JP6462172B1 - Heat sink and manufacturing method thereof - Google Patents

Abstract

A heat dissipation plate having a low thermal expansion coefficient and a high thermal conductivity, having a clad structure of a Cu-Mo composite material and a Cu material.
In the sheet thickness direction, Cu layers and Cu-Mo composite layers are alternately stacked to form three or more Cu layers and two or more Cu-Mo composite layers, The outermost layer is a heat radiating plate made of a Cu layer, and the Cu-Mo composite layer has a plate thickness cross-sectional structure in which a flat Mo phase is dispersed in a Cu matrix. Compared to a heat sink with a three-layer clad structure having the same plate thickness and density, the thermal conductivity in the plate thickness direction is increased because the coefficient of thermal expansion is low and the thickness of the outermost Cu layer is reduced.
[Selection] Figure 1

Description

  The present invention relates to a heat radiating plate used for efficiently dissipating heat generated from a heating element such as a semiconductor element and a method for manufacturing the same.

  In order to efficiently dissipate heat generated from the semiconductor element from the semiconductor device, a heat sink (heat sink) is used. This heat sink is required to have high thermal conductivity in terms of its function, and since it is joined to a semiconductor, ceramic circuit board, metal package member, etc. by soldering or brazing, it has a thermal expansion coefficient close to that of the member to be joined (low heat Expansion coefficient).

Conventionally, a Mo-Cu composite material has been used as a heat dissipation plate having high thermal conductivity and low thermal expansion coefficient (for example, Patent Document 1). In general, the Mo-Cu composite material used for the heat sink is formed by pressing Mo powder or a mixed powder of Mo powder and Cu powder into a green compact, and this green compact is subjected to reduction sintering as necessary. Thereafter, Cu infiltration or densification treatment is performed to obtain a Mo—Cu composite material, and the Mo—Cu composite material is rolled. Since Mo hardly dissolves with Cu, this Mo-Cu composite material has a two-phase structure of Mo and Cu, and a heat sink utilizing the characteristics of Mo having a low thermal expansion coefficient and Cu having a high thermal conductivity. be able to.
Patent Document 2 shows a heat dissipation plate based on the Mo-Cu composite material as described above, in which a Cu plate is pressure-bonded to both surfaces of a Mo-Cu composite material obtained through a specific rolling process. The heat radiating plate has a higher thermal conductivity than the [Cu / Mo / Cu] clad material and is excellent in press punchability.

Moreover, it is qualitatively known that the coefficient of thermal expansion is reduced by rolling, and the Mo—Cu composite material is manufactured through the rolling process as described above. Conventionally, since Mo particles are hard and primary particles are small, it is considered that they are not easily deformed by rolling. For this reason, rolling of a Mo—Cu composite material is performed by warm rolling at about 200 to 400 ° C. (Patent Document 1). Further, Patent Document 2 discloses a manufacturing method in which warm rolling is performed by primary rolling and cold rolling is performed by secondary rolling. However, even in this manufacturing method, Mo particles are hardly deformed. On the premise, warm rolling (primary rolling) is an essential process.
In recent years, heat dissipation of a heat sink has become more important due to higher output of semiconductors. On the other hand, there is a high need for miniaturization of semiconductor modules, and the heat radiating plate is also required to radiate heat from a smaller area. Therefore, heat dissipation in the thickness direction is more important than heat dissipation in the plate surface direction.

Japanese Patent Laid-Open No. 11-307701 JP 2001-358266 A

However, according to the study by the present inventor, the heat dissipation plate described in Patent Document 2 is certainly superior in thermal characteristics (low thermal expansion) compared to the Mo—Cu composite material alone described in Patent Document 1. In the clad structure in which the Mo—Cu composite material and the Cu material are laminated, the thermal characteristics (particularly, superior to the Cu / (Cu—Mo) / Cu structure described in Patent Document 2) It has been found that there is an optimum cladding structure that can obtain a thermal conductivity in the plate thickness direction.
Accordingly, an object of the present invention is to provide a heat dissipation plate having a low thermal expansion coefficient and a high thermal conductivity, which has a clad structure of a Mo—Cu composite material and a Cu material.
Another object of the present invention is to provide a manufacturing method capable of stably and inexpensively manufacturing a heat sink having such excellent heat characteristics.

  In contrast to the clad material having a Cu / (Cu—Mo) / Cu structure described in Patent Document 2, the present inventors have made the Mo—Cu composite material and the Cu material multi-layered, so that the Mo -Conclusion of five or more layers under the idea of increasing the restraint of the Cu layer by the Cu composite layer and reducing the coefficient of thermal expansion, but when the thermal conductivity is particularly improved Did not think. However, as a result of investigation, contrary to such initial expectation, the number of layers of the Mo—Cu composite material and the Cu material is set to 5 or more, that is, Cu / (Cu—Mo) / Cu / ( By using a Cu-Mo) / Cu structure (five-layer clad structure) or Cu / (Cu-Mo) / Cu / (Cu-Mo) / Cu / (Cu-Mo) / Cu structure (seven-layer clad structure) Compared with Cu / (Cu—Mo) / Cu structure (three-layer clad structure) having the same plate thickness and density, not only the in-plane thermal expansion coefficient is lowered, but also the thermal conductivity in the plate thickness direction is considerable. I found it to be higher. It has also been found that the thermal conductivity is particularly improved by making the thickness of the outermost Cu layer smaller than the thickness of the intermediate Cu layer. Further, when manufacturing the above clad material, the material is cold-rolled at a high pressure reduction rate (total reduction rate) or warm-rolled at a temperature of about 250 ° C. or less at which the surface is not significantly oxidized, It has been found that the expansion rate is more effectively lowered.

The present invention has been made on the basis of the above-described findings and has the following gist.
[1] In the plate thickness direction, Cu layers and Cu-Mo composite layers are alternately stacked to form three or more Cu layers and two or more Cu-Mo composite layers. The outer layer is a heat sink made of a Cu layer,
The Cu—Mo composite layer has a plate thickness cross-sectional structure in which a flat Mo phase is dispersed in a Cu matrix.
[2] In the heat sink of [1], the thickness t ₁ of the outermost Cu layer (1a) on both sides and the thickness t ₂ of the Cu layer (1b) of the intermediate layer satisfy t ₁ ≦ t ₂ . A heat sink characterized by that.

[3] The heat radiating plate according to the above [2], wherein the thickness t ₁ and the plate thickness T of the outermost Cu layers (1a) on both sides satisfy t ₁ /T≦0.2. .
[4] In the heat radiating plate of the [2] or [3], the thickness _{t 2} is _t 1 of the Cu layer of the outermost layer of double-sided Cu layer of (1a) thickness _{t 1} and the intermediate layer (1b) <t ₂ is satisfied.
[5] In the heat sink of [4] above, the total number of Cu layers and Cu—Mo composite layers is 9 or more, and the plurality of intermediate Cu layers (1b) have a thickness of radiator plate, wherein a Cu layer (1b) as the thickness t ₂ is thicker closer to the center.
[6] In the heat dissipation plate of any one of [1] to [5], the Cu—Mo composite layer includes a plurality of unit Cu—Mo composite layers via a joining Cu layer having a thickness of 75 μm or less. A heat sink having a laminated structure.

[7] The radiator plate according to any one of [1] to [6], wherein the Cu-Mo composite layer has a Cu content of 10 to 50% by mass.
[8] The radiator plate according to any one of [1] to [6], wherein the Cu-Mo composite layer has a Cu content of 20 to 30% by mass.
[9] In the heat dissipation plate of any one of [1] to [8], the thermal conductivity in the thickness direction is 200 W / m · K or more, and the average in-plane thermal expansion coefficient from 50 ° C. to 800 ° C. is 8 A heat radiating plate characterized by being 0.0 ppm / K or less.
[10] The radiator plate according to any one of the above [1] to [9], wherein a plating film is formed on one side or both sides of a radiator plate body composed of a laminated Cu layer and a Cu-Mo composite layer. A heat sink.

[11] A method for manufacturing a heat radiating plate according to any one of [1] to [10], wherein a Cu—Mo composite material (a) and a Cu material having a plate thickness cross-sectional structure in which a Mo phase is dispersed in a Cu matrix After laminating (b) and diffusion-bonding the laminated body, cold rolling (x) is performed to thereby form a Cu-Mo composite layer made of Cu-Mo composite (a) and Cu made of Cu material (b). A method of manufacturing a heat sink, characterized in that a heat sink having laminated layers is obtained.
[12] In the manufacturing method of [11], the Cu—Mo composite material (a) is formed by pressing a mixed powder of Mo powder and Cu powder into a green compact, and reducing the green compact. A heat-radiating plate manufacturing method characterized by being obtained through a step of sintering in a neutral atmosphere or vacuum to form a sintered body.
[13] In the manufacturing method of [11], the Cu—Mo composite material (a) is formed by pressing a mixed powder of Mo powder and Cu powder into a green compact, and reducing the green compact. A heat-radiating plate manufacturing method characterized by being obtained through a step of sintering in a neutral atmosphere or vacuum to form a sintered body and a step of densifying the sintered body.

