and the wafer may not be easily peeled off.

  Moreover, it is preferable that the shear contact | adhesion power of the surface at the side which contact | connects the adhesive sheet 2 of the contact | adherence layer 31 is 35N or more. If it is less than 35 N, the wafer may be peeled off together with the adhesive sheet when a force is applied in the shearing direction of the wafer (horizontal to the wafer surface), and the wafer may be damaged. In the present invention, the shear adhesion force is a value measured between the adhesion layer 31 and the mirror surface of the silicon wafer, and the adhesion layer 31 is applied to a known glass plate having a size of 30 mm long × 30 mm wide × 3 mm thick. Attaching and placing on a mirror wafer made of silicon, when a load of 900 g is applied to the entire glass plate and adhesion layer 31 for 5 seconds, and the glass plate is pressed in parallel with the mirror wafer when it starts moving The load is measured.

Further, the adhesion strength of the adhesion layer 31 is desirably 2 N / 25 mm or less. If the value exceeds this value, the adhesion between the adhesion layer 31 and the base material of the pressure-sensitive adhesive sheet 2 becomes too large, and a blocking state may occur, and there is a possibility that peeling by suction cannot be performed. In the present invention, the adhesion strength refers to the peel strength when the adhesion layer 31 is attached to the mirror surface of the wafer and peeled off.

  For such an adhesion layer 31, for example, a film made of the elastomer is prepared in advance by a calendar method, a press method, a coating method, a printing method, or the like, and the elastomer film is bonded to the upper surface of at least the side wall 35 of the jig base 30. By doing so, the partition space 37 is formed. As a method for adhering the adhesion layer 31, a method of adhering via an adhesive made of acrylic resin, polyester resin, epoxy resin, silicone resin or elastomer resin, or heat sealing when the adhesion layer 31 is heat-sealable The method to adhere | attach is mentioned.

  The surface of the adhesion layer 31 may be subjected to non-adhesion treatment, and in particular, only the adhesion layer surface above the protrusion 36 that comes into contact with the adhesive sheet 2 when deformed into an uneven shape is subjected to non-adhesion treatment. It is preferable. When treated in this way, before the adhesion layer 31 is deformed, the adhesion layer surface is in close contact with the pressure-sensitive adhesive sheet 2 at the non-adhesion-treated portion, and the adhesion layer 31 deformed into a concavo-convex shape is the surface above the protrusion 36, that is, Since only the non-adhesive convex surface is in contact with the semiconductor chip, the semiconductor chip can be removed from the adhesion layer 31 more easily. Non-adhesive treatment methods include, for example, a method in which air in the partition space 37 is sucked by a vacuum device to deform the adhesion layer 31 into a concavo-convex shape, and the convex tip is physically roughened by a grindstone roller or the like, UV The method of processing, the method of laminating non-adhesive rubber, the method of coating non-adhesive paint, etc. are mentioned. Further, the non-adhesive portion may be formed in a cross pattern so as to pass through the center of the adhesion layer 31 instead of the convex portion. The surface roughness of the non-adhesive part is preferably an arithmetic average roughness Ra of 1.6 μm or more, and more preferably 1.6 to 12.5 μm. By roughening the non-adhesive portion with the surface roughness in the above range, the adhesion layer 31 does not deteriorate, and the semiconductor chip can be easily detached from the adhesion layer 31.

<Adhesive sheet>
In this invention, in order to protect the circuit surface of a semiconductor wafer, as shown in FIG. 10, you may adhere the adhesive sheet 2 to the circuit surface of the semiconductor wafer 1 as needed. The pressure-sensitive adhesive sheet 2 includes a base material and a pressure-sensitive adhesive layer, and more preferably has an intermediate layer between the base material and the pressure-sensitive adhesive layer.

