History Of The Incandescent Lamp - By John W. Howell And Henry Schroeder (1927)

You are reading: Chapter 4: The Vacuum, "Getters" and the Gas Filled Lamp




The vacuum was one of the elements of Edison's original lamp and still is in the majority of lamps made today. In the early days of lamp manufacture all lamps were exhausted on Sprengel mercury pumps, which consisted of a glass "fall" tube down which mercury was allowed to fall. The fall tube was connected at the top to a branch tube, one end of which was connected to the lamp to be exhausted, the other end being connected to the upper reservoir of mercury. The mercury trapped bubbles of air from the lamp and its weight forced the bubbles down and out of the end of the fall tube, which dipped into the lower reservoir of mercury. The mercury from the lower reservoir was pumped back to the upper by an Archimedes screw pump.

A very high degree of vacuum is necessary in a vacuum lamp, but, since it is impossible to produce an absolute vacuum, the degree of vacuum is measured by the pressure of the residual gases in the bulb. Atmospheric pressure at sea level is about fifteen pounds per square inch which is equal to the weight of a column of mercury about 760 millimeters high. The degree of vacuum is measured in microns, a micron being one-thousandth of a millimeter, and in modern lamps the degree of vacuum is often less than one micron mercury pressure, or about a millionth of that of the atmosphere at sea level.

In June, 1881, it took five hours to exhaust a lamp, each operator taking care of about fifty pumps, with one lamp on each pump. The chief difficulty was then and still is getting the moisture, in the form of water vapor, out of the lamp bulb. This moisture adheres to the surface of the glass and the glass must be heated to liberate it. No matter how hot the bulb is heated, more moisture will be liberated if the bulb is heated still hotter, so the bulbs must be heated during exhaustion, or just before it, hotter than they will ever become in use. In practice they are heated to about 300° C.


Mercury, dropping down the "fall" tube, trapped bubbles of air and so exhausted a lamp. This originally required five hours. Improvements reduced the time to thirty minutes.

After the moisture is liberated it must be removed from the bulb. The mercury pumps would not draw it out, so from the beginning it was absorbed by phosphoric anhydride which was held in a small glass cup attached to the pump. In the early days this phosphorus cup was not close enough to the lamp and the absorption of moisture had to take place through five or six inches of glass tubing. This was one reason for the long time required to exhaust lamps in 1881. Later this condition was improved, the dryer being put as close as possible to the lamp, about 2½ inches, and this shortened the time of exhaust a great deal.

The vacuum pump itself was much improved, larger tubing being used so that the pump required three times as much mercury to operate it. The contraction which limited the flow of mercury to the pump was changed from glass to iron. Glass contractions got dirty and gradually reduced the flow of mercury, but iron did not get dirty and kept the pumps working at full capacity. All these changes ultimately reduced the time required to exhaust a lamp to thirty minutes.

The copper plated filament connections and carbon paste connections liberated a good deal of gas when heated. Before the vacuum became good there came a stage in which it was conductive. In this condition, when the filament was burned at high temperature, this cross current, passing through the partial vacuum and creating a blue glow in the bulb, heated the filament connections red hot and drove the gas out. The carbon filaments themselves gave out very little gas when heated.

The condition of the vacuum in the Sprengel mercury pumps was indicated by the size of the bubbles of gas passing down the fall tube of the pump. As the vacuum improved, these bubbles got smaller and smaller until they could not be seen. This condition was known as a solid tube, the tube being filled with mercury with no bubbles showing in it. This was the indication of a good vacuum and the lamp was then sealed off. A solid tube indicated a vacuum of one thousandth of an inch (about forty microns) of mercury pressure or less. These mercury pumps were used from the beginning until 1896, when the Malignani chemical exhaust process was introduced.

Malignani Chemical Exhaust

Arturo Malignani was a home made engineering genius. He lived in the town of Udine, in the northern part of Italy, right at the foot of the Austrian Alps, where he built an electric lighting plant for the town and made his own electric lamps. He did not have mercury pumps and his mechanical pump would not make a good enough vacuum, about one millimeter (1000 microns) mercury pressure being the best he could get. He made the great discovery that, when he had this poor vacuum in a lamp, if he liberated phosphorus vapor in the lamp while the filament was burning at high incandescence and the bulb was full of blue glow, the glow suddenly disappeared and a high vacuum was produced.


This chemical method of improving a relatively poor vacuum, quickly obtained by a piston vacuum pump, to the high degree required in a lamp, reduced the time of exhaust from half an hour to less than two minutes.

Malignani painted the inside of the exhaust tube of the lamp with red phosphorus. Then, when the filament was raised to high incandescence and the bulb was full of blue glow, he closed the connection between the pump and the lamp and heated the exhaust tube enough to vaporize the phosphorus. This drove the vapor inside the lamp, the blue glow disappeared and a good vacuum resulted.

This invention revolutionized the art of lamp exhaustion. The General Electric Company bought Malignani's U.S. patent, and adopted this method of exhausting lamps. It enabled one operator with one pump to exhaust a lamp a minute with more uniform result than could be produced by the old method which required thirty minutes.


Before the expiration of the Edison lamp patent in 1894, the Waring Electric Lamp Company marketed lamps called "Novak Lamps" which, instead of having a high vacuum, had a small amount of bromine vapor in the bulbs, usually about 1½ millimeters (1500 microns) mercury pressure. This bromine very materially reduced the discoloration, and that which did occur was greenish and not black. It is believed that the carbon molecules thrown off from the incandescent filament combined with the bromine vapor to form the greenish compound on the bulb. The bromine was gradually used up and, after a few hundred hours burning, the lamp had good vacuum. This bromine was what is now called a "getter" and this was the first, though unrealized use of a getter.

