WO1998012748A1 - Junction semiconductor module - Google Patents

Abstract

A first radiating plate is fixed on one major surface of a power semiconductor device. An insulating board is fixed to the other major surface, and a second radiating plate to the insulating board. An insulating and non-pressurized semiconductor module in which the heat is radiated from the two major surfaces of a power semiconductor device is provided.

Description

 Specification

 Technical field of junction type semiconductor module

 The present invention relates to a junction type semiconductor module having at least one power semiconductor element built in a resin case and having a heat sink on the bottom and top surfaces of the resin case. Background art

 The power semiconductor package consists of an internal semiconductor element and an external mounting surface.

(Bottom) can be classified into an insulation type package that is electrically insulated from the other, and a non-insulation type package whose bottom side is electrically connected to the internal semiconductor device. In this case, an absolute type that does not need to consider the potential of the package is more convenient. The other classification is pressurized type and non-pressurized type (terminal type, screw type). The pressurized type employs a press-contact structure inside the semiconductor package. In both cases, the non-pressurized type, which does not require a configuration to apply external pressure to the package, is more convenient to use. In other words, of the combinations of these two categories, the most convenient is an insulated non-pressurized package.

In power semiconductor devices, since the voltage and current handled are large, the amount of heat generated increases secondarily. In power semiconductor devices, a semiconductor main junction is provided in the bulk of silicon and current flows in the vertical direction, thereby effectively utilizing the entire silicon. In the insulation displacement package, the insulation displacement electrodes press the semiconductor from both sides, so that heat can be dissipated from both sides without any special measures. Excluding the crimp-type package, the heat generated in the silicon is generally Only the surface, that is, one side of the semiconductor element, emits to the outside, so the Icheon efficiency is lower than the electrical surface. If resin is used, it is easy to release heat from both sides of the semiconductor element. One example is disclosed in JP-A-6-252299. However, as described above, since the power semiconductor element generates a large amount of heat, it is not possible to use a resin whose thermal conductivity is two orders of magnitude smaller than that of a metal. In addition, since the electrode area is large because a large current is handled, it is difficult to add a special structure for heat dissipation to the surface of the semiconductor element as disclosed in Japanese Patent Application Laid-Open No. 5-121601.

 Yoneda et al. "GTO Thyristor Power Brick" Mitsubishi Electric has an example of an insulation type package that does not require externally pressurizing the package while having a press-fit type mounting structure suitable for heat dissipation on both sides inside the package. It is disclosed in Technical Report Vol. 6 7 No. 9 PP. 85-89. If the number of elements inside the package is small (two in this example), as in this example, it is possible to use a press-contact structure inside. However, as the number of elements increases, it becomes difficult to perform uniform pressing with a press-contact type package. An example in which a relatively large number of elements are mounted is disclosed in JP-A-52-113681.

Transistors, IGBTs (Insuia Led Gale Biological Transistors), etc., have a single element size limit compared to thyristors, G-cho 0 (Gate Turn Off Thyristor), etc. A module structure that implements must be adopted. The internal structure of an insulated non-pressurized module on which a large number of elements are mounted is, for example, as disclosed in Japanese Patent Application Laid-Open No. 63-226045, on an insulating substrate bonded to a metal base plate. Usually, a semiconductor element is mounted and mounted by wire bonding and soldering technology. Wire bonding and soldering technology can be applied to a thyristor, GTO, etc., even when a large number of devices are mounted to form a module. In many cases, a structure to be mounted by using the above is adopted.

 As described above, the insulation type module that has an internal structure that can be mounted even if the number of power semiconductor elements to be stimulated is large. Not disclosed.

 An object of the present invention is to provide an insulation type non-pressurized junction type semiconductor module which can radiate heat from the bottom and top surfaces of the semiconductor module. Disclosure of the invention

 The junction type semiconductor module according to the present invention is of an insulating type, and at the same time, emits heat generated by the power semiconductor element from the first and second ripening plates disposed on both sides of the power semiconductor element. .

 More specifically, the present bonded semiconductor module includes a plurality of power semiconductor elements, a plurality of insulating substrates to which the power semiconductor elements are fixed via a bonding material containing metal, and a power semiconductor element. At least one first heat sink fixed via a bonding material containing metal, and at least one second heat sink fixed on the plurality of insulating substrates via a bonding material containing metal The junction type semiconductor module according to the present invention preferably has a resin case for housing the power semiconductor element. The first radiator plate is located on the upper surface of the resin case, and the second radiator plate is located on the lower surface of the resin case.

In the junction type semiconductor module according to the present invention, various power semiconductor elements such as an IGBT, a power MOS FET, a power transistor, and a salista can be applied. Various metal materials such as copper, copper alloy, aluminum, and aluminum alloy can be used as the material of the heat sink. In addition, various joining materials including metals, such as solder and silver solder, are used as joining materials. Applicable.

 According to the present invention, the power generated by the power semiconductor element can be sufficiently released from the module from the first and first heat sinks arranged on both sides of the power semiconductor element. Therefore, a non-pressurized type, that is, a junction type semiconductor module which is a large-capacity and easy-to-use insulating type is realized.

 Other features of the present invention will become apparent from the following detailed description. BRIEF DESCRIPTION OF THE FIGURES

 FIG. 1 is a perspective view of an IGBT module according to the present invention.

 FIG. 2 is a perspective view of a semiconductor module according to the present invention.

 FIG. 3 is a sectional view of a semiconductor module according to the present invention.

 FIG. 4 is a perspective view of a semiconductor unit according to the present invention.

 FIG. 5 is a front view of a semiconductor module according to the present invention.

 FIG. 6 is a front view of a conventional semiconductor module.

 FIG. 7 is a schematic diagram of the heat flow of the semiconductor module according to the present invention.

 FIG. 8 is a cross-sectional perspective view of the-part of the flat conductor module according to the present invention.

 FIG. 9 is a schematic view of the thermal expansion of the semiconductor module according to the present invention.

FIG. 10 is a schematic diagram of a cut-off fibre of the IGBT unit in the embodiment of FIG. 1 ₍ FIG. 11 is a cross-sectional schematic diagram of a freewheel diode unit in the embodiment of FIG. 1).

 FIG. 12 is a perspective view of an IGBT chip in the embodiment of FIG.

 FIG. 13 is an explanatory view of a process for manufacturing an IGBT module according to the present invention. FIG. 14 is a perspective view of another IGBT module according to the present invention.

FIG. 15 is a partial cross-sectional perspective view of the external electrodes of the IGBT module of FIG. FIG.

 FIG. 16 is an explanatory view of the process of manufacturing the IGBT module shown in FIG.

 FIG. 17 is a perspective view of another IGBT module according to the present invention.

 FIG. 18 is a cross-sectional view of the semiconductor unit in the embodiment of FIG. FIG. 19 is a sectional view of another semiconductor unit according to the present invention.

 FIG. 20 is a perspective view of another IGBT module according to the present invention.

 FIG. 21 is a perspective view of another IGBT module according to the present invention.

 FIG. 22 is a perspective view of an inverter device according to the present invention.

 FIG. 23 is a perspective view of still another IGBT module according to the present invention. FIG. 24 is a cross-sectional view of the IGBT module of FIG.

 FIG. 25 is an explanatory view of the process of manufacturing the IGBT module shown in FIG. 23.

Best mode for carrying out the explanation

 (Example 1)

 First, main features of the present invention will be described with reference to an embodiment of the present invention. 2 and 3 show one embodiment of the junction type semiconductor module according to the present invention. The module of the present invention comprises a resin case in which the bottom surface is metal and the side surfaces are electrical insulators, and the upper surface is formed of a divided metal.

(External electrode terminals), and the space between the metal and the resin case is filled with resin. The bottom metal base 201 is electrically insulated from the internal circuitry. The first function of the metal base 201 is to mechanically support the module, and the second function is to protect the inside of the module from outside air. The third function of the metal base 201 is to conduct heat generated inside the module to the outside of the module. The resin case 202 electrically insulates the inside and outside of the module and protects the inside of the module from outside air. In addition, there are three functions: to secure the insulation distance between the external electrode terminal (gold bending on the upper surface) and the gold bending base 201.

 The semiconductor module of the present invention includes an external cathode electrode 203 occupying a large area on the upper surface. The external force source electrode 203 and the external anode electrode 204 are terminals electrically connected to a circuit inside the module. 2 and 3, only the external force source electrode 203 and the external anode electrode 204 are shown, and the gate electrode and the like are omitted to simplify the drawings. The external cathode electrode 203 of the semiconductor module of the present invention is larger than the terminal of the conventional semiconductor module. If only electrical requirements are met, the same size as the external anode electrode 204 may be used. This is because the currents flowing through the two are equal. The reason for enlarging the external cathode electrode 203 is to provide the external cathode electrode 203 with a function of guiding the heat generated inside the module to the outside similarly to the metal base 201. To release the heat, a larger heat transfer area is required if only electricity is passed.

 As shown in Fig. 3, inside the semiconductor module of the present invention, units having the same shape are arranged. Figure 3 shows three units. Each unit is provided with a ceramic insulating plate 310 having a high thermal conductivity as an electrical insulator on the bottom surface, and a metal on-chip buffer plate 304 that conducts electricity and heat well on the upper surface. Each unit conducts heat up and down, but does not conduct electricity downward.

