JP2010283169A - Method of manufacturing semiconductor device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a method of manufacturing a semiconductor device that is more inexpensive than before with conventional normal general module constitution, and secures bonding reliability of a solder bonding part. <P>SOLUTION: This invention relates to the method of manufacturing the semiconductor device, including: a circuit board P formed by bonding metal plates to both surfaces of an insulating substrate; one semiconductor element bonded to an external surface of one metal plate through first solder; and a base plate for heat dissipation, bonded to an external surface of the other metal plate through second solder. The method includes steps of: manufacturing the circuit board P in such a way that the ratio (a) of the sum t<SB>M</SB>of plate thicknesses of both metal plates M1, M2 to the plate thickness t<SB>C</SB>of the insulating substrate C falls within a predetermined range for securing durability of solder H1 and solder H2 which are lead-free solder of the same kind, to temperature stress; and carrying out, under an identical heating condition, processing for bonding the semiconductors S, S' to the external surface of one metal plate M1 of the circuit board P through the first solder H1 concurrently with processing for bonding the base plate B to the external surface of the other metal plate M2 through the second solder H2. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention provides a circuit board formed by bonding metal plates to both surfaces of an insulating substrate, at least one semiconductor element bonded to the outer surface of one metal plate via a first solder, and the other metal plate. The present invention relates to a method of manufacturing a semiconductor device including a heat dissipation base plate joined to an outer surface via a second solder.

  The power module used in the semiconductor device, such as a motor drive control system, requires sufficient insulation and heat dissipation as the output increases. Therefore, the first is bonded to one surface using a ceramic insulating substrate. Tin having a relatively low melting temperature so as not to remelt the lead-based solder joint to the second copper plate joined to the copper plate with a lead-type solder having a high melting temperature and then joined to the other surface of the insulating substrate. The base plate for heat dissipation is joined with lead eutectic solder.

  On the other hand, in recent years, lead-free solder materials have been promoted to protect the environment, but the lead-free solder has a melting temperature higher than that of tin-lead eutectic solder. Since the soldering temperature is lower than the melting temperature, when solder bonding is performed with the above-described conventional method using this lead-free solder, the solder bonding portion on the semiconductor element side that has been bonded first is reheated by the heat during the bonding process of the base plate for heat dissipation. There has been a problem that the bonding reliability is remarkably lowered by melting.

  Further, since lead-free solder is harder than lead-based or tin-lead-based solder, there is a problem that once a crack occurs, the progress is rapid and the durability is inferior.

  Therefore, in order to solve the former problem, a method of joining the semiconductor element and the heat radiating base plate using two kinds of lead-free solders having different melting temperatures (for example, refer to Patent Document 1) or the latter problem. In order to solve this problem, a technique is known in which a rare metal such as bismuth or indium is added to lead-free solder to increase durability (for example, see Patent Document 2 below). Furthermore, a special technique is known in which the structure of the power module is significantly changed, and the solder joints are hardened with a mold resin and sealed with resin.

JP 2006-237057 A JP 2007-141948 A

  However, in the method of Patent Document 1, it is necessary to specially select and prepare two kinds of lead-free solders having different melting temperatures, and the two kinds of solders are used for joining a semiconductor element to an insulating substrate and dissipating heat. It is necessary to use differently for the base plate joining, and the handling thereof is complicated as a whole, and in the method of Patent Document 2 above, it is necessary to specially use an expensive rare metal as an additive to lead-free solder. In either method, there is a problem that the material cost and the management cost are soaring and the manufacturing process is forced to change, resulting in a significant increase in cost.

  The present invention has been made in view of such circumstances, and provides a method of manufacturing a semiconductor device that has a conventional and ordinary general module configuration, is cheaper than the conventional one, and can ensure the bonding reliability of the solder joint portion. For the purpose.

In order to achieve the above object, the present invention provides at least one semiconductor element bonded to the outer surface of one metal plate via a first solder and bonded to the outer surface of the other metal plate via a second solder. In a method for manufacturing a semiconductor device provided with a heat dissipation base plate,
The circuit board is manufactured such that the ratio of the sum of the thicknesses of the one and the other metal plates to the thickness of the insulating substrate is within a predetermined range that can ensure durability against temperature stress of each solder. A step of bonding the semiconductor to the outer surface of the one metal plate of the circuit board via a first solder, and bonding the base plate to the outer surface of the other metal plate via the second solder. A first feature is that the first and second solders are made of the same kind of lead-free solder material.

