Glass Blowing for Vacuum Devices

Glass to Metal Seals

When constructing glass vacuum devices, glass to metal seals are required to get electrical connections through the glass envelope. The problem is that the metal must have a co-efficient of expansion which is very similar to the glass. This is particularly important with soda-lime glass. It is also important that the glass bonds well to the metal. Up to about 1915 this was achieved using platinum wire. Platinum has an expansion co-efficient of 8.8x10^-6C^-1 which is a good match to soda-lime glass, which has an expansion co-efficient of 9.0 to 10.0x10^-6C^-1. Platinum is also resistant to oxidation and glass adheres well to it. The main disadvantage is the high cost of platinum.

Dumet Seals

In 1911, Fink and Eldred introduced a revolutionary new type of seal. Nickel/Iron alloys can be made with a wide range of expansion co-efficients which cover that of glass. The problem is that seals made with Nickel/Iron wire are often porous. It also suffers a chemical reaction with the glass during the sealing operation which causes gas bubbles to form. Copper seals well to glass, but has the wrong expansion co-efficient. The step which Fink and Eldred made was to make a wire which has a Nickel/Iron core bonded to a sleeve of copper. The thicknesses and expansion co-efficients are chosen so that the radial expansion matches that of the glass. The axial expansion is less than the glass, but the stress is taken up by the copper sheath. This type of seal is now known as a Dumet seal. It is pronounced 'dew-met'. Dumet wire is suitable for use with soda-lime glass and lead glass. It is not suitable for use with borosilicate glass. It is used in CRTs, vacuum tubes, fluorescent tubes and filament lamps. Dumet seals can be easily identified by their red appearance within the glass. See United States patents 1083070 and 1498908.

Seals Using Recovered Valve Pins

For experimental work, valve pins can be recovered from the more modern vacuum tubes. I have re-used seals from low power radio valves. I think these valves use soda-lead or soda-potash-lead glass which has a very similar expansion co-efficient to soda-lime glass. These are Dumet seals and are necked where they pass through the glass to improve mechanical strength. They can be removed from the glass by breaking up the valve base and then crushing the glass around each seal using a vice. This is rather dangerous and goggles must be used. An additional hazard of breaking up old valves is that they contain poisonous materials such as barium oxide, which can be fatal if ingested or inhaled and can also seriously irritate the skin. On removal of a pin, most of the red material remains adhered to the glass. The necked portion then has the copper surface exposed. These pins can be successfully re-sealed into soda-lime glass.

I have found that they can be sealed in more easily if they are pre-beaded with glass. This can be done with a glass rod which has been drawn out thin. I hold the pin with a pair of pliers and heat the glass and the pin at the same time. I wrap one end of the thin rod around the necked part of the pin and then draw the rod away so that it detaches. I turn the gas up high and rotate the pin in the flame until the glass melts into a round bead. To seal in the pin, a small area of the wall of the glass tube or envelope can be blown out while heating with a very sharp flame. The pin and the blown out side of the tube then have to be heated at the same time. I use the bunsen type burner for this. I then push the softened bead into the hole in the tube. It then needs to be annealed and allowed to cool slowly.

Seals Using Nickel/Iron Wire

Nickel/iron wire can be purchased in small quantities from specialist material companies such as Goodfellow.. The cost for short lengths is rather high. I have successfully used 0.5mm diameter Ni48/Fe52 wire for filament lamp glass to metal seals. The most obvious problem is the development of bubbles in the seal. However, this type of seal can sometimes be porous, leading to premature failure of the device.

I have found that when sealing nickel/iron wire into glass, the formation of bubbles can act as a guide to the correct temperature. The bubbles start to form when the glass begins to fuse to the wire. This means that nickel/iron wire seals with no bubbles are usually leaky.

Seals Using Dumet Wire

Dumet wire is difficult to obtain in small quantities. I eventually purchased a 3kg reel of 0.35mm diameter borated Dumet wire from JLC Electromet in India.

I have since been contacted by Kamal Baid of JLC Electromet who has kindly offered to provide technical support to anybody doing scientific work involving Dumet wire. Email ksbaid@jlcelectromet.com