Page images


DISCUSSION OF THIS PAPER IS INVITED. It should preferably be presented in person at the New York meeting, February, 1916, when an abstract of the paper will be read. If this is impossible, then discussion in writing may be sent to the Editor, American Institute of Mining Engineers, 29 West 39th Street, New York, N. Y., for presentation by the Secretary or other representative of its author. Unless special arrangement is made, the discussion of this paper will close Apr. 1, 1916. Any discussion offered thereafter should preferably be in tho form of a new paper.

Recrystallization of Cold-Worked Alpha Brass on Annealing*



(New York Meeting, February, 1916)

DURING the past year considerable work dealing with the mechanical properties and microstructure following the anneal under uniform conditions of certain types of commercial rolled brass has been done in this laboratory. Since these experiments were conducted with the active coöperation of the Bridgeport Brass Co., involving free use of its product, and have thus furnished information which is characteristic of some of its special mixtures, full publication of the numerical relationships and comparative results encountered does not seem justifiable. Some of the more general features of this work are, however, available and it is our present purpose to use the material in question, along with some results of special tests, in discussing the characteristics of recrystallization as related to the degree of hardening by strain and the ordinary annealing variables. We desire to make acknowledgment of the many courtesies extended by Mr. Charles Ferry, Metallurgist of the Bridgeport Brass Co., and the value of his coöperation, which has greatly stimulated the study of brass in this laboratory.

A clear and systematic presentation of the principal relations between temperature and time of anneal, the changing physical properties and microstructure and the degree of previous reduction by cold-rolling, was first made, in the case of rolled brass and copper, by the French engineer, C. Grard, 1 in 1909, although much fragmentary information of this general character had appeared earlier and the main facts involved were undoubtedly known to many brass specialists through their own private research. From this work, it appears that in the cartridge mixture (67 per cent. copper, 33 per cent. zinc) the properties imparted by coldworking within the limits, 13 to 75 per cent. reduction of area by rolling

* Metallographic communication from the Hammond Laboratory of the Sheffield Scientific School, Yale University.

† Assistant Professor of Metallurgy, Sheffield Scientific School, Yale University.

Metallurgical Department, Bridgeport Brass Co. 1 C. Grard: Cuivre et Laitons à Cartouches, Révue d'Artillérie, Feb.-Apr., 1909.


(Original Arca - Final Area

0.13 to 0.75), are not affected by annealOriginal Area ing for a 50-min. period at any temperature below 200°C. At or in the vicinity of this temperature, metal which had been reduced 75 per cent.

owed a sensible decrease in tensile strength and a corresponding increase in elongation, while metal which had been reduced 13 per cent. showed no change. An intermediate value of reduction, 56 per cent., the one which Grard evidently used in most of his tests, gave earliest evidences of change at about 250°. Thus, the effect of anneal in its incipient stages depends somewhat upon the previous treatment of the metal. As the temperature of anneal is increased, the change in properties becomes more marked in each case, until, at about 325°, the individual curves coincide, and from this point, the final condition produced by anneal is independent of the previous reduction within the limits investigated. With regard to peculiarities of inflexion, Grard distinguishes a number of zones corresponding to regions of permanent, or of changing properties in one sense or another. In accepting such conclusions, the empirical character of the work must be borne in mind. Grard has shown, as indicated above, that the degree of reduction by rolling determines the temperature range of permanency in the original strain-hardened condition. The reasons for this and the prevailing possibilities of variation will at least be indicated from a general point of view in the present paper.