The semiconductor module of the present invention has a form in which a plurality of units are sandwiched between a gold erasure base 201 and an external force source electrode 203. Semiconductor modules that use conventional wire bonding have a single base under the module and there are no large components on the force. Compared to the module using wire bonding, the internal structure of the module according to the present invention is vertically symmetric. High in nature. Such vertical symmetry is a feature of the configuration that conducts heat vertically. Since the cathode electrode 203 has a smaller area than the metal base 201, the heat radiation effect is smaller than that of the metal base 201. However, it uses the same material with high thermal conductivity (for example, copper, A1, composite metal, etc.) as the metal base 20], and is several times thicker than ordinary electrodes. Excellent heat conduction in the vertical and horizontal directions. Therefore, external force source electrode

Heat can also be effectively released from 203.

 The structure of each semiconductor unit will be described with reference to FIG. One semiconductor unit shown in FIG. 4 has a vertically stacked structure including a ceramic insulating plate 310 on the bottom surface and a buffer plate 304 on the upper surface. In the vertical direction, there is no factor that narrows the conduction path both electrically and thermally. One semiconductor unit is substantially all metal except for the ceramic insulating plate 310 and the silicon chip 301 on the bottom surface, and therefore has little electrical and mature resistance. Above and below the silicon chip 301, a buffer plate on the chip made of a metal with a thermal expansion coefficient close to that of silicon and good thermal and electrical conductivity (eg, W, Mo, composite metal, etc.) Solder 04 and buffer plate under chip 3 05. Since the structure is sandwiched between the upper and lower sides by a material close to the thermal expansion coefficient of silicon, the structure is always reinforced from the upper and lower sides of the silicon, and the thermal stress on silicon due to temperature changes is there. The difference in the size of the shock-absorbing plate between the upper side (force side) and the lower side (anode side) of the silicon chip 301 is because the outer periphery of the upper surface of the silicon chip 301 has a withstand voltage. This is because the electrode has been bonded to avoid such a part because of the structure of the element (eg, FLR (Field Lighting Ring)). Silicon chip

The upper side of 301 is the only upper buffer plate 304. The lower side (anode side) of the silicon chip is below the lower buffer plate 105 and the internal anode. There are electrodes 308. Since this is slightly larger and thicker than the silicon chip 301, it has the effect of spreading the heat generated in the silicon chip 301 and is effective in lowering the thermal resistance. On the side surface of the internal anode electrode 308, there is provided an anode terminal arm 401 for connection to the anode electrode connection wiring 310. This arm has an elongated structure because it is used to guide electricity and not to ripen it.

As shown in FIG. 4, the ceramic insulating plate 310 is larger than the internal anode electrode 308 in order to secure an insulation distance for electrically insulating the internal circuit from the outside of the module. This is because, in general, an insulation distance of 1 mm is required to ensure 100 V insulation. This distance is not preferred for miniaturization of the module, the junction semiconductors module according to the present invention, even if ensuring this distance, using a fifth diagram and Figure 6 for which _c can be reduced overall dimensions Will be explained. FIG. 5 is a diagram in which only the parts necessary for explanation in the junction type semiconductor module of the present invention are extracted. FIG. 6 shows a conventional wire bonding structure. Both figures are relatively the same scale so that the dimensions can be compared. In both figures, the illustration of solder layers that are not directly related to the description is omitted. In FIG. 5, since the insulation distance is set around the ceramic insulating plate 504 for each unit, the unit interval is widened. On the other hand, in the structure of FIG. 6, two silicon chips 501 are mounted on one ceramic insulating plate 602, so that the two silicon chips 501 There is no wasted space for insulation as shown in Fig. 5. Nevertheless, the width of the wire bonding structure according to the conventional technique shown in FIG. 6 is larger. The reason is that the wire bonding structure requires a counter electrode (the internal force source electrode 605 in Fig. 6) for fixing the other end of the wire bonded to the silicon chip. Because it is. This electrode must have an insulation distance between the adjacent electrode, ie, the internal anode and the electrode 601. This distance is slightly smaller than the absolute distance provided around the ceramic insulating plate 504 or the ceramic insulating plate 602, but still needs to be a distance commensurate with the withstand voltage of the silicon chip 501. It is. The internal force source electrode 605 has a cathode terminal arm for connection to the outside of the module.

There is also room for 606, so extra space is needed. In contrast, the structure of the present invention shown in FIG. 5 is connected to the external cathode electrode 507 directly above the silicon chip 501 without using wires, so that the horizontal direction as shown in FIG. No wasted space is generated. As shown by the thick arrows in the figure, the cathode flow flows in the vertical direction in FIG. 5, and flows in the horizontal direction in FIG. In the conventional technology, the current can flow sideways, so it can be taken as space is increased.

 In the case of a semiconductor unit on which a plurality of silicon chips are mounted, by laying out the wiring between the silicon chips in parallel, it is possible to provide a structure that does not generate a wasteful space in the horizontal direction for wiring between the chips.

 The heat flow of the semiconductor module according to the present invention will be described with reference to FIG. Also in Fig. 7, the indication of solder layers that are not directly related to the description of U is omitted. In the figure, arrows indicate the flow of heat generated in the silicon chip 70 i. The heat generated from the silicon chip 701 flows up and down. Silicon chip

The heat flowing downward from 70 1 flows almost vertically through the buffer 70 2 below the chip. This is because the size of the buffer plate under the chip 70

When the thermal conductivity is not so large compared to that of 701, which is indicated by the same size in the figure, and the buffer under the chip 702 is W or M0 due to the restriction of thermal expansion coefficient, the thermal conductivity Lower internal anode electrode 7 0 3 (usually copper or A 1) This is because it is as small as a fraction of. At the internal anode electrode 703, heat spreads to the left and right. Since the internal anode electrode 703 is thick, and copper or A1 with high thermal conductivity is used as the material, the thermal conductivity of the ceramics insulating plate 704 downstream of the heat is low. This is because the anode electrode is smaller than 03. In the ceramic insulating plate 704, as opposed to the internal anode electrode 703, the material is thin and the metal base 705 downstream of heat has a large thermal conductivity, so that ripening does not spread. In the metal base 705, the force of the member for mounting the module, which is usually mounted via thermal grease having a lower thermal conductivity than metal, spreads the heat further.

 The heat flowing upward from the silicon chip 701 flows almost vertically through the on-chip buffer plate 706. This is because the size of the buffer on the chip 706 is smaller than the size of the silicon chip 701, and because the buffer on the chip 70 ΰ uses W or Μ ο due to the restriction of the coefficient of thermal expansion, This is because the rate is smaller than the external force source electrode 707 above it. At the external force source electrode 707, ripening spreads right and left. This is because the external force source electrode 707 is thick, and copper or A 1 having high thermal conductivity is used as the material.

The heat radiation outside the semiconductor module will be described with reference to FIG. FIG. 8 shows a central part of an inverter device for obtaining three-phase alternating current from direct current, comprising six junction type semiconductor modules of the present invention. In the junction type semiconductor module of the present invention, As a switching element, an IGBT (Insulated Gate Bipolar Transistor), which consists of a drive unit with a MOS structure and a main conduction unit with a bipolar structure, is mounted. External anode electrode 8 0 1 External force source electrode 8 0 2, resin case 8 0 3, 3 modules with metal base 8 0 4 horizontal, 2 vertical, a total of 6 modules, metal base Arrange on the mounting plate 8 1 5. The heat generated from the module is The air is released from the fins 8 16 on the metal base mounting plate attached to the metal base mounting plate 8 15 through the outside. In Fig. 8, the fins 8 16 on the metal base mounting plate are provided on the back of the metal base mounting plate 8 15 to provide a ripening path to the outside air.However, cooling using a liquid such as water or Freon is used. It may be a system. In addition, a heat pipe may be used to radiate heat away from the semiconductor module.

 Connect the three external anode electrodes 800 in the upper row of the module with the DC positive busbar 805, and connect the positive side of the DC power supply. In addition, the three external force source electrodes 802 arranged in the lower row of the module are connected by the DC minus side bus bar 806, and the negative side of the DC power supply is connected. Next, the upper external force electrode 802 and the lower external AND electrode 80] of the upper and lower modules on the left end are connected with the U-phase bus bar 80Ω. This is the U-phase output terminal for three-phase AC. Similarly, the upper external cathode electrode 802 and the lower external anode electrode 801 of the upper and lower modules at the center are connected to the V-phase bus bar 811, and the upper and lower modules at the right end The upper external force source electrode 802 and the outer external anode electrode 801 are connected by a W-phase bus bar 813. These forces become the output terminals of the V-phase and W-phase of the two-phase AC.

 In order to ripen the lower three modules, attach the heat dissipation fin 807 on the DC minus busbar to the DC minus busbar 806. Because the part hidden under the fin is difficult to see, the fin is cut away. There is a cutting part of the heat dissipation fin on the DC minus side busbar. The heat of the upper three modules is dissipated by the U-phase bus bar — Top ripening fins 8 10, V-hook bus bar fins 8 12, and W-phase bus bar 一 top fins 8 14 I do.

