JP2014138042A - Semiconductor device, and method of manufacturing the same - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a technology capable of forming a plurality of desired dimples in a short cycle time.SOLUTION: A method of manufacturing a semiconductor device includes following steps of: (a) emitting laser light 21a from a laser oscillator 21; (b) branching the laser light 21a into a plurality of optical paths axial symmetrical about an optical axis of the laser light 21a, by using a beam splitter 22; and (c) guiding the laser light 21a branched into the plurality of optical paths while maintaining the axial symmetry by using a lens 23 and a movable lens 24, condensing it onto a surface of the heat sink 13, and forming a plurality of dimples 18 on the surface by using an interference pattern generated by the light condensation.

Description

  The present invention relates to a semiconductor device including a semiconductor element, a connected member connected to the semiconductor element, and a mold resin that seals the semiconductor element, and a method for manufacturing the semiconductor device. A plurality of dimples are formed on at least one surface of the connected member.

  In order to improve the adhesion between the semiconductor element and a member to be connected (for example, a lead frame) connected to the semiconductor element and the mold resin, the surface of the semiconductor element and the member to be connected is subjected to uneven processing or roughening treatment. It has been proposed to form a plurality of dimples. Such dimples need to be formed in a region designed in advance.

  Conventional techniques related to unevenness processing or roughening treatment include dimple processing or V-groove processing for pressing a mold with an uneven pattern formed on a processing surface such as a lead frame (hereinafter referred to as “mold processing”). Called). As another conventional technique, there is a roughening process (hereinafter, “chemical roughening”) in which a work surface is chemically dissolved using a chemical solution. Another conventional technique is blasting, that is, roughing using a method in which a projection material such as alumina or silicon carbide is accelerated by compressed air or the like and collides with a work surface (hereinafter referred to as “mechanical roughing”). )

  However, according to “mold processing”, although it is excellent in running cost and tact (process work time), it is necessary to prepare an expensive mold for each type (product) to be processed. In addition, it is necessary to change the mold according to the product type, and as a result, not only does it take a lot of time to replace heavy objects or adjust the position of the mold, but it also causes variations between processing lots. It was also. Further, even when a slight change is made in the pattern such as dimples, there is a problem that additional work on the mold by the mold maker is required, and this additional work takes time and cost. Furthermore, even if fine and high-density dimples or V-shaped grooves are formed in the mold in order to improve the adhesion to the mold resin, the workability of the work material to the fine shape of the mold is not sufficient. Pressing is difficult, and necessary dimples and V-grooves cannot be formed with high accuracy. In addition, there is a problem that warpage is likely to occur.

  Next, according to “chemical roughening”, although fine and complicated dimples can be uniformly formed, since a method such as a spray method or a dipping method is generally used, a dimple is selectively formed. It was necessary to prepare a chemical-resistant mask material for each type of processed material. In addition, water treatment facilities for obtaining pure water such as precipitation, filtration, adsorption, and ion exchange, regeneration treatment facilities using hydrochloric acid and caustic soda to regenerate deteriorated ion exchange resins, components contained by chelating agents (metals) Large-scale infrastructure facilities such as wastewater treatment facilities such as recovery of wastewater and storage facilities for used waste liquids are necessary, and there has been a problem in securing costs and land.

  In order to perform chemical roughening selectively, a dry film is pasted with a laminator ⇒ pattern laser exposure ⇒ developer spray ⇒ water washing ⇒ chemical roughening ⇒ water washing ⇒ stripping solution spray ⇒ water washing ⇒ (rust prevention treatment + Many processes such as washing with water ⇒ air knife ⇒ drying were generally required. In addition, as a result of the necessity of an analytical instrument for performing chemical solution management associated therewith, there has been a problem in that many devices, workers, analysts, etc. are required, which takes time and costs. Further, in terms of quality, since it is a wet process, there is a problem that problems such as a so-called watermark are likely to occur.

  Next, according to "mechanical roughening", although relatively fine and complex dimples can be formed, a mask material having a high wear resistance for selectively forming dimples is processed for each type of product to be processed. It was necessary to prepare. In addition, as with “chemical roughening”, large-scale infrastructure facilities and analytical equipment are required, and problems such as watermarks are likely to occur. Further, as problems inherent to mechanical roughening, there are problems that a projection material such as aluminum oxide and silicon carbide remains on the surface to be processed and that a strong warp tends to occur.

