US 3649224 A
A tungsten wire, previously drawn to a predetermined diameter, is stretched at a stress between its elastic limit and about 95 percent of its ultimate tensile strength, thereby imparting a permanent elongation to the wire. The stressed wire is then coiled into filaments which, on recrystallization, have improved nonsag properties.
Claims available in
Description (OCR text may contain errors)
United States Patent Anderson et al.
1451 Mar. 14, 1972 METHOD OF MAKING NONSAG FILAMENTS FOR ELECTRIC LAMPS Warren A. Anderson; Wilfrid G. Matheson, both of Marblehead; Lester W. Strock, Salem, all of Mass.
Assignee: Sylvania Electric Products Inc.
Filed: Apr. 18, 1968 Appl. No.2 722,214
us. or ..29/1s2.s, 29/4205, 75/206, 75/207, 75/214 Int. Cl ..C22c 1/08, C22C 27/00 Field of Search 158/126; 75/206, 01; 29 529, 5 29/1825 [5 6] References Cited UNITED STATES PATENTS 3,278,281 10/1966 Ehringer ..75/206 X Primary Examiner-Carl D. Quarforth Assistant Examiner-R. E. Schafer Attorney-Norman J. OMalley and Laurence Burns ABSTRACT 7 Claims, No Drawings METHOD OF MAKING NONSAG FILAMENTS FOR ELECTRIC LAMPS BACKGROUND OF THE INVENTION 1. Field of the Invention The invention relates to tungsten filaments for electric lamps and particularly to filaments for operation at incandescent temperatures.
2. Description of the Prior Art Tungsten wire for filament use is normally made from a blend of at least 96 percent tungsten powder and various dopant materials. The dopants, examples of which are potassium, silicon and aluminum, are included in order to produce a long interlocking crystalline structure in a filament after the filament has been flashed, that is, recrystallized.
An ingot is pressed from a measured quantity of the blend and then resistively heated at a high temperature to increase its strength and density. The ingot is then mechanically worked into an elongated rod by swaging operations alternated with strain-relieving annealing steps, as is conventional in the art. Wire is produced from the swaged rod by drawing as, for example, is shown in U.S. Pat. No. 3,262,293 to Maclnnus, entitled Method of Manufacturing Wire. After the wire has been drawn to a desired diameter, it is wound on spools.
The wire thus produced has a fibrous structure, which is a result of the swaging and drawing operations, alternated with annealing steps. A fibrous structure is desirable since it results in ductility and workability of the wire and permits it to be coiled, and coil-coiled, into various filamentary shapes.
After the filament has been coiled and formed into its final desired shape, the fibrous wire structure is converted into a crystalline structure by flashing, that is, heating to a high temperature, generally above 2,000 C. Ideally, the preferred crystal structure for optimum resistance to sag consists of a long single crystal or, at least, relatively long interlocking crystals, the grain boundaries of which run lengthwise with the wire. The dopants, previously mentioned, aid in the grain growth which yields the desired crystalline structure.
However the filaments so produced do not always have adequate resistance to sagging during lamp operation. This is especially true in high-wattage, high-intensity lamps, such as quartz-halogen, where the normal filament operating temperature is but a few hundred degrees centigrade below the melting point of tungsten and where the wire size of the filament is relatively large, that is, above about l mils diameter.
It is felt that one of the reasons for poor sag resistance is due to the fact that mechanical deformation or cold work is not being uniformly distributed over the wire cross section as it is drawn through a die. In a typical case, the relative diameter of the die is such that the wire drawn therethrough is decreased about to 30 percent in area. Since it is mainly the surface of the wire that is worked by the die, most of its effect is confined to the surface of the wire and the immediately underlying portions thereof. The center of the wire is subjected to little mechanical work in comparison to the wire surface. This nonuniformity of mechanical work throughout the wire cross section is not completely alleviated by subsequent operations. As a result, the subsequent grain growth, on recrystallization, does not always yield a crystalline structure which is adequately resistant to sag.