(Base material)
If the base material used for the adhesive sheet 2 is a resin sheet, it will not be selected and can be used. As such a resin sheet, for example, polyolefin such as low density polyethylene, linear low density polyethylene, polypropylene, polybutene, ethylene vinyl acetate copolymer, ethylene (meth) acrylic acid copolymer, ethylene (meth) acrylic acid Examples thereof include ethylene copolymers such as ester copolymers, polyesters such as polyethylene terephthalate and polyethylene naphthalate, and resin films such as polyvinyl chloride, acrylic rubber, polyamide, urethane, and polyimide. The substrate may be a single layer of these or a laminate. Moreover, the sheet | seat which gave the process of bridge | crosslinking etc. may be used. Further, the substrate may be formed by curing and curing a curable resin, or may be formed by forming a thermoplastic resin.

The thickness of the substrate is preferably 30 to 1000 μm, more preferably 50 to 800 μm, and particularly preferably 80 to 500 μm.
20-60 mN / m is preferable and, as for the said base material, the surface energy of the surface contact | abutted to the contact | adherence layer 31 of the fixed jig 3 is more preferable, and 25-50 mN / m is more preferable. When the surface energy is within the above range, the adhesion with the adhesion layer 31 is optimum, and blocking and the wafer are not dropped due to insufficient adhesion.

  The substrate preferably has a surface roughness (arithmetic average roughness Ra) of 1.0 μm or less, more preferably 0.2 μm or less. When the arithmetic average roughness Ra is in the above range, there is no factor for reducing the adhesion between the adhesion layer 31 and the pressure-sensitive adhesive sheet 2, so that a stable adhesion can be obtained.

  The substrate is made of a resin film and is not particularly limited as long as the above physical properties are satisfied. Even if the resin itself exhibits the above physical properties, by adding other additives or performing surface treatment, It may be a physical property.

Furthermore, the above-mentioned resin may contain additives such as inorganic fillers such as calcium carbonate, silica and mica, metal fillers such as iron and lead, and colorants such as pigments and dyes.
The base material can be manufactured by casting a liquid resin (resin before curing, resin solution, etc.) in a thin film on a casting process sheet, and then forming the film by a predetermined means. According to such a manufacturing method, the stress applied to the resin during film formation is small, and the formation of fish eyes is small. In addition, the uniformity of the film thickness is high, and the thickness accuracy is usually within 2%. Further, as another production method, a method of producing a substrate as a single layer film by extrusion molding by a T-die or an inflation method or a calendar method can be mentioned.

(Adhesive layer)
The pressure-sensitive adhesive layer can be formed of various conventionally known pressure-sensitive pressure-sensitive adhesives. Such an adhesive is not limited at all, but an adhesive such as rubber-based, acrylic-based, silicone-based, or polyvinyl ether is used. In addition, an energy ray curable adhesive, a heat-foaming adhesive, or a water swelling adhesive can be used. In particular, in the present invention, an energy ray curable adhesive, particularly an ultraviolet curable adhesive is preferably used.

  The energy beam curable pressure-sensitive adhesive generally comprises an acrylic pressure-sensitive adhesive and an energy beam polymerizable compound as main components. As the energy beam polymerizable compound used in the energy ray curable pressure-sensitive adhesive, low molecular weight compounds having at least two photopolymerizable carbon-carbon double bonds in a molecule that can be three-dimensionally reticulated by light irradiation are widely used. Specifically, trimethylolpropane triacrylate, pentaerythritol triacrylate, pentaerythritol tetraacrylate, dipentaerythritol monohydroxypentaacrylate, dipentaerythritol hexaacrylate or 1,4-butylene glycol diacrylate, 1,6-hexane Diol diacrylate, polyethylene glycol diacrylate, oligoester acrylate, urethane acrylate and the like are used.

  The blending ratio of the acrylic pressure-sensitive adhesive and the energy beam polymerizable compound in the energy ray curable pressure sensitive adhesive is 10 to 1000 parts by weight, preferably 20 to 100 parts by weight of the energy ray polymerizable compound with respect to 100 parts by weight of the acrylic pressure sensitive adhesive. It is desirable to use 500 parts by weight, particularly preferably in an amount ranging from 50 to 200 parts by weight. In this case, the obtained pressure-sensitive adhesive sheet has a large initial adhesive strength, and the pressure-sensitive adhesive strength greatly decreases after irradiation with energy rays. Therefore, peeling at the interface between the semiconductor chip and the energy ray curable pressure-sensitive adhesive layer after the back surface grinding is facilitated.

