Power conversion device and method for manufacturing same

Abstract

A power conversion device (1000) according to the present disclosure comprises: a power module (100) that has a circuit board (200) in which a circuit pattern (201) is formed on the upper surface of an insulating layer (202) and a metal layer (203) is formed on the lower surface of the insulating layer (202) such that the insulating layer (202) is interposed between the circuit pattern (201) and the metal layer (203), and that has a semiconductor element (10) mounted on the circuit board (200); and a cooler (1) that is bonded to the metal layer (203) via a bonding member (7). The cooler (1)-side surface of the circuit board (200) is a curved surface that protrudes toward the cooler (1) side, and the bonding member (7) is thinnest at the position where the cooler (1)-side surface of the circuit board (200) is closest to the cooler (1) side. The thinnest portion (300) of the bonding member is at a position separated from the center position of the metal layer (203) when the power conversion device (1000) is viewed in a plan view. Furthermore, the semiconductor element (10) is disposed on the circuit board (200) on an extending line (301) drawn from the thinnest portion of the bonding member in the thickness direction thereof.

Description

Power conversion device and manufacturing method thereof

[0001] The present disclosure relates to a heat dissipation structure for a power conversion device and a manufacturing method thereof.

[0002] In conventional power conversion devices, in order to achieve both high reliability and high heat dissipation, the solder joint between the circuit board mounting the semiconductor device and the cooler is configured by warping the circuit board surface so that it forms a convex curve toward the cooler, and the solder layer is thin at the center of the board and thick at the periphery of the board. This is thought to improve heat dissipation in the center of the board where the solder layer is thin, and to extend the life of the solder at the periphery of the board where the solder layer is thick due to stress relaxation.

[0003] JP 2008-227336 A

[0004] In such a power conversion device, the thermal resistance is low in the area where the solder layer is thin in the center of the board, but is high in the area where the solder layer is thick around the periphery of the board. Therefore, when it is desired to place a semiconductor element around the periphery of the board where the solder layer is thick, there is a problem that the temperature of the semiconductor element becomes high during operation due to the high thermal resistance.

[0005] The power conversion device according to the present disclosure includes a power module having a semiconductor element mounted on a circuit board, and a cooler that dissipates heat generated by the power module. The circuit board has an insulating layer sandwiched between the insulating layer and a circuit pattern formed on the upper surface of the insulating layer, and a metal layer formed on the lower surface of the insulating layer. The cooler is joined to the metal layer, which is a component of the circuit board, via a bonding member. The surface of the circuit board facing the cooler is convexly curved toward the cooler, and the bonding member is thinnest at the position where the surface of the circuit board facing the cooler is closest to the cooler. The thinnest part of the bonding member is located away from the center of the metal layer when the power conversion device is viewed from above. The semiconductor element is further arranged on the circuit board on an extension line drawn in the thickness direction from the thinnest part of the bonding member.

[0006] In addition, the manufacturing method of a power conversion device according to the present disclosure is a manufacturing method of a power conversion device including a power module having a circuit board in which a circuit pattern is formed on the upper surface of an insulating layer sandwiched between the insulating layer and a metal layer is formed on the lower surface of the insulating layer, a semiconductor element mounted on the circuit board, and a cooler joined to the metal layer via a joining member, and includes a step of forming the cooler-side surface of the circuit board so as to be a convex curved surface toward the cooler, and a step of arranging the semiconductor element on the circuit board on an extension line drawn in the thickness direction from a portion other than the central part of the metal layer, wherein the joining member is thinnest at the position where the cooler-side surface of the circuit board is closest to the cooler side, and the thinnest part of the joining member is located away from the central position of the metal layer when the power conversion device is viewed in a plane.

[0007] According to the configuration of the present disclosure, even if the semiconductor element group is arranged in a location other than the center of the circuit board surface, it is possible to mitigate the temperature rise of the semiconductor elements during operation.

[0008] 1 is an oblique external view of the power converter according to the first embodiment; FIG. 2 is an oblique external view of the power converter according to the first embodiment, with the mold resin 2 shown in FIG. 1 not shown; FIG. 3 is an enlarged view of the first circuit pattern 201a shown in FIG. 2 for the power converter according to the first embodiment; FIG. 4 is a cross-sectional view of the power converter according to the first embodiment, taken along the line A-A in FIG. 2; FIG. 5 is an oblique view of the power converter according to the first embodiment, with only the metal layer 203 and the bonding member 7 shown; 13 is an enlarged view of a portion of the circuit pattern 201 in FIG. 13 for the power conversion device according to the third embodiment. FIG. 14 is a perspective view of the power conversion device according to the third embodiment, showing a center point 600 of the semiconductor element group. FIG. 15 is a cross-sectional view of the power conversion device according to the fourth embodiment. FIG. 16 is a cross-sectional view of the power conversion device according to the fifth embodiment.

[0009] Hereinafter, embodiments will be described in detail with reference to the drawings. Note that the embodiments described below are merely examples, and the scope of the present disclosure is not limited to the embodiments described below.

[0010] 1 is an external view of a power conversion device according to a first embodiment of the present disclosure. A power module 100 is disposed on an upper surface of a cooler 1. The interior of the power module 100 is covered with a molded resin 2, which is a component thereof. Terminals 3 protrude from the molded resin 2.

[0011] FIG. 2 is an oblique external view of the power conversion device according to the first embodiment, with the mold resin 2 not shown. Inside the power module 100, there is a circuit board 200 on which a semiconductor element 4 is mounted. From the oblique external view of FIG. 2, among the components of the circuit board 200, a circuit pattern 201 and an insulating layer 202 can be seen. The circuit pattern 201 is composed of a first circuit pattern 201a and a second circuit pattern 201b, each of which has two semiconductor elements 4 attached thereto. The semiconductor element 4 is a vertical element, and its two plate-shaped main surfaces serve as electrodes for carrying a main current. The terminals 3 are electrically connected to the circuit pattern 201 or the semiconductor element 4. Two of the terminals 3 are bonded to the circuit pattern 201, and the remaining terminal is connected to the semiconductor element 4 by a wiring member 5, which is a wire. A component 6 is also attached to the circuit pattern 201.

[0012] The arrangement, types, and numbers of the components described above are merely examples and do not limit the scope of the present embodiment.

[0013] The cooler 1 has a function of dissipating heat generated inside the power module 100 during operation to the outside, and also functions as part of the housing. The cooler 1 is made of, for example, aluminum, aluminum alloy, copper, copper alloy, or metal matrix composite (MMC). An example of MMC is AlSiC.

[0014] The mold resin 2 has the function of sealing and protecting the terminals 3, semiconductor elements 4, wiring members 5, and circuit board 200 inside the power module 100. The mold resin 2 is typically an epoxy resin containing silica filler.

[0015] The terminals 3 are used for electrical connection between the inside and outside of the power module 100. The terminals 3 are made of a metal material with good conductivity, such as copper.

[0016] The semiconductor element 4 is a switching element such as an IGBT or a MOSFET, or a rectifying element (diode), and has metal such as aluminum attached to its main surface as an electrode. In the case of a power device, it is often a vertical device in which current flows perpendicular to the main surface of the element. It is connected to the circuit board 200 by a bonding material (not shown) such as solder. The semiconductor element 4 is made of a material such as Si, SiC, GaN, or Ga2O3.