  Therefore, as a technique different from the above technique, for example, Patent Documents 1 to 4 propose a technique for forming dimples (grooves) or the like using a thermal action generated by irradiating laser light. For example, a dimple (groove) or the like is formed by narrowing the laser spot diameter with a lens and scanning the surface to be processed with a galvano mirror or the like.

  According to such laser processing, it is only necessary to change the control program for operating the laser for each type, so that a mold or mask material for each type is not required. In addition, since the dimples can be formed with high density and high accuracy without being affected by the malleability of the material, the adhesion between the semiconductor element and the mold resin can be increased. Furthermore, since infrastructure equipment and analysis equipment required for chemical roughening and mechanical roughening are unnecessary, there is a feature that a space required for the apparatus can be reduced.

Japanese Patent No. 3682369 Japanese Patent No. 4472773 JP-A-8-236684 JP 7-326699 A

  However, in the system using laser processing, there is a problem that the laser processing apparatus is expensive in order to form a large number of dimples (grooves) that can obtain a high anchor effect with a short tact.

  Specifically, in order to obtain a high anchor effect by dimples, (a) increasing the dimple aspect ratio (= dimple depth / dimple diameter) and (b) dimple density (= It is necessary to increase the dimple opening area / (dimple opening area + area of the region where no dimples are formed).

  However, even if the laser beam is sufficiently narrowed with a lens or the like, the spot diameter is about several tens of microns to several millimeters and is relatively large. Therefore, the diameter of the dimple formed in substantially the same shape is relatively large. It has become. Therefore, in order to satisfy the condition (a) while maintaining a short tact, it is necessary to form dimples deeply with a short tact. Therefore, parallel processing by a plurality of laser processing apparatuses or high processing capability is expensive but expensive. A high power laser device was required. Also, in condition (b), since the density of dimples and the increase in tact are proportional to each other, in order to satisfy condition (b) while maintaining a short tact, a plurality of laser processing apparatuses are used. Parallel processing or a high-power laser device with high processing capability but high cost is required.

  Therefore, the present invention has been made in view of the above-described problems, and an object thereof is to provide a technique capable of forming a plurality of desired dimples with a short tact.

  A method of manufacturing a semiconductor device according to the present invention is a method of manufacturing a semiconductor device comprising a semiconductor element, a connected member connected to the semiconductor element, and a mold resin that seals the semiconductor element and the connected member. And (a) a step of emitting laser light from a laser oscillator, and (b) a step of branching the laser light into a plurality of optical paths that are axially symmetric with respect to the optical axis of the laser light using a beam splitter. (C) The laser beam branched into the plurality of optical paths is guided using a lens while maintaining the axial symmetry, and is condensed on the surface of at least one of the semiconductor element and the connected member. And forming a plurality of dimples on the surface using an interference pattern generated by the light collection.

  According to the present invention, the branched laser light is condensed on the surface of at least one of the semiconductor element and the connected member, and a plurality of dimples are formed on the surface using the interference pattern generated thereby. Therefore, since a plurality of dimples having a small diameter can be formed by parallel processing, these can be formed with a short tact.

1 is a cross-sectional view showing a configuration of a semiconductor device according to a first embodiment. FIG. 6 is a perspective view illustrating the method for manufacturing the semiconductor device according to the first embodiment. 3 is a diagram showing a heat sink manufactured by the method of manufacturing a semiconductor device according to the first embodiment. FIG. It is a figure which shows the evaluation result of the anchor effect by a dimple. It is a figure which shows the evaluation result of the retention effect of the brazing member by a dimple. It is a figure which shows the evaluation result for suppressing generation | occurrence | production of a burr | flash. It is a figure which shows the semiconductor element manufactured by the manufacturing method of the semiconductor device which concerns on a modification.

<Embodiment 1>
FIG. 1 is a cross-sectional view showing a configuration of a semiconductor device 1 according to Embodiment 1 of the present invention. The semiconductor device 1 includes a semiconductor element 11 and a connected member connected to the semiconductor element 11 (here, lead frames 12a and 12b electrically connected to the semiconductor element 11 and the semiconductor element 11 thermally. The heat sink 13 is connected to dissipate heat from the semiconductor element 11, and the mold resin 14 seals the semiconductor element 11 and the connected member.