  The energy beam curable pressure-sensitive adhesive may be formed of an energy beam curable copolymer having an energy beam polymerizable group in the side chain. Such an energy beam curable copolymer has the property of having both adhesiveness and energy beam curable properties.

By adding a photopolymerization initiator to the energy ray curable pressure-sensitive adhesive, it is possible to reduce the polymerization curing time by light irradiation and the light irradiation amount.
Examples of such photopolymerization initiators include photoinitiators such as benzoin compounds, acetophenone compounds, acylphosphine oxide compounds, titanocene compounds, thioxanthone compounds, and peroxide compounds, and photosensitizers such as amines and quinones. Specifically, 1-hydroxycyclohexyl phenyl ketone, benzoin, benzoin methyl ether, benzoin ethyl ether, benzoin isopropyl ether, benzyldiphenyl sulfide, tetramethylthiuram monosulfide, azobisisobutyronitrile, dibenzyl, diacetyl, β- Examples include crawl anthraquinone.

  The amount of the photopolymerization initiator used is preferably 0.05 to 15 parts by weight, more preferably 0.1 to 10 parts by weight, particularly preferably 0.5 to 5 parts by weight, based on 100 parts by weight of the total pressure-sensitive adhesive. Part.

  A crosslinking agent may be used to partially crosslink the polymer component in the pressure-sensitive adhesive. As this crosslinking agent, for example, an epoxy crosslinking agent, an isocyanate crosslinking agent, a methylol crosslinking agent, a chelate crosslinking agent, an aziridine crosslinking agent, and the like are used.

  The acrylic energy ray-curable pressure-sensitive adhesive as described above has a sufficient adhesive force to the semiconductor wafer before the energy ray irradiation, and the adhesive force significantly decreases after the energy ray irradiation. That is, the adhesive sheet 2 and the semiconductor wafer are brought into close contact with each other with sufficient adhesive force before the energy beam irradiation, and the surface can be protected. After the energy beam irradiation, it can be easily peeled off from the ground semiconductor chip.

Although the thickness of an adhesive layer is based also on the material, it is about 3-100 micrometers normally, Preferably it is about 10-50 micrometers.
(Middle layer)
When bumps having a large unevenness are formed on the circuit surface of the wafer, an intermediate layer is provided between the base material and the adhesive layer in the adhesive sheet 2 in order to cause the adhesive layer to follow the uneven surface of the bumps. It may be provided.

  The material of the intermediate layer is not particularly limited as long as the above physical properties are satisfied. For example, various pressure-sensitive adhesive compositions such as acrylic, rubber-based, and silicone-based materials, ultraviolet curable resins, and thermoplastic elastomers are used.

  The total thickness of the intermediate layer and the pressure-sensitive adhesive layer is appropriately selected in consideration of the bump height of the adherend to which the pressure-sensitive adhesive sheet 2 is adhered, the bump shape, the pitch of the bump interval, and the like. The total thickness of the intermediate layer and the pressure-sensitive adhesive layer is desirably selected to be 50% or more, preferably 100 to 200% of the bump height. When the total thickness of the intermediate layer and the pressure-sensitive adhesive layer is selected in this way, the pressure-sensitive adhesive sheet 2 follows the unevenness on the circuit surface, and the unevenness difference can be eliminated. For this reason, the thickness of the intermediate layer is also in the range of 5 to 500 μm, for example, depending on the unevenness of the circuit surface.

(Method for producing adhesive sheet)
The pressure-sensitive adhesive sheet 2 is formed by coating the pressure-sensitive adhesive with a suitable thickness on a substrate according to a generally known method such as a knife coater, roll coater, gravure coater, die coater, reverse coater and the like, and drying the pressure-sensitive adhesive layer. It is obtained by forming and then bonding a release sheet on the pressure-sensitive adhesive layer as necessary. On the contrary, after forming the pressure-sensitive adhesive layer on the release-treated surface of the release sheet, the pressure-sensitive adhesive sheet 2 may be obtained by pasting with a base material.