A fin is provided on the bus bar to reach the external cathode electrode 802 The heat in the yule can be effectively released to the outside air. ¾If the busbar has a sufficient surface area for ripening and has sufficient heat radiation capability by itself, there is no need to attach a fin. Of course, liquid such as water or chlorofluorocarbon may be circulated in the bus bar for cooling, or a heat pipe or the like may be used for cooling. In any case, heat is easily conducted because heat is efficiently guided from the module to the terminals.

 In the junction type semiconductor module according to the present invention, the upper heat sink may be electrically insulated from the inside. In this case, the members that are in contact with the bottom and top surfaces of the module only need to have a heat dissipation function, and the structure around the module is simplified as compared with the configuration shown in Fig. 8, so that the transmission efficiency as a whole system is improved. Can be higher.

 It is important to match the expansion coefficients of the materials that make up the bonded conductor module. The configuration of the junction type semiconductor module of the present invention will be described with reference to FIG.

In the structure of the present invention, the silicon chip 901 has a thermal expansion coefficient of about 3 ppm, while the external cathode electrode 903, the internal anode electrode 905, and the metal base 907 The thermal expansion unbalance occurs in the lateral direction in the unit due to the large thermal expansion coefficient of the material composing it. For example, copper is about 17 pm and A1 is about 24 ppm. The absolute value of the stress based on the difference between the coefficients of thermal expansion is proportional to the product of the difference between the difference between the coefficients of thermal expansion, the temperature difference, and the length of the contact between the materials (the diagonal length of the soldered part in the structure of the present invention). Therefore, the thermal stress generated in the semiconductor unit of the present invention, in which the lateral dimension of the unit is slightly larger than the size of the silicon chip 911, has no large-area soldered portion. Target small. Furthermore, since the upper and lower structures viewed from the silicon chip 91 are roughly symmetrical, the bending stress on the chip caused by the bimetallic action Also less.

 The semiconductor unit has a layered structure in the vertical direction, and in the unit longitudinal direction, almost no thermal stress is generated due to the difference in thermal expansion coefficient.

 The horizontal arrow in FIG. 9 indicates the horizontal thermal expansion of the entire module. Each unit is vertically sandwiched between an external cathode electrode 903 and a metal base 907. These materials include copper or A 1, silicon carbide, silicon, or germanium dispersed in copper or A 1, low thermal expansion and high thermal conductivity, and copper or A 1 laminated with W or Mo. You may. It is important to make the material of the external force source electrode 903 and the metal base 907 the same in order to reduce the thermal stress. If the material of the external force source electrode 903 and the metal base 907 are the same, the upper and lower members of the semiconductor unit expand and contract by the same amount, and each semiconductor unit simply moves horizontally in parallel. There is no thermal stress. In the junction type semiconductor module structure of the present invention, since the heat generated in the silicon chip 901 flows roughly balanced in the vertical direction, the temperature of the external force source electrode 903 and the temperature of the metal base 907 are increased. The imbalance is small.

 The vertical force of three arrows in FIG. 9 represents the thermal expansion in the vertical direction of each semiconductor unit. Since the constituent materials of each semiconductor unit are substantially the same and the dimensions are almost the same between the units, the coefficient of thermal expansion in the vertical direction of the unit is the same for each unit. Therefore, even if the upper and lower parts are sandwiched by the external force source electrode 903 and the metal base 907, there is little fear that stress will be generated between the units in the vertical direction.

It is important that the internal filling resin 909 has a thermal expansion coefficient close to the vertical thermal expansion of the unit. If the thermal expansion in the vertical direction of the filling resin 909 and the whole unit is not matched, the thermal stress between the resin and the unit Occurs. Aligning the thermal expansion of the resin case 908 with the external force source electrode 903 and the metal base 907 is also effective in suppressing the occurrence of stress in the horizontal direction. The strictness is smaller than the requirement of the thermal expansion coefficient for the resin 909.

 Resins generally have a higher coefficient of thermal expansion than metals. Therefore, a resin having a low thermal expansion such as silica can be mixed with the resin to make the thermal expansion coefficient of the resin close to that of copper. Therefore, when the external force source electrode 903 and the metal base 907 are made of copper or A1, matching with the resin is obtained, and the thermal stress of the entire module can be reduced. The thermal expansion coefficient of the upper and lower portions 1 of each semiconductor unit is close to the thermal expansion coefficient of the inner anode electrode 905 which is the thickest member. By using copper or Λ1 for the internal anode electrode 905, the thermal expansion coefficient of the filling resin 909 can be adjusted.

 In the junction type semiconductor module of the present invention, since the area of each soldered portion is small, it is easy to maintain good soldering quality. Also, since the number of chips per unit is small, the quality of each chip can be determined before assembling into a module.

 Hereinafter, the present invention will be further specifically described with reference to other examples.

 (Example 2)

 Second Embodiment A second embodiment will be described with reference to FIGS. 1 and 10 to 13. FIG. 1 is an oblique view of the junction type 1 G-mode module of the present embodiment. In the figure, the resin case〗 02 is shown transparent so that the inside can be seen, and the resin filled in the case is omitted.

The module has six IGBT units 113 and three freewheel diode units 119 cut on a Ni-plated copper base 101. Connect all 9 units on a total of 9 units _U external Ano cathode electrode external cathode cathode electrode 1 0 3 in the form rests for 1 (M is partially covers the top of the external force source cathode electrode 1 0 3. External force saw de electrodes i 0 Filled resin (not shown) is inserted between the notch of No. 3 and the external anode electrode i 04 covering the notch to keep the two insulated. As long as insulation does not occur when a predetermined voltage is applied, there is no specific distance between the electrodes, that is, the vertical distance between the external force source electrode〗 0 3 and the external anode electrode 104 is The dielectric strength of the resin to be filled can be approached, but it is necessary to secure the space and creepage distance specified on the outer wall of the module specified by the standards of the Japan Electrical Manufacturers' Association. 】 At 500 V, both AC and DC have a clearance of 20 ι, creepage distance Therefore, the above-mentioned control is applied to the lateral distance between the external force electrode 103 and the external anode electrode 104. In order to make the electrode 103 as large as possible, a circular hole may be formed in the external force source electrode 103, and an external anode electrode 104 may be arranged at the center of the hole. Space is required around the entire area to secure the insulation distance, which increases wasted space, and in the case of a circular hole, the external anode electrode is used during assembly.

It is necessary to mount the external cathode electrode 103 before 104, which reduces the degree of freedom in the assembly process. For the above reasons, in this embodiment, one side of the external force source electrode 103 is cut out, and the external anode electrode 104 is covered therewith.

Three IGBT units 113 along the front row and three rows next to the back row are freewheel diode units 119 in the center row. is there. From the gates of the three IGBT units 113 in the back row and the IGBT chips 106 of the three IGBT units 113 in the front row, It is connected to the external gate electrode 105 through the gate connection wiring 123. In this example, a copper wire insulated with a fluororesin was used for the gate collective wiring 123. A resin case 102 is provided on the outer periphery of the semiconductor module, and the module is filled with a filling resin (not shown) whose coefficient of thermal expansion is adjusted to be approximately the same as copper. I have.

 In the IGB Dip Tip 106, the bottom is the anode (collector) and the top is the power source (emitter). In the free wheel diode tip 107, the bottom is the power source and the top is the anode. It is. IG B and the diode are connected in anti-parallel.

A current path to the external anode electrode 104 will be described. The internal anode electrode 111 and the internal anode electrode 117 of each unit are connected to an anode assembly plate A120 extending in the front-back direction. Of the three anode assembly plates A120, only the rightmost one can be seen from the side of the external cathode electrode 103. The anode assembly plate A 120 is a 1 GB unit at the back 1 13, a freewheel diode unit 1 19 at the center, and one iGBT unit 1 1 3 at the front. Can drain current. By arranging three pieces on the left and right, nine units are connected. Next, as a member connecting the three anode plywoods A120, there is an anode assembly board B122. It runs left and right through the space between the 1 GBT unit 113 in the back row and the freewheel diode unit i 19 in the center row. As another member connecting the three anode aggregate plates A 120, there is an anode aggregate plate C〗 22. It runs left and right in the space between the freewheeling diode unit 119 in the center row and the IGBT unit 113 in the front row. The current of the left anode plate A 120 and the central anode plate collected by the anode plate B 122 and the anode plate C 122, respectively. The current of A120 finally collects on the rightmost anode collecting plate A120, and reaches the external anode electrode 104.

 In power semiconductor modules, matching of the inductive component of the wiring path is important. In the structure of this embodiment, the induction component of the path leading to the external force source electrode 103 is sufficiently small, and there is no problem with the matching of the induction component. However, there is an imbalance in the path leading to the external anode electrode 104. Since the rightmost column is closest to the external anode electrode 104, the inductive component is small and the leftmost column is largest. In each row, the induction component of the center freewheel diode unit 119 is smaller. There is no problem because this imbalance occurs on the anode side. The imbalance on the anode side has a smaller effect on the operation of the element than on the cathode side. Since a voltage is applied between the force source and the gate to drive the cathode, the potential of the cathode must be stable. Therefore, it is necessary to minimize the unbalance on the cathode side. In comparison, the anode is relatively insensitive to imbalance. In this embodiment, the imbalance between the IGBT units 113 is small because the imbalance between the 1 GBT units 113 is small and the freewheel diode unit 119 is arranged at the center. I got it.