In the principal set of curves, Grard adopted a period of 30 min. in which to complete the anneal at any given temperature. From his temperature-time curves it appears that in all cases the rate of change of physical properties, which is greatest in the earlier stages of the anneal, has become very slow at the expiration of the whole period. Moreover, as the temperature of anneal is raised, a decreasing fraction of the total 30-min. period suffices to produce most of the annealing effect. That this is a generally prevalent condition was shown by T. K. Rose 2 who, in studying the annealing properties of a variety of coinage alloys by the scleroscopic method, makes the following statement: “The annealing action at any given temperature is most rapid at first and gradually dies away, so that a state of equilibrium is not attained in any short space of time, except perhaps at high temperatures. For example, equilibrium apparently was not attained in the gold-copper, silver-copper, or nickel-copper alloys, or in pure nickel, in 16 days at 300°. At high temperatures, almost the whole result is obtained instantaneously, though a little extra softness is got by prolonging the time. A few minutes at such a temperature cannot be replaced, for practical purposes, even by hours of annealing at much lower temperatures.”

This work of Grard has been widely circulated among brass metallur

2 T. K. Rose: On the Annealing of Coinage Alloys, Journal of the Institute of Metals, vol. viii, pp. 86-125 (1912).

gists through the medium of reprint, translation, and review. Doubtless it has supplied information of direct numerical value in some quarters, while in others it has served as a welcome guide in the development and coördination of testing data. The latter type of service appears to us to be the most desirable function of publications of this sort. The individual brass mill may be urged to develop its own data of standardization, or control, whereby all local matters of manufacturing practice, peculiarities or specialties in composition, and other individual features, may receive consideration in their proper environment, leaving broader conceptions subject to development and modification by exchange of ideas through the scientific press.

With these ideas in mind, the following notes are selected from work which we have done, with the hope that they may be of some service to those who are interested in the annealing characteristics of brass. In these notes some addition to the commonly available information on brass metallurgy has been made chiefly along the following lines:

(a) By presenting a few carefully conducted physical tests to show the general direction and magnitude of changes which are effected by heating cold-rolled strips below an effective annealing temperature.

(6) By determining the periods of time necessary to produce a measurable amount of softening in one kind of strain-hardened material at a succession of temperatures below the region of rapid effects.

(C) By emphasizing the micrographic aspects of strain-hardening and recrystallization in metal which has received both light and heavy reductions by rolling.

(d) By discussing the relations between temperature and time of anneal, degree of deformation, and structural alteration in alpha brass, with an attempt to give some general explanation of the facts involved.

(e) By showing comparisons between the ordinary physical properties and the grain size, both given as functions of the annealing temperature for a fixed period of anneal. Changes Produced in Cold-Rolled Strips by Heating One-Half Hour at 200°C.

Bengough's experiments on 70/30 brass, while devoted largely to the question of "burning," extend over a wide enough range to furnish a broad basis of comparison with those of Grard. Both observers have recorded a similar sequence of changes in the strength-properties after a constant period of anneal at different temperatures, except that, according to Bengough, the metal becomes several per cent. stronger after anneal at temperatures in the vicinity of 300°, while Grard, in his curves, gives no indication of an increase in tensile strength before the normal (reverse) effect, due to anneal at higher temperatures, sets in.

It may be urged that variations of this order may be expected, in


Bengough and Hudson: Heat Treatment of Brass, Journal of the Institute of Metals, vol. iv, pp. 92–127 (1910).

view of unavoidable slight imperfections in the test pieces, or on account of testing errors, irrespective of any actual annealing effect. It is obviously necessary to obtain the mean values from a moderately large number of tests before reasonably certain conclusions can be drawn. We are not certain whether Bengough used several test pieces for each experiment, or relied upon a single piece to represent each anneal. In any event, it it true that, in all three series embracing low-temperature anneals, viz., copper 69.4 per cent., 4-in.round bar; copper 71 per cent., 34-in. round bar, and copper 71 per cent., Y8-in. wire, somewhat higher tensile strength was obtained in samples annealed between 285° and 320° (different series), than in the original hard-rolled material. A critical examination of Grard's data is not possible, since the actual values are not tabulated or shown in his diagrams.

