JP4265247B2 - High heat dissipation alloy, heat dissipation plate, package for semiconductor element, and manufacturing method thereof - Google Patents

Description

【００２８】
２．粒径および含有量の検討
ここで、平均粒径が０．１μｍのＷ粉末と平均粒径が０．５μｍのＣｕ粉末を用いて、Ｃｕ粉末の含有量を１０，１５，２０，２５，３０，３５，４０，４５，５０質量％になるように変化させて、上述と同様にして焼結体（高放熱性合金）１０ｂを用いた放熱板を作製し、これらを放熱板ａ１，ａ２，ａ３，ａ４，ａ５，ａ６，ａ７，ａ８，ａ９とした。
【００２９】
同様に、平均粒径が０．５μｍのＷ粉末と平均粒径が２．５μｍのＣｕ粉末を用いて、Ｃｕ粉末の含有量を１０，１５，２０，２５，３０，３５，４０，４５，５０質量％になるように変化させて、上述と同様にして焼結体（高放熱性合金）１０ｂを用いた放熱板を作製し、これらを放熱板ｂ１，ｂ２，ｂ３，ｂ４，ｂ５，ｂ６，ｂ７，ｂ８，ｂ９とした。また、平均粒径が０．９μｍのＷ粉末と平均粒径が４．５μｍのＣｕ粉末を用いて、Ｃｕ粉末の含有量を１０，１５，２０，２５，３０，３５，４０，４５，５０質量％になるように変化させて、上述と同様にして焼結体（高放熱性合金）１０ｂを用いた放熱板を作製し、これらを放熱板ｃ１，ｃ２，ｃ３，ｃ４，ｃ５，ｃ６，ｃ７，ｃ８，ｃ９とした。
【００３０】
一方、平均粒径が１．５μｍのＷ粉末と平均粒径が１．５μｍのＣｕ粉末を用いて、Ｃｕ粉末の含有量を１０，１５，２０，２５，３０，３５，４０，４５，５０質量％になるように変化させて、上述と同様にして焼結体（高放熱性合金）１０ｂを用いた放熱板を作製し、これらを放熱板ｘ１，ｘ２，ｘ３，ｘ４，ｘ５，ｘ６，ｘ７，ｘ８，ｘ９とした。さらに、比較のためにＣｕの含有量が１０，１５，２０質量％になるように変化させた溶浸材を作製し、これらを放熱板ｙ１，ｙ２，ｙ３とした。なお、溶浸材においては、Ｃｕの含有量は２０質量％が限度であった。
【００３１】
ついで、これらの各放熱板ａ１〜ａ９，ｂ１〜ｂ９，ｃ１〜ｃ９，ｘ１〜ｘ９，ｙ１〜ｙ３の熱伝導率（Ｗ／ｍＫ）と熱膨張係数（ｐｐｍ／Ｋ）を測定した。この場合、熱伝導率（Ｗ／ｍＫ）は熱定数測定装置（アルバック理工（株）製）を用いてレーザーフラッシュ法により求め、Ｃｕ含有量（質量％）を横軸にプロットし、熱伝導率（Ｗ／ｍＫ）を縦軸にプロットすると、図３に示すような曲線ａ（ａ１〜ａ９），ｂ（ｂ１〜ｂ９），ｃ（ｃ１〜ｃ９），ｘ（ｘ１〜ｘ９），ｙ（ｙ１〜ｙ３）が得られた。
【００３２】
また、熱膨張係数（ｐｐｍ／Ｋ）を差動トランス法により求め、Ｃｕ含有量（質量％）を横軸にプロットし、熱膨張係数（ｐｐｍ／Ｋ）を縦軸にプロットすると、図４に示すような曲線ａ（ａ１〜ａ９），ｂ（ｂ１〜ｂ９），ｃ（ｃ１〜ｃ９），ｘ（ｘ１〜ｘ９），ｙ（ｙ１〜ｙ３）が得られた。ついで、これらの結果から、熱膨張係数（ｐｐｍ／Ｋ）を横軸にプロットし、熱伝導率（Ｗ／ｍＫ）を縦軸にプロットすると、図５に示すような曲線ａ（ａ１〜ａ９），ｂ（ｂ１〜ｂ９），ｃ（ｃ１〜ｃ９），ｘ（ｘ１〜ｘ９），ｙ（ｙ１〜ｙ３）が得られた。
【００３３】
図３において、溶浸材から曲線ｙは均一なＣｕのマトリクスが形成されないために、熱伝導率が一番低いことが分かる。そして、曲線ｘと曲線ａ，ｂ，ｃとを比較すると、曲線ａ，ｂ，ｃにおいてはＣｕの含有量が増大するに伴って熱伝導率が向上するが、曲線ｘにおいてはＣｕの含有量がある一定以上になると熱伝導率がそれほど向上しないことが確認できた。これは、曲線ｘにおいてはＷ粒子の平均粒径が１．５μｍと大きいため、焼結密度を上げるために高温での焼結が必要となる。このため、Ｗ粒子が成長することによりＷ粒子同士が固まった状態で接合し、図１（ｄ）に示すような結合状態が形成されなく、したがって、Ｃｕのチャネリング効果が減少して熱伝導率が低下したと考えられる。
【００３４】
一方、曲線ａ，ｂ，ｃの場合は、Ｗ粒子の平均粒径が１．０μｍ未満と小さく、かつこの非常に微小なＷ粒子同士が結合して、図１（ｄ）に示すような結合状態となるように熱処理が施されている。これにより、Ｗスケルトン（Ｗの三次元網目状構造の骨格）が形成されて、これがＣｕのマトリクス中に独立して分布している。このため、Ｃｕのチャネリング効果で熱伝導率が大きくなったと考えられる。この場合、図３において、曲線ｃよりも曲線ｂの方が上方に位置し、曲線ｂよりも曲線ａの方が上方に位置することから、Ｗ粒子の平均粒径（Ｃｕ粒子はその５倍）が小さくなるほど熱伝導率が向上ということができる。
【００３５】
なお、平均粒径が０．１μｍ未満のＷ粒子を作製するのは困難なため、Ｗ粒子の平均粒径は０．１μｍ以上で、１．０μｍ未満にするのが望ましいということができる。この場合、Ｃｕ粒子の平均粒径はＷ粒子の平均粒径の５倍以上の大きさにする必要があるが、Ｃｕ粒子の平均粒径が５μｍ以上になると、Ｃｕのマトリクス中でＷの偏析が起こり、組織が不均一となって熱伝導率も減少する。このため、Ｃｕ粒子の平均粒径はＷ粒子の平均粒径の５倍以上で、５μｍ未満にするのが望ましいということができる。また、Ｃｕ粒子の含有量を多くすればするほど熱伝導率が向上するが、熱伝導率が２５０Ｗ／ｍＫ（Ｋ＝２５０Ｗ／ｍＫ）以上であれば充分に熱伝導性が良好な放熱板ということができるので、Ｃｕ粒子の含有量は２５質量％以上にするのが望ましい。
【００３６】
図４において、曲線ｘと曲線ａ，ｂ，ｃとを比較すると、曲線ａが一番下方に位置して、これの上に曲線ｂ、曲線ｃが位置し、一番上に曲線ｘが位置していることが分かる。このことは、Ｗ粒子の平均粒径（Ｃｕ粒子はその５倍）が小さくなるほど熱膨張係数が低下することを意味している。これは、曲線ｘにおいては、Ｗ粒子の平均粒径が１．５μｍと大きいために高温での焼結が必要となって、Ｗ粒が成長することによりＷ粒子同士が固まった状態で接合している。このため、Ｗスケルトン（Ｗの三次元網目状構造の骨格）が形成されなくて、熱膨張係数が増加したと考えられる。
【００３７】
一方、曲線ａ，ｂ，ｃにおいては、Ｗ粒子の平均粒径が１．０μｍ未満と小さいために、微小なＷ粒子同士が結合してＷスケルトン（Ｗの三次元網目状構造の骨格）が形成されている。このため、このＷの三次元網目状構造の骨格が熱膨張を規制するように作用して、熱膨張係数が低下したと考えられる。このことからも、Ｗ粒子の平均粒径は０．１μｍ以上で、１．０μｍ未満で、Ｃｕ粒子の平均粒径はＷ粒子の平均粒径の５倍以上で、５μｍ未満にするのが望ましいということができる。また、Ｃｕ粒子の含有量を多くすればするほど熱膨張係数が増加するが、熱膨張係数が１２ｐｐｍ／Ｋ（α＝１２ｐｐｍ／Ｋ）以下であればガラスやセラミックの熱膨張係数に近似するため、Ｃｕ粒子の含有量は５０質量％以下にするのが望ましい。