The 1 GBT unit 113 has one IGB chip 1 ◦6. It is a square with 15 sides. The top and bottom of the IGBT chip 106 are sandwiched between the M0 buffer plate 108 on the chip and the M0 buffer plate 110 below the chip. <The Mo buffer plate 108 on the chip is a square with a side of 1 O mm. , Thickness l min, under chip M0 buffer plate 110 is a square with 16mm on each side and thickness is arbitrary <All are Ni plated and soldered IGBT chip It is connected to 106. The difference between the upper and lower plates is that there should be no metal on the periphery (termination area) of the top (front) surface of the IGBT chip 106 Because there is no. The periphery of the upper surface (surface) of the IGBT chip 106 is a portion where an electric field applied to the chip appears on the surface. If the electric field is not generated uniformly, an abnormal discharge will occur. Therefore, special measures such as FLR (Field Limi ting Ring) have been applied to this part. Therefore, a good conductor of electricity that disturbs the electric field must not be close to the upper marauder in this part.

 Conventional silicon chips based on wire bonding often use A1 (or A1 alloy with some added elements) for the electrodes on the chip surface. However, in the structure of the present invention, since both sides of the silicon chip are soldered, the chip surface requires the same solderable metal film as the rear surface. In this embodiment, both surfaces of the IGB chip 10β have a multilayer structure of four layers of A1-Ti-Ni-gold.

 Below the Mo buffer plate 110 below the chip is the internal anode electrode 111. This is a copper plate with a side of 20 mm and a thickness of 2 mm. An alumina insulating plate 112 is provided below the internal anode electrode 111, and is electrically insulated from the copper base 101.

 Copper block for height adjustment, placed on the Mo buffer plate 108 on the chip

1 109 will be described. The size is 10 mm on each side and the thickness is i O mni. In order to satisfy the above insulation distance standard, the copper base under the module must be

It is necessary to secure the distance between 101 and the external electrode group on the module. Therefore, the height of the module must be larger than a certain value. In order to secure the distance in the module height direction, a regulating river block is required in the module. It may cause the thermal resistance increases in the module, _u height adjusting copper block 1 0 9 thickness desirably kept to the minimum necessary for this block, § Roh one de ^ plywood A 1 2 0, Anode collecting plate B 1 2 and anode collecting plate C 1 2 2, that is, internal anode electrode 1 1 1, or Or while maintaining the vertical distance between the ground wire connecting the internal anode electrode 117 and the external anode electrode 104 and the bottom of the external force electrode 103. In addition, a sufficient space between the IGBT chip 106 and the free wheel diode chip 107 in the area for ensuring the withstand voltage (the above-described termination portion provided with FLR, etc.) is sufficiently secured in the vertical direction.

 The freewheel diode unit 119 has one freewheel diode chip 107 mounted thereon. The size is a square of 15 sides. The upper and lower sides of the freewheel diode chip 107 are sandwiched between a Mo buffer board 114 above the chip and a Mo buffer board 116 below the chip. The Mo buffer plate 114 on the chip is a square with a side of 1 O inm and a thickness of l n, and the Mo buffer plate 116 below the chip is a square with a side of 16 and a thickness of one. All are plated with Ni and adhered to the freewheel diode chip 107 by soldering. The difference between the dimensions of the upper and lower plates is due to the pressure resistance at the periphery of the top surface (front surface) of the freewheel diode chip 107. Since a strong electric field is applied to this part, keep good conductors of electricity and ions away.

 The freewheel diode chip 107, like the] GBT chip 106, also had a four-layer, multi-layer electrode consisting of A] -Ti-Ni-gold for soldering on the front side.

The freewheel diode unit 119 is the same as the IGB unit 113 except that there is no gate hole in the Mo buffer plate 114 on the chip. The IGBT unit 113 will be described with reference to FIG. 10, and the freewheel diode unit 1] 9 will be described with reference to FIG. At the center of the IGBT chip 106 is a gate terminal. In order for the gate signal to spread throughout the chip, it is best to have a gate terminal in the center of the chip. Get — Solder the gate wiring 1 0 7 to the gate terminal with gate river solder 1 0 8. Furthermore, a hole was made in the center of the Mo buffer plate 108 on the chip, and a groove was dug in the surface of the copper block for height adjustment 109 that was in contact with the Mo buffer plate 108 on the chip. Pull out from the side of height adjustment / H copper block 109 through wiring 107.

On the other hand, the freewheel diode chip 107 has no gain. Therefore, the Mo buffer plate 1 14 on the chip has no hole and the copper block 1 15 for height adjustment has no groove. The difference between the IGBT unit 113 and the freewheel unit unit 119 It is. Since the freewheel diode chip 107 is mounted with the anode facing upward, the power source of the freewheel diode chip 107 is connected to the internal anode pole 117. The called anodic to force saw de is connected, 1 GBT chip 1 0 6 internal anodic electrode 1 1 1 for, _c first 2 FIG is order to unify the name, IGBT In the figure where the chip 106 is viewed obliquely from above, the IGBT chip anode surface 124 is hidden below. When the GBT chip force source surface 1203 is roughly divided, it is divided into two areas. The circuit forming area 122 1 in the center and the resistance area 1... In the peripheral area are secured areas 102. Since the electric field leaks into the space above the chip in the withstand voltage securing region 1222, it is necessary not to dispose a conductor above. On chip 緩衝 ο Buffer plate

The reason why it is smaller than 108 and the height adjusting copper block 109, ',', IG Τ {chip} 06 is to keep the open side of the withstand voltage securing area 122. A gate electrode 1206 is provided at the center of the circuit forming region 1201. Gate electrode

Cathode electrodes 125 are arranged in the circuit formation region 1201 on the periphery II of the circuit 206. In the freewheel diode tip 107, both the anode electrode and the cathode electrode are on one side. G13 Τ In the tip 106. The cathode electrode 125 has a strip shape as shown in the figure, and a MOS region for gate drive is provided near the surface of the circuit forming region 1221, in which several are arranged. This is because MOS is susceptible to external forces, and electrodes that are susceptible to external forces are not provided directly above MOS. That is, a MOS portion is formed in a strip shape, and a force source electrode 125 is formed in the remaining strip portion where no MOS is present.

 A method of manufacturing the junction type IGB T module of this embodiment will be described with reference to FIG. In the figure, the left half shows the manufacturing process, and the right half shows the state after each step is completed (before the next step is started).

 (1) Unit assembly

 A 1GB chip 106 and a freewheel diode chip 107, which have been evaluated in advance for static characteristics and evaluated as good, are mounted on a semiconductor unit. The figure shows an IG13 @ unit 113 with an IGBT chip 106 mounted. In this embodiment, the copper block for height adjustment 109 and the copper block for height adjustment

A semiconductor unit is mounted with 1 1 5 on top. In this process, up to the height adjustment copper block 109 is mounted. Each layer is bonded by soldering. IG Β 6 There are six soldering points in unit 113 and five in freewheel diode unit 119. These are collectively performed in the present embodiment. Of course, soldering may be performed several times.

 (2) Unit sorting

Sorting on a chip basis, which is a feature of the present invention, is performed. Even before the semiconductor unit is mounted, the static characteristics can be measured by inserting a needle for measurement into the chip. Static characteristics are measurement items that do not involve changes in current or voltage, such as withstand voltage and on-voltage. On the other hand, dynamic characteristics such as switching characteristics and safe dynamic The measurement of the working area or the measurement of heating the chip is not possible unless the supply of current including the conductive component is normal. Conventionally, measurement was performed after assembling the semiconductor module. In the present invention, all the items 0 are measured in the process before assembling the components into the module, and the sorting can be performed in units of units. In the figure, one IGB unit and one free-wheel diode unit selected from the selection are displayed.

 (3) Unit layout

 The selected units are arranged on the copper base 101. The soldering using the solder i 301 under the alumina insulating plate may be performed in this step, or may be performed in the next step or in the next step. In this step, the method of soldering the solder under the alumina insulating plate 1301 can be reliably manufactured without the risk of displacement. Conversely, if the three steps are soldered all together in the next step (1), the number of steps is small and the cost is advantageous. However, since the external cut electrode 103 hides the inside of the module, the testability deteriorates. In the present embodiment, in consideration of the inspectability, the intermediate between the two is avoided, and it is not necessary to perform the soldering at the same time as the soldering of the cathode. We are going to solder.

