JP3494357B2 - Semiconductor device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To secure electrical and mechanical connection between a bump electrode and a pad electrode, even substances where electrically insulating fine particles such as silica, alumina, or the like dispersed in large quantity into a resinous adhesive are used for connecting a semiconductor pellet with a wiring board. SOLUTION: In this semiconductor device, a semiconductor pellet 1 where a bump electrode 3 is made and a wiring board 4 where a flat pad electrode 8 is made, corresponding to the bump electrode 3 of the semiconductor pellet 1 at least, on an insulating substrate 5, are faced to each other via a resinous adhesive 9 where fillers 12 for reduction of thermal expansion coefficient are dispersed, and several electrodes are superposed on top of one another and are pressurized to electrically connect each electrode and also to bond the facing opposite faces of the semiconductor pellet 1 and the wiring board 4. In this case, the bump electrode 3 is composed of a metal softer than the fillers wiring the adhesive, and the peripheral section adjacent to the pressure contact end face is bulged by crushing, while pressure-welding this bump electrode 3 to the pad electrode 8 and recessing the surface of the pad electrode 3.

Description

Description: BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a semiconductor device having a structure in which a semiconductor pellet and a wiring board are opposed to each other via a resin-based adhesive, and respective electrodes are pressed and electrically connected. The present invention relates to a semiconductor device. 2. Description of the Related Art Among electronic circuit devices, portable electronic circuit devices such as video cameras and notebook personal computers are required to be smaller and lighter, and the electronic components used therein are further downsized. Is required. In mechanical devices controlled by the electronic control unit, high-precision and high-speed control is possible by directly incorporating a control unit into a drive mechanism such as a motor in a work position. When there are restrictions on the mounting dimensions, miniaturization and high-density mounting are strongly required. In order to respond to such demands, a method of reducing the size by changing the shape of the electronic component or reducing the size of the electronic component body and a method of integrating a plurality of functional components even if the external dimensions are the same or large are adopted. There is a method of substantially reducing the size. In addition, the electronic component main body is directly mounted on a wiring board to increase the mounting density and achieve downsizing. FIG. 4 shows a semiconductor device having a high-density mounting structure. In the figure, reference numeral 1 denotes a semiconductor pellet, and a semiconductor substrate 2
Although not shown, a number of electronic circuit elements are formed and connected internally, and a number of bump electrodes 3 connected to the main parts of the electronic circuit are formed on one main surface thereof. 4 is a wiring board,
A conductive pattern 6 for wiring is formed on an insulating substrate 5 of ceramic, glass, epoxy resin, glass epoxy resin, or the like, and the conductive pattern 6 is covered with a resist film 7. A portion corresponding to the electrode 3 is opened and a part of the conductive pattern 6 is exposed to form a pad electrode 8. Reference numeral 9 denotes a resin-based resin that is supplied in advance to a region surrounded by the pad electrode 8 on the wiring substrate 4 and is expanded by the semiconductor pellet 1 disposed opposite to the wiring substrate 4 in a state where the bump electrode 3 and the pad electrode 8 are positioned. As an adhesive, a thermosetting epoxy resin or the like is used. In this semiconductor device, when the semiconductor pellet 1 is pressurized and the bump electrode 3 and the pad electrode 8 are pressed and electrically connected to each other, the adhesive 9 supplied in advance on the wiring board 4 is filled between the electrode overlap portions. However, since the opposing surfaces of the bump electrode 3 and the pad electrode 8 are not completely flat and not parallel, the adhesive material is expelled from between the electrodes when the polymerizing pressure is applied. There is no problem with the electrical connection. In particular, as shown in FIGS. 5 to 7, a metal ball 11a is formed at the end of a metal wire 11 inserted into the capillary 10, and the metal ball 11a is pressed against the semiconductor substrate 2 at the lower end of the capillary 10 to raise the capillary 10 After the wire 11 is extended by a predetermined length, the wire 11 is clamped above the capillary 10 and the capillary 11 is further raised to cut the wire 11 to form the bump electrode 3. Therefore, when the bump electrode 3 is pressed against the pad electrode 8, it is crushed from the small-diameter end of the bump electrode 3. When viewed from the pad electrode 8 side, the bump electrode 3 is tightly polymerized while pushing the adhesive 9 outward from the center of the electrode 8. Therefore, there is little possibility that the adhesive remains between the overlapping portions. FIG. 4 shows a wiring board 4 made of a resin-based insulating substrate which is lighter than ceramic or glass.