【００３８】
図５において、曲線ｙは一番下に位置し、その上に曲線ｘが位置し、その上に曲線ａ，ｂ，ｃが位置していることが分かる。また、曲線ｘにおいては、熱膨張係数が大きくなっても、それに伴って熱伝導率が大きくはならず、Ｃｕの添加量を増大させて熱伝導率を向上させるという目的が達成できないことが分かる。
一方、曲線ａ，ｂ，ｃにおいては、熱膨張係数が大きくなるに伴って熱伝導率も大きくなり、Ｃｕの添加量を増大させて熱伝導率を向上させるという目的が達成できることが分かる。
【００３９】
この場合、曲線ａが一番上方に位置して、これの下に曲線ｂが位置し、その下に曲線ｃが位置していることから、Ｗ粒子の平均粒径（Ｃｕ粒子はその５倍）が小さくなるほど、Ｃｕの添加量を増大させて熱伝導率を向上させるという目的が達成できることが分かる。このことからも、Ｗ粒子の平均粒径は０．１μｍ以上で、１．０μｍ未満で、Ｃｕ粒子の平均粒径はＷ粒子の平均粒径の５倍以上で、５μｍ未満にするのが望ましいということができる。そして、Ｃｕ粒子の含有量においては、熱伝導率が２５０Ｗ／ｍＫ以上になる２５質量％以上で、熱膨張係数が１２ｐｐｍ／Ｋ以下になる５０質量％以下にするのが望ましい。
【００４０】
３．ハイパワー高周波トランジスタの実装評価
ついで、上述のようにして作製された放熱板に発熱量が大きいハイパワー高周波トランジスタ（出力が１５０Ｗ以上のもの）を実装して、この放熱板の放熱性の評価を行った。そこで、図６および図７に示すように、まず、上述のようにして作製された放熱板１０ｂを用いて半導体素子用パッケージ２１を用意し、このパッケージ２１の上にアルミナ回路基板２２をＰｂＳｎ半田２７により接合した。ついで、アルミナ回路基板２２の所定の位置にハイパワー高周波トランジスタ２３を配設するとともに、アルミナ回路基板２２の両側に一対のリード２４，２４を接続した。そして、アルミナ回路基板２２の接続部とハイパワー高周波トランジスタ２３の接続部との間をワイヤーボンディング２５により接続した。この後、これらの上にカバー２６を接合して高周波トランジスタモジュール２０を作製した。
【００４１】
この場合、平均粒径が０．９μｍのＷ粉末と平均粒径が４．５μｍのＣｕ粉末を用い、Ｃｕ粉末の含有量が２５質量％になるような半導体素子用パッケージ２１を用いた高周波トランジスタモジュール２０を試料Ａとし、Ｃｕ粉末の含有量が５０質量％になるような半導体素子用パッケージ２１を用いた高周波トランジスタモジュール２０を試料Ｂとした。同様に、平均粒径が０．５μｍのＷ粉末と平均粒径が２．５μｍのＣｕ粉末を用い、Ｃｕ粉末の含有量が２５質量％になるような半導体素子用パッケージ２１を用いた高周波トランジスタモジュール２０を試料Ｃとし、Ｃｕ粉末の含有量が５０質量％になるような半導体素子用パッケージ２１を用いた高周波トランジスタモジュール２０を試料Ｄとした。
【００４２】
同様に、平均粒径が０．１μｍのＷ粉末と平均粒径が０．５μｍのＣｕ粉末を用い、Ｃｕ粉末の含有量が２５質量％になるような半導体素子用パッケージ２１を用いた高周波トランジスタモジュール２０を試料Ｅとし、Ｃｕ粉末の含有量が５０質量％になるような半導体素子用パッケージ２１を用いた高周波トランジスタモジュール２０を試料Ｆとした。同様に、平均粒径が１．５μｍのＷ粉末と平均粒径が１．５μｍのＣｕ粉末を用い、Ｃｕ粉末の含有量が２５質量％になるような半導体素子用パッケージ２１を用いた高周波トランジスタモジュール２０を試料Ｘとし、Ｃｕ粉末の含有量が５０質量％になるような半導体素子用パッケージ２１を用いた高周波トランジスタモジュール２０を試料Ｙとした。
【００４３】
この後、これらの試料Ａ，Ｂ，Ｃ，Ｄ，Ｅ，Ｆ，Ｘ，Ｙをそれぞれヒートシート２８を介してアルミニウム製放熱フィン２９の上に配置した。ついで、これらを外部温度（Ｔａ）が２５℃の温度雰囲気中に配置した後、ハイパワー高周波トランジスタ２３の出力を１００Ｗ〜３００Ｗまで変化させながら、ＭＩＬ−ＳＴＤ−８８３Ｃに準拠して、アルミナ回路基板２２とハイパワー高周波トランジスタ２３との接合部の温度（Ｔｊ）求めた。その結果、図８に示すような結果が得られた。なお、この種のハイパワー高周波トランジスタ２３においては、接合部温度（Ｔｊ）が１５０℃を超えた状態で使用を続けると、この素子２３が破壊したり、素子２３の寿命が著しく低下するため、接合部温度（Ｔｊ）は１３０℃以下になるように放熱するのが望ましい。
【００４４】
図８の結果から明らかなように、試料Ｙのように、Ｃｕ粉末の含有量が５０質量％であっても、出力が１５０Ｗを超えると接合部温度（Ｔｊ）は１５０℃以上になって、素子２３とパッケージ２１の熱膨張係数差から生じる応力により、素子２３やアルミナ回路基板２２が損傷した。一方、試料Ａにおいては、Ｃｕ粉末の含有量が２５質量％（０．９μｍＷ，４．５μｍＣｕ）であっても、素子２３の出力が２００Ｗまでは接合部温度（Ｔｊ）を１３０℃以下に維持できた。また、Ｃｕ粉末の含有量が５０質量％（０．９μｍＷ，４．５μｍＣｕ）の試料Ｂ、およびＣｕ粉末の含有量が２５質量％（０．５μｍＷ，２．５μｍＣｕ）の試料Ｃにおいては、素子２３の出力が３００Ｗまでは接合部温度（Ｔｊ）を１３０℃以下に維持できた。
【００４５】
また、Ｃｕ粉末の含有量が５０質量％（０．５μｍＷ，２．５μｍＣｕ）の試料Ｄ、およびＣｕ粉末の含有量が２５質量％（０．１μｍＷ，０．５μｍＣｕ）の試料Ｅにおいては、素子２３の出力が３００Ｗまでは接合部温度（Ｔｊ）を１２０℃以下に維持できた。さらに、Ｃｕ粉末の含有量が５０質量％（０．１μｍＷ，０．５μｍＣｕ）の試料Ｆにおいては、素子２３の出力が３００Ｗまでは接合部温度（Ｔｊ）を１１０℃以下に維持できた。
【００４６】
【発明の効果】
上述したように、本発明においては、タングステン粒子１２が連続して結合した三次元網目状構造の骨格となっているため、銅１１の含有量が２５〜５０質量％と多くなっても、熱膨張係数を１２ｐｐｍ／Ｋ以下に維持できるようになる。そして、平均粒径が１．０μｍ未満のタングステン粒子１２同士が結合した三次元網目状構造の骨格を形成し、この骨格がこれらの粒子の平均粒径の５倍以上で、５．０μｍ未満の銅粉末１１が溶融したマトリックス内に分布している。これにより、銅のチャネリング効果が発揮できて、熱伝導性に優れたタングステン−銅合金が得られるようになる。
【００４７】
なお、上述した実施形態においては、主にＷＣｕ合金について説明したが、ＭｏＣｕ合金についてもＷＣｕ合金の場合と同様である。また、上述した実施形態においては、半導体素子としてハイパワー高周波トランジスタを用いる例について説明したが、半導体素子として、半導体レーザなどの他の半導体素子に適用できることも明らかである。
【図面の簡単な説明】
【図１】 本発明の高放熱性合金が得られる過程を模式的に示す断面図である。
【図２】 脱バインダ処理工程および焼結工程での昇温パターンを示す図である。
