JP7418779B2 - Electrode materials for electrical discharge machining or heat sinks for semiconductors, and methods of manufacturing them - Google Patents

Description

The present invention relates to an electrode material for electric discharge machining or a heat sink for a semiconductor, and a method for manufacturing the same.

Electrical discharge machining is a machining method in which a pulsed arc discharge is generated between an electrode material and a workpiece material, and the high heat and shock waves generated during this process repeatedly melt and remove the workpiece material. It is. Therefore, ideal electrical discharge machining is performed under machining conditions that suppress the amount of melting of the electrode material itself in order to suppress consumption of the electrode material itself. Furthermore, since complex shapes are machined in electric discharge machining, machinability of the electrode material itself is also important.

Electrode material properties suitable for these conditions include higher thermal conductivity and electrical conductivity than the material to be cut, as well as high arc wear resistance and machinability. Examples of practical materials include silver-tungsten alloys and copper-tungsten alloys, which are a combination of silver and copper, which have high thermal conductivity and electrical conductivity, and tungsten, which has the highest melting point among practical metals. Generally, expensive silver-tungsten alloys are avoided and copper-tungsten alloys are often used.

Arc wear resistance is low for silver and copper, but high for tungsten in a non-oxidizing atmosphere. On the other hand, tungsten is a brittle material that is difficult to cut, whereas silver and copper are highly malleable and easy to process. For these reasons, copper-tungsten alloys containing tungsten and copper in an amount of 30 to 35% by mass of the total mass of the alloy are often used in electrical discharge machining electrodes.

However, when a copper-tungsten alloy is used for the electrode, although it is excellent in terms of electrode wear, the processing speed is not fast. To increase the machining speed, it is necessary to increase the arc discharge energy per pulse and increase the number of pulses per unit time, but when performing electric discharge machining under these conditions, an abnormality called concentrated discharge may occur. Electric discharge occurs, which not only makes it impossible to obtain high machining accuracy but also reduces machining speed.

In other words, in electrical discharge machining, surface machining only progresses when uniformly distributed electrical discharge is generated on the machined surface, but if the electrical discharge is uneven, machining will occur in areas where electrical discharge is concentrated, but in areas where it is not. Since the machining does not proceed, the machining process does not proceed as only holes are formed locally on the entire machined surface.

To improve this problem, the copper-tungsten alloy contains additives such as barium oxide (BaO), which is a low work function compound, to lower the work function of the electrode material and improve arc starting performance. There is also known a material that suppresses concentrated discharge and increases machining speed by doing so (see, for example, Patent Document 1).

Copper-tungsten alloys are used not only as electrodes for electrical discharge machining as described above, but also as heat sinks for semiconductors, contacts, etc. In semiconductor heat sinks, the tungsten typically contains 10-20% by weight of copper based on the total weight of the alloy, and in contacts, the tungsten typically contains 20-40% by weight of copper based on the total weight of the alloy.

Furthermore, while the melting point of tungsten is 3400°C, the melting point of copper is as low as 1083°C, and the sintering temperature at which the copper-tungsten alloy is densified by sintering is higher than the melting point of copper. Therefore, the manufacturing process for copper-tungsten alloys is usually to create a tungsten porous body (tungsten skeleton) with a predetermined porosity, and then melt and infiltrate it with copper (infiltration method). There are many things. However, in the case of copper-tungsten alloys containing 30% by mass or more of copper, handling of the porous tungsten material is difficult because the porosity has reached a certain level at the time of molding. A problem arises. In such cases, copper powder is mixed with tungsten powder in an amount of 5 to 10% by mass of the total mass to create a copper-tungsten porous body, and copper is melted and infiltrated into the porous body (partial infiltration method). There are many things to do. It can also be manufactured by mixing predetermined amounts of tungsten powder and copper powder and performing liquid phase sintering (mixing method). However, although copper and tungsten have good wettability, they do not form a solid solution with each other, so densification by a dissolution-precipitation mechanism (Ostwald growth) does not occur, and therefore tungsten particles hardly become coarse.

In the case of copper-tungsten alloys containing 10 to 20 mass% copper in tungsten, heating the tungsten porous body to a predetermined porosity causes the tungsten crystal grains to firmly adhere to each other, resulting in tungsten crystals. The formation of a skeletal structure between grains contributes to improving the mechanical strength of the copper-tungsten alloy and suppressing the coefficient of linear expansion. In this case, the tungsten powder used is often doped with 0.1-0.3% by mass of nickel in liquid form with a soluble compound such as nickel nitrate. Nickel, which is fixed or solid-solved on the surface of tungsten particles after removing nitric acid groups through thermal decomposition, activates the tungsten particles and shrinks them at a relatively low temperature (activation sintering) to achieve a desired porosity economically. Not only can copper and nickel form a solid solution, but it also contributes to improving the wettability when infiltrating copper into porous tungsten materials, reducing internal defects such as pores and poor infiltration. This results in a significant decrease in thermal conductivity and electrical conductivity.

By the way, the copper-tungsten alloy used in electrodes for electric discharge machining contains 30 to 35% by mass of copper, so it is produced by a partial infiltration method or a mixing method. When manufactured using these methods, tungsten crystal grains hardly stick to each other, resulting in a structure in which the surfaces of the tungsten crystal grains are covered with molten copper. In this case, the mechanical strength of the copper-tungsten alloy not only depends on the adhesion strength between copper crystal grains, but also the binding force due to the adhesion of tungsten crystal grains, which are a material with a low coefficient of linear expansion, is reduced. , the contribution of copper crystal grains, which is a material with a high coefficient of linear expansion, increases. As a result, when a copper-tungsten alloy is used as an electrode for electrical discharge machining, it undergoes a thermal cycle in which it is heated and cooled in short cycles due to pulsed arc discharge, resulting in repeated rapid expansion and contraction, which shortens the life of the electrode. This was one of the reasons for the reduction.

On the other hand, the characteristics required of a copper-tungsten alloy used in a semiconductor heat sink are high thermal conductivity and a coefficient of linear expansion matching that of a ceramic package. High thermal conductivity is easily achieved by increasing the copper content in copper-tungsten alloys, but adding too much copper, a material with a high coefficient of linear expansion, to tungsten, a material with a low coefficient of linear expansion, results in a failure in ceramic packages. Matching with the coefficient of linear expansion becomes impossible.

In other words, it is desirable that the linear expansion coefficient required for the electrode material for electrical discharge machining be as small as possible, but it is important that the linear expansion coefficient required for the semiconductor heat sink material matches the ceramic package to be brazed.

It is said that the change in the coefficient of linear expansion in a composite material is based on a composite law that depends on the content ratio of the contained components. However, in the case of a copper-tungsten alloy, as the copper content increases, the adhesion of tungsten crystal grains to each other decreases, and the restraining force that suppresses copper's high coefficient of linear expansion rapidly decreases. This is because the tungsten crystal grains and copper crystal grains that make up the copper-tungsten alloy do not form a solid solution with each other, but are simply composited together. This is because it covers the Heating when making a tungsten porous body or a copper-tungsten porous body with a predetermined porosity promotes the fixation of tungsten crystal grains, but in order to contain a large amount of copper, a large porosity is required. Since this is necessary, it results in a decrease in fixing strength due to a decrease in heating temperature.

Looking at the elongation rates of copper-tungsten materials produced by conventional infiltration methods, partial infiltration methods, and mixed methods, the elongation rates change linearly below 500℃, but above 500℃, the elongation rates change to the right. A tendency to increase gradually can be seen. This shows that there is a deviation from the ideal compound law in the temperature range of 500°C or higher. This phenomenon is thought to be due to the fact that at temperatures above 500°C, the restraining force of tungsten crystal grains decreases, and the contribution of copper, which is a material with a high coefficient of linear expansion, to the elongation rate increases.

In other words, a positive deviation from the composite law in the temperature range of 500°C or higher in copper-tungsten materials produced by conventional methods shortens the life of the electrode material for electrical discharge machining due to the promotion of expansion-contraction behavior. Furthermore, in order to match the coefficient of linear expansion between copper-tungsten alloys used for semiconductor heat sinks and ceramic packages, this means that it is necessary to incorporate further reductions in copper content, as required by the composite rule, into material design. There is.

Furthermore, if the content of nickel, which reduces thermal conductivity, is eliminated, extremely high molding pressure and sintering temperature are required to obtain a porous tungsten body with a predetermined porosity, and copper infiltration becomes necessary. Internal defects due to defects tend to remain.