  In the first embodiment, the upper surface of the semiconductor element 11 is connected (joined) to the lead frame 12a via a solder layer 15 that is a brazing member, and is connected to the lead frame 12b via a wire 16. . The lower surface of the semiconductor element 11 is connected to the heat sink 13 through a solder layer 17 that is a brazing member. Further, the heat sink 13 has a quadrangular shape in plan view, and dimples to be described later are formed on the peripheral edge portion (edge portion) thereof.

  Next, the material of each component will be described. As the semiconductor element 11, silicon carbide, gallium nitride, gallium oxide, or the like is used in addition to silicon that is generally used. Here, the semiconductor element 11 is demonstrated as what is comprised from silicon carbide.

  The lead frames 12a and 12b include Cu (oxygen-free copper), Cu-Fe-P, Cu-Fe-Zn-P, Cu-Ni-Si-Mg, Cu-Ni-Si-Sn-Zn, Cu -Cr-Sn-Zn, Fe-Ni, etc. are applied. Here, the case where the lead frames 12a and 12b are made of Cu (oxygen-free copper) will be described as an example.

As the heat sink 13, in addition to metals such as Cu (oxygen-free copper), Al (aluminum), W, Mo, Cu—Mo, and Cu—W, ceramics such as metallized AlN and Al ₂ O ₃ are applied. . Here, a case where the heat sink 13 is made of Cu (oxygen-free copper) will be described as an example.

  As the mold resin 14, an epoxy resin, a polyimide resin, a phenol resin, an unsaturated polyester resin, or the like is applied. Note that alumina, silica, silicon nitride, boron nitride, titanium oxide, silicon carbide, or the like may be applied as a filler for the mold resin 14.

  As the solder layers 15 and 17, Sn—Ag—Cu, Sn—Ag—Sb, Sn—Cu, Sn—Sb, Bi—Sn, Sn—In, Sn—Ag, Sn—Pb, Pb—Sn—Sb, Sn-Pb-Bi, Sn-Pb-Cu, Sn-Pb-Ag, Pb-Ag, or the like is applied. As the wire 16, Au (gold), Cu (copper), Al (aluminum), or the like is applied.

<Manufacturing method>
Next, regarding the method for manufacturing the semiconductor device according to the first embodiment, a method for forming dimples on the surface of a workpiece will be described. Here, description will be made assuming that the workpiece is the heat sink 13.

  FIG. 2 is a perspective view showing the configuration of the laser processing apparatus 2 that forms the dimples 18 on the surface of the heat sink 13. The laser processing apparatus 2 includes a laser oscillator 21, a beam splitter 22, lenses (here, a lens 23 and a movable lens 24), and a stage 25 on which the heat sink 13 is placed.

  The laser oscillator 21 emits laser light 21a by generating pulsed light. The wavelength of the laser beam 21a emitted from the laser oscillator 21 should be selected according to the reflectance of the workpiece. For example, when the workpiece is made of oxygen-free copper as in the first embodiment, since the reflectance of oxygen-free copper with respect to red to line red light is high, the laser beam 21a has a wavelength Is preferably 533 nm or less. Although details will be described later, the pulse width of the pulsed light generated by the laser oscillator 21 is preferably picoseconds to femtoseconds so that generation of burrs (metal foreign matter) can be suppressed.

  Laser light 21 a emitted from the laser oscillator 21 is incident on the beam splitter 22. The laser oscillator 21 and the beam splitter 22 are arranged so that the optical axis of the laser beam 21a and the optical axis of the beam splitter 22 coincide with each other and the incident direction of the laser beam 21a and the incident surface of the beam splitter 22 are perpendicular to each other. The position of is adjusted.

  The beam splitter 22 branches the laser light 21a emitted from the laser oscillator 21 into a plurality of optical paths (here, four optical paths) that are axially symmetric with respect to the optical axis of the laser light 21a. For example, a diffractive beam splitter that has a light transmission member having a diffraction grating pattern formed on the surface thereof and divides light by the diffraction is applied to the beam splitter 22. The number of branches (number of optical paths) of the beam splitter 22 is preferably about 3 to 4, and the intensity of each branched light 22a is preferably equal.

  In the following description, each laser beam 21a of a plurality of optical paths branched by the beam splitter 22 is referred to as “branched light 22a”.