【図３】 ＷＣｕ合金のＣｕの含有量（Ｃｕ濃度）に対する熱伝導率の関係を示す図である。
【図４】 ＷＣｕ合金のＣｕの含有量（Ｃｕ濃度）に対する熱膨張係数の関係を示す図である。
【図５】 ＷＣｕ合金の熱膨張係数に対する熱伝導率の関係を示す図である。
【図６】 ＷＣｕ合金を用いた放熱板に高周波トランジスタが配置されて半導体素子用パッケージが形成された状態を示す斜視図である。
【図７】 図６の半導体素子用パッケージにカバーを装着して放熱試験を行う状態を模式的に示す図であって、図７（ａ）は図６の半導体素子用パッケージにカバーを装着した状態での図６のＡ−Ａ断面を示す図であり、図７（ｂ）は図７（ａ）のパッケージの下部に放熱フィンを装着した状態を示す断面図である。
【図８】 高周波トランジスタの出力（Ｗ）と、高周波トランジスタとセラミック基板との接合部温度（Ｔｊ）の関係を示す図である。
【図９】 従来例の溶浸材からなるＷＣｕ合金を模式的に示す断面図である。
【図１０】 従来例の粉末法によるＷＣｕ合金を模式的に示す断面図である。
【図１１】 銅の含有量が多い場合の従来例の粉末法によるＷＣｕ合金の製造過程を模式的に示す断面図である。
【図１２】 銅の含有量が少ない場合の従来例の粉末法によるＷＣｕ合金の製造過程を模式的に示す断面図である。
【符号の説明】
１０…混合粉末体、１０ａ…スケルトン、１０ｂ…放熱板、１１…銅（Ｃｕ）粉末、１２…タングステン（Ｗ）粉末、２０…高周波トランジスタモジュール、２１…半導体素子用パッケージ、２２…アルミナ回路基板、２３…ハイパワー高周波トランジスタ、２４…リード、２５…ワイヤーボンディング、２６…カバー、２７…半田、２８…ヒートシート、２９…アルミニウム製放熱フィン[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a high heat dissipation alloy composed of tungsten-copper alloy or molybdenum-copper alloy having a low thermal expansion coefficient and high heat dissipation, a heat dissipation plate using this high heat dissipation alloy, a package for a semiconductor element using this heat dissipation plate, And a manufacturing method thereof.
[0002]
[Prior art]
Conventionally, a semiconductor device such as a semiconductor package, a high-frequency transistor module, or an optical communication module has been provided with a heat radiating member called a heat sink so that the heat generated in the semiconductor package, the high-frequency transistor module, or the optical communication module can be efficiently removed from the system. To dissipate (to exhaust heat). Such a heat sink has high thermal conductivity and good thermal conductivity, and since it is bonded to a ceramic substrate or a glass substrate, the thermal expansion coefficient is the thermal expansion coefficient (4 to 4) of ceramic or glass. It is required to approximate 12 ppm / K). At present, a tungsten-copper alloy (or molybdenum-copper alloy) is used as a material having such characteristics that the thermal conductivity and the thermal expansion coefficient conflict with each other.
[0003]
As a heat sink made of such a tungsten-copper alloy, FIG. 9 (note that FIG. 9 (a) schematically shows a state before infiltrating W into the sintered body, and FIG. 9 (b) shows infiltration. Infiltration in which copper (Cu) 32 is infiltrated (impregnated) into the pores 31a of the sintered body 31 made of tungsten (W), as shown in FIG. A heat sink 30 made of a sintered alloy is used. By the way, the heat sink 30 which consists of such an infiltration sintered alloy is manufactured in the following procedures, for example. First, a tungsten (W) powder is preliminarily blended with an organic binder to form a raw material mixture, and this raw material mixture is pressed by a mold press to form a thin plate-shaped molded body. The molded body is degreased and sintered to form a porous sintered body (W skeleton) 31, and copper (Cu) 32 is infiltrated (impregnated into the pores 31 a of the sintered body (W skeleton) 31. ).