  When forming the intermediate layer, after applying the resin for forming the intermediate layer on the substrate, the resin is dried or cured by a required means to form the intermediate layer, and the intermediate layer is adhered to the intermediate layer by the above method. By forming the agent layer, the pressure-sensitive adhesive sheet 2 with an intermediate layer is obtained. Also, an intermediate layer is formed on the release surface of the release sheet, transferred to the substrate, and the adhesive layer formed on another release sheet is bonded to the surface of the intermediate layer from which the release sheet has been peeled, with an intermediate layer The pressure-sensitive adhesive sheet 2 may be obtained.

<Transfer tape>
The semiconductor chip 13 is transferred to the transfer tape 5 so that the semiconductor chip 13 separated on the fixing jig 3 or the adhesive sheet 2 can be picked up. This transfer tape 5 is composed of a base material and an adhesive layer in the same manner as the adhesive sheet 2, and a releasable adhesive sheet that exhibits a peeling force enough to pick up the transferred chip 13, and the adhesive force is reduced by irradiation with energy rays. Alternatively, an energy ray-curable pressure-sensitive adhesive sheet that can disappear is preferably used. Such a transfer tape 5 can be the same as a so-called commercially available dicing sheet.

<Semiconductor chip manufacturing method>
In the semiconductor chip manufacturing method according to the present invention, first, a groove 11 having a cutting depth shallower than the thickness of the wafer 1 is formed along the circuit on the circuit surface side of the wafer 1 on which a plurality of circuits are formed ( FIG. 4). The groove 11 is formed so as to partition a plurality of circuits formed on the surface of the wafer 1. The depth of the groove 11 is not particularly limited as long as it is a little deeper than the thickness of the target chip. The grooves 11 are formed using a dicing blade of a dicing apparatus.

  Next, the fixing jig 3 is attached to the surface on which the groove 11 is formed so that the surface and the adhesion layer 31 are in contact with each other (FIG. 5). Thereafter, the back surface (grinding surface) 12 of the wafer 1 is ground using a grinder to reduce the thickness of the wafer, and finally, the wafer 1 is divided into individual chips 13. That is, back grinding is performed until the bottom of the groove 11 is removed, and the wafer is chipped for each circuit (FIG. 6). Furthermore, a plurality of chips having a predetermined thickness (hereinafter also referred to as “chip group”) can be obtained by performing back surface grinding as necessary. During this grinding process, the jig base 30 is sucked and fixed to the processing table. However, the processing table is fixed to the through hole 38 by closing the suction surface of the processing table that contacts the through hole 38 of the jig base 30. The gas in the partition space 37 cannot be sucked through. Since the fixed jig becomes an extremely rigid support, the wafer 1 can be ground to an extremely thin thickness of, for example, 100 μm or less, particularly 50 μm or less.

  The method for picking up the obtained chip group is not particularly limited. For example, the following method can be preferably employed. The pick-up transfer tape 5 is attached to the ground surface (back surface) while keeping the divided state of the divided chips 13 (FIG. 7). The transfer tape 5 has a larger area than the chip group, and its periphery is fixed by a frame 51.

  Next, the vacuum device 4 sucks the gas in the partition space 37 from the through-hole 38 of the fixed jig 3 to deform the adhesion layer 31 into an uneven shape, and the chip 13 is formed only on the upper surface of the side wall 35 and the upper surface of the convex portion of the adhesion layer 31. Then, it is brought into contact with the fixed jig 3 (FIG. 8). Thereby, the contact | adhesion power of the chip | tip 13 and the fixed jig | tool 3 falls, and it can remove from the fixed jig | tool 3 easily, and the chip | tip 13 is transcribe | transferred to the transfer tape 5 (FIG. 9).