When the Edison patent expired, the manufacture of this lamp was abandoned in favor of the high vacuum lamp. The Novak lamp was invented by John Waring, who supervised its manufacture. He was a man of unusual character and promise, and his death, which was due to an explosion in his laboratory, was deeply regretted by all who knew him.

The word "getter" is now applied to any active agent used inside the bulb, either to assist in getting a vacuum, or to improve the quality of the lamp, usually by preventing the blackening of the bulb.

The use of phosphorus to improve the vacuum, invented by Malignani, has already been described. The phosphorus was never called "getter" in connection with Malignani's chemical exhaust.

In 1908 or 1909, John T. Marshall invented the present day method of exhausting tungsten filament lamps without lighting the filament. He coated the filament and mount by dipping the mounts in a mixture of phosphorus and water. After the lamps were sealed off, the filament was burned at high incandescence; a blue glow appeared, and a good vacuum resulted.

Phosphorus is now used as a getter to assist in getting the vacuum in all vacuum lamps, it being applied to the filament as a coating. After sealing off the lamp, this coating of getter is vaporized by flashing the filament to a bright incandescence. At this time, a blue glow appears in the bulb for about a second. Disappearance of the blue glow is always associated with "clean-up", or reduction in pressure. The explanation of the formation of blue glow is somewhat as follows: Under the influence of the voltage applied across the filament, electrons emitted from the negative end of the filament are accelerated with appreciable velocity toward the positive end. If there are present sufficient residual gas molecules, a large number of collisions occur between electrons and gas molecules with the consequence that the molecules are ionized; that is, they are dissociated into an electron and a positively charged residue (positive ion). These separated parts naturally tend to recombine, and during the process of recombination, radiation is emitted, which is perceived as blue glow.

The action of phosphorus in getting the vacuum, or in "cleaning-up" the lamp as it is called, is not perfectly understood, but it is believed to act in two ways. The phosphorus vapor combines chemically with oxygen and water vapor, and the products of these combinations are carried to the bulb and held there. It is also believed that the phosphorus, which has condensed on the bulb under the conditions which exist when the filament is at intensive incandescence and blue glow is in the lamp, absorbs other gases with which it does not combine chemically, and holds them on the bulb. These gases may be subsequently liberated in their original condition if the lamp bulb is heated hot enough to vaporize the phosphorus.

Many experiments have been tried to determine the nature of the action of these getters, and some evidence, although it is not conclusive, has been obtained. The residual gases in vacuum lamps after sealing off are, on the average, about 25 microns. In gettered lamps, this pressure will be reduced to less than one micron when the lamp has burned for a few seconds, provided the applied voltage is high enough. In lamps below forty volts, the reaction is much slower, and may be of a different nature. There is evidence that the reaction of the getter with the residual gas is not predominantly chemical, since such getters as phosphorus, silica, aluminum or manganous oxide will, when applied to the filament as a getter, clean up such gases as hydrogen, nitrogen, carbon monoxide, carbon dioxide, oxygen, water vapor, and to a lesser degree, argon, at about the same rate, and to about the same residual pressure, regardless of which getter or gas is used. Chemical action probably takes place between phosphorus and oxygen, and also water vapor. In addition to any action which may take place in the vapor phase, probably the most important reaction takes place on the bulb wall. When the lamps with phosphorus, silica, aluminum or manganous oxide getter are burned, the getter having been thrown to the bulb wall and the gas having been cleaned up, a second clean-up can be obtained by admitting a small quantity of gas and again flashing the lamp. Only a limited quantity of gas can be made to disappear on a single coating of getter in this manner. It is believed that phosphorus has a continuing action during the life of the lamp in case any water vapor is liberated in its interior.

The general conclusion is that the vacuum clean-up must be largely an absorption, by the getter on the bulb wall, of gases activated in some manner by an electrical discharge of sufficiently high voltage. It is believed that the action of the phosphorus getter is solely to get and keep the vacuum, as it acts on gases only.

There is another kind of getter used in all tungsten filament vacuum lamps, the action of which is not to get or maintain the vacuum, but is to reduce the blackening of the lamp. This getter is usually a fluoride, and is now applied to the filament as a coating. In practice, it is mixed with phosphorus, and the mixture is put on the filament.

When the lamp is flashed after exhaustion the getter vaporizes and condenses on the bulb, where it remains. During the life of the lamp, molecules of tungsten fly from the hot filament to the bulb and slowly blacken the bulb, but the coating of getter reduces this blackening very much.

The reaction to prevent blackening by fluoride is shown to take place on the bulb wall by placing a small piece of glass, about the size of a dime, on the inside of the lamp, and letting it protect one small spot on the bulb from a deposit of getter while the lamp is being flashed. The small piece of glass is then removed to some other location and the lamp burned for several hours. The spot where the glass rested when the lamp was flashed will blacken much more rapidly than the remainder of the bulb due to the absence of getter at that point. This black spot has sharply defined edges, and has the same shape and size as the protecting glass. One explanation of prevention blackening is due to some optical experiments at the Philips lamp factories at Eindhoven, Holland. The result of these experiments indicate that the particles of vaporized tungsten are held in a sort of colloidal suspension in the getter, and in this condition will not absorb so much light as if allowed to agglutinate and form a continuous layer. Some engineers think that this is due, in part at least, to chemical action.