[0017] The wiring member 5 is a wire made of aluminum or copper. Note that the wiring member is not limited to a wire, and may be a plate-like member such as a ribbon or a copper plate.

[0018] The components 6 include a thermistor, a resistor and a capacitor for forming a snubber circuit, and the like.

[0019] In the circuit board 200, a circuit pattern 201 is formed on the upper surface of the insulating layer 202, with the insulating layer 202 sandwiched therebetween, and a metal layer 203 is formed on the lower surface of the insulating layer 202 (the metal layer 203 is not shown in FIGS. 1 and 2 ). The circuit pattern 201, the insulating layer 202, and the metal layer 203 are all required to have heat dissipation properties, which allow them to transfer and dissipate heat generated when the semiconductor element 4 operates to the cooler 1. The circuit pattern 201 also functions as a path for electrical signals and is made of a material such as copper or aluminum. The insulating layer 202 has the function of ensuring electrical insulation in addition to heat dissipation properties and is made of a ceramic material such as alumina, silicon nitride, or aluminum nitride, or a resin material. The metal layer 203 is formed on the insulating layer 202 on the opposite side of the circuit pattern 201 for heat dissipation purposes and is made of a material such as copper or aluminum.

[0020] FIG. 3 is an enlarged view of a portion of the first circuit pattern 201a of FIG. 2 for the power conversion device according to the first embodiment. Two semiconductor elements 4 are mounted on the first circuit pattern 201a. Because the two semiconductor elements 4 are formed side by side on the same circuit pattern, they are treated as a single semiconductor element group 10. The semiconductor element group 10 is depicted as a three-dimensional shape indicated by dotted lines. In this manner, the semiconductor element group 10 is defined as a three-dimensional shape defined by the outer edges of the semiconductor elements 4, which are constituent elements, when viewed as a single block. Note that, not limited to this embodiment, in the invention according to the present disclosure, if multiple semiconductor elements 4 form a structural unit on the same circuit pattern 201, they are treated as a semiconductor element group 10. In this case, the number of semiconductor elements 4 may be three or more.

[0021] Fig. 4 is a cross-sectional view of the power conversion device according to the first embodiment, taken along the line A-A in Fig. 2. Components not shown in the oblique external views of Figs. 1 and 2 include a metal layer 203 and a bonding member 7. The circuit board 200 has an insulating layer 202 sandwiched between them, with a circuit pattern 201 formed on the upper surface of the insulating layer 202 and a metal layer 203 formed on the lower surface of the insulating layer 202. The cooler 1 is bonded to the metal layer 203 (a component of the circuit board 200) via the bonding member 7.

[0022] The joining member 7 is required to have a heat dissipation property that allows the heat generated by the semiconductor element 4 and transmitted from the circuit board 200 to be further transmitted to the cooler 1 and dissipated. The joining member 7 can be made of a material such as solder, sintered silver, sintered copper, or a conductive adhesive.

[0023] As shown in FIG. 4 , the surface of the circuit board 200 facing the cooler 1 is curved convexly toward the cooler 1. Therefore, the thickness of the bonding member 7 bonding the circuit board 200 to the cooler 1 varies depending on the position, with the thickness being thinnest where the surface of the circuit board 200 facing the cooler 1 is most convex. In other words, the thinnest portion 300 of the bonding member is located at the position where the surface of the circuit board 200 facing the cooler 1 is closest to the cooler. This thinnest portion 300 of the bonding member (the area marked with a star in FIG. 4 ) is located in a different area from the central portion 400 of the metal layer (the area marked with a black circle in FIG. 4 ). The central portion of the metal layer refers to the area of ​​the metal layer in the thickness direction that passes through its geometric center point. To help understand this, FIG. 5 shows an oblique view of only the metal layer 203 and bonding member 7 in embodiment 1. The surface of the metal layer 203 facing the joining member 7 is a convex curved surface toward the joining member 7, and a central portion 400 of the metal layer is shown as the geometric center point of the metal layer 203. In other words, the thinnest portion 300 of the joining member is located away from the central position of the metal layer 203 when the power conversion device 1000 is viewed from above. Note that the dashed dotted lines (dashed dotted lines indicated by reference numerals 301 and 401) shown in Figures 4 and 5 are perpendicular lines drawn to the power conversion device 1000.

[0024] The semiconductor element group 10 is mounted on the circuit board 200 on an extension line 301 drawn in the thickness direction from the portion where the bonding material is thinnest (see FIG. 4 ). In the heat dissipation path from the semiconductor element group 10 to the cooler 1, the path passing through the portion 300 where the bonding material is thinnest in the thickness direction is the path with the smallest thermal resistance. The heat dissipation path refers to the path along which heat generated by the semiconductor element 4 or the semiconductor element group 10 is diffused through the circuit board 200 toward the cooler 1. In this embodiment, the portion 300 where the bonding material is thinnest is located in a portion other than the central portion 400 of the metal layer, so the thermal resistance is minimized in the path passing through the portion other than the central portion 400 of the metal layer. In this embodiment, by arranging the semiconductor element group 10 in a location on the circuit board surface where the thermal resistance is minimized, the temperature rise of each semiconductor element 4 during operation can be mitigated. In this embodiment, arranging the semiconductor element group 10 in a portion other than the central portion of the circuit board not only increases the layout flexibility of the semiconductor elements but also suppresses localized temperature rise of the power conversion device.

[0025] Considering the cooling efficiency as described above, the closer the semiconductor element group 10 is positioned on the circuit board surface to the path with the smallest thermal resistance (the path passing through the thinnest portion 300 of the bonding material in the thickness direction), the higher the cooling efficiency. Therefore, as shown in FIG. 4 , it is preferable that the distance between an extension line 301 drawn from the thinnest portion of the bonding material in the thickness direction and the center point 600 of the semiconductor element group is shorter than the distance between an extension line 401 drawn from the center of the metal layer in the thickness direction and the center point 600. The distance between each extension line and the center point can be calculated by the method for calculating the distance between a point and a line (the shortest distance between a point and any point on the line). Here, the center point 600 of the semiconductor element group is the point at which the line connecting the geometric centers 500 of the respective semiconductor elements 4 is equally divided. This definition is used because the temperature of each semiconductor element 4 is essentially designed to be uniform. To help understand this, FIG. 6 shows an oblique view of the center point 600 of the semiconductor element group in embodiment 1. Each of the two semiconductor elements 4 has a geometric center point, and the point at which the straight line created by connecting these center points equally divides the line is the center point of the semiconductor element group 10. In an aspect different from this embodiment, when the semiconductor element group 10 includes three or more semiconductor elements 4, a polygonal figure is created by connecting the geometric center points of the semiconductor elements 4. In this case, the center of the semiconductor element group 10 is the geometric center point of the polygon.

[0026] It is even more preferable that the center point 600 of the semiconductor element group is located on an extension line 301 drawn in the thickness direction from the thinnest portion of the bonding member. In this case, the cooling efficiency of the semiconductor element group 10 as a whole is further improved.

[0027] Some semiconductor element characteristics, such as on-resistance, exhibit temperature dependency, and if there is variation in the temperature of each semiconductor element, the on-resistance will also vary, causing a current imbalance. If the design is based on this assumption, the element area will increase, leading to increased costs. The configuration of this embodiment makes it possible to uniformize the heat dissipation of each semiconductor element 4, thereby reducing costs.