  The semiconductor chip 13 thus transferred to the transfer tape 5 is picked up from the transfer tape 5 by a conventionally known method, and a semiconductor device is manufactured through a normal process. When the transfer tape 5 is formed of an energy ray curable adhesive, the semiconductor chip 13 is transferred to the transfer tape 5 by irradiating energy rays from the base side of the transfer tape 5 to reduce the adhesive force of the transfer tape 5. Can be easily picked up from.

  In addition, when the unevenness of the circuit surface is large and the wafer 1 cannot be firmly fixed by the adhesion layer 31, the adhesive sheet 2 is attached to the circuit surface on which the groove 11 is formed (FIG. 10), and the base material surface of the adhesive sheet 2 The fixing jig 3 is attached to (Fig. 11). When bumps such as electrodes are formed on the circuit surface and the size of the unevenness is larger, the pressure-sensitive adhesive sheet 2 having an intermediate layer is used.

Thereafter, the wafer 1 fixed to the fixing jig 3 via the adhesive sheet 2 is also subjected to back surface grinding in the same manner as described above, and the separated chips are transferred to the transfer tape 5 (FIGS. 12 to 15). ). Next, the adhesive sheet 2 is removed from the semiconductor chip 13. When the pressure-sensitive adhesive layer is formed of an energy ray-curable pressure-sensitive adhesive, the pressure-sensitive adhesive sheet 2 is removed from the chip 13 by irradiating energy rays from the substrate side of the pressure-sensitive adhesive sheet 2 to reduce the adhesive force of the pressure-sensitive adhesive layer. It can be easily peeled off.

  Further, instead of the above process, after the fixing jig 3 is removed from the base material surface of the adhesive sheet 2 before the transfer tape 5 is attached to the chip group, the chip group is transferred to the transfer tape 5, and then the adhesive tape is adhered. The order in which the sheets 2 are peeled may be used.

[Example]
EXAMPLES Hereinafter, although an Example demonstrates this invention, this invention is not limited at all by this Example. Moreover, it evaluated by the following method.

  (Bending elastic modulus) The bending elastic modulus of the jig base was measured by a bending test method (three-point bending test method) defined in JIS K6911. Further, the flexural modulus of the adhesion layer was measured according to ASTM D747-70.

(Tensile strength at break) Measured in accordance with JIS K7127 at a specimen type 2 and a tensile speed of 200 mm / min.
(Tensile rupture stress) Measured according to JIS K7127 at test piece type 2 and a tensile speed of 200 mm / min.

  (Shearing adhesion) Adhesion layer is pasted on a well-known glass plate having a size of 30 mm in length × 30 mm in width × 3 mm in thickness, and placed on a mirror wafer made of silicon, and 900 g on the whole of the glass plate and the adhesion layer. Was applied for 5 seconds, and when the glass plate was pressed in parallel with the mirror wafer, the load when it started to move was measured. When the load did not start at 35N, the result was “over 35N”, and no measurements were made above this.

  (Adhesion force) Adhesion force is obtained by cutting a single film constituting the adhesion layer 31 into a width of 30 mm, adhering it to a mirror surface of a wafer with a rubber roller, and leaving it for 20 minutes, then at a speed of 300 mm / min and an angle of 180 °. The peel strength when peeled was evaluated.

  (Surface roughness) Based on JIS B0601-2001, arithmetic average roughness Ra was measured with a surface roughness meter (trade name SURFPACK SV-3000, manufactured by Mitutoyo Corporation).

(Create fixed jig)
A 0.7 mm thick polycarbonate sheet (flexural modulus 2.3 GPa) is cut into a circle with a diameter of 202 mm, and protrusions with a height of 0.1 mm and a diameter of 0.2 mm are arranged on one side with a pitch of 1.0 mm. The outer peripheral portion was molded by a hot press method so as to be a side wall having a width of 1.0 mm and a height of 0.1 mm. Further, a through hole having a diameter of 1 mm was provided in a portion having no projection at a position 1 cm inside from the side wall using a drilling machine, and a jig base was prepared. Also, as an adhesion layer, an ethylene methyl methacrylate resin having a thickness of 100 μm, a tensile breaking strength of 9 MPa, a tensile breaking elongation of 750%, a flexural modulus of 27 MPa, a shear adhesion strength of more than 35 N, and an adhesion strength of less than 0.1 N / 25 mm (less than the lower limit of measurement). A film made of Sumitomo Chemical Co., Ltd. (trade name: ACRIFTH WH303) was bonded to the side wall of the jig base and the upper surface of the protrusion with a modified silicone adhesive to prepare a fixed jig having a diameter of 202 mm.