It is an object of this invention to improve the nonsag properties of tungsten filaments by improving the manufacturing process of the tungsten wire.
SUMMARY OF THE INVENTION According to our invention, doped tungsten wire is stretched, subsequent to its final drawing operation, at a stress at least exceeding the elastic limit of the wire, in order to cause a permanent elongation. Preferably, the wire should be permanently deformed to at least about a 0.1 percent increase in length. The upper stress limit is that which breaks the wire. For practical consideration, however, we prefer not to exceed about percent of the ultimate tensile strength, in order to minimize the possibility of wire fracture during stretching.
The process is applicable to wire sizes normally used in lamp filaments, extending from about lor 2-mil wire for lowwattage lamps to about 30- or 50-mil wire such as is used in 10,000-watt lamps.
Filaments produced from the stretched wire were markedly improved in resistance to sag when compared with identical filaments made from the same wire, but which had not been stretched. For example, two groups of filaments were prepared from a spool of l9-mil tungsten wire, part of which had been stretched at 95 percent of ultimate tensile strength. The filaments were identical, with the exception that one group was made from the unstretched wire and the other from the stretched wire. The filaments were coiled coils and had 10 secondary turns, spaced about 1.7 mm. apart. 5,000-watt lamps were manufactured, utilizing the two groups of filaments, and operated at normal voltage, with the filaments in a vertical position.
After 8 hours of operation, the unstretched filaments had sagged noticeably; the lower three secondary turns were less than l mm. apart and the upper three turns were 3 mm. apart. In marked contrast, the stretched wire filaments had not sagged after hours of operation and had maintained their original spacing between secondary turns.
A comparison of the flashed structure of the two wires showed the stretched wire to have a superior interlocking crystalline structure. The grains were larger and longer than those of the unstretched wire, and had fewer transverse grain boundaries. Specifically, the stretched wire had an average of 2.5 grain boundaries per wire diameter, versus 5.0 for the unstretched wire. In addition, while the unstretched wire had many unlocked tiny grains near its surface, the stretched wire had very few. The improved crystalline structure, and grain substructure, of the stretched wire are primarily responsible for increased resistance to sag of the filaments.
A comparison of the stress-strain curves of the two wires, before flashing, showed them to have about the same tensile strength, but the stretched wire had an elastic limit about 70 percent higher. It is felt that this is due to a great increase in, and rearrangement of, the dislocations in the wire structure, resulting from the stretching operation. This is also an indication of the amount of cold work energy that is in the wire.
The improvements in crystallized filaments made from stretched wire may be explained by describing the general manufacturing process, from the blend of tungsten powder and dopants to the finished wire.
As previously mentioned, a thoroughly admixed blend of at least 96 percent tungsten powder and various dopants is pressed into a bar or ingot at pressures of 10 to 20 tons per square inch. In a typical case, the mixture comprised 99 percent tungsten, 0.5 percent potassium chloride, 0.4 percent silica and 0.1 aluminum oxide. The pressed ingot is very brittle and has insufficient strength to be mechanically worked. Therefore, it is fused, or sintered, at a high temperature, about 2,800 C., generally by passing electric current therethrough. The sintered ingot is still quite brittle but has sufficient strength to be mechanically worked.
Its structure consists mainly of equiaxed small grains and it must be heated to ductility during the subsequent swaging operation to prevent cracking and splitting. Temperatures above l,000 C. are usually employed, at least during the initial swaging steps. Reduction in wire diameter for each pass through a swaging die is about 10 to 20 percent. Thus, in a typical case, for an ingot having a l-inch-square cross section, about 12 passes are required to reduce the cross section of the rod to about 0.150 inches. Due to the nature of the swaging operation, in which the surface of the rod is subjected to hammer blows which decrease the diameter of the rod, the center of the rod is worked considerably less than the surface thereof; thus, the center tends to remain fine grained while the grains near the surface tend to elongate. To relieve stresses and soften the tungsten, the rod is annealed at temperatures above 2.000 C. two or three times during swaging. in the finished swaged rod. the original equiaxed grains of the sintered bar are broken up and elongated in the direction of the rod axis and are eventually only discernible as a bundle of fibers. It is this fibrous structure that gives the rod. and wire tlrawn therefrom. its ductility.
