JP5368357B2 - Electrode member and semiconductor device using the same - Google Patents

Description

  The present invention relates to a structure of an electrode member for performing electrical connection between a semiconductor element and a wiring member, and particularly to a configuration suitable for a power semiconductor device that handles a large current.

Among semiconductor devices, power semiconductor devices are used to control and rectify relatively large power in vehicles such as railway vehicles, hybrid cars, and electric vehicles, home appliances, and industrial machines. Therefore, a semiconductor element used for a power semiconductor device is required to be energized at a high current density exceeding 100 A / cm ² . Therefore, in recent years, silicon carbide (SiC), which is a wide band gap semiconductor material, has attracted attention as a semiconductor material that replaces silicon (Si), and a semiconductor element made of SiC can operate at a current density exceeding 500 A / cm ^2. It is. Further, SiC is capable of stable operation even at a high temperature of 150 ° C. to 300 ° C., and is expected as a semiconductor material capable of achieving both high current density operation and high temperature operation.

  On the other hand, in semiconductor devices including those for electric power, the lower electrode surface of the semiconductor element is solder-bonded to the circuit pattern on the insulating substrate, and an aluminum wire is ultrasonically bonded to the upper electrode surface to form a power supply path for the semiconductor element. It has generally been done. However, since the linear expansion coefficient of aluminum is about 22 ppm / K and the linear expansion coefficient of Si, SiC, etc. used as a semiconductor element is greatly different from 3-5 ppm / K, thermal stress accompanying repeated temperature changes. As a result, it is known that a crack develops inside the wire and eventually breaks. This is hereinafter referred to as a power cycle life because durability against breakage can be evaluated by applying an intermittent pulse current load called a power cycle test.

  Since the power cycle life is a phenomenon related to cracks caused by thermal stress, it greatly depends on the temperature change amount of the semiconductor element. In addition, the power cycle life tends to be short as the maximum temperature of the semiconductor element increases. As described above, since a wide band gap semiconductor is expected to operate at a high temperature, a wiring structure that can withstand a high maximum temperature and a large amount of temperature change is essential as compared with a Si device. This is because, for example, a wiring structure provided with a maximum temperature of 125 ° C. and a temperature change of 80 K with respect to a conventional Si semiconductor has a maximum temperature of 175 ° C., which is an operating condition in a wide band gap semiconductor. When the temperature change amount is 130K, the power cycle life is reduced to 1/10 or less. In other words, a wiring structure such as ultrasonic bonding of aluminum wires that has been used for conventional Si semiconductors is sufficient for severe temperature conditions that enable the function of a wide bandgap semiconductor in a power semiconductor device. A long life could not be obtained.

  Therefore, a semiconductor device has been proposed in which a buffer plate close to the linear expansion coefficient of the semiconductor element is inserted in the joint between the flat lead member and the semiconductor element instead of the wire, and the thermal stress generated in the joint during thermal cycling is alleviated. (For example, see Patent Document 1 or Patent Document 2). Further, a power semiconductor module has been proposed in which an electrode for bonding a wire is arranged on a semiconductor element via a conductive resin having a low elastic modulus so as to relieve thermal stress applied between the semiconductor element and the electrode. (For example, refer to Patent Document 3).

Japanese Patent Laid-Open No. 11-163045 (paragraph 0034, FIG. 2) JP 2000-332067 (paragraphs 0007, 0012 to 0014, FIG. 1) Japanese Patent Laid-Open No. 11-163045 (paragraphs 0018 to 0019, FIG. 1)

  However, if the linear expansion coefficient of the buffer member is close to that of the semiconductor element, the life against repeated thermal stresses of the semiconductor element, the buffer member, and the joint is increased, but it is difficult to suppress the thermal stress between the buffer member and the lead member. It becomes. Conversely, when the linear expansion coefficient of the buffer member is made closer to the lead member, the repeated thermal stress life of the lead member joint becomes longer, but it becomes difficult to suppress the thermal stress between the semiconductor element and the buffer member. In other words, in order to optimize the power cycle life, it is necessary to select a point at which the two lifespans are almost equal from this trade-off relationship. This means that the power cycle life has a limit point due to the linear expansion coefficient trade-off. Under such circumstances, it has been impossible to obtain a power cycle performance more than 10 times that is necessary for high-temperature operation in a wide band gap semiconductor. In addition, when a conductive resin having a low elastic modulus is inserted, the resin is easily deteriorated at a high temperature, so that the reliability is lowered, and since the resistance is higher than that of the metal member, a power loss is generated and the efficiency is lowered. There was a problem.

  The present invention has been made to solve the above-described problems, and an object of the present invention is to obtain a highly reliable semiconductor device even when the maximum temperature reached is high and the temperature change amount is large.

  The electrode member of the present invention is an electrode member for electrically connecting the semiconductor element and the wiring member, and the second metal is formed in a region having a predetermined length from one end side in the length direction of the belt-shaped first metal plate. A U-shaped member in which the first metal plate is positioned outside the second metal plate by folding the band-like material on which the plate is bonded to the other end side of the region where the second metal plate is bonded. In the surface that is the outer shape of the U-shape, the bonding surface with the semiconductor element is in a predetermined length region from the one end side to the bent portion, and the predetermined surface from the other end side to the bent portion. A joining surface with the wiring member is provided in a length region, and the linear expansion coefficient of the first metal plate is closer to the linear expansion coefficient of the wiring member than the linear expansion coefficient of the semiconductor element, and the second The strip-like material line in the region where the metal plates are bonded together Expansion coefficient is close to the linear expansion coefficient of the semiconductor element than the linear expansion coefficient of the wiring member, it is characterized.

