CHAPTER THREE

DEVELOPMENT OF FILAMENTS

In an incandescent lamp the current passing through the resistance of the filament heats it to an almost white heat and in this condition it radiates light. The hotter it is heated the more light it radiates; also the hotter it is heated the sooner it wears out. Edison decided that to be satisfactory a lamp should last 600 hours, so lamps were rated to operate at a temperature at which the filament would last 600 hours. As filaments were improved in quality the operating temperature was raised in such proportion that there would be no change in life. Each increase in filament temperature improved the efficiency of the lamp, causing it to give more light for each unit of electricity used, so from the beginning it has been the endeavor to improve the filament so that it could be safely operated at higher temperature.

The Carbon Filament

For about 26 years all incandescent lamps had carbon filaments made by carbonizing cellulose paper, bamboo or cotton. All cellulose is composed largely of carbon combined with other elements, principally hydrogen and oxygen. When cellulose is slowly heated in a closed furnace, away from air, it is decomposed, the hydrogen and oxygen and some of the carbon is driven out and the carbon skeleton remains. This carbon skeleton is the filament. It is very dense and hard and much like anthracite coal.

Edison's first commercial lamps had a filament of carbonized paper which was rather porous and fragile. When the lamp factory was started in 1880, the lamps were made with filaments of carbonized bamboo which was very hard and strong. Much had to be learned about carbonizing bamboo, it shrinks from 20 to 30 per cent during the process, and it must be free to shrink but must not be allowed to distort. If the shrinkage is too much restrained, weaknesses in the filament will result. The atmosphere surrounding the filament during the carbonizing and cooling must be a reducing atmosphere, free from air. The temperature of the carbonization, especially when the decomposition of the cellulose is going on, must be very slowly raised or the filaments will be stuck together because of too rapid distillation of the hydrocarbons.

CARBONIZING FURNACES

After reaching 600 deg. F. the temperature may be rapidly increased until the crucibles are white hot. As the crucibles containing the filaments cool down they must be surrounded by a reducing gas to prevent air reaching the filaments.

Edison sent several men all over the world to get samples of different bamboos. In the summer of 1880, William H. Moore went to China and Japan. He sent great bales of samples to Menlo Park and after careful tests, a certain variety and growth of Japanese bamboo called "Madake" was found to be the best. Moore was instructed to arrange for the cultivation and shipment of this, so he got a Japanese farmer to do it. The farmer displayed such ingenuity in fertilizing and cross fertilization that the product was constantly improved. It was used until 1894.

In December, 1880, John C. Brauner was sent to South America. He travelled over two thousand miles on foot and by canoe in the wilds of southern Brazil and secured a great variety of specimens of bamboo. None, however, was found to be superior to the Japanese bamboo then being used.

Another expedition was sent to Cuba and Jamaica, the trip taking two months. Three men explored the Florida swamps for five months. None, however, found samples as good as the Japanese variety.

A few years later (1887) two men, Frank McGowan and C. F. Hanington, went to Brazil and up the Amazon River for 2300 miles. There the two separated, McGowan exploring Peru, Ecuador and Colombia, Hanington went down the Amazon River again, up the La Plata River and through Uruguay, Argentine and Paraguay. McGowan's trip was particularly dangerous as he went through a comparatively wild and unknown country filled with hostile natives.

The last trip of this kind was made by James Ricalton who went completely around the world, the trip taking exactly one year. He was unable, however, to find a fiber better than that being obtained from Japan.

Clamps

Prior to 1881 the filament was fastened to the leading-in wires by delicate clamps. This is why the joint between the filament and leading-in wire is often called the clamp. From 1881 to 1886, this connection between the filament and the leading-in wire was copperplated. To keep the copper from being melted by the hot filament the ends of the latter were made large enough to radiate the heat and so keep the temperature down. The filaments were cut with these large ends on them; this increased the expense and trouble of making them and prevented any adjustment of length after they were cut.

Differences in dimensions due to shrinkage or cutting inaccuracies made these filaments quite different in voltage, so that in any lot of lamps made, the voltage of individual lamps would vary 15 or 20 per cent. To utilize these lamps, electric lighting plants were arranged to be operated at different voltages. Plants operating all the way from 95 to 125 volts were thus established all because it was impossible to make all the lamps of the desired voltage which was 110.

About 1886 carbon paste was adopted for making the connection between the filament and the leading-in wire. Since this paste joint would not melt or be injured by the hot filament, enlarged ends were no longer necessary on the filaments. This reduced the cost and simplified the manufacture of filaments, permitting any adjustment of their length which was desirable after carbonization. Although this somewhat reduced the variation in voltages of lamps made, the difference was still so considerable that each lamp had to be photometered to determine its voltage at its proper candle power.

The carbon paste used in these lamps was first made by mixing graphite and india ink, and later by mixing graphite with caramelized sugar and gum arabic. Paste for large size filaments was made of graphite, soft coal and coal tar pitch. The joints containing pitch were heated red hot before sealing the filament in the bulb to reduce the pasted joint to coke.

Treated Carbon Filaments

For over ten years the filaments used in all Edison lamps were carbonized bamboo. Other lamp manufacturers used an additional process called "treating" which was patented by Sawyer and Man. In this treating operation the filaments were held by clamps in a bottle which was connected on one side to a vacuum pump and on the other to a bottle containing gasoline. The vacuum pump first drew the air out of the bottle containing the filament and then drew gasoline vapor into it. Electric current was then passed through the filament, heating it to a very high temperature, the gasoline vapor in contact with the filament was decomposed and a layer of graphitic carbon was deposited on the filament. This process was capable of nice adjustment and gave the filament just the resistance desired. The graphitic coating also gave the filament a much better light radiating characteristic and considerably reduced the variation of voltages in the lamps. This patent expired in 1893 and after that Edison lamp filaments were so treated.

Later an automatic treating machine was developed by John W. Howell. In this machine the operator made no adjustments, only putting filaments in the bottle and taking them out. With this machine the gasoline was held in an underground tank outside the building, pipes bringing only the gasoline vapor indoors. Thus the danger of fire was removed, which was always present when each operator had a two-quart bottle of gasoline on the table beside her, as was previously the case.

The quality of the treated carbon filament depended upon the amount of gasoline vapor in the bottle and the temperature of the filament during treating. The amount of vapor in the bottle was measured by a "dose" bottle which was connected first to the vacuum pump, then to the gasoline vapor supply which filled it with vapor, and then to the treating bottle which had been exhausted of air and into which the dose bottle emptied its dose of vapor. The electric current was adjusted to maintain the filament at an approximately constant temperature during the treating operation, which required about 3½ seconds. During this time the resistance of the filament was reduced to one-third of its resistance before treating.

In this treating machine there were four treating bottles which were used in regular order. Stoppers, through which extended clamps which held the filaments and connected them to the electric current, fitted the bottles. When a filament was placed in a bottle, the latter was connected to a vacuum pump Which pumped the air out of it. Then the bottle was connected with the dose bottle which gave it the correct amount of gasoline vapor. Electric current was then passed through the filament, treating it to the proper resistance, at which point the current was cut off by an automatic device.

Air was then admitted to the bottle, the filament taken out and a new filament put in its place. All this, except putting the filament in the clamps and removing it, was done automatically by means of two flat rotary valves, invented by Mr. Howell. He also invented the mechanism which operated them.

TREATING CARBON FILAMENTS, 1893

The carbon filament was materially improved by coating it with graphite. This was done by heating the filament by passing current through it for a few seconds in gasoline vapor.

Squirted Cellulose Carbon Filament

In the Spring of 1888, Leigh S. Powell, an Englishman, developed a process he had originated for preparing cellulose for filaments. Sir Joseph W. Swan had some time previously invented a process along very similar lines. The two processes, although the same in principle, consisting as they did of projection of a solution containing cellulose through a nozzle into a setting liquid, were very different as regards the materials needed and the operations and apparatus employed.