[0028] Although the power conversion device according to the present embodiment is a molded-type power module, the power conversion device according to the present disclosure may also be a case-type module as shown in Fig. 7. The case-type module has a structure in which a frame-shaped case 8 is provided around the circuit board 200 and a sealing material 9 such as silicone gel or potting resin is dropped instead of a molding resin to seal the circuit board 200. In the example of Fig. 7, the terminals 3 pass through the inside of the case 8 and protrude to the top surface of the case 8 to form a terminal block.

[0029] A method for manufacturing the power converter according to this embodiment will be described below.

[0030] Three aspects of manufacturing methods 1 to 3 will be described below. In manufacturing methods 1 to 3, plate-shaped wiring members 501 to 503 are used as wiring members to warp the circuit board 200 (to make the surface of the circuit board 200 on the cooler 1 side a convex curved surface toward the cooler 1 side). Note that the power conversion device according to this embodiment may be manufactured by methods other than these manufacturing methods, and the wiring members used are not limited to plate-shaped wiring members but may also be wires or ribbons. Furthermore, although each manufacturing method describes a case where there are multiple semiconductor elements 4, there may also be only one semiconductor element 4.

[0031] <Manufacturing Method 1> First, a method for manufacturing the circuit board 200 will be described. The circuit board 200 is manufactured by brazing an insulating layer 202 such as ceramic and a conductive material for the circuit pattern to a metal layer 203, and then etching the brazed conductive material to form the circuit pattern 201. The circuit board may be manufactured by methods other than those described above, such as forming the insulating layer 202 and the conductive layer that will become the circuit pattern 201 by thermal spraying.

[0032] Next, a plurality of semiconductor elements 4 (semiconductor element group 10) are bonded with a solder layer and mounted on the circuit board 200. The semiconductor element group 10 is arranged at a position off the center of the surface of the circuit board.

[0033] Furthermore, a plate-shaped wiring member 501 is bonded to the surface of the semiconductor element group 10 opposite the circuit board 200 side. FIG. 8 shows this positional relationship. In this case, it is assumed that the semiconductor elements 4 are connected in parallel. The semiconductor elements 4 may also be connected in series, in which case the plate-shaped wiring member 501 has a structure that insulates the terminals of the semiconductor elements 4. Specifically, a structure in which slits are provided in the portions to be insulated and these portions are filled with an insulator is conceivable. Furthermore, the plate-shaped wiring member 501 has a linear expansion coefficient greater than that of the entire circuit board 200. The bonding of the semiconductor elements 4 and the plate-shaped wiring member 501 is performed by the following steps S1 to S3.

[0034] First, in the bonding material application step of step S1, a sinterable metal bonding material is applied, for example, by screen printing, onto the plate-shaped wiring member 501. A printing mask and a squeegee are used for printing to apply the sinterable metal bonding material, and a paste-like sinterable metal bonding material is supplied to the surface of the printing mask, and the paste is applied onto the plate-shaped wiring member 501 by scraping it off.

[0035] Subsequently, in the mounting step of step S2, the surface of the plate-shaped wiring member 501 coated with the sinterable metal bonding material is placed on the surface of the semiconductor element group 10 opposite the circuit board 200 side.

[0036] In the pressure bonding process of step S3, the structure formed in the previous processes is heated to an appropriate temperature (e.g., 80°C, 30 minutes) and dried to remove the organic solvent component of the sinterable metal bonding material. Then, while pressing the semiconductor element 4 down to pressurize the sinterable metal bonding material, the structure is heated to the temperature required for bonding (e.g., 200°C to 350°C, 30 minutes), and pressure (e.g., 10 MPa) is applied to the structure. As a result, the sinterable metal bonding material sinter-bonds the bonding surfaces of the plate-shaped wiring member 501, the bonding surfaces of the semiconductor element 4, and the metal microparticles. After pressing, the temperature is lowered from the process temperature to room temperature.

[0037] After steps S1 to S3, the cooler 1 is joined to the surface of the circuit board 200 opposite the semiconductor element group 10 via a joining member 7 such as solder. Finally, the terminals 3 and components 6 are also joined to the circuit board 200 or the plate-like wiring member 501 by soldering, and then sealed with mold resin 2 by transfer molding. In this way, the power conversion device according to this embodiment is manufactured.

[0038] In step S3 described above, a process of lowering the temperature from the process temperature to room temperature occurs when the plate-shaped wiring member 501 is attached to the semiconductor element group 10. During this process, warping occurs in the circuit board 200, and the surface of the circuit board 200 facing the cooler 1 can be controlled to have a curved shape that is convex toward the cooler 1. Because the plate-shaped wiring member 501 has a larger linear expansion coefficient than the circuit board 200, the plate-shaped wiring member 501 shrinks more than the circuit board 200 during the temperature lowering process. This allows the shape of the warpage to be controlled by the amount of stress applied to both the circuit board 200 and the plate-shaped wiring member 501. Specifically, this can be achieved by adjusting the thicknesses of the circuit board 200 and the plate-shaped wiring member 501, or by adjusting the difference in the linear expansion coefficients of the two by selecting materials. This can be designed using a general structural analysis simulator.

[0039] Furthermore, because the plate-shaped wiring member 501 is positioned at a position offset from the center of the main surface of the circuit board 200 via the semiconductor element group 10 (see FIG. 8 ), the mounting in this positional relationship allows the center of warpage to be controlled to occur at a position offset from the center of the circuit board 200 during the cooling process from the process temperature. Because the plate-shaped wiring member 501 has a larger linear expansion coefficient than the circuit board 200, the plate-shaped wiring member 501 shrinks more than the circuit board 200 during the cooling process. The resulting stress on both the wiring member and the circuit board 200 can be used to control the position of warpage. Specifically, this can be achieved by adjusting the thickness of the circuit board 200 and the plate-shaped wiring member 501 or adjusting the difference in linear expansion coefficients between the two by selecting the appropriate materials. This can be designed using a general structural analysis simulator. Accordingly, the thinnest portion 300 of the bonding member can be formed at a position offset from the center 400 of the metal layer.

[0040] In this way, by using the plate-shaped wiring member 501 in addition to the circuit pattern 201 as a path for electrical signals, the surface of the circuit board 200 on the cooler 1 side can be locally warped during the manufacturing process so as to form a convex curved surface toward the cooler 1 side, and the power conversion device according to this embodiment can be manufactured.

[0041] <Manufacturing Method 2> First, a method for manufacturing the circuit board 200 will be described. The circuit board 200 is manufactured by brazing an insulating layer 202 such as ceramic and a conductive material for the circuit pattern to a metal layer 203, and then etching the brazed conductive material to form the circuit pattern 201. The circuit board may be manufactured by methods other than those described above, such as forming the insulating layer 202 and the conductive layer that will become the circuit pattern 201 by thermal spraying.

[0042] Next, a plurality of semiconductor elements 4 (semiconductor element group 10) are bonded with a solder layer and mounted on the circuit board 200. The semiconductor element group 10 is arranged at a position off the center of the surface of the circuit board.