Special Publication No. 35-8046

In recent years, especially regarding electrode materials for electrical discharge machining, there has been a demand for not only increasing processing speed but also further reducing electrode wear from the viewpoint of electrode replacement costs due to the automation of electrical discharge machines. It cannot be said that this method has sufficiently improved the amount of electrode material consumed and the processing speed. On the other hand, with regard to semiconductor heat sinks, there has been a demand for higher thermal conductivity in order to cope with the increase in heat generation due to higher integration of semiconductors. In addition, if it is possible to improve the structure of a copper-tungsten alloy containing a large amount of copper (for example, 30% by mass or more), which has a high thermal conductivity and a large coefficient of linear expansion, and to suppress an excessively large coefficient of linear expansion, It can be applied not only to electrodes for electrical discharge machining, but also to semiconductor heat sinks, contacts, etc.

The present invention is an attempt to solve the above-mentioned problems, and includes an electrode material for electric discharge machining that suppresses material consumption and improves machining speed, or a material that has high thermal conductivity by containing more copper than before. An object of the present invention is to provide a heat sink for a semiconductor that has a suppressed coefficient of linear expansion despite having a high coefficient of thermal expansion.

As a result of extensive research, the present inventors have discovered that the tungsten crystal is composed of 15-45% by mass of copper, 0.3-3% by mass of alkaline earth metal oxide, and the balance is tungsten, and firmly adheres between tungsten crystal grains. By using a copper-tungsten alloy whose coefficient of linear expansion at 800℃ is smaller than that at 500℃, we have created an electrode material for electrical discharge machining that reduces material consumption and improves machining speed. It has been found that by containing a large amount of copper, it is possible to obtain a semiconductor heat sink with a suppressed coefficient of linear expansion despite having high thermal conductivity. The present invention has been completed based on such knowledge through further research. That is, the present invention includes the following configurations.

Item 1. An electrode material for electric discharge machining or a heat sink for semiconductors made of a copper-tungsten alloy,
The copper-tungsten alloy is a copper-tungsten alloy containing 15 to 45% by mass of copper and 55 to 85% by mass of tungsten, with the total amount of the copper-tungsten alloy being 100% by mass,
The copper-tungsten alloy has tungsten crystal grains having an average grain size of 0.6 to 16 μm,
The copper-tungsten alloy is an electrode material for electric discharge machining or a heat sink for semiconductors, and has a linear expansion coefficient at 800°C that is smaller than a linear expansion coefficient at 500°C.
Item 2. Item 2. The electrode material for electrical discharge machining or the heat sink for semiconductors according to Item 1, wherein the copper-tungsten alloy has tungsten crystal grains firmly fixed to each other.
Item 3. The discharge according to item 1 or 2, in which the shrinkage rate after heat treatment is 0.05% or less when subjected to a heat treatment test in which the temperature is raised to 900°C at a rate of 10°C/min and then naturally cooled to room temperature. Electrode material for processing or heat sink for semiconductors.
Item 4. An electrode material for electrical discharge machining made of a copper-tungsten alloy,
The copper-tungsten alloy contains 15 to 45 mass % copper, 54.7 to 82 mass % tungsten, and 0.3 to 3 mass % alkaline earth metal, with the total amount of the copper-tungsten alloy being 100 mass %. It is a copper-tungsten alloy containing oxides,
The copper-tungsten alloy has tungsten crystal grains having an average grain size of 2 to 16 μm,
The copper-tungsten alloy is an electrode material for electrical discharge machining, and has a linear expansion coefficient at 800°C that is smaller than that at 500°C.
Item 5. 5. The electrode material for electric discharge machining according to item 4, wherein the copper-tungsten alloy has strong adhesion of tungsten crystal grains to each other.
Item 6. The discharge according to item 4 or 5, in which the shrinkage rate after heat treatment is 0.05% or less when subjected to a heat treatment test in which the temperature is raised to 900°C at a rate of 10°C/min and then naturally cooled to room temperature. Electrode material for processing.
Section 7. A method for manufacturing an electrode material for electric discharge machining or a heat sink for a semiconductor according to any one of items 1 to 3, comprising:
A process of mixing raw material powder containing copper powder and tungsten powder and subjecting it to discharge plasma sintering under solid phase sintering conditions,
A manufacturing method, wherein the tungsten powder has an average particle size of 0.6 to 16 μm.
Section 8. Item 7, wherein in the raw material powder, the content of the copper powder is 15 to 45% by mass, and the content of the tungsten powder is 55 to 85% by mass, with the total amount of the raw material powder being 100% by mass. Production method.
Item 9. A method for producing an electrode material for electric discharge machining according to any one of items 4 to 6, comprising:
A step of mixing raw material powder containing copper powder, tungsten powder, and alkaline earth metal tungstate powder and subjecting it to discharge plasma sintering under solid phase sintering conditions,
A manufacturing method, wherein the tungsten powder has an average particle size of 2 to 16 μm.
Item 10. In the raw material powder, the content of the copper powder is 15 to 45% by mass, the content of the tungsten powder is 54.7 to 82% by mass, and the content of the alkaline tungstate is 15 to 45% by mass, with the total amount of the raw material powder being 100% by mass. Item 9. The manufacturing method according to Item 9, wherein the content of the similar metal powder is 0.3 to 3% by mass.

According to the present invention, an electrode material for electric discharge machining that suppresses material consumption and improves machining speed, or contains a larger amount of copper than before, suppresses the coefficient of linear expansion despite having high thermal conductivity. It is possible to provide a heat sink for a semiconductor with a high temperature.

Composition images based on backscattered electron images of crystal structures of samples of Example 1 and Comparative Examples 1 to 3 (500x and 2000x), results of elemental analysis using energy dispersive X-ray analysis, and Example 1 and Comparative Examples 1 to 3. These are composition images based on backscattered electron images of electron micrographs (2000x and 5000x) of fractured surfaces of samples of Comparative Examples 1 and 3. These are composition images based on backscattered electron images of electron micrographs (500x and 2000x) of crystal structures of samples of Examples 2, 4, and 6, and the results of elemental analysis using energy dispersive X-ray analysis. 2 is a graph showing linear expansion coefficients of samples of Example 1 and Comparative Example 1. 3 is a graph showing linear expansion coefficients of samples of Examples 2, 4, and 6 and Comparative Examples 4, 5, and 7. It is a graph showing the relationship between Cu composition and linear expansion coefficient when SPS treatment is performed and temperature is changed. 3 is a graph showing the relationship between temperature and elongation rate of samples of Examples 2, 4, and 6 and Comparative Examples 4 to 7. 2 is a graph showing the relationship between temperature and elongation rate of samples of Example 2 and Comparative Examples 4 and 5. 3 is a graph showing the relationship between temperature and elongation rate of samples of Example 6 and Comparative Examples 4 to 6.

In this specification, "contain" is a concept that includes all of "comprise," "consist essentially of," and "consist of." Furthermore, in this specification, when a numerical range is expressed as A to B, it indicates a range of A or more and B or less.

1. Electrode material for electric discharge machining or heat sink for semiconductors The electrode material for electric discharge machining or the heat sink for semiconductors in the first aspect of the present invention (hereinafter sometimes simply referred to as "electrode material or heat sink") is made of copper-tungsten alloy. An electrode material for electrical discharge machining or a heat sink for semiconductors, wherein the copper-tungsten alloy contains 15-45% by mass of copper and 55-85% by mass, with the total amount of the copper-tungsten-based alloy being 100% by mass. It is a copper-tungsten alloy containing tungsten, the copper-tungsten alloy has tungsten crystal grains having an average grain size of 0.6 to 16 μm, and the copper-tungsten alloy has a coefficient of linear expansion at 800°C. It is smaller than the linear expansion coefficient at 500℃.

Further, the electrode material for electrical discharge machining in the second aspect of the present invention is an electrode material for electrical discharge machining made of a copper-tungsten alloy, wherein the copper-tungsten alloy has a total amount of the copper-tungsten alloy. It is a copper-tungsten alloy containing 15 to 45 mass % copper, 54.7 to 82 mass % tungsten, and 0.3 to 3 mass % alkaline earth metal oxide, and the copper-tungsten The average grain size of tungsten crystal grains in the copper-tungsten alloy is 2 to 16 μm, and the coefficient of linear expansion of the copper-tungsten alloy at 800°C is smaller than that at 500°C.