(Wafer tip dicing process)
Half cut using a dicing machine (DFD-6561, manufactured by DISCO Corporation) so that the groove depth is 120 μm with a chip size of 10 mm × 10 mm on the mirror surface side of a silicon wafer (200 mm diameter, 750 μm thickness) Dicing was performed. Subsequently, the adhesion layer surface of the fixed jig was bonded to the mirror surface of the wafer by a vacuum bonding apparatus (manufactured by Shibaura Mechatronics) so that the respective centers coincided, and the wafer was fixed to the fixed jig.

  This is supported and fixed by suction to a processing table of a wafer grinding apparatus (trade name DFG-840, manufactured by DISCO Corporation) that does not have a delivery mechanism for dicing, and backside grinding is performed until the thickness of the wafer reaches 100 μm. The wafer was separated into pieces. In addition, the suction surface of the processing table located in the through hole of the fixing jig was partially blocked with an adhesive tape so that the gas in the partitioning space of the fixing jig was not sucked by suction when the processing table was fixed.

  Next, a transfer tape (Lintec UV curable dicing tape, trade name Adwill D650) is applied to the ground surface (chip surface) of the wafer by a wafer mounter with a peeling mechanism (Adwill RAD-2500F / 8MUL, manufactured by Lintec). Affixed to the ring frame. Subsequently, a nozzle of a vacuum pump was connected to the opening of the through hole of the fixed jig, and the gas in the partition space of the fixed jig was sucked to deform the adhesion layer. As a result, the adhesive layer of the fixed jig easily peeled from the chip group, and the chip group was transferred from the fixed jig to the transfer tape.

  As described above, by using the fixed jig, the first dicing process can be performed without causing breakage of the chips or disorder of the arrangement. Furthermore, the pre-dicing process could be performed using a wafer grinding apparatus that is not equipped with a special delivery device. In addition, during these steps, the silicon wafer (chip group) could be delivered without damaging it.

  Instead of the silicon wafer of Example 1, a wafer having simulated high bumps in which ink dots (height 100 μm, diameter 100 to 200 μm, pitch 1 mm) are formed on the mirror surface of a silicon wafer having a diameter of 200 mm and a thickness of 750 μm. Using. Further, as a pressure-sensitive adhesive sheet having an intermediate layer, a UV-curable protective pressure-sensitive adhesive sheet (trade name Adwill E8310LS342F, intermediate layer thickness 110 μm, pressure-sensitive adhesive layer thickness 40 μm) manufactured by Lintec Corporation was used. Further, the same devices and materials as those in Example 1 were used unless otherwise specified.

  Half-cut dicing was performed on the bump side of the wafer having bumps using a dicing apparatus so that the chip size was 10 mm × 10 mm and the groove depth was 120 μm. Subsequently, an adhesive sheet was attached to the bump surface using a tape laminator (Adwill RAD3500 / m12 manufactured by Lintec Corporation), and the adhesive sheet was cut along the contour of the silicon wafer. Furthermore, the adhesion layer surface of the fixing jig was attached to the base material surface of the pressure-sensitive adhesive sheet with a vacuum bonding apparatus so that the respective centers coincided, and the wafer was fixed to the fixing jig.

  This was supported and fixed by suction on a processing table of a wafer grinding apparatus, and backside grinding was performed until the wafer thickness became 100 μm, thereby dividing the wafer into individual pieces. In addition, the suction surface of the processing table located in the through hole of the fixing jig was partially blocked with an adhesive tape so that the gas in the partitioning space of the fixing jig was not sucked by suction when the processing table was fixed.