[0043] Furthermore, the circuit board 200 and the plate-shaped wiring member 502 are joined so as to bridge the circuit pattern 201 on the circuit board 200. This positional relationship is shown in Figure 9. The plate-shaped wiring member 502 has two protrusions 512 for bridging and joining the circuit pattern 201. Here, the plate-shaped wiring member 502 has a linear expansion coefficient greater than that of the entire circuit board 200. The joining of the circuit board 200 and the plate-shaped wiring member 502 is carried out by the following steps S4 to S6.

[0044] First, in the bonding material application step of step S4, a sinterable metal bonding material is applied onto the protruding portion 512 of the plate-shaped wiring member by, for example, a screen printing method. A printing mask and a squeegee are used for printing to apply the sinterable metal bonding material, and a paste-like sinterable metal bonding material is supplied to the surface of the printing mask and then scraped off to apply the paste onto the protruding portion 512 of the plate-shaped wiring member.

[0045] Subsequently, in the mounting process of step S5, the protruding portion 512 of the plate-shaped wiring member is placed on the circuit board 200 from the surface coated with the sinterable metal bonding material so as to bridge the circuit pattern 201 on the circuit board 200.

[0046] In the pressure bonding process of step S6, the structure formed in the previous processes is heated to an appropriate temperature (e.g., 80°C, 30 minutes) and dried to remove the organic solvent component of the sinterable metal bonding material. Then, while pressing the semiconductor element 4 down to pressurize the sinterable metal bonding material, the structure is heated to the temperature required for bonding (e.g., 200°C to 350°C, 30 minutes), and pressure (e.g., 10 MPa) is applied to the structure. As a result, the sinterable metal bonding material sinter-bonds the bonding surfaces of the protrusions 512 of the plate-shaped wiring member, the bonding surfaces of the circuit pattern 201, and the metal microparticles. After pressing, the temperature is lowered from the process temperature to room temperature.

[0047] After steps S4 to S6, the cooler 1 is joined to the surface of the circuit board 200 opposite the semiconductor element group 10 via a joining member 7 such as solder. Finally, the terminals 3 and components 6 are also joined to the circuit board 200 or the plate-like wiring member 502 by soldering, and then sealed with mold resin 2 by transfer molding. In this way, the power conversion device according to this embodiment is manufactured.

[0048] In step S6 described above, a process of lowering the temperature from the process temperature to room temperature occurs when the plate-shaped wiring member 502 is attached to the circuit board 200. During this process, warping occurs in the circuit board 200, and the surface of the circuit board 200 facing the cooler 1 can be controlled to have a curved shape that is convex toward the cooler 1. Because the plate-shaped wiring member 502 has a larger linear expansion coefficient than the circuit board 200, the plate-shaped wiring member 502 shrinks more than the circuit board 200 during the temperature lowering process. This allows the shape of the warpage to be controlled by the amount of stress applied to both the circuit board 200 and the plate-shaped wiring member 502. Specifically, this can be achieved by adjusting the thicknesses of the circuit board 200 and the plate-shaped wiring member 502, or by adjusting the difference in the linear expansion coefficients of the two by selecting materials. This can be designed using a general structural analysis simulator.

[0049] Furthermore, by bridging the plate-shaped wiring member 502 at a position off the center of the main surface of the circuit board 200 (see FIG. 9 ), the center of warpage can be controlled to occur off the center of the circuit board 200 during the cooling process from the process temperature after mounting in this positional relationship. Because the plate-shaped wiring member 502 has a higher linear expansion coefficient than the circuit board 200, the plate-shaped wiring member 502 shrinks more than the circuit board 200 during the cooling process. The resulting stress on both the wiring member and the circuit board 200 can be used to control the position of warpage. Specifically, this can be achieved by adjusting the thickness of the circuit board 200 and the plate-shaped wiring member 502 or adjusting the difference in linear expansion coefficients between the two by selecting the right materials. This can be designed using a general structural analysis simulator. Accordingly, the thinnest portion 300 of the bonding member can be formed off the center 400 of the metal layer.

[0050] In this way, by using the plate-shaped wiring member 502 in addition to the circuit pattern 201 as a path for electrical signals, the surface of the circuit board 200 on the cooler 1 side can be locally warped during the manufacturing process so as to form a convex curved surface toward the cooler 1 side, and the power conversion device according to this embodiment can be manufactured.

[0051] <Manufacturing Method 3> First, a method for manufacturing the circuit board 200 will be described. The circuit board 200 is manufactured by brazing an insulating layer 202 such as ceramic and a conductive material for the circuit pattern to a metal layer 203, and then etching the brazed conductive material to form the circuit pattern 201. The circuit board may be manufactured by methods other than those described above, such as forming the insulating layer 202 and the conductive layer that will become the circuit pattern 201 by thermal spraying.

[0052] Next, a plate-shaped wiring member 503 is bonded to this circuit board 200. The plate-shaped wiring member 503 is bonded to a position that is off the center of the circuit board surface. Here, the plate-shaped wiring member 503 has a linear expansion coefficient greater than that of the entire circuit board 200. Bonding of the circuit board 200 and the plate-shaped wiring member 503 is carried out by the following steps S7 to S9.

[0053] First, in the bonding material application step of step S7, for example, a sinterable metal bonding material is applied by screen printing onto the plate-shaped wiring member 503. A printing mask and a squeegee are used for printing to apply the sinterable metal bonding material, and a paste-like sinterable metal bonding material is supplied onto the surface of the printing mask, and the paste is applied onto the plate-shaped wiring member 503 by scraping it off.

[0054] Subsequently, in the mounting step of step S8, the plate-shaped wiring member 503 is placed on the surface of the circuit board 200 with the surface coated with the sinterable metal bonding material facing up.

[0055] In the pressure bonding process of step S9, the structure formed in the previous processes is heated to an appropriate temperature (e.g., 80°C, 30 minutes) and dried to remove the organic solvent component of the sinterable metal bonding material. Then, while pressing the semiconductor element 4 down to pressurize the sinterable metal bonding material, the structure is heated to the temperature required for bonding (e.g., 200°C to 350°C, 30 minutes), and pressure (e.g., 10 MPa) is applied to the structure. As a result, the sinterable metal bonding material sinter-bonds the bonding surfaces of the plate-like wiring member 503, the bonding surfaces of the circuit board 200, and the metal microparticles to each other. After pressing, the temperature is lowered from the process temperature to room temperature.

[0056] Thereafter, a plurality of semiconductor elements 4 (semiconductor element group 10) are bonded to the plate-like wiring member 503 with a solder layer and mounted. This positional relationship is shown in FIG. 10. In this case, it is assumed that the semiconductor elements 4 are connected in parallel. The semiconductor elements 4 may also be connected in series, in which case the plate-like wiring member 503 has a structure that insulates the terminals of the semiconductor elements 4. Specifically, a structure in which slits are provided in the portions to be insulated and the portions are filled with an insulator is conceivable.

[0057] After steps S7 to S9, the cooler 1 is joined to the surface of the circuit board 200 opposite the semiconductor element group 10 via a joining member 7 such as solder. Finally, the terminals 3 and components 6 are also joined to the circuit board 200 or the plate-like wiring member 503 by soldering, and then sealed with mold resin 2 by transfer molding. In this way, the power conversion device according to this embodiment is manufactured.