  In addition to the configuration of the first feature, the present invention has a second feature that the ratio is 1.5 or more and 5.5 or less.

  According to the third aspect of the present invention, in addition to the first or second feature, the third and second solders are SnCu-based, SnAg-based, or SnAgCu-based alloys, and do not include a rare metal. Features.

  In the present invention, in addition to any of the configurations of the first to third features, the treatment is performed in a reflow furnace, and the reflow temperature at that time is set to 240 ° C. or more and 320 ° C. or less. This is the fourth feature.

  In the present invention, “the same kind of lead-free solder material” refers to a lead-free solder material composed of an alloy having the same main component. For example, a lead-free solder material selected from SnCu-based, SnAg-based or SnAgCu-based alloys is used. It is done. The “same kind of lead-free solder material” of the present invention may or may not contain an additive, but if included, the composition and amount of the additive contained in the The first and second solders may be the same or different. In any case, the first and second solders have a melting point difference within 20 ° C. (that is, a range in which heating conditions do not need to be changed as in Patent Document 1 (Japanese Patent Application Laid-Open No. 2006-237057)). It is desirable to select the composition and amount of the additive for the second solder.

  According to the first and second features of the present invention, the durability reliability of the solder joint portion for joining the first and second metal plates on both sides of the circuit board in the semiconductor device to the element and the heat radiating base plate, respectively. Can be sufficiently secured without using a special technique such as resin sealing and without using a special solder material to which an expensive additive such as a rare metal is added. In addition, since it is possible to perform joint processing using the same kind of solder at both solder joint portions, the joint processing work can be easily and quickly performed efficiently. Therefore, even if the same kind of lead-free solder is used for both solder joints, it is possible to achieve material cost reduction and cost reduction by reducing processes while ensuring durability reliability of the joints.

  In addition, according to the third feature of the present invention, it is possible to use an inexpensive lead-free solder that is conventionally used, thereby further reducing the cost.

  Further, according to the fourth feature of the present invention, the entire solder is uniformly melted in a reflow furnace and each solder is simultaneously heated and melted at a heating temperature at which durability reliability of the joint portion with the semiconductor element is obtained. Bonding can be performed.

Sectional drawing of the principal part of the power module which concerns on one Example of this invention Exploded view of the main part of the power module Graph showing the results of crack growth analysis by heat of solder on the base plate side for heat dissipation Graph showing the relationship between generated strain and temperature cycle obtained using Coffin-Manson rule by analysis of thermal fatigue test on solder joint of the power module Graph showing the relationship between average thermal expansion coefficient α _{ave of} circuit board and cycle number A graph showing the relationship between the average thermal expansion coefficient α _ave of the circuit board and the ratio a of the sum t _M of the thicknesses of both copper plates in the circuit board to the thickness t _C of the insulating board A graph showing the relationship between the thickness of the copper plate on the circuit board and the number of cycles

  Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings.

  1 and 2, a power module PM as a semiconductor device includes first and second copper plates as first and second metal plates on both surfaces of an insulating substrate C mainly made of a ceramic material such as silicon nitride. A circuit board P formed by integrally joining M1 and M2, respectively, at least one semiconductor element S, S ′ joined to the outer surface of the first copper plate M1 via a first solder H1, and a second copper plate M2 And a heat-radiating base plate B made of copper, which is joined to the outer surface via the second solder H2, and each of the solders H1 and H2 is made of a lead-free solder material.

  The structure of the circuit board P is basically the same as a conventionally known DCB board.

  The first and second solders H1 and H2 are the same kind (the same in this embodiment) of lead-free solder materials, for example, SnCu-based, SnAg-based or SnAgCu-based alloys, which do not contain any additives. An inexpensive lead-free solder material is used that does not contain or includes at least one of Ni, Co, or Ge, which is an expensive additive to increase durability For example, rare metals such as bismuth and indium are not added. These lead-free solders have a melting temperature higher than that of tin-lead eutectic solder and lower than that of lead-based solder.

  Therefore, when these SnCu-based, SnAg-based or SnAgCu-based lead-free solder materials are used as the first solder H1 on the semiconductor element S, S 'side, the operating temperature of the element (160 ° C or lower) is reached. On the other hand, it has a sufficiently high melting temperature (about 220 ° C.), and does not remelt due to heat generated during use of a semiconductor element as in the case of a conventional lead-based high melting point solder. As a result of the analysis by the present inventor, when the second solder H2 is used, it is expected that the durability against heat is expected to be higher than when the conventional tin-lead eutectic solder is used.