  According to the electrode member of the present invention, the bonding surface with the semiconductor element is close to the linear expansion coefficient of the semiconductor element, the bonding surface with the wiring member is close to the linear expansion coefficient of the wiring member, and the laminated structure interface in the electrode member is Since a current flows between both junction surfaces without passing through, a highly reliable semiconductor device can be obtained even when the maximum temperature reached is high and the temperature change amount is large.

It is a figure for demonstrating the structure of the power semiconductor device and electrode member concerning Embodiment 1 of this invention. It is a fragmentary sectional view for demonstrating the method to manufacture a power semiconductor device using the electrode member concerning Embodiment 1 of this invention. It is sectional drawing for demonstrating the structure of the 1st modification of the electrode member concerning Embodiment 1 of this invention. It is sectional drawing for demonstrating the structure and manufacturing method of the 2nd modification of the electrode member concerning Embodiment 1 of this invention. It is a figure for demonstrating the structure of the 3rd modification of the electrode member concerning Embodiment 1 of this invention. It is sectional drawing for demonstrating the structure of the electrode member concerning Embodiment 2 of this invention. It is sectional drawing for demonstrating the structure of the electrode member concerning Embodiment 3 of this invention.

Embodiment 1 FIG.
1 and 2 are diagrams for explaining a semiconductor device electrode member and a power semiconductor device according to a first embodiment of the present invention. FIG. 1 shows a semiconductor device electrode member (of a power semiconductor device). FIG. 1A is a top view, FIG. 1B shows a semiconductor element to which a semiconductor element is bonded) and a bonding wire portion connected to the semiconductor element through the electrode member. FIG. 1A is a cross-sectional view taken along the line AA of FIG. 1A, and FIG. 1C is a cross-sectional view showing a state of a clad material before molding an electrode member. FIG. 2 is a diagram for explaining a method of bonding wires to electrode members as a method of manufacturing the power semiconductor device according to the first embodiment. 3-5 is sectional drawing which shows the structure of the electrode member concerning the modification of this Embodiment. First, the configuration and operation of the power semiconductor device and the electrode member will be described.

  As shown in FIG. 1, the power semiconductor device 10 has a circuit pattern 2 bonded with a brazing material (not shown) on a circuit surface 1f of an insulating substrate 1 made of a ceramic material such as aluminum nitride, silicon nitride, or alumina. Has been. The circuit pattern is made of a conductive material such as copper or aluminum or an alloy material containing them as a main component. Further, the surface of the circuit pattern 2 is formed with a plating film such as nickel in consideration of oxidation prevention and wettability of the solder material. Moreover, although not shown in figure, the heat sink may be formed in the surface on the opposite side to the circuit surface 1f of the insulating substrate 1 (surface in which 2r is formed in the figure).

In FIG. 1, the semiconductor element 3 is connected to the circuit pattern 2a via the solder 5b. The semiconductor element 3 may be a general element based on a silicon wafer. In the present invention, the semiconductor element 3 has a wider band gap than silicon such as silicon carbide (SiC), gallium nitride (GaN), or diamond. It is intended to be applied to a band gap semiconductor material, and is particularly applied to a semiconductor element using silicon carbide. Device types include switching elements such as IGBTs (Insulated Gate Bipolar Transistors) and MOSFETs (Metal Oxide Semiconductor Field-Effect-Transistors), or rectifying elements such as diodes. In the case of a MOSFET, a drain electrode is formed on the surface of the semiconductor element 3 on the circuit pattern 2 side. The gate electrode and the source electrode are formed on the surface opposite to the drain electrode (upper side in the figure), but the upper electrode is formed in order to make the characteristics of the embodiment of the present invention easier to understand. Only the source electrode through which a large current flows will be described on the surface. Note that a composite metal film is formed on the surface of the drain electrode to improve the bonding with the solder 5a. On the surface of the source electrode, a thin electrode film such as aluminum having a thickness of several μm and a thin film layer such as titanium, molybdenum, nickel, and gold (not shown) are formed. The first joining surface _6t1 of the electrode member 6 is joined to the upper surface of the semiconductor element 3 via the solder 5a. Note that the solder 5a and the solder 5b may be a bonding material that is not classified into solder, such as a sinterable filler mainly composed of silver or a brazing material.

The electrode member 6 is formed by bonding and integrating at least two types of metal plates having different linear expansion coefficients, and one of the two types of metal plates 6 _MC is bonded to the bonding wire 4 rather than the semiconductor element 3. It is a material with a close linear expansion coefficient, and is a highly conductive material mainly composed of copper or copper. The other type of metal plate 6 _MB is a material having a linear expansion coefficient closer to the semiconductor element 3 than the bonding wire 4 or lower than that of the semiconductor element 3, and molybdenum or an alloy containing molybdenum, Alternatively, it is a so-called low thermal expansion material such as Invar alloy.