[0058] In step S9 described above, a process of lowering the temperature from the process temperature to room temperature occurs when attaching the plate-shaped wiring member 503 to the circuit board 200. During this process, warping occurs in the circuit board 200, and the surface of the circuit board 200 facing the cooler 1 can be controlled to have a curved shape that is convex toward the cooler 1. Because the plate-shaped wiring member 503 has a larger linear expansion coefficient than the circuit board 200, the plate-shaped wiring member 503 shrinks more than the circuit board 200 during the temperature lowering process. This allows the shape of the warpage to be controlled by the amount of stress applied to both the circuit board 200 and the plate-shaped wiring member 503. Specifically, this can be achieved by adjusting the thicknesses of the circuit board 200 and the plate-shaped wiring member 503, or by adjusting the difference in the linear expansion coefficients of the two by selecting materials. This can be designed using a general structural analysis simulator.

[0059] Furthermore, because the plate-shaped wiring member 503 is positioned at a position offset from the center of the main surface of the circuit board 200 via the semiconductor element group 10 (see FIG. 10 ), the mounting in this positional relationship allows the center of warpage to be controlled to occur at a position offset from the center of the circuit board 200 during the cooling process from the process temperature. Because the plate-shaped wiring member 503 has a larger linear expansion coefficient than the circuit board 200, the plate-shaped wiring member 503 shrinks more than the circuit board 200 during the cooling process. The resulting stress on both the wiring member and the circuit board 200 can be used to control the position of the warpage. Specifically, this can be achieved by adjusting the thickness of the circuit board 200 and the plate-shaped wiring member 503 or adjusting the difference in linear expansion coefficients between the wiring member and the circuit board 200 by selecting the appropriate material. This can be designed using a general structural analysis simulator. Accordingly, the thinnest portion 300 of the bonding member can be formed at a position offset from the center 400 of the metal layer.

[0060] In this way, by using the plate-shaped wiring member 503 in addition to the circuit pattern 201 as a path for electrical signals, the surface of the circuit board 200 on the cooler 1 side can be locally warped during the manufacturing process so as to form a convex curved surface toward the cooler 1 side, and the power conversion device according to this embodiment can be manufactured.

[0061] Furthermore, in the power conversion device manufactured by this manufacturing method, the plate-shaped wiring member 503 is attached between the semiconductor element 4 and the circuit board 200, so the heat generated by the semiconductor element 4 is diffused in the in-plane direction of the plate-shaped wiring member 503 and follows the heat dissipation path, thereby achieving the effect of further reducing thermal resistance.

[0062] A power converter according to a comparative example will be described in comparison with the power converter according to the present embodiment. FIG. 11 is a cross-sectional view of the power converter according to the comparative example. A circuit board 210 has an insulating layer 212 sandwiched therebetween, with a circuit pattern 211 formed on the upper surface of the insulating layer 212 and a metal layer 213 formed on the lower surface of the insulating layer 212. A cooler 21 is bonded to the metal layer 213 (a component of the circuit board 210) via a bonding member 27. A semiconductor element 24 is mounted on the circuit board 210. The surface of the circuit board 210 facing the cooler 21 is curved convexly toward the cooler 21 side and has a warp. The warp of the circuit board 210 can be achieved by making the circuit pattern 211 and the metal layer 213 different thicknesses or by using different materials for the circuit pattern 211 and the metal layer 213. Thermal stress generated in the circuit board 210 causes the circuit board 210 to warp toward the metal layer 213. In this case, because the thermal stress occurring in the circuit board 210 is isotropic, the center of the warp of the circuit board 210 will be formed in the central portion 410 of the metal layer (the area marked with a ● in FIG. 11 ). Accordingly, the portion 310 where the bonding member is thinnest (the area marked with a ☆ in FIG. 11 ) will also be formed in the central portion 410 of the metal layer.

[0063] In the power converter according to the comparative example, the heat dissipation path from the semiconductor element 24 to the cooler 21 has the smallest thermal resistance when passing through the portion 310 of the bonding material in the thickness direction. The heat dissipation path refers to the path along which heat generated by the semiconductor element 24 is diffused through the circuit board 210 toward the cooler 21. In the power converter according to the comparative example, the portion 310 of the bonding material in the thinnest direction is located in the central portion 410 of the metal layer, so the thermal resistance is minimized when passing through the central portion 410 of the metal layer. Therefore, in order to prevent the temperature from rising due to the heat generated by the semiconductor element 24, the semiconductor element 24 must be located in the central portion of the main surface of the circuit board 210 (on the extension of the central portion 410 of the metal layer in the thickness direction), which limits the layout. On the other hand, if the semiconductor element 24 were located outside the central portion of the main surface of the circuit board 210, the temperature would rise locally in the power converter.

[0064] Second Embodiment FIG. 12 is a cross-sectional view of a power conversion device according to the second embodiment. In the power conversion device according to the second embodiment, one semiconductor element 4 is mounted on a circuit board 200. The circuit board 200 has an insulating layer 202 sandwiched between them, with a circuit pattern 201 formed on the upper surface of the insulating layer 202 and a metal layer 203 formed on the lower surface of the insulating layer 202. The cooler 1 is joined to the metal layer 203 (a component of the circuit board 200) via a joining member 7. The surface of the circuit board 200 facing the cooler 1 is curved convexly toward the cooler 1 side. Therefore, the thickness of the joining member 7 joining the circuit board 200 and the cooler 1 varies depending on the position, and is thinnest at the most convex portion of the surface of the circuit board 200 facing the cooler 1. In other words, the thinnest portion 300 of the joining member is located at the position where the surface of the circuit board 200 facing the cooler 1 is closest to the cooler. The thinnest portion 300 of the bonding material (the area marked with a star in FIG. 12 ) is located in a different area from the central portion 400 of the metal layer (the area marked with a ● in FIG. 12 ). Here, the central portion of the metal layer refers to a region of the metal layer in the thickness direction that passes through its geometric center point. This definition is the same as that described in FIG. 5 in the first embodiment. The semiconductor element 4 is disposed on the circuit board 200 on an extension line 301 drawn in the thickness direction from the thinnest portion of the bonding material. In other words, the thinnest portion 300 of the bonding material is located away from the center of the metal layer 203 when the power conversion device 1000 is viewed from above. The dashed-dotted lines in FIG. 12 (the dashed-dotted lines indicated by symbols 301 and 401) are perpendicular lines drawn to the power conversion device 1000.

[0065] In the heat dissipation path from the semiconductor element 4 to the cooler 1, the path passing through the thinnest portion 300 of the bonding member in the thickness direction is the path with the smallest thermal resistance. The heat dissipation path refers to the path along which heat generated by the semiconductor element 4 is diffused through the circuit board 200 toward the cooler 1. In this embodiment, the thinnest portion 300 of the bonding member is located in a portion other than the central portion 400 of the metal layer, so the thermal resistance is minimized in the path passing through the portion other than the central portion 400 of the metal layer. In this embodiment, by arranging the semiconductor element 4 in a location on the circuit board surface where the path with the smallest thermal resistance exists, it is possible to mitigate the rise in temperature during operation. Furthermore, in this embodiment, arranging the semiconductor element 4 in a portion other than the central portion of the circuit board not only increases the degree of freedom in the layout of the semiconductor element, but also suppresses localized rise in temperature of the power conversion device.