In electrode materials for electric discharge machining, the copper content is generally about 30 to 35% by mass. The purpose of this copper content is to ensure electrical conductivity and thermal conductivity higher than that of the workpiece material, as well as machinability of the electrode material itself, but it also causes a decrease in arc wear resistance. According to the present invention, for example, according to the manufacturing method described below, plastic deformation of tungsten particles occurs at the same time as necking between tungsten particles occurs during sintering, and a three-dimensional network structure of tungsten crystal grains with high arc wear resistance is created. It is formed. During electric discharge machining, the surface of the electrode material is expanded and contracted due to intense thermal cycles due to pulsed arc discharge, and shock waves due to arc discharge preferentially consume copper, which has a low melting point, high coefficient of linear expansion, and is highly resistant to arc consumption. At the same time, since the tungsten crystal grains are not tightly fixed to each other, the tungsten grains may be shed, resulting in large consumption. However, in the electrode material of the present invention, tungsten crystal grains, which have a high melting point, a low linear expansion coefficient, and good arc wear resistance, are firmly attached to each other to form a three-dimensional network structure, which suppresses electrode wear. At the same time, it became possible to increase the processing speed. The same holds true when such a copper-tungsten alloy is used as a semiconductor heat sink. In the case of semiconductor heat sinks, instead of arc wear resistance, it is important to reduce the coefficient of linear expansion so that it matches the coefficient of linear expansion of the ceramic package. Since tungsten crystal grains with a small coefficient of linear expansion are firmly attached to each other to form a three-dimensional network structure, the binding force of the copper crystal grains with a large coefficient of linear expansion is improved, making it possible to improve the copper-tungsten system for semiconductor heat sinks. It becomes possible to contain more copper in the alloy than before. As a result, it becomes possible to create a semiconductor heat sink with higher thermal conductivity than before. Therefore, in the present invention, it is possible to cause strong adhesion between tungsten crystal grains over a wider range of copper content than before, and assuming that the total amount of copper-tungsten alloy is 100% by mass, the copper content is The amount is 15 to 45% by weight, preferably 20 to 40% by weight. When the copper content is less than 15% by mass, strong adhesion between tungsten particles becomes dominant, but the formation of a strong tungsten skeleton structure inhibits plastic deformation, making densification difficult, and the thermal conductivity decreases. , the electrical conductivity is inferior to that of the workpiece material, and the machinability decreases, making it difficult to process the electrode material itself. In addition, if the copper content exceeds 45% by mass, the binding force of the copper crystal grains decreases due to a decrease in the adhesion of tungsten crystal grains to each other, and the arc wear resistance of the electrode material decreases, making it difficult to suppress electrode wear. In addition to this being impossible, the coefficient of linear expansion increases rapidly.

In the copper-tungsten alloy contained in the electrode material of the present invention, the tungsten content is, for example, 55 to 85% when the total amount of the copper-tungsten alloy is 100% by mass when no alkaline earth metal oxide is included. % by mass, preferably 60-80% by mass. If the tungsten content is less than 55% by mass, material consumption cannot be suppressed. Furthermore, if the tungsten content exceeds 85% by mass, the thermal conductivity, electrical conductivity, etc. will be poor. On the other hand, when an alkaline earth metal oxide is included, the content is 54.7 to 82% by mass, preferably 59.5 to 77.5% by mass, based on the total amount of copper-tungsten alloy as 100% by mass. If the tungsten content is less than 54.7% by mass, material consumption cannot be suppressed. Furthermore, if the tungsten content exceeds 82% by mass, the thermal conductivity, electrical conductivity, etc. will be poor.

In the present invention, when a copper-tungsten alloy is used as an electrode material, the copper-tungsten alloy may contain an alkaline earth metal oxide, which is a low work function compound. The alkaline earth metal oxide is not particularly limited and includes magnesium oxide, calcium oxide, strontium oxide, barium oxide, and the like. These alkaline earth metal oxides can be used alone or in combination of two or more.

In the copper-tungsten alloy contained in the electrode material of the present invention, the content of alkaline earth metal oxide is 0.3% by mass when the total amount of copper-tungsten alloy is 100% by mass from the viewpoint of processing speed and material consumption. -3% by mass is preferred, and 0.5-2.5% by mass is more preferred. Since alkaline earth metal oxides are low work function compounds, they have excellent thermionic emission properties, but because they are insulators, they do not conduct electricity, and because they are oxides, they have lower thermal conductivity than metals. Furthermore, excessive inclusion of alkaline earth metal oxides in the electrode material causes an increase in electrical resistance, and the electrode material itself generates heat due to Joule heat, promoting electrode wear.

In the present invention, when a copper-tungsten alloy is used as a heat sink, the copper-tungsten alloy may contain an iron group metal in order to improve the wettability of the tungsten crystal grain surface and densification during sintering. You can also do it. The iron group metal is not particularly limited and includes iron, cobalt, nickel, and the like. These iron group metals can be used alone or in combination of two or more.

In the copper-tungsten alloy contained in the heat sink of the present invention, from the viewpoint of wettability and densification during sintering, the content of iron group metal is 0.05 to 0.05% by mass, assuming the total amount of copper-tungsten alloy is 100% by mass. 0.5% by mass is preferred, and 0.1 to 0.3% by mass is more preferred.

Iron group metals can be added as metal powder by mixing them together with tungsten powder or copper powder, or by doping tungsten powder in the form of liquid nitrate dissolved in alcohol, etc., and heating it in a hydrogen atmosphere. It may also be contained as a doped tungsten powder for producing a tungsten skeleton for infiltration by decomposing and reducing nitric acid groups.

Among iron group metals, when nickel is added, densification is particularly improved by activated sintering of tungsten.

On the other hand, added iron group metals are expected to have the effect of reducing internal defects by improving the wettability of copper on the surface of tungsten crystal grains and improving densification during sintering, but they tend to reduce thermal conductivity and electrical conductivity. There is.

According to the present invention, it is possible to obtain a dense copper-tungsten alloy with no internal defects without the addition of iron group metals, and even without the addition of iron group metals, it is possible to obtain higher thermal conductivity and electrical conductivity. It is possible to achieve this.

Note that the copper-tungsten alloy included in the electrode material or heat sink of the present invention may also contain inevitable metal elements in addition to the above-mentioned copper, tungsten, alkaline earth metal oxides, and iron group metals. Examples of such unavoidable metal elements include chromium, molybdenum, aluminum, and the like. These inevitable metal elements may be contained alone or in combination of two or more. Even if these unavoidable metal elements are contained in small amounts, they do not affect the wear rate and processing speed of the electrode material, as well as the thermal conductivity and linear expansion coefficient of the heat sink, and must be contained within the range that produces the effects of the present invention. I can do it. Specifically, the total amount of copper-tungsten alloy is 100% by mass, the content of copper is 15 to 45% by mass, the content of alkaline earth metal oxides is 0.3 to 3% by mass, and the content of iron group metals. It is preferable that the amount is 0.05 to 0.5% by mass, and the balance is tungsten and unavoidable metal elements.

The copper-tungsten alloy contained in the electrode material or heat sink of the present invention is preferably dense and has tungsten crystal grains firmly fixed to each other. If internal defects such as pores remain in the electrode material, the internal defects such as pores will be transferred to the workpiece material when die carving is performed by electrical discharge machining. The presence of internal defects such as pores will lead to a decrease in thermal conductivity and electrical conductivity, and if they are exposed on the surface of the product, when plating is applied to the surface of the heat sink, the plating solution that has soaked into the pores will be removed when drying. It may appear as a stain or become a starting point for corrosion.

Among the crystal grains of each metal element, tungsten crystal grains are preferably large in the electrode material. Since the tungsten crystal grains are large, the tungsten crystal grains are firmly fixed to each other, and it is possible to suppress shedding of the tungsten crystal grains during electrical discharge machining, thereby suppressing consumption or linear expansion coefficient of the electrode material. This is because the mechanical strength of tungsten crystal grains is higher in the grain than in the grain boundaries, and the larger the tungsten crystal grains, the greater the mean free path of the copper phase. This is because from this viewpoint, an improvement in thermal conductivity can be expected due to an improvement in thermal diffusivity. Incidentally, if the material is not densified, the thermal conductivity will decrease significantly due to a decrease in thermal diffusivity due to internal defects and a decrease in density. Generally, specific heat capacity, thermal diffusivity, and thermal conductivity are called three thermal constants, and thermal conductivity is calculated by multiplying specific heat capacity, thermal diffusivity, and density. Like the linear expansion coefficient, the specific heat capacity is determined by a compound law based on the material composition, but the thermal diffusivity is greatly affected by internal defects and the formation of solid solutions. Note that when the tungsten crystal grains are small, densification is promoted, but the contact area between the tungsten crystal grains decreases, and the adhesion strength tends to decrease. In the case of general continuous arc discharge, the discharge portion melts to form a molten pool, and the electrode material is subject to constant arc wear resistance based on the composition of the electrode material itself. On the other hand, in the case of electric discharge machining, arc starts are repeated many times within a short period of time because the workpiece is machined by pulsed arc discharge. In the case of such an unsteady discharge form, the load is applied to the discharge starting point rather than the formation of a molten pool, and if the bond between crystal grains is weak, electrode wear due to shedding of crystal grains along grain boundaries becomes dominant, and the electrode Material consumption rate increases. From this point of view, the average grain size of tungsten crystal grains in the electrode material for electric discharge machining is preferably 2 to 16 μm, more preferably 3 to 10 μm, and even more preferably 4 to 8 μm. In addition, in order to further optimize the relationship between electrode life and machinability, it is also possible to adjust the size of the tungsten crystal grains as appropriate. On the other hand, in the case of semiconductor heat sink applications, densification is prioritized since there is no need to consider prevention of shedding of tungsten crystal grains, and the average grain size of tungsten crystal grains may be 0.6 to 2 μm. That is, for use as a heat sink for semiconductors, the average grain size of tungsten crystal grains is preferably 0.6 to 16 μm, and for use as an electrode material for electrical discharge machining, the average grain size of tungsten crystal grains is preferably 2 to 16 μm. Note that the average grain size of the tungsten crystal grains is measured based on a composition image of a backscattered electron image obtained by observing the electrode material of the present invention with an electron microscope (SEM).