As shown in FIG. 1 (c), the electrode member 6 is a two-layer clad material composed of a metal plate _6MC and a metal plate _6MB , and the side on which the metal plate _6MB is laminated only on one end _6E1 side. The lay clad material 6B is bent into a U shape (or a U shape) so that the metal plate 6 _MC side is on the outside. Of the electrode member 6, the first joint surface 6 _t1 is the outer surface of the U-shape in the region 6 _FB to lower metal plate 6 _MB of linear expansion coefficient of the end 6 _E1 side of the clad material are integrated Due to the influence of the metal plate 6 _MB , the linear expansion coefficient is closer to the linear expansion coefficient of the semiconductor element 3 than the linear expansion coefficient of the metal plate 6 _MC itself, and is as small as the linear expansion coefficient of the semiconductor element 3. Yes. That is, in the electrode member 6, the region 6 _FB portion in which the metal plate 6 _MB is integrated functions as a so-called buffer member for the semiconductor element 3. On the other hand, the metal plate 6 _MB having a low coefficient of linear expansion on the other end 6 _E2 side of the clad material is not integrated, in other words, the U-shaped outside of the portion having only the highly conductive metal plate 6 _MC. linear expansion coefficient of the second bonding surface 6 _t2 is a side, the linear expansion coefficient of the metal plate 6 _MC, that is, become a linear expansion coefficient comparable to the wiring member such as a bonding wire and the lead member, the cushioning member for the wiring member Function as.

Then, an aluminum or copper wire 4 is joined so as to bridge the joint surface 6 _{t 2} on the other end 6 _{E 2} side of the electrode member 6 and the other circuit pattern 2 b on the insulating substrate 1, and the source of the semiconductor element 3 The electrode can be electrically connected to an external circuit (not shown). Usually, a plurality of wires 4 are arranged side by side to ensure current capacity. Although not shown in the drawing, a large number of wires are generally arranged by changing the size of the loop shape of the wires 4. Moreover, not only a wire but a plate-like or ribbon-like lead member may be used.

In any configuration, for example, when a current of 50 amperes is applied to the electrode member 6, if a copper plate having a width of 4 mm and a thickness of 0.3 mm is used for the _MC portion of the metal plate 6, the cross-sectional area becomes 1.2 mm ^2. Sufficient current capacity can be secured against heat generated by energization.

Next, the operation will be described.
When the power semiconductor device 10 is driven, a current flows through various elements in the power semiconductor device 10 including the semiconductor element 3. At this time, a power loss corresponding to the electrical resistance is converted into heat, and heat is generated. Arise. At this time, since the current changes greatly with time, the temperature distribution accompanying heat generation and heat conduction is not steady but transient, and the time during which the semiconductor element 3 reaches the maximum temperature is also short. . Therefore, in the configuration in which the bonding portion between the wire 4 and the bonding surface _6t2 is separated from the surface of the semiconductor element 3 that is a heat generating source as in the electrode member 6 according to the first embodiment, heat conduction from the semiconductor element 3 is performed. Is delayed, and the temperature rise at the joint between the wire 4 and the joint surface _6t2 is suppressed. For this reason, compared with the case where a wiring member such as a wire is joined immediately above the semiconductor element, or the case where the wiring member is joined on the buffer member joined on the semiconductor element, the temperature change amount and the maximum at the joint with the wire Achieving temperature is suppressed and thermal stress can be reduced.

Moreover, the linear expansion coefficient of the joint surface 6 _t1 joined to the semiconductor element 3 can be controlled by changing the ratio of the thickness of the low thermal expansion material 6 _MB and the good conductive material 6 _MC . For example, the metal plate 6 _MB is integrated. When the linear expansion coefficient of the portion 6 _FB is close to the linear expansion coefficient of the semiconductor element 3 and the difference between the two linear expansion coefficients is less than 5 ppm / K within the operating temperature range, the direction parallel to the solder joint surface The shear stress of the material is remarkably reduced, and cracks that develop parallel to the joint surface are less likely to occur. In that case, the solder may crack in the thickness direction due to the power cycle load, but the crack in the thickness direction is different from the bonding surface direction, and the influence on electric resistance and heat conduction may be small. That is, since the ratios of fixing, energization, and heat dissipation of the semiconductor element are small, it is possible to withstand a power cycle load having a high maximum temperature and a large amount of temperature change.

Moreover, the characteristic configuration of the present invention, among the electrode member 6, in that a bonding surface 6 _t2 for wire end 6 _E2 side low thermal expansion material 6 _MB is not integrated. This will be described in detail below. When copper is used for the good conductive material 6 _MC , nickel plating is applied to prevent oxidation, and a general aluminum wire is joined, the difference in linear expansion coefficient between the two is less than 8 ppm / K. The stress reduction in this case can realize a power cycle life of 40 times or more compared to the case where the wire is directly bonded onto the semiconductor element 3. Further, since ultrasonic bonding is not performed directly on the semiconductor element 3, it is not necessary to consider the element destruction phenomenon caused by propagation of ultrasonic energy to the semiconductor element 3. Therefore, by making the magnitude of the ultrasonic energy larger than the upper limit value at which element destruction does not occur, the wire can be greatly deformed and the wire bonding area can be expanded, thereby further improving the power cycle life. I can do it.