Is used, since the difference in the coefficient of thermal expansion between the semiconductor pellet 1 and the insulating substrate 5 is large, the wiring substrate 4 warps when the temperature rises due to repeated operation and shutdown, and the warped wiring substrate 4 returns to a flat state due to the temperature drop. Repeat that. The repetition causes a crack at the bonding interface of the adhesive material 9, and the crack grows to lower the moisture resistance. When the peeling reaches the electrode overlap portion, the deformation stress directly acts on the electrode overlap portion to cause the peeling. There is a problem that the electrical connection is impaired. Therefore, fine particles of silica or alumina are dispersed in a base resin made of an epoxy resin, so that the coefficient of thermal expansion is
The deformation is alleviated by using an adhesive adjusted to have an intermediate coefficient of thermal expansion. As such an adhesive having a low coefficient of thermal expansion, for example, an adhesive containing 50 to 70% by weight of a filler having a particle size of 1 to 20 μm is generally used. This allows
The reliability of a semiconductor device which consumes large power and has a large temperature rise during operation can be improved. On the other hand, in a semiconductor device using an adhesive having a low coefficient of thermal expansion, the adhesive 9 is completely dissolved with fuming nitric acid, the wiring substrate 4 is fixed, and the semiconductor pellet 1 is fixed. The semiconductor pellet 4 was separated from the wiring substrate 4 by applying a force in parallel with the substrate 4 by applying a block to the substrate 4 and the peeled surface of the bump electrode 3 was observed. As shown in FIG. It was found that the filler 12 composed of fine particles such as silica and alumina dispersed in the adhesive 9 was bitten. Specifically, after crimping, the diameter is 50 μm,
In a semiconductor pellet having 150 bump electrodes with a height of 40 μm, 90 to 95 bump electrodes bite the filler, and the occurrence rate was about 60 to 63%. Also, the bitten filler is concentrated almost at the center of the electrode overlap,
It occupied 10 to 18% of the polymerization area. Therefore, in this semiconductor pellet, the conductive area was reduced by the non-conductive filler in 6 to 11% of the total connection area of the electrodes. The reason that the filler is concentrated and bitten into the central portion of the electrode overlap portion is that when the semiconductor pellet 1 is opposed to the wiring board 4 and pressed, the adhesive 9 previously supplied on the wiring board 4 is pressed and spreads. However, since the interval between the semiconductor pellet 1 and the wiring board 4 is several tens of μm and the adhesive 9 is completed in a very short time from the start of spreading, the filler between the electrodes 3 and 8 hardly flows and the tip of the bump electrode 3 Is sandwiched between the pad electrodes 8, and the filler at the periphery is pushed away together with the resin when the tip of the bump electrode 3 is crushed and flattened, and is removed from the electrode overlap portion. . The filler caught in the electrode overlap portion is embedded in the electrode, but its coefficient of thermal expansion is different from that of the electrode material and does not adhere to the electrode material, so it repeatedly operates in an environment where the temperature changes rapidly. In this case, a microcavity is generated between the electrode material and the filler. Therefore, when the amount of the bitten filler increases, a crack grows from the cavity, which may eventually make the electrical connection between the electrodes unstable. was there. To solve such a problem, Japanese Patent Laid-Open Publication No.