  That is, in FIG. 3, the heat dissipation base plate B is joined to two copper plates of different thicknesses (0.3 mm and 0.5 mm) with tin-lead eutectic solder or the lead-free solder of the present invention. Shows the results of analyzing the relationship between the number of test cycles when a temperature cycle test (TCT) is performed at a predetermined cycle temperature variation range and the crack growth length of the solder joint. When using lead-free solder, the crack propagation length of the solder joint is shorter at the same temperature cycle than when using tin-lead solder, that is, the durability of the solder joint against temperature stress is reduced. Clearly high.

  The temperature cycle test (TCT) used in this example is based on the temperature cycle test in the semiconductor reliability standard established by the JETA, that is, the Japan Electronics and Information Technology Industries Association. When mounted, the test is performed under the conditions determined by the requirements for the number of times to turn on and off in one run and the material characteristics, and the cycle temperature variation range during the test is in the range of −40 ° C. to 105 ° C. Is set to

  As the semiconductor elements S and S ′, conventionally known power semiconductor elements, for example, various elements such as IGBT, MOS-FET, or FWD, are used, which are mainly made of silicon.

  In addition, in the circuit board P, the thickness t1 of the insulating substrate C is equal to the sum t1 of the thicknesses of the first and second copper plates M1 and M2 in order to ensure sufficient durability against the temperature stress of the solders H1 and H2. The ratio a to t2 is set so as to be within a predetermined range (a range of 1.5 or more and 5.5 or less as will be described later) in which durability against temperature stress of the solders H1 and H2 can be secured.

  The present inventor has clarified that the ratio a is an important parameter for ensuring the durability against the temperature stress of each solder H1, H2 from the analysis result such as a fatigue test of the solder joint. The analysis method will be described below.

  First, an analysis model of each component of the circuit board P, the heat radiating base plate B, each of the solders H1 and H2, and the elements S and S ′ is created, and then a thermal fatigue test of the solder joint portion of the semiconductor device is performed. Analysis conditions are set by setting a cycle temperature change width in accordance with a predetermined temperature condition of the temperature cycle test (TCT) described above.

  Next, structural analysis was carried out by actual experiments or computer simulations (simulations), and the relationship between the generated strain at the joints of each solder H1, H2 and the number of cycles in the temperature cycle test was determined. Based on the -Manson rule, it is obtained as shown in FIG. 4, and the durability reliability of the solder joint is evaluated and judged from this relationship. According to FIG. 4, the smaller the generated strain (that is, temperature stress) of the solder joint in each temperature cycle, the smaller the life cycle number (that is, the test until the crack propagation length of the joint reaches the specified limit). It is clear that the durability reliability of the solder joint becomes higher as the number of cycles) increases.

  Based on the analysis result, the thickness of each part of the power module PM (for example, the thickness of the elements S, S ′, the thickness of each solder H1, H2, The thicknesses of the copper plates M1 and M2, the difference between the thicknesses of the copper plates M1 and M2, the thickness of the base plate B, etc.) and the contribution factors such as the thermal expansion coefficient are analyzed. Of these, it was found that the contribution ratios of the thicknesses of the copper plates M1 and M2 were the highest, and the contribution ratios of the thicknesses of the other parts were relatively small.

  Further, according to the analysis result, the first solder H1 on the element side has durability reliability when the average thermal expansion coefficient of the circuit board P is small (that is, as it becomes closer to silicon which is the main material of the elements S and S ′). The second solder H2 on the base plate side becomes higher in durability reliability when the average thermal expansion coefficient of the circuit board P is larger (that is, as it becomes closer to copper as a constituent material of the base plate B). However, this is considered to be due to the fact that the strain generated in each solder H1, H2 with temperature change depends on the difference in the thermal expansion coefficient between the structures sandwiching the solder.

Then, next, when the relationship between the average thermal expansion coefficient α _{ave of} the circuit board P and the life cycle number of the solder joint portion was examined, the result as shown in FIG. 5 was obtained. According to this, the durability reliability of the first solder H1 on the element side becomes higher as the average thermal expansion coefficient α _ave of the circuit board P becomes smaller, while the durability reliability of the second solder H2 on the base plate side is It was confirmed that the average thermal expansion coefficient α _ave of the circuit board P decreases as the average thermal expansion coefficient α _ave decreases, and the two tend to conflict. Therefore, it is necessary to select the thicknesses of the first and second copper plates M1 and M2 based on a specific region of the average thermal expansion coefficient α _ave that satisfies the durability reliability of the solders H1 and H2.

