In this paper, it is proposed to outline a method for the determination of melting points of those metals and alloys having high fusion temperatures. The application of the method as used to determine the melting points of alloys of tungsten and molybdenum will be given. It will be seen from these results that tungsten and molybdenum form a completely isomorphous series. This is also verified by the accompanying photomicrographs.

Tungsten and molybdenum have been found to crystallize in the same system, namely, isometric, and the crystal units are cubes. These determinations were made by examining etching pits in the pure metals.

Equipment Required

Fig. 1 shows a partial section of the apparatus used for determining the melting points of the alloys. W1 is a water-jacketed metal housing with a mercury seal m2 at the bottom, h1 is a hydrogen inlet and h2 a hydrogen outlet, c1 is the top electrode which is water-cooled, as indicated by W2. This electrode is supported by the post p, to which it is fastened by means of an adjustable sleeve. C2 is the bottom electrode which floats in a bath of mercury m1. This mercury is contained in the casting k, which is water-cooled, as shown by W3. The whole apparatus is mounted on the base b made of electrical insulating material. The electrodes are suitably connected to a current regulator, and the voltmeter and ammeter are properly inserted in the circuit.

Method of Making Determinations

Wires (S), 2¾ in. between electrode contacts and 0.030 in. in diameter, were inserted in the electrodes and hydrogen was allowed to flow through the inclosed chamber until the air was expelled. Electric current was then passed through the wire, starting with low amperage and gradually increasing until fusion occurred. The time of fusion is easily determined because the circuit is broken and the ammeter ceases to register. The readings of the voltmeter and ammeter are recorded with each increase in electric current so the fusion wattage can be accurately calculated.

Besides using, samples of the same size, other conditions must be maintained constant; for example, the flow of hydrogen through the inclosed chamber, the flow of water through each of the water jackets, and the temperature of the water at the intake. The electrical resistance of the electrodes and their connections is so small that it can be neglected in the calculation for drop of potential when compared to the drop of potential between the two electrodes.

Calibration of the Apparatus using Samples of Known Melting Points

For calibrating the apparatus wires of pure platinum, pure molybdenum and pure tungsten were used. These wires were all of the same size and same length. The number of watts to fuse each is shown in Table 1.

By considering the fusion wattage of tungsten as 100 per cent., the fusion wattage of molybdenum is 22.2 per cent, and of platinum, 8 per cent.

The curve in Fig. 2 is obtained by plotting the following temperatures against the corresponding fusion wattage percentages:

The temperature 3,300° C. is recommended as the most probable figure for the melting point of tungsten by Dr. Langmuir, and also by the Nela Laboratories.

If a sample of unknown melting point, having the same diameter and same length as the standardizing samples, be fused in the apparatus, its fusion temperature can be interpolated from its fusion wattage, from the curve shown in Fig. 2. A discussion will be found below outlining the possible sources of error in the determination of melting points by this method.

Results Obtained with Tungsten-Molybdenum Alloys

The results are shown in Table 3, and the points on the equilibrium diagram are shown in Fig. 3.

The solidus and liquidus curves are drawn tentatively, it being, of course, impossible to determine either of these curves by this method.

The point of breakdown of the alloys under the action of the electric current probably lies between the solidus and the liquidus. Most of the points shown in the diagram in Fig. 3 lie between the proposed solidus and liquidus curves. The fact that two points lie outside of the area included between these two curves may be explained by saying that the accuracy of any of the determinations may be subject to an error of from 30 to 50° C.

As an example to show how the temperature is obtained from the fusion wattage, let us consider the alloy containing 50 per cent, tungsten and 50 per cent, molybdenum. The fusion wattage of this alloy was 710, and the fusion wattage of tungsten was 1,800.

710/1,800 x 100 = 39.4

This alloy fuses, therefore, at 39.4 per cent, of the fusion wattage of, tungsten. Referring to Fig. 2, it will be seen that 39.4 per cent, of the fusion wattage of tungsten corresponds to a temperature of about 2,890° C.

The equilibrium diagram suggests that tungsten and molybdenum form a series of alloys which are completely soluble in each other, both in the liquid and solid states, and that the melting point of any alloy of the series will lie between the melting points of the end members. The photomicrographs confirm this indication.

Mennicke observes. that two compounds between tungsten and molybdenum occur, namely, W2M03 and WMo. The former would contain 79.3 per cent, tungsten and the latter 65.7 per cent. Samples closely approximating these analyses, as well as those containing more and less tungsten, have solid-solution structures. All of the evidence of the melting points indicates that tungsten and molybdenum form no inter-metallic compounds.

Microscopic Examination of Tungsten and Molybdenum Alloys

Fig. 4 is a photomicrograph of nearly pure tungsten at a magnification of 320 diameters. The impurity in this sample is non-metallic. A few globules of this can be seen in the micrograph. This sample contains by analysis about 99.8 tungsten.