In this embodiment, three rows of semiconductor units are arranged as shown in FIG. Units of the same type are arranged side by side. However, Fig. 13 shows two rows of different types of units side-by-side to avoid complicating the diagram. <In other words, in Fig. 13, the left is the IGBT unit 113 and the right is the right. Was set to the freewheel diode unit 119. Therefore, FIG. 13 does not represent the force-specific cross section based on the cross section obtained by cutting out a part of FIG. 1 as it is. (4) Anode and gate connection

 External anode electrode 104, external gate electrode 105, anode assembly plate A120, anode assembly plate B121, anode assembly plate C122, and The resin case 102 in which the gate collective wiring 123 is integrally molded is mounted on the copper base 101. At this time, the anode terminal arm bonding portion 1302 and the gate wiring bonding portion 1303 are bonded by soldering, and the resin case bonding portion 1304 is bonded by an adhesive. In this embodiment, soldering of the solder under the alumina insulating plate 1301 is performed at the same time.

 (5) Cathode connection

 Adhere external force source electrode 103. After this step, the external cathode electrode 103 almost hides the inside of the module, so it is better not to use soldering for other steps. Since the external anode electrode 104 covers a part of the external force electrode 103, the external force electrode 103 is sunk under the external anode electrode 104 and placed. It is necessary. Figure 1 shows the end of the process.

 (6) Resin filling

 The module is completed by pouring the filling resin 13 06. In this example, a resin containing silicon oxide as a main component and a filler for adjusting a thermal expansion coefficient mixed therein was used. The resin is colored black to avoid malfunction due to light entering.

 (Example 3)

Third Embodiment A third embodiment will be described with reference to FIGS. FIG. 14 is an oblique view of another junction type IGBT module according to the present invention. In the figure, the resin case 1402 is made transparent so that the inside can be seen, and the resin filled in the case is omitted. The semiconductor module according to the present embodiment is the same as the module according to the second embodiment in both the withstand voltage and the ia flow rate, but is different from the second embodiment in the material of the base and the electrodes. In this embodiment, the Λ1 base 1401 was used. The material of the external force source electrode 144, the external anode electrode 144, and the external gate electrode 144 was also set to A1 in order to match the thermal expansion coefficient in the lateral direction. A1 has the advantage of being lighter and cheaper than copper. On the other hand, there is a disadvantage that the electric conductivity and the heat conductivity are smaller than those of copper. Regarding electrical conductivity, it is not a problem because the dimensions of the electrodes are naturally large. Also, since the thermal conductivity of A 1 is higher than that of other members, there is no particular problem regarding the thermal conductivity. However, thermal expansion is a problem. In other words, A 1 has a larger coefficient of thermal expansion than copper, so the difference between the maturing coefficient of silicon and that of silicon increases. The Cabinet addressed the question as follows. In other words, for the module as a whole, the lower base and the upper electrode are made of the same material, so that the expansion coefficients in the horizontal direction are made uniform and the mismatch of the thermal expansion coefficient in large dimensions is avoided. An increase in the difference in thermal expansion coefficient from silicon was avoided as follows. That is, copper was used for the members constituting the IGBT unit 1407 and the freewheel diode unit 144 located near the silicon chip. As a result, the difference in thermal expansion coefficient as viewed from the silicon chip was the same as in Example 2. The coefficient of thermal expansion of the resin (not shown) filled inside was adjusted to copper to match the vertical expansion of the module in the vertical direction;

This is because the overall thermal expansion coefficient of the IGBT unit 1407 and the freewheel diode unit 1408 is close to that of the most widely used copper. In terms of thermal and electrical performance, it is preferable that the members in each unit having a high heat flow and current density are not A1, but copper as in the second embodiment. From the viewpoint of protecting the silicon chip, the heat of the resin case I 402 The expansion coefficient was also adjusted to copper. Finally, in the module according to the present embodiment, the coefficient of thermal expansion in the vertical direction was set to copper, and the coefficient of thermal expansion in the left, right, front and rear directions was set to A 1.

 The difference from the second embodiment is that the external cathode electrode 144 and the external anode electrode 144 overlap. In Example 2, the filling resin

1 3 () 6 has a structure to enter. On the other hand, in this embodiment, an alumina insulating plate 1406 is sandwiched in place of the resin. A1 has a lower thermal conductivity than copper, so to compensate for this, an external anode electrode

A part of what was transmitted to the external cathode electrode 1403 was also transmitted to 1444 to reduce the total thermal resistance.

This part is shown enlarged in Fig. 15. The figure shows a section taken along the plane passing through the center of the external force source electrode 1403 and the external anode lightning pole 144. An alumina insulating plate 144 is sandwiched between the external force electrode 1403 and the external anode electrode 144. Ceramics typified by alumina have a two-digit higher maturation conductivity than organic resins, so the IGBT unit 1407 and freewheel diode unit 1408 in the module The heat flowing into the external force source electrode 1403 is received by the alumina insulating plate 144 and can be transferred to the external anode electrode 144. Furthermore, when filling the gap with resin, the resin cannot be filled well unless the gap is at least about 1 band.However, when the insulating plate is sandwiched, the thickness should be at most about 0.5 mm. Is enough. The effects of both the thermal conductivity and the thickness of the insulator are synergistic, and the thermal resistance in this area is reduced. As a result, the external anode electrode 144 is thermally coupled, and the external anode electrode 144 can also contribute to the ripening. In this embodiment, the external anode electrode 144 is electrically connected to the anode, and thermally connected to the force source. Is wearing.

 A method for manufacturing the junction type semiconductor module of this embodiment will be described with reference to FIG. In the figure, the left half shows the manufacturing process, and the right half shows the state after each step is completed (before the next step is started).

 (1) Unit assembly and (2) Unit sorting

 The above two steps are the same as in Example 2.

 (3) Unit layout

 The selected units are arranged on the A1 base 1401. The unit mounting solder 1601 is placed between the A1 base 1401 and the 1GBT unit 1407 or the freewheel diode unit 1408.

 (4) Gate connection

 External gate electrode 1 4 0 5, anode assembly plate A 1 0 9, anode assembly plate B 1 4 10, anode assembly plate C 1 4 11 1, and gate assembly wiring 1412 The integrally molded resin case 1402 is mounted on the A1 base 1401. As in the second embodiment, each wiring member integrated with the resin case 1402 is connected to each unit by the anode terminal arm bonding portion 1602 and the gate wiring bonding portion 1603, and at the same time, the unit is connected. The unit solder is melted, and each unit is bonded to the A] base 1401. The resin case 1402 is bonded to the A1 base 1401 at a resin case bonding portion 1604 using an adhesive. This embodiment is different from the second embodiment in that the resin case 1442 does not have the external anode electrode 144. Therefore, inspection of each soldering point is easier than in the second embodiment.

 () Cathode connection

Alumina insulating plate] The external cathode electrode 1403 formed integrally with the external anode electrode 1404 via the 106 is connected to the IGBT unit] 407 and Place it on the freewheel die unit 408 and bond it with the external cathode electrode bonding part 165. Before this step, the external anode electrode 144 is bonded with solder with the alumina insulating plate 144 interposed therebetween. In this step, bonding between the anode collective plate Λ149 and the external anode electrode 0404 is also performed simultaneously by the external anode electrode bonding portion 166 電極. FIG. 14 shows the end point of this process.

 (6) Resin filling

 Finally, the filling resin〗 607 is poured to complete the module. As already mentioned, the thermal expansion coefficient of the filled resin 167 was 1 GB T unit

Since the vertical thermal expansion coefficient (approximately equal to copper) of 107 and freewheel diode unit 108 is matched, the vertical thermal stress generated by temperature fluctuations during use is small.

 (Example 4)

 Embodiment 4 will be described with reference to FIGS. 17 to 18. FIG. 17 is a perspective view of another junction type IGB T module according to the present invention as viewed obliquely. The figure shows a stage in the middle of the manufacturing process where no resin case and no external electrodes are mounted so that the inside can be seen.

 The chip configuration in the module is the same as in the second and third embodiments. In other words, there are a total of nine chips, six IGBT chips 102 and three three freewheel diode chips 1703. The structure of the external anode electrode is the same as that of the second embodiment.

This embodiment is different from the second and third embodiments in that each unit is composed of three silicon chips. Two IGBT chips 1702 sandwich the freewheel diode chip 1703.By increasing the number of chips in the unit, the overall structure of the module is simplified. And the module becomes smaller. In the present embodiment, the wiring between the chips in the module is a parallel wiring, and an increase in the size in the left, right, front and rear directions and a complicated structure in the semiconductor unit are avoided.

 In this embodiment, the reason why the size of the semiconductor module is reduced will be described. A distance must be provided around the outer periphery of each unit to ensure insulation. Because this distance is constant, increasing the number of chips per unit, and consequently reducing the number of units in the module, reduces the overall module size. The dimensions of the semiconductor module of the present embodiment are the same as those of the second embodiment, and the distance for insulation around each unit is 4 mm. Therefore, in Example 2, since three units are arranged in the horizontal direction and the front-rear direction, the total distance for insulation is 24 mm. On the other hand, in this embodiment, since the units are arranged in a horizontal line, the lateral direction is the same as that of the second embodiment, and the force is 24 units for 3 units, and the unit is 1 unit in the front-rear direction. Is only 8 mm. It shrinks 16 mm in the front-rear direction compared to the second embodiment. Since the wiring in the unit is arranged in parallel, there is no increase in dimensions due to wiring, and the decrease in insulation distance leads to a reduction in overall dimensions.