Our method of testing the point at issue was as follows: Strips of the dimensions, 0.1 by 0.5 by 8 in., were sawed from a cold-rolled sheet of 70/30 brass* which had received a dead soft anneal in the mill previous to the final reduction of 40 per cent. Five of these strips were tested in a 50,000-lb. Riehle universal testing machine in the cold-rolled condition. Another set of five was tested after annealing for a period of one-half hour at 200°. The material under present consideration cannot be heated for a half-hour period at temperatures in the neighborhood of 300°, viz., at temperatures similar to those given by Bengough, without appreciable softening and visible recrystallization (true annealing effect). One of our tests at 300° showed a softening of four points on the scleroscope scale and abundant recrystallization, as illustrated in Fig. A, Plate I. The main reason for this difference lies in the moderate initial reductions by cold-working received by Bengough's rods, as compared with the heavy initial reduction received by our strips. While Bengough does not give the reductions used in his material, the above conclusion is justified by the materially lower tensile strength and higher elongation of his bars, as compared with our strips; all tested in the cold-rolled condition.

The annealing was conducted in an electric resistance furnace of rectangular box-shaped type containing a central tube of alundum (14-in. long by 1 in. bore) closely wound with "Nichrome" ribbon cemented into place with alundum cement. To avoid rapid radiation from this heating unit, it was embedded in a mixture of granular fire clay and silica.

Two test strips were treated simultaneously (except in the annealing of the fifth strip). The strips were clamped in a vise and a hole large enough to accommodate the thermo-couple tube was drilled at one end in such manner as to bore both pieces simultaneously for a distance of approximately 12 in. in the direction of their length. Two wire clasps held the strips in this position and served as a support for the metal when placed

* 来

Analysis: Copper, 69.35; iron, 0.04; tead, 0.02 per cent.

in the center of the furnace tube. All bars for tests and micrographic observations were annealed in this way. After anneal, the drilled ends were cut off and micrographic observations made in the immediate vicinity of the region where the thermo-couple junction had previously rested. It is to be noted that the junction was virtually imbedded in the metal at this point.

Owing to increased radiation at the ends of the furnace, the temperature of the tube at the center was found by preliminary experiments to be about 10° higher than at the ends. This difference is small and is actually less in the bars themselves, as their high heat conductivity tends toward equalization of temperature, i.e., flattens the thermal gradient.

Probably the most effective method of flattening the thermal gradient in a wire-wound tube furnace is to make use of independent heating coils around the ends which may be run at a temperature just high enough to compensate for the increased radiation at the ends. Gray' has obtained ideal conditions with this type of construction. By decreasing the distance between turns from center to ends, a compensating effect may be secured, but with any given winding the desired effect will be secured only when the furnace is running at certain temperatures; above and below this range, the ends will be over- and under-compensated, respectively.

Gray's method complicates the construction of the furnace and requires independent circuits along with independent control, while the other method requires a number of different resistors to be used for different temperature ranges and there is great difficulty in properly spacing the windings. We have sought to improve the gradient without introducing another element of control and without greatly complicating the construction of the furnace. Lately, we have used furnaces in which the thickness of the heat-insulating covering around the tube (granular fire-clay) increases from the center to the ends, thus permitting more rapid loss of heat in the region of the center. In principle, this is similar to varying the space between turns, but it is far easier to accomplish, and any adjustment may be changed in a few minutes. This modification merely requires the construction of false inner sides of sheet iron which are bent to give the proper inclination from center to ends and slipped within the furnace box (made of asbestos wood). The granular fire-clay is then filled in with a scoop. It is clear that, according to this construction, the thickness of the heat-insulating covering varies from the center of the tube to the ends, but the depth is constant, viz., the correction applies to the sides of the tube but not to the top and bottom. While it would be preferable to make the correction concentric with the tube, this would entail greater complication and less flexibility. These

* A. W. Gray: Production of Temperature Uniformity in an Electric Furnace, Bulletin Bureau of Standards, vol. x, pp. 451–473 (1914).

« PreviousContinue »