  The lenses (the lens 23 and the movable lens 24) collect a plurality of branched lights 22a by overlapping them on the surface of the heat sink 13 in a single point.

  Specifically, the lens 23 makes the optical paths of the plurality of branched lights 22a parallel to the optical axis of the laser light 21a. Then, the movable lens 24 condenses the plurality of branched lights 22a from the lens 23 in a single spot on the surface of the heat sink 13 where the dimples are to be formed. The movable lens 24 is movable in the direction of arrow A shown in FIG. 2 so that it can flexibly correspond to the thickness of the workpiece such as the heat sink 13.

  At the location of the heat sink 13 (hereinafter also referred to as “laser spot 27”) where the plurality of branched lights 22a are superimposed in a single point by the lens 23 and the movable lens 24, the plurality of branched lights 22a interfere with each other. Then, an interference pattern including a plurality of bright and dark portions smaller in size than the laser spot 27 is generated. By using this interference pattern, a plurality of dimples 18 having substantially the same shape as the plurality of bright portions are formed almost simultaneously on the surface of the heat sink 13 (laser spot 27). That is, a plurality of dimples 18 are formed by parallel processing.

  Here, as the interference pattern, a pattern in which a plurality of bright portions are arranged in a grid and each has a round shape is used. According to such a manufacturing method, a plurality of dimples 18 having a diameter smaller than the diameter of each branched light 22a (laser light 21a) can be formed on the surface of the heat sink 13 by parallel processing.

  It should be noted that the light intensity, the polarization angle, and the optical path length of each branched light 22a from branching to condensing by the beam splitter 22 must be aligned with high accuracy. In particular, when the optical path length is deviated, the interference required for dimple formation may not occur even if the light branched by the beam splitter 22 is collected. Therefore, the optical path lengths of the branched lights 22a are accurately aligned. It is necessary. This is particularly important when using laser light 21a having a narrow pulse width (for example, the pulse width is about picosecond to femtosecond).

  Therefore, in the first embodiment, the lens 23 and the movable lens 24 are so guided that the plurality of branched lights 22a are guided while maintaining the above-described axial symmetry and are superimposed on the surface of the heat sink 13 in a single point. Is arranged. Specifically, a lens having an axially symmetric shape is used for each of the lens 23 and the movable lens 24, and a plurality of branched light beams 22a are transmitted on the same circumference around the axis of each lens. A lens 23 and a movable lens 24 are arranged. By arranging in this way, the thickness of the portion where each branched light 22a passes through the lens can be made the same, that is, the distance that each branched light 22a propagates in the lens medium can be made the same. It is possible to accurately align the optical path lengths of the branched lights 22a.

  If the pitch between the dimples 18 (the pitch between the bright portions of the interference pattern) is d, the wavelength of the branched light 22a is λ, and the incident angle of the branched light 22a with respect to the surface of the workpiece is θ, the pitch d Is expressed by the following equation (1).

  Therefore, in order to reduce the pitch between the dimples 18, the incident angle θ may be reduced (the angle formed by the traveling directions of the branched lights 22a is reduced) or the wavelength of the branched light 22a may be reduced.

  The lens 23 may be configured by combining a plurality of lenses so that monochromatic aberration can be reduced, or a telecentric lens may be applied so as to reduce the influence of axial positional deviation. Similarly, the movable lens 24 may be configured by combining a plurality of lenses, or a telecentric lens may be applied. Further, when a mirror or the like is used using light having a relatively small pulse width as described above, it is preferable to use a mirror or the like that has been coated for the purpose of group delay dispersion control.

  The stage 25 can move the heat sink 13 in the directions of arrows B and C shown in FIG. Here, the stage 25 on which the heat sink 13 is placed is moved so that the one-point laser spot 27 passes through the peripheral edge of the heat sink 13. By this movement, a plurality of dimples 18 are formed at the peripheral edge of the heat sink 13.

  Here, the scanning method in which the optical path of the laser beam 21a (branched light 22a) is fixed and the workpiece is moved has been described. However, the present invention is not limited to this, and a scanning method may be used in which a workpiece is fixed and the optical path of the laser light 21a (branched light 22a) is moved by a galvanometer mirror or the like.