A third class of getters was formerly used to a considerable extent, but is little used now. These getters were placed in a cavity in the glass filament support, where the heat caused them to give off a continuing supply of gas. Different materials have been used, some of which give off gases, such as oxygen, or a halogen gas, which combine with the vaporized tungsten to form a light colored deposit on the bulb. Barium chlorate is an example of an oxygen getter. This is used today in vacuum series lamps, which is the only use today of a getter of this class.

A little oxygen has a beneficial getter action in either carbon, tantalum or tungsten lamps. In all these lamps, oxidized copper supports which slowly liberated oxygen, gave better results than supports made of non-oxidized metal.

Other getters of this third class yield gases which are halogen compounds, and which have a regenerative action, combining with the vaporized tungsten, carrying it back and depositing it on the filament.

Tungsten oxychloride is a regenerative getter. Potassium thallium chloride is another. The latter was used commercially for a long time on some lamps. When the temperature condition of the getter was right, the lamps remained clear and did not change in resistance or candle power. But temperature conditions varied so much that the lamps gave variable results, and the getters are no longer used.

Getters are also used in gas-filled lamps. Phosphorus is used in gas-filled lamps, being applied to the filament as in vacuum lamps, and, when vaporized, combining with the water vapor and oxygen in the lamp, thus purifying the gas with which the lamp is filled. Carbon and carbon compounds are also used as getters in gas-filled lamps. They are applied to the filaments in the same manner as a phosphorus getter. Both getters take care of water vapor and oxygen— the phosphorus possibly has a continuing action during the life of the lamp. The action of the carbon, however, will cease when all of the carbon has been removed from the filament.

Barium ozoamid is used by one European lamp manufacturer in gas-filled lamps. It prevents blackening by the liberation of nascent nitrogen which is very active in combining with vaporized tungsten and water vapor. This getter is decomposed by heat when, after the lamp has been exhausted, it is flashed high in the gas.

Skaupy's Getter

Franz Skaupy, an Austrian chemist, invented the use of getters in metal filament lamps to lessen the blackening of the bulb caused by the deposit of the filament material on the bulb. Skaupy's idea was to use chemicals in the lamp which would convert this black deposit into one of a lighter shade, so that less light would be cut off from the filament during the life of the lamp. The chemicals as used by Skaupy did not improve the vacuum; on the contrary a gas was purposely formed in the bulb as the lamps burned. This was opposite to what lamp engineers considered desirable, as the belief was that nothing should be done to impair the vacuum. Skaupy's invention is, therefore, all the more meritorious. He applied for a U.S. patent which was granted in November, 1915.

In Skaupy's getter certain chemical compounds of the halogen group of elements (fluorine, bromine, iodine, chlorine) are put inside the bulb and remain there after the manufacture of the lamp has been completed. These compounds will break up when heated, releasing some of the atoms of the halogen element used, the rate of release depending on the temperature and pressure.

For example, with thallic chloride (which was commercially used) chlorine gas is evolved which will combine with tungsten, forming tungsten chloride, which is lighter in color than tungsten itself. If the chlorine gas is evolved at the proper rate by heating the thallic chloride to the proper temperature, it will combine with the tungsten which evaporates from the filament as the lamp burns, without attacking the tungsten filament itself. If it evolved too slowly, the deposit will contain black tungsten, as an insufficient quantity of chlorine gas is evolved to combine with all of the tungsten which vaporizes. If evolved too fast, the tungsten filament itself will be attacked, thereby shortening the life of the lamp. It is, therefore, important that the thallic chloride getter be kept at a given temperature as the lamp burns.

This was accomplished by inserting the getter in a cavity in the end of the glass arbor supporting the filament anchors, the upper end of the arbor being made of glass tubing. The getter was held in place by glass wool, and the end of the tube constricted to prevent the getter and wool from dropping out.

In actual practice a double halogen compound was used, potassium thallic chloride, a chemical combination of two salts, potassium chloride and thallic chloride. Thallic chloride readily absorbs water vapor and was apt to do so before it was put in the lamp. Water vapor is very detrimental in a lamp, as it causes the lamp to blacken rapidly. The double chloride compound does not easily absorb water vapor.

The use of getters was particularly desirable in the larger sizes of lamps, since such lamps blacken to a greater extent during their life than those of the lower wattages. This is because the relation of bulb surface to filament surface becomes smaller in the higher wattage lamps, causing a denser deposit on the bulb. Skaupy's getter was, therefore, used on the 100-watt and larger sizes of 110-volt types of lamps. It was also found that his getter was so active, that it was impractical to use it in lower wattage lamps, as it could not be prevented from attacking the filament.

It is an expensive manufacturing proposition to make a hollow arbor to hold the getter. This method and its location was found to be the only practical one with this getter in order that it should reach the proper temperature. Many investigations were made to see if other chemical compounds could be used with simpler manufacturing construction, or to permit taking advantage of similar chemical reactions in lower wattage lamps.


The chemicals called "getters", in the hollow end of the glass rod supporting the filament, vaporized as the lamp burned, reducing the blackening of the bulb.

Dr. Fink's Potassium Iodide Getter

Dr. Colin G. Fink invented a getter which was used in 1912 in the smaller sizes of lamps, namely those of 15 to 40 watts for 110-volt circuits. It consisted of potassium iodide mixed with water, a drop of which was put on the end of the glass arbor holding the filament, after which the drop was dried by baking the mounted filament in an oven before the mount was sealed in the bulb.