In the present embodiment, not only the reduction of the dimensions but also the simplification of the structure is realized. The simplification of the structure according to the present embodiment will be described. Comparing FIG. 17 with FIGS. 1 and 14, it is clear that the anode wiring in the module from the semiconductor unit to the external electrodes is simplified. Since there is essentially no difference between the anode wiring in FIGS. 1 and 14, the differences between FIGS. 1 and 17 will be described. In Fig. 1, there are six internal anode electrodes 111 of IGBT unit 113 and three internal anode electrodes 117 of freewheel diode unit 119, which is the same as the number of chips. In order to lead these to one external node '¾ pole 104, A120, Anode plywood B122 and Anode assembly C122 were required. On the other hand, in this embodiment, since there are only three internal anode common electrodes 1707, there is only the anode collective plate 1711. In this embodiment, the simplification can be achieved because a parallel circuit of three nodes of anodes is completed in each unit.

The reason why the unit is constituted by three chips in this embodiment is not necessarily the optimum number of chips in the _c semiconductor unit. This is because there are benefits due to less and benefits due to more. However, if units having different numbers of chips are mixed in the same module, the degree of freedom in unit arrangement and the degree of chip selection are reduced, and the balance between current and heat generation is unfavorably deteriorated. Therefore, in the present embodiment, one module is composed of nine chips, and the number of chips per unit is one or three. Next, which chip is mounted on the unit will be described. IGBT chip 1BT0 2 only unit and free wheel diode chip】 ΊIt is also possible to use 0 3 only unit. However, in this embodiment, a unit in which the IGBT chip 1702 and the freewheel diode chip 1703 are mixed is used. List the reasons.

 (1) All units are the same, which is convenient for both selecting the units to be mounted on the module and arranging the units.

(2) The IGBT chip 1Ί02 and the freewheel diode chip 1703 are not energized at the same time. Energization is always one side. Therefore, the mixed unit of the IGBT chip 1702 and the freewheel diode chip 1703 has less current flow. In addition, the flow to each unit Current is the same at all times, always balanced and electrically balanced.

 (3) Since the IGBT chip 1Ί02 and the freewheel diode chip 1703 are not energized at the same time, a mixed unit of 13 IG chips 17 チ ッ プ 02 and freewheel diode chip 1 103 Has a better thermal balance. In particular, in the case of motor drive, current mainly flows through the IGBT chip 1702 when the rotation speed increases, and current mainly flows through the free wheel diode chip 1703 when the rotation speed decreases. At a somewhat longer time interval, the two naturally differ. Therefore, a mixed unit of 1 () 8 chips 1Ί02 and a freewheel diode chip 1Ί03 is more advantageous from a mature balance with a long time constant. A free wheel diode chip 1Ί03 is placed in the center to improve the balance in the unit.

 In each unit, a simple parallel circuit was used to minimize the main current flow in the horizontal direction.

The structure of the semiconductor unit will be described with reference to FIG. FIG. 18 is a cross section of the unit of FIG. 17 as viewed from the left. Here, the indication of the solder layer to which each member is adhered is omitted. A free wheel diode chip 1Ί03 is placed in the center of the unit, and IGBT chips 1702 are placed on the left and right. The bottom Mo buffer plate 1706 under each chip is the same member in all three places. The components on each chip are different between the IGBT chip 102 and the free-wheel diode chip 1703 due to the presence of gates and the difference in chip thickness. Regarding the gate, a through hole was made in the center of the Mo buffer plate 1704 on the chip, and a groove for taking out the gate wiring 1712 was formed at the bottom of the height adjustment copper block 1705. dig. This groove Is dug perpendicular to the paper in the figure and from the paper toward the front. Further, when the IGBT chip 1Ί02 and the free wheel diode chip 1703 are each designed for optimization, the IGBT chip 1702 becomes thicker than the free wheel diode chip 1703. Regarding this difference in chip thickness, the Mo buffer plate 1704 on the chip on the IGBT chip 1Ί02 and the Mo buffer plate 1709 on the chip on the freewheel diode chip 1703 The difference in thickness is absorbed.

 The internal anode common electrode 177 under the chip has a function of not only collecting the current of the three chips but also dispersing the heat generated from the three chips. Copper with high thermal conductivity is used and thickened, and on the downstream side there is alumina, which has a lower thermal conductivity than copper, so that heat flows easily in the horizontal direction. By mixing the IGBT chip 102 and the freewheel diode chip 1703 in one unit, electrical and thermal dispersion is achieved.

 (Example 5)

FIG. 19 shows a fifth embodiment. This embodiment is different from embodiment 4 shown in FIG. 18 in that a copper block 1901 for height adjustment is connected at the upper part and integrated. The height adjustment copper block 1901 is integrated to expand the heat flow path before heat flows into the external force source electrode (not shown). The upper part of the copper block for height adjustment】 901 touches an external cathode electrode (not shown). As described in FIG. 7, the heat that has passed through the on-chip buffer plate 706 spreads laterally in the external force source electrode, and the thermal resistance of the total decreases. The same is realized in this embodiment. Therefore, the basic flow does not differ between the fourth embodiment and the present embodiment. However, by spreading the heat of the three chips in advance within the height adjustment copper block 19 Decrease. In particular, since the left and right IGBT chips 1702 and the center free-wheel diode chip 1703 do not conduct electricity at the same time, the heat diffusion effect of this connection is effective in lowering the total thermal resistance. Function. In other words, when the left and right IGBT chips 1702 are heated, the copper block for height adjustment above the freewheel diode chip 1 03 in the center

19001 functions as an area to spread the heat, and conversely, the freewheeling diode tip 1703 in the center (left and right when hot) 90 1 functions as an area for spreading heat. The advantage of the multi-chip simulation is further emphasized by the configuration of the present embodiment.

(Example 6)

Embodiment 6 will be described with reference to FIG. FIG. 20 is a perspective view of a part of the IGBT unit in another junction type IGBT module according to the present embodiment. The overall structure of the module is the same as in the second embodiment. In the figure, the portion below the IGBT chip 201 is omitted because it is the same as in the second embodiment. The feature of this embodiment is the position of the gate of the IGBT chip 2001. Gate pad 209 on chip Except for the withstand voltage area 209 around the chip, it is arranged at the end of the circuit formation area of the 1 GBT chip 201. In any of the means, operation, and embodiments, in the present invention, the gate is disposed at the center of the chip. This is because it is appropriate to provide a gate pad at the center of the chip for the convenience of voltage transfer within the chip ₍ however, when mounted in a module, the gate in the center of the chip is inconvenient.

Therefore, in the present embodiment, the gate pad 208 is arranged around the circuit forming area as a structure convenient for mounting in a module. As a result, the unit structure is simpler than when the gate is pulled out from the center of the chip. Has been Specifically, there is not a hole in the center of the M 0 buffer plate 2 on the chip, but a cut is made in the circumference IS, M 0 on the chip rather than making a hole. Processing of the buffer plate 200 is easy. Further, even if the Mo buffer plate 200 is provided on the chip, the gate solder 2007 on the gate pad 2008 can be seen from the outside. Inspection of the adhesion state of the gate wiring 206, which is almost impossible in other embodiments, can be easily performed. In the figure, a cut is also made in the lower part of the height adjustment copper block 203, but this cut may not be necessary depending on how the gate wiring 206 is taken out. In that case, processing of the copper block 203 is further facilitated.

 In this embodiment, a cut is made in the Mo buffer plate 200 on the chip, and the force source pad on the chip (Mo buffer on the chip with the solder 204 on the chip) is used. By reducing the size of the portion that is bonded to the plate 200), it is possible to make the Mo buffer plate 2002 with no cut. This will further simplify the structure. Although the wiring of the gate is slightly complicated, two gate pads 208 may be arranged on opposite sides of the chip. Considering the propagation of gate voltage in the chip, it is better to arrange two gate pads 208 on opposite sides of the chip. In trade-off with electrical characteristics, you can select whether to use one or two gate pads.

 (Example 7)

Example 7 will be described with reference to FIG. FIG. 21 is a perspective view of a part of an IGBT unit in another junction type IGBT module according to the present embodiment. The overall structure of the module is the same as in the second embodiment. In the figure, the part above the same IGBT chip 2 101 as in Example 2 is shown. The portion below the minute and internal anode electrodes 210 is omitted. The feature of this embodiment is the tip portion 210 of the anode terminal arm 210. At this point, the anode terminal arm 210 is bonded to the anode collective plate A (not shown) by soldering. As described in the second embodiment, the modular semiconductor device according to the present invention has many soldering points. In order to solder all of them well, various production techniques are required. The present embodiment is one of them.

 The end of the anode terminal arm 210 has a recess. Work so that the tip of anode collective plate A (not shown) fits into this part. When soldering the anode terminal arm tip 210, many members are soldered at the same time. First, the anode terminal arm tip 210 and the anode assembly A are fitted together. Take steps to align other components. Since it is fitted, even if a slight left and right front and rear force is applied when joining other members, the anode terminal arm tip 210 and the anode assembly plate A do not separate.

 In addition, a solder is previously inserted into the recess formed in the tip 210 of the anode terminal arm. As can be seen from the figure, the end of the anode terminal arm 2 107 forms a type of container, so there is no problem of solder spilling before soldering or flowing out during soldering.