In Swan's process nitro-cellulose (gun cotton) was dissolved in acetic acid.

SQUIRTING THE CELLULOSE CARBON FILAMENT, 1894

Cotton was dissolved in a hot zinc chloride solution, the syrup being squirted through a die into alcohol to harden the thread formed. This thread was then washed, dried, wound on forms to give it the desired shape, cut off in bunches and carbonized.

After squirting the solution through a small orifice into alcohol and washing the thread so formed, it was necessary to denitrate the thread before it could be carbonized. In Powell's process the danger of using and denitrating the gun cotton was eliminated. Cotton was dissolved in a hot zinc chloride solution to form a syrup which was squirted through a die into alcohol. The alcohol solidified the squirted thread and dissolved out some of the zinc chloride, the rest of the zinc chloride being washed out with several changes of water. The thread was then wound on drums and dried. It was then a strong, smooth, round, structureless, cellulose thread which was wound on forms to give it the desired shape, cut off in bunches, packed in crucibles and carbonized.

With this squirted cellulose, filaments of any desired length could be made, whereas with bamboo the length was limited to the distance between the joints of the cane and was not long enough for treated filaments of the desired dimensions. The treated squirted cellulose oval anchored filaments were the best carbon filaments ever made, their commercial adoption in this country beginning about 1894.

TREATED SQUIRTED CELLULOSE CARBON LAMP, 1894

The lower specific resistance of this filament required that its length be increased, the filament having a loop which was anchored to the stem.

Aggregate Improvement of the Carbon Filament

The lamps commercially sold in 1881 produced, when new, 1.68 lumens per watt. Lumens per watt is the term now used to express the efficiency of a lamp. A lumen is the amount of light in a beam having a cross-section of one square foot at a distance of one foot from a light source of one candle power. If a light source of one spherical candle power be placed at the center of a sphere of one foot radius, it will give one lumen on each square foot of surface of the sphere. As there are 12.57 square feet of surface on a sphere of one foot radius, one spherical candle power will give 12.57 lumens. Therefore, any light source will give 12.57 lumens for each spherical candle power; that is, the number of lumens given by any lamp is 12.57 times its spherical candle power. Carbon lamps were rated in horizontal candle power and the ratio of their horizontal candle power to their spherical candle power varied considerably. To determine their lumens, their spherical candle power must first be determined, which, multiplied by 12.57, gives their lumens.

The efficiency of 1.68 lumens per watt was steadily improved, first by improved methods of carbonizing and exhausting, then by surfacing the filament with asphalt, then by further improvements in vacuum production, including the Malignani chemical exhaust, and finally by the hydrocarbon treating process and the squirted cellulose filament. These improvements cannot be separately valued, but the carbon lamp of 1906, which is practically the same as the few now made, gave 3.4 lumens per watt. If the 1906 carbon lamp were burned at the same efficiency as that of the lamp of 1881, it would last 139 times as long, so it may be said that the quality of the 1906 lamp was 139 times better than that of the 1881 lamp.

The GEM or Metallized Carbon Filament

Dr. Willis R. Whitney, head of the Research Laboratory of the General Electric Company at Schenectady, had developed an electric resistance furnace. This consisted of a carbon tube, about three inches in diameter, inside of which articles to be heated could be placed. A heavy current of several thousand amperes was passed through the tube, heating it to a very high temperature, estimated to be about 3500 deg. C., which is about 500 deg. below the melting point of carbon and about 1650 deg. above the operating temperature of the carbon filament.

To give an idea of the terrifically high temperature reached by this electric furnace, the writer once looked directly into the open end of one of the tubes when it was fully heated and, when the eyes were adjusted to the task, held a 50-volt carbon filament lamp directly between the eye and the hot interior of the tube. The voltage on the lamp was then slowly raised and, when the voltage on the 50-volt filament was over 100 volts, the filament looked like a dark line on the background of the hot tube.

Dr. Whitney's original experiments were based on the idea that previous carbon filaments still retained small traces of such ash oxides as silica and alumina, substances which are not readily reduced by carbon at lamp temperature. It was evident that bulb blackening of carbon lamps might be due to the reaction of heat on carbon dioxide by which carbon monoxide and carbon are formed. The conditions of a lamp were such that this carbon could be deposited on the glass and the monoxide could react again with the filament to give more dioxide. In this way a steady blackening of glass could proceed indefinitely. The application of excessive temperatures to the filaments in vacuo could not succeed in removing the ash oxide because the carbon would itself vaporize too much, but it was evident that the filaments could not vaporize inside a highly heated carbon tube, while the oxides would be reduced by such excessive temperatures. The effect actually produced of changing the nature of the graphite coating in the treated filament was not anticipated.

The highest temperature reached during the time a carbon filament is carbonized is about 2700 deg. and is, therefore, considerably below that which Dr. Whitney was able to obtain with his furnace.

ELECTRIC RESISTANCE FURNACE, 1905

Dr. W. R. Whitney invented the GEM lamp which had a carbon filament subjected to the high temperature of an electric resistance furnace which he also invented. The GEM lamp was 25 per cent more efficient than the regular carbon lamp.

Having some filaments on hand, he decided to try the experiment of heating these already carbonized filaments to see if they could be improved. After subjecting them to the high temperature, he made them into lamps in his laboratory and life tested them. They gave surprisingly good results.

He ordered some filaments from Harrison to repeat the experiment, but these failed to give good results. A second lot of filaments sent him were no better. Upon investigation it was found that he had thought that the filaments, which he had on hand and which gave good results, were untreated filaments, whereas they were really treated filaments, so that he had ordered untreated filaments from Harrison with which to repeat his experiments. He thereupon obtained some treated filaments from Harrison and this time he repeated his original success.

These treated filaments, after being subjected to the high temperature of the electric furnace, were very much blistered, as if gases had come out from within the filament. It was found that these blisters disappeared if the untreated filament were first heated in the electric furnace, then treated and then again heated in the furnace. A lamp with this filament was developed and called the GEM or metallized carbon filament lamp and was put on the market in 1905. Dr. Whitney obtained a patent on it in March, 1909, the original application for which was made in February, 1904.

It was operated at 25 per cent higher efficiency than the regular carbon lamp, or 4.25 lumens per watt for the GEM compared with 3.40 for the regular carbon lamp. The same life results (600 hours) were obtained with both lamps. If the GEM lamp were operated at the same efficiency as the regular carbon lamp, it would last 4¾ times as long, hence, its quality may be said to be 4¾ times as good.

The resistance characteristic of an ordinary treated carbon filament is "negative", that is, its resistance decreases with increases in temperature. Metals have a "positive" characteristic and the resistance of the GEM filament increases with increases in temperature, similar to that of metals. This is why the new filament was called the metallized carbon or GEM (General Electric Metallized) filament.

The chief change in the physical properties of the GEM compared with the carbon filament, which made it possible to operate it safely at a higher temperature (about 1900 deg. C.) and so give a greater efficiency, was the change in the treated coating of the filament which is called the "shell". This shell is graphite, both before and after firing in the electric furnace, as has been determined by chemical test. Furthermore, it has the greasy feel of graphite and gives the characteristic pencil mark of graphite on white paper. The shell after firing is a purer graphite, as its specific gravity is much higher and it is much tougher and more flexible than before. The shell can be pulled off the core (the base filament) in short tubular sections. This fired shell, if pressed flat, will spring back to its original form when the pressure is removed, whereas the unfired shell will break with very little pressure. The unfired shell has a negative resistance characteristic up to a certain temperature, after which it has a slightly positive characteristic. The fired shell has a much lower cold resistance and a decidedly positive characteristic at all temperatures.