[0004]
However, in the heat sink 30 made of the above-described infiltrated sintered alloy, it is difficult to always infiltrate a certain amount of Cu 32 because it is difficult to control the holes 31a of the W skeleton 31. Even if the air holes 31 a of the W skeleton 31 can be controlled, it is difficult to hold the Cu 32 in the W skeleton 31 after the Cu 32 is infiltrated. The problem of elution to the outside occurred.
[0005]
Therefore, a heat sink made of a tungsten-copper alloy manufactured by a powder method has been proposed in Patent Document 1 (Japanese Patent Laid-Open No. 10-245603). As shown in FIG. 10, the heat sink 40 proposed in Patent Document 1 is a mixture in which copper powder 41 and tungsten powder 42 are sufficiently mixed by a mechanical alloy method, and then at a temperature equal to or higher than the melting point of copper 42 in a reducing atmosphere. It is manufactured by sintering.
[Patent Document 1]
JP-A-10-245603
[0006]
[Problems to be solved by the invention]
However, the tungsten-copper alloy proposed in the above-mentioned Japanese Patent Application Laid-Open No. 10-245603 has a problem that the sintered density is not improved because it is sufficiently mixed by the mechanical alloy method. This is because the average particle diameter of the tungsten particles 42 of the raw material is as large as 2 to 3 μm. Therefore, when the content (concentration) of the copper 41 is 25% by mass or more, as shown in FIG. Since the probability of being adjacent to each other decreases, the copper 41 is first melted and sintered before the sintering of the tungsten 42 starts. For this reason, as shown in FIG.11 (b), the void | hole 43 was formed at the time of sintering and the problem that a sintering density did not improve was produced.
[0007]
On the other hand, when the content (concentration) of copper 41 is less than 25 mass%, as shown in FIG. 12A, the probability that the tungsten 42 is adjacent to each other increases, so the sintering of the tungsten 42 is started first. Thereafter, the copper 41 is melted and sintered. For this reason, as shown in FIG.12 (b), it can prevent that a void | hole is formed at the time of sintering, and a sintering density improves. However, since the sintering temperature is increased, tungsten particles grow during sintering, causing a problem that the channeling effect by copper is reduced and the thermal conductivity is lowered.
[0008]
Further, when the sintering temperature is increased to increase the sintering density, there arises a problem that copper flows out of the system. Furthermore, when the average particle size of the tungsten particles as the raw material is reduced and the average particle size of the copper particles is reduced, the probability that tungsten is adjacent to each other decreases, and the probability that copper is adjacent to each other increases. . For this reason, before the tungsten particles are sintered, the copper is first melted, and there is a problem that the sintered density does not increase.
[0009]
Accordingly, the present invention has been made to solve the above-described problems, and is easy to manufacture and has a low heat expansion coefficient and a high heat dissipation property, and is a high heat dissipation alloy made of tungsten-copper alloy or molybdenum-copper alloy. An object of the present invention is to provide a heat radiating plate using the high heat radiating alloy, a package for a semiconductor device using the heat radiating plate, and a method of manufacturing the same.
[0010]
[Means for Solving the Problems]
  In order to achieve the above object, the high heat dissipation alloy of the present invention comprises a tungsten-copper alloy or molybdenum-copper alloy containing 25 to 50% by mass of copper, and tungsten particles or molybdenum particles are continuously bonded. A skeleton having a three-dimensional network structure is formed, copper is filled in these skeletons, and the average particle diameter of the raw material powder of tungsten particles or molybdenum particles is 0.1 μm or more and less than 1.0 μm, and copper The average particle size of the raw material powder is 5 times or more the average particle size of the raw material powder of tungsten particles or molybdenum particles and less than 5.0 μmAnd a skeleton having a three-dimensional network structure in which tungsten particles or molybdenum particles are continuously bonded such that the cross-sectional length of the bonded portion of the tungsten particles or molybdenum particles is ½ or less of the particle size of these particles. A three-dimensional network in which tungsten particles or molybdenum particles are continuously bonded so that the cross-sectional length of the bonded portion of the tungsten particles or molybdenum particles is ½ or less of the particle size of these particles. A skeleton with a structure is formedIt is characterized by that.
[0011]
Here, in the tungsten-copper alloy or molybdenum-copper alloy, the thermal conductivity increases as the proportion of copper increases, but the thermal expansion coefficient also increases. This is because copper is a metal excellent in heat conduction, so that the coefficient of thermal expansion increases as the proportion of copper increases. For this reason, in order to make a tungsten-copper alloy or molybdenum-copper alloy having a low thermal expansion coefficient, it is necessary to regulate the ratio of copper. In general, when the thermal expansion coefficient is 12 ppm / K or less, the thermal expansion coefficient approximates that of ceramic or glass. For this reason, it is desirable that the copper content of the tungsten-copper alloy or molybdenum-copper alloy is 20% by mass or less with respect to the mass of the tungsten-copper alloy or molybdenum-copper alloy.
[0012]
However, in the present invention, since the skeleton has a three-dimensional network structure in which tungsten particles or molybdenum particles are continuously bonded, even if the copper content increases to 25 to 50% by mass, the thermal expansion coefficient Can be maintained at 12 ppm / K or less. Then, a skeleton having a three-dimensional network structure in which tungsten particles having an average particle diameter of less than 1.0 μm or molybdenum particles are bonded to each other is formed, and the skeleton is 5 times or more the average particle diameter of these particles. Copper powder of less than 0 μm is distributed in the molten matrix. Thereby, the channeling effect of copper can be exhibited, and a tungsten-copper alloy or a molybdenum-copper alloy having excellent thermal conductivity can be obtained.
[0013]
In this case, in order to increase the probability that the tungsten particles or the molybdenum particles are adjacent to each other, the average particle size of the copper particles needs to be 5 times or more the average particle size of the tungsten particles or molybdenum particles. However, when the average particle diameter of the copper particles is 5 μm or more, segregation of tungsten or molybdenum occurs in the copper matrix, the structure becomes non-uniform, and the thermal conductivity decreases. For this reason, it is desirable that the average particle diameter of the copper particles be 5 times or more the average particle diameter of the tungsten particles or molybdenum particles and less than 5 μm.
[0014]
A three-dimensional network structure is formed by continuously bonding tungsten particles or molybdenum particles so that the cross-sectional length of the bonded portion of tungsten particles or molybdenum particles is 1/2 or less of the particle size of these particles. It is desirable that This is because when the cross-sectional length of the bonded portion exceeds 1/2 of the particle size of these particles, the heat flow path between the tungsten particles or molybdenum particles is obstructed, and the channeling effect of copper cannot be exhibited. As a result, the thermal resistance increases, which is not preferable.