[14] In the manufacturing method of [11], the Cu—Mo composite material (a) is a step of pressing a Mo powder or a mixed powder of Mo powder and Cu powder into a green compact; It was obtained through a step of sintering the body in a reducing atmosphere or vacuum to obtain a sintered body, and a step of impregnating the sintered body with Cu melted in a non-oxidizing atmosphere or vacuum. A manufacturing method of a heat sink characterized by being.
[15] The method for manufacturing a heat sink according to any one of [11] to [14], wherein the rolling reduction of cold rolling (x) is 70 to 99%.
[16] The method for manufacturing a heat sink according to [15], wherein the rolling reduction of cold rolling (x) is 90 to 96%.
[17] The method for manufacturing a heat sink according to any one of [11] to [16], wherein the cold rolling (x) is performed by cross rolling.

[18] In the manufacturing method of [11], the Cu—Mo composite material (a) is formed by pressing a mixed powder of Mo powder and Cu powder into a green compact, and reducing the green compact. A step of sintering in a neutral atmosphere or vacuum to obtain a sintered body, a step of densifying the sintered body, and a step of rolling (y) the densified Cu-Mo composite material A method of manufacturing a heat sink, characterized by being obtained through the process.
[19] In the manufacturing method of [11] above, the Cu—Mo composite material (a) is a step of pressing a Mo powder or a mixed powder of Mo powder and Cu powder into a green compact; A step of sintering the body in a reducing atmosphere or vacuum to form a sintered body, a step of impregnating the sintered body with Cu melted in a non-oxidizing atmosphere or vacuum, and impregnating the Cu A heat-radiating plate manufacturing method characterized by being obtained through a step of rolling (y) the Cu-Mo composite.
[20] In the production method of [18] or [19], the total rolling reduction of the Cu-Mo composite (a) obtained by combining the cold rolling (x) and the rolling (y) is 70 to 99%. The manufacturing method of the heat sink characterized by these.

[21] In the production method of [20], the total rolling reduction of the Cu—Mo composite (a) obtained by combining the cold rolling (x) and the rolling (y) is 90 to 96%. Manufacturing method of heat sink.
[22] The method for manufacturing a heat sink according to any one of [18] to [21], wherein the rolling (y) is performed by cross rolling.
[23] In the manufacturing method according to any one of [18] to [22] above, when the Cu—Mo composite material (a) is unidirectionally rolled by rolling (y), in cold rolling (x), Cu— A method of manufacturing a heat sink, comprising rolling a Mo composite material in a direction orthogonal to a rolling direction of rolling (y).
[24] In the production method according to any one of [11] to [23], the Cu—Mo composite material (a) is a laminate of a plurality of unit Cu—Mo composite materials (a _u ). The manufacturing method of a heat sink.
[25] In the manufacturing method according to any one of [11] to [23], the Cu—Mo composite material (a) includes a plurality of unit Cu—Mo composite materials (a _u ) via a Cu thin plate for bonding. A method of manufacturing a heat sink, characterized by being laminated.

[26] In the manufacturing method according to any one of [11] to [25], the Cu material (b) is a laminate of a plurality of unit Cu materials (b _u ). Method.
[27] The method for manufacturing a heat sink according to any one of [11] to [26], wherein the Cu-Mo composite material (a) has a Cu content of 10 to 50% by mass.
[28] The method for manufacturing a heat sink according to any one of [11] to [26], wherein the Cu-Mo composite material (a) has a Cu content of 20 to 30% by mass.
[29] In the manufacturing method of [27], the Cu-Mo composite material (a) has a Cu content of less than 20 mass% and is a combination of cold rolling (x) and rolling (y). A manufacturing method (including a manufacturing method in which the rolling (y) of the Cu—Mo composite material (a) is not performed) in which the total rolling reduction of (a) is 70% or more,
The manufacturing method of a heat sink characterized by performing the following (1) or / and (2) warm rolling.
(1) Perform warm rolling instead of cold rolling (x).
(2) Rolling (y) is performed by warm rolling.

[30] In the manufacturing method according to [28], the total rolling reduction of the Cu—Mo composite (a) obtained by combining cold rolling (x) and rolling (y) is 96% or more (provided that Cu -Including a manufacturing method in which the rolling (y) of the Mo composite (a) is not performed.
The manufacturing method of a heat sink characterized by performing the following (1) or / and (2) warm rolling.
(1) Perform warm rolling instead of cold rolling (x).
(2) Rolling (y) is performed by warm rolling.
[31] The method according to any one of [11] to [30], wherein a plating film is formed on one side or both sides of a heat sink main body composed of a laminated Cu—Mo composite layer and a Cu layer. The manufacturing method of the heat sink.
[32] A semiconductor package comprising the radiator plate of any one of [1] to [10].
[33] A semiconductor module comprising the semiconductor package of [32].

  The heat sink of the present invention has excellent thermal characteristics such as low thermal expansion coefficient and high thermal conductivity. Moreover, according to the manufacturing method of this invention, the heat sink which has such the outstanding thermal characteristic can be manufactured stably and at low cost.

Explanatory drawing which shows typically the plate | board thickness cross section of the heat sink of this invention which has a 5 layer clad structure (FIG. 1 (A)) and a 7 layer clad structure (FIG. 1 (B)). The graph which shows the thermal characteristic (The thermal conductivity of a plate | board thickness direction, the average thermal expansion coefficient in the plate surface from 50 degreeC to 800 degreeC) of the heat sink of an Example. The graph which shows the heat characteristic (heat conductivity of a plate | board thickness direction, the average thermal expansion coefficient in the plate surface from 50 degreeC to 400 degreeC) of the heat sink of an Example. The graph which shows the thermal characteristic (The thermal conductivity of a plate | board thickness direction, the average thermal expansion coefficient in the plate surface from 50 degreeC to 800 degreeC) of the heat sink of an Example. The graph which shows the heat characteristic (heat conductivity of a plate | board thickness direction, the average thermal expansion coefficient in the plate surface from 50 degreeC to 400 degreeC) of the heat sink of an Example. The graph which shows the thermal characteristic (The thermal conductivity of a plate | board thickness direction, the average thermal expansion coefficient in the plate surface from 50 degreeC to 800 degreeC) of the heat sink of an Example. The graph which shows the heat characteristic (heat conductivity of a plate | board thickness direction, the average thermal expansion coefficient in the plate surface from 50 degreeC to 400 degreeC) of the heat sink of an Example. The heat sink] is a view showing a relationship between the thickness t ₁ and a ratio t _{1 /} T and the plate thickness direction of the thermal conductivity of the plate thickness T of the outermost layer of the Cu layer

  The heat sink of the present invention is composed of three or more Cu layers and two or more Cu-Mo composite layers by alternately stacking Cu layers and Cu-Mo composite layers in the thickness direction. The outermost layer on both sides is a heat sink made of a Cu layer, and the Cu-Mo composite layer has a plate thickness cross-sectional structure in which a flat Mo phase is dispersed in a Cu matrix. FIG. 1 schematically shows a thickness cross section of a heat sink of the present invention having a five-layer clad structure (FIG. 1A) and a seven-layer clad structure (FIG. 1B). In the figure, 1a is the outermost Cu layer on both sides, and 1b is the intermediate Cu layer.

  The Cu—Mo composite layer and the Cu layer of the heat dissipation plate of the present invention are formed by diffusion bonding the laminated Cu—Mo composite material and the Cu material, and have a diffusion bonding portion between both layers. However, since Cu of both members (Cu and Cu material of a Cu-Mo composite material) is diffusion bonded, a sound diffusion bonded portion can be obtained. For example, considering the case of clad Mo (Mo material) and Cu (Cu material), since Mo and Cu are not alloyed, the bonding of both members is not diffusion bonding but mechanical bonding. Then, oxide films and fine voids are likely to remain at the bonding interface, and cracks and the like are likely to occur starting from these. On the other hand, as in the present invention, Cu of both members (Cu and Cu material of the Cu-Mo composite material) is diffusion bonded so that no oxide film or fine voids remain at the bonding interface. Sound joints can be obtained.