  This means that a thick wire that requires large ultrasonic energy at the time of bonding can be applied. For example, a diameter of 500 μm or more that has been difficult to apply to an electrode surface having a gate wiring in the past. The aluminum wire can be joined with sufficient ultrasonic energy to conduct a large current. Similarly, an aluminum ribbon having a width of 2 mm or more and a thickness of 200 μm or more that requires large ultrasonic energy can be bonded. In addition to aluminum, the ribbon material may be copper or a clad material of copper and aluminum. Furthermore, if a copper wire is used for the copper electrode member, the wire bonded portion can be made of the same material, so that a bonded body with little deterioration can be formed.

  Next, a method for manufacturing the semiconductor device 10 according to the first embodiment will be described with reference to FIG. The point in the assembly process of the power semiconductor device 10 according to the first embodiment is a wire bonding process to the electrode member 6 provided on the semiconductor element (semiconductor chip) 3. As described above, solder, a silver adhesive, a silver sintered body, or the like can be used for joining the circuit pattern 2 and the semiconductor element 3 of the insulating substrate 1 and the semiconductor element 3 and the electrode member 6. The bonding material may be changed depending on the bonding location, or may be the same. Moreover, all joining except a wire may be performed simultaneously, and you may perform separately.

After the joining of the electrode member 6 to the semiconductor element 3 is completed, the wire 4 is joined. Usually, ultrasonic bonding is used for wire bonding, and it is necessary that the position of the wire bonding portion does not shift due to ultrasonic vibration of the wire bonder tool. In the electrode member 6 shown in the first embodiment, when the thickness of the plate material exceeds 0.5 mm, ultrasonic bonding is possible as it is. On the other hand, for example, when the thickness of the electrode member 6 is 0.5 mm or less in order to reduce the thermal stress on the solder 5a between the semiconductor element 3 or the current amount, the ultrasonic bonding is performed. Insufficient strength to do. In that case, by inserting the vacuum chuck table 7 as shown in FIG. 2 on the rear surface _6t2R joint surface 6 _t2, the fixation of the electrode member 6 from the back _6t2R side of the wire bonding surface 6 _t2 by the electrode member 6 by vacuum suction Do. Furthermore, in order to ensure the fixation, it is effective to press the electrode member 6 against the end of the wire bonding surface 6 _{t 2} using the pressing claw 8 and sandwich and fix the electrode member 6 to the vacuum chuck table 7. If the wire bonding strength is sufficient, the vacuum suction function of the vacuum chuck table 7 may be omitted to simplify the mechanism. Wire bonding is performed by a wire bonder using ultrasonic bonding, and the vacuum chuck table 7 and the presser claw 8 are removed after the bonding is completed.

After the wire 4 is joined, the end 6 _E2 on which the wire joining surface 6 _t2 of the electrode member 6 is installed is unstable in a single-side supported state. Usually, however, the entire semiconductor device 10 is molded with a heat-resistant resin. The space below the joint surface is filled with resin, and the wire can be stably supported even in an environment where vibration is received.

Modification 1 of Embodiment 1
In the first embodiment shown in FIG. 1, an example in which the two-layer sidelay clad material 6B is used as the electrode member 6 is shown. However, in order to prevent the stress of the bimetal, it is symmetrical in the thickness direction as shown in FIG. You may make it take a three-layer structure so that it may become. FIGS. 3 (a) Yoshirubeden shows a cross section of the plate member 16B which constitute a 3-layer clad by inserting a low thermal expansion material _{16 MB} in the middle portion in the thickness direction at one end _{16 E1,} FIG. 3 (b) plate material 16B The cross section of the electrode member 16 bent into a U-shape is shown so that the member 16 _MC is on the outside (although it is outside on either side in this modification 1). In the first modification, due to the influence of the low expansion member 16 _MB , the linear expansion coefficient of the buffer portion 16 _FB provided with the first joint surface 16 _t1 is closer to the linear expansion coefficient of the semiconductor element than the wire. , since the same member of the linear expansion coefficient is adapted to be disposed at symmetrical positions in the thickness direction of the plate 16B, the force acting on the joint surface 16 _t1 side and the back 16 _T1R side becomes equal, such as a so-called bimetal The occurrence of warpage is suppressed.

  In the first modification, the partially clad (inlay clad) plate material 16B is shown, but the two side lay clad materials as shown in the first embodiment are overlapped so that the low expansion material faces each other. You may make it form integrally.

Modification 2 of Embodiment 1
Further, in the first embodiment and the first modified example, by using the partial clad material, the linear expansion coefficient is different between the one end side and the other end side. However, in this modified example, the configuration in the thickness direction of the plate material is different. An example in which the electrode member is configured using a clad material that is uniform in the extending direction of the surface of the plate material, a so-called overlay clad material will be described. 4A and 4B are diagrams for explaining an electrode member and a method for forming the electrode member according to the second modification. FIG. 4A is a cross section of the plate material, and FIG. 4B is a cross section of the etched plate material. FIG. 4C shows a cross section of the electrode member. In the figure, the plate member 26B is integrated by the low expansion member _{26 MB} from both sides in the thickness direction so as to sandwich in good conductive member _{26 MC1, 26 MC2,} in the extending direction of the surface is a uniform three-layer clad material Material is there. For this clad material 26B, as shown in FIG. 4 (a), a portion corresponding to the other end _{26 E2} the curve E-E is removed by etching, among the three layers, one good conductive member _{16 MC1} Only leave to leave. Then, as shown in FIG. 4B, when the etched plate material 26 _BE is bent into a U shape so that the good conductive member 16 _MC1 side is on the outside, the electrode member 26 as shown in FIG. Can be formed.