Potassium iodide is not as active as thallic chloride, but was commercially suitable for the lower wattage lamps, although it was impractical for use on 60-watt and larger lamps. During the life of the lamp, the iodide is decomposed by the heat from the filament, iodine vapor being released which combines with the vaporizing tungsten, forming a light colored deposit in the bulb.

Needham's Getter

Harry H. Needham, of the General Electric Company, invented a getter which was more active than Dr. Fink's, but less so than Skaupy's, and was suitable for 25- to 60-watt 110-volt lamps in which it was used. A patent was applied for in October, 1912, and granted in June, 1916, covering the method of application and use of double halogen salts, such as cryolite, which is a combination of sodium and aluminum fluoride, and which was commercially used.

The double salt was mixed with a binder such as water glass, a drop of which was put on the anchors supporting the filament, care being exercised that the getter did not touch the filament, as otherwise it would cause the filament to fail at the point of contact.


This method of application greatly simplified the lamp construction. The chemicals used made the getter practicable in smaller sizes of lamps.

The getter was then dried by baking the filament mounts in an oven before they were put in the bulb. During the life of the lamp, the heat from the filament decomposed the cryolite releasing fluorine gas, which combined with the vaporizing tungsten, forming a light colored deposit.

Red phosphorus was mixed with this getter, the lamp being exhausted and sealed off without lighting the filament. After the base had been put on, the lamp was slowly lighted for the first time by gradually increasing the voltage applied to it. This is called "flashing", and by this means the red phosphorus in the getter became heated and vaporized, improving the vacuum in accordance with Malignani's scheme as previously described.

Friederich's Oxygen Getter

Ernst Friederich, a German, invented a getter consisting of an oxygen compound, barium chlorate being commercially used. He applied in June, 1913, for a patent in this country, which was granted in September, 1917. The barium chlorate was later mixed with manganese dioxide which acted as a catalyzer; that is, it assisted in breaking up the barium chlorate so that it would give up oxygen gas when heated. Red phosphorus was also mixed with this getter as in Needham's scheme, but the getter was located in the hollow end of the glass arbor supporting the filament anchors as in Skaupy's construction. It was used in lamps of 150 watts and above and is now used in vacuum series lamps. The oxygen gas, released by the heat of the filament, which decomposed the barium chlorate, combined with the vaporizing tungsten, forming a light colored deposit.

Gill's Invisible Getter

Frederic W. Gill, of the General Electric Company, applied for a patent in June, 1915, which was granted in November, 1918, covering a getter which could be applied directly to the filament. The first getter commercially used, superseding Needham's getter, was ordinary table salt (sodium chloride) dissolved in water and sprayed on the mount. Red phosphorus was included in the getter as in Needham's scheme. When the lamp was lighted for the first time, the sodium chloride immediately vaporized from the filament and condensed on the walls of the bulb in an invisible layer.

Care had to be exercised not to spray the mount too much, as too great an amount of the solution would cause the bulb to become iridescent. This led to the development of another method now used, also covered by patent, of putting the getter on the filament in such a way that the amount put on could be more accurately controlled.

A fluid mixture consisting of either sodium iron fluoride or cryolite (sodium aluminum fluoride) is made with red phosphorus and gun cotton, the latter dissolved in alcohol, ether and amyl-acetate. The drawn tungsten wire, before it is put on the anchors, is run through this paste, which forms a coating on the wire. The coated wire is then run through a plain solution of gun cotton to give it a further protective coating which dries and hardens on the wire.


Dr. Irving Langmuir joined the staff of the Research Laboratories of the General Electric Company at Schenectady in 1909, while they were in the midst of Dr. Coolidge's invention of ductile tungsten and its application to incandescent lamps. One of the troubles, as has been explained, was the curious phenomenon of "offsetting", a tendency for the filament to divide into little sections of short length which slid sidewise over each other. Dr. Langmuir undertook a study of this phenomenon, which led him to a study of the gas given off by a tungsten filament at very high temperatures.

The necessity of a high degree of vacuum appeared to be of even greater importance in a tungsten than in a carbon filament lamp. The candle power given out by a lamp during its life decreases as it burns, the decrease being mainly due to blackening of the bulb caused by material leaving the filament, depositing on the inner surface of the bulb, and thus shutting off the light emitted by the filament. It was believed that the blackening of the bulb might be caused by slight traces of gases in the bulb, the rapid motion of the gas molecules striking the surface of the filament causing its disintegration. Some engineers thought that the disintegration might be due to chemical or electrical action of the gases and others thought it might be due to true evaporation.

Study of the Residual Gases in a Vacuum Lamp

It appeared, therefore, that a study of these traces of gases in the bulb was desirable, as their elimination might improve the lamp. Attempts to improve the lamp by obtaining a better vacuum than usual had not been very successful, and while it appeared that in operating a filament at its normal temperature, the vacuum gradually improved to a point better than that directly obtainable by any known method of exhaust, there were clear indications that undue blackening was caused by imperfect exhaust. The faint traces of residual gases were in such minute quantities that their pressure was less than that possible of measurement with the most sensitive vacuum gauge. The failure to improve the lamp by a new method of exhaust might mean that the vacuum had not been improved, since the pressures were too low to measure.