Firing the core drives out most of its mineral ash constituents and so prevents blistering of the finished filament. The ash content is more volatile than carbon. This ash content (as well as the carbon of the filament) vaporizes in an ordinary carbon lamp during its burning life, condensing on the bulb, and forms part of the discoloration on the bulb. Owing to the small amount of ash present in the GEM filament the lamp maintains its candlepower during life much better than the regular carbon lamp, due to the lesser blackening of the bulb. The untreated carbon filament is shiny black, the treated carbon is shiny gray and the GEM filament is dull gray in color. By this means it is possible to distinguish these lamps from one another.

The first GEM lamps for 110-volt service, put on the market in 1905, had two single hairpin filaments connected in series. Later it became possible to make GEM lamps having a single oval filament for use on 110 volts, these being put on the market in 1909. Lamps were made in sizes from 30 to 250 watts but, with the introduction of the tungsten filament lamp in 1907, the higher wattage sizes soon disappeared from use. The 50-watt lamp was the most popular size and was marketed until 1918, when the manufacture of all GEM lamps ceased.

GEM series lamps were made for street lighting but they also quickly disappeared, as did the 30- and 60-volt GEM lamps for train lighting service, with the advent of the tungsten filament. GEM lamps for 220-volt service were not manufactured.

The Osmium Filament

Dr. Carl Auer Von Welsbach, who had produced the Welsbach gas mantle, invented the first commercial metal filament lamp, the Osmium lamp, but it was used only in Europe and in very limited quantities.

Osmium is an extremely rare and expensive metal, costing much more than platinum, which itself is over five times as expensive as gold. It is non-ductile and exceedingly brittle and so cannot be drawn into wire. Von Welsbach applied in this country in August, 1898, for patents on the lamp and processes for making the filament, the patents being granted in November, 1910. The filament was made by mixing powdered osmium with a binder, such as syrup of sugar, the resulting paste being squirted by pressure through a die. The thread formed was heated to carbonize the binder and and current then passed through it in moist hydrogen gas. The current heated the thread to a high temperature which decomposed the water vapor, the oxygen of which combined with the carbon binder forming carbonic acid gas. The particles of osmium remaining were then sintered together by the high temperature, forming the filament.

OSMIUM LAMP, 1899-1906

A few of these lamps were made in Europe. They were considerably more efficient than the carbon lamp, but on account of the scarcity of osmium, the filament material, it was impossible to make them in large quantities.

The filament was extremely fragile and, as its resistance was very low, at first only low voltage lamps were made to burn two or more in series on 110-volt circuits. Later a few 110-volt lamps were made. Osmium melts at about 2500 deg. C., which is much below the melting point of carbon, but the filament can be operated at a higher temperature than that permissible with carbon for the same life, as it does not vaporize so easily. This made it possible to operate the lamp at 5.9 lumens per watt, which is about 75 per cent more efficient than the carbon lamp.

This extremely high (at that time) efficiency lamp was a tremendous improvement, and even with its fragility, would have formed a great step forward in the lamp art if it could have been produced in large quantities. The world was ransacked for osmium. Expeditions were sent out to explore wild territory, engineers being hired to go out with pack mules to traverse unknown country far away from places man had ever visited. Even as late as the summer of 1903, the Canadian Northwest was being explored, but with all these efforts and expenditures, the best that could be done was to obtain but a small quantity of the rare metal.

A few thousand lamps were made, and these were generally not sold, but rented so that the burned out lamps could be obtained to recover the osmium left in them. They were put on the market about 1899, and only used in a few installations in Berlin and Vienna, where the lamps were made. Manufacture of the lamp was abandoned in 1906, when the tungsten lamp appeared. Osmium lamps were not marketed in this country.

The Tantalum Filament

The metallic substance, known as tantalum, one of the elements, was discovered over a hundred years ago, about 1802. It is practically unaffected by various chemicals, an early writer stating that "even when in the midst of an acid it is unable to take the liquid unto itself". It was named after the fabled Tantalus, who was condemned to stand up to his chin in water which constantly eluded his lips when he attempted to quench his tormenting thirst.

Dr. Werner Von Bolton, a Russian chemist, in the employ of the Siemens & Halske Company, a large electrical manufacturer in Germany, discovered about 1902 that this metallic substance really contained a considerable amount of oxide of tantalum. He removed the oxide in the metal by placing some of it between the poles of an electric arc in vacuum, a vacuum pump removing the oxygen as fast as it was released. He later found that at first he did not obtain pure tantalum because what he got was an extremely hard metal, so hard that it was impossible for a diamond drill rotating 5000 times a minute for three days to drill a hole through a sheet of it only one millimeter thick. This extreme hardness was due to impurities which disappeared when he employed electrodes of the first lot of tantalum he made. The pure metal, however, is still hard, about equal to that of the hardest steel, but it is ductile so that it can be drawn out into a fine wire, having a tensile strength of about 100,000 lb. per sq. in.

TANTALUM LAMP, 1906

This lamp had a filament of the metal tantalum, and was much more efficient than the carbon lamp. It disappeared from use in 1913.

Tantalum is about twice as heavy as iron, having a specific gravity of 14.5, that is, it is 14½ times as heavy as distilled water at ordinary temperature. Its melting temperature is high, about 2850 deg. C., but while this is considerably below that of carbon, Dr. Von Bolton found that it could be operated as a lamp filament at somewhat higher temperature than that permissible with the GEM lamp for the same life because it vaporized less easily. This made it possible for him to produce a tantalum lamp to operate at 4.8 lumens per watt. It had a quality value nearly 2¾ times that of the GEM lamp, and if the two were operated at the same efficiency, the tantalum lamp would live 2.71 times as long as the GEM lamp. Dr. Von Bolton applied for a U.S. patent in May, 1902, which was granted in April, 1906.

Tantalum has a relatively low electrical resistance, so the filament for a 110-volt lamp had to be long and thin. The 44-watt lamp originally made had a wire filament 1.8 thousandths of an inch in diameter and about twenty inches long. For comparison the 50-watt carbon lamp filament is four thousandths of an inch in diameter and about nine inches long. A human hair is about three thousandths of an inch in diameter.

The tantalum lamp was put on the market in this country in 1906. The original 44-watt lamp was later changed to 40 watts, and an 80-watt lamp added for 110 volt circuits. It was also supplied in round bulbs, and lamps for 30, 60 and 220 volt service were also made. It was found that while good life results were obtained on direct-current circuits, the filament, when burned on alternating current, rapidly crystallized and so did not last long. As direct current is supplied by lighting companies in only a few cities, the use of the lamp was limited, the greater portion of electric current supplied being alternating. The lamp disappeared from the market in 1913.

The Tungsten Filament

The metal tungsten, an element, was discovered in 1781, and for more than a century and a quarter was known to chemists as an entirely intractable metal, existing only as a powder of hard, brittle particles or as a rough, more or less fused mass, incapable of being forged or worked in any way. It was used only in alloys, notably in tungsten steel, making the steel extremely hard, and as a constituent of chemical compounds.

It is extremely heavy, nearly twice as heavy as lead. It is now known to have a specific gravity of 19.1; prior to its use as a filament, authorities stated it to be from about 17.2 to 17.6. It has a melting temperature of about 3400 deg. C., a temperature at which asbestos and fire brick would melt like wax in a furnace. But little of the properties of the metal itself were known until it was used in a lamp, one authority even stating as late as 1903, that its melting temperature was 1500 deg. The operating temperature of a treated carbon filament is about 350 deg. higher than this.

The name tungsten is derived from the Swedish "Tung" meaning heavy and "Sten" meaning stone. Its chemical symbol "W" is derived from Wolff, one of the early experimenters on the metal.

Tungsten is plentiful, being obtained from various ores, such as Wolframite, a tungstate of iron and manganese, and Sheelite, a tungstate of calcium. Ores are mined in Colorado, California, New Mexico, China, Korea, and many other places. The ore is usually purified to the oxide, which is a yellow powder resembling sulphur. There are lower oxides which are bluish and brown. The oxides are further reduced to tungsten, which appears as a fine gray black powder.