Note that the sizes of other metal crystal grains, that is, copper crystal grains, alkaline earth metal oxide crystal grains, and iron group metals are not particularly limited. However, from the viewpoint of obtaining homogeneous dispersibility without segregation, the average grain size of copper crystal grains is preferably 25 to 50 μm, and the average grain size of alkaline earth metal oxide crystal grains is preferably 0.1 to 200 μm, and 50 to 50 μm. It can also be 150 μm. Note that the average particle diameters of the copper crystal grains and the alkaline earth metal oxide crystal grains are measured based on a composition image of a backscattered electron image obtained by observing the electrode material of the present invention with an electron microscope (SEM). On the other hand, iron group metals are often contained in the form of liquid dope with nitrates, and are homogeneously dispersed or solidly dissolved on the surface of tungsten powder.

In the electrode material or heat sink of the present invention, it is preferable that metal crystal grains, especially tungsten crystal grains, form necking and firmly adhere to each other. In this way, since the tungsten crystal grains are firmly fixed to each other, the tungsten crystal grains and/or copper crystal grains are prevented from falling out due to thermionic emission, shock waves, and thermal expansion that occur at the time of arc start, and the electrode Material consumption can be suppressed. Normally, copper-tungsten alloys used as electrode materials for electrical discharge machining contain 30 to 35% by mass of copper in tungsten. Since there are many powder particles around, adhesion between tungsten powder particles is rare, and when the copper powder particles are melted by heating during sintering, they cover the surfaces of the tungsten particles and inhibit the formation of necking between the tungsten particles. On the other hand, in the copper-tungsten alloy of the present invention, tungsten crystal grains are firmly fixed to each other.

Comparing the two in terms of coefficient of linear expansion and temperature, in the case of conventional copper-tungsten alloys, the restraint on copper particles with a large coefficient of linear expansion is weak due to inhibition of network formation by tungsten particles, which are substances with a low coefficient of linear expansion, and the heating temperature is low. As the temperature increases, the coefficient of linear expansion of the copper-tungsten alloy increases upwardly at 500°C or higher due to copper, which is a material with a high coefficient of linear expansion. On the other hand, in the case of the copper-tungsten alloy included in the electrode material or heat sink of the present invention, the strong three-dimensional network structure of tungsten crystal grains with a low coefficient of linear expansion strongly restrains copper particles with a large coefficient of linear expansion. , suppressing the increase in the coefficient of linear expansion of copper-tungsten alloy. As a result, the copper-tungsten alloy contained in the electrode material or heat sink of the present invention is a material whose linear expansion coefficient at 800°C is smaller than that at 500°C. Specifically, the linear expansion coefficient at 800°C is preferably 80 to 99% (particularly 85 to 98%) of the linear expansion coefficient at 500°C. In addition, in the region below 500° C., a composite law of linear expansion coefficient based on the content ratio of tungsten particles and copper particles generally holds.

In electrical discharge machining electrodes, machining is performed by generating a pulsed arc discharge on the electrode surface, so if the bonding strength of metal crystal grains is weak due to rapid heating at the start of the arc, grains will fall off. In this case, the copper crystal grains on the electrode surface undergo rapid thermal expansion and melting, and because they have low resistance to arc consumption, they are preferentially consumed. Tungsten crystal grains, which have a high melting point, low coefficient of linear expansion, and excellent arc wear resistance, suppress electrode wear, but if the tungsten crystal grains are not firmly fixed to each other, they easily fall off. Further, since the necking portion of the tungsten crystal grain, that is, the tungsten crystal grain boundary, has a weaker bonding force than the inside of the tungsten grain, it is advantageous to use coarse-grained tungsten particles.

Furthermore, the inside of the surface layer where arc discharge occurs is also heated by Joule heat . In this case, if the tungsten crystal grains are not firmly fixed, the tungsten crystal grain boundaries will be destroyed by the copper crystal grains having a large coefficient of linear expansion, promoting electrode wear.

On the other hand, a heat sink is assembled together with a semiconductor into a ceramic package and soldered using silver solder or the like. At this time, if the linear expansion coefficients of the heat sink and the ceramic package are not matched, brazing cannot be performed. The melting point of silver is 961.8°C, and the melting point of silver solder is lower, around 700-800°C. Matching of the linear expansion coefficients of the heat sink and the ceramic package is required up to the brazing temperature at which silver is soldered.

There are two terms for thermal expansion: linear expansion coefficient (also called thermal expansion coefficient, thermal expansion coefficient) and volumetric expansion coefficient.Thermal expansion in the longitudinal direction of the sample is measured and the volumetric expansion is measured. It shows that there are cases. Normally, thermal expansion refers to the coefficient of linear expansion, which is easy to measure. Further, the linear expansion coefficient includes an average linear expansion coefficient and an instantaneous linear expansion coefficient. The average coefficient of linear expansion refers to the average value of the coefficient of linear expansion between a predetermined reference temperature and the measured temperature, and the instantaneous coefficient of linear expansion refers to the instantaneous coefficient of linear expansion at a certain measurement temperature. When simply referred to as linear expansion coefficient, it refers to the average linear expansion coefficient.

Strictly speaking, the matching of the linear expansion coefficients of the heat sink and the ceramic package means that the linear expansion coefficients are matched throughout the entire process, from the heating up to the silver brazing temperature to the cooling down to room temperature. However, the temperature cooling process after the brazing filler metal has bonded the heat sink and ceramic package is particularly important. If the linear expansion coefficient of the heat sink deviates from the range of linear expansion coefficients allowed by the ceramic package, brazing defects such as warping or peeling will occur.

Generally, in the case of copper-tungsten alloys made by the infiltration method, the tungsten crystal grains form a strong three-dimensional network, so the linear expansion coefficient increases slightly as it is heated. It tends to change. On the other hand, the coefficient of linear expansion of copper-tungsten alloys made by the partial infiltration method or the mixing method decreases with increasing temperature because the tungsten crystal grains are weakly fixed or are covered with molten copper phase. The contribution of the linear expansion coefficient becomes larger and tends to gradually increase.

The generally accepted composite law for the coefficient of linear expansion of composite materials was maintained in the range from room temperature to 500°C for copper-tungsten alloys.

On the other hand, in the case of the copper-tungsten alloy of the present invention, even though it contains a large amount of copper, the adhesion of tungsten crystal grains to each other is strong, so it cannot be made by the infiltration method, partial infiltration method, or mixing method. Unlike copper-tungsten alloys, it exhibits a unique behavior in which the rate of increase decreases at temperatures above 500°C, making it suitable for use in heat sinks.

As an idea to increase the amount of copper while keeping the linear expansion coefficient to the same level as the infiltration method, attempts were made to infiltrate copper into a tungsten material with a honeycomb structure or small holes. As a result, although the effect of suppressing the coefficient of linear expansion was recognized, when a thermal cycle load was applied, the copper phase part gradually swelled up, causing damage to the semiconductor, making it impossible to apply it.

On the other hand, in the case of the copper-tungsten alloy of the present invention, the adhesion of tungsten crystal grains to each other forms a three-dimensional network structure even when viewed microscopically, so that partial swelling of the copper phase can be suppressed. is possible.

In addition, in copper-molybdenum alloys or copper-tungsten alloys, when the infiltrated body or sintered body is rolled, the molybdenum particles or tungsten particles plastically deform each other to form a flat shape. By increasing the contact area with particles, it is possible to suppress the coefficient of linear expansion in a high temperature region. However, since plastic deformation of molybdenum particles or tungsten particles is required, a corresponding degree of processing is required, and the rolled material is limited to a thin plate, and anisotropy in the coefficient of linear expansion occurs.

Copper-tungsten alloys produced by conventional methods such as infiltration, partial infiltration, and mixing methods experience dimensional shrinkage of 0.11% in heat treatment tests heated to 900°C, but the copper-tungsten alloy of the present invention In this case, there is almost no dimensional shrinkage. Specifically, when a heat treatment test was conducted under conditions of heating to 900℃ at a temperature increase rate of 10℃/min in a reducing atmosphere and cooling naturally, the shrinkage rate after heat treatment was 0.05% or less (especially 0.001-0.03%).