The residual gases in the bulb, after it has been exhausted to about one micron or less, were found to consist of water vapor, oil (hydrocarbon) vapors, carbon monoxide, carbon dioxide and hydrogen. By operating a filament above its normal temperature, more gases come out, and it was found that these gases not only came from the filament but the heat caused gases to come out of the anchors, leading-in wires and the glass as well. The actual gases from the filament were found to be small in quantity, as later work showed that the apparently inexhaustible supply of gas from within the filament was produced by its decomposing the water and hydrocarbon vapors present at extremely low pressures in the bulb. The actual gases from within the filament were mainly carbon monoxide and small amounts of hydrogen and carbon dioxide. The gases from the anchors and leading-in wires were also small in quantity. If the bulbs were externally heated so as to obtain higher temperatures than that received from the filament, large quantities of gases were driven out from the glass. These gases were mainly water vapor, a small amount of carbon dioxide and a still smaller amount of nitrogen.

The determination of these gases was a great achievement, as it was necessary for Dr. Langmuir to devise special apparatus to make qualitative and quantitative analyses for the determination of five different gases from but one cubic millimeter of total volume. Heretofore, it had been impossible to make determinations when such small quantities were involved.

Small quantities of various gases, up to about a tenth of a millimeter pressure, were then put into lamps to study their effect. Hydrogen was found to dissociate, that is, the molecules of hydrogen broke up into their two atoms, in which condition the gas is chemically very active. Dry hydrogen did not have the slightest tendency to blacken the bulb. Oxygen combined with the hot filament, forming an oxide which coated the bulb with an invisible layer, but it did not cause blackening. Nitrogen did not attack the filament, but it combined with the tungsten which evaporated from the filament, changing the deposit from black to brown. Carbon monoxide behaved almost exactly like nitrogen and thus could not be responsible for blackening. Carbon dioxide attacks the filament, producing an oxide of tungsten, the carbon dioxide reducing to the monoxide, but without blackening the bulb.

Water vapor, even at very low pressures, was found to produce blackening. This was surprising, as neither of its constituents, hydrogen and oxygen, acting alone, produces blackening. The explanation seems to be that the water vapor, coming in contact with the hot filament is decomposed, the oxygen combining with the tungsten which deposits on the bulb. The chemically active atomic hydrogen formed attacks the tungsten oxide deposit, and reduces it to metallic tungsten, forming water vapor again. This cycle may be repeated indefinitely, so that a small quantity of water vapor will quickly blacken the bulb.

The effects of many other gases and vapors were studied, among which were chlorine, bromine, iodine, sulphur, phosphorus, phosphine, hydrochloric acid, methane, argon, etc., but in no case did these gases produce blackening, provided great care was taken to have them extremely dry. The behavior of argon was interesting. At pressures above five microns and below one micron, a glow occurs in the bulb, current flowing from one filament leg to the other through the gas. This so-called "Edison effect" caused the bulb to blacken rapidly. The small amount of argon which might exist in an ordinary lamp could not, however, produce such blackening.

This study led Dr. Langmuir to the conclusion that if the blackening of the bulbs of ordinary lamps was caused by imperfect vacuum, it must be due to water vapor. He devised new methods of producing extremely high vacua, improving the vacuum from a millionth to much less than a billionth of an atmosphere. Extra precautions were taken to remove all traces of water vapor and, to make sure that water vapor was not evolved by the bulb becoming hot, lamps were even run with the bulbs completely immersed in liquid air during their entire life.

The unexpected result of his work was that with all these precautions, the lamps were not materially better than the best lamps regularly made in the factory. There seemed to be no hope of improving the lamp by getting a better vacuum, the vacuum in the ordinary lamp being good enough. The investigations did show however, that, excepting water vapor, the presence of gas in small quantities in tungsten filament lamps did not seem to cause blackening and that the only one of the suggested causes of blackening which had not been investigated was the true evaporation of the filament. This he probably never would have discovered if he had not endeavored to find an explanation for the various phenomena found, rather than by trying to look for a definite object.

To test out the theory that true evaporation is the cause of blackening, Dr. Langmuir made many experiments to determine the rate of loss of weight of tungsten filaments operated at various temperatures. The actual results agreed remarkably well with the theoretical figures, which indicated that blackening of well-made tungsten filament lamps is caused by true evaporation of the filament.

Introduction of Gases at Atmospheric Pressure

It was possible that the presence of a chemically inert gas inside the bulb would reduce the rate of evaporation of the filament provided the phenomenon was simply one of evaporation. This is somewhat similar to the effect air pressure has on the boiling point of water. At sea level, water boils at 212 deg. F.; at high altitudes where the air pressure is less, the boiling temperature is less. Thus if a gas pressure were put in the lamp it might retard the evaporation of the filament, though it had usually been found that the presence of a considerable amount of gas caused an increase in the rate of disintegration of a heated metal. Dr. Langmuir had shown that low pressures of gases, except water vapor and argon, did not produce blackening of the bulb, and, therefore did not produce disintegration in the ordinary sense, and that hydrogen, nitrogen, argon and mercury vapor seemed chemically inert towards tungsten at high temperatures.

To test this out, a tungsten filament lamp was made which was filled with carefully dried and purified hydrogen at atmospheric pressure. The filament was run at the same temperature as that of vacuum lamps operating at one watt per candle. The loss of heat, due to its conduction away from the filament by the gas, was so great that 17 watts were required for each candle power (less than 0.6 lumens per watt) actually produced in this lamp. The heat is conducted away from the filament by its contact with the gas, on the same principle that causes the handle of a poker to become hot when the other end is put into a fire. Furthermore, the heated gas rises, circulating in the bulb and forming convection currents, thereby rapidly transferring the heat to the upper part of the bulb. This is why so much more electrical energy (watts) had to be put into the filament to maintain it at the same temperature as that obtainable in vacuum, where these heat losses do not occur.