  In this case, since an overlay clad material having a large circulation amount and high bonding reliability of each layer can be used, variation in the linear expansion coefficient of the buffer portion 26FB is small, and the designed linear expansion coefficient can be easily obtained. Therefore, the quality is stabilized when used in the semiconductor device 10.

Modification 3 of Embodiment 1
Further, in the electrode member according to the present embodiment, since the portion where the bonding surface with the semiconductor element is provided and the portion where the bonding surface with the wiring member is provided are separated, the shape of the bonding surface with the wiring member However, there is no restriction on the shape of the semiconductor element. Therefore, the size of the joint surface with the wiring member can be made larger than the size of the semiconductor element. FIGS. 5A and 5B are diagrams for explaining the electrode member and the electrode member forming method according to the third modification. FIG. 5A is a plan view of the plate material, and FIG. FIG. 5C shows a plane of the electrode member after processing. In the figure, the plate member 36B is low expansion member 36 _MB are integrated, the width of the first mating surface 36 _t1 is provided at one end portion 36 _E1 side, but in accordance with the electrode surface of the semiconductor element to be joined, The width on the other end 36 _E2 side where the second bonding surface 36 _t2 is provided is wider than the width on the one end 36 _E1 side, and the width of the bonding surface 36 _t2 is larger than the electrode surface of the semiconductor element. .

Since the size of the second bonding surface 36 _t2 is made larger than the size of the semiconductor element as in the third modification, the number of wires that are restricted by the area of the semiconductor element as in the conventional case is set to the second number. It becomes possible to increase corresponding to the size of the joint surface 36 _t2 . For this reason, the amount of current per wire is reduced with respect to the current flowing through the semiconductor element, heat generation due to energization of the wire is reduced, and the maximum temperature of the wire is lowered, thereby improving the power cycle life. It becomes. This also extends the life of fatigue failure due to loop deformation in the energization / non-energization cycle of the wire, which is the second factor of the power cycle life.

The size on the other end _36E2 side where the second joining surface _36t2 is provided may project not only in the width but also in the length direction, and the projecting direction may be biased to one side.

<Example of adjustment of linear expansion coefficient>
In the above-described embodiment and modification, the linear expansion coefficient of the joint surface 6 _t1 joined to the semiconductor element 3 is controlled by changing the ratio of the thickness of the low thermal expansion material 6 _MB of the clad material and the good conductive material 6 _MC. Here are some examples: For example, a thickness ratio T _C1 : T _B : T _C2 of a three-layer structure in which an Invar alloy (thickness T _B ) is sandwiched between copper (thickness T _C ) as in Modification 1 shown in FIG. 3 is 1: 3. When set to 1, the linear expansion coefficient as a plate material in the three-layer structure portion 16 _FB is approximately 7 ppm / K.

Also, the second bonding surface that becomes the bonding surface with the wire is perpendicular to the first bonding surface that becomes the bonding surface with the semiconductor element. For example, if the electrode member is bent like an L shape, Although it is not impossible to function as an electrode member, it becomes a plane parallel to the electrode surface in consideration of connecting the wire to the joint portion in the surface extending direction of the circuit pattern provided on the insulating substrate 1. Is preferred. However, it is not necessary to be completely parallel, and it may be formed in a so-called U-shape that is closer to parallel than vertical. In this case, the wire bonder and the ribbon bonder may be improved based on a general apparatus, so that an increase in cost can be suppressed. Further, in the U-shape, the back surface 6 _{t1R of} the first joint surface 6 _t1 that is the joint surface with the semiconductor element 3 and the back surface of the second joint surface 6 _t2 that is the joint surface with the wiring member 4. It is _desirable that there is a predetermined gap between the surface _6t2R and the semiconductor element at the time of wire bonding. However, even when there is no gap, there is an effect of extending the power cycle life after joining, and therefore, bending may be performed to the extent that there is no gap.