[0015]
  Further, in the method for producing a high heat dissipation alloy of the present invention, tungsten fine powder and copper fine powder having an average particle size larger than this, or molybdenum fine powder and copper fine powder having an average particle size larger than this are mixed, A mixing step in which the surface of the copper fine powder is coated with tungsten fine powder or molybdenum fine powder, and the mixture is heated at a temperature below the melting point of copper (1083 ° C.) to make the fine tungsten powder or fine molybdenum powder Start sinteringIn addition, the sintering temperature and the sintering time are adjusted so that the cross-sectional length of the bonded portion of the tungsten particles or molybdenum particles is ½ or less of the particle size of these particles.Skeleton forming step (first sintering step) and after the skeleton forming step, a sintering step in which copper is melted by heating at a temperature not lower than the melting point of copper (1083 ° C.) and not higher than the melting points of tungsten and molybdenum (1300 ° C.). (Second sintering step).
[0016]
After mixing the tungsten (or molybdenum) fine powder and the copper fine powder having an average particle size larger than this to obtain a mixture in which the surface of the copper fine powder is coated with the tungsten (or molybdenum) fine powder, When heated at a temperature below the melting point of copper (1083 ° C.), sintering of tungsten (or molybdenum) fine powders started, and tungsten (or molybdenum) skeleton (tungsten (or molybdenum) particles were continuously bonded. A skeleton of a three-dimensional network structure) is formed. Thereafter, when the copper is melted by heating at a temperature not lower than the melting point of copper (1083 ° C.) and not higher than the melting point of tungsten (or molybdenum) (1300 ° C.), the copper moves to the space of the three-dimensional network structure. It will be filled.
[0017]
As a result, a sintered body having a sintered density of 98.5% or more and close to the theoretical density can be obtained. In this case, in the mixing step, when the tungsten (or molybdenum) fine powder and the copper fine powder having a larger average particle size are mixed without mechanical alloying, the surface of the copper fine powder becomes the tungsten (or molybdenum) fine powder. A mixture coated with is obtained. Note that “without mechanical alloy” means that the tungsten (or molybdenum) fine powder and the copper fine powder are in an independent particle state and do not come into an entangled state until they are united by plastic deformation. .
[0018]
  When mixing, if tungsten (or molybdenum) fine powder and copper fine powder are sufficiently mixed together by plastic deformation or mechanical alloy, copper sintering occurs first in the sintering process, making it difficult to increase the sintering density. Desirable becauseYes.
[0019]
DETAILED DESCRIPTION OF THE INVENTION
Next, an embodiment of the high heat dissipation alloy of the present invention, a heat dissipation plate using the heat dissipation alloy, and a package for a semiconductor element using the heat dissipation plate will be described with reference to FIGS. FIG. 1 is a cross-sectional view schematically showing the process of obtaining the high heat dissipation alloy of the present invention. FIG. 2 is a view showing a temperature rising pattern in the binder removal process and the sintering process. FIG. 3 is a diagram showing the relationship of the thermal conductivity with respect to the Cu content (mass%) of the obtained WCu alloy. FIG. 4 is a diagram showing the relationship of the thermal expansion coefficient with respect to the Cu content (mass%) of the WCu alloy.
[0020]
FIG. 5 is a diagram showing the relationship of the thermal conductivity to the thermal expansion coefficient of the WCu alloy. FIG. 6 is a perspective view showing a state where a package for a semiconductor element is formed by arranging a high-frequency transistor on a heat sink using the obtained WCu alloy. FIG. 7 is a diagram schematically showing a state in which a cover is attached to the semiconductor element package of FIG. 6 and a heat dissipation test is performed. FIG. 7A is a state in which the cover is attached to the semiconductor element package of FIG. FIG. 7B is a cross-sectional view showing a state in which a radiation fin is attached to the lower part of the package of FIG. 7A. FIG. 8 is a graph showing the relationship between the output (W) of the high frequency transistor and the junction temperature (Tj) between the high frequency transistor and the ceramic substrate.
[0021]
1. Production of heat sink (high heat dissipation alloy plate)
First, electrolytic Cu powder 11 having an average particle size of 0.5 μm or more and less than 5.0 μm and W powder (or Mo powder) 12 having an average particle size of 0.1 μm or more and less than 1.0 μm were prepared. Then, the Cu powder 11 is weighed to 25 to 50% by mass and the W powder (or Mo powder) 12 is weighed to 75 to 50% by mass and put into a Henschel mixer, and then these powders are mixed thoroughly. did. In this case, it is necessary not to use a mixer with a mechanical alloy such as a ball mill or an attritor that gives plastic deformation to the powder. By such mixing, as shown in FIG. 1A, a mixed powder body 10 in which the surface of the Cu powder 11 is coated with the W powder (or Mo powder) 12 is formed.
[0022]
Next, the mixed powder body 10 was molded into a predetermined shape (for example, a plate-like body having a width of 15 mm, a length of 35 mm, and a thickness of 2 mm) by a dry compaction method or an injection molding method. . Here, when applying the dry compaction method, first, a wax liquid in which paraffin wax was dissolved in methyl alcohol was added to the mixed powder body 10 to obtain a slurry. In this case, the paraffin wax content was 1% by mass with respect to the mass of the mixed powder body 10. Thereafter, the resulting slurry can be formed into a spherical shape by a spray dryer and then formed by molding with a dry powder press.
[0023]
On the other hand, when applying the injection molding method, first, a thermoplastic resin and an organic binder were added to the obtained mixed powder body 10 to form a feedstock, and then pelletized by a pelletizer. Next, the obtained pellets are put into an injection molding machine and molded into a predetermined shape. In this case, polypropylene (PP), polystyrene (PS), polyethylene (PE), acrylic, POM resin, or the like was used as the thermoplastic resin. Moreover, as the organic binder, paraffin wax, carnapa wax, beeswax or the like was used, and the mixing ratio was 48 vol% with respect to the volume of the mixed powder body 10.
[0024]
Next, after the molded body obtained by the above-described dry compaction method or injection molding method is placed in a debinder device (not shown), nitrogen gas is introduced into the debinder device at a flow rate of 1 liter / min. It was an atmosphere. And it heated at the temperature increase rate of 1 degree-C / min (0-> t1 of FIG. 2), and hold | maintained the temperature of 500 degreeC for 2 hours (t1-> t2 of FIG. 2). Thus, the molded body was subjected to binder removal processing. By this binder removal treatment, the organic binder such as the paraffin wax described above is volatilized (removed). Then, after cooling to room temperature (t2-> t3 of FIG. 2), it removed from the binder removal apparatus.