The further reduction of the rod is effected by drawing it through dies. The diameter at which the change from swaging to wire drawing is made usually depends on the size of the cross section of the original ingot and also on the size of the finished wire that is required. With ingots having a cross section about 1 inch square. the change from swagmg to drawing is generally made between 0.120 and 0.200 inches; for ingots about one-half inch square. the change is made between 0.060 and 0.120 inches.
Generally. the wire is reduced about to 30 percent in area for each pass through a drawing die. Although the cold work imparted to the tungsten by the die is concentrated at or hear the surface of the wire. as in swagmg. the wire is more severly worked by drawing than by swaging. Thus. the coldworked region resulting from wire drawing extends deeper. proportionately. toward the center of the wire than that from waging. However. the amount of cold work is not uniform throughout the cross section of the wire. The microstructure of the drawn wire is fibrous. the fibers being longer. narrower and more compact than those in the swaged rod and microscopic examination does not show the difference in cold work that exists throughout the cross section. However. X-ray patterns show that the structure at the center of the wire is deformed less than that near the surface of the wire. In addition. when such wire is subjected to various recrystallization temperatures, it is seen that the temperature of incipient recrystallization differs between the surface and core regions; there is also a difference in the rate of crystal growth.
As wire is drawn to finer sizes. say 1 to 2 mils. it is subjected to more and more cold working, i.e.. drawing, and the difference in cold work that exists between the core and surface of the wire decreases. This improvement in the uniformity of the cold work throughout the wire cross section can be seen by the X-ray studies and recrystallization tests mentioned above. However. a small amount of nonuniformity still exists in the finer wires. even though its effect on the structure of the recrystallized wire may be negligible. that is. the difference in amount of sag between filaments made from stretched wire and unstretched wire may be too small to be directly measurable. However. for wire sizes of about 5 to [0 mils and higher. the cold work difference between the core and surface lS significant and results in recrystallized structures that have measurable differences in sag in filaments made therefrom when compared with filaments made from wire that has been stretched according to this invention.
it is particularly important that the tungsten wire process according to this invention incorporate dopants. as previously mentioned. Wire made from pure tungsten. that is. without dopants. yields an unsatisfactory structure on recrystallization. The structure consists mainly of unlocked equiaxed crystals having many grain boundaries transverse to the axis of the wire. Such a structure has poor sag resistance. since the crystals can readily slip against each other under filament operating conditions. However. dopants. such as the aluiminum. silica and potassium previously mentioned. inhibit the formation of equiaxed crystals. at recrystallization. and promote the formation of the long interlocking crystal structure which is necessary to prevent filament sag. It is not completely understood how the dopant material effects the desired structure. since most of the material volatilizes at the high temperatures involved in fusion. swaging, annealing and drawing. However. enough residual dopants. or the effects thereof. remain distributed along the boundaries of the fibers to result in the desired elongated crystals at the proper recrystallization temperatures. as mentioned above.
A significant difference in the effects of drawing wire and stretching wire according to this invention can be shown by the amount of force required for each. It is known that at least 0 percent of the work done in drawing a wire through a die is spent in overcoming friction in the die. This means that the t'orce required to pull the wire through the die is more nearly proportioned to its diameter (or its surface area) than to its cross-sectional area. However. in stretching, the force required to stress the wire beyond its elastic limit is proportional to its cross-sectional area. The result is that stretching imparts cold work that is more uniformly distributed throughout the cross section of the wire than does drawing.
it is possible to explain the effect of wire stretching in terms of dislocations in the microstructure of the tungsten. Tungsten wire. as swaged. annealed and drawn. normally contained about 10 dislocations per square centimeter in random disarray. Upon straining and deformation of the wire. the dislocation density increases. At low strains the immobile dislocation density, and particularly that of immobile dislocation loops. ll'lCl'CflSCS at a faster rate than does the density of mobile dislocations.