As described above, the electrode member 6 (16, 26, 36, 46) according to the first embodiment of the present invention is an electrode member for electrically connecting the semiconductor element 3 and the wiring member 4, The first metal plate 6 in the length direction of the band-shaped first metal plate 6 _{MC in} the length direction is a band-shaped material 6 _{B in} which the second metal plate 6 _MB is bonded to a region of a predetermined length from the _E1 side, and the second metal plate 6 _MB is The first metal plate 6 _MC is located outside the second metal plate 6 _MB by bending a portion on the other end 6 _E2 side of the bonded region 6 _FB so that the first metal plate 6 _MC is located outside the second metal plate 6 _MB. In the formed surface, the bonding surface 6 _t1 to the semiconductor element 3 is in a region having a predetermined length from the one end 6 _E1 side to the bent portion 6 _C, and a region having a predetermined length from the other end 6 _E2 side to the bent portion 6 _C. Is provided with a joint surface _6t2 with the wiring member 4, and the first metal plate _6MC The coefficient of linear expansion close to the coefficient of linear expansion of the wiring member 4 than the linear expansion coefficient of the semiconductor element 3, the linear expansion coefficient of the belt-shaped member 6 _B of the second metal plate 6 _MB is glued area 6 _FB is wiring Since the configuration is such that the linear expansion coefficient of the semiconductor element 3 is closer than the linear expansion coefficient of the member 4, the bonding surface 6 _t1 with the semiconductor element 3 is close to the linear expansion coefficient of the semiconductor element 3 and is bonded to the wiring member 4. higher maximum temperature because the surface 6 _t2 is close to the linear expansion coefficient of the wiring member 4, even if the temperature variation is increased, the power cycle life increased dramatically, moreover, the first metal plate 6 _MC overall length Are exposed to the outside where the joint surface 6 _t1 and the joint surface 6 _t2 are formed, so that current flows between the joint surfaces without passing through the joint surface of the laminated structure, which is a joint interface of dissimilar metals in the band-shaped material. Since it was allowed to flow, It not affected even if the change in conductivity due to thermal stress, it is possible to obtain a highly reliable semiconductor device with high efficiency. In addition, since the region where the second metal plate 6 _MB is bonded is not bent, stress during bending may be applied to the interface between the first metal plate 6 _MC and the second metal plate 6 _MB. Therefore, the bonding strength can be maintained.

In particular, in the region 16 _FB in which the second metal plate 16 _MB of the band-shaped material 16B is bonded, the third metal plate 16 _MC2 having the same linear expansion coefficient as the first metal plate 16 _MC1 is _replaced with the second metal plate 16. _Since it is stuck on the _MB , it does not warp like a bimetal when the temperature changes, and stable bonding can be maintained.

Furthermore, there is a gap of a predetermined distance between the surface 6 _{t1R on} the back side of the bonding surface 6 _t1 to the semiconductor element 3 and the surface 6 _{t2R on} the back side of the bonding surface 6 _t2 to the wiring member 4 in the strip 6B. As a result, the second bonding surface _6t2 faces the same direction as the bonding surface of the semiconductor element 3, and the same wire bonder is used regardless of the presence or absence of the electrode member 6 or the same method without using a special bonding method. Wiring members can be joined by a joining method. Further, ultrasonic vibration during wire bonding is not directly transmitted to the semiconductor element 3, and strong bonding is possible with minimal influence on the semiconductor element.

Furthermore, the width W _E2 of the other end 36 _E2 side of the strip material 36B is. Thus wider than the width W _E1 of the one end 36 _E1 side, such as a wire without the size limitations of the electrode surface of the semiconductor element By increasing the number of wiring members or increasing the size of the lead material, the current density in the wiring members can be reduced, heat generation can be suppressed, and the life can be further improved.

A first metal plate 6 _MC, the same as the main constituent material of the wiring member such as wire bonding and lead material, for example, when made of aluminum or copper, and the second joint surface 6t2 and wiring member 4 are joined in the same material As a result, a strong bond capable of reducing the influence of thermal stress as much as possible can be obtained.

Embodiment 2. FIG.
In the electrode member according to the second embodiment, the space formed inside the U shape of the electrode member according to the first embodiment is filled with resin. Other components such as application to the semiconductor device 10 are the same as those in the first embodiment, and thus description thereof is omitted.

FIG. 6 shows a cross section of the electrode member 216 according to the second embodiment, and the U-shaped portion of the electrode member is the same as that of the electrode member 16 according to the first modification of the first embodiment. Then, the resin pellets void filled with filler _{9 F} in between the back surface _{216 T2R} backside _{216 T1R} and second mating surface _{216 t2} of the first joint surface _{216 t1} is an inner space of the U-shaped electrode member 216 9 is inserted. Resin pellets 9 contains ceramic particles such as metal particles or resin particles or silica as the filler 9 _F. The resin 9 _R may be a thermosetting resin such as a heat-resistant epoxy resin, and is molded as a pellet having a shape that can be inserted into the space below the wire bonding surface 216 _t2 . The resin is preliminarily set in a semi-cured state, and after being inserted under the wire bonding surface, is pressurized and heated to adhere (completely cure) the back surface 216 _t1R and the back surface 216 _t2R . At this time, the presence of the filler 9 _F, by the effect which had been semi-cured prior to pressurization, the deformation of the pellets 9 itself is suppressed, it is possible to maintain an approximately adhesion previous shape. Needless to say, the pressure applied at the time of bonding is appropriately adjusted within a range in which the pellets 9 are not crushed. Also, among the space inside the U-shaped electrode member 216 in the figure, the resin 9 on the semi-circular portion in contact with the corner portion 216 _C is not filled, the space of the semi-circular portion may be filled.

  Thus, the configuration of the electrode member 216 as shown in FIG. 6 is completed, but this step may be before or after the step of bonding the electrode member 216 to the semiconductor element 3. In the latter case, the resin 9 needs to be a material that maintains adhesion even after a high-temperature process (heating for melting the solder) at the time of solder joining between the semiconductor element 3 and the electrode member 216.

When the electrode member 216 according to the second embodiment is used, the other end portion 216 _E2 provided with the wire bonding surface 216 _t2 is fixed by the resin 9 at the time of wire bonding, so that even when the plate member of the electrode member 216 is thin There is no need to use a vacuum chuck table or holding claws. The flatness and horizontality of the wire bonding surface 216 _t2 necessary for the wire bonding may be ensured by a necessary tolerance by adjusting the pressure mechanism in the bonding process after the resin pellet 9 is inserted.