[0025]
Next, the binder-treated compact was put into a sintering furnace and heated in a hydrogen stream (1 to 10 liters / minute) at a heating rate of 5 ° C./minute (t3 → t4 in FIG. 2). 950 ° C., which is a temperature below the melting point of Cu, was held for 1 hour (t4 → t5 in FIG. 2). Thereby, as shown in FIG. 1 (b), the W powder particles (or Mo powder particles) 12 are started to bond (sinter) to each other, and these W powder particles (or Mo powder particles) 12 continue. Thus, a skeleton 10a having a three-dimensional network structure skeleton is formed. In this case, as shown in FIG. 1 (d), the heating temperature is set so that the cross-sectional length y of the bonded portion is 1/2 or less (y ≦ (1/2) x) of the diameter x of these particles 12. It is necessary to adjust the heating time. This is because when the cross-sectional length y of the joint portion exceeds 1/2 of the diameter x of the particle 12, the heat flow path between the W powder particles (or Mo powder particles) 12 is obstructed and the channeling of Cu is performed. This is because the effect cannot be exhibited, and as a result, the thermal resistance becomes large, which is not preferable.
[0026]
Thereafter, the temperature was increased at a rate of temperature increase of 5 ° C./min (t5 → t6 in FIG. 2), and heating was performed until the maximum temperature not lower than the melting point of Cu (1083 ° C.) as shown in Table 1 below. After maintaining this maximum temperature for 2 hours (t6 → t7 in FIG. 2), the temperature was lowered to room temperature at a rate of temperature decrease of 10 ° C./min (t7 → t8 in FIG. 2). As a result, the Cu powder 11 is melted and the liquid Cu moves to the voids of the skeleton 10a in which the skeleton of the three-dimensional network structure is formed. Thereby, as shown in FIG.1 (c), the heat sink which consists of the sintered compact (high heat dissipation alloy) 10b whose Cu filling density is 98.5% or more (close to the theoretical density) is formed. Become. As shown in Table 1 below, the optimum maximum sintering temperature needs to be increased as the Cu content decreases and the particle size of the W powder (Mo powder) increases. There is. In addition, the average particle diameter of Cu powder was made into 5 times the average particle diameter of W powder (Mo powder).
[0027]
[Table 1]

[0028]
2. Examination of particle size and content
Here, using a W powder having an average particle size of 0.1 μm and a Cu powder having an average particle size of 0.5 μm, the content of Cu powder is 10, 15, 20, 25, 30, 35, 40, 45, A heat sink using a sintered body (high heat dissipation alloy) 10b was prepared in the same manner as described above by changing to 50% by mass, and these were used as heat sinks a1, a2, a3, a4, a5, a6. , A7, a8, a9.
[0029]
Similarly, using W powder with an average particle diameter of 0.5 μm and Cu powder with an average particle diameter of 2.5 μm, the content of Cu powder is 10, 15, 20, 25, 30, 35, 40, 45, In the same manner as described above, heat sinks using the sintered body (high heat dissipation alloy) 10b were produced, and these were changed to heat sinks b1, b2, b3, b4, b5, b6. , B7, b8, b9. Also, using W powder with an average particle size of 0.9 μm and Cu powder with an average particle size of 4.5 μm, the content of Cu powder is 10, 15, 20, 25, 30, 35, 40, 45, 50. In the same manner as described above, heat sinks using the sintered body (high heat dissipation alloy) 10b were prepared, and these heat sinks were c1, c2, c3, c4, c5, c6. It was set as c7, c8, c9.
[0030]
On the other hand, using W powder with an average particle size of 1.5 μm and Cu powder with an average particle size of 1.5 μm, the content of Cu powder is 10, 15, 20, 25, 30, 35, 40, 45, 50. In the same manner as described above, heat sinks using the sintered body (high heat dissipation alloy) 10b were prepared, and these were changed to heat sinks x1, x2, x3, x4, x5, x6. x7, x8, and x9. Further, for comparison, infiltrates were prepared so that the Cu content was 10, 15, and 20% by mass, and these were used as the heat sinks y1, y2, and y3. In the infiltrated material, the Cu content was limited to 20% by mass.
[0031]
Subsequently, the thermal conductivity (W / mK) and the thermal expansion coefficient (ppm / K) of each of these radiator plates a1 to a9, b1 to b9, c1 to c9, x1 to x9, and y1 to y3 were measured. In this case, the thermal conductivity (W / mK) is obtained by a laser flash method using a thermal constant measuring device (manufactured by ULVAC-RIKO, Inc.), and the Cu content (mass%) is plotted on the horizontal axis. When (W / mK) is plotted on the vertical axis, curves a (a1 to a9), b (b1 to b9), c (c1 to c9), x (x1 to x9), y (y1) as shown in FIG. ~ Y3) was obtained.
[0032]
Further, when the thermal expansion coefficient (ppm / K) is obtained by the differential transformer method, the Cu content (% by mass) is plotted on the horizontal axis, and the thermal expansion coefficient (ppm / K) is plotted on the vertical axis, FIG. Curves a (a1 to a9), b (b1 to b9), c (c1 to c9), x (x1 to x9), and y (y1 to y3) as shown were obtained. Then, from these results, when the thermal expansion coefficient (ppm / K) is plotted on the horizontal axis and the thermal conductivity (W / mK) is plotted on the vertical axis, a curve a (a1 to a9) as shown in FIG. , B (b1 to b9), c (c1 to c9), x (x1 to x9), and y (y1 to y3) were obtained.
[0033]
In FIG. 3, it can be seen from the infiltrant that the curve y has the lowest thermal conductivity because a uniform Cu matrix is not formed. When the curve x is compared with the curves a, b, and c, the thermal conductivity improves as the Cu content increases in the curves a, b, and c, but the Cu content in the curve x. It was confirmed that the thermal conductivity does not improve so much when the value exceeds a certain value. This is because, in the curve x, the average particle size of W particles is as large as 1.5 μm, so that sintering at a high temperature is required to increase the sintering density. For this reason, the W particles grow to join together in a state where the W particles are solidified, and the bonded state as shown in FIG. 1D is not formed. Therefore, the channeling effect of Cu is reduced and the thermal conductivity is reduced. Is thought to have declined.
[0034]
On the other hand, in the case of curves a, b, and c, the average particle size of W particles is as small as less than 1.0 μm, and these very fine W particles are bonded to each other as shown in FIG. Heat treatment is performed so as to be in a state. As a result, a W skeleton (a skeleton of a three-dimensional network structure of W) is formed and distributed independently in the Cu matrix. For this reason, it is considered that the thermal conductivity is increased due to the channeling effect of Cu. In this case, in FIG. 3, the curve b is positioned higher than the curve c, and the curve a is positioned higher than the curve b. It can be said that the smaller the), the better the thermal conductivity.