The heat sink of the present invention having a clad structure of five or more layers as described above and the outermost layers on both sides being Cu layers (for example, a heat sink of Cu / (Cu—Mo) / Cu / (Cu—Mo) / Cu structure). Has a higher thermal conductivity than the heat dissipation plate having a Cu / (Cu—Mo) / Cu structure disclosed in Patent Document 2, which is considered to be due to the following reasons. That is, in the case of a clad structure in which Cu layers and Cu—Mo composite layers are alternately laminated and the outermost layers on both sides are made of Cu layers, the thermal conductivity is outer layer (Cu layer)> inner layer (Cu—Mo composite layer). Therefore, the heat flowing into the outer layer (Cu layer) is reflected / scattered at the interface between the outer layer and inner layer and the heat flow is disturbed, so the heat does not transfer well to the inner layer (Cu-Mo composite layer) side, and the outer layer / It is considered that a high heat transfer resistance is generated by the interface between the inner layers, and the thermal conductivity in the thickness direction is lowered accordingly. The decrease in thermal conductivity in the plate thickness direction due to such causes depends on the thickness of the outermost Cu layer, and the thinner the outermost Cu layer, the less heat is reflected and scattered at the interface with the inner layer. Therefore, the degree to which the thermal conductivity is reduced is reduced. Therefore, when the heat sink of the present invention having a clad structure of five or more layers is compared with the heat sink of the three-layer clad structure described in Patent Document 2, if the plate thickness and density are the same, the heat sink of the present invention. Since the thickness of the outermost Cu layer is thinner in the plate, it is considered that the thermal conductivity in the plate thickness direction is higher than that of the heat sink with a three-layer clad structure. Further, in the case of a clad structure of five layers or more, the outermost Cu layer can be made thinner by increasing the thickness of the intermediate Cu layer. By making it smaller than the thickness of the Cu layer, the thermal conductivity in the plate thickness direction can be made higher.
In addition, the heat dissipation plate of the present invention has the same thickness and density because the number of layers of the Mo—Cu composite material and the Cu material is increased so that the Cu layer is restricted by the Mo—Cu composite layer. If so, the coefficient of thermal expansion is lower than that of the heat sink with a three-layer clad structure.

  The number of stacked layers in the clad structure (the total number of Cu layers and Cu—Mo composite layers) is not particularly limited, and the larger the number of stacked layers, the lower the coefficient of thermal expansion, and the higher the hardness and the lower the ductility. The thinner the composite layer, the better the press workability, which is advantageous for press work. When heat enters the heat sink, when the outermost layer is a Cu layer, heat enters due to the high heat conduction of Cu, but as described above, the interface with the Cu-Mo composite layer having the following low thermal conductivity. Therefore, the amount of heat entering the Cu—Mo composite layer is limited. In addition, when heat is transferred from the Cu layer to the Cu-Mo composite layer on the lower layer side, heat reflection and scattering occur at the interface as well, but the amount of heat has already been limited, and the heat was limited. Since the amount of heat is transmitted, there is little decrease in thermal conductivity at the interface. Therefore, even if the number of stacked layers is 7 or more, if the ratio of the thickness of the outermost Cu layer is small (generally, the ratio of the thickness of the Cu layer of the outermost layer is small for 7 or more layers), the number of stacked layers increases. Although the thermal conductivity in the thickness direction tends to decrease slightly, it can contribute to the decrease in thermal expansion coefficient and press workability, the thickness ratio of each layer of the Cu-Mo composite layer is reduced, and the heat transfer resistance of the layer is also reduced. Considering what to do, it can be said that there is no particular problem. Therefore, there is no special limitation on the number of layers, and the number of layers may be determined according to the application and product thickness. For example, when Invention Example 1 (5 layers) and Invention Example 11 (7 layers), Invention Example 2 (5 layers) and Invention Example 12 (7 layers) of Examples described later are compared, respectively, 7 layers are more heat-resistant. As shown in FIG. 8, the conductivity is higher in Invention Example 11 than in Invention Example 1, and in Invention Example 12 than in Invention Example 2, respectively. This is probably because the thickness ratio is small.

  Although there is no restriction | limiting in particular in Cu content of a Cu-Mo composite layer, Generally 10-50 mass% is suitable. The higher the Cu content, the better the cold rolling property when cold rolling is performed at a high pressure rate, and the effect of reducing the thermal expansion coefficient by cold rolling at a high pressure rate is likely to be obtained. On the other hand, in terms of enhancing the effect of restraining the thermal expansion of the Cu layer of the intermediate layer (the effect of physically restraining the Cu layer of the intermediate layer from both sides), not only the rolling reduction ratio but also the Mo content is large. However, the thermal conductivity is in a trade-off relationship, and if the Mo content is too large, cold rolling becomes difficult. For this reason, about 10-50 mass% of Cu content of a Cu-Mo composite layer is preferable. Further, from the viewpoint of the thermal characteristics of the heat sink, the Cu content of the Cu—Mo composite layer is preferably 30% by mass or less, while the Cu content of the Cu—Mo composite layer (Cu—Mo composite material) is low. If it is less than 20% by mass, there is a possibility of causing a problem in cold rollability. From the viewpoint of the heat characteristics of the heat sink and the cold rollability, the Cu content of the Cu—Mo composite layer is 20 to 30% by mass. More preferably, it is about.

  The Cu-Mo composite layer may have a structure composed entirely of an integral Cu-Mo composite, but a structure in which a plurality of unit Cu-Mo composite layers are stacked via a very thin bonding Cu layer. It is good. If the thickness of the bonding Cu layer is about 75 μm or less, it hardly affects the thermal characteristics of the heat sink. Therefore, the thickness is preferably 75 μm or less, and more preferably 25 μm or less. . In addition, this Cu layer for joining comprises a part of Cu-Mo composite layer, therefore, it differs from the Cu layer laminated | stacked alternately with a Cu-Mo composite layer in the heat sink of this invention. This Cu layer is not included.

  As will be described later, the heat sink of the present invention is manufactured by alternately laminating the Cu-Mo composite material (a) and the Cu material (b), diffusion bonding the laminated body, and then rolling. The Cu—Mo composite material (a) used in this production may not be a single plate material but may be composed of a plurality of laminated thin Cu—Mo composite materials (unit Cu—Mo composite material). This is because the Cu—Mo composite material may become thin when the rolling reduction ratio is increased. When it is composed of a plurality of thin unit Cu-Mo composites laminated with a Cu-Mo composite (a), especially when the Cu content of the Cu-Mo composite is relatively low, the unit Cu-Mo composite In order to enhance the bondability between the unit Cu-Mo composites, a plurality of unit Cu-Mo composites are laminated via Cu thin plates (including Cu foil) (that is, a thin Cu plate is formed between each unit Cu-Mo composites). It is preferable to perform diffusion bonding via the Cu thin plate. The Cu layer for joining in the Cu-Mo composite layer of the heat sink described above is obtained by further thinning the Cu thin plate by rolling. The bonding Cu layer constituting the Cu-Mo composite layer is a very thin intermediate layer, and therefore has a heat transfer resistance that is negligibly small and hardly affects the thermal characteristics of the heat sink. That is, the heat characteristics of the heat radiating plate having the bonding Cu layer in the Cu-Mo composite layer and the heat radiating plate not having the bonding Cu layer are almost the same.

  FIG. 2 and FIG. 3 show the thermal characteristics of some of the heat radiating plates of the examples described later. FIG. 2 shows the thermal conductivity in the plate thickness direction (thermal conductivity at room temperature). ) And the average thermal expansion coefficient in the plate surface from 50 ° C. to 800 ° C., FIG. 3 shows the thermal conductivity in the thickness direction (thermal conductivity at room temperature) and the average thermal expansion in the plate surface from 50 ° C. to 400 ° C. Each rate is shown. Here, the coefficient of thermal expansion in the plate surface is measured by a push rod type displacement detection method. For example, “the average coefficient of thermal expansion in the plate surface from 50 ° C. to 400 ° C.” is 50 ° C. and 400 ° C. The difference in elongation was obtained, and the value was divided by a temperature difference of 350 ° C. (= 400 ° C.-50 ° C.). Similarly, the in-plane average thermal expansion coefficient from 50 ° C. to 800 ° C. was determined. Further, the thermal conductivity in the plate thickness direction (thermal conductivity at room temperature) was measured by a flash method. The method for measuring and calculating the thermal characteristics is the same for the thermal characteristics shown in FIGS.

  2 and 3, a heat dissipation plate made of a single Cu—Mo composite material (Comparative Examples 7 to 10 and 13) and a heat dissipation made of a three-layer clad material of Cu / (Cu—Mo) / Cu structure of Patent Document 2. Thermal characteristics of the plates (Comparative Examples 1 and 2) and the heat sinks (Inventive Examples 1, 2, 11, and 12) made of the five-layer and seven-layer clad materials of the present invention are shown. In the figure, what is surrounded by a circle and connected by an arrow is a heat sink having substantially the same density. According to this, when comparing the thermal characteristics of heat sinks having substantially the same density, the heat sink of Cu / (Cu—Mo) / Cu structure in Patent Document 2 is a heat sink of a single Cu—Mo composite material. In comparison, the thermal conductivity in the plate thickness direction is slightly lower, but the in-plane thermal expansion coefficient is greatly reduced. And with respect to the thermal characteristics of the heat sink of this Cu / (Cu—Mo) / Cu structure, the heat sink of the present invention has a lower thermal expansion coefficient in the plate surface and has a thermal conductivity in the plate thickness direction. It is high.