In the selection, the life cycle number of the solder joint part is required as the solder joint part of the power module in the temperature cycle test (TCT) as a guideline for satisfying the durability reliability of the first and second solders H1 and H2. It is assumed that the condition of a predetermined required number of cycles or more (1000 cycles or more in this embodiment) is satisfied. Average thermal expansion coefficient alpha _ave satisfying this condition the circuit board P, in Figure 5, be in the range of 7.5~12ppm / ℃, average thermal expansion coefficient alpha _ave is The use of the circuit board P is in this range, The durability reliability of both the solders H1, H2 can be satisfied at the same time, and the required level of durability reliability as a power module can be satisfied.

By the way, if the ratio of the sum t _M of the thicknesses of the first and second copper plates M1 and M2 to the thickness t _C of the insulating substrate is a, a = t _M / t _C ,
The average thermal expansion coefficient of the circuit board P is α _ave , the thermal expansion coefficient and Young's modulus of the ceramic insulating substrate C are α _C and E _C , respectively, and the thermal expansion coefficient and Young's modulus of the copper material of the copper plates M1 and M2 Α _ave can be expressed by the following equation, where α _M and E _M respectively.

α _ave = (a · α _M · E _M + α _C · E _C ) / (aE _M + E _C )
This equation can be expressed in the graph of FIG. 6 as the relationship between the ratio a and the average thermal expansion coefficient α _ave of the circuit board P. In this graph, the range of the average thermal expansion coefficient α _ave of the circuit board P that satisfies the durability reliability of the first and second solders H1 and H2 as described above (7.5 to 12 ppm / ° C.). The range of the ratio a that satisfies the above is 1.5 to 5.5.

  Therefore, if the thicknesses of the first and second copper plates M1 and M2 are selected so as to satisfy the range of the ratio a, the durability reliability of the first and second solders H1 and H2 can be satisfied at the same time. The thickness of the first and second copper plates M1 and M2 can be selected.

For example, in a standard power module PM in which the thickness t _C of the insulating substrate is 0.32 mm, the ratio a (= t _M / t _C ) is 1.5 ≦ a ≦ 5.5. When applied to calculate the sum t _M of the thicknesses of the first and second copper plates M1 and M2, it is in the range of 0.48 mm to 1.76 mm. On the other hand, FIG. 7 shows the first power module PM with the sum t _M of the thicknesses of the first and second copper plates M1 and M2 as the horizontal axis and the number of temperature cycles in the temperature cycle test (TCT) as the vertical axis. The graph which calculated | required the life cycle number of 2nd solder H1, H2 experimentally is shown. In this graph, when the sum t _{M of the} thicknesses of the first and second copper plates M1 and M2 is in the range of 0.48 mm to 1.76 mm (shown as an OK region in FIG. 7), the first and second solders. It is well known that the number of life cycles of H1 and H2 is equal to or greater than a predetermined required number of cycles (1000 in the embodiment), and each solder joint has sufficient durability reliability.

  When the power module PM is manufactured by soldering the semiconductor elements S, S ′ and the heat radiating base plate B to the copper plates M1, M2 on one side and the other side of the circuit board P configured as described above, the circuit board P The process of joining the semiconductors S and S ′ to the outer surface of the first copper plate M1 on the element side via the first lead-free solder H1, and the second lead-free solder H2 on the outer surface of the second copper plate M2 on the base plate side Thus, the process of joining the base plate B for heat dissipation is performed simultaneously under the same heating conditions.

  These soldering processes are performed in a conventionally known reflow furnace (not shown) as a processing furnace, and the reflow temperature at that time uniformly melts each of the solders H1 and H2 in the reflow furnace, and the semiconductor element. The heating temperature at which the durability reliability of the joint portion with S and S ′ is obtained is set to 240 ° C. or more and 320 ° C. or less. With such a reflow temperature setting, the solders H1 and H2 are simultaneously melted at a heating temperature at which the entire solders H1 and H2 are uniformly melted in the reflow furnace and durability of the joints with the semiconductor elements S and S ′ is obtained. By heating and melting, both solder joints can be accurately joined.