 By adopting the tip processing described in the present embodiment also in other bonding portions, more reliable soldering can be realized.

 (Example 8)

Example 8 will be described with reference to FIG. FIG. 22 shows an inverter device for driving a motor according to the present invention. The junction type semiconductor module 222 shown in the figure is a module according to the second embodiment. Joined half The conductor module 2201 is hidden behind an external anode electrode (upper anode busbars 2211ΰ to 2218, and output busbars 221 3 to 2215) on the back side of the figure. It is placed on the anode side water-cooled heat sink 222 in the direction in which the external gate electrode 222 is arranged on the near side. In the present embodiment, two junction type semiconductor modules 222 were connected in parallel to form one switching unit from the relationship with the output current. Also, considering the voltage jump at the time of switching, etc., the withstand voltage of the junction type semiconductor module 222 is larger than the voltage between the DC plus bus bar 222 and the DC minus bus bar 202. Therefore, it was not a three-level but a two-level inverter. As a result, a total of four junction-type semiconductor modules 2221, two parallel in the upper part and two in the lower part, are in charge of one phase. As inverters, a total of 12 junction semiconductor modules 222 generate three-phase alternating current from direct current.

The electrical configuration will be described. The leftmost U-phase is described on behalf of each phase. The U-phase upper anode bus bar 2 2 16 force and the voltage (shown below) are placed on the external anode electrode (below the U-phase upper anode bus bar 2 2 16) of the upper junction type semiconductor module 222. Z) is fixed at. The U-phase upper anode busbar 2 2 16 is connected to the external anode electrodes of the two modules, so that the front side is forked. The back side merges into one to connect to the DC plus busbar 222. In the figure, the junction is not visible because it is directly below the busbar 222 on the DC plus side. On the external cathode electrode 222 of the upper junction type semiconductor module 222, the back side of the U-phase output busbar 222 is fixed by a bolt (not shown). To connect the external cathode electrodes of both modules in parallel, the external cathode electrodes of two modules are attached to one plate. It is shaped like The U-phase output busbar 2 13 in front of the external cathode electrode 2 2 2 7 is the external gate electrode 2 2 1 2 below and the external gate wiring 2 2 2 In order to avoid contact with 5, the width is reduced by avoiding over the external gate electrode 2 2 1 2. The front side of the U-phase output bus bar 2213 is bifurcated and fixed to the external anode electrode of the lower junction type semiconductor module 2221 with a bolt (not shown). In the center of the U-phase output bus bar 222, there is a U-phase output terminal 222 for the U-phase output. On the outer cathode electrode 222 of the lower junction type semiconductor module 222, the back side of the U-phase lower force source bar 222 is fixed with bolts (not shown). I have. The front side of the lower U-phase force bus bar 2 2 2 2 has a small width avoiding the outer gate electrode 2 2 1 2, similarly to the U-phase output bus bar 2 2 13. Then, a direct-current minus busbar 2203 is fixed thereon with bolts (not shown). The external gate electrode 222 is connected to an external gate wiring 222 volts (not shown). The external gate wiring 2 2 5 connected to the two parallel junction type semiconductor modules 220 1 is connected at the center of both modules using the gap on the front side of both modules, and connected to the front side of the inverter. Drawn out. This portion is the junction gate wiring 222. Here, one merged gate wiring 222 pulled out from the upper two junction-type semiconductor modules 2221, and one pulled out from the lower junction-type semiconductor module 2221 Two confluent gate wirings, 2 2 2 6, appear. Bolt (not shown) DC positive busbar 2202 on U-phase upper anode busbar 2 2 16, V-phase upper anode busbar 2 2 17, and W-phase upper anode busbar 2 2 18 Install and connect to the positive side of the power supply. U-phase lower cathode busbar 2 2 2 2, V-phase lower cathode busbar 2 2 2 3 Attach the DC negative busbar 2203 to the W phase lower cathode busbar 2222 with a bolt (not shown), and connect to the negative side of the power supply.

The thermal configuration will be described. The heat generated from the anode side flows into the anode-side water-cooled heat sink 222 and is radiated outside the inverter device. In order to improve the thermal contact between the bottom surface of each bonded semiconductor module 222 and the anode-side water-cooled heat sink 222, a thermal conductive grease is sandwiched between them. The heat generated from the force source is described. In the upper junction type semiconductor module 222, the U-phase upper power source water-cooled heat sink fixed on the U-phase output bus bar 222, busbar for the V-phase output 2220 V-phase upper cathode water-cooled heat sink fixed on 2214, and W-phase upper power sink water-cooled heat sink fixed on W-phase output busbar 2 2 15 The heat is radiated to the outside of the inverter device. Each heat sink is provided with two water pipe connections for power source water cooling heat sinks 2 2 1 1, and water is supplied by electrically insulating pipes. An external conductive grease containing fine silver particles is used at the contact between the external cathode electrode 2 227 and each output bus bar 221 3 to 221 5 to allow current and heat to flow. Sandwich. Since there is no need to supply current to the contact between each output ffl bus bar 2 2 1 3 to 2 2] 5 and each upper cathode water-cooled heat sink 2 205 to 220 7, the anode Sandwich the same thermal grease as that of the grease side. The heat generated from the external cathode electrodes 2 227 of the six lower junction-type semiconductor modules 220 1 is transferred to the lower cathode bus bars 2 2 2 to 2 2 4 of each lower cathode bus bar. It is led to a water-cooled heat sink 222 to 222 and exits the inverter. Also here, the details are omitted because they are the same as the upper module. The heat sink on the force source side was manufactured separately from the electrodes. Thermally integrated Although it is more advantageous, it is because priority was given to ease of production. For the same reason, all six water-cooled heat sinks on the cathode side have the same shape. One of the reasons for this is that the heat transfer section (external cathode electrode 222) has the same shape in both the lower module and the lower module.

 In this embodiment, a structure in which heat is taken out by a water-cooled heat sink on both the anode side and the power source side is adopted. Depending on the installation location and the heat generated from the module, other cooling means such as air cooling and heat pipes can be adopted. On the anode side, since there is no restriction on wiring, a cooling structure using a direct air-cooled heat pipe is easy.

 (Example 9)

The ninth embodiment will be described with reference to FIGS. 23 to 25. FIG. 23 is a perspective view of still another junction type IGBT module according to the present embodiment. The features of the junction type semiconductor module according to the present embodiment are that copper bases dedicated to heat transfer are provided on the upper and lower sides, and external electrode terminals are provided on the side surface of the resin case. The bottom of the module is a lower copper base 2301. This is the same as in Example 2 except that a thick copper plate is plated with Ni. The lower copper base 2301, which is electrically insulated from the internal circuit, has an upper copper base 2302 on the top surface of the _t module. This is basically the same as the lower copper base 2301, and is also electrically isolated from the internal circuitry. The difference from the lower copper base 2301 is that a resin filling hole 2307 is opened. The external electrode terminals are located almost at the center of the side surface of the resin case 2303 in the vertical direction, and are connected to the external anode electrode 230, external gate electrode 230, and external force source electrode 230. It is arranged in the order of. The height of the resin case 2303 is about 80 mm to secure the creepage and clearance from the lower copper base '2301 and upper copper base 2302 to each terminal. . other In this embodiment, the distance between each terminal and the base can be secured over the entire height of the side surface of the case. In this embodiment, however, each terminal and the lower This is because both the copper base 2301, the distance between each terminal and the upper copper base 2302 must be secured.

 FIG. 24 shows a cross section taken along a plane passing through the center of the external force source electrode 2305 of this embodiment. In this cross section, two units equipped with one IGBT chip 2401 are lined up. Unlike the other embodiments described above, in this embodiment, the upper surface of the module is also electrically insulated from the inside, so that the structure of the unit is different from the other embodiments.

The structure of the semiconductor unit of this embodiment will be described. The bottom surface of the unit is a ceramic insulating plate 2416 under the anode made of aluminum. This bottom surface and the lower copper base 2301 are bonded with solder 217 below the ceramic insulating plate below the anode. Above the anode ceramic insulating plate 2416, there is an internal anode electrode 2 4 13. The ceramic insulating plate 2416 under the anode and the internal anode electrode 2413 are bonded together with the solder 2415 under the anode electrode. An anode terminal arm 2 4 1 4 protrudes from the internal anode electrode 2 4] 3. This is the electrical connection point on the node side of the unit. Above the internal anode electrode 2 4 13, there is an under chip buffer plate 2 4 1 1 composed of M 0, which is bonded with solder 2 4 12 for the under chip buffer plate. Since copper has a larger coefficient of thermal expansion than silicon, it has a role of buffering with silicon by sandwiching Mo with an intermediate coefficient of thermal expansion between the two. There is an IGBT chip 2401. Between these two, there is a solder 2401 below the chip, which plays the role of bonding.Around the top surface of the 1GBT chip 2401, FLR is applied to withstand voltage. Therefore, the upper half of the unit is mounted avoiding this part. Have been. Above the IGBT chip 2401, there is an on-chip buffer plate for buffering the difference in thermal expansion coefficient, which is bonded with on-chip solder. An internal cathode electrode 2405 is provided on the buffer board on chip 2403, and is bonded with solder 2404 for the buffer board on chip. Further, a power source terminal arm 2406 extends from the internal power source electrode 2405, and constitutes a power source side electrical connection point of the unit. A hole is formed in the center of the buffer plate on chip 2403 and the internal force source electrode 2405, and a gate wiring 2418 passes through the center. A ceramic insulating plate 2408 on the cathode is bonded on the internal cathode electrode 2405 with solder 2407 on the force cathode electrode. The ceramic insulating plate 2408 on the cathode is the uppermost component of the unit of the present embodiment. The ceramic upper insulating plate 2408 is adhered to the upper copper base 2302 with solder 409 above the ceramic upper insulating plate.