  4 and 5 are graphs of FIG. 2 and FIG. 3 in which a comparative example of a Cu—Mo composite material having a different Cu content is added, and FIG. 4 shows the thermal conductivity in the plate thickness direction (at room temperature). Figure 5 shows the thermal conductivity in the plate thickness direction (thermal conductivity at room temperature) and the plate surface from 50 ° C to 400 ° C. The average coefficient of thermal expansion is shown respectively. The broken line in the figure shows that the Cu—Mo composite material alone has a tendency that the lower the Cu content (the higher the Mo content), the lower the thermal conductivity in the plate thickness direction and the lower the in-plane thermal expansion coefficient. ing. And as shown by the arrow in a figure, with respect to the tendency of the thermal characteristic of such Cu-Mo composite material single-piece | unit, the heat sink of the Cu / (Cu-Mo) / Cu structure of patent document 2 (Comparative example 1). The thermal characteristics of 2) are shifted to high thermal conductivity (thermal conductivity in the plate thickness direction) and low thermal expansion coefficient (in-plane thermal expansion coefficient), but the thermal characteristics of the heat sink of the present invention are: Furthermore, it has shifted to the high thermal conductivity (thermal conductivity in the plate thickness direction) and low thermal expansion rate (in-plane thermal expansion rate) side.

  FIGS. 6 and 7 are graphs of FIGS. 4 and 5 in which other examples of the invention in which the thickness of the outermost Cu layer and the Cu content of the Cu-Mo composite layer are different are added. 6 shows the thermal conductivity in the thickness direction (thermal conductivity at room temperature) and the average thermal expansion coefficient in the plate surface from 50 ° C. to 800 ° C., and FIG. 7 shows the thermal conductivity in the thickness direction (at room temperature). ) And the average in-plane thermal expansion coefficient from 50 ° C. to 400 ° C., respectively. According to this, the heat sink of the present invention is a single Cu-Mo composite material having the same plate thickness and density regardless of the difference in the thickness of the outermost Cu layer or the Cu content of the Cu-Mo composite layer. Compared with the thermal characteristics of the heat sink (Comparative Examples 1 and 2) of Cu / (Cu-Mo) / Cu structure of Patent Document 2 and high thermal conductivity (thermal conductivity in the plate thickness direction) and low thermal expansion coefficient (plate In-plane thermal expansion coefficient).

Based on the principle described above, the heat dissipation plate of the present invention has a higher thermal conductivity in the plate thickness direction as the thickness of the outermost Cu layer is smaller. From this viewpoint, it is preferable that the thickness t ₁ and the plate thickness T of the outermost Cu layer 1a on both sides satisfy t ₁ /T≦0.2.
8, the heat radiating plate of Example organize the relationship between the ratio t _{1 /} T and the plate thickness direction of the heat conductivity of the thickness t ₁ and the thickness T of the outermost layer of the Cu layer 1a (see FIG. 1) In the figure, what is connected by a solid line is a heat dissipation plate having substantially the same density. According to this, it can be seen that the smaller the ratio of the thickness t ₁ of the outermost Cu layer, the higher the thermal conductivity in the plate thickness direction, and t ₁ /T≦0.2 is preferable.

From the same viewpoint as above, it is preferable that the thickness t ₂ of the Cu layer 1b having a thickness of t ₁ and the intermediate layer of Cu layer 1a is the outermost layer of the double-sided satisfies t ₁ ≦ t _2. As described above, the thickness t ₁ of the outermost Cu layer 1a on both sides is preferably as thin as possible because the thermal conductivity can be increased. When t ₁ > t ₂ , the thickness of the outermost Cu layer of the three-layer clad structure is approached, and the effect of improving the thermal conductivity in the present invention is reduced.
Still Preferred conditions, it is preferable that the thickness t ₂ of the Cu layer 1b having a thickness of t ₁ and the intermediate layer of Cu layer 1a is the outermost layer of the double-sided satisfies t _{1 <t 2.} Further, in the case of a heat sink having a total number of Cu layers and Cu-Mo composite layers (number of stacked layers) of 9 or more (a heat sink having three or more Cu layers 1b as an intermediate layer), it is preferred that greater thickness _{t 2} closer Cu layer 1b. These reasons are considered as follows.

In the material of thickness L, the heat flow when heat flows in the thickness direction is expressed by the following equation.
Heat flow q (W) = CA (θ ₁ −θ ₂ ) [θ; temperature, C; thermal conductance from point 1 to point 2, A: cross-sectional area of material through which heat flows]
C = λ / L [λ: thermal conductivity (W / m · K), L: material thickness (m)]
The thermal conductance is the amount of heat that flows per fixed area and fixed time when the temperature difference between both surfaces of the material is 1 ° C., and expresses the ease of heat transfer. Here, the heat transfer resistance R is the reciprocal of C.
The heat transfer resistance R _CLAD of the entire five-layer clad material is given by the following equation.
R _CLAD = (L ₁ / λ _Cu ) + (L ₂ / λ _Cu-Mo ) + (L ₃ / λ _Cu ) + (L ₄ / λ _Cu-Mo ) + (L ₅ / λ _Cu ) + R ₁₂ + R ₂₃ + R ₃₄ + R ₄₅
= R ₁ + R ₂ + R ₃ + R ₄ + R ₅ + R ₁₂ + R ₂₃ + R ₃₄ + R ₄₅
Here, L _{1 to} L ₅ are the thicknesses of the first to fifth layers, λ _Cu is the thermal conductivity of the Cu layer, λ _Cu-Mo is the thermal conductivity of the Cu-Mo composite layer, R ₁ , R ₂ , R ₃ , R ₄ , R ₅ are the heat transfer resistance of each layer, R ₁₂ , R ₂₃ , R ₃₄ , R ₄₅ are the heat transfer resistance at the interface of each layer, ₁₂ , ₂₃ , ₂₃ , ₄₅ are from the top Each layer of is shown.
Here, R ₁₂ , R ₂₃ , R ₃₄ , and R ₄₅ are not the material but the degree of heat flow turbulence due to heat reflection and heat scattering at the interface, that is, the load (resistance).

Here, considering the heat flow from the first layer to the fifth layer of the five-layer clad material, the lower R ₁ Cu layer of the first layer (outermost layer) to the higher R ₂ Cu—Mo composite of the second layer. The heat flow is reduced when entering the body layer, and in the first Cu layer, not only the original heat transfer resistance R ₁ but also the heat transfer resistance of R ₁₂ at the interface is added. When the thickness L _{1 of the} first layer (outermost layer) Cu layer is small, R ₁ also becomes small, and the amount added to the Cu layer by heat reflection or heat scattering decreases, and R ₁₂ also decreases accordingly. . As the thickness approaches zero, R ₁ and R ₁₂ also approach zero. The interface between the second Cu-Mo composite layer and the third Cu layer is that heat enters the lower layer from the high heat transfer resistance layer, and the Cu phase in the Cu-Mo composite layer Since the Cu layer is completely diffusion-bonded and integrated, and the Cu is continuous, R ₂₃ may be considered to be almost zero. R ₄₅ may be considered to be zero as well. A third layer of Cu layer heat transfer resistance R ₃ of is added the heat transfer resistance R ₃₄ minutes at the interface between the fourth layer Cu-Mo composite layer. However, since the heat flow after being squeezed through the second Cu-Mo composite layer flows to the third Cu layer, the thickness L _{1 of the first} Cu layer is equal to the third Cu layer. even the same as the thickness L ₃ of the layer, R ₃₄ is smaller than R _12. More specifically (however, the heat flow in the description is a tentative value), assuming that the Cu-Mo composite layer and the Cu layer of the five-layer cladding material have the same thickness, the first heat flow of 100 Enters the first Cu layer, the _first Cu layer has a heat transfer resistance of (L ₁ / λ _Cu ) + R ₁₂ , and the heat flow is limited to 80. Thereafter, the second Cu—Mo composite layer has a heat transfer resistance of (L ₂ / λ _Cu—Mo ) + R ₂₃ (R ₂₃ ≈0), the heat flow is reduced to 60, and the third Cu layer Enter the layer entrance. Since R ₃₄ when entering from the third Cu layer to the fourth Cu—Mo composite layer is not the heat transfer resistance from the heat flow 100 but the heat transfer resistance from the heat flow 60, R ₁₂ > R ₃₄ . From the above, the heat transfer resistance R _CLAD of the entire cladding material is reduced by making the outermost Cu layer thinner than the inner (intermediate) Cu layer.

For the reason described above, in the case of a heat radiating plate having nine or more layers (a heat radiating plate having three or more intermediate Cu layers 1b), the thickness of the Cu layer is the inside of the plate (the plate It is preferable to increase the thickness as it goes to (thickness center). Furthermore, the combination of the Cu layer and the Cu—Mo composite layer is thicker from the combination of the thin Cu layer on the outer layer side (heat inlet side) and the thin Cu—Mo composite layer toward the inside of the plate (plate thickness center). Since it is considered that reflection and scattering at the interface of the heat flow are reduced by the combination, not only the thickness of the Cu layer but also the thickness of the Cu-Mo composite layer is inside the plate (plate thickness center). It is preferable to increase the thickness as the time goes.
In the five-layer clad material (invention example) of Examples described later, the thickness t ₁ of the outermost Cu layer 1a on both sides and the thickness t _{2 of the} intermediate Cu layer 1b are t ₁ <t _2. However, the degree of t ₁ <t ₂ is that t ₁ / t ₂ ≦ 0.4 for invention examples 3 to _{10 and} 13 to 21, and t ₁ / t ₂ ≦ 0.4 for invention examples t ₁ to t _{1 for} invention examples 3 to 8, 13 to 17, / T ₂ ≦ 0.1, Invention Examples 3 to 6, 13 to 16, and 19 to 21 are t ₁ / t ₂ ≦ 0.06.