Early Suggested Uses of Tungsten in Incandescent Lamps

As a matter of record it is interesting to note that Turner D. Bottome, an American, applied for a patent in September, 1887 (granted in April, 1889), which discloses a process consisting of saturating carbon filaments with solution containing a tungsten compound, baking the filaments and reducing the tungsten compound to tungsten metal. This process was to be repeated as often as necessary in order to obtain the proper amount of tungsten in the carbon filament. Bottome's idea was that by adding tungsten to the carbon it would produce an additional hardness to the filament such as is conferred upon steel by the addition of tungsten. The scheme was never used. Such a filament, if operated above the normal temperature of the carbon lamp, would rapidly blacken the bulb with a deposit of carbon.

Alexandre De Lodyguine, a Russian, suggested the use of tungsten and other materials to make up a composite filament in patents he applied for in 1893 and 1894. At this time Edison's basic carbon lamp patent had been sustained in the courts, and the Westinghouse Company was trying to develop a lamp that would not infringe this patent. De Lodyguine was retained by the Westinghouse Company to do this, and put in two years of intensive work but without success.

De Lodyguine's idea was to build up a high resistance coating or shell on a platinum or carbon core, thereby making a high resistance composite filament. The shell could consist of molybdenum, tungsten, rhodium, iridium, ruthenium, osmium or chromium. The scheme was never used, as with a platinum core, the platinum would melt, soak through the shell and vaporize quickly, blackening the bulb if it were operated above the filament temperature of a carbon lamp. Platinum melts about a hundred degrees below the operating temperature of the carbon lamp. With a carbon core the same difficulty would occur as in Bottome's scheme.

Invention of the Tungsten Filament Lamp

Alexander Just and Franz Hanaman, in 1902, were laboratory assistants to the professor of chemistry in the Technical High School in Vienna. Just was making use of his spare time by working in another laboratory trying to develop an incandescent lamp having a filament of boron. His means were very limited, his whole income being about $55 per month. In August, 1902, he got his co-worker Hanaman, whose monthly income was even less, to assist him. The two conceived the idea of trying to produce a tungsten filament lamp and they worked on both the boron and tungsten lamps for about two years. The boron lamp was a failure.

They first started experiments on the tungsten lamp by exposing a carbon filament at high temperature to the vapor of tungsten oxychloride in the presence of a small quantity of hydrogen. Their theory was that a complex chemical reaction takes place, depositing the tungsten of the oxychloride in place of the carbon, and that this reaction continues until the carbon of the filament has been entirely replaced by tungsten.

Their aim was to make a pure tungsten filament and, as they knew that tungsten was brittle and unworkable so that it could not be drawn out into a wire, they thought this carbon replacement method would finally produce a tungsten filament. This effort was a failure for the reason that the first thin coating of tungsten on the carbon filament prevents further action between the carbon and tungsten oxychloride vapor. This filament merely became one having a carbon core and a tungsten shell, and when operated at a temperature above that of the ordinary carbon lamp, the carbon would dissolve through the tungsten, vaporize, and quickly blacken the bulb, as in Bottome's scheme.

They were using a paste containing graphite and a binding material, such as coal tar, to fasten the filament to the leading-in wires. They found that much of the black deposit in the bulb came from this paste, so they heated the pasted joints in hydrogen gas and found that the blackening was very materially reduced. This led them to believe that there must be some de-carbonizing process going on. Being chemists they came to the conclusion that some oxidizing substance was acting as a go-between between the carbon and hydrogen. The hydrogen gas they obtained was produced by the action of hydrochloric acid and zinc, and they found that it contained a considerable amount of water vapor. They therefore reasoned that the high temperature decomposed the water vapor, the oxygen combining with the carbon.

Finally they evolved a process of making a substantially pure tungsten filament by coating a fine carbon filament with tungsten deposited by heating the carbon filament in a vapor of tungsten oxychloride as previously described. The coated filament was then heated to high temperature by passing current through it in an atmosphere of neutral gases which would not react on it chemically. This heating made the carbon core dissolve into the tungsten shell surrounding it, the carbon then being removed by another heating in an atmosphere of water vapor and hydrogen. Later the first heating was dispensed with, the second heating accomplishing the results obtained by the original first heating.

Another process was evolved by them, which was commercially used in this country for several years. It produced what was called the "pressed" filament and consisted of mixing tungsten powder with an organic binding material, of which there are several that can be used. In the commercial process, a very fine grained tungsten powder was mixed with a solution of sugar and gum arabic to make a thick paste. This paste was squirted under high pressure through a diamond die and caught in loops on a piece of cardboard. Tungsten is so hard that it will soon wear out any other than a diamond die. The loops were baked enough to partly carbonize the binder and then were passed through a "forming" machine in which electric current of increasing amount was passed through them while they were in an atmosphere of hydrogen and nitrogen which contained some moisture. This removed the binder and left substantially pure tungsten in the filaments.

Just and Hanaman found that the substantially pure tungsten filament they were able to make could be operated at about 7¾ lumens per watt and yet give good life results. This was an enormous improvement over all previous lamps made. Their financial resources by this time were so depleted that they did not have sufficient money to apply for patents to protect their invention in all the various European countries. They finally were able to borrow $60 from a chemical manufacturer in Vienna with which to apply for British and French patents, which were filed on Nov. 4, 1904. They found it difficult to obtain financial assistance to develop their invention further, but they finally induced a carbon lamp manufacturer in Ujpest, Hungary, to try out their lamp. Much further experimental work had to be done before the lamp could be produced commercially, the lamps being put on the market in Europe in limited quantities in September, 1906. They used the carbon filament displacement method at first, later the pressed filament. In July, 1905, they applied for a patent in this country.

MULTIPLE TUNGSTEN FILAMENT LAMP, 1907

This lamp was originally nearly three times as efficient as the carbon lamp.

The General Electric Company bought Just and Hanaman's American patent rights and after much development work, marketed, early in 1907, a street series and 100-volt multiple lamp. The filament was rather fragile and the lamps had to be handled carefully. Notwithstanding the fragility, their high efficiency made them a great commercial success, the tungsten filament making the greatest advance ever made in the quality value of the vacuum incandescent lamp. If the 100-watt, 110-volt tungsten lamp of 1907, having an efficiency of 7.85 lumens per watt, were operated at the same efficiency as that of the tantalum lamp, it would last 27.1 times as long, making its advance over the tantalum lamp just ten times the advance of the tantalum over the GEM lamp. Nearly half a million tungsten filament lamps were sold during the first year, 1907.

Tungsten has a low electrical resistance, lower than tantalum, about half that of platinum and very much lower than that of carbon. However, when heated, tungsten increases greatly in resistance and even though the carbon filament decreases in resistance when heated, the tungsten filament in a lamp must be much longer and thinner than that in a carbon lamp. The 40-watt, 110-volt, vacuum tungsten filament lamp has a filament very nearly two feet long and about 1.6 thousandths of an inch in diameter. In order to get this long tungsten filament in a bulb, several hairpin loops of the pressed tungsten filaments were mounted on a spider, and connected in series with each other to get the requisite resistance for 110-volt circuits.

Series lamps were also put on the market which quickly displaced the carbon and GEM lamps used in street lighting. They not only consumed less energy for the same candle power given by the other lamps, but made it possible to greatly increase the lamp capacity of the constant current transformers used. As a result, larger sizes and greater numbers of street lights began to be used.

The low resistance of tungsten made lower voltage lamps commercially feasible, so that in lighting trains 30 and 60-volt tungsten lamps immediately displaced the lamps formerly used. The lighting of automobiles with 6-volt lamps operating on storage batteries soon replaced the oil and acetylene lamps formerly used. Flashlights received a tremendous boom, as the 2½ and 3½-volt tungsten filament lamps tripled the capacity of the small dry batteries used.

