In my initial attempts at making homemade incandescent lamps (see Homemade Light Bulbs (failed initial attempts)), I didn't know the proper way to work with leaded glass. I worked it as if it were a borosilicate glass with a low melting point, so I was always using a soft flame, and tried not to heat it to aggressively due to the higher thermal expansion. However the glass would always turn dark, no matter how little I tried to heat it, and I was frequently developing cracks. After receiving some feedback by glassblowers and reading a few books, it turns out a more neutral/oxidizing flame is required when working with leaded glass. Using a reducing flame will allow metallic lead to come out of the glass, causing it to darken (and even start to conduct).

I now work with a very oxygenated flame (an oxygen supply to the torch is required, air will not work even though the flame may be hot enough to melt the glass), and work the glass at the outer part of the flame. As the glass warms in this flame, it will discolor yellow before it gets hot enough to incandesce, but it will cool to a perfectly clear glass. If the glass is brought too far into the flame, patches of opaque glass will start to incandesce where the metallic lead is coming through, but simply moving the glass further down the flame will allow the lead to re-oxidize.

In order to make the glass to metal seal, Dumet wire is used. The glass tube is first heated enough that it may be squeezed onto the Dumet wires, then the seal is thoroughly heated so the glass and Dumet wire are glowing orange hot. If the seal is not heated enough, as the seal cools the wire will separate from the glass, and a leak will form. This is visible when closely examining the cooled seal. In a good Dumet seal, the glass fully wets the wire, and the wire will have a reddish color inside the glass (this can be seen by examining any commercial incandescent lamp).

Shown below is one of my successful incandescent lamps.

The lamp works well when connected to a 9-volt battery:

Construction:

The lamp is made from a single piece of 7 mm lead glass tubing. First a bulb is blown in the center of a tube. The tubing at one end is scored and cleaved about 1cm from the bulb, leaving enough length for the glass to metal seal. Contaminates from the score do not need to be cleaned from this end, as it will not be sealed to anything. The opposite end is heated in a flame and allowed to thicken such that the inner diameter becomes narrower, while keeping the outer diameter the same (this step should actually be performed right after blowing the bulb, while the glass is still hot; I simply forgot to do this step until later). This thickening of the tube wall is to facilitate vacuum sealing later, and allows for a stronger seal.

A piece of borated Dumet wire is cut, and the ends are sanded to reveal a bright copper finish. The ends are then folded over, and a tungsten filament is crimped on the end.

The Dumet wire with the filament attached is placed in the bulb, and the glass is softened and pressed together. The glass and Dumet wire are heated further until they glow orange hot, allowing for a good glass to metal seal.

The bulb is attached to a vacuum pump, and evacuated for several minutes. At this stage the bulb can be leak checked (see Leak Checking). The filament is then flashed several times, burning off contaminates, and revealing a bright clean tungsten filament as long as no large leaks are present. Argon could be used to backfill the bulb then pumped out to achieve a better vacuum. Afterwards the bulb can be filled with ~5 Torr of argon to reduce filament evaporation and extend the life of the lamp.

Finally, the thickened portion of glass tubing is heated (while the vacuum pump is still running), and allowed to slowly collapse in on itself. The glass should not be allowed to get so hot that the vacuum pulls in a bubble of glass, as this will result in a weaker joint. The completed light bulb is then pulled away.

Bulb Blackening

In a standard incandescent lamp, a small amount of inert gas is added to help cool the surface of the filament and reduce evaporation, extending the life of the lamp. If the bulbs here which only contain a vacuum are run very brightly, the tungsten will quickly evaporate and coat the inside of the bulb, blackening the lamp. In the images below, the lamp which lit nicely at about 10 volts (shown in the images above) was run at about 20 volts (which is four times as much power) for about a minute. Running the filament at these higher temperatures greatly reduced the effective life of the bulb (running even hotter would simply burn out the filament). The filament will still evaporate when the lamp is run at lower powers, but keeping the filament cooler greatly reduces the rate at which it evaporates.

Having a small amount of water trapped in the bulb will have a similar effect, known as the water cycle. When water gets near a hot filament, it dissociates and the oxygen reacts with the tungsten. The tungsten oxide boils off the filament and condenses to a cooler part of the lamp, and the freed hydrogen reacts with the tungsten oxide to reform water, and tungsten metal is left deposited. For this reason it is very important that water is removed. A getter may be used to trap any water remaining after tip-off.