RU2564685C1 - Heat fusion method - Google Patents

Abstract

FIELD: electricity.

SUBSTANCE: heat fusion method using electrode materials of silicon semiconductor substrate with thermal compensator is performed by aluminium/silumin through the silicon oxide layer which is grown up on the surface of the semiconductor substrate.

EFFECT: providing small depth of penetration of electrode material into the semiconductor substrate.

2 cl, 2 dwg

Description

The invention relates to a method for fusing silicon semiconductor wafers and semiconductor substrates with thermocompensating electrodes with electrode materials.

Since silicon is brittle and has a low CTE (coefficient of thermal expansion), the silicon semiconductor wafer of a high-power device is connected to a copper collector through an intermediate thermocompensating electrode made of metal tungsten or molybdenum (W and Mo have good thermal conductivity and electrical conductivity; KTP W, Mo, and Si differ slightly).

An example of the manufacture of a diode structure on silicon. On one side, a p-type silicon wafer is fused with an Au-Sb electrode alloy foil. On the opposite side, an Au-B electrode alloy is melted over the entire plate area to create an ohmic contact. To reduce the thermal stresses arising in silicon during the alloying of gold alloys, tungsten plates are melted to Au - electrodes, the thermal expansion coefficient of which is approximately equal to the thermal expansion coefficient of silicon [1].

Variant of manufacturing powerful semiconductor rectifiers: the active part of the device is made in the form of a set (“sandwich”), the elements of which are arranged in the following order - mounting plate Mo or W + aluminum layer + n-type silicon disk + antimony-doped gold layer + second mounting plate (W or Mo). A set of these elements is heated at a temperature of ~ 750 ° C. In this case, diffusion of impurities in silicon occurs and a pn junction is formed [2].

Gold (silver) is alloyed with antimony or boron (usually ~ 1%) to create a pn junction or to provide a low ohmic transition resistance of the metal-silicon compound. When pure gold (silver) is fused into silicon, a transition layer of high-resistance resistance is formed, which prevents the passage of electric current (deterioration of the current-voltage characteristics of a semiconductor device).

In valves consisting of silicon wafers with molybdenum disks soldered on both sides, pores are produced at the soldering sites, which worsen electrical characteristics. One disk is soldered to the plate with Al-Si eutectic solder, the other is Ag - 2% Pb - 1% Sb solder. This disc is nickel-plated for better wetting with solder. Soldering is carried out in vacuum at 900 ° C. Metallographic analysis showed that the pores are obtained between the plate and the nickel-plated disk due to the formation of the Ni-Si compound. This phase is obtained due to the reaction between Ni and Si, dissolving in the solder. Porosity does not occur at the sites of soldering of metallic molybdenum (without coating) with Al-Si solder [3].

Note. The galvanically deposited nickel on the surface of Mo (W) is preliminarily burned into the substrate at high temperature. During the fusion of silicon with a temperature compensator, due to the metallurgical (chemical) processes that occur, the nickel sublayer often completely peels off the Mo (W) surface.

Bell Telephone has developed an electrolytic method for applying gold to molybdenum without depositing an intermediate nickel coating. The need for coating molybdenum with gold is due to the lack of the possibility of direct connection of molybdenum with silicon using an Au-Sb or Au-B alloy. It was necessary to first precipitate nickel on the surface of molybdenum, which had a negative effect on the performance of semiconductor devices.

Exposure of hydrogen peroxide to molybdenum on it produces a thin porous oxide film. During the electrolytic process, gold is deposited both on molybdenum, penetrating through the pores in the oxide film, and so on the film itself. Subsequent reduction of the oxide with hydrogen at a temperature of 900 ° C provides mechanical adhesion of the Mo-Au interface. An additional gold coating may be applied to the resulting layer. [4] [5]

Silver alloy composition: 0.01% Si (Ge); 0.2% Pd (Pt); Ag - the rest - can be used for thermal silvering of temperature compensators without first applying a nickel sublayer to the surface of W or Mo for the subsequent fusion of these electrodes with silicon. A portion of a silver alloy in a hydrogen atmosphere at a temperature of 1100 ° C completely spreads over the surface of metallic Mo or oxidized (to dark purple) W. [6]

The high temperature of fusion of silicon with a temperature compensator adversely affects the characteristics of a semiconductor device.

An electrode solder of the composition Pb-Jn-Ni (lead content 50–99%) wets silicon at a temperature of 825 ° C [7].