With further straining, the dislocations tend toward grouping due to interaction of mobile dislocations with immobile dislocation loops. As the number of dislocations continues to increase. the groups. or tangles. which were formed earlier. now transform into long skeins of tangled dislocations which interconnect to form cells. Thus. upon straining of the wire beyond its elastic limit. a well-defined cellular network of skeins is formed. At this point. the dislocation density has increased to an average of about 10 dislocations per square centimeter and the altered grain structure. mentioned above. is obtained because of the effect of stretching on the disloca- T1011 mechanism and. in turn. its effect on grain growth.
The dislocation tangles are the predecessors of the subgrain boundaries of the new grain growth which occurs during flashing and which promotes the formation of the improved nonsag crystalline structure. mentioned above.
The process may also be used in the manufacture of fila- "nents made from alloys of tungsten. such as tungsten and rhenium. or thoriated tungsten. where dopants are incorporated into the initial blend in order to promote the forma- :ion ot'a nonsag interlocking crystal structure.
DESCRIPTlON OF THE PREFERRED EMBODIMENT The ultimate tensile strength of conventionally manufac- :ured l9-mil tungsten wire. wound on a spool. was determined by removing three short lengths from the spool and measuring them in an lnstron tensile tester. The average of the three sam ples was 43 kilograms.
92-meter length of the wire was then dereeled and removed from the spool. One end of the wire was clamped in a use and the other end was passed over a pulley. about 90 me- :ers away. and attached to a spring scale. A load of 40.8 kilograms was applied to the scale for 2 minutes. The wire stretched a maximum of 91 centimeters. or 1 percent. during the application of the load. Upon removal of the load. the wire relaxed to a permanent stretch or deformation of S3 centimeters. equivalent to 0.50 percent of the original length.
The stretched wire was then rewound on a spool to facilitate forming coils. that is. filaments. therefrom. Coiled-coiled filaments for a 5.000-watt quartz-halogen lamp, such as is shown in US. Pat. application Ser. No. 680,893. filed Nov. 6. 1967 by Peterson and assigned to the instant assignee. were wound from the stretched wire. The wire was primary coiled on a 29- mil molybdenum mandrel. annealed, secondary coiled on a L97-mil and sintered. After acid treatment to remove the molybdenum mandrel. the filament was supported on a tungsten and flashed rod and at a temperature over 2,000 C. for about l0 minutes in vacuum to convert the wire structure from fibrous to crystalline. Inserts were attached to each end or the filament for mounting and the filament. in final form. had an overall length of mm.. a body length of 30 mm.. and 10 secondary turns spaced 1.7 mm.. apart. When tested in the lamps. as described above in the summary of the invention,
the stretched wire filaments showed no noticeable sag after 120 hours while unstretched wire filaments sagged considerably after only 8 hours.
Other lengths of l9-mil wire from different tungsten lots were similarly stretched, at 40.8 kilograms, and the maximum elongation during stress, as well as the permanent elongation, are shown in Table I.
A continuous process that was used to impart a permanent deformation to wire from a spool involved stretching the wire between two pulleys. The apparatus used comprised a spindle for supporting a spool of wire, two rotatable pulleys separated from each other, and a second spindle on which a takeup reel was mounted. The second pulley had a circumference about onefourth or one-half percent greater than the first pulley. During operation, both pulleys were rotated in the same direction and at the same angular speed. Wire from the delivery spool was first looped around the first pulley, then looped around the second pulley and finally fastened to the takeup reel. As the pulleys were rotated, the wire therebetween was stretched an amount approximately equal to the difference in circumference to the two pulleys, and was then wound on the takeup reel. In order to increase the amount of elongation, as by a factor of two or three, two or three loops, respectively, could be wrapped around each pulley. Nineteen-mil wire that was stretched on such apparatus which was set at a 1% percent elongation yielded similar improved filaments as did the first simple stretching method. The permanent deformation of the 1.2 percent stretched wire was not measured but it was estimated to be between 0.4 and 0.7 percent.