  In the joining process of the solders H1 and H2, for example, the base plate of the circuit board P is sandwiched with a thin plate-like second lead-free solder H2 formed in a predetermined shape at a predetermined position on the base plate B for heat dissipation. The outer surface of the second copper plate M2 on the side is placed, and the semiconductor element S is sandwiched at a predetermined position on the outer surface of the first copper plate M1 on the element side of the circuit board P with a thin plate-like first lead-free solder H1 formed in advance. , S 'are placed in a reflow furnace heated to a predetermined reflow temperature in advance, and heated for a predetermined time to melt each solder H1, H2, and then taken out from the reflow furnace to cool and solidify. So that the soldering process is performed.

Thus, in the circuit board P of the power module PM of the present embodiment, the sum t _M of the thicknesses of the first and second copper plates M1 and M2 on both sides of the insulating substrate C with respect to the plate thickness t _C of the insulating substrate C. The ratio a (= t _M / t _C ) is predetermined in order to ensure durability against temperature stress of the first and second solders H1 and H2 made of the same kind (same in this embodiment) of lead-free solder materials. Is set within a limited range of 1.5, that is, 1.5 or more and 5.5 or less. Moreover, both the solder joints have a sufficiently high melting temperature (about 220 ° C.) with respect to the operating temperature of the elements S and S ′ (160 ° C. or less), so that they can be re-melted due to heat generated during use of the elements. In addition, by using lead-free solder, which is expected to have higher durability against heat than when tin-lead solder is used, these are reflow temperatures (that is, 240 ° C or higher and 320 ° C or lower) in a reflow furnace. It is possible to perform the joining process at once.

  Therefore, the first and second copper plates M1 and M2 on both sides of the circuit board P and the elements S, S ′ and the heat radiating base plate B are joined by first and second solders H1 and H2, respectively. Durability and reliability can be sufficiently secured without using a special technique such as resin sealing and without using a special solder material to which an expensive additive such as a rare metal is added. Moreover, since both the solder joints can be jointly processed, the joint processing operation can be easily and quickly performed efficiently as a whole.

  Thus, an inexpensive lead-free solder material that is conventionally used at the joint between the solders H1 and H2 (that is, an inexpensive SnCu-based, SnAg-based or SnAgCu-based alloy that does not contain a rare metal as an additive). Even if a lead-free solder material) is used, it is possible to achieve a reduction in material cost and cost reduction by reducing processes compared to conventional devices while ensuring sufficient durability reliability of the solder joints.

  Although the embodiments of the present invention have been described above, the present invention is not limited to the above-described embodiments, and various design changes can be made without departing from the present invention described in the claims. Is possible.

  For example, in the above embodiment, a copper plate is used as the metal plate used as the heat radiating base plate. However, in the present invention, the base plate may be composed of a composite material such as copper, aluminum, tungsten, and molybdenum.

B .... Heat dissipation base plate C ... Insulating substrate H1 ... First solder H2 ... Second solder M1 ... First copper plate M2 as first metal plate ... Second metal Second copper plate P as a plate... Circuit board S, S ′... Semiconductor element a... Ratio t _C .. thickness t _{M of} insulating substrate ... first and second metal plates ( Sum of thickness of first and second copper plates)

Claims (4)

A circuit board (P) formed by bonding metal plates (M1, M2) to both surfaces of the insulating substrate (C) and a first solder (H1) to the outer surface of one metal plate (M1). A semiconductor device comprising at least one semiconductor element (S, S ′) and a heat radiating base plate (B) joined to the outer surface of the other metal plate (M2) via a second solder (H2). In the manufacturing method,
The ratio (a) of the sum (t _M ) of the thicknesses of the one and the other metal plates (M1, M2) to the thickness (t _C ) of the insulating substrate (C) is determined by each solder (H1, H2 ) Manufacturing the circuit board (P) so as to be within a predetermined range in which durability against temperature stress can be secured;
The process of joining the semiconductor (S, S ′) to the outer surface of the one metal plate (M1) of the circuit board (P) via the first solder (H1) and the outer surface of the other metal plate (M2) And simultaneously performing the process of joining the base plate (B) via the second solder (H2) under the same heating conditions,
The method for manufacturing a semiconductor device, wherein the first and second solders (H1, H2) are made of the same kind of lead-free solder material.   2. The method of manufacturing a semiconductor device according to claim 1, wherein the ratio (a) is 1.5 or more and 5.5 or less.   The semiconductor device according to claim 1, wherein the first and second solders (H 1, H 2) are SnCu-based, SnAg-based, or SnAgCu-based alloys and do not contain a rare metal. Method.   The method of manufacturing a semiconductor device according to claim 1, wherein the process is performed in a reflow furnace, and a reflow temperature at that time is set to 240 ° C. or more and 320 ° C. or less. .

Images