The use of silver alloy composition: 0.5% Ni; 0.5% Sb; 2% Pb; 15% Jn; Ag - the rest allows you to reduce the soldering temperature to 780 ° C (in a vacuum of 10 ^-4 mm Hg). [8]

The recrystallized layer formed upon the fusion of indium into p-type silicon has n-type conductivity. Perhaps this is due to a sharp difference in the atomic radii of silicon and indium and the appearance of a broken crystal lattice in the crystallized silicon layer. [9] [10]

In all the examples cited, the thermocompensator is fused to the surface of a semiconductor wafer having the same conductivity type (“n” or “p”) on the entire plane. Known devices (for example, triacs and thyristors with reverse conductivity), where under the same plane of the semiconductor substrate are formed (diffusion) region of both "n" and p-type.

Creating low-impedance non-rectifying contacts on thin diffusion layers of silicon devices: a thin film (200-300Å) of gold is sprayed onto the silicon surface. Then, without stopping the deposition of Au, an Ag film is deposited with a thickness of 5 μm. From the start of the Ag sputtering process, Au sputtering is stopped until the Ag film thickness reaches a predetermined value. After that, Au is again sprayed, stopping Ag. At the end of the deposition process, contacts are melted at a temperature above the eutectic point of Au-Si (370 ° C), but below the eutectic of Ag-Si (830 ° C). [eleven].

The method is based on the fact that gold and silver form a solid solution between themselves, and depending on the content of components (Ag and Au), the Ag Au Si alloy has a different eutectic temperature. An alloy containing 75% Au and 25% Ag has a eutectic temperature with silicon of 490 ° С, and for a 50% Au and 50% Ag alloy, the eutectic temperature is 570 ° С [12].

The same technique is used by spraying a silver alloy of the composition: 0.5% Ni; 2.5% Ge; 27% Cu; 70% Ag (~ 5 μm), which is then fused into the substrate at a temperature of 700 ° C. Gold foil solder the temperature compensator with the surface of metallized silicon [13] [14].

All these are quite laborious processes and are associated with the use of precious metals (Ag and Au).

In mass industrial production, they try to use silumin (aluminum alloy with silicon) electrode solders to connect the temperature compensator (Mo and W) and silicon, which has different types of conductivity on one surface. Al-Si alloy (silumin) has a eutectic temperature of 577 ° C with a silicon content of 12.5% in the eutectic. The fusion is carried out with the smallest possible thin foil of silumin (to fill the soldered seam) at the lowest possible temperature (~ 600 ° C) and a long exposure time to form a Si-Mo (W) brazed joint. They try to saturate the surface of the n-type silicon region with a phosphorus donor (diffusion) as much as possible to protect against the possible transition of the n-type silicon into the p-type (as a result of fusion of the aluminum acceptor into silicon).

The smallest possible soldering modes cause an increased percentage of rejects (non-soldering). There is an overcompensation from the fusion of acceptor-aluminum into n-silicon (increased transient ohmic resistance).

Before fusing the temperature compensator with silicon, the brazed surfaces, namely silicon, aluminum (silumin), molybdenum (tungsten), are first etched to remove possible contaminants. But in an airy atmosphere, etched surfaces are again coated with oxide films. And if in the fusion medium (vacuum, hydrogen) the oxide films on the surface of the temperature compensator (W, Mo) are reduced to pure metal, the reduction (dissociation) of oxide films on the surface of silicon and aluminum (silumin) does not occur (vacuum or hydrogen for silicon and aluminum in fusion modes play the role of a protective environment against further oxidation).

Since the KTP of alumina is much smaller than the KTP of aluminum, as the temperature rises (when soldering) the oxide film breaks on the aluminum surface, pure aluminum deoxidizes the silicon oxide and melts into the semiconductor wafer.

But the solder foil (aluminum, silumin) is not able to deoxidize a thickened layer (up to ~ 1 μm) of borosilicate (phosphorosilicate) oxide formed on the surface of the semiconductor substrate at the stages of diffusion processes (creating the necessary semiconductor structure).

It is known that Schottky diodes (having relatively low values of direct voltage drop) are made by sputtering a metal onto an n-type silicon surface. But in an airy atmosphere, a silicon surface is never free of oxides. Those. the structure of a Schottky diode has the form: Si-SiO ₂ -Μ (Μ is a metal). When summing up the difference in electric potentials, the electrons from the conductor overcome the potential barrier created, including by the oxide film, passing into the metal. When changing the direction of the current, the electrons do not transfer from the metal to the semiconductor (if the breakdown voltage of the barrier layer is not exceeded).