Similar effects can be applied to silver-tungsten materials, copper-molybdenum materials, silver-molybdenum materials, silver-tungsten carbide-cobalt materials, and the like.

The electrode material or heat sink of the present invention may be composed only of the above-mentioned copper-tungsten alloy, or may contain some impurities in addition to the copper-tungsten alloy. Examples of such impurities include chromium, molybdenum, aluminum, and the like. The content of such impurities is not particularly limited as long as the effects of the present invention are achieved, and the content of the above-mentioned copper-tungsten alloy is 99.5 to 100 mass % (particularly 99.7 to 99.9 mass %), with the electrode material of the present invention being 100 mass %. % by mass) and preferably contains 0 to 0.5 mass % (particularly 0.1 to 0.3 mass %) of impurities.

The density of the electrode material or heat sink of the present invention is not particularly limited, but from the viewpoint of particularly suppressing consumption of the electrode material, further improving machining speed and machinability, or suppressing plating defects of the heat sink, from 13.5 to 14.5 g/cm ³ is preferable, and 13.9 to 14.0 g/cm ³ is more preferable. The density of the electrode material of the present invention is measured by a calculation method based on size and mass measurement or by an underwater method (Archimedes method).

The relative density of the electrode material or heat sink of the present invention is not particularly limited, but if the relative density is low, the adhesion strength between tungsten crystal grains tends to decrease, and if the relative density is low, internal defects such as pores are likely to occur. If pores remain on the surface of the electrode material or heat sink, defects are likely to be transferred to the workpiece material during die engraving, and they may become a starting point for stains and corrosion when plated. From this point of view, the relative density is preferably 95% or more, more preferably 97% or more. The relative density of the electrode material or heat sink of the present invention is the theoretical density determined from the density and content ratio of the compound constituting the electrode material, or the density determined by a calculation method based on dimension and mass measurement, or the underwater method (Archimedes method). Calculated by dividing by and multiplying by 100.

The hardness (Vickers hardness; Hv10) of the electrode material or heat sink of the present invention is not particularly limited, but from the viewpoint of particularly suppressing consumption of the electrode material and further improving machining speed and machinability, it is 170 to 200. is preferable, and 180 to 195 is more preferable. The hardness of the electrode material of the present invention is measured by a Vickers hardness test based on JIS Z 2244.

The conductivity of the electrode material or heat sink of the present invention is not particularly limited, but when the conductivity of annealed standard annealed copper (volume resistivity ρ ₀ = 1.7241×10 ^-2 μΩm) is defined as 100% IACS, the conductivity of the electrode material or heat sink of the present invention is The conductivity is preferably 40 to 55%, more preferably 42 to 50%, and even more preferably 44 to 48%, when calculated as E _c = ρ ₀ /ρ×100, where ρ is the volume resistivity of the electrode material. The conductivity of the electrode material of the present invention is measured using a conductivity measuring device.

The electrode material for electric discharge machining or the heat sink for semiconductors of the present invention that satisfies these conditions dramatically improves the machining speed, makes the material particularly difficult to wear out, and suppresses the coefficient of linear expansion.

2. Method for manufacturing an electrode material for electric discharge machining or a heat sink for semiconductors The method for manufacturing an electrode material for electric discharge machining or a heat sink for semiconductors of the present invention is not particularly limited, and for example, raw material powders containing copper powder and tungsten powder are mixed. and includes a step of performing discharge plasma sintering treatment under solid phase sintering conditions.

The copper powder used as the raw material powder is not particularly limited, but it is preferable to use a copper powder of a size that forms copper crystal grains having the above average grain size by subjecting the raw material powder to discharge plasma sintering. Specifically, the average particle diameter of the copper powder is an electrode material for electrical discharge machining that is made possible by solid-phase sintering using electrical discharge plasma sintering, which causes tungsten crystal grains to firmly adhere to each other, thereby further suppressing consumption of the electrode material and improving machining speed. From the viewpoint of easy obtaining, the average particle diameter is preferably 13 to 100 μm, more preferably 25 to 50 μm. The average particle diameter of the copper powder is measured using a particle size distribution measuring device.

The tungsten powder used as the raw material powder is not particularly limited, but it is preferable to use tungsten powder of a size that forms tungsten crystal grains having the above-mentioned average particle size by subjecting the raw material powder to discharge plasma sintering. Specifically, the average particle diameter of tungsten powder is It is preferably 2 to 16 μm, more preferably 3 to 10 μm, and even more preferably 4 to 8 μm. In addition, in order to further optimize the relationship between electrode life and machinability, it is also possible to adjust the size of the tungsten crystal grains as appropriate. Furthermore, when giving priority to densification for heat sinks, the average grain size of tungsten crystal grains may be 0.6 to 2 μm. That is, for use as a heat sink for semiconductors, the average grain size of tungsten crystal grains is preferably 0.6 to 16 μm, and for use as an electrode material for electrical discharge machining, the average grain size of tungsten crystal grains is preferably 2 to 16 μm. In addition, if priority is given to the consumability or machinability of the electrode material depending on the electrical discharge machining conditions or the electrical discharge machine used, optimization can be achieved by appropriately adjusting the average particle diameter of the tungsten powder. can. When the average particle size of the tungsten powder is large, consumption of the electrode material can be suppressed, but machinability is reduced. On the other hand, when the average particle diameter of the tungsten particles is small, the wear resistance of the electrode material decreases, but the machinability improves. The average particle diameter of the tungsten powder is measured using a particle size distribution measuring device.

In solid-phase sintering by spark plasma sintering of the present invention, copper particles exist around tungsten particles, but since tungsten and copper do not form a solid solution, growth of tungsten crystal grains and copper crystal grains hardly occurs. do not have. However, when nickel is added to promote densification, activated sintering occurs, which promotes densification, but leads to coarsening of tungsten crystal grains and a decrease in the electrical conductivity and thermal conductivity of the electrode material. It is preferable to limit the nickel content to 0.5% by mass or less for ease of use.

When the copper-tungsten alloy of the electrode material of the present invention contains an alkaline earth metal oxide, it is preferable that the raw material powder contains an alkaline earth metal tungstate powder. The "tungstic acid" in the alkaline earth metal tungstate powder used as the raw material powder includes not only orthotungstic acid but also pyrotungstic acid. There are no particular restrictions on the alkaline earth metal tungstate powder used as the raw material powder, and magnesium tungstate (MgWO ₄ , MgW ₂ O ₇ , etc.), calcium tungstate (CaWO ₄ , CaW ₂ O ₇ , etc.), tungstate, etc. Examples include strontium (SrWO ₄ , SrW ₂ O _7, etc.), barium tungstate (BaWO ₄ , BaW ₂ O _7, etc.). These alkaline earth metal tungstate powders can be used alone or in combination of two or more.

The size of the alkaline earth metal tungstate powder used as the raw material powder is not particularly limited, but it is possible to form alkaline earth metal oxide crystal grains having the above average particle size by subjecting the raw material powder to discharge plasma sintering. It is preferred to use alkaline earth metal tungstate powder of a suitable size. Specifically, the average particle diameter of the alkaline earth metal tungstate powder was determined by solid-phase sintering using spark plasma sintering, which caused the tungsten crystal grains to firmly adhere to each other, further suppressing the wear of the electrode material and further improving the processing speed. From the viewpoint of easy obtaining of electrode material for electric discharge machining, the average particle diameter is preferably 0.1 to 200 μm, more preferably 1 to 10 μm. Note that it can also be set to 50 to 200 μm, particularly 50 to 100 μm. The average particle diameter of the alkaline earth metal tungstate powder is measured using a particle size distribution measuring device.

The amount of each raw material powder used is not particularly limited. When the electrode material or heat sink of the present invention is obtained, it is preferable to use the above-mentioned content. Specifically, if the total amount of raw material powder is 100% by mass, and if it does not contain alkaline earth metal tungstate powder, the content of copper powder is 15 to 45% by mass (particularly 20 to 40% by mass), and the remainder is It is preferable to use tungsten, and when it contains alkaline earth metal tungstate powder, the content of copper powder is 15 to 45% by mass (especially 20 to 40% by mass), and the content of alkaline earth metal tungstate powder is It is preferable that 0.3 to 3% by mass (particularly 0.5 to 2.5% by mass) and the remainder be tungsten.

The electrode material for electric discharge machining or the heat sink for semiconductors of the present invention can be produced by mixing the raw material powders described above and subjecting them to a discharge plasma sintering process (hereinafter also referred to as "SPS treatment") by solid-phase sintering. can be obtained.