[0066] Furthermore, adjusting the position of the active area of ​​the semiconductor element 4 so that the extension line 301 passes through it further enhances cooling efficiency. The active area refers to the heat-generating region of the semiconductor element where structural components such as transistors and diodes are located. Cooling efficiency is further enhanced when the center point of the active area of ​​the semiconductor element is located on the extension line 301. Since the active area is a region that has a three-dimensional spatial extent as the heat-generating region of the semiconductor element, the center point of the active area refers to the geometric center point of this three-dimensional region. This arrangement not only improves the cooling efficiency of the semiconductor element, but also provides the additional benefit of reducing temperature variation within the chip surface. Note that recent power semiconductor elements, particularly Si-IGBTs, which are larger than 10 mm x 10 mm, benefit more from the effect of reducing temperature variation within the chip surface than smaller semiconductor elements.

[0067] Specific examples of the materials of the components in this embodiment are the same as those in the first embodiment.

[0068] 13 is a top view of a power conversion device according to embodiment 3. In the power conversion device according to embodiment 3, six semiconductor elements 4 are arranged on a circuit pattern 201.

[0069] 14 is an enlarged view of a portion of the circuit pattern 201 of FIG. 13 for the power conversion device according to the third embodiment. The six semiconductor elements 4 arranged on the circuit pattern 201 are formed side by side on the same circuit pattern, and are therefore treated as a single semiconductor element group 10. The semiconductor element group 10 is depicted as a three-dimensional shape indicated by dotted lines. In this manner, the semiconductor element group 10 is defined as a three-dimensional shape defined by the outer periphery of the semiconductor elements 4, which are constituent elements, when viewed as a single block as a whole.

[0070] 15 is a view of the power conversion device according to the third embodiment as viewed from the cross section B-B of FIG. 13 . The circuit board 200 has an insulating layer 202 sandwiched between them, with a circuit pattern 201 formed on the upper surface of the insulating layer 202 and a metal layer 203 formed on the lower surface of the insulating layer 202. The cooler 1 is joined to the metal layer 203 (a component of the circuit board 200) via a joining member 7. The surface of the circuit board 200 facing the cooler 1 is curved and convex toward the cooler 1. Therefore, the thickness of the joining member 7 joining the circuit board 200 and the cooler 1 varies depending on the position, and the thickness is thinnest at the most convex portion of the surface of the circuit board 200 facing the cooler 1. In other words, the thinnest portion 300 of the joining member is located at the position where the surface of the circuit board 200 facing the cooler 1 is closest to the cooler. The thinnest portion 300 of the bonding member (the area marked with a star in FIG. 15 ) is located in a different area from the central portion 400 of the metal layer (the area marked with a ● in FIG. 15 ). The central portion of the metal layer refers to a region of the metal layer in the thickness direction that passes through its geometric center point. This definition is the same as that described in FIG. 5 in the first embodiment. Furthermore, the semiconductor element group 10 is mounted on the circuit board 200 on an extension line 301 drawn in the thickness direction from the thinnest portion of the bonding member. In other words, the thinnest portion 300 of the bonding member is located away from the center of the metal layer 203 when the power conversion device 1000 is viewed from above. Note that the dashed dotted lines shown in FIG. 15 (the dashed dotted lines indicated by symbols 301 and 401) are perpendicular lines drawn to the power conversion device 1000.

[0071] In the heat dissipation path from the semiconductor element group 10 to the cooler 1, the path passing through the portion 300 where the bonding material is thinnest in the thickness direction is the path with the smallest thermal resistance. The heat dissipation path refers to the path along which heat generated by the semiconductor element 4 or the semiconductor element group 10 diffuses through the circuit board 200 toward the cooler 1. In this embodiment, the portion 300 where the bonding material is thinnest is located in a portion other than the central portion 400 of the metal layer, so the thermal resistance is minimized in the path passing through the portion other than the central portion 400 of the metal layer. In this embodiment, by arranging the semiconductor element group 10 in a location on the circuit board surface where the path with the smallest thermal resistance exists, it is possible to mitigate the rise in temperature of each semiconductor element 4 during operation. Furthermore, in this embodiment, arranging the semiconductor element group 10 in a portion other than the central portion of the circuit board increases the degree of freedom in the layout of the semiconductor elements and also suppresses localized rise in temperature of the power conversion device.

[0072] Considering the cooling efficiency as described above, the closer the semiconductor element group 10 is positioned on the circuit board surface to the path with the smallest thermal resistance (the path passing through the thinnest portion 300 of the bonding material in the thickness direction), the higher the cooling efficiency. Therefore, as shown in FIG. 15 , the distance between an extension line 301 drawn from the thinnest portion of the bonding material in the thickness direction and the center point 600 of the semiconductor element group is preferably shorter than the distance between an extension line 401 drawn from the center of the metal layer in the thickness direction and the center point 600. The distance between each extension line and the center point can be calculated using the method for calculating the distance between a point and a line (the shortest distance between a point and any point on the line is calculated). The center point 600 of the semiconductor element group is the geometric center point of a figure created by connecting the geometric centers 500 of the semiconductor elements. This definition is used because the temperature of each semiconductor element 4 is designed to be essentially uniform. To help understand this, FIG. 16 shows an oblique view of the center point 600 of the semiconductor element group in embodiment 3. The six semiconductor elements 4 are arranged in two rows of three, and a rectangular shape is created by connecting the geometric center points of each element. The geometric center point of this rectangle is the center point 600 of the semiconductor element group.

[0073] It is even more preferable that the center point 600 of the semiconductor element group is located on an extension line 301 drawn in the thickness direction from the thinnest portion of the bonding member. In this case, the cooling efficiency of the semiconductor element group 10 as a whole is further improved.

[0074] Some semiconductor element characteristics, such as on-resistance, exhibit temperature dependency, and if there is variation in the temperature of each semiconductor element, the on-resistance will also vary, causing a current imbalance. If the design is based on this assumption, the element area will increase, leading to increased costs. The configuration of this embodiment makes it possible to uniformize the heat dissipation of each semiconductor element 4, thereby reducing costs.

[0075] Specific examples of the materials of the components in this embodiment are the same as those in the first embodiment.

[0076] 17 is a cross-sectional view of a power conversion device according to embodiment 4. In the power conversion device according to embodiment 4, the surface of circuit board 200 on the cooler 1 side is curved with two protrusions facing the cooler 1 side, and therefore there are two portions 300 where the bonding member is thinnest.

[0077] Additionally, since there are two portions 300 where the bonding material is thinnest, there are also two extension lines 301 drawn in the thickness direction of the portions where the bonding material is thinnest. A plurality of semiconductor elements 4 are mounted on each of the two extension lines 301 on the circuit board 200. Each of the plurality of semiconductor elements 4 is treated as a semiconductor element group 10. That is, in this embodiment, there are two semiconductor element groups 10. The semiconductor element group 10 is defined as having a shape defined by the outer edge when the semiconductor elements 4, which are components, are viewed as a single block as a whole. This definition is as explained using the figures in the first and third embodiments.