The thickness of each of the Cu-Mo composite layer and the Cu layer, the layer thickness ratio of the Cu-Mo composite layer and the Cu layer, the thickness of the heat sink, etc. are not particularly limited. In order to prevent warping or distortion during practical use, the structure is symmetrical in the thickness direction with the Cu layer at the center in the thickness direction as the center (structure in which the thickness of the Cu layer and the Cu-Mo composite layer is symmetrical). Is preferred. Further, the thickness of the heat sink is often around 1 mm, but is not particularly limited.
Although the lower limit is not particularly thickness t ₁ of the outermost Cu layers 1a, the manufacture of the thickness t ₁ is extremely small as the cladding material is difficult, also, the thickness of the Cu layer of the intermediate layer is larger Since the coefficient of thermal expansion becomes high, about 0.01 mm is the practical lower limit.

Regarding the layer thickness ratio between the Cu-Mo composite layer and the Cu layer, when the layer thickness ratio of the Cu layer is large with respect to the Cu-Mo composite layer, the thermal conductivity increases, but it depends on the Cu-Mo composite layer. Since the restraint of the intermediate Cu layer becomes weak, the coefficient of thermal expansion becomes high. On the other hand, when the layer thickness ratio of the Cu layer is small, the coefficient of thermal expansion becomes low, but the thermal conductivity becomes low. Therefore, the layer thickness ratio between the Cu-Mo composite layer and the Cu layer may be appropriately selected according to the thermal characteristics (thermal conductivity, thermal expansion coefficient) to be obtained. From the viewpoint of lowering the coefficient of thermal expansion at (), it is better not to make the Cu layer too thick with respect to the Cu-Mo composite layer.
Further, since the Cu content of the Cu-Mo composite layer or the layer thickness ratio of the Cu-Mo composite layer and the Cu layer is linked to the density of the heat sink, this density is about 9.25 to 9.55 g / cm ^3. It is preferable that it is about 9.30-9.45 g / cm < ³ >.

  The heat sink of the present invention is manufactured by performing diffusion bonding after a Cu-Mo composite material and a Cu material manufactured in advance are subjected to diffusion bonding, and rolling may also be performed in the manufacturing process of the Cu-Mo composite material. Therefore, the whole is a rolled structure, and the Mo phase dispersed in the Cu matrix of the Cu-Mo composite layer has a flatly stretched form, and usually the Mo phase in the plate thickness cross-sectional structure. The aspect ratio (aspect ratio in the rolling direction) exceeds 2. Here, the aspect ratio is the major axis / minor axis (length ratio) of the Mo phase in the sheet thickness sectional structure in the rolling direction. For example, the sheet thickness sectional structure in the rolling direction (ion milled sheet) A thick cross-sectional structure) is observed with an SEM or the like, and the major axis / minor axis of each Mo phase included in an arbitrary visual field can be obtained and defined by an average value thereof.

  In addition, the Mo phase dispersed in the Cu matrix of the Cu-Mo composite layer is a flat stretched form depending on the Mo content of the Cu-Mo composite layer, the form of rolling (unidirectional rolling, cross rolling), and the like. For example, when the Mo content of the Cu-Mo composite layer is relatively low, the flatly stretched Mo phase has a form that is close to an individual island, but has a large Mo content. Then, the flatly stretched Mo phases are connected to each other, and the absence of stripes in which such a Mo phase and a Cu matrix are mixed has a marbled form (rolling structure). Therefore, in the latter case, the aspect ratio clearly exceeds 2, but there are cases where it cannot be specifically quantified.

Since the semiconductor package to which the heat sink of the present invention is mainly applied repeats the operation and the rest of the semiconductor, it is from room temperature (may be about −50 ° C. in a cold region) to about 200 ° C. during semiconductor operation. Repeat until the temperature rises. For this reason, the heat sink needs to have a low coefficient of thermal expansion in order to cope with thermal fatigue. In addition, it is important that the thermal expansion coefficient is as low as about 800 ° C. for use in brazing and about 400 ° C. for use in soldering. On the other hand, the heat sink needs to have a high thermal conductivity, in particular, a high thermal conductivity in the thickness direction in order to obtain high heat dissipation.
Although the heat sink of the present invention has excellent thermal characteristics that have both high thermal conductivity and low thermal expansion coefficient, specifically, the thermal conductivity in the plate thickness direction (thermal conductivity at room temperature) is It is preferably 200 W / m · K or more, and more preferably 250 W / m · K or more. Moreover, it is preferable that the plate | board surface average thermal expansion coefficient from 50 degreeC to 800 degreeC is 10.0 ppm / K or less, and it is more preferable that it is 8.0 ppm / K or less.

The heat sink of the present invention may be plated with Ni plating or the like on the surface thereof for the purpose of anticorrosion or for joining with other members (brazing joining or soldering joining). In this case, the plating film is formed with a thickness that does not significantly affect the thermal characteristics of the heat sink. There is no special restriction | limiting in the kind of plating, For example, Ni plating, Au plating, Ag plating etc. can be applied, and plating chosen from these can be given individually or in combination of 2 or more layers. A plating film may be provided only on one side (one surface of both Cu layers which are outermost layers) of a heat sink, and may be provided on both surfaces of a heat sink.
Depending on the material of the heat sink, Cu plating may be applied as the base for improving the plating property when plating such as Ni plating on the surface of the heat sink. Since the outer layer is a Cu layer, it is not necessary to perform such base plating.

Next, the manufacturing method of the heat sink of the present invention described above will be described.
In one embodiment of the heat sink manufacturing method of the present invention, a Cu-Mo composite material (a) and a Cu material (b) having a plate thickness cross-sectional structure in which a Mo phase is dispersed in a Cu matrix are stacked, and this stacked body. Then, cold rolling (x) is performed to obtain a heat sink in which a Cu—Mo composite layer made of Cu—Mo composite (a) and a Cu layer made of Cu material (b) are laminated. Here, the Cu-Mo composite material (a) is manufactured in advance, but this Cu-Mo composite material (a) is a method in which rolling is not performed (for example, the methods (i) to (iii) described later). ) Or a method of rolling (y) (for example, methods (iv) and (v) described later).
Moreover, in other embodiment of the manufacturing method of the heat sink of this invention, when Cu content of Cu-Mo composite material (a) is comparatively low, in order to prevent the crack by cold rolling, etc. Perform warm rolling of 1) and / or (2). This manufacturing method will be described in detail later.
(1) Perform warm rolling instead of cold rolling (x).
(2) Rolling (y) is performed by warm rolling.

The thicknesses of the Cu—Mo composite material (a) and the Cu material (b) are appropriately selected according to the thicknesses of the Cu—Mo composite layer and the Cu layer of the heat sink to be manufactured.
In addition, although Cu-Mo composite material (a) and Cu material (b) may each be comprised with a single board | plate material, several thin Cu-Mo composite material which laminated | stacked Cu-Mo composite material (a) was used. (Unit Cu-Mo composite material (a _u )) may be used, or a plurality of thin Cu materials (unit Cu material (b _u )) in which the Cu material (b) is stacked may be used. This is because the Cu—Mo composite material and the Cu material may be thin when the rolling reduction ratio is increased. Therefore, in that case, (1) the Cu—Mo composite material (a) composed of a plurality of unit Cu—Mo composite materials (a _u ) and the single Cu material (b) are laminated, A Cu-Mo composite material (a) and a Cu material (b) made of a plurality of unit Cu materials (b _u ) are laminated, and (3) Cu— made of a plurality of unit Cu-Mo composite materials (a _u ). A laminate is formed by either laminating the Mo composite (a) and the Cu material (b) made of a plurality of unit Cu materials (b _u ), and the laminate is diffusion-bonded.

Here, when configuring in Cu-Mo composite material as described above plurality of unit Cu-Mo composite material obtained by laminating _(a) (a _u), the unit Cu-Mo composite material _{(a u)} if and bondability In order to increase the thickness, a plurality of unit Cu—Mo composite materials (a _u ) are laminated via Cu thin plates (including the case of Cu foil) (that is, between each unit Cu—Mo composite material (a _u )). A thin Cu plate is interposed), and diffusion bonding is preferably performed through this Cu thin plate. The Cu layer for joining in the Cu-Mo composite layer of the heat sink described above is obtained by further thinning the Cu thin plate by rolling. Therefore, this Cu thin plate is preferably of such a thickness that the thickness of the bonding Cu layer in the Cu—Mo composite layer of the heat radiating plate is 75 μm or less (more preferably 25 μm or less).
Although there is no restriction | limiting in particular in the method of performing the diffusion bonding of a laminated body, The diffusion bonding by discharge plasma sintering (SPS) and hot press is preferable.
The following can be used for the Cu-Mo composite material (a). Further, as the Cu material (b), a pure Cu plate (including a pure Cu foil) is usually used.