 Next, the outer circumference and the current path will be described. A resin case 2303 is adhered on the lower copper base 2301 with an adhesive (not shown). The external cathode electrode 2305 is arranged substantially at the center. The external gate electrode 230 and the external anode electrode 230

(Back), all of which are located at the same distance from the lower copper base 2301 and the upper copper base 2302 as the external cathode electrode 2305, It is hidden by 2305 and cannot be seen in this section ¾. The external cathode electrode 2305 is connected to the external cathode electrode mounting part via the cathode terminal connecting wire 2411 from the cathode terminal arm 2406 of each unit.

The force source current collected at 2 4 2 2 flows. Similarly, an anode current flows from the external anode electrode 2304 (not shown) to the external anode electrode mounting portion (not shown), and passes through the anode electrode connection wiring 240. This leads to the anode terminal arm 2 4 1 4 of each unit. Also, the gate signal passes from the external gate electrode 2306 to the external gate electrode mounting portion (not visible), passes through the gate collective wiring 2421, and the gate signal of each unit. It leads to wiring 2 4 1 8.

 Heat flows up and down from both sides of the IGBT chip 2401, and passes through the upper ceramic insulating plate 2408 and the lower ceramic insulating plate 2410 of each unit. Lead to copper base 2302 and lower copper base 2301.

 The structural features of this embodiment will be described. In this embodiment, the unit is mounted on the lower copper base 2301, and the upper copper base 2302 is mounted thereon. In order to reduce the thermal resistance, the members from the lower copper base 2301 to the upper copper base 2302 must be securely bonded. Therefore, the upper copper base 2302 is placed on each unit. To ensure this, there is a gap between the upper copper base 2302 and the resin case 2303. This gap also has the function of bleeding air when filling the inside with the filling resin 242. Next, a description will be given of the resin filling hole 2307. The resin is filled in a state where all the members are assembled. At that time, the resin is charged through this hole. When explaining the process using Fig. 25, the function of this hole will be mentioned.

 The manufacturing method of this embodiment will be described with reference to FIG. In the figure, the left half shows the manufacturing process, and the right half shows the state after the end of each process (before the start of the next process).

 (1) Unit assembly

The IGB chip 2401 and the freewheel diode chip 2501, which have been evaluated as having good quality by examining static characteristics in advance, are mounted on the unit. In the figure, This shows a unit equipped with a 1G13T chip 2401. The feature of the unit of this embodiment is that it has a cathode terminal arm 2406 and a ceramic insulating plate 2408 on a force source. Although the solder layer is omitted in this figure, this unit has six solder layers. In the present embodiment, these soldering operations are collectively performed. Of course, soldering may be performed several times.

 (2) Unit sorting

 Sorting in chip units is performed in the same manner as in the second embodiment. At the same time, measurement of the switching characteristics and the safe operation area will be performed by the unit. In the cross section shown in FIG. 24, each of the two units had the IG chip 241 mounted thereon. As in the other embodiments described so far, in this embodiment, the I〇8 chip 2401 and the freewheel diode chip 2501 are mounted in the same module, and both are connected in anti-parallel. Connecting. In Fig. 25, the selected non-defective IG Β chip 241 and the freewheel diode chip 2501 are displayed one by one.

 (3) Unit layout

 The selected units are arranged on the lower copper base 2301. Soldering with a ceramic under the ceramic insulating plate under the ground may be performed in this step, or may be performed in the next step or in the next step. In the present embodiment, unlike the second embodiment, the soldering of the ceramic under the ceramic insulating plate 2417 is performed in this step as a reliable method with a small number of steps.

 (4) External electrode terminal connection

First, the resin case 2303 is placed on the lower copper base 2301 via the resin case bonding portion 2505. Next, the anode electrode connection wiring 2420 The anode terminal arm 2 4 1 4 is connected to the external anode electrode 2 3 4 (not shown) by using a wire. Next, the power source terminal arm 2406 and the external power source electrode 2305 are connected using the power source electrode connection wiring 2419. Lastly, the gate wiring 2418 is connected to the external gate electrode 2306 (not shown) using the gate collective wiring 2421. Thereafter, solder is potted to the anode terminal arm bonding section 2503, the cathode terminal arm bonding section 2502, and the gate wiring bonding section 2504. Place in a furnace and simultaneously bond the soldering points and the resin case bonding portion 2505 with an adhesive.

 (5) Upper copper base adhesive

 Glue the upper copper base 2302. After this step, the inside of the module becomes almost invisible due to the upper copper base 2302, so it is necessary to complete the internal inspection before this step. The same force is applied to other processes. If there are many soldering points as in this embodiment, the solder at the previously bonded point may re-melt during the work, so the It is necessary to be careful not to give vibration.

 (6) Resin filling

Finally, the filling resin 2 4 2 3 is poured to complete the module. The resin is poured through the resin filling hole 2307 opened in the upper copper base 2302. In this process, it is difficult to observe how much resin is filled in the module. Therefore, a method is used in which the required resin volume is determined in advance, and a predetermined volume of resin is poured. During the filling of the filling resin 2 4 2 3, the air inside the module is released by using the small gap between the resin case 2 3 0 3 and the upper copper base 2 3 2, so the filling is smooth Proceed to As in Example 2, a resin in which a thermosetting resin containing silicon oxide as a main component and a filler for adjusting a thermal expansion coefficient is mixed is used. did.

 In the above embodiment, the description of the combination of the IGBT and the diode as the junction-semiconductor module has been described. Since the IGBT is a bipolar transistor having a MOS drive unit, the operation of the main circuit is a normal bipolar transistor. Same as transistor. Therefore, the IGBT of the above embodiment is replaced with a current-driven bipolar transistor, and the structure and the manufacturing method are not changed even when a combination of a current-driven bipolar transistor and a diode is used, and the present invention is applied to a bipolar transistor. Applicable.

 The IGCT is a device in which the transistor portion of the IGBT is replaced with a thyristor, and operates closer to the IGBT than a normal bipolar transistor. Therefore, the present invention can be applied to a semiconductor module in which IGCT and a diode are combined.

 According to the present invention, it is possible to provide a structure of a non-welding type module that is advantageous when the number of elements is increased, and that draws heat not only from the bottom surface but also from the top surface of the module.

Claims

The scope of the claims

 1. a plurality of power semiconductor elements;

 A plurality of insulating substrates to which the power semiconductor element is fixed via a bonding material containing metal;

 At least one first heat sink fixed to the power semiconductor element via a bonding material containing metal,

 At least one second heat sink, which is fixed to the plurality of insulating substrates via a bonding material containing metal,

A junction type semiconductor module having:

 2. The junction type semiconductor module according to claim 1, wherein at least one piece is provided between the power semiconductor element and the first heat sink or between the power semiconductor element and the insulating substrate. A metal member is interposed.

 3. The junction type semiconductor module according to claim 2, wherein the metal member is an electrode electrically connected to the power semiconductor element.

 4. The junction type semiconductor module according to claim 3, further comprising an external electrode terminal electrically connected to the metal member.

 5. The junction type semiconductor module according to claim 2, wherein the metal member is a buffer plate.

 6. The junction type semiconductor module according to claim 5, wherein the buffer plate is a tungsten plate or a molybdenum plate.

 7. The junction type semiconductor module according to claim 2, wherein there are a plurality of the metal members.

8. The junction type semiconductor module according to claim 6, wherein the plurality of metal members include one of a tungsten plate and a molybdenum plate and one of a copper plate and an aluminum plate.

9. The junction semiconductor module according to claim 1, wherein the first heat sink is an external electrode terminal.

 10. The junction type semiconductor module according to claim 1, wherein the material of the first heat sink and the material of the second heat sink are substantially the same.

 11. The junction type semiconductor module according to claim 10, wherein the material is selected from the group consisting of copper, copper alloy, aluminum, and aluminum alloy.

12. The bonded semiconductor module according to claim 1, further comprising a resin case, wherein the resin case houses the power cow conductor element, and wherein the first heat sink is located on an upper surface of the resin case. The second heat sink is located on the bottom surface of the resin case.

 13. The bonded semiconductor module according to claim 12, further comprising an external electrode terminal, wherein the external electrode terminal is located on a side surface of the resin case.

 14. The junction type semiconductor module according to claim 1, wherein the insulating substrate is a ceramic substrate.

 15. The bonding semiconductor module according to claim 1, wherein the bonding material is solder.