This hydrogen filled lamp burned for more than 360 hours without showing any blackening of the bulb, but the loss of heat was so great, and so much more electrical energy was required to maintain the proper temperature, that it was impractical from a commercial standpoint. Subsequently it was found that while the heat conductivity of hydrogen is high compared with other gases, the amount of electrical energy required to operate the lamp was abnormally great, because at high temperatures the hydrogen molecules break up into their two atoms (as Dr. Langmuir had previously discovered), absorbing an added amount of electrical energy.

Experiments were then tried with tungsten filaments in mercury vapor at atmospheric pressure and the heat loss by convection was found to be extremely small in comparison with hydrogen, so small that the filament could be operated for about a minute at 21½ lumens per watt. The experiments showed that the presence of mercury vapor very greatly retarded the rate of evaporation of the filament. Nitrogen at atmospheric pressure was next tried and found to be entirely inert towards the high temperature tungsten filament. Comparatively so little heat was taken away from a large diameter filament operating close to its melting temperature that it could be operated for a moment at 22 lumens per watt. At the melting temperature of tungsten, it would theoretically give an efficiency of about 25 lumens per watt in vacuum. The rate of evaporation of tungsten at high temperature in nitrogen was also found to be less than in vacuum.

A slight increase in the temperature of a filament, which requires but a small increase in electrical energy (watts) will make a great increase in the amount of light it gives. This, however, is done at a sacrifice to the life of the lamp. The rate of evaporation of the filament at a given temperature having been found to be less in gas than in vacuum, the next thing to be determined was whether or not the filament could be operated in gas at a higher temperature (and so obtain more light for the same life) than possible in vacuum and yet not require more watts for the actual candle power obtained. In other words, was it possible that a gas-filled lamp, having the handicap of large heat losses by convection, could be made more efficient than a vacuum lamp not having this handicap, both lamps having the same life. A careful study was therefore undertaken of the laws of heat convection from filaments at high temperatures in various gases, since the knowledge on this subject was extremely meager.

Study of the Dissipation of Heat from Hot Wires

Dr. Langmuir had studied abroad in 1903-5 and had made some researches on the effect of highly heated platinum wires in dissociating steam, and other vapors and gases. He had become interested in the laws governing the dissipation of heat from hot wires, and when he returned to this country he continued this investigation, but had little opportunity to experiment until he entered the Research Laboratories of the General Electric Company. Some experiments he conducted at Pittsfield, in 1911 on electric heating devices broadened his knowledge.

Further experiments were then made to determine the laws of heat convection, by operating platinum wires in air, carbon dioxide, and hydrogen, and tungsten wires in hydrogen, nitrogen, mercury vapor and argon. He found that the heat loss varies with the temperature, according to a simple function of the heat conductivity of the gas, and also varies with the diameter of the wire according to a rather complicated formula. From this he derived an equation by which he could calculate the heat losses from a wire at any given temperature in various gases. This showed that the heat lost by convection increases (around high temperatures) rather slowly with increases in temperature in the case of nitrogen and mercury vapor, but increases very rapidly in the case of hydrogen. It also showed that the heat loss from very small wires, such as those of about a thousandth of an inch in diameter, was not very different from wires of several times this diameter. This was most unexpected; one would think that if the size and, therefore, surface of the wire were doubled, the rate at which heat was lost would be doubled, but this is not true.

Dr. Langmuir's explanation of this is that a wire or filament, in the case of a lamp, seems to hold a layer of hot gas, about a sixth of an inch in thickness, which adheres to it, the thickness of this gas film being independent (within certain limits) of the diameter of the filament. Halving the diameter of the filament, therefore, does not halve the thickness of the filament with its gas film. For example, a filament two-sixths of an inch in diameter would, with its gas film, have a diameter of four-sixths of an inch. A filament of one-sixth of an inch in diameter, 50 per cent less than that of the former, has a diameter of three-sixths of an inch with its gas film which is 25 per cent less than that of the former (four-sixths as compared with three-sixths). Hence the effective cooling surface of a thin filament is relatively greater than that of a thick filament. From this it will be seen that with small wires more heat is proportionally lost than with large wires.

The increase in temperature necessary, due to the presence of gas in the bulb at about atmospheric pressure, in order that a gas-filled lamp could operate at the same efficiency as a vacuum lamp, would, therefore, be very much greater with thin filaments than with thick ones. Thus in nitrogen, Dr. Langmuir estimated that a filament of 1.1 thousandths of an inch in diameter (the size of tungsten filament of a 25-watt, 110-volt vacuum lamp) would have to operate at about 2600 deg. C. to give nine lumens per watt, the present efficiency of the 25-watt vacuum lamp. The 25-watt vacuum lamp now operates at about 2050 deg. giving a life of a thousand hours, and if operated at 2600 deg. would last about half an hour. It will be seen, therefore, that the reduction in rate of evaporation (and consequent increase in life) due to the gas would have to be very great to overcome the handicap imposed by the gas in lamps having small diameter filaments.

On the other hand, a filament of 13 thousandths of an inch in diameter, which would be the size of the filament in a 1000-watt, 110-volt vacuum lamp, if such a lamp were made, would only have to operate in nitrogen at 2300 deg. as compared with 2200 deg. C. in vacuum for the same efficiency. A 1000-watt, vacuum lamp operated at 2200 deg. would give a thousand hours life and if operated at 2300 deg. would last about sixty hours. Thus with thick filament lamps the handicap of the gas should not be so great.