Tungsten Lamp Patent Granted to Just and Hanaman

There were two other inventors, who had applied before Just and Hanaman, to the Patent Office in Washington for patents covering a tungsten lamp filament. One was Von Bolton, the inventor of the tantalum lamp, whose application was dated November 10, 1904, and the other was Dr. Hanz Kuzel, a German, who applied January 4, 1905. Just and Hanaman filed their application on July 6, 1905.

Von Bolton's application covered various metals, among which tungsten was mentioned, which were to be melted and could be fashioned into filaments by a drawing process. He had discovered that the supposedly non-ductile metal tantalum, if purified, became ductile and could be drawn into a wire and would make a good lamp filament. He did not know that any other metal would make a good lamp filament, but there was a large group of metals whose properties were little known and whose adaptability to the lamp art was not even known at all. He appears to have thought that possibly some of these other metals might be made ductile if purified and thus make good lamp filaments and to have wondered if his success with tantalum might not be repeated with some other metal by some other inventor. Desirous of forestalling such other inventor, Von Bolton filed his speculative patent application.

Up to this time it had been impossible to produce ductile tungsten so that it could be drawn into a wire by any known process. The Patent Office, therefore, questioned the operativeness of Von Bolton's application. As will be shown, a brilliant invention was later made by another inventor by which tungsten could be drawn into a wire by an entirely new process. This new process was not covered by Von Bolton's application.

The Siemens & Halske Company, Von Bolton's employer, had in 1903 abandoned his theory of the ability to draw tungsten. They had, in that year, obtained an English patent covering a process of making a tungsten filament by means of an alloy of tungsten and nickel, drawing this alloy into wire and then removing the nickel. In this patent it stated the impossibility of directly making a tungsten filament and spoke of tungsten as a non-ductile refractory metal.

Dr. Kuzel's application covered a process of making a filament from any one of fourteen metals, among which tungsten was included. The process consisted of reducing these metals to a colloidal condition which, when made into a paste with water (no organic binder being used), was squirted through a die to form a thread. The tungsten particles of the thread were then sintered together to form the filament.

It then appeared to be only a question of a proven priority date of invention as to which of the two parties, Just and Hanaman or Kuzel, would be granted the patent. Evidence was introduced to the U.S. Patent Office that Just and Hanaman had filed applications for their French and British Patents on November 4, 1904. This was prior to the U.S. application of both Kuzel (January, 4, 1906) and Von Bolton (November 10, 1904). In July, 1911, the Assistant Commissioner of Patents handed down a very thorough and extended decision on the patent interference, and the patent was granted to Just and Hanaman in February, 1912.

The Trade Mark Mazda

The trade mark MAZDA was adopted by the General Electric Company late in 1909, but is now used by more than one manufacturer. It is not the name of a thing but the mark of a research service rendered to the manufacturer by the Research Laboratories of the General Electric Company at Schenectady, New York. It comprises not only the incandescent lamp research work done by these laboratories and the data obtained from the testing and inspection work done throughout the company, costing over a million dollars a year, but also, the accumulation of scientific and practical data from laboratories, factories, etc., all over the world. The results are transmitted to the manufacturers entitled to this service, with such aid and information as will assist them to improve the quality of their lamps.

A MAZDA lamp is, therefore, the product of the latest and best method of incandescent lamp making. The filaments of all MAZDA lamps are at present made of tungsten, but when any material more suitable for the purpose is discovered or developed, it will be used.

Persian mythology gives to their ancient god of light the name Ahura Mazda, and to the Persians, light was knowledge. MAZDA service therefore, very fittingly stands for the accumulation and transmission to lamp manufacturers of the knowledge which will enable them to produce the best light.

The Drawn Tungsten Wire Filament

As has been stated, tungsten was known to be a very hard, non-ductile and brittle metal which could not be drawn into a wire. Many scientists were misled into the belief that if it were purified it would become ductile as Von Bolton found to be the case with tantalum. Prior to 1906, it was the universal opinion that tungsten could not be made ductile. It was known that when heated to very high temperatures it could be bent, but when cool it was always brittle.

Dr. William D. Coolidge, of the Research Laboratories of the General Electric Company at Schenectady, began an investigation of the subject in 1906. He first produced tungsten as pure as he could get it, and then deliberately added various impurities to study their effect. These experiments led him to believe that in the case of tungsten it was not the presence of impurities which made the metal brittle, but that the brittleness was an inherent characteristic of the metal itself. His first discovery, which later gave him the clue which he afterwards so brilliantly followed, consisted in finding that tungsten, carefully prepared in a particular way, could be hammered at certain temperatures and that by so hammering, the material could be considerably elongated and its form changed. While the metal which was thus hammered was brittle when allowed to cool, nevertheless Dr. Coolidge had done something which no one else had ever done and it encouraged him to continue.

At this point, he discovered a new process for getting tungsten into a dense coherent form. This process consisted in incorporating tungsten powder with a ductile metal alloy of cadmium, bismuth and mercury, squirting the mixture through a suitable die and then, by heat treatment, removing the foreign ingredients and sintering the tungsten powder. This, so-called, amalgam process was subsequently used in preparing thick tungsten filaments from which the first tungsten wire was drawn. As the amalgam process gave better squirted filaments, in the large sizes, than were at the time obtainable in any other way, it was intensively developed by Dr. Coolidge in the laboratories and later became the standard factory process for the production of high wattage and series lamp filaments.

Early in 1907, Dr. Coolidge again took up the hot working of tungsten, experimenting with a small rolling mill such as is used by jewelers. He heated the rolls, a most unusual operation, to a temperature of about 300 degrees Centigrade and passed amalgam process tungsten filaments between the hot rolls, obtaining an appreciable lengthening of the filaments. Before this time he had discovered that he could bend amalgam process filaments into special shapes by the application of proper but relatively low temperatures, going so far as to coil the filament into a spiral whose internal diameter was no greater than that of a knitting needle. This in itself was a valuable achievement, as such concentrated filaments are of value in focusing types of lamps such as those used in automobile headlights.

His next work, done late in 1907 and early in 1908, consisted in squeezing thick tungsten filaments between hot blocks of tungsten steel whose working faces had been ground parallel and hardened. An appreciable extension of the filament was obtained and when such a hot-worked filament was broken in two and one part was heated above the equiaxing temperature, measurements showed that the part which had been hot-worked and not equiaxed was stronger than the other part in the sense that it would stand cold bending through an arc of smaller radius.

Dr. Coolidge had, then, learned that suitably prepared amalgam process filaments could be bent, rolled and pressed at temperatures at which hardened alloy steel tools would not lose their temper. The hot-pressing experiments had also shown an improvement in mechanical strength resulting from such hot working.

He next decided to try hot-drawing some filaments and, guided by his earlier hot-working experience, he recognized the need of heating the die, that portion of the filament which was in tension, and the jaws of the pliers holding the end of the filament. The openings in the dies naturally were smaller than the filament, but the difference, called "the draft", had to be very small, a fraction of a thousandth of an inch, as otherwise the filament invariably broke.

In order to introduce the filament into the opening in the die, the entering end was pointed by a process which he had previously invented which consisted in electrolyzing it in a concentrated aqueous solution of potassium cyanid. This method, unlike the ordinary electrolysis of tungsten, reduced the diameter without rendering the surface pitted and porous, and hence without needless weakening of the filament at the point where it was to be grasped by the hot pliers. The die was heated by a special gas burner; that portion of the filament between the die and the pliers, pulling the filament through, was heated by a hot body of metal underneath; the pliers were heated by gas; and that portion of the filament back of the die on the entering side was, in some cases, heated by a gas heated metal under and partially surrounding the filament.

In this way, in the fall of 1908, pieces of pressed tungsten filament were successfully drawn through many dies, each but little smaller than the previous one, and then it was found that a wonderful thing had been accomplished, the tungsten had lost its brittleness. The tungsten had actually become bendable, and even ductile, when cold.