No. 04 (prior art), an electrical connection between a bump electrode and a pad electrode is temporarily connected with a conductive adhesive containing silver powder having rubbery elasticity, and the adhesive is heated and cured. After that, a semiconductor device in which a semiconductor pellet and a wiring board are connected to each other with a thermosetting epoxy resin is disclosed. However, since an adhesive supply step and a heating step are respectively increased, it cannot be adopted immediately from the viewpoint of cost. Was. SUMMARY OF THE INVENTION The present invention has been proposed for solving the above problems, and is intended to position a large number of bump electrodes formed on a semiconductor pellet and a flat pad electrode on a wiring board. Then, the semiconductor pellet and the wiring board are opposed to each other via a resin-based adhesive in which a filler for reducing the coefficient of thermal expansion is dispersed, and the respective electrodes are polymerized and pressurized to electrically connect the respective electrodes. In the semiconductor device in which the opposing surfaces of the substrate are bonded, the bump electrode is made of a metal softer than the filler in the adhesive, and the bump electrode is pressed against the pad electrode so that the surface of the pad electrode is concavely curved. And a semiconductor device characterized in that a peripheral surface portion adjacent to the press-contact end surface is swelled by crushing. [0005] The semiconductor device according to the embodiment of the present invention constitutes the bump electrode at a softer filler in the adhesive metal, and is concave curved pad electrode surface is pressed against the bump electrode to the pad electrode In addition, the peripheral surface portion adjacent to the press-contact end surface is swelled by crushing, but gold which is sufficiently softer than the filler in the adhesive is used as the bump electrode material of the semiconductor pellet, and the semiconductor pellet is furthermore flexible. It is preferable to use a wiring board in which pad electrodes are formed on an insulating substrate having a property, and further, nickel and gold are sequentially laminated on exposed portions of a conductive pattern in which pad electrodes are made of copper. Further, the contact interface between the bump electrode and the pad electrode is an annular press contact interface that is inclined. In this case, by setting the area of the annular pressure contact interface between the bump electrode and the pad electrode to 20% or more of the area of the pressure contact end face, even if the filler in the adhesive is caught between the electrodes, the distribution of the filler at the annular pressure contact interface is reduced. it can. An embodiment of the present invention will be described below with reference to FIG.
In the figure, the same components as those in FIG. 4 are denoted by the same reference numerals, and duplicate description will be omitted. The semiconductor device shown in FIG. 1 is largely different from the semiconductor device shown in FIG.
Pressed in, while the pad electrode 8 is concave curved, only that swelled by crushing a peripheral surface portion adjacent to the pressure contact edge. In order to improve the swelling, the bump electrode 3 is made of an adhesive 9.
It is composed of a metal softer than the filler inside. Also,
A flexible insulating substrate 5 is used for the wiring substrate 4 so that the pad electrode 8 is easily concavely curved by the bump electrode 3 made of the soft metal. More specifically, as shown in FIG. 2, the insulating substrate 5 made of epoxy resin has a thickness of 18 μm.
The conductive pattern 6 formed by laminating and etching the copper foil is covered with a resist film 7, a main part of the conductive pattern 6 is opened, and a part of the conductive pattern 6 is exposed. A nickel layer 13 having a thickness of 0.03 to 1.0 μm is formed on the nickel layer 13.
4 are formed to form conductive pads 8. Copper or nickel constituting the pad electrode 8 is a bump electrode 3 made of gold.
, It is usually hard to deform even when pressed. The wiring board 4 uses a flexible insulating board 5.
To improve the pressure contact between the electrodes 3 and 8, the wiring board 4 is
When heated to 0 to 250 ° C., the insulating substrate 5 is further softened,
The lower surface of the periphery of the pad electrode 8 also softens. On the other hand, when the bump electrode 3 having a hemispherical or parabolic tip is brought into contact with the pad electrode 8 and pressed, the pressing force is concentrated on the tip of the bump electrode 3, and the pad electrode 8 is transferred to the insulating substrate 5. It is concavely curved like a push-in funnel. That is, as shown in FIG. 3, the parabolic tip 3a of the bump electrode 3 comes into contact with the pad electrode 8 as shown by the dotted line and advances in the direction of the arrow A shown by the dotted line by the pressing force. Solid line B