These calculations did not prove that the introduction of gas would produce a better lamp, but indicated that if it were possible at all, it could be done more easily by using large diameter filaments. There was nothing to indicate how great the reduction in evaporation would be, so that it would have to be found by experiment, but the calculations showed in what manner the experiments should be conducted.

Experimental Gas-filled Tungsten Filament Lamps

Dr. Langmuir made two sets of thick filament lamps, one set operating in nitrogen at atmospheric pressure and the other set in vacuum. In both sets the filaments were operated at the same efficiency in order that the life results could be compared.

The nitrogen-filled lamps were failures. This was very discouraging and probably to the ordinary experimenter would have ended the investigation. But Dr. Langmuir had built up a theory that the nitrogen-filled lamps should be better, his former experiments showing that the evaporation of the filaments in gas was less than in vacuum, although its extent had not been determined. It seemed as if the extent should be great enough to overcome the handicap of the extra heat losses due to the gas, provided thick filaments were used. This faith encouraged him to continue his research, but before trying the experiments again he carefully examined the lamps he had tested to see if he could find any clue that might show him the cause of their failure.

He noticed that the deposit of evaporated material from the filament was located at the upper part of the bulb where it was expected to be, having been carried there by the circulating currents of gas in the bulb, but the deposit was black instead of brown. He had found in previous experiments with nitrogen that the otherwise black deposit of tungsten was changed to brown, owing to the formation of tungsten nitride. It seemed strange that this had not happened in these lamps, so he concluded that there must have been some trace of water vapor in the nitrogen gas which was responsible, in spite of the extraordinary precautions he had taken to prevent its presence.


This lamp, invented by Dr. Irving Langmuir, was twice as efficient in the larger sizes as the vacuum lamp. The bulb was filled with nitrogen gas at about atmospheric pressure. The filament was coiled.

He then repeated the experiments, taking still greater precautions to eliminate any water vapor, and this time his experiments were successful. The filaments he used were relatively very large, two to four hundredths of an inch in diameter, requiring from 20 to 60 amperes of current. Such lamps for 110-volt circuits would consume from about 2000 to 6000 watts, and would be very large compared with the ordinary vacuum tungsten filament lamps of 25, 40 and 50 watts used in the home, and large even when compared with the biggest vacuum lamp then made for commercial lighting, which consumed 500 watts.

Dr. Langmuir then conceived the idea that the effect of a large filament might be obtained by properly coiling a small one. In designing such coiled filaments, it was evidently desirable to wind the filament on as large a mandrel as possible to obtain the advantage of the large diameter. It was also desirable to have the coils as close together as possible. Tungsten is a relatively soft material at the operating temperatures of these lamps. If too large a mandrel were used, the weight of the filament would pull out the helix very materially in a few hours, so that the heat lost by convection would be increased. This sagging of the wire might also allow the lower turns of the coil to touch each other and short circuit, so the spacing between turns of the coil must not be too small. Careful experiment showed that certain mandrel sizes and spacings gave the best results. He then was able to make a gas-filled lamp, taking a little less than ten amperes, and consuming 1000 watts on 110-volt circuits, which was twice as efficient as a vacuum lamp for the same life. Further experience made it possible to produce a 750-watt lamp and these lamps were put on the market late in 1913. Dr. Langmuir applied for a patent in April, 1913, which was granted in the same month of 1916.

In order to distinguish the vacuum from the gas-filled lamp, the former is called a MAZDA B lamp and the latter a MAZDA C lamp. If these designating letters after the trade mark MAZDA had been desirable at the time the pressed filament lamp was being commercially made, the pressed filament lamp would have been known as a MAZDA A lamp.


When Mr. Edison visited the Research Laboratories at Schenectady in 1922, Dr. Langmuir showed him a 30,000-watt lamp he had made for experimental purposes. This is the largest lamp ever made, giving 100,000 candle power.

Exhaustion of Gas-filled Lamps

The moisture exhaustion problem is present in the gas filled lamp to as great, if not to a greater, degree than in the vacuum lamp. It is just as necessary to get rid of this moisture in the gas-filled lamp and it is more difficult because the blue glow of ionization which is such a great help in clearing up the moisture with phosphorus in the vacuum lamp does not appear in the gas-filled lamp. Other means, which are not so simple and easy as the clean up with phosphorus, must be used to get rid of the moisture. Washing out the moisture with dry gas is the most practical and is now used in regular factory practice. Several washings are necessary; dry air can be used for the first washings and dry nitrogen for the later ones.

High vacuum pumps are not necessary in exhausting gas-filled lamps because the washing out removes all the air and other gases and vapors without requiring a high vacuum at any time. The pumps used in this work have large capacity and produce a vacuum of about two-tenths of an inch (about 800 microns). When these lamps are sealed off they contain a sufficient amount of argon (with about 15 per cent of nitrogen) to make the pressure inside the lamp equal to atmospheric pressure when the lamp is burning.

Although phosphorus does not clean up the moisture in a gas-filled lamp as it does in a vacuum lamp, it has a good effect in taking care of the moisture which is left after the lamp is sealed off. It is put on the filament in gas-filled lamps as in vacuum lamps. Carbon and carbon compounds are also used, these and phosphorus possibly having a continuing action during the life of the lamps in taking care of water vapor and oxygen. The action of carbon, however, will cease when all of the carbon has been removed from the filament.

Previous Attempts to Make Gas-filled Lamps

Mention has been made that several Russian scientists had attempted fifty years ago to make lamps having a graphite burner operating in nitrogen gas. In 1878-9, Sawyer had tried the same thing, as has been stated, and failed. Even Edison had tried the use of nitrogen in the experimental lamps he made in the early eighties, after he had invented his practical vacuum lamp, and he also failed.

