VACUUM BRAZING OF METAL WITH CERAMICS*)

Recently, especially in connection with the development of magnetrons, a need has arisen for insulating materials possessing good dielectric properties, a high melting temperature, and easy amenability to mechanical machining. Such materials make it possible to achieve high precision in assembly, and also make it possible to accept high temperatures present in certain vacuum devices. For a long time the idea of using ceramics for these purposes has been discussed in various countries. To do this, one must be able to braze metals to ceramics.

In the article under review, Neal Williams presents data on work carried out in this direction by his group. In order to braze metal to ceramics, it is first necessary to obtain a certain chemical affinity between the surfaces being brazed. This was achieved by metallizing the ceramic (AlSiMag No. 243) by a special method, namely by sintering a mixture of molybdenum and iron powders with the ceramic under the action of high temperature. The metallized surface thus obtained was then sintered with a special nickel mixture. The metallization process was as follows: the ceramic was cleaned by immersion in nitric acid and then thoroughly washed. A mixture was prepared of 96% Mo and 4% Fe in powder form. 40 g of this mixture was mixed with 100 g of binder (10% nitrocellulose and 90% ethyl acid). The resulting composition was applied in a thin layer ($\sim 0.1$ mm) to the surface of the ceramic. The ceramic was placed in the flame of an oxyhydrogen torch (70% $N_2$ and 30% $H_2$) and subjected to a temperature of 1400° C. The metallized surface obtained in this way was then, in the same order, coated with a nickel composition (powdered nickel—40 g and the same binder—100 g). Firing in this case was carried out in a hydrogen flame to a temperature of 1000° C.

The ceramic surface treated in this way can be brazed to the surface of the required metal. Williams’s group used Driver-Harris alloys No. 52 and No. 14 for this purpose. As the solder they used Handy and Harman silver solder “BT.” Since the coefficient of expansion of alloy No. 52 is greater, and that of No. 14 less, than the coefficient of expansion of the ceramic, equal to $10.5 \cdot 10^{-6}$, No. 52 was used for external joints and No. 14 for internal joints. In this case, at the moment of heating, the gap between the metal and the ceramic increased, the solder filled it, and upon subsequent cooling a more compacted layer was obtained.

The brazing itself was carried out either in the flame of a hydrogen torch or with the aid of a molybdenum induction coil placed over the joint.

The joints obtained were vacuum-tight—no leak could be detected at a vacuum of $5 \cdot 10^{-7}$ mm Hg. They withstood temperatures up to 600° C without deterioration of their vacuum properties. Magnetrons with these joints were tested

* Neal Williams, The Review of Scientific Instruments 18, No. 6, June (1947).

did not leak during three to four months of operation. The author of the article considers it necessary to make several comments concerning the technique of this type of brazing. It is noted that, in attempting to metallize the surface of ceramics in a flame using hydrogen alone, it was not possible to obtain products of proper quality. The suggestion is made that the presence of a small amount of oxygen (\(\sim 0.25\%\)) in the nitrogen cylinder has a beneficial effect on the formation of the silicate mixture. It is assumed that a mixture of 70% Mo and 30% Fe (instead of 96% Mo and 4% Fe) permits brazing without tinning the metallized surface of the ceramic. In this case the mechanical strength of the joint was established, but the vacuum tightness was not checked.

It is noted that only ceramic cylinders with wall thicknesses up to 1.25 mm were tested for vacuum tightness. No data are given for smaller thicknesses.

Finally, it is indicated that Kovar, using the same method, was brazed to AlSiMag No. 35 ceramics.

V. Fedorov