When performing SPS treatment, it is preferable to heat the raw material powder. The heating temperature in this case is 950 to 1050, from the viewpoint that it is easy to obtain an electrode material for electrical discharge machining or a heat sink for semiconductors in which tungsten crystal grains are firmly fixed to each other, and it is possible to further suppress material consumption and coefficient of linear expansion and further improve machining speed. ℃ is preferable, and 1000 to 1030 ℃ is more preferable. Furthermore, for the same reason, the heating rate is preferably 30 to 100°C/min, more preferably 45 to 55°C/min. Note that the sintering temperature during discharge plasma sintering is preferably equal to or lower than the melting point of copper (1083° C.) in order to cover the surface of the tungsten powder particles with molten copper and to make it more difficult to inhibit the adhesion of tungsten crystal grains to each other.

When performing the SPS treatment, it is preferable to perform it under pressure. The pressure in this case is not particularly limited, and from the viewpoint that it is easier to obtain an electrode material for electric discharge machining or a heat sink for semiconductors in which tungsten crystal grains are firmly fixed to each other, material consumption and coefficient of linear expansion can be suppressed, and machining speed can be further improved. , 30 to 100 MPa is preferable, and 45 to 55 MPa is more preferable.

The atmosphere for SPS treatment is not particularly limited, but from the viewpoint that it is easier to obtain an electrode material for electric discharge machining or a heat sink for semiconductors in which tungsten crystal grains are firmly fixed to each other, material consumption and coefficient of linear expansion can be suppressed and processing speed can be further improved. A reducing atmosphere such as a hydrogen atmosphere or carbon monoxide, a vacuum atmosphere, an inert atmosphere such as a nitrogen gas atmosphere, or an argon gas atmosphere is preferable.

The time of SPS treatment is not particularly limited, but from the viewpoint that it is easier to obtain an electrode material for electric discharge machining or a heat sink for semiconductors in which tungsten crystal grains are firmly fixed to each other, material consumption and coefficient of linear expansion can be suppressed, and processing speed can be further improved. 30 to 60 minutes is preferred, and 35 to 45 minutes is more preferred.

Effects similar to solid phase sintering using the spark plasma sintering method can also be achieved using the hot press method or the hot isostatic pressing method (Canning HIP method), which involves filling a sealed container with a mixed powder and forming it while vacuuming. However, since the formation of necking is more pronounced in the discharge plasma sintering method, the fixing force of tungsten crystal grains is not as good as that in the discharge plasma sintering method. The reason for this is that in the case of the hot press method and hot isostatic pressing method, homogeneous heat and pressure are applied to the entire grain, whereas in the case of the discharge plasma sintering method, the former This is because, in addition to this effect, the generated plasma preferentially melts the contact areas of crystal grains, forming strong necking.

Next, looking at the obtained infiltrated body or sintered body, in the conventional infiltration method or partial infiltration method, in addition to the amount of copper required for infiltration, excess copper is added to prevent copper from breaking. There are many things to do. These excess copper covers the surface of the infiltrated body, but tends to segregate in some areas due to the temperature distribution in the infiltration furnace and the surface tension of copper. This excess copper phase has a larger coefficient of linear expansion than the copper-tungsten material, which causes the infiltrated body to warp during cooling. In addition, even in the conventional sintering method using the mixing method, a portion of the molten copper seeps out to the surface, forming an excess copper phase and deforming the sintered body.

When extracting copper-tungsten material from these infiltrated bodies or sintered bodies by grinding or honing, it is necessary to take into consideration the amount of deformation, which leads to a reduction in material yield.

On the other hand, sintered bodies obtained by solid-phase sintering such as the hot press method or the SPS method have no excess copper phase, and therefore have a good material yield. On the other hand, when hot isostatic pressing (Canning HIP) is used for solid phase sintering, there is a filling container around the copper-tungsten material that is deformed due to pressure, resulting in a decrease in material yield. It is difficult to prevent.

This is the warping of the copper-tungsten material due to excess copper input, but as the thickness of the excess copper phase increases, shrinkage of copper occurs in the copper phase near the interface between the copper-tungsten material and the excess copper phase. A nest-based pore will be formed. If such pores are formed, the cooling step following the infiltration step under a hydrogen atmosphere can be avoided by switching the atmosphere from hydrogen to nitrogen.

Next, copper-tungsten materials obtained from conventional infiltrated bodies or sintered bodies still have room for densification. This is thought to be because hydrogen incorporated into the copper-tungsten material and the copper phase, which has a coefficient of linear expansion larger than that of tungsten, generate tensile residual stress and fine shrinkage cavities during the cooling process.

When such a copper-tungsten material is subjected to heat treatment at a solid-state sintering temperature, internal energy is released and density is increased and dimensional shrinkage occurs.

On the other hand, copper-tungsten materials obtained by the hot press method, SPS method, and hot isostatic pressing method (Canning HIP method), which do not use hydrogen that is pressurized until the cooling process and is likely to be occluded, are heat treated. There's no need.

The present invention will be specifically explained based on examples, but the present invention is not limited to these examples.

[density]
The density of each sample was measured by a calculation method or an underwater method (Archimedes method). Note that the density of the sample prepared by the infiltration method or the like is measured after removing the excess copper phase.

[Relative density]
The density measured as above is defined as the sintered density (actual density), and the density theoretically calculated by the compound law assuming that no pores or defects are included is defined as the theoretical density, and both densities are calculated as [sintered density / theoretical density × 100(%)] to calculate the relative density.

[Vickers hardness (Hv10)]
The Vickers hardness (Hv10) of each sample was measured in accordance with JIS Z 2244.

[conductivity]
The volume resistivity of each sample was measured using a conductivity measuring device (Sigma Test 2067). Then, when the conductivity of annealed standard annealed copper (volume resistivity ρ ₀ = 1.7241×10 ^-2 μΩm) is defined as 100% IACS, the volume resistivity of the sample of the present invention is ρ, and the conductivity E _c = ρ Calculated as ₀ /ρ×100.

[Electrode consumption]
For the same workpiece having a certain thickness, the electrode wear length (area) was compared when the penetrating process was completed. We determined that the shorter the electrode wear length, the better, and the following:
○: Less electrode consumption than Example 1 (excellent)
△: The amount of electrode consumption is the same as in Example 1. ×: The amount of electrode consumption is greater (inferior) than in Example 1.
It was evaluated as follows.

[Machining time]
We compared the machining time required to complete penetrating machining for the same workpiece with a certain thickness. We determined that the shorter the processing time, the better, and the following:
○: Less processing time than Example 1 (excellent)
△: Processing time is the same as Example 1 ×: Processing time is longer than Example 1 (inferior)
It was evaluated as follows.

[Linear expansion coefficient, elongation rate and total elongation]
The linear expansion coefficient, elongation rate, and total elongation were measured by horizontal differential expansion measurement (using TD5200S manufactured by Mac Science Co., Ltd., in a 100% hydrogen atmosphere, heating rate: 5° C./min, load: 10 gf).

[Thermal conductivity]
The specific heat capacity and thermal diffusivity were measured in the atmosphere at room temperature (25°C) using a laser flash method thermal constant measuring device (Model TC-7000, manufactured by Shinku Riko Co., Ltd.). By multiplying the specific heat capacity by the thermal diffusivity and further by the separately measured density, the thermal conductivity is calculated according to [thermal conductivity=specific heat capacity×thermal diffusivity×density].

[Example 1: 29.15Cu-69.15W- _1.7BaWO4 ]
Cu powder, W powder (average particle size 4.00 to 7.99 μm), and BaWO ₄ powder were weighed to a predetermined ratio and dry mixed using a shaker mixer to obtain a mixed powder. This mixed powder was filled into a graphite die mold, and the die was sandwiched between upper and lower punch molds and placed in the apparatus chamber. A vacuum atmosphere of approximately 10 ^-1 Pa is created inside the chamber using a rotary pump, and then, while applying a pressure of 50 MPa to the mixed powder, the mixed powder is heated with electricity at a heating time of 50°C/min from room temperature to 1030°C, and the temperature reaches the maximum temperature. After holding at 1030°C for 30 minutes, the furnace was cooled, and when the temperature reached 300°C or lower, the load pressure was released, the inside of the chamber was opened to atmospheric pressure, and the sintered body was taken out. The heating temperature was measured using a thermocouple installed in a graphite die.

[Comparative example 1: Ba-based compound addition 30Cu-70W]
Copper-tungsten alloy NEL150 (nickel-containing Cu-W alloy; not subjected to SPS treatment) manufactured by Allied Materials Co., Ltd. was used as the electrode material for electrical discharge machining in Comparative Example 1. Note that the electrode material for electric discharge machining of Comparative Example 1 is a material that has not been subjected to SPS treatment during the manufacturing process.

