Several early inventors developed incandescent lamps that contained a gas filling. Those early efforts did not result in truly successful products and the aim of this writing is to comment on the features needed in such a lamp for it to be successful. This can be done by comparing early lamp designs with the later successful design of Irving Langmuir.

The necessary attributes of an efficient gas-filled lamp can be determined from the work performed by Nobel Laureate Irving Langmuir. Langmuir's biography and technical articles dealing with incandescent lamps are available in Volumes 12 and 2, respectively, of The Collected Works of Irving Langmuir, C. Guy Suits, General Editor, Pergamon Press, New York, 1960.

In the year 1911 Irving Langmuir began a scientific investigation to understand the effects of the interaction of gases with incandescent filaments. Platinum and tungsten filaments were used with gases such as hydrogen, nitrogen and air. Total power loss from a filament was determined to be mainly via radiation and convection. End losses through the lead wires, being rather small, are excluded from discussion here. Radiation losses could be calculated easily but the convection loss needed to be understood. The convection loss from a filament was found to be dependent on what Langmuir termed the "shape factor" of the filament as well as the thermal conductivity of the fill gas. Very close to the filament in an ordinary household lamp the heat is conducted away from the filament, just as heat is conducted along a poker in a fire. Out away from the filament the heat is than convected away. The region about a filament through which conduction takes place is a few millimeters thick. In the lamp industry this region is referred to as the Langmuir film or sheath. The air

Nobel Laureate Dr. Irving Langmuir (1881-1957)

flow about a vertically or horizontally oriented filament can be visualized by a schlieren technique. It should be mentioned that this boundary layer, through which conduction takes place, is present about any body that is heated relative to the surrounding gas atmosphere. For example, one can sometimes detect the boundary layer and shimmering air flow about a heated furnace when sunlight casts shadows of the furnace on a wall.

Langmuir's study of conduction-convection losses from heated bodies was most extensive and detailed. The conclusion drawn from that work was that the long filament length had to be effectively shortened to reduce the losses. This was accomplished by coiling the tungsten wire. Successful lamps have been made by double coiling and triple coiling. The conduction-convection loss is relatively insensitive to the diameter of the resulting coil. While coiling had the effect of reducing gas losses, another important effect occurred. Bulb blackening was reduced considerably because some of the diffusing tungsten atoms that evaporated during normal lamp usage would diffuse back to the filament and deposit themselves on the filament instead of the inside of the glass bulb. Another effect also occurred because of the gas filling, which initially was nitrogen but eventually was a nitrogen-argon mixture. Lamps were filled roughly at 80% of atmospheric pressure. During normal operation of household lamps the operating pressure then rose to about one atmosphere (750 Torr). After some of the evaporating filament atoms reached the outer boundary of the Langmuir layer the convection currents would carry most of them upwards to deposit on only a portion of the glass surface rather than the entire bulb, as in vacuum lamps. This reduced the overall light output loss.

A view of the boundary layer about a vertically operated coiled-coil filament operating in a lamp with a cold fill pressure of 600 Torr xenon is shown below.

The gas loss from a vertically operated filament is less than that from one operated horizontally. Manufacturing considerations generally dictated when design changes could be implemented. For example, the change from a horizontally oriented coiled-coil filament in the 120-volt, 100-watt household lamp didn't occur until the mid 1950s.

A nice treatment of early so-called gas-filled lamps can be found in the book by Arthur A. Bright Jr., The Electric-Lamp Industry: Technological Change and Economic Development from 1800 to 1947. Start in the index, page 518, under "gas-filled incandescent lamps." Bright gave the following examples of "gas-filled lamps" where the year, inventor, filament material and gas type are given:

1840, W. R. Grove, platinum, air
1845, J. W. Starr, platinum, air
1856, C. de Changy, platinum, air
1859, M. G. Farmer, platinum, air
1872, A. M. Lodyguine, graphite, nitrogen
1875, S. A. Kosloff, graphite, nitrogen
1878, St. George Lane-Fox, platinum-iridium, air or nitrogen
1879, Sawyer-Man, carbon, nitrogen
1894, The Star Electric Lamp Co., carbon, heavy hydrocarbons in "New Sunbeam" lamp
1894, Waring Electric Co., carbon, low pressure bromine in "Novak" lamp
1901, A. E. G., carbon, low pressure carbon monoxide
1908, Hopfelt, carbon, mercury

Even if someone had developed a coiled carbon filament lamp employing an inert fill gas it would not have been successful because of the relatively high vapor pressure of carbon at operating temperatures. The situation required a ductile metal filament of low vapor pressure that could be coiled into a small space. Tungsten filled that requirement; however, another problem existed at that time that required its solution before Langmuir's lamp could be manufactured for all wattage ratings. Drawn tungsten wire, with its low vapor pressure, solved many of the design requirements but when it was first developed the wire tended to sag and offset. The work of Aladar Pacz about 1915, which resulted in a non-sagging and non-offsetting wire, assured the success of Langmuir's coiled tungsten filament nitrogen-filled lamp.