Figs. 5 to 7 inclusive are photomicrographs at 320 diameters of the various alloys of tungsten and molybdenum, as indicated in the accompanying descriptions. It can be readily seen that all of these alloys are solid solutions.

Judging from the micrograph (not here reproduced) alone, of an alloy of 20 per cent, tungsten and 80 molybdenum, the black areas might be considered to be due to a second phase; in fact, these regions represent areas rich in-tungsten. Under high power (1,200 diameters) the separation of the two phases could readily be detected. That they were not in

equilibrium was also demonstrated by an additional heating which completely homogenized the alloy. In this connection, it would probably be advisable to mention the manner in which these alloys were made.

The oxides of both tungsten and molybdenum were obtained in a very pure powdered state, and reduced to powdered metal by hydrogen.

The alloys were made up by thoroughly mixing the tungsten and molybdenum powders in the proper amounts. The mixed powders were then pressed into briquets which were sintered at about 1,300° C. in an atmosphere of hydrogen and were then heated by electric current to about 100° C. below the fusion points (which were previously determined with a separate series of briquets of these alloys) for a period of 12 min. The alloys, after receiving the above treatment, were completely crystalline. They may be ground by an emery wheel to a certain diameter, and the fusion points determined on these samples. They may also be heated and rolled or swaged to any desired diameter. The alloys used in these experiments were swaged to 0.030 m. diameter, the original cross-section of the briquets being about 1/16 sq. in.

The micrographs, with the exception of Fig. 6 (which is a section of

0.080 in. wire), represent longitudinal sections of these 0.030 in. diameter wires after fusion. The portions shown in these micrographs represent sections which have not been fused but which have been heated to temperatures very near their melting points. It will be evident that fusion of the wire will take place at a point approximately equidistant between the two electrodes and that adjacent portions of the wires would be heated to very near the melting point. In these sections, therefore, the solution of the tungsten and molybdenum must take place while both metals are in the solid state, or at least while the tungsten-rich portion is in the solid state. In case solution and diffusion have not been complete, two phases, one rich in tungsten, the other rich in molybdenum, will be present. In some of the briquets, the solution was not complete during the first heating, but was complete after the second heating. All evidences of the existence of two phases were removed when sufficient time was given the samples at temperatures near their melting points.

These samples were also examined after fusion in the parts which had been molten. There was no evidence of a second phase in any of these samples; that is, they were all true solid solutions.

Fig. 8 represents a micrograph of pure molybdenum. By analysis, it contains about 99.9 per cent, molybdenum. Its purity is also suggested by the micrograph.

These alloys are all readily etched by boiling hydrogen peroxide. Etching pits in the various alloys were produced by a comparatively long attack—say, 3 or 4 min. in boiling hydrogen peroxide. All of the evidence from the etching pits points to the conclusion that the crystal units are cubes. The alloys of tungsten and molybdenum, when made up according to the above description, seem to form etching pits very much more readily than either of the pure metals. This might be attributed to the existence of small particles of either the tungsten-rich or the molybdenum-rich portions, which, by the difference in composition, would facilitate solution at that point, thus easily forming the etching pits.

It is of interest in this connection to note that the average grain as seen in the micrographs represents about 1,000 of the original particles of the powdered metals. This free grain growth is positive evidence of free solution of the metals in each other. Had the metals remained as separate phases, each, would have impeded the coalescence of the grains of the other.

Probable Accuracy of Results

The quantity of heat represented by the fusion wattage is independent of the quantity of heat actually- necessary to raise a mass of metal, similar to that used in these samples, up to its melting point. It depends on the ability of the sample under investigation to dissipate heat in the apparatus. The thermal balance may be expressed by the equation:

Heat added electrically = heat dissipated by radiation, convection and conduction.

The conditions obtaining within the apparatus are such that if the fusion,of tungsten is represented by 1,800 watts, 1,750 watts could be dissipated for a long period of time without fusion of the tungsten.

The probable errors in the determinations are outlined below:

Errors due to:

1. Differences in emissivity of the various alloys under investigation.
2. Differences in their specific heats, heat capacities and heat conductivities.
3. The selective volatilization of molybdenum.
4. Variations in the flow of water through the water jackets, of the temperature of the water at the intake, and of the flow of hydrogen through the inclosed chamber.
5. Slight variations in the diameter of the wires.
6. Possible changes of melting points due to chemical or physical combination of alloys with hydrogen.
7. The personal equation in drawing the per cent, of tungsten fusion wattage-temperature curve.
8. The readings of the electrical instruments.

The differences in emissivity of the various alloys will make but a slight difference in the fusion wattage, for the reason that only about one-twentieth of the total heat dissipated is lost by radiation, and the electrodes combined.

The errors due to No. 2 will be very slight, for the same reason.