[Comparative example 2: Ba-based compound addition 30Cu-70W]
Copper-tungsten alloy CE-08 manufactured by Fuji Dice Co., Ltd. was used as the electrode material for electrical discharge machining in Comparative Example 2. Note that the electrode material for electrical discharge machining of Comparative Example 2 is a material that has not been subjected to SPS treatment during the manufacturing process.

[Comparative example 3: Cr-based compound addition 30Cu-70W]
A copper-tungsten alloy manufactured by Silverloy Co., Ltd. was used as the electrode material for electrical discharge machining in Comparative Example 3. Note that the electrode material for electrical discharge machining of Comparative Example 3 is a material that was not subjected to SPS treatment during the manufacturing process.

[Example 2: 30Cu-70W, 50MPa]
Cu powder and W powder (average particle size 2.00 to 3.99 μm) were weighed out at a predetermined ratio (mass ratio of Cu and W: 30:70), dry mixed using a shaker mixer for 2 hours, and then mixed. It was made into a mixed powder. This mixed powder was filled into a graphite die mold, and the die was sandwiched between upper and lower punch molds and placed in the apparatus chamber. Using a rotary pump, create a vacuum atmosphere inside the chamber of about 10 ^-1 Pa. Then, while applying a pressure of 50 MPa to the mixed powder, the temperature was raised from room temperature to 800°C over 12 minutes, and then over 5 minutes. The temperature was raised to 900℃, then electrical heating was carried out to raise the temperature to 1030℃ over 13 minutes, the maximum temperature was maintained at 1030℃ for 30 minutes, the furnace was cooled, and when the temperature reached 300℃ or less, the load pressure was reduced. The chamber was opened and the inside of the chamber was opened to atmospheric pressure, and the sintered body was taken out. The heating temperature was measured using a thermocouple installed in a graphite die.

[Example 3: 30Cu-70W, 80MPa]
A sintered body was produced in the same manner as in Example 2 except that the load during electrical heating was 80 MPa instead of 50 MPa.

[Example 4: 25Cu-75W, 50MPa]
A sintered body was produced in the same manner as in Example 2, except that Cu powder and W powder (average particle size 2.00 to 3.99 μm) were weighed at a predetermined ratio (mass ratio of Cu and W 25:75). .

[Example 5: 25Cu-75W, 80MPa]
A sintered body was produced in the same manner as in Example 2 except that the load during electrical heating was 80 MPa instead of 50 MPa.

[Example 6: 20Cu-80W, 80MPa]
Cu powder and W powder (average particle size 2.00 to 3.99 μm) were weighed to a predetermined ratio (mass ratio of Cu and W 20: 80), and the load during electrical heating was 80 MPa instead of 50 MPa, and the temperature was 900°C. A sintered body was produced in the same manner as in Example 2 except that the temperature was raised from 1030° C. to 1030° C. over 26 minutes.

[Example 7: 20Cu-80W, 80MPa]
Cu powder and W powder (average particle size 1.50 to 1.99 μm) were weighed to have a predetermined ratio (mass ratio of Cu and W 20:80), and the load during electrical heating was 80 MPa instead of 50 MPa, and the temperature was 900°C. A sintered body was produced in the same manner as in Example 2 except that the temperature was raised from 1030° C. to 1030° C. over 26 minutes.

[Comparative example 4: 30Cu-70W, partial infiltration method]
A copper-tungsten skeleton with a predetermined porosity is produced by mixing W powder (average particle size 4.00 to 7.99 μm) with 10% by mass of copper powder and granulating the mixed powder, pressing, degreasing, and solid phase sintering. Then, cut copper strips were added and infiltrated. It was adopted as a regular product for electrical discharge machining or contact points.

[Comparative example 5: 20Cu-80W, coarse grain type, infiltration method]
A tungsten skeleton with a predetermined porosity is produced by pressing, degreasing, and solid-phase sintering W powder (average particle size 4.00 to 7.99 μm), which is doped with 0.3% by mass of nickel nitrate and thermally decomposed nitric acid groups. Cut copper strips were added and infiltrated. It was adopted as a regular product for semiconductor heat sinks.

[Comparative Example 6: 20Cu-80W, Canning HIP method]
Cu powder and W powder (average particle size 1.50 to 1.99 μm) were dry mixed, placed in an iron container, and subjected to canning HIP treatment while vacuum degassing.

[Comparative Example 7: Sapphire standard]
Standard accessories for horizontal differential dilatation measurements were used.

[Evaluation results]
Composition images of backscattered electron images using electron micrographs (500x and 2000x magnification) of crystal structures of samples of Example 1 and Comparative Examples 1 to 3 regarding electrode material applications for electrical discharge machining, and elemental analysis using energy dispersive X-ray analysis. The results and electron micrographs (2000x and 5000x) of the fractured surface are shown in Figure 1. Samples for crystal structure photography are mirror-polished after being embedded in resin, and then etched with Murakami's reagent to clarify grain boundaries. In addition, in Figure 1, in Comparative Examples 1 and 2, the white particles are tungsten, the black particles are copper, and the gray particles are barium compounds, and in Comparative Example 3, the white particles are tungsten, the black particles are copper, and particles with a stronger black color. is a chromium oxide, and the color of the particles becomes whiter as the atomic number increases. As a result, compared to Comparative Example 2, the tungsten crystal grains in Example 1 were larger in size (average grain size 4.00 to 7.99 μm) and more angular, and necking between the tungsten crystal grains was firmly formed. , it can be seen that the adhesion strength between tungsten crystal grains is high. On the other hand, when comparing the sizes of tungsten crystal grains in Example 1 and Comparative Example 1, it is observed that Example 1 is slightly larger and more angular, but the difference is slight. However, from observation of the fracture surface, the tungsten particles in Comparative Examples 1 and 3 were rounded, whereas this was not observed in Example 1, and the tungsten crystal grains were angular, and the tungsten crystal grains were not close to each other. It can be seen that the necking formation is well formed. Furthermore, when comparing the fracture locations, in Example 1 it was confirmed that the fracture occurred from the tungsten-tungsten crystal grain boundary, whereas in Comparative Examples 1 and 3, the tungsten crystal grain It is observed that the particles are shed and broken. This suggests that Comparative Example 1 may have been produced by a mixing method, and it was also found that there was almost no necking formation of tungsten crystal grains.

Next, FIG. 2 shows electron micrographs (500x and 2000x) of the crystal structures of the samples of Examples 2, 4, and 6 for semiconductor heat sink applications. The sample was embedded in resin, polished to a mirror surface, and then etched with Murakami's reagent to clarify the grain boundaries. In FIG. 2, white particles are tungsten, black particles are copper, and the color of the particles becomes whiter as the atomic number increases. As a result, in all samples, although the size of the tungsten crystal grains was smaller than that in Example 1 (average grain size 2.00 to 3.99 μm), they were also angular, and necking between the tungsten crystal grains was firmly formed. It can be seen that the adhesion strength between tungsten crystal grains is high.

In the case of Example 1, coarse-grained tungsten powder was used from the viewpoint of electrode wear, but in Examples 2, 4, and 6, fine-grained tungsten powder was used with emphasis on increasing the sintered density, so the overall crystal structure was It was getting finer.

Next, Table 1 shows the results of the density, relative density, hardness, electrical conductivity, electrode wear and processing time of the samples of Example 1 and Comparative Examples 1 to 3. As a result, even though the density, relative density, hardness, and electrical conductivity are the same, when coarse-grained tungsten powder is used and subjected to SPS treatment, the tungsten crystal grains deform plastically compared to other comparative examples. Moreover, the electrodes were firmly fixed and the wear of the electrodes was suppressed, and the wear of the electrodes was about half that of Comparative Examples 1 to 3. Furthermore, the processing time could be reduced by about 10% compared to Comparative Examples 1 to 3.

Next, Table 2 shows the results of the densities and relative densities of the samples of Examples 2 to 7. As a result, in all samples, a considerable degree of density could be secured by SPS treatment. Since Cu softens at around 500°C, it is assumed that the relative density increases due to sintering (neck formation) between tungsten grains at high temperatures. It can also be seen that densification progresses by increasing the load during SPS treatment from 50 MPa to 80 MPa and by reducing the average particle size of the tungsten powder from 2.00 to 3.99 μm to 1.50 to 1.99 μm.

Next, the linear expansion coefficients of the samples of Example 1 and Comparative Example 1 are shown in FIG. As a result, in Comparative Example 1, the linear expansion coefficient at 800°C was larger than that at 500°C, whereas in Example 1, the linear expansion coefficient at 800°C was smaller than the linear expansion coefficient at 500°C. (approximately 95%). According to Figure 1, there is a slight difference between the microstructure photographs of Example 1 and Comparative Example 1, so the difference in linear expansion coefficient behavior is between the SPS treatment and the conventional method (which appears to be a mixed method based on the observation of the fracture surface). It is thought that this was caused by the difference.