  As described above, it is qualitatively known that the Cu-Mo composite material has a reduced coefficient of thermal expansion by rolling, and the Cu-Mo composite material is also rolled in the prior art. Since the Mo particles are hard and the primary particles are small, it is considered that the Mo particles are difficult to be deformed by rolling. For this reason, the rolling of the Cu—Mo composite material is performed exclusively by warm rolling at about 200 to 400 ° C. Yes. Moreover, although the method of implementing cold rolling by secondary rolling about 65 mass% Mo-35mass% Cu composite material is also proposed, warm rolling is performed by primary rolling.

  However, in contrast to the conventional recognition as described above and the manufacturing method based thereon, when rolling of Cu—Mo composite material (particularly Cu—Mo composite material with a Cu content not so low) is performed by warm rolling, Mo Since the deformation of the particles does not proceed properly, the effect of reducing the coefficient of thermal expansion is poor. On the other hand, when rolling is performed by cold rolling, the deformation of the Mo particles proceeds appropriately and the coefficient of thermal expansion is effectively improved. It turns out that it falls. In addition, when the Cu content of the Cu-Mo composite material is relatively low (for example, less than 20% by mass), when cold rolling is performed, there is a risk of causing cracks or the like depending on the rolling reduction. In some cases, it is better to make the rolling warm. However, when the Cu content is 20% by mass or more and the rolling reduction is not extremely high, the Cu-Mo composite material is rolled only by cold rolling. However, it was found that a good rolled sheet can be obtained without causing large cracks. Thus, it is thought that the reason why the plastic deformation mode of the Mo particles is greatly different between the warm rolling and the cold rolling is as follows.

  When rolling a Cu-Mo composite, due to the difference in yield stress between Mo and Cu, at the initial stage of rolling, the Mo particles are not deformed, but the relative position in the Cu matrix is changed, and the rolling progresses and the plate thickness is increased. When the Mo particles come into contact with each other in the direction, the deformation tends to occur. In cold rolling, work hardening of Cu occurs, so that Mo particles can be deformed by the Cu phase as the rolling progresses, and therefore the deformation of Mo particles proceeds appropriately. Conceivable. On the other hand, since the relative position change of the Mo particles in the Cu matrix becomes easier in the warm rolling and the work hardening of Cu is less likely to occur, the action of deforming the Mo particles by the Cu phase is less than that in the cold rolling. Therefore, it is considered that the deformation of the Mo particles does not proceed properly.

The difference in the plastic deformation mode of Mo particles in such a Cu-Mo composite becomes more prominent as the Cu content of the Cu-Mo composite increases. In addition to the fact that the work hardening of Cu cannot be used in warm rolling, the amount of Cu tends to change the relative position in the Cu matrix, while in the cold rolling, the Cu is not easily used. This is considered to be due to the greater influence of Cu work hardening. Cu has a high thermal conductivity but also a high coefficient of thermal expansion. Therefore, if the Cu content of the Cu-Mo composite is increased, a problem is likely to occur in terms of the coefficient of thermal expansion, but cold rolling is performed at a predetermined high pressure ratio. Thus, it was found that the coefficient of thermal expansion of the Cu—Mo composite material can be effectively reduced. In addition, as will be described later, even when warm rolling is incorporated in a part of rolling and cold rolling and warm rolling are used in combination, the effect of lowering the thermal expansion coefficient can be expected by the cold rolling.
Further, even when the Cu content of the Cu—Mo composite material is relatively small (for example, Cu content of 30% or less), the same effect as described above can be obtained although the degree thereof is relatively small. On the other hand, when the Cu content of the Cu—Mo composite material is relatively small, since the constraint by Mo is strengthened as described above, the effect of reducing the thermal expansion coefficient from this surface can be expected.

The Cu-Mo composite material (a) is manufactured in advance, but as the Cu-Mo composite material (a), for example, one obtained by any of the following methods (i) to (iii) Can be used.
(I) through a step of pressing a mixed powder of Mo powder and Cu powder to form a green compact, and a step of sintering the green compact in a reducing atmosphere or vacuum to form a sintered body Obtained Cu-Mo composite (a)
(Ii) a step of pressing a mixed powder of Mo powder and Cu powder into a green compact, a step of sintering the green compact in a reducing atmosphere or vacuum to form a sintered body, Cu-Mo composite material (a) obtained through the step of densifying the sintered body
(Iii) A step of pressure-molding Mo powder or a mixed powder of Mo powder and Cu powder to form a green compact, and sintering the green compact in a reducing atmosphere or vacuum to obtain a sintered body Cu-Mo composite material (a) obtained through a process and a process of impregnating the sintered body with Cu melted in a non-oxidizing atmosphere or in a vacuum

Since the Cu-Mo composite material (a) obtained by any one of the above methods (i) to (iii) is not cold-rolled, the cold rolling (x) of the clad material is a reduction. It is desirable to roll at a rate of 70 to 99%, more preferably 80 to 99%, particularly preferably 90 to 96%. This rolling reduction is also the rolling reduction of the Cu—Mo composite material (a). Thus, the effect of lowering the thermal expansion coefficient is obtained by cold rolling at a high pressure reduction rate, and if the reduction rate is excessively high, the thermal conductivity tends to decrease. %, Preferably 96%, it is possible to effectively reduce the thermal expansion coefficient while suppressing a decrease in thermal conductivity. Cold rolling (x) is performed in multiple passes.
Cold rolling (x) may be unidirectional rolling, but the in-plane anisotropy is reduced by reducing the difference in coefficient of thermal expansion between two directions (X-axis direction and Y-axis direction) orthogonal to each other in the plate surface. In order to reduce, cross rolling which performs rolling in two orthogonal directions may be performed. Here, rolling in two orthogonal directions may be performed at different reduction ratios, but when it is desired to obtain a rolled sheet having uniform thermal characteristics with no difference in thermal expansion coefficient between the X-axis direction and the Y-axis direction. It is preferable to roll at the same rolling reduction.

Moreover, as a Cu-Mo composite material (a), you may use what was obtained by the method of the following (iv) or (v).
(Iv) a step of pressing a mixed powder of Mo powder and Cu powder into a green compact, a step of sintering the green compact in a reducing atmosphere or vacuum to form a sintered body, Cu-Mo composite material (a) obtained through a step of densifying the sintered body and a step of rolling (y) the densified Cu-Mo composite material
(V) A step of pressure-molding Mo powder or a mixed powder of Mo powder and Cu powder to form a green compact, and sintering the green compact in a reducing atmosphere or vacuum to obtain a sintered body Obtained through a step, a step of impregnating the sintered body with Cu melted in a non-oxidizing atmosphere or in a vacuum, and a step of rolling (y) the Cu-Mo composite material impregnated with Cu. Cu-Mo composite (a)

  Rolling (y) can be performed by cold rolling. Even when the Cu content of the Cu-Mo composite material (a) is 30% by mass or less, the rolling (y) can be performed by cold rolling, but in some cases, it may be performed by warm rolling. The rolling (y) may be unidirectional rolling, but the in-plane anisotropy is reduced by reducing the difference in thermal expansion coefficient between two directions (X-axis direction and Y-axis direction) orthogonal to each other in the plate surface. In order to reduce, cross rolling which performs rolling in two orthogonal directions may be performed. Here, the rolling in the two orthogonal directions may be performed at different reduction ratios, but the Cu—Mo composite material (a) having uniform thermal characteristics with no difference in thermal expansion coefficient between the X-axis direction and the Y-axis direction. When it is desired to obtain the same, it is preferable to roll at the same rolling reduction.

  Since the Cu-Mo composite material (a) obtained by the above method (iv) or (v) is subjected to rolling (y), cold rolling (x) of the clad material is cold rolling. The total rolling reduction of the Cu-Mo composite (a) combining (x) and rolling (y) is 70 to 99%, more preferably 80 to 99%, and particularly preferably 90 to 96%. It is desirable to roll at. The reason is the same as above. For the same reason as the above-described cross rolling, when the Cu—Mo composite material (a) is unidirectionally rolled by rolling (y), the cold rolling (x) is performed by rolling the Cu—Mo composite material ( You may roll in the direction orthogonal to the rolling direction of y).

Moreover, in the manufacture of the heat sink of the present invention, when the Cu content of the Cu-Mo composite (a) is relatively low, it prevents cracks caused by cold rolling depending on the total rolling reduction of the material. Therefore, it is preferable to use a production method incorporating warm rolling (however, including a production method in which the rolling (y) of the Cu-Mo composite (a) is not performed). In this production method, for example, It is preferable to perform warm rolling under the following conditions.
That is, the total reduction ratio of the material (total reduction ratio combining the reduction ratio of the Cu-Mo composite alone and the reduction ratio of the Cu-Mo composite during rolling of the clad material) is 70% or more, and the Cu-Mo When the Cu content of the composite (a) is less than 20 mass%, it is preferable to perform the following (1) or / and (2) warm rolling, particularly when the Cu content is 15 mass% or less. The following (1) and (2) warm rolling is preferably performed. Even when the Cu content of the Cu-Mo composite (a) is 20 to 30 mass% and the total rolling reduction of the material is particularly high (for example, a total rolling reduction of 96% or more), the following (1) or / And it is preferable to perform the warm rolling of (2).
(1) In place of the cold rolling (x), warm rolling is performed.
(2) The rolling (y) is performed by warm rolling.