  According to the manufacturing method of the semiconductor device according to the first embodiment as described above, the branched laser beam 21a is condensed on the surface of the connected member (here, the heat sink 13), and the interference pattern generated thereby. Are used to form a plurality of dimples 18 on the surface. Therefore, since the plurality of dimples 18 can be formed by parallel processing, these can be formed in a short time. Further, since the dimple 18 having a smaller diameter than the conventional dimple can be formed, the dimple aspect ratio (= dimple depth / dimple diameter) can be increased without forming the dimple deeply. . Therefore, also from this viewpoint, the plurality of dimples 18 having an appropriate shape can be formed in a short time. As a result, according to the semiconductor device manufacturing method of the first embodiment, the tact time can be shortened without using a plurality of laser processing apparatuses for performing parallel processing or an expensive high-power laser apparatus. be able to.

  FIG. 3 is a diagram illustrating the heat sink 13 manufactured by the above-described manufacturing method. FIGS. 3A, 3B, and 3C are a plan view, an enlarged plan view, and a plan view of the heat sink 13, respectively. Each corresponds to an enlarged sectional view.

  In the semiconductor device 1 (FIG. 1) including the heat sink 13 in which a plurality of dimples 18 are formed as shown in FIG. 3, due to the anchor effect of the mold resin 14 that has entered the dimple 18, Adhesion is improved. Therefore, the semiconductor device 1 with improved stress resistance can be realized. Moreover, since the semiconductor device 1 according to the first embodiment includes the semiconductor element 11 (FIG. 1) composed of silicon carbide, the operating temperature is made higher than that of the semiconductor device including the semiconductor element composed of silicon. be able to.

  In most cases where the mold resin 14 is peeled off from the peripheral portion of the heat sink 13 on which the dimples 18 are formed, minute peeling generated at the corners of the peripheral portion gradually progresses to other portions of the peripheral portion. It was found by the investigation that it was caused by Therefore, as shown in FIG. 3, the inner edge (boundary) of the peripheral edge (predetermined region) is preferably defined by a smooth curve 13a. If the dimple 18 is formed in this way, for example, the corner portion of the peripheral portion is replaced with the R shape 13b, and thus the peel resistance can be improved.

  In addition, as described above, according to the plurality of dimples 18, it is possible to improve the adhesion that the mold resin 14 adheres to the heat sink 13. On the other hand, it is investigated that the solder member of the solder layer 17 that has been dissolved and is liquid during manufacture is difficult to wet in the region where the plurality of dimples 18 are formed, that is, the wettability of the solder member is low. I understood. Therefore, as shown in FIG. 3, it is preferable that the plurality of dimples 18 surround the bonding region 13 c to be bonded to the solder layer 17 on the surface of the heat sink 13 and be formed on the outer peripheral portion adjacent to the bonding region 13 c.

  If the dimples 18 are formed in this manner, the adhesion between the mold resin 14 and the heat sink 13 can be improved in the outer peripheral portion, while the solder layer 17 (the brazing member) is connected to the heat sink 13 in the joining region 13c. Since the wettability of the solder layer 17 and the heat sink 13 can be maintained, the wettability can be maintained. As a result, it is possible to realize the semiconductor device 1 in which heat dissipation by the heat sink 13 is reliably performed. Further, by forming a plurality of dimples 18 on the outer peripheral portion that surrounds and adjoins the joining region 13c, the liquid brazing member melted at the time of manufacture leaks out from the joining region 13c and is joined to the joining region 13c. The shortage can be suppressed. That is, since the brazing member can be retained in the joining region 13c, it is possible to suppress the generation of voids in the solder layer 17. Also from this viewpoint, it is possible to realize the semiconductor device 1 in which heat dissipation by the heat sink 13 is reliably performed.

<Dimple shape and manufacturing conditions>
<Anchor effect by dimple>
In order to enhance the anchor effect by the dimple 18, it is necessary to make the shape of the dimple 18 difficult to remove the mold resin 14 that has entered. Therefore, the manufacturing conditions were changed to form dimples 18 having various shapes, and each of them was evaluated. The evaluation results are described below.

  FIG. 4 is a diagram showing the results of evaluating the anchor effect for three types of dimples having the same diameter but different depths, that is, different aspect ratios (= depth of the dimple central portion / dimple opening diameter). In this evaluation, a laser oscillator 21 having a wavelength of 533 nm, a pulse width of 4 nm, and an output of 6 W was used, and a plurality of dimples having a diameter of 25 μm were formed on the surface of the oxygen-free copper member with vertical and horizontal intervals of 50 μm. Then, after molding resin is formed on the surface on which the dimples are formed and a sample piece having a bottom area of 10 square millimeters is formed, the contact surface is subjected to a test fracture with a pressure in a shear direction, and the interface state is changed. Observed.