Langmuir and McKay report that at 2,800° K., the heat loss from a tungsten filament in hydrogen at 750 mm. pressure, is 10½ times greater than that due to radiation; and at 3,500°K., the heat loss due to hydrogen is 11½ times that due to radiation. The conditions obtaining in the apparatus used by the writer were such that the losses due to the hydrogen are considerably more than is reported by these authors.

Dr. Irving Langmuir suggests errors due to the selective volatilization of molybdenum. The apparent high melting points of the alloys containing 5 per cent, tungsten, 95 per cent, molybdenum, and 5 per cent, molybdenum, 95 per cent, tungsten, may be partially due to this selective volatilization. The enrichment of the alloys in tungsten, due to this cause, could not have been more than 3 per cent, in any case, as was indicated by the diameter measurements of the wires before and after fusion.

The errors due to No. 4 are probably greater than those due to all other variations combined. Errors due to these causes, however, are not necessary if the proper precautions are taken. These precautions would consist of supplying the water for the water jackets from a constant-level and constant-temperature tank. The same conditions should be fulfilled with the hydrogen.

To find out the approximate magnitude of these errors, samples of the same wires were fused on different days, so that the valve-settings for jacket water and hydrogen had to be re-made and re-adjusted.

Several determinations showed a maximum variation in the melting point of molybdenum, as interpolated from the curve in Fig. 2, of about 40° C., and a maximum variation in the melting point of tungsten of about 10° C. The results on tungsten seem very favorable when it is considered that some methods for determining its melting point may vary 15.0° C.

Allowing for other errors, encountered in the determinations, it is reasonably accurate to assume that the maximum errors toward the molybdenum side of the diagram, are about 50° C., and toward the tungsten side of the diagram, 30°C. Allowing for errors of this magnitude, the points in Fig. 3 would fit nicely between the solidus and liquidus curves.

Errors due to slight variations in the diameters of the wires will be very small, owing to the fact that the ability of small wires to dissipate heat in a gas depends largely upon the existence of a relatively thick film of gas surrounding the wire. The thickness of this gas film is dependent largely upon the pressure of the gas, and not upon the diameter of the wire. The fact that these errors were slight was confirmed by an experiment, and was also predicted by Dr. Langmuir from his extensive work along this line.

The suggestion was made by Dr. G. K. Burgess that hydrogen might, by reason of either chemical or physical combination with the alloys, change their melting points. The writer has no knowledge concerning this. It might be well to add, however, that the handling of this type of alloys at high temperatures is done almost entirely in hydrogen.

To ascertain the approximate errors due to the drawing of the curve in Fig. 2, it was plotted independently on two sheets of paper. The maximum difference between the two curves was about 10° C. at about 60 per cent, of the tungsten fusion wattage. To facilitate the interpolation of temperatures from fusion wattages, this curve was drawn on a large piece of cross-section paper, so that the reading errors would be reduced to a minimum.

The wattage, as determined by volts and amperes, was, of course, quite accurate, so errors from this source were negligible.

Concerning the probable accuracy of results obtained by this method, Dr. Irving Langmuir writes: “I think the method you are using for estimating the temperature of filaments should give reasonably accurate results.” He points out, among other things, the desirability of making chemical analyses after heating, to determine the actual composition of the alloy at the time of fusion. Dr. G. K. Burgess writes: “This method seems to me susceptible of very considerable accuracy.”

The author is convinced that the method is susceptible of greater accuracy than was obtained; that, with careful attention given to the construction and operation of the apparatus, melting points can be determined within 10° C. The method promises to offer solutions for several equilibrium diagrams of the higher melting-point metals.

Crystal System of Tungsten and Molybdenum

Fig. 9 shows etching pits in tungsten, magnified 775 diameters. As can readily be seen from the micrograph, these etching pits represent sections of cubes cut by a plane parallel to a face.

Fig. 10 shows triangular etching pits which are formed by a plane cutting three faces of a cube.

Fig. 11 represents the intersection of the six faces of a cube by a plane. By changing the direction of the illumination, the writer was able to examine the bottom of this etching pit and could easily see that it was a section of a cube.

It was noticed in some of the micrographs that the etching pits instead of having straight line sides had curved sides. Goldschmidt thinks that these curved fines are due to the convection currents set up in the etching solution by the unequal rate of attack at various points. The solution will be used up most rapidly at places where the greatest amount of surface is exposed. This will cause a flow of fresh solvent by diffusion. The corners of the etching pits will offer greater resistance to the flow of convection currents than the adjacent portions, and hence will not receive so much fresh solution. The lines bounding the; polygons are thus dissolved at unequal rates and become curved instead of straight.

A great many etching pits in all conceivable positions were examined microscopically. From these examinations; the author has very little hesitancy in saying that both tungsten and molybdenum crystallize in the isometric system and the form of their crystal units is the cube.