Edison knew that nitrogen would cool the filament, and tried to compensate for this by using a filament of smaller cross-section. He did not know why he failed, but found out that his gas-filled lamp lasted only one twentieth as long as a vacuum lamp at the same efficiency. Even after Dr. Langmuir's success, the Research Laboratory of the General Electric Company was unable to produce a gas-filled carbon lamp as good as a vacuum carbon lamp.

The "Novak" lamp, previously mentioned, which was made for a while in 1892, and which contained bromine gas at a pressure of about two one-thousandth parts of the atmosphere, cannot be construed as a gas-filled lamp. The Courts decided it was a vacuum lamp and therefore infringed upon Edison's basic vacuum lamp patent. Gettered lamps as originally made, in which a slight trace of gas was generated as the lamps burned, cannot be said to be gas-filled lamps either as the vacuum in such gettered lamps is at least one-thousandth part of atmospheric pressure, whereas the gas in the lamp invented by Dr. Langmuir is at about atmospheric pressure.

In this connection Dr. Langmuir found that there was no material advantage in having the gas pressure in the bulb much greater than that of the atmosphere. Even if it were desirable, there might be danger of the lamp's exploding. The gas is put in the bulb at slightly less than atmospheric pressure, so that when the lamp is lighted and becomes heated, the gas expands to a pressure about equal to that of the atmosphere.

Commercial Developments of the Gas-filled Lamp

The first commercial lamps, those of 1000 and 750 watts for 110-volt circuits, were made with round bulbs. The circulating currents of gas in the bulb in rising made the base quite hot, the heat being conducted to the socket holding the lamp. In order to lower the temperature of the base and socket, the bulb shape was changed by putting a tubular glass neck on the upper part of the bulb, a mica disk keeping the gas from circulating in the neck. The simpler shaped straight sided bulb was adopted soon after, which later was changed to pear shape.

As the art of making MAZDA C lamps progressed, it became possible to make smaller sizes. In July, 1924, 500 and 400-watt lamps for 110-volt circuits were developed, these lamps, on account of their smaller diameter filaments, not being quite as efficient as the larger sizes. They were, however, considerably more efficient than the same size of MAZDA B lamps, which then disappeared from the market.


A glass neck was put on the bulb, a mica disk preventing the circulating hot gas from reaching the base.

Series MAZDA C lamps were also made which displaced the vacuum lamps formerly used. These (as well as the former vacuum lamps) were more efficient with the 6.6-ampere filament in the ordinary sizes used, so the 6.6-ampere circuit for street lighting became the standard.

Lamps for 220-volt circuits were developed but, of course, could not be made in as small a size as those for 110 volts, as the 220-volt filament is smaller in diameter than that for 110 volts for a given wattage. Concentrated filament lamps for projection service were also developed for such uses as floodlighting, motion picture projection, etc.


Straight sided bulb used, a mica disk deflecting the circulating hot gas away from the base.

The efficiency of the MAZDA C lamp is so high and the simplicity and convenience of the incandescent lamp is so great, that the carbon arc lamp was gradually displaced and has now practically disappeared from use. The only other forms of electric illuminants now in use are the magnetite arc lamp used in street lighting, and the Cooper-Hewitt mercury vapor arc, often used in photography. The magnetite arc gives a brilliantly luminous white light. The mercury arc is valuable in photography on account of the high actinic value of its light, to which the photographic negative is particularly sensitive.


Pear shaped bulb about as now used.

Still lower wattage MAZDA C lamps for 110-volt circuits were developed, the 200- and 300-watt lamps being put on the market in October, 1914. Argon gas with a small amount of nitrogen was and is now used on account of its lower heat conductivity, with consequently less cooling of the filament. The lamps are therefore more efficient and it is possible to produce lower wattage MAZDA C lamps which are more efficient than MAZDA B lamps of this wattage and voltage. While Dr. Langmuir had found that pure argon in a lamp is a conductor of electricity, so that current would arc across from one end of the filament to the other, the lamp thus short circuiting, it was also found that such conditions were eliminated by adding about fifteen per cent of nitrogen gas to the argon. Argon is one of the constituents of the air, but is present only in small quantities, about one-half of one per cent. The necessity for developing a process to obtain argon in reasonable quantities caused some time to elapse before the gas became available in sufficient amounts to make an argon-filled lamp commercial. This gas has made it practicable to make lamps consuming a current as low as half an ampere. On 110-volt circuits the 50-watt lamp is therefore available, the minimum wattage, of course, decreasing as the voltage decreases. Thus on 60-volt circuits, 25-watt lamps can be had; 15 watts on 30 volts; etc. This limit of half an ampere does not quite apply on very low voltages, as in such cases the filament is much shorter, and therefore the amount of heat conducted away by the leading-in wires becomes proportionally greater, so that the minimum size increases with very low voltage lamps.

On 6-8-volt automobile lighting circuits the 21 candlepower MAZDA C headlight lamp is now standard, it being a legal requirement to use this lamp in certain states. The lamp consumes about 2½ amperes.

It is uneconomical to use MAZDA C lamps of smaller sizes than those given above, because, while it is possible to make them, their efficiency would be no better than that of a vacuum lamp for the same life. They can be made to give a higher efficiency, but their life would be correspondingly shortened. As the art progresses it may be possible some day to make still smaller MAZDA C lamps which would be more efficient than the same wattage size of MAZDA B lamps and yet have the same life.