[0078] The circuit board 200 has an insulating layer 202 sandwiched between them, with a circuit pattern 201 formed on the upper surface of the insulating layer 202 and a metal layer 203 formed on the lower surface of the insulating layer 202. The cooler 1 is bonded to the metal layer 203 (a component of the circuit board 200) via a bonding member 7. The surface of the circuit board 200 facing the cooler 1 is curved convexly toward the cooler 1. Therefore, the thickness of the bonding member 7 bonding the circuit board 200 to the cooler 1 varies depending on the position, and the thickness is thinnest where the surface of the circuit board 200 facing the cooler 1 is most convex. In other words, the thinnest portion 300 of the bonding member is located at the position where the surface of the circuit board 200 facing the cooler 1 is closest to the cooler. This thinnest portion 300 of the bonding member (the area marked with a star in FIG. 17 ) is located in a different area from the central portion 400 of the metal layer (the area marked with a black circle in FIG. 17 ). The central portion of the metal layer refers to a region of the metal layer in the thickness direction that passes through its geometric center point. This definition is the same as that described in FIG. 5 in the first embodiment. Furthermore, the semiconductor element group 10 is mounted on the circuit board 200 on an extension line 301 drawn in the thickness direction from the portion where the bonding material is thinnest. In other words, the thinnest portion 300 of the bonding material is located away from the center of the metal layer 203 when the power conversion device 1000 is viewed from above. Note that the dashed dotted lines shown in FIG. 17 (the dashed dotted lines indicated by symbols 301 and 401) are perpendicular lines drawn to the power conversion device 1000.

[0079] In the heat dissipation path from the semiconductor element group 10 to the cooler 1, the path passing through the portion 300 where the bonding material is thinnest in the thickness direction is the path with the smallest thermal resistance. The heat dissipation path refers to the path along which heat generated by the semiconductor element 4 or the semiconductor element group 10 diffuses through the circuit board 200 toward the cooler 1. In this embodiment, the portion 300 where the bonding material is thinnest is located in a portion other than the central portion 400 of the metal layer, so the thermal resistance is minimized in the path passing through the portion other than the central portion 400 of the metal layer. In this embodiment, by arranging the semiconductor element group 10 in a location on the circuit board surface where the path with the smallest thermal resistance exists, it is possible to mitigate the rise in temperature of each semiconductor element 4 during operation. Furthermore, in this embodiment, arranging the semiconductor element group 10 in a portion other than the central portion of the circuit board increases the degree of freedom in the layout of the semiconductor elements and also suppresses localized rise in temperature of the power conversion device.

[0080] Considering the cooling efficiency as described above, the closer the semiconductor element group 10 is positioned on the circuit board surface to the path with the smallest thermal resistance (the path passing through the thinnest portion 300 of the bonding material in the thickness direction), the higher the cooling efficiency. Therefore, as shown in FIG. 17 , it is preferable that the distance between an extension line 301 drawn from the thinnest portion of the bonding material in the thickness direction and the center point 600 of the semiconductor element group is shorter than the distance between an extension line 401 drawn from the center of the metal layer in the thickness direction and the center point 600. The distance between each extension line and the center point can be calculated by the method for calculating the distance between a point and a line (the shortest distance between a point and any point on the line is calculated). Furthermore, the center point 600 of the semiconductor element group is the geometric center point of a figure created by connecting the geometric centers 500 of the semiconductor elements. This is the same as in the third embodiment.

[0081] It is even more preferable that the center point 600 of the semiconductor element group is located on an extension line 301 drawn in the thickness direction from the thinnest portion of the bonding member. In this case, the cooling efficiency of the semiconductor element group 10 as a whole is further improved.

[0082] Some semiconductor element characteristics, such as on-resistance, exhibit temperature dependency, and if there is variation in the temperature of each semiconductor element, the on-resistance will also vary, causing a current imbalance. If the design is based on this assumption, the element area will increase, leading to increased costs. The configuration of this embodiment makes it possible to uniformize the heat dissipation of each semiconductor element 4, thereby reducing costs.

[0083] By using the configuration of this embodiment, the semiconductor element 4 or the semiconductor element group 10 can be arranged in a plurality of locations, thereby improving the degree of freedom in layout.

[0084] In this embodiment, the thinnest portions 300 of the joining member are two locations, but the thinnest portions 300 of the joining member may be three or more locations.

[0085] Specific examples of the materials of the components in this embodiment are the same as those in the first embodiment.

[0086] Fifth Embodiment Fig. 18 is a cross-sectional view of a power conversion device according to a fifth embodiment. In the power conversion device according to the fifth embodiment, two spacers 701 and 702 of different heights are arranged on a surface of the cooler 1 facing the circuit board 200, with the height of the spacer 701 being greater than the height of the spacer 702. By arranging the circuit board 200 on these spacers, the main surface of the circuit board 200 is inclined with respect to the main surface of the cooler 1. The main surface refers to a surface of a thin plate-shaped object (which includes both the cooler 1 and the circuit board 200) that has a much larger area than the other surfaces. In other words, the main surface of the circuit board 200 being inclined with respect to the main surface of the cooler 1 means that the main surfaces of the two are not parallel to each other.

[0087] Furthermore, a plurality of semiconductor elements 4 are mounted on the circuit board 200. These semiconductor elements 4 are treated as one semiconductor element group 10. The semiconductor element group 10 is defined as having a shape defined by the outer edge when the semiconductor elements 4, which are the components, are regarded as a single block as a whole. This definition is as explained using the drawings in the first and third embodiments.

[0088] The circuit board 200 has an insulating layer 202 sandwiched between them, with a circuit pattern 201 formed on the upper surface of the insulating layer 202 and a metal layer 203 formed on the lower surface of the insulating layer 202. The cooler 1 is bonded to the metal layer 203 (a component of the circuit board 200) via a bonding member 7. The surface of the circuit board 200 facing the cooler 1 is curved convexly toward the cooler 1. Therefore, the thickness of the bonding member 7 bonding the circuit board 200 to the cooler 1 varies depending on the position. In other words, the thinnest portion 300 of the bonding member is located at the position where the surface of the circuit board 200 facing the cooler 1 is closest to the cooler. This thinnest portion 300 of the bonding member (the area marked with a star in Figure 18 ) is located in a different area from the central portion 400 of the metal layer (the area marked with a black circle in Figure 18 ). The central portion of the metal layer refers to the area of ​​the metal layer in the thickness direction that passes through its geometric center point. This definition is the same as that explained in Fig. 5 in the first embodiment. Furthermore, the semiconductor element group 10 is mounted on the circuit board 200 on an extension line 301 drawn in the thickness direction from the thinnest part of the bonding member. In other words, the thinnest part 300 of the bonding member is located at a position away from the center position of the metal layer 203 when the power conversion device 1000 is viewed from above. Note that the dashed dotted lines shown in Fig. 18 (the dashed dotted lines indicated by symbols 301 and 401) are perpendicular lines drawn down to the power conversion device 1000.

[0089] According to the configuration of this embodiment, by using two spacers 701 and 702 of different heights to tilt the main surface of the circuit board 200 relative to the main surface of the cooler 1, unlike other embodiments, even if the center of the warp of the circuit board 200 is in the center of the circuit board surface as in the conventional case, the part 300 where the bonding member is thinnest can be formed in a part other than the center part 400 of the metal layer.