In the configuration of the electrode member 216 according to the second embodiment, the resin 9 deteriorates due to the holding at a high temperature by the operation and the power cycle load, and the inner surface of the resin 9 and the electrode member 216 (216 _t1R , 216 Separation may proceed on the adhesive surface with _t2R ). However, the resin 9 is intended for fixing at the time of wire bonding, and the entire semiconductor device 10 is fixed by the mold resin and does not move. Therefore, even if peeling occurs after that, Does not affect connection reliability. Therefore, the effect of extending the power cycle life of the wire joint can be obtained as in the first embodiment.

As described above, according to the electrode member 216 according to the second embodiment of the present invention, the back surface 216 _t1R of the bonding surface 216 _t1 to the semiconductor element 3 and the back surface of the bonding surface 216 _t2 to the wiring member 4 are provided. Since the resin pellet 9 is filled in the gap with the surface 216 _t2R , the other end 216 _E2 is fixed by the resin 9 at the time of wire bonding, so even if the plate material of the electrode member 216 is thin, the vacuum There is no need to use a chuck table or a holding claw, and it can be easily joined without using a special joining method.

Embodiment 3 FIG.
In the electrode member according to the third embodiment, the electrode member according to the second embodiment is filled with a solder material instead of the resin filled in the space formed inside the U-shape. Since other components such as application to the semiconductor device 10 are the same as those in the first and second embodiments, the description thereof is omitted.

FIG. 7 shows a cross-section of the electrode member 316 according to the third embodiment, and the U-shaped portion of the electrode member is the electrode according to the first modification of the first embodiment as in the second embodiment. This is the same member as the member 16. The gap between the back surface 316 _{t1R of} the first joint surface 316 _{t1 and} the back surface 316 _t2R of the second joint surface 316 _t2 that is the U-shaped inner space of the electrode member 316 is filled with the solder material 15. It is a feature. The solder material 15 may be the same material as or different from the solder 5 used for joining the circuit pattern 2 a and the semiconductor element 3 described in FIG. 1 and joining the semiconductor element 3 and the electrode member 6. However, it is necessary not to melt when the semiconductor device 10 is used. Before the wire 4 is bonded, it is supplied in advance under the wire bonding surface, heated and melted, and bonded to the back surface 316 _t1R and the back surface 316 _t2R . At this time, it goes without saying that the back surface 316 _t1R and the back surface 316 _t2R , which are joint surfaces with the solder 15, or nickel plating (not shown) is appropriately applied to the entire inside of the U shape.

In order to stay in the region where the solder 15 is to be joined at the time of heating and melting and to contact the joint surfaces 216 _t1R and 216 _t2R , it is necessary to use the effect of staying between the upper and lower joint surfaces due to the surface tension of the solder 15. In order to surely obtain the effect, it is desirable that the height H _S of the space below the wire bonding surface is 1.0 mm or less. Further, in order to prevent the wire bonding surface 316 _{t2 from} being deformed due to the shrinkage of the solder 15 during cooling, when the copper is used for the good conductive member 316 _MC , a thickness of 0.3 mm or more is secured at the other end 316 _E2 . It is necessary to use 0.5 mm or more. Also, among the space inside the U-shaped electrode member 316 in the figure, although the solder 15 in a semi-circular portion in contact with the corner portion 316 _C is not filled, the space of the semi-circular portion may be filled.

By using this configuration, the other end portion 316 _E2 provided with the wire bonding surface 316 _t2 is fixed by the solder 15 at the time of wire bonding, so there is no need to use a vacuum chuck table or a holding claw. In addition, although the crack progresses inside the solder 15 due to the power cycle load or the separation occurs at the joint interface, the current path required for the electrode member 316 is the good conductive member 316 _MC portion, and therefore the influence on the electrical characteristics. There are few. In addition to the crack propagation, the solder material may cause a creep phenomenon, but the influence on the semiconductor element junction is small. Further, since the linear expansion coefficient of the solder 15 is close to that of aluminum, the effect of extending the power cycle life of the wire joint can be obtained as in the first embodiment.

As described above, according to the electrode member according to the third embodiment of the present invention, the back surface 316 _t1R of the bonding surface 316 _t1 with the semiconductor element 3 and the back surface of the bonding surface 316 _t2 with the wiring member 4 316 _{Since the} solder 15 is filled in the gap with the _t2R , the other end portion 316 _E2 is fixed by the solder 15 at the time of wire bonding. Therefore, even when the plate member of the electrode member 316 is thin, the vacuum chuck table It is not necessary to use a presser claw and can be easily joined without using a special joining method.

  In each of the above embodiments, the semiconductor element 3 functioning as a switching element (transistor) or a rectifying element (diode) is shown as being formed of silicon carbide, but is not limited thereto. It may be formed of silicon (Si) that is generally used. However, when using silicon carbide, gallium nitride-based material, or diamond that can form a so-called wide gap semiconductor having a larger band gap than silicon, the effects of the present invention can be further exhibited as described below. Can do.