As described above, since the relative position change of the Mo particles in the Cu matrix becomes easier and the work hardening of Cu hardly occurs in the warm rolling, the Mo particles are deformed by the Cu phase as compared with the cold rolling. In the case of a Cu-Mo composite with a low Cu content, the distance between the Mo particles is not sufficient, and the rate of decrease in the coefficient of thermal expansion due to rolling tends to be lower than that in cold rolling. Since the relative position change between the Cu phase and the Mo particles is less likely to occur due to the shortened length of the Mo particles, the Mo particles are easily deformed. For this reason, even if warm rolling is performed under the above conditions, there is no great difference from the case of cold rolling. A heat sink having thermal properties is obtained.
The warm rolling is preferably performed at a temperature of about 200 to 300 ° C. When the temperature of the warm rolling exceeds 300 ° C., Mo is easily oxidized to form a surface oxide, which tends to cause problems such as peeling during rolling and adversely affecting product quality.
In addition, when performing the warm rolling of any one of the above (1) and (2), considering the rollability according to the Cu content or thickness of the Cu-Mo composite material (a), either one of them Is selected.

Next, the steps of the methods (i) to (v) for obtaining the Cu—Mo composite material (a) will be described.
In the following description, the step of pressing the Mo powder or a mixed powder of Mo powder and Cu powder to form a green compact is the step (A), and the green compact is sintered in a reducing atmosphere or vacuum. The step of forming a sintered body is step (B), the step of impregnating the sintered body with Cu melted in a non-oxidizing atmosphere or vacuum is the step (C1), and the step of densifying the sintered body Step (C2), and the step of rolling (y) the Cu-Mo composite material subjected to Cu infiltration or densification treatment is referred to as Step (D).

In the step (A), Mo powder or a mixed powder of Mo powder and Cu powder is pressure-formed into a green compact according to a conventional method. In the manufacturing method of the Cu-Mo composite (a) described above, Cu is infiltrated after the green compact is sintered (step (C1)) and Cu is infiltrated after the green compact is sintered. In some cases, the densification treatment is performed (step (C1)). In the latter case, an amount of Cu powder corresponding to the Cu content of the Cu-Mo composite (a) is blended.
The purity and particle size of the Mo powder and Cu powder are not particularly limited. Usually, the Mo powder having a purity of 99.95% by mass or more and an FSSS average particle size of about 1 to 8 μm is used. The Cu powder is usually pure Cu such as electrolytic copper powder or atomized copper powder and has an average particle diameter D50 of about 5 to 50 μm.

In step (A), Mo powder or a mixed powder of Mo powder and Cu powder is filled into a mold, and pressure is adjusted while adjusting the pressure according to the target value of the filling property of the mixed powder to be used and the molding density of the green compact. Molding to obtain a green compact.
In the step (B), the green compact obtained in the step (A) is sintered in a reducing atmosphere (such as a hydrogen atmosphere) or in a vacuum to obtain a sintered body. This sintering may be performed under normal conditions. In the case of a green compact mixed powder of Mo powder and Cu powder, the temperature is about 900 to 1050 ° C. (preferably 950 to 1000 ° C.) for about 30 to 1000 minutes. It is preferable to carry out under the condition of holding. In the case of a Mo powder green compact, it is preferably carried out at a temperature of about 1100 to 1400 ° C. (preferably 1200 to 1300 ° C.) for about 30 to 1000 minutes.

In the step (C1), the sintered body (porous body) obtained in the step (B) is impregnated with Cu melted in a non-oxidizing atmosphere or vacuum (Cu infiltration) to form a Cu-Mo composite material. (A) is obtained. When this step (C1) is performed, a desired Cu content is obtained by Cu infiltration.
Infiltration of Cu may be performed under normal conditions. For example, a Cu plate or Cu powder is disposed on the upper surface and / or lower surface of the sintered body, and held at a temperature of about 1083 to 1300 ° C. (preferably 1150 to 1250 ° C.) for 20 to 600 minutes. The non-oxidizing atmosphere is not particularly limited, but a hydrogen atmosphere is preferable. Further, from the viewpoint of improving workability after infiltration, infiltration in a vacuum is preferable.

Here, when the step (B) and the step (C1) are sequentially performed, a sintering temperature is first set in a state in which a Cu plate or Cu powder for Cu infiltration is arranged on the green compact obtained in the step (A). The step (B) may be carried out by heating to a temperature, and then the temperature may be raised to the Cu infiltration temperature to carry out the step (C1).
The Cu—Mo composite material (infiltrated) obtained in this step (C1) is subjected to surface grinding (for example, in order to remove excess pure Cu remaining on the surface prior to cold rolling in the next step. It is preferable to perform surface grinding using a milling machine or a grindstone.

Moreover, in the process (C2) performed instead of the process (C1), the sintered body obtained in the process (B) is densified to obtain the Cu—Mo composite material (a). In this case, after the sintering in the step (B), the temperature is further raised to perform a treatment for dissolving Cu (treatment for holding at about 1200 to 1300 ° C. for about 20 to 120 minutes), and then densification in the step (C2). Processing may be performed.
This densification treatment requires high temperature and pressure, and can be performed by methods such as hot pressing, spark plasma sintering (SPS), and hot rolling. By this densification treatment, the voids in the sintered body are reduced and densified to increase the relative density.
In the step (D), the Cu-Mo composite material obtained in the step (C1) or (C2) is rolled at a predetermined reduction rate for the purpose of reducing the thermal expansion coefficient of the Cu-Mo composite material (a). (Y) is applied.
In addition, before rolling the Cu-Mo composite material obtained at the process (C1) or (C2), you may perform a homogenization aging heat processing at the temperature of about 800-1000 degreeC as needed.

  The heat sink of the present invention can be made into a product by performing cold-rolling or warm-rolling, or by performing a softening aging heat treatment. Further, if necessary, the surface may be further subjected to plating such as Ni plating for the purpose of improving the corrosion resistance assumed to be used as a semiconductor pedestal and the performance against electric corrosion. In this case, the plating film is formed with a thickness that does not significantly affect the thermal characteristics of the heat sink. There is no special restriction | limiting in the kind of plating, For example, Ni plating, Au plating, Ag plating etc. can be applied, and plating chosen from these can be given individually or in combination of 2 or more layers. Plating may be performed only on one side (one surface of both Cu layers which are outermost layers) of a heat sink, or may be performed on both surfaces of a heat sink.

  The heat sink of the present invention can be suitably used for semiconductor packages such as ceramic packages and metal packages provided in various semiconductor modules, and high heat dissipation and durability can be obtained. In particular, although it has a high thermal conductivity, a low thermal expansion coefficient is maintained even after being exposed to a high temperature exceeding 800 ° C., so there is no problem in applications such as brazing joining where the joining temperature is increased to 750 ° C. or higher. Applicable.

(1) Manufacturing conditions of Cu-Mo composite material Mixed powder obtained by mixing Mo powder (FSSS average particle diameter: 6 μm) and pure Cu powder (average particle diameter D50: 5 μm) at a predetermined ratio into a mold (50 mm × 50 mm) The resulting green compact was pressed into a green compact having a thickness corresponding to the rolling reduction ratio in the subsequent cold rolling. The green compact was sintered (1000 ° C., 600 minutes) in a hydrogen atmosphere to obtain a sintered body. Next, a pure Cu plate is placed on the upper surface of the sintered body, heated to 1200 ° C. in a hydrogen atmosphere (holding time 180 minutes) to dissolve the pure Cu plate, and the sintered body is impregnated with the dissolved Cu. Thus, a Cu—Mo composite material having a predetermined Cu content was obtained. After removing Cu remaining on the surface using a milling machine, the Cu-Mo composite material was subjected to unidirectional rolling (y) (cold rolling) at a predetermined reduction rate to produce a Cu-Mo composite material. .