[0090] In the heat dissipation path from the semiconductor element group 10 to the cooler 1, the path passing through the portion 300 where the bonding material is thinnest in the thickness direction is the path with the smallest thermal resistance. The heat dissipation path refers to the path along which heat generated by the semiconductor element 4 or the semiconductor element group 10 diffuses through the circuit board 200 toward the cooler 1. In this embodiment, the portion 300 where the bonding material is thinnest is located in a portion other than the central portion 400 of the metal layer, so the thermal resistance is minimized in the path passing through the portion other than the central portion 400 of the metal layer. In this embodiment, by arranging the semiconductor element group 10 in a location on the circuit board surface where the path with the smallest thermal resistance exists, it is possible to mitigate the rise in temperature of each semiconductor element 4 during operation. Furthermore, in this embodiment, arranging the semiconductor element group 10 in a portion other than the central portion of the circuit board increases the degree of freedom in the layout of the semiconductor elements and also suppresses localized rise in temperature of the power conversion device.

[0091] Considering the cooling efficiency as described above, the closer the semiconductor element group 10 is positioned on the circuit board surface to the path with the smallest thermal resistance (the path passing through the thinnest portion 300 of the bonding material in the thickness direction), the higher the cooling efficiency. Therefore, as shown in FIG. 18 , it is preferable that the distance between an extension line 301 drawn from the thinnest portion of the bonding material in the thickness direction and the center point 600 of the semiconductor element group is shorter than the distance between an extension line 401 drawn from the center of the metal layer in the thickness direction and the center point 600. The distance between each extension line and the center point can be calculated by the method for calculating the distance between a point and a line (the shortest distance between a point and any point on the line is calculated). Furthermore, the center point 600 of the semiconductor element group is the geometric center point of a figure created by connecting the geometric centers 500 of the semiconductor elements. This is the same as in the third embodiment.

[0092] It is even more preferable that the center point 600 of the semiconductor element group is located on an extension line 301 drawn in the thickness direction from the thinnest portion of the bonding member. In this case, the cooling efficiency of the semiconductor element group 10 as a whole is further improved.

[0093] Some semiconductor element characteristics, such as on-resistance, exhibit temperature dependency, and if there is variation in the temperature of each semiconductor element, the on-resistance will also vary, causing a current imbalance. If the design is based on this assumption, the element area will increase, leading to increased costs. The configuration of this embodiment makes it possible to uniformize the heat dissipation of each semiconductor element 4, thereby reducing costs.

[0094] In this embodiment, the spacers 701 and 702 are arranged inside the joining member 7, but these spacers do not necessarily need to be inside the joining member 7 and may be exposed to the outside of the joining member 7.

[0095] Furthermore, in this embodiment, two spacers 701 and 702 having different heights are arranged, but three or more spacers having different heights may be arranged.

[0096] Specific examples of the materials of the components in this embodiment are the same as those in the first embodiment.

[0097] REFERENCE SIGNS LIST 1 Cooler 2 Molded resin 3 Terminal 4 Semiconductor element 5 Wiring member 6 Component 7 Bonding member 10 Semiconductor element group 100 Power module 200 Circuit board 201 Circuit pattern 201a First circuit pattern 201b Second circuit pattern 202 Insulating layer 203 Metal layer 300 Thinnest portion of bonding member 400 Central portion of metal layer 501, 502, 503 Plate-shaped wiring member 701, 702 Spacer 1000, 1001 Power conversion device

Claims

1. A power conversion device including: a power module having a circuit board in which a circuit pattern is formed on an upper surface of an insulating layer with an insulating layer sandwiched therebetween and a metal layer is formed on a lower surface of the insulating layer; and a semiconductor element mounted on the circuit board; and a cooler joined to the metal layer via a joining member, the surface of the circuit board facing the cooler is a convex curved surface facing the cooler, the joining member is thinnest at a position where the surface of the circuit board facing the cooler is closest to the cooler, the thinnest portion of the bonding member is located away from a center position of the metal layer when the power conversion device is viewed from above, The semiconductor element is disposed on the circuit board on an extension line drawn in the thickness direction from the thinnest portion of the joining member. Power conversion device.

2. A plurality of the semiconductor elements form a semiconductor element group, a distance between an extension line drawn from the thinnest portion of the bonding member in the thickness direction and the center point of the semiconductor element group is shorter than a distance between an extension line drawn from the center of the metal layer in the thickness direction and the center point of the semiconductor element group. The power conversion device according to claim 1 .

3. A plurality of the semiconductor elements form a semiconductor element group, The center point of the semiconductor element group is on an extension line drawn in the thickness direction from the thinnest part of the bonding member. The power conversion device according to claim 1 .

4. one semiconductor element is mounted on the circuit board, The active area of ​​the semiconductor element is located on an extension line drawn in the thickness direction from the thinnest part of the bonding member. The power conversion device according to claim 1 .

5. one semiconductor element is mounted on the circuit board, The center point of the active area of ​​the semiconductor element is on an extension line drawn in the thickness direction from the thinnest part of the bonding member. The power converter according to claim 4.

6. A plurality of the semiconductor elements form a semiconductor element group, a plurality of sets of the semiconductor element groups are mounted, and a plurality of thinnest portions of the bonding member are present; the thinnest portions of the bonding members are present corresponding to the respective groups of semiconductor elements; The power conversion device according to claim 1 .

7. It further includes a plurality of spacers of different heights, the plurality of spacers having different heights are disposed between the circuit board and the cooler; The main surface of the circuit board is inclined with respect to the main surface of the cooler. The power conversion device according to any one of claims 1 to 5.

8. A method for manufacturing a power conversion device including: a power module having a circuit board in which a circuit pattern is formed on an upper surface of an insulating layer with an insulating layer sandwiched therebetween and a metal layer is formed on a lower surface of the insulating layer; and a semiconductor element mounted on the circuit board; and a cooler joined to the metal layer via a joining member, forming a surface of the circuit board facing the cooler to have a convex curved surface facing the cooler; a step of arranging the semiconductor element on the circuit board on an extension line drawn from a portion other than the central portion of the metal layer in the thickness direction thereof; the joining member is thinnest at a position where the surface of the circuit board facing the cooler is closest to the cooler, The thinnest portion of the joining member is formed at a position away from the center position of the metal layer when the power conversion device is viewed from above. A method for manufacturing a power conversion device.

9. It has a plate-shaped wiring member, the linear expansion coefficient of the plate-shaped wiring member is greater than the linear expansion coefficient of the circuit board; a step of attaching the plate-shaped wiring member to the semiconductor element; a step of warping the circuit board by lowering the temperature from the process temperature for the attachment. The method for manufacturing the power converter according to claim 8.

10. It has a plate-shaped wiring member, the linear expansion coefficient of the plate-shaped wiring member is greater than the linear expansion coefficient of the circuit board; a step of attaching the plate-shaped wiring member to the circuit board so as to bridge the circuit pattern; a step of warping the circuit board by lowering the temperature from the process temperature for the attachment. The method for manufacturing the power converter according to claim 8.

11. It has a plate-shaped wiring member, the linear expansion coefficient of the plate-shaped wiring member is greater than the linear expansion coefficient of the circuit board; attaching the semiconductor element to the circuit board via the plate-shaped wiring member; a step of warping the circuit board by lowering the temperature from the process temperature for the attachment. The method for manufacturing the power converter according to claim 8.