[0035]
Since it is difficult to produce W particles having an average particle size of less than 0.1 μm, it can be said that the average particle size of W particles is preferably 0.1 μm or more and less than 1.0 μm. In this case, the average particle size of the Cu particles needs to be 5 times or more the average particle size of the W particles. However, when the average particle size of the Cu particles is 5 μm or more, the segregation of W in the Cu matrix. Occurs, the structure becomes non-uniform and the thermal conductivity decreases. For this reason, it can be said that it is desirable that the average particle diameter of Cu particles is 5 times or more the average particle diameter of W particles and less than 5 μm. Further, the heat conductivity is improved as the content of Cu particles is increased. However, if the heat conductivity is 250 W / mK (K = 250 W / mK) or more, the heat dissipation plate is sufficiently good in heat conductivity. Therefore, the content of Cu particles is preferably 25% by mass or more.
[0036]
In FIG. 4, when the curve x is compared with the curves a, b, and c, the curve a is located at the bottom, the curves b and c are located above it, and the curve x is located at the top. You can see that This means that the thermal expansion coefficient decreases as the average particle size of the W particles (Cu particles are five times as large). In the curve x, since the average particle size of W particles is as large as 1.5 μm, sintering at a high temperature is necessary, and the W particles grow and join together in a state where the W particles are solidified. ing. For this reason, it is considered that the W skeleton (skeleton of W three-dimensional network structure) was not formed and the thermal expansion coefficient increased.
[0037]
On the other hand, in the curves a, b, and c, since the average particle diameter of W particles is as small as less than 1.0 μm, fine W particles are bonded to each other to form a W skeleton (skeleton of W three-dimensional network structure). Is formed. For this reason, it is considered that the skeleton of the three-dimensional network structure of W acts to regulate the thermal expansion, and the thermal expansion coefficient is lowered. Therefore, it is desirable that the average particle size of W particles is 0.1 μm or more and less than 1.0 μm, and the average particle size of Cu particles is 5 times or more the average particle size of W particles and less than 5 μm. It can be said. Further, the thermal expansion coefficient increases as the content of Cu particles is increased. However, if the thermal expansion coefficient is 12 ppm / K (α = 12 ppm / K) or less, it approximates the thermal expansion coefficient of glass or ceramic. The Cu particle content is desirably 50% by mass or less.
[0038]
In FIG. 5, it can be seen that the curve y is located at the bottom, the curve x is located thereon, and the curves a, b, and c are located thereon. In addition, in the curve x, it can be seen that even if the thermal expansion coefficient increases, the thermal conductivity does not increase accordingly, and the purpose of increasing the thermal conductivity by increasing the amount of Cu added cannot be achieved. .
On the other hand, in curves a, b, and c, it can be seen that the thermal conductivity increases as the coefficient of thermal expansion increases, and the purpose of increasing the thermal conductivity by increasing the amount of Cu added can be achieved.
[0039]
In this case, the curve a is located at the uppermost position, the curve b is located below it, and the curve c is located below it. It can be seen that the purpose of increasing the thermal conductivity by increasing the amount of Cu added can be achieved as the value of ()) decreases. Therefore, it is desirable that the average particle size of W particles is 0.1 μm or more and less than 1.0 μm, and the average particle size of Cu particles is 5 times or more the average particle size of W particles and less than 5 μm. It can be said. And as for content of Cu particle | grains, it is desirable to set it as 50 mass% or less from which thermal conductivity becomes 25 ppm or more which becomes 250 W / mK or more and thermal expansion coefficient becomes 12 ppm / K or less.
[0040]
3. Mounting evaluation of high power high frequency transistor
Next, a high-power high-frequency transistor (with an output of 150 W or more) having a large calorific value was mounted on the heat sink manufactured as described above, and the heat dissipation of this heat sink was evaluated. Therefore, as shown in FIGS. 6 and 7, first, a semiconductor element package 21 is prepared using the heat sink 10 b manufactured as described above, and an alumina circuit board 22 is placed on the package 21 with PbSn solder. 27. Next, a high power high frequency transistor 23 was disposed at a predetermined position on the alumina circuit board 22, and a pair of leads 24, 24 were connected to both sides of the alumina circuit board 22. And the connection part of the alumina circuit board 22 and the connection part of the high power high frequency transistor 23 were connected by wire bonding 25. Thereafter, the high frequency transistor module 20 was fabricated by bonding the cover 26 on these.
[0041]
In this case, a high frequency transistor using a semiconductor device package 21 using a W powder having an average particle size of 0.9 μm and a Cu powder having an average particle size of 4.5 μm and a Cu powder content of 25 mass%. The module 20 was a sample A, and the high-frequency transistor module 20 using the semiconductor element package 21 such that the Cu powder content was 50 mass% was a sample B. Similarly, a high-frequency transistor using a semiconductor device package 21 in which a W powder having an average particle size of 0.5 μm and a Cu powder having an average particle size of 2.5 μm are used and the Cu powder content is 25% by mass. The module 20 was a sample C, and the high-frequency transistor module 20 using the semiconductor element package 21 in which the Cu powder content was 50% by mass was the sample D.
[0042]
Similarly, a high-frequency transistor using a semiconductor device package 21 using a W powder having an average particle size of 0.1 μm and a Cu powder having an average particle size of 0.5 μm and a Cu powder content of 25% by mass. The module 20 was a sample E, and the high-frequency transistor module 20 using the semiconductor element package 21 in which the Cu powder content was 50 mass% was a sample F. Similarly, a high-frequency transistor using a semiconductor device package 21 using W powder having an average particle size of 1.5 μm and Cu powder having an average particle size of 1.5 μm and having a Cu powder content of 25 mass%. The module 20 was a sample X, and the high-frequency transistor module 20 using the semiconductor element package 21 with a Cu powder content of 50 mass% was a sample Y.
[0043]
After that, these samples A, B, C, D, E, F, X, and Y were arranged on the aluminum radiation fins 29 through the heat sheets 28, respectively. Next, after placing them in an atmosphere having an external temperature (Ta) of 25 ° C., the output of the high-power high-frequency transistor 23 is changed from 100 W to 300 W, and in accordance with MIL-STD-883C, an alumina circuit board The temperature (Tj) of the junction between 22 and the high power high frequency transistor 23 was determined. As a result, a result as shown in FIG. 8 was obtained. In this type of high-power high-frequency transistor 23, if the device is used in a state where the junction temperature (Tj) exceeds 150 ° C., the device 23 is destroyed or the life of the device 23 is significantly reduced. It is desirable to dissipate heat so that the junction temperature (Tj) is 130 ° C. or lower.
[0044]
As is clear from the results of FIG. 8, even when the content of the Cu powder is 50% by mass as in the sample Y, when the output exceeds 150 W, the junction temperature (Tj) becomes 150 ° C. or higher. The element 23 and the alumina circuit board 22 were damaged by the stress generated from the difference in thermal expansion coefficient between the element 23 and the package 21. On the other hand, in sample A, even when the Cu powder content is 25 mass% (0.9 μmW, 4.5 μmCu), the junction temperature (Tj) is maintained at 130 ° C. or lower until the output of the element 23 is 200 W. did it. In Sample B with a Cu powder content of 50 mass% (0.9 μmW, 4.5 μmCu) and Sample C with a Cu powder content of 25 mass% (0.5 μmW, 2.5 μmCu), the element The junction temperature (Tj) could be maintained at 130 ° C. or lower until the output of 23 was 300 W.