Next, the results of the linear expansion coefficients of the samples of Examples 2, 5, and 7 and Comparative Examples 4 to 7 are shown in Table 3 and FIGS. 4 to 5. For samples prepared by the infiltration method or the partial infiltration method, linear expansion coefficients were observed to increase linearly with increasing temperature, but in the temperature range of 500°C or higher, the linear expansion coefficients increased steadily. On the other hand, the slope of the SPS sintered body became gentle from around 600℃, and the coefficient of linear expansion in the high temperature range became low. This is a feature that cannot be seen in products produced by conventional mixing methods, infiltration methods, and partial infiltration methods. In addition, in the case of the sapphire standard sample of Comparative Example 7, it increases linearly, so it is understood that the decrease in linear expansion coefficient in the high temperature range of the SPS sintered body is not a measurement error caused by the measuring device. can. Moreover, from FIG. 5, the behavior of the coefficient of linear expansion of the SPS sintered body changes almost linearly depending on the change in copper content. Furthermore, when comparing the linear expansion coefficients at 500°C and 800°C of samples prepared by the infiltration method or the partial infiltration method, the linear expansion coefficient at 800°C is higher. On the other hand, the SPS sintered body has a higher coefficient of linear expansion at 500°C. Therefore, it is assumed that the decrease in the linear expansion coefficient in the high temperature range is a behavior unique to SPS sintering. The reason for this behavior is that a strong three-dimensional network of tungsten crystal grains was formed by the solid-phase sintering SPS process. In other words, in SPS treatment, tungsten crystal grains are more firmly fixed to each other than in the conventional tungsten skeleton, so it has been found that the binding force against copper, which has a high coefficient of linear expansion, is high. Furthermore, the constraint on the coefficient of linear expansion depends on the amount of copper contained.

Next, the elongation percentage results of the samples of Examples 2, 5, and 7 and Comparative Examples 4 to 7 are shown in FIGS. 6 to 8. As a result, in Comparative Examples 4 and 5, which were manufactured using the infiltration method or the partial infiltration method, the elongation rate increased almost linearly as the temperature increased, and increased upward in the temperature range of 500°C or higher. . On the other hand, in Examples 2, 5, and 7, the increase in elongation rate was suppressed in the high temperature range (700 to 900°C). This suggests that in Examples 2, 5, and 7, the tungsten particles were more firmly stuck to each other (necking) due to the SPS treatment. In addition, in the case of Comparative Example 6 that adopted the Canning HIP method, the increase in elongation rate in the high temperature range (700 to 900°C) was also suppressed, but in the case of Example 7 that adopted the SPS treatment, the increase in the elongation rate in the high temperature range ( The increase in elongation rate at temperatures (700 to 900℃) was further suppressed. For this reason, it is suggested that by employing SPS treatment, the tungsten grains are more firmly stuck to each other (necking) than when the Canning HIP method is employed. This is thought to be because in the case of the Canning HIP method, solid-phase sintering is performed by heating and hydrostatic pressure, whereas in the SPS processing method, necking formation by plasma is further taken into consideration.

Semiconductor heat sinks are coated with nickel plating on the surface to improve their corrosion resistance, but copper-tungsten alloys made by conventional infiltration methods are known to shrink during the nickel plating baking process. Countermeasures are being taken, such as making products larger in anticipation of the shrinkage rate. This shrinkage behavior will cause dimensional defects in the product if the above-mentioned measures are not taken when it is necessary to manufacture a very large product. Table 3 shows the shrinkage rate caused by heating conditions in which the copper-tungsten alloy is heated at a rate of 10°C/min and cooled naturally when it reaches 900°C. Comparative Example 5 produced by the conventional infiltration method had a shrinkage of 0.11%, whereas Example produced by SPS sintering had a shrinkage rate of only 0.01%.

Copper-tungsten alloys fabricated by SPS treatment rarely exhibit such shrinkage behavior, indicating that precision machining is possible without the above-mentioned measures.

The presence or absence of sample shrinkage during the heat treatment test is likely caused by the behavior of the copper phase, which makes a large contribution to the thermal expansion and contraction of copper-tungsten alloys. Although the sample as a whole contracts during the cooling process, it is thought that the copper phase, which has a large coefficient of linear expansion, causes tensile residual stress in the copper-tungsten alloy compared to the tungsten phase, which has a small coefficient of linear expansion. It is thought that this tensile residual stress caused microscopic defects in the copper-tungsten alloy and created room for shrinkage during the heat treatment test. This phenomenon occurs in infiltration methods, partial infiltration methods and mixing methods which are processed under atmospheric pressure. On the other hand, in the SPS method and Canning HIP method, which are processed under pressure, it is thought that the tensile residual stress that occurs during cooling is canceled by the pressure from the surroundings, so that the sample does not shrink. .

Next, Table 4 shows the measurement results of the thermal conductivity and electrical conductivity of the samples of Examples 2, 4, 5, and 7. For reference, the thermal conductivity and electrical conductivity of tungsten and copper are quoted from the Metal Data Book.

In Examples 2, 4, and 5 where the copper content is up to 25% by mass, the thermal conductivity differs from the theoretical value depending on the volume ratio, but it decreases significantly in Example 7 where the copper content is 20% by mass. Therefore, the difference from the theoretical value due to the volume ratio is large. This tendency is the same for electrical conductivity. The reason for this is thought to be insufficient densification of the sample. Referring to Table 2, the relative densities of Examples 2, 4, 5, and 7 were 98.5%, 97.7%, 98.0%, and 97.4%, respectively, and it was confirmed that this was due to insufficient densification of the samples.

Claims (10)

An electrode material for electric discharge machining or a heat sink for semiconductors made of a copper-tungsten alloy,
The copper-tungsten alloy consists of 15 to 45% by mass of copper, with the remainder being tungsten and unavoidable metal elements, and the content of tungsten is 100% by mass. is a copper-tungsten alloy with a content of 55 to 85% by mass,
The copper-tungsten alloy has tungsten crystal grains having an average grain size of 0.6 to 16 μm, and the copper-tungsten alloy has a coefficient of linear expansion at 800°C smaller than a coefficient of linear expansion at 500°C. electrode material or heat sink for semiconductors. The electrode material for electric discharge machining or the heat sink for a semiconductor according to claim 1, wherein the copper-tungsten alloy forms necking between tungsten crystal grains. The electric discharge machining material according to claim 1 or 2, wherein the shrinkage rate after heat treatment is 0.05% or less when subjected to a heat treatment test in which the temperature is raised to 900 °C at a rate of 10 °C/min and then naturally cooled to room temperature. Electrode materials or heat sinks for semiconductors. An electrode material for electrical discharge machining made of a copper-tungsten alloy,
The copper-tungsten alloy contains 15 to 45 mass % copper, 0.3 to 3 mass % alkaline earth metal oxide, and the remainder tungsten and A copper-tungsten alloy consisting of inevitable metal elements and having a tungsten content of 54.7 to 82% by mass,
The copper-tungsten alloy has tungsten crystal grains having an average grain size of 2 to 16 μm, and the copper-tungsten alloy has a coefficient of linear expansion at 800°C smaller than that at 500°C. electrode material. The electrode material for electric discharge machining according to claim 4, wherein the copper-tungsten alloy forms necking between tungsten crystal grains. The electrical discharge machining material according to claim 4 or 5, wherein the shrinkage rate after heat treatment is 0.05% or less when subjected to a heat treatment test in which the temperature is raised to 900 °C at a rate of 10 °C/min and then naturally cooled to room temperature. electrode material. A method for manufacturing an electrode material for electric discharge machining or a heat sink for semiconductor according to any one of claims 1 to 3, comprising:
A process of mixing raw material powder containing copper powder and tungsten powder and subjecting it to discharge plasma sintering under solid phase sintering conditions,
A manufacturing method, wherein the tungsten powder has an average particle size of 0.6 to 16 μm. According to claim 7, in the raw material powder, the content of the copper powder is 15 to 45% by mass, and the content of the tungsten powder is 55 to 85% by mass, assuming that the total amount of the raw material powder is 100% by mass. manufacturing method. A method for producing an electrode material for electrical discharge machining according to any one of claims 4 to 6,
A step of mixing raw material powder containing copper powder, tungsten powder, and alkaline earth metal tungstate powder and subjecting it to discharge plasma sintering under solid phase sintering conditions,
A manufacturing method, wherein the tungsten powder has an average particle size of 2 to 16 μm. In the raw material powder, the content of the copper powder is 15 to 45% by mass, the content of the tungsten powder is 54.7 to 82% by mass, and the content of the alkaline earth metal tungstate powder is 0.3 to 3% by mass. The manufacturing method according to claim 9, which is mass %.