The Drawing of Ordinary Ductile Metals

Ordinary ductile metals, such as wrought iron, copper, silver, gold, etc., may exist in either one of two states which are known as the "crystalline" state and the "strain-hardened" state. The crystalline state is the natural condition of the metal and is that in which it exists after it has cooled from a molten state. Under the microscope, and sometimes by the naked eye, the metal will be seen to be composed of an aggregate of crystals. Ordinary workable metals are ductile in this crystalline state. In the strain-hardened state, these crystals have been changed into fibers, threads or plates, or in some other way have been strained and distorted out of their original crystalline form.

The change from the crystalline to the strain-hardened state is produced by mechanical working at low temperatures such as by drawing the metal into wire, which is ordinarily done at room temperature. As the crystals are deformed by working, the metals become hard and springy and their workability decreases. If the strain-hardened (sometimes called "hard-drawn") fibrous metal be heated to a certain temperature, different for each metal but always below its melting temperature, and maintained long enough at this temperature, the fibers break up and recrystallize. This temperature is called the metal's "annealing" temperature and with ordinary metals it restores its ductility. Thus in drawing ordinary metals they become hard and difficult of further working. They are then annealed, bringing them back to their original ductile condition.

Ductility or its absence is a specific property of a metal, not entirely dependent upon hardness or softness, strength or weakness, nor on any other single property. For example, at room temperature, manganese steel is very hard, very strong, and very ductile; certain heat treated steels are hard, very strong, and non-ductile; copper is very soft, weak, and very ductile; lead is very soft, very weak, and only slightly ductile; and antimony is soft, weak, and non-ductile.

BRITTLE TUNGSTEN, CRYSTALLINE STATE

This photo micrograph shows the normally crystalline state of tungsten in which condition it is brittle.

Tungsten Ductile in Fibrous State

Under the microscope the structure of Dr. Coolidge's ductilized tungsten filament was fibrous, while that of the original brittle filament was crystalline. This is just the opposite of what had been found in the ordinary ductile metals. He had "ductilized" a non-ductile metal and, as he later discovered, had increased its strength enormously. Samples of drawn tungsten wire of one-thousandth of an inch in diameter show a tensile strength of 600,000 to 650,000 pounds per square inch. The tensile strength of this drawn tungsten is more than thirty times that of the original sintered tungsten, no other material showing any such increase in strength as this.

DUCTILE TUNGSTEN, FIBROUS STATE

When tungsten is carefully prepared in a certain manner and worked at certain temperatures, the crystals are deformed into fibers and the metal becomes ductile.

A striking feature is that no such process as that developed by Dr. Coolidge has ever been able to increase the ductility of any other metal, and no mechanical process whatever had previously produced ductility in any metal which was non-ductile.

Dr. Coolidge also later found out that the ductile tungsten he had produced would, if heated to a certain high temperature, again become brittle. This might be called its annealing temperature, although an annealing temperature produces ductility in ordinary metals.

Development of the Commercial Drawn Tungsten Wire Process

While Dr. Coolidge had finally been able to make a small piece of tungsten ductile, it required much more investigation and experiment to repeat the accomplishment on a large enough scale to make the process commercially practical. In fact, as will be shown, many obstacles appeared which for some time seemed insurmountable, and it required about two years of painstaking effort and skill before the desired result was obtained. The difficulties and discouragements he met with were at times almost heart breaking.

The first piece of ductile tungsten he had produced was made from an "ingot" (if so ponderous a name can be used) consisting of a pressed tungsten filament 25 one-thousandths of an inch in diameter. In order to obtain an ingot or slug, of a reasonable size, he first tried to press dry tungsten powder together without a binder. He used a steel mould filled with tungsten powder and tried to form the slug by pressure applied at the end. This was the natural thing to do, but instead of producing a homogeneous slug, he obtained one with a plate-like structure.

He next tried using a mould in which the pressure was applied at the side, but the resulting slug contained what he called "corner cracks". These cracks caused much difficulty and it was only after an extended study of the effect of the amount of pressure used, the method of applying the pressure, the design of the mould and many experiments on various lubricating substances which could be used on the surfaces of the mould, that he was able to make slugs free from mechanical faults. The slugs finally produced were so fragile that they could only be handled by sliding them carefully along a smooth surface. The next step was to give them some mechanical strength, which was accomplished by baking them in a tube in a stream of hydrogen.

This baking was only a preliminary stage; it was necessary to heat the slugs to a very high temperature to cause the tungsten powder to sinter together. This was done by passing a heavy current through them, like the sintering operation in making pressed filaments, but here new problems arose requiring the development of a special bottle.

In this heating operation the slug was mounted vertically and at first a rigid clamp was attached to each end, the current passing in at one clamp and out at the other. The slug was surrounded by a metal treating bottle and a stream of hydrogen gas passed through the bottle to protect the tungsten from oxidizing. When the slug was heated it shrank and usually broke in two or pulled out at one end from one of the clamps. The bottle was full of hydrogen and a certain amount of air was drawn in by the first cooling resulting from the shutting off of the current. Hydrogen and air form an explosive mixture and the result was usually a violent explosion, the very hot tungsten slug igniting the mixture, and the bottle being blown to the ceiling.

To overcome the difficulty the expedient was tried of giving the slug a slight partial treatment, reclamping it, giving it a further slight treatment, and so on, but dangerous explosions still occasionally occurred. The problem was finally solved by suspending the slug by the upper clamp, the lower clamp dipping in mercury which was kept cool by water flowing through a copper tube. The mercury would conduct current to the lower clamp and allow the slug to shrink, the apparatus being so designed that the shrinkage did not cause the lower clamp to leave the mercury.

Serious difficulty arose from another cause. The slug would occasionally break near the upper end or pull out of the upper clamp. The upper end of the slug in falling would often strike the inner surface of the bottle, forming a severe arc, and often melting a hole through the inner layer of the bottle, which was a double walled affair, cooled by water flowing between the walls. Such conditions were finally overcome by using springs instead of bolts in the clamps.

Another serious difficulty remained, however. There was a good deal of oxidization of the slug while in the bottle, the cause of which was not clear for a long time. It was finally found that it was due to the fact that when the slug was at a high temperature, the convection currents in the hydrogen gas around it were so vigorous that they extended down to the mouth of the bottle and caused air to be drawn in. To obviate this, the mouth of the bottle was allowed to dip into mercury filling a circular depression in a metal plate, which made an effective seal.

All this required several months of work, and it turned out that all the slugs produced were entirely brittle, not only when cold but also when hot, and so could not be worked. This was so discouraging that it then seemed impossible to start with a slug of anything but minute size.

He then tried, with the help of an expert, skilled in electric furnace practice, to produce a slug of tungsten by heating the metal in the high temperature of an electric arc furnace. But this did not help, for when he attempted to work the slug it cracked all to pieces.

Feeling that he was making so little headway on the direct attack, Dr. Coolidge decided to drop work on tungsten for a time and to try hot working large masses of molybdenum. The latter metal has some of the properties of tungsten, but possesses some slight inherent ductility; so he hoped that the presumably simpler problem of working molybdenum might teach him something which would help him to work tungsten. All this effort on the hot working of the larger metal masses so far had taken over a year of his time.

He then returned to his sintered tungsten slugs and tried hammering them hot by hand on an anvil, but could make no progress. Fearing that the failure was caused by his own lack of skill, he called in two expert blacksmiths. He found that it was possible to hammer the slug a little, certain blows being successful, but with others the slug would break to pieces.

He then tried his jeweler's rolling mill again, using exceedingly small drafts, but even then the slugs cracked badly. He found, however, that the work was being cooled at the point where it should remain hot, so he built a special rolling mill in which a current of about a thousand amperes passed from one roll across the tungsten to the other roll. This heated the slug at the point where it was being worked, and with it he made a little headway, but was not able to work a tungsten rod down to such a size that it could be drawn through a die.