Filaments for standard lOO-watt incandescent lamps were made from 2.5-mil tungsten wire, stretched 1.0 percent on the apparatus mentioned above, and compared with filaments made from unstretched wire. The former had 14 percent longer life and 8 percent less sag than the latter under normal life testing of the lamps.
Since tungsten wire for incandescent lamps usually has about 2 /2 to 4 percent elongation at the point of breaking during tensile testing, and is also quite uniform in tensile strength for wire made by a particular process, the amount .of stretching can be controlled by either controlling the stretching force or the elongation.
A particular 6-mil tungsten wire was tested and stretched according to this invention. The ultimate tensile strength of the unstretched wire was 5.6 kilograms and its elastic limit was reached at 2.9 kilograms or 52 percent of the ultimate tensile strength. At a stress of 3.2 kilograms, the permanent deformation was less than 0.1 percent. At 3.6 and 5.4 kilograms, the permanent deformation was 0.1 and 0.5 percent respectively. As the ultimate tensile strength was approached, the permanent deformation rapidly increased, to a maximum 2.4 percent just before rupture. At rupture, the total elongation was 2.8 percent.
In contrast to these results, 6-mil wire which had been previously stretched for 5 minutes at a stress of 5.4 kilograms, had the same ultimate tensile strength, 5.6 kilograms, but an increased elastic limit, 5.0 kilograms, which is 70 percent higher than that of the unstretched wire. As mentioned previously, this increase in elastic limit is due to the increase in, and rearrangement of, the dislocations in the wire substructure. The total elongation of the previously stretched wire, at rupture, was only 1.8 percent, much less that that of the unstretched wire.
Another test of l9-mil wire was made to determine the ef fect of different amounts of stretching on the flashed crystalline structure of the wire. One sample was stretched 0.9 percent at percent of ultimate tensile strength; a second was stretched 1.1 percent at percent of ultimate tensile strength; and a third was stretched 1.2 percent on the apparatus mentioned. All three samples were flashed at 2,200 C., along with an unstretched wire sample, used as a control. All three stretched samples had better nonsag interlocking crystalline structures than did the control, that is, longer, and fewer, interlocking crystals with substantially longitudinal grain boundaries. The grain structure was improved with increased elongation at stretching and the 1.2 percent stretched sample approached crystal structure.
The preferred method of practicing the invention would be on equipment of the type previously mentioned, wherein wire from a spool could be stretched on two driven pulleys having unequal circumferences. Such a spool could accommodate a continuous length of thousands of meters of wire. The minimum length of wire to which the process is applicable is determined by the total length of the desired filament which is coiled from such wire and would be that length, say, 10 or 20 mm., plus a few millimeters at each end for clamping.
Although the process has been particularly applied to improving the nonsag properties of incandescent filaments, we have noted other improvements in filaments made from wire processed according to this invention. For example, such filaments twist and distort less than conventional filaments when they are flashed in a lamp after mounting.
It is apparent that modifications can be made within the spirit and scope of the instant invention but it is our intention, however only to be limited by the scope of the claims.
1. In a process for manufacturing wire for electric lamp filaments the steps which comprise: preparing an ingot of doped tungsten powder; fusing, swaging and drawing said ingot into a wire; and stretching the wire beyond its elastic limit but below its tensile strength to a permanent elongation of at least 0.1 percent.
2. The process according to claim 1 wherein the force used in stretching wire is in the range between its elastic limit and about 95 percent of its ultimate tensile strength.
3. The process according to claim 1 including the step of stretching said wire between two pulleys having unequal circumferences.
4. The process according to claim 2 including the steps of forming coiled filaments from said stretched wire and recrystallizing said filaments.
5. The process according to claim 4 wherein said filaments are made from said wire having a diameter of about 1 to 50 mils.
6. The process according to claim 4 wherein the dopants for said tungsten powder include potassium, silica and aluminum.
7. Tungsten filaments manufactured according to the process of claim 4.