A semiconductor structure with “n” and “p” types of silicon conductivity on the same surface is proposed, fused with a thermal compensator using silumin, but an oxide film (~ 1 μm) can be grown on silicon (or left after the diffusion process). The oxide film is wetted by solder, but prevents the acceptor-aluminum from penetrating deep into the semiconductor substrate. The oxide film does not prevent the opening of the device and the flow of direct current through it. Fusion is carried out in conventional (non-critical) soldering modes, with technological adjustment of the temperature and time of the process.

Two versions are available:

a) coating with oxide the entire plane of the plate (Fig. 1);

c) oxide coating only the n-type surface (complete oxidation, using photolithography, the oxide is etched off the p-type silicon surface) (Fig. 2)

Explanation 1 — temperature compensator (Mo, W); 2 - solder (Al, silumin); 3 - oxide layer; 4 - region of n-type silicon; 5 - emitter shunts; 6 - region of p-type silicon.

In the case of the embodiment of FIG. 2 requires individual conditions for the composition of the solder (silumin). Al-Si alloy has a melting point of 577 ° C with a silicon content in the eutectic of 12.5% (according to other sources, l = 11.7%). With a reduced silicon content (5.10%), the solder consists of aluminum interspersed with the eutectic Al-Si alloy (12.5% Si). With an increase in silicon content (15.20%), the solder consists of a eutectic Al-Si alloy interspersed with silicon.

In the process of soldering, when the temperature rises above 577 ° C, the eutectic Al-Si alloy begins to dissolve the adjacent silicon of the semiconductor substrate according to the state diagram [15]. But if the solder is melted into silicon in the region of emitter shunts (having a small area), the alloy outside the shunts is forced to saturate with silicon from the shunt region, melting far into the thickness of the silicon semiconductor substrate, reducing the depth of the p-type region. This can be both useful and harmful to the characteristics of a semiconductor device. If increased alloying of the surface of the shunts with an acceptor (aluminum) is necessary without the danger of deep penetration of silicon, the manufactured solder (silumine) should contain more silicon than its possible complete dissolution (according to the state diagram) at the soldering temperature. The basis of the solder (eutectic alloy) when the soldering temperature rises (heating the furnace) will dissolve silicon from its own composition and from the adjacent surface of the semiconductor substrate.

Ultimate solubility of silicon in solder depending on temperature: 577 ° С - 12.5%; 600 ° C - 13.5%; 650 ° C - 17%; 700 ° C - 20%; 750 ° C - 24%; 800 ° C - 28%. [fifteen]

If deep fusion is required, silumin with a low silicon content or pure aluminum should be used as solder.

It is proposed to use this fusion method to connect the temperature compensator to the surface of the n-silicon diode structure (by order of consumers).

At first glance, the proposed method contradicts common sense: a dielectric oxide film does not interfere with the passage of an electric current of considerable strength (tens and hundreds of amperes). Reality: the laws of quantum physics begin to manifest themselves in thin films on the surface of a semiconductor.

Information sources

1. The patent of Germany No. 1110323.

2. US patent No. 3160798.

3. "Metallurgy" 1964.70 No. 422, p. 267-270.

4. "Semiconductor Product", 1965, vol 8, Χ, No. 10, p. 64.

5. “Ε E E”, 1965, vol 13, Χ, No. 10, p. thirty.

6. Copyright certificate of the USSR No. 254299.

7. US Patent No. 2879457.

8. Copyright certificate of the USSR No. 254784.

9. Japanese Patent No. 855 publ. 01/27/66.

10. US Patent No. 3285791 I. US Patent No. 3028663.

12. Japan Patent No. 23269 publ. 10/19/64.

13. USSR copyright certificate No. 415117.

14. USSR copyright certificate No. 1273224.

15. Hansen M, Anderko K. “Structures of double alloys”, M., “Metallurgizdat”, 1962, v. 1, p. 150-152.

Claims (2)

1. The method of fusion of a silicon semiconductor substrate with a temperature compensator by the electrode material, characterized in that to ensure a small depth of penetration of the electrode material into the semiconductor substrate over a large area, the fusion is carried out by aluminum / silumin through a layer of silicon oxide grown on the surface of the semiconductor substrate. 2. The fusion method according to claim 1, characterized in that to create zones of direct soldering of the silicon surface, before fusion in local places, remove the layer of silicon oxide from the surface of the semiconductor substrate.

Images