He then went to see a manufacturer of swaging machines. These machines have two small hammers which operate at high speed as the machine is rotated, striking blows on anything placed between them. The hammers have a recess in them leaving an opening through which the rod to be swaged is fed. The minimum size of this recess determines the diameter of the rod after it has passed through the machine, the hammers being usually called swaging dies. This manufacturer had built a few machines for hot hammering steel, but on account of difficulties, the work was confined to short lengths of large cross section and the machines were not adapted to hot hammering long lengths of small cross section.

He next visited another concern where swaging machines were being used for the cold swaging of needles, but no one seemed to think that the machines were suitable for hot working rods of small diameter. Nevertheless, he obtained one of these machines, but when he tried it, even with molybdenum, the metal went all to pieces in the first two or three dies; Another difficulty was that as the dies rotated about the work, they tended to take the work with them and twist it off. He tried increasing the speed of the machine, but this only intensified the trouble and as the material was so hard, the hammering not only cracked the material but even the dies themselves.

The operating principle of the machine consisted of striking a large number of overlapping blows to produce a smooth surface on the material worked as it was slowly passed through the dies. Having found that this procedure led only to failure, he decided to strike out for himself in an opposite direction. He found that with the ordinary swaging die each blow abstracted a certain amount of heat from the tungsten. The next blow, struck practically in the same place, hit the spot of tungsten that had become chilled below the most favorable temperature and cracked it. He designed some special dies that had but a small working face, and by feeding the rod through the machine at fairly high speed, he was able to prevent the blows from overlapping. This helped tremendously, and by specially shaping the face of the dies he was finally able, to eliminate the trouble of twisting the work.

As a result he was enabled to carry a molybdenum (not tungsten) rod successfully through several dies without cracking, but then another difficulty appeared. He had been holding the heated rod in a pair of tongs, thrusting it into the swaging machine as rapidly as possible for half its length, and then withdrawing it. As a result, a number of blows which overlapped each other struck the middle of the rod, chilling it and producing cracks. To overcome this, he provided a very powerful brake by which he was able to stop the machine very suddenly when he had thrust the rod in as far as he thought desirable, thus slowing down the hammering action of the machine before he slowed down the motion of the rod. He then withdrew the rod, reheated it, and thrust the opposite end in the swaging machine. Later, however, as the art advanced as a result of his researches, it became possible to get along without the brake.

Finally, he was able by this hot swaging process to reduce his original tungsten slugs, which were about ¼ to 3/8 of an inch square and six inches long, to a rod having a diameter of about 1/8 of an inch. However, from this point on, his difficulties increased enormously as the size of the rod decreased. As the rod decreased in diameter its length of course increased, increasing the number of blows that had to be struck, and a single blow struck under unfavorable conditions was sufficient to crack or break the rod. At this point the problem of bridging the interval between 1/8 of an inch (or 125 mils— a mil is a thousandth of an inch) to 30 mils seemed almost impossible with a swaging machine. His original piece of ductile tungsten was made from a pressed filament of 25 mils drawn down through diamond dies, and diamond dies larger than 30 mils were not available. He tried chilled iron dies, using swaged molybdenum, but after passing it through several dies, it split up badly. He then tried drawing hot molybdenum through the dies but found it destroyed them.

His next step was to obtain a much smaller swaging machine that could be driven at a higher rate of speed, which permitted the work to be fed faster into the machine, the working face of the dies being still further reduced. A small tube furnace was placed in front of the machine and the work was fed from this directly into the machine by a pair of rolls running at a uniform rate of speed. As a result he was finally able to swage molybdenum, and later tungsten, down to 30 mils.

During this whole process the workability of the tungsten was being improved and at 30 mils it was found to be ductile. From this point on he was then able to draw the tungsten, which owing to its reduced size may now be called wire, through diamond dies by methods he had already used. This consisted of using hot dies; heating the wire; aqua dag lubricant which, at the suggestion of one of his assistants, was baked on the wire; small drafts; and a gradual reduction of temperature as the work proceeded.

While he had now been able to make a fine drawn tungsten wire from a relatively large tungsten slug in the laboratory and with much patience, it did not mean that a commercial process had been developed to manufacture wire on a large scale. There were difficulties which had to be met, some tungsten slugs seemed capable of being mechanically worked, while others did not. Two lines of research were started, mechanical and chemical, both being carried along together.

The tungsten slugs had been heated in a gas forge. He tried heating them in an atmosphere of hydrogen and devised a special iron tube furnace for the purpose. This helped, as it was believed that the rods took up carbon or oxygen from the furnace gases and thereby lost much of their workability. Another difficulty then arose. Small shiny spots appeared on the slugs after they had been sintered in the treating bottle. After the first swaging, the metal in the neighborhood of these spots was found to be brittle. Samples of this brittle metal were analyzed and found to contain iron, which, it was decided, must have come from the walls of the iron tube furnace. This difficulty was overcome by placing the rods in carriers and embedding them in powdered silica. A vigorous stream of hydrogen gas was passed through the iron tube and in this way the vapor of any iron evaporated from the walls of the furnace was prevented from reaching the tungsten.

But there still was a lack of uniformity in the behavior of the various slugs. Some of them, as they came from their first heating in the iron tube furnace, shrank more than others when being sintered in the treating bottle, those which had shrunk least being the more easily worked. He thought that the ones that shrank most must have taken up some other impurity from the furnace, so he devised an electrically heated porcelain tube furnace. The first slugs fired in this furnace looked much better than anything seen up to that time. He then went on a vacation, leaving instructions to press up, heat in the porcelain tube furnace and sinter in the treating bottle a hundred slugs which were to be ready for hot working experiments on his return.

On his return he found that none of the slugs could be worked, all breaking up in either the first or second swaging die. This puzzled him greatly, but he finally decided, after much investigation, that the trouble was caused by the presence of oxygen in the sintered slugs. He was using very fine tungsten powder which oxidizes to such an extent that it normally absorbs a relatively considerable amount of oxygen before it is moulded in the press. During the first heating of the slug in the iron tube furnace, the rods had been packed in silica and this finely divided material had hindered the escape of water vapor resulting from the action of hydrogen on the oxide of tungsten. The long continued heating in this atmosphere of water vapor had materially coarsened the tungsten powder and thus had made it possible to remove the oxygen before sintering had taken place. The porcelain furnace merely removed the oxygen from the surface of the slug, and after this had happened the surface sintered over imprisoning the balance of the oxygen. Dr. Coolidge then made some relatively coarse tungsten powder by melting tungsten oxide in a crucible, crushing the resulting mass, and reducing the oxide to tungsten. He then found that another chemist in the laboratory had, for some other purpose, heated some tungsten oxide in a "Battersea" crucible and also reduced it to a coarse tungsten powder. With these coarse powders he was able to get good results from slugs treated in the porcelain furnace, provided they were kept out of contact with the porcelain tube. Otherwise they would take up the glaze from the tube, which caused a network of fine cracks to develop on the surface of the rods after they had been partially worked.

The problem of producing tungsten wire in quantity had now been solved, but lamp filaments made from this early wire "offset" badly when burned on alternating current. This was a difficulty which had also been found in the tantalum lamp. It was discovered, however, that wire made from the coarse tungsten powder produced from the oxide heated in the Battersea crucible did not offset. Dr. Coolidge reached the conclusion that the tungsten had absorbed certain substances from the Battersea crucible which had some effect on the offsetting. After much experiment he found it possible to prevent offsetting by directly mixing the tungsten powder with small amounts of certain other substances.

DR. COOLIDGE AND MR. EDISON, 1922

Dr. Coolidge showed Mr. Edison, when he visited the Research Laboratory in Schenectady in 1922, the swaging machine which he had developed and with which he was able to make tungsten ductile on a commercial scale.

The Temperature of the Working

The temperature at which tungsten is worked is an important part of Dr. Coolidge's invention. With other metals, except in the special case of molybdenum, working below the annealing temperature will always reduce ductility. The reverse is true with tungsten; its ductility is created by working it below its annealing temperature.