  For observation, an optical microscope and an optical roughness measuring instrument were used. When the mold resin remained in the dimples, the cohesive failure was observed when the mold resin remained, and when the resin was not, the cohesive failure rate (= number of cohesive failure occurred / (Number of cohesive failure and interface failure)) was calculated. In addition, it means that an anchor effect is so strong that the numerical value of a cohesive failure rate is high.

  As can be seen from the results shown in FIG. 4, the plurality of dimples 18 are preferably formed so that each of them has an aspect ratio of 0.3 or more. In particular, the plurality of dimples 18 are more preferably formed so that the aspect ratio of each of them is 0.5 or more. If the dimples 18 are formed in this manner, the rate at which the cohesive failure of the mold resin 14 occurs can be increased rather than the interface failure, so that the anchor effect can be enhanced (adhesion strength at the interface can be increased).

<Dwelling effect of brazing member by dimple>
FIG. 5 shows the retention effect of the brazing member by dimples on nine types of dimples having different dimple diameters and densities (= dimple opening area / (dimple opening area + area of dimple not formed area)). It is a figure which shows the result. In this evaluation, a plurality of dimples were formed on the surface of the oxygen-free copper member using the laser oscillator 21 under the same setting conditions as when the anchor effect was evaluated. Next, Sn—Ag—Cu solder was dropped onto the surface on which dimples were formed in a reducing atmosphere, and the wetting angle (contact angle) was observed.

  In FIG. 5, the ◯ mark indicates that the retention effect is strong (low wettability) and the wetting angle is obtuse, the X mark indicates that the retention effect is weak (high wettability) and the wetting angle is acute, and the Δ mark indicates that In the middle, the wetting angle is around a right angle.

  As can be seen from the results shown in FIG. 5, the plurality of dimples 18 are preferably formed so that each opening has a diameter of 0.25 mm or less and a density of 0.17 or more. . In particular, the plurality of dimples 18 are more preferably formed so that the diameter of each opening is 0.12 mm or less and the density is 0.35 or more. By forming the dimples 18 in this way, it is possible to enhance the retention effect of the brazing member by the dimples 18, that is, the effect of suppressing the spreading of the brazing member of the solder layer during manufacturing.

<Inhibition of burr (metal foreign matter) generation and manufacturing conditions>
Next, the results of evaluating the relationship between the suppression of the generation of burrs associated with the formation of dimples and the manufacturing conditions will be described. FIG. 6 is a diagram showing the results of evaluating the generation of burrs for a plurality of laser beams having different pulse widths. In this evaluation, a plurality of dimples were formed on the surface of the oxygen-free copper member. Next, the height of burrs near the dimple opening was measured using a three-dimensional roughness measuring instrument.

  As can be seen from the results shown in FIG. 6, the laser oscillator 21 preferably emits the laser light 21a by generating pulsed light having a pulse width of 20 ns or less. According to such a laser beam 21a, the thermal action on the workpiece is reduced, so that the generation of burrs can be suppressed. As a result, the occurrence of process defects due to the burr can be reduced. Further, since the thermal action is reduced, it is possible to suppress the occurrence of thermal warping and distortion of the workpiece.

<Modification>
In the above description, the formation of the plurality of dimples 18 on the heat sink 13 has been described. However, the present invention is not limited to this, and a plurality of dimples 18 may also be formed on the surface of the semiconductor element 11 in order to improve the adhesion between the semiconductor element 11 and the mold resin 14.

  FIG. 7 is a view showing a silicon wafer 31 having a plurality of semiconductor elements 11 formed on the surface. FIGS. 7A and 7B are a plan view and an enlarged plan view of the silicon wafer 31, respectively. It corresponds.

  The silicon wafer 31 is cut along the dicing line 32 by dicing (not shown), so that the plurality of semiconductor elements 11 are separated. Here, the laser spot 27 is scanned on the dicing line 32 in a state where the diameter of the laser spot 27 is adjusted to be slightly wider than the interval between the adjacent semiconductor elements 11. Thereby, the number of scans of the laser spot 27 necessary for forming the dimple 18 on the peripheral edge of the semiconductor element 11 can be reduced. Similar to the heat sink 13, the inner edge (boundary) of the peripheral edge (predetermined region) where the plurality of dimples 18 are formed is preferably defined by a smooth curve 13a. If comprised in this way, peeling resistance can be improved like the above-mentioned.