Next, the transfer tape was affixed to the ground surface (chip surface) of the wafer by a wafer mounter and fixed to the ring frame. Subsequently, a nozzle of a vacuum pump was connected to the opening of the through hole of the fixed jig, and the gas in the partition space of the fixed jig was sucked to deform the adhesion layer. As a result, the adhesive layer of the fixed jig easily peeled from the chip group, and the chip group with the adhesive sheet was transferred from the fixed jig to the transfer tape. Furthermore, the adhesive sheet is peeled off using the wafer mounter peeling mechanism,
Only the chip group was attached to the transfer tape.

  As described above, by using the fixed jig, the first dicing process can be performed without causing breakage of the chips or disorder of the arrangement. Furthermore, the pre-dicing process could be performed using a wafer grinding apparatus that does not have a special delivery device. In addition, during these steps, the silicon wafer (chip group) could be delivered without damaging it.

FIG. 1 shows a schematic sectional view of an example of a fixed jig used in the present invention. FIG. 2 shows a schematic top view of the jig base constituting the fixed jig used in the present invention. FIG. 3 shows a schematic cross-sectional view of a jig base constituting the fixed jig used in the present invention. FIG. 4 shows one step of the method of manufacturing a semiconductor chip according to the present invention. FIG. 5 shows one step of the method of manufacturing a semiconductor chip according to the present invention. FIG. 6 shows one step of the method of manufacturing a semiconductor chip according to the present invention. FIG. 7 shows one step of the method of manufacturing a semiconductor chip according to the present invention. FIG. 8 shows one step of the method of manufacturing a semiconductor chip according to the present invention. FIG. 9 shows one step of the method of manufacturing a semiconductor chip according to the present invention. FIG. 10 shows one step of the method of manufacturing a semiconductor chip according to the present invention. FIG. 11 shows one step of the method of manufacturing a semiconductor chip according to the present invention. FIG. 12 shows one step of the method of manufacturing a semiconductor chip according to the present invention. FIG. 13 shows one step of the method of manufacturing a semiconductor chip according to the present invention. FIG. 14 shows one step of the method of manufacturing a semiconductor chip according to the present invention. FIG. 15 shows a step of the method of manufacturing a semiconductor chip according to the present invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... Semiconductor wafer 11 ... Groove 12 ... Grinding surface 13 ... Semiconductor chip 2 ... Adhesive sheet 3 ... Fixing jig 30 ... Jig base 31 ... Adhesion layer 35- .... Side wall 36 ... Projection 37 ... Partition space 38 ... Through hole 4 ... Vacuum device 5 ... Transfer tape 51 ... Frame

Claims (3)

(I) forming a groove having a depth of cut shallower than the thickness of the wafer along the circuit on the circuit surface side of the wafer on which a plurality of circuits are formed, and (II) laminating a fixing jig on the circuit surface. And (III) grinding the back surface of the wafer until it reaches the groove to divide the wafer into chips.
The fixture jig has a plurality of protrusions on one side and a jig base having a side wall substantially the same height as the protrusions on the outer periphery of the one surface, and a surface having the protrusions of the jig base. And an adhesion layer laminated on the upper surface of the side wall,
On the surface having the protrusions of the jig base, a partition space is formed by the adhesion layer, the protrusions, and the side walls,
The jig base is provided with at least one through-hole penetrating the outside and the partition space,
A method of manufacturing a semiconductor chip, wherein the adhesion layer is laminated on the circuit surface. After dividing the wafer into chip groups in the step (III), a transfer tape fixed to the frame is attached to the back surface of the chip group,
Then, the chip group is removed from the adhesion layer and transferred to the transfer tape by sucking the gas in the partition space from the through hole and deforming the adhesion layer into an uneven shape. A method for producing a semiconductor chip as described in 1. above.   3. The method of manufacturing a semiconductor chip according to claim 1, wherein the adhesion layer is laminated on a circuit surface of a wafer via an adhesive sheet.

Images