Whenever any of the other metals has been worked hot, above its annealing temperature, it has been because it was easier and cheaper to give it the desired form at that temperature, or it was desired to secure the superior mechanical properties associated with the "fine grained" structure, or by working it at or above its annealing temperature, it was possible to work and anneal at the same time thus avoiding a special annealing process.

The actual annealing temperature of tungsten becomes lower the greater the amount it is worked below its annealing temperature. Tungsten can be worked above its annealing temperature, but its ductility is destroyed and it will revert to its crystalline state, becoming brittle when cool. The working range is from about 1650 degrees C., a high white heat, down to about 350 degrees C., which is below a dull red heat; the more the metal is worked, the lower the working temperature. The initial working operations must be carried on at high temperature, otherwise the tungsten would break in pieces on account of its brittleness at low temperature. The high temperature also reduces its hardness.

The fact that heating to temperatures above the annealing temperature destroys the effect of previous working was utilized in a curious and interesting way. After it had been discovered how to make an ingot of tungsten that could be swaged and worked down to small diameters, there was found a tendency for the tungsten to split longitudinally at some stage of the process, usually before the wire had been brought down to the desired size. It split up into a bundle of fibers almost like the fibers of a hemp rope. A certain amount of working the tungsten is good, but too much working spoils it, so the object was to subject it to that certain amount of working and then stop. As various sizes of wire are necessary for the various wattages and voltages of lamps, the exact amount of working was accomplished by the simple expedient of reducing the standard size ingot to one of a particular size under conditions which would not change its internal structure to any considerable extent, and then properly work it down to the size of wire desired.

DRAWN TUNGSTEN WIRE LAMP, 1911

Dr. Coolidge's invention of drawn tungsten wire materially simplified the manufacture of the tungsten filament lamp and greatly increased its ruggedness.

Announcement and Adoption of Drawn Tungsten Wire

Dr. Coolidge's success in being able to make ductile tungsten was announced in March, 1910. In 1914, he was awarded the Rumford Medal by the American Academy of Arts and Sciences for his scientific triumph. This is perhaps the highest recognition of the sort to which an American scientist can aspire, the medal being granted for the most important discovery or useful improvement in heat or light. A patent was granted to Dr. Coolidge in December, 1913.

The making of tungsten filaments was changed over to the drawn wire process, beginning with the latter part of 1910. In the early part of 1911, the drawn wire lamps were put on the market. Over half a million dollars' worth of the pressed filament apparatus had to be scrapped, as well as nearly another half million dollars' value of unsold pressed filament lamps.

Drawn tungsten wire filaments are very strong, and consequently the lamp is very sturdy, a marked improvement over the fragile pressed filament lamp. The lamp is, therefore, much more practicable and the breakage in shipment is reduced. The enormous increase in the strength of the filament greatly increased the use of the lamp under such severe conditions as those met within its application to automobiles, street railway and steam railroad cars, etc.

Drawn wire filaments are much cheaper to make than pressed filaments, so that it became possible to materially reduce the price of the lamps. Drawn wire can be readily coiled, which greatly simplified the manufacture of concentrated filament lamps for focusing purposes.

Ductile tungsten can be drawn to such an exact diameter and cut to the desired length so accurately that practically all lamps made are of the voltage and efficiency for which they are designed. In fact the variation in voltage is so small that these lamps are not photometered, as all previous kinds of lamps had to be, to determine their voltage. Sample lamps are constantly tested for voltage and efficiency, and these tests show that lamps as made today vary less in voltage and efficiency than previous lamps, even after the latter had been tested and sorted for voltage. Thus the necessity for a multiplicity of voltages because of the variations in lamps was eliminated, all lamps could, if desired, be made for a single voltage. As it seemed impractical for all plants to readjust their voltage to one standard, three standard voltages, 110, 116 and 120 volts, have been adopted. At present more than 90 per cent of the standard lighting lamps are of these three voltages, and the stock necessary to properly supply the demand has been greatly simplified.

In the usual sizes of lamps the filament is of such a small diameter, a few thousandths of an inch, that it is impossible to determine the diameter accurately by a micrometer. It is accurately measured, however, by weighing a definite length, a few inches of the wire, in a sensitive torsion balance, which will determine its weight and hence, by calculation from its specific gravity, its diameter, to within three millionths of an inch.

OFFSET TUNGSTEN FILAMENT

After a filament has been lighted for the first time it has a crystalline structure. If the faces of the crystals fall in one plane across the diameter of the filament, offsetting may occur.

THORIA PREVENTS OFFSETTING

This high magnification photomicrograph shows the thoria globules which tend to key the crystals together, preventing offsetting.

Non-Sag Drawn Tungsten Wire

When a lamp is first lighted, the long fibrous grains of the drawn wire, heated above their annealing temperature, are changed to the equiaxed grains of an annealed metal. These grains, during this transition, absorb each other, gradually increasing in size until further growth is retarded or stopped. The cessation of growth of the grains may be attributed to several causes, not the least of which is the presence of impurities in the metal.

The crystals composing the wire are, to the best of present knowledge, held together by amorphous tungsten. This material acts as a binder to hold them together and in place, but, at very high temperatures, it is not as rigid as the crystals themselves, consequently the positions of the latter may become altered.

Should the faces of one or more crystals fall in one plane across the diameter of the filament, offsetting will occur, that is, sections of the filament will slide sidewise and it will soon bum out due to the decrease in cross-section at this point.

Sag Wire

Non-Sag Wire

CRYSTAL GROWTH IN TUNGSTEN FILAMENTS

By removing nearly all the slight amount of impurities in a tungsten filament, the crystals become long and overlap and so make a sag-resisting wire which does not offset. The sag resisting feature is of great advantage in coiled filament lamps.

Fairly small crystals with minute particles of thoria, which is hard at the high temperature, will retard offsetting, the thoria particles tending to key the crystals together. The thoriated wire, however, bends easily at this high temperature due to the fact that the proportion of amorphous material present is greater than that in a large grained wire. Thus, if the crystals could be allowed to grow to a large size, there would be relatively less of the amorphous tungsten present and, if these crystals overlapped and interlocked with each other, a wire should be produced which would remain stiff and would not readily sag nor offset at high temperature.

Dr. Aladar Pacz, of the General Electric Company, made a study of this. He reasoned that if it were possible to get rid of the minute impurities and eliminate the use of thoria, a wire of large overlapping crystals might be obtained which would neither sag nor offset. It might be possible to get rid of these impurities by purposely inserting certain substances. These substances should not vaporize at the relatively low temperatures at which the moulded tungsten slug is given its preliminary heating in hydrogen gas, but should readily vaporize at the relatively high temperature at which the slug is sintered in the treating bottle by a heavy electric current. Thus the substances, coming out of the slug as it is being sintered, might carry with them the minute impurities it was desirable to get rid of.

Dr. Pacz tried mixing various substances with the tungsten powder, and after many experiments finally evolved a process, by which tungsten is produced in apparently such a pure state that the wire filament, made by Dr. Coolidge's process, when lighted for the first time, immediately crystallizes in such large overlapping crystals that it does not materially sag or offset. The crystals are many hundred times larger than those of thoriated wire.

In a MAZDA C (gas-filled) lamp it is most important that the coiled wire should not sag materially. If it does, part of the helix opens up, allowing the gas to more readily circulate between the turns of the helix and thus cool the wire to a greater extent. This lowers the temperature of the filament, reducing its candle power and efficiency. Certain turns of the helix tend to sag together, and if the turns touch each other they short circuit themselves. Dr. Pacz's invention was, therefore, of great value in the MAZDA C lamp, so that it not only considerably improved the maintenance of candle power of the lamp during its life but also increased its average efficiency throughout life.

Non-sag wire is used only in coiled filament lamps. In the straight filaments used in some vacuum lamps, the bends of the filament around the anchors operate at a much lower temperature due to the conduction of heat away from the filament at these places. As a consequence the filament does not sag at the bends.