The tip of the bump electrode 3 shown in FIG. 1C is crushed and flattened as the concavely curved portion becomes deeper and wider as shown in FIG. At the same time, the crushed portion of the bump electrode 3 swells in the direction of the arrow D in the drawing to have a large diameter, fills the concave curved portion, and presses not only the end surface of the bump electrode 3 but also its peripheral surface with the pad electrode 8. On the other hand, among the fine fillers 12 such as silica and alumina dispersed in the adhesive material 9, at the moment when the bump electrode 3 of the pressurized semiconductor pellet 1 comes into contact with the pad electrode 8, the fine filler 12 remains between the electrodes 3 and 8. Filler 12 (in FIG. 3,
(Indicated by the symbol E) is confined between the electrodes 3, 8. On the other hand, the bump electrode 3 has its lower end peripheral surface bulged due to the crushing deformation of the front end portion and is pressed from the lower end to the concave curved surface of the pad electrode 8 so that the adhesive material 9 located between the electrodes is removed from the pressed surface. At the same time, the filler 12 is removed from the crimping surface. In this way, a swelling of the peripheral surface due to crushing of the tip portion of the bump electrode 3 forms an inclined annular pressure contact interface between the electrodes 3 and 8 as indicated by the reference symbol F. When the semiconductor pellet 1 once adhered was peeled off from the wiring board 4 and the amount of filler remaining between the electrodes was observed. At the tip of the bump electrode 3, the same amount of filler as the conventional one remained at the center of the bump electrode 3. On the other hand, no filler could be confirmed at all electrodes at the annular pressure-welding interface F. When the state of the connection interface between the electrodes 3 and 8 was confirmed by Auger electron spectroscopy analysis, it was confirmed that a deep layer of a gold-nickel alloy was formed at the tip of the bump electrode 3 and at the peripheral surface adjacent to the tip. Was confirmed, and it was confirmed that the crimping condition was also good. This annular press-contact interface F is larger in diameter than the flat press-contact portion at the lower end of the bump electrode 3. For example, when the flat crimp end face at the tip of the bump electrode 3 has a diameter of 50 μm, the width of the annular press-contact interface is equal to that of the pellet 1. Since the pressure can be set to 3 to 20 μm depending on the pressing force and the material of the wiring board 4, the pressure contact area of each of the electrodes 3 and 8 can be increased by 24 to 160% with respect to the pressure contact area of only the tip of the bump electrode 3, so that the filler is caught. Because it can be crimped in a much larger area than the area,
Electrical and mechanical connection between the electrodes 3 and 8 can be improved. Since the semiconductor device according to the present invention can be manufactured in exactly the same steps as the conventional device, the manufacturing is also easy. As described above, according to the present invention, a large amount of electrically insulating fine particles such as silica and alumina are dispersed in a resin adhesive for connecting a semiconductor pellet and a wiring board. Even if a bump electrode is used, electrical and mechanical connection between the bump electrode and the pad electrode can be ensured.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a side sectional view showing an embodiment of the present invention. FIG. 2 is an enlarged side sectional view of a main part of a semiconductor device. FIG. 3 illustrates a connection state between a bump electrode and a pad electrode. FIG. 4 is a side sectional view showing an example of a conventional semiconductor device. FIG. 5 is a side sectional view showing an example of a manufacturing process of a pad electrode of the device. FIG. FIG. 7 is a side sectional view showing an example of a manufacturing process of a pad electrode. FIG. 8 is a side sectional view of an electrode overlapping portion for explaining a problem of the apparatus. Semiconductor pellet 3 Bump electrode 4 Wiring board 5 Insulating substrate 8 Pad electrode 9 Resin-based adhesive 12 Filler

Claims (1)

(57) [Claim 1] A semiconductor pellet having a large number of bump electrodes formed on one main surface and a flat pad electrode on an insulating substrate corresponding to at least the bump electrode of the semiconductor pellet. The formed wiring board is opposed via a resin-based adhesive in which a filler for reducing the coefficient of thermal expansion is dispersed, and the respective electrodes are polymerized and pressurized to electrically connect the respective electrodes and to connect the semiconductor pellet and the wiring. In the semiconductor device in which the opposing surfaces of the substrate are bonded, the bump electrode is made of a metal softer than the filler in the adhesive, and the bump electrode is pressed against the pad electrode so that the surface of the pad electrode is concavely curved, and the pressing end face is formed. Has a ring-shaped pressure contact interface in which the bump electrode and the pad electrode are swelled by crushing the adjacent peripheral surface portion.
And a band-shaped rib at the above-mentioned annular pressure-contact interface between the bump electrode and the pad electrode.
A width of the pressing portion is 6% to 40% of a diameter of the press contact end face .