[0045]
In the sample D having a Cu powder content of 50% by mass (0.5 μmW, 2.5 μmCu) and the sample E having a Cu powder content of 25% by mass (0.1 μmW, 0.5 μmCu), the element The junction temperature (Tj) could be maintained at 120 ° C. or lower until the output of 23 was 300 W. Furthermore, in the sample F having a Cu powder content of 50 mass% (0.1 μmW, 0.5 μmCu), the junction temperature (Tj) could be maintained at 110 ° C. or lower until the output of the element 23 was 300 W.
[0046]
【The invention's effect】
As described above, in the present invention, since the tungsten particles 12 have a three-dimensional network structure in which the tungsten particles 12 are continuously bonded, even if the content of the copper 11 is increased to 25 to 50% by mass, The expansion coefficient can be maintained at 12 ppm / K or less. Then, a skeleton having a three-dimensional network structure in which tungsten particles 12 having an average particle diameter of less than 1.0 μm are bonded to each other is formed, and the skeleton is 5 times or more the average particle diameter of these particles and less than 5.0 μm. The copper powder 11 is distributed in the molten matrix. Thereby, the channeling effect of copper can be exhibited and the tungsten-copper alloy excellent in thermal conductivity can be obtained.
[0047]
In the above-described embodiment, the WCu alloy has been mainly described, but the MoCu alloy is the same as that of the WCu alloy. In the above-described embodiment, an example in which a high-power high-frequency transistor is used as a semiconductor element has been described. However, it is obvious that the semiconductor element can be applied to other semiconductor elements such as a semiconductor laser.
[Brief description of the drawings]
FIG. 1 is a cross-sectional view schematically showing a process of obtaining a high heat dissipation alloy of the present invention.
FIG. 2 is a diagram showing a temperature rising pattern in a binder removal processing step and a sintering step.
FIG. 3 is a graph showing the relationship of thermal conductivity with respect to the Cu content (Cu concentration) of a WCu alloy.
FIG. 4 is a diagram showing the relationship of the coefficient of thermal expansion to the Cu content (Cu concentration) of a WCu alloy.
FIG. 5 is a diagram showing the relationship of thermal conductivity to the thermal expansion coefficient of a WCu alloy.
FIG. 6 is a perspective view showing a state in which a high frequency transistor is arranged on a heat sink using a WCu alloy to form a package for a semiconductor device.
7 is a diagram schematically showing a state in which a cover is attached to the semiconductor element package of FIG. 6 and a heat dissipation test is performed, and FIG. 7A is a view showing that the cover is attached to the semiconductor element package of FIG. 6; It is a figure which shows the AA cross section of FIG. 6 in a state, FIG.7 (b) is sectional drawing which shows the state which attached the radiation fin to the lower part of the package of Fig.7 (a).
FIG. 8 is a graph showing the relationship between the output (W) of a high-frequency transistor and the junction temperature (Tj) between the high-frequency transistor and a ceramic substrate.
FIG. 9 is a cross-sectional view schematically showing a WCu alloy made of a conventional infiltrant.
FIG. 10 is a cross-sectional view schematically showing a conventional WCu alloy by a powder method.
FIG. 11 is a cross-sectional view schematically showing a production process of a WCu alloy by a powder method of a conventional example when the content of copper is large.
FIG. 12 is a cross-sectional view schematically showing a production process of a WCu alloy by a powder method of a conventional example when the content of copper is low.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 10 ... Mixed powder body, 10a ... Skeleton, 10b ... Heat sink, 11 ... Copper (Cu) powder, 12 ... Tungsten (W) powder, 20 ... High frequency transistor module, 21 ... Semiconductor device package, 22 ... Alumina circuit board, 23 ... High-power high-frequency transistor, 24 ... Lead, 25 ... Wire bonding, 26 ... Cover, 27 ... Solder, 28 ... Heat sheet, 29 ... Aluminum radiating fin

Claims (6)

A heat-dissipating alloy comprising a tungsten-copper alloy or a molybdenum-copper alloy containing 25 to 50% by mass of copper,
A skeleton of a three-dimensional network structure in which tungsten particles or molybdenum particles are continuously bonded is formed, and copper is filled in these skeletons.
The average particle size of the raw material powder of the tungsten particles or molybdenum particles is 0.1 μm or more and less than 1.0 μm,
The copper raw material powder has an average particle size of not less than 5 times the average particle size of the tungsten particle or molybdenum particle raw material powder and less than 5.0 μm ,
A skeleton having a three-dimensional network structure in which the tungsten particles or the molybdenum particles are continuously bonded so that the cross-sectional length of the bonded portion of the tungsten particles or the molybdenum particles is ½ or less of the particle size of these particles. A heat-dissipating alloy characterized in that is formed . A heat sink made of a metal material,
The heat dissipation plate according to claim 1 , wherein the metal material is the high heat dissipation alloy according to claim 1 . A package for a semiconductor device using a heat sink,
3. The semiconductor device package according to claim 2 , wherein the heat radiating plate is a heat radiating plate according to claim 2 , wherein a high frequency transistor having an output of 150 W or more is disposed on the heat radiating plate. A method for producing a high heat dissipation alloy comprising a tungsten-copper alloy or a molybdenum-copper alloy containing 25 to 50% by mass of copper,
Mix tungsten fine powder with copper fine powder with a larger average particle diameter, or molybdenum fine powder with copper fine powder with a larger average particle diameter, and cover the surface of the copper fine powder with tungsten fine powder or molybdenum fine powder. A mixing step to obtain a mixed mixture;
The mixture is heated at a temperature equal to or lower than the melting point of copper to start sintering of the tungsten fine powders or the molybdenum fine powders , and the cross-sectional length of the bonded portion of the tungsten particles or molybdenum particles is equal to the particle size of these particles. A skeleton forming step of adjusting the sintering temperature and the sintering time to be ½ or less ,
A method for producing a highly heat-dissipating alloy, comprising: a sintering step of melting copper by heating at a temperature not lower than the melting point of copper and not higher than the melting points of tungsten and molybdenum after the skeleton forming step. 5. The high heat dissipation alloy according to claim 4 , wherein in the mixing step, the tungsten fine powder and the copper fine powder or the molybdenum fine powder and the copper fine powder are mixed so as not to be accompanied by a mechanical alloy. Method. A method for producing a heat dissipation plate made of a metal material comprising a step of sintering after filling a mixed powder into a plate-shaped mold and forming into a flat plate shape by compaction press molding,
A method for manufacturing a heat sink, wherein the metal material is formed by the method for manufacturing a high heat dissipation alloy according to claim 4 or 5 .

Images