(2) Manufacturing conditions for each specimen (2.1) Example of the present invention A Cu-Mo composite material and a pure Cu plate having a predetermined plate thickness obtained as described above were prepared as Cu / (Cu-Mo) / Cu / ( A five-layer structure of Cu-Mo) / Cu or a seven-layer structure of Cu / (Cu-Mo) / Cu / (Cu-Mo) / Cu / (Cu-Mo) / Cu is laminated, and this laminate is subjected to discharge plasma. Using a sintering (SPS) apparatus (“DR.SINTER SPS-1050” manufactured by Sumitomo Coal Mining Co., Ltd.), diffusion bonding was performed under the conditions of 950 ° C., holding for 18 minutes, and pressure of 20 MPa. Subsequently, the Cu-Mo composite material is rolled (cold rolled) in the direction orthogonal to the rolling direction of the rolled (y) at the same rolling reduction as the rolled (y) (cold rolled), and the heat sink of the present invention example (Plate thickness 1 mm) was manufactured.
(2.2) Comparative Example A heat radiating plate of a comparative example (plate thickness of 1 mm) under the same conditions as those of the present invention example except that the Cu-Mo composite material and the pure Cu plate have a three-layer structure of Cu / (Cu-Mo) / Cu. (Comparative Examples 1, 2, and 11).
Moreover, the said Cu-Mo composite material single-piece | unit was also used as the heat sink (plate thickness 1mm) of the comparative example (comparative examples 3-10, 12-14).

(3) Measurement of thermal characteristics For each specimen, the coefficient of thermal expansion in the plate surface was measured by a push rod displacement detection method, and the difference in elongation between 50 ° C-400 ° C and 50 ° C-800 ° C was divided by the temperature difference. Then, the average in-plane thermal expansion coefficient from 50 ° C. to 400 ° C. and the average in-plane thermal expansion coefficient from 50 ° C. to 800 ° C. were determined. Further, the thermal conductivity in the plate thickness direction (thermal conductivity at room temperature) was measured by a flash method.
(4) Evaluation of thermal characteristics Tables 1 to 6 show the thermal characteristics of each specimen together with the manufacturing conditions. According to this, it can be seen that the thermal conductivity in the thickness direction of the present invention example is greatly increased compared to the comparative example.

Claims (30)

In the plate thickness direction, Cu layers and Cu-Mo composite layers are alternately stacked to form three or more Cu layers and two or more Cu-Mo composite layers, and the outermost layers on both sides are Cu layers. A heat sink consisting of layers,
The Cu-Mo composite layer has a plate thickness cross-sectional structure in which a flat Mo phase is dispersed in a Cu matrix,
The outermost layer of the Cu layer of the double-sided thickness _{t 1} and the intermediate layer Cu layer of (1a) the thickness _{t 2} of the (1b) satisfies the _{t 1} ≦ _{t 2,} and Cu layer of the outermost layer of double-sided (1a) radiator plate thickness _{t 1} and the thickness T of which satisfies the _t 1 / T ≦ 0.04. Heat radiating plate according to claim 1, the thickness _{t 1} and the thickness T of the Cu layer of the outermost layer of double-sided (1a) is characterized by satisfying the _t 1 / T ≦ 0.02. The heat dissipation according to claim 1 or 2 , wherein the Cu-Mo composite layer has a structure in which a plurality of unit Cu-Mo composite layers are stacked via a joining Cu layer having a thickness of 75 µm or less. Board. The heat sink according to any one of claims 1 to 3 , wherein the Cu-Mo composite layer has a Cu content of 10 to 50 mass%. The heat sink according to any one of claims 1 to 3 , wherein the Cu-Mo composite layer has a Cu content of 20 to 30% by mass. The plate thickness direction of the thermal conductivity of 200 W / m · K or more, more of claims 1 to 5 in which the plate plane average thermal expansion coefficient of up to 800 ° C. from 50 ° C. to equal to or less than 10.0 ppm / K The heat sink of crab. The heat sink according to any one of claims 1 to 6 , wherein a plating film is formed on one side or both sides of a main body of the heat sink composed of the laminated Cu layer and Cu-Mo composite layer. A method of manufacturing a heat sink according to any one of claims 1 to 7
A Cu-Mo composite material (a) and a Cu material (b) having a plate-thickness cross-sectional structure in which a Mo phase is dispersed in a Cu matrix are laminated, and after the diffusion body is diffusion bonded, cold rolling (x) is performed. By this, the manufacturing method of the heat sink characterized by obtaining the heat sink with which the Cu-Mo composite layer by Cu-Mo composite material (a) and the Cu layer by Cu material (b) were laminated | stacked. The Cu-Mo composite material (a) includes a step of pressing a mixed powder of Mo powder and Cu powder to form a green compact, and sintering the green compact in a reducing atmosphere or vacuum. The method of manufacturing a heat sink according to claim 8 , wherein the heat sink is obtained through a step of forming a ligation. The Cu-Mo composite material (a) includes a step of pressing a mixed powder of Mo powder and Cu powder to form a green compact, and sintering the green compact in a reducing atmosphere or vacuum. The method for manufacturing a heat sink according to claim 8 , wherein the heat sink is obtained through a step of forming a sintered body and a step of densifying the sintered body. The Cu-Mo composite material (a) includes a step of pressing a Mo powder or a mixed powder of Mo powder and Cu powder into a green compact, and sintering the green compact in a reducing atmosphere or in vacuum. to a step of a sintered body, according to claim 8, wherein said sintered body is obtained through a step of impregnating a molten Cu or in vacuum in a non-oxidizing atmosphere Manufacturing method of heat sink. The method for manufacturing a heat sink according to any one of claims 8 to 11 , wherein the rolling reduction of cold rolling (x) is 70 to 99%. The manufacturing method of the heat sink according to claim 12 , wherein the rolling reduction of cold rolling (x) is 90 to 96%. Method for manufacturing a heat sink according to any one of claims 8 to 13, characterized in that cold rolling (x) with a cross rolling. The Cu-Mo composite material (a) includes a step of pressing a mixed powder of Mo powder and Cu powder to form a green compact, and sintering the green compact in a reducing atmosphere or vacuum. It is obtained through a step of forming a sintered body, a step of densifying the sintered body, and a step of rolling (y) the densified Cu-Mo composite material. The manufacturing method of the heat sink of Claim 8 . The Cu-Mo composite material (a) includes a step of pressing a Mo powder or a mixed powder of Mo powder and Cu powder into a green compact, and sintering the green compact in a reducing atmosphere or in vacuum. A sintered body, a step of impregnating the sintered body with Cu melted in a non-oxidizing atmosphere or in a vacuum, and rolling (y) the Cu-Mo composite material impregnated with Cu. It is obtained through the process to give, The manufacturing method of the heat sink of Claim 8 characterized by the above-mentioned. The heat sink according to claim 15 or 16 , wherein the total rolling reduction of the Cu-Mo composite (a) obtained by combining cold rolling (x) and rolling (y) is 70 to 99%. Method. The method for manufacturing a heat sink according to claim 17 , wherein the total rolling reduction of the Cu-Mo composite (a) obtained by combining cold rolling (x) and rolling (y) is 90 to 96%. Method for manufacturing a heat sink according to any of claims 15 to 18, characterized in that rolling (y) in the cross rolling. When the Cu-Mo composite material (a) is unidirectionally rolled by rolling (y), in cold rolling (x), the Cu-Mo composite material is rolled in a direction orthogonal to the rolling direction of rolling (y). method for manufacturing a heat sink according to any of claims 15 to 19, characterized in. The method for manufacturing a heat sink according to any one of claims 8 to 20 , wherein the Cu-Mo composite material (a) is a laminate of a plurality of unit Cu-Mo composite materials (a _u ). Cu-Mo composite material (a) is in any one of claims 8 to 20, wherein the plurality of unit Cu-Mo composite material _{(a u)} is formed by laminating via the Cu thin plate for joining The manufacturing method of the heat sink of description. The method for manufacturing a heat sink according to any one of claims 8 to 22 , wherein the Cu material (b) is a laminate of a plurality of unit Cu materials (b _u ). The Cu-Mo composite material (a) has a Cu content of 10 to 50% by mass, and the method for manufacturing a heat sink according to any one of claims 8 to 23 . The Cu-Mo composite material (a) has a Cu content of 20 to 30% by mass, and the method for manufacturing a heat sink according to any one of claims 8 to 23 . Cu content of Cu-Mo composite material (a) is less than 20 mass%, and the total rolling reduction of Cu-Mo composite material (a) combining cold rolling (x) and rolling (y) is 70% or more. A manufacturing method (including a manufacturing method in which the rolling (y) of the Cu—Mo composite material (a) is not performed),
The method for producing a heat sink according to claim 24 , wherein the warm rolling of (1) and / or (2) below is performed.
(1) Perform warm rolling instead of cold rolling (x).
(2) Rolling (y) is performed by warm rolling. A manufacturing method in which the total rolling reduction of the Cu-Mo composite (a) obtained by combining the cold rolling (x) and the rolling (y) is 96% or more (provided that the rolling of the Cu-Mo composite (a) (y) Including a manufacturing method that does not perform
The method for manufacturing a heat sink according to claim 25 , wherein the warm rolling of (1) and / or (2) below is performed.
(1) Perform warm rolling instead of cold rolling (x).
(2) Rolling (y) is performed by warm rolling. The method of manufacturing a heat sink according to any one of claims 8 to 27 , wherein a plating film is formed on one side or both sides of a heat sink main body composed of the laminated Cu-Mo composite layer and the Cu layer. Semiconductor package characterized by comprising a heat dissipation plate according to any one of claims 1-7. 30. A semiconductor module comprising the semiconductor package according to claim 29 .

Images