  Since switching elements and rectifier elements (semiconductor elements 3 in each embodiment) formed of wide band gap semiconductors have lower power loss than elements formed of silicon, higher efficiency can be achieved in switching elements and rectifier elements. As a result, the power semiconductor device 10 can be made highly efficient. Furthermore, since the withstand voltage is high and the allowable current density is high, the switching element and the rectifying element can be downsized. By using the downsized switching element and rectifying element, the power semiconductor device 10 is also small. Can be realized. Further, since the heat resistance is high, it is possible to operate at a high temperature, and it is possible to reduce the size of the heat dissipating fins of the heat sink and the air cooling of the water-cooled portion, so that the power semiconductor device 10 can be further reduced in size.

  On the other hand, when operating at a high temperature as described above, the temperature difference during stop and drive increases, and the amount of current handled per unit volume increases due to high efficiency and downsizing. Therefore, the temperature change with time and the spatial temperature gradient increase, and the thermal stress between the semiconductor element and the wiring member may also increase. However, one end side where the bonding surface of the electrode member with the semiconductor element is arranged like the present invention has a linear expansion coefficient closer to the semiconductor element, and the other end where the bonding surface with the wiring member such as a wire is arranged. Since the coefficient of linear expansion is closer to the wiring member on the side, the thermal stress at each joint is relieved, so the power cycle life can be improved even if miniaturization and higher efficiency are promoted by taking advantage of the characteristics of wide band gap semiconductors. Therefore, it is easy to obtain the power semiconductor device 10 having a long and high reliability. That is, by exhibiting the effect of the present invention, the characteristics of the wide band gap semiconductor can be utilized.

  Note that both the switching element and the rectifying element may be formed of a wide band gap semiconductor, or one of the elements may be formed of a wide band gap semiconductor. Different materials may also be used for the wiring members such as wires and leads. In that case, if the linear expansion coefficient on one end side and the other end side is changed in accordance with the type and material of the element and wiring member, that is, in accordance with the linear expansion coefficient of the semiconductor element and wiring member, the power cycle life is further increased. Can be improved.

DESCRIPTION OF SYMBOLS 1 Insulation board | substrate (1f circuit surface), 2 Circuit pattern (2a, 2b), 3 Semiconductor element, 4 Wire (wiring member), 5,15 Solder (5a, 5b),
6 Electrode member (6B belt-like material), 7 Vacuum chuck table, 8 Holding claw, 9 Filling resin, 10 Power semiconductor device
Subscript FB: region where second metal plate is bonded (buffer region), MC: good conductive member (first metal plate), MB: low expansion member (second metal plate), t1: first Bonding surface, t2: second bonding surface, E1: one end, E2: other end The tenth digit difference indicates a modification in the first embodiment, and the hundredth digit indicates a difference in configuration according to the embodiment. .

Claims (11)

An electrode member for electrically connecting the semiconductor element and the wiring member,
The other end side of the band-shaped material in which the second metal plate is bonded to the region of a predetermined length from the one end side in the length direction of the band-shaped first metal plate, with respect to the region where the second metal plate is bonded The first metal plate is bent into a U-shape that is located outside the second metal plate,
On the surface that is the outer side of the U-shape, a bonding surface with the semiconductor element is in a region of a predetermined length from the one end side to the bent portion, and a region of a predetermined length from the other end side to the bent portion. A joint surface with the wiring member is provided,
The linear expansion coefficient of the first metal plate is closer to the linear expansion coefficient of the wiring member than the linear expansion coefficient of the semiconductor element,
The linear expansion coefficient of the strip in the region where the second metal plate is bonded is closer to the linear expansion coefficient of the semiconductor element than the linear expansion coefficient of the wiring member,
An electrode member. The front Symbol opposite the first surface that is glued to the metal plate of the second metal plate, that the third metal plates of the same linear expansion coefficient as the first metal plate is combined Ri Zhang The electrode member according to claim 1.   3. The gap between the back surface of the bonding surface with the semiconductor element and the back surface of the bonding surface with the wiring member in the belt-shaped material is a predetermined gap. An electrode member according to 1.   The electrode member according to claim 3, wherein a width of the belt-like material on the other end side is wider than a width on the one end side.   The electrode member according to claim 3 or 4, wherein a resin pellet is filled in the gap of the predetermined interval.   The electrode member according to claim 3 or 4, wherein a gap between the predetermined intervals is filled with a solder material. A semiconductor element mounted on a circuit pattern formed on an insulating substrate;
The electrode member according to any one of claims 1 to 6, wherein a bonding surface with the semiconductor element is bonded to a surface opposite to a mounting surface of the semiconductor element with the circuit pattern.
A wiring member joined to a joint surface of the electrode member with the wiring member;
A semiconductor device comprising: A difference between the linear expansion coefficient of the region where the second metal plate of the strip-shaped material is bonded and the linear expansion coefficient of the semiconductor element;
The difference between the linear expansion coefficient of the first metal plate and the linear expansion coefficient of the wiring member is
8. The semiconductor device according to claim 7, wherein both are 8 ppm / K or less within the operating temperature range of the semiconductor device.   The semiconductor device according to claim 7, wherein the first metal plate is made of the same metal material as a main constituent material of the wiring member.   10. The semiconductor device according to claim 7, wherein the semiconductor element is made of a wide band gap semiconductor material.   The semiconductor device according to claim 10, wherein the wide band gap semiconductor material is any one of silicon carbide, gallium nitride, and diamond.

Images