  Although not shown, a plurality of dimples 18 may be formed on the surfaces of the lead frames 12a and 12b in order to improve the adhesion between the lead frames 12a and 12b and the mold resin 14. Similar to the heat sink 13, the plurality of dimples 18 may be formed on the outer periphery of the surface of the lead frame 12 a that surrounds and adjoins the bonding region bonded to the solder layer 15 (the brazing member). Preferably, the plurality of dimples 18 are formed on the outer periphery of the surface of the lead frame 12b that surrounds and is adjacent to the bonding region bonded to the wire 16.

  If formed in this manner, the adhesiveness between the mold resin 14 and the lead frames 12a and 12b can be improved in the outer peripheral portion, as described above, while the solder layer 15 and the wires 16 are connected in the bonding region. Adhesive strength (bondability, wire bondability) with the lead frames 12a and 12b can be ensured. As a result, it is possible to realize the semiconductor device 1 in which the lead frames 12a, 12b and the like are reliably electrically connected. Moreover, since the liquid brazing member melt | dissolved at the time of manufacture can suppress leaking outside from a joining area | region, it can suppress that a void generate | occur | produces in the solder layer 15 grade | etc.,. From this point of view, the semiconductor device 1 in which the lead frames 12a, 12b and the like are securely electrically connected can be realized.

  In the present invention, the embodiments can be appropriately modified and omitted within the scope of the invention.

  DESCRIPTION OF SYMBOLS 1 Semiconductor device, 11 Semiconductor element, 12a, 12b Lead frame, 13 Heat sink, 13a Curve, 13c Joining area, 14 Mold resin, 15, 17 Solder layer, 16 Wire, 18 Dimple, 21 Laser oscillator, 21a Laser light, 22 Beam Splitter, 23 lenses, 24 movable lenses.

Claims (8)

A method for manufacturing a semiconductor device comprising: a semiconductor element; a connected member connected to the semiconductor element; and a mold resin that seals the semiconductor element and the connected member.
(A) emitting laser light from a laser oscillator;
(B) branching the laser light into a plurality of optical paths that are axially symmetric with respect to the optical axis of the laser light using a beam splitter;
(C) The laser beam branched into the plurality of optical paths is guided using a lens while maintaining the axial symmetry, and is condensed on the surface of at least one of the semiconductor element and the connected member. And a step of forming a plurality of dimples on the surface using an interference pattern generated by the light collection. A method of manufacturing a semiconductor device according to claim 1,
In the step (c), the plurality of dimples are formed in a predetermined region on at least one surface of the semiconductor element and the connected member, and a boundary of the predetermined region is defined by a smooth curve Device manufacturing method. A method of manufacturing a semiconductor device according to claim 1,
The connected member includes a lead frame connected to the semiconductor element via a wire or a brazing member,
In the step (c), the plurality of dimples are formed on an outer peripheral portion of the surface of the lead frame that surrounds and is adjacent to a bonding region bonded to the wire or the brazing member. Production method. A method of manufacturing a semiconductor device according to claim 1,
The connected member includes a heat sink connected to the semiconductor element via a brazing member,
In the step (c), the plurality of dimples surrounds a bonding region to be bonded to the brazing member on the surface of the heat sink, and is formed in an outer peripheral portion adjacent to the bonding region. A method of manufacturing a semiconductor device according to any one of claims 1 to 4,
In the step (c), the plurality of dimples are formed on the surface made of copper so that each opening has a diameter of 0.25 mm or less and a density of 0.17 or more. A method for manufacturing a semiconductor device. A method of manufacturing a semiconductor device according to any one of claims 1 to 5,
In the step (c), the plurality of dimples are formed so that each of them has an aspect ratio of 0.3 or more. A method for manufacturing a semiconductor device according to any one of claims 1 to 6,
In the step (a), the laser oscillator emits the laser light by generating pulsed light having a pulse width of 20 ns or less. A semiconductor device formed by the method for manufacturing a semiconductor device according to claim 1,
The semiconductor device, wherein the semiconductor element is made of silicon carbide.

Images