Copper Refining: Explained Step-by-Step

In refining copper, the metal is melted down in a reverberatory furnace in a more or less oxidizing atmosphere and then further subjected to an oxidizing smelting in order to eliminate the common impurities, most of which have a stronger affinity for oxygen than has copper. In these operations some of the copper is oxidized to cuprous oxide and dissolved by the metal bath. When the quantity of dissolved cuprous oxide has reached about 6 per cent, the metal is said to have been brought to “ set-copper.” A button-sample will show a depressed surface and, when broken, a single bubble at the apex of the depression; the fracture will be brick-red and dull. It is essential to carry the oxidation to this point in order to know that the impurities have been oxidized as far as it is possible under the given working-conditions. Nearly all the cuprous oxide of the set-copper is now reduced to the metallic state by poling, when “ tough-pitch ” copper will be obtained. A button-sample will show a flat surface. Upon breaking, it will be found that the former bubble has disappeared and that the fracture has become rose-colored and shows a silky luster. The quantity of cuprous oxide allowed to remain in the copper will vary with the impurities still present in the metal and with the degree of pitch that it is desired to reach. It is essential for the general physical and the mechanical properties of the resulting copper that such impurities as arsenic, antimony, bismuth, lead shall be present in the oxidized state, as they are then less harmful than when present in the metallic state. Refiners commonly distinguish “ ingot- or cake-pitch ” and “ wire- bar pitch;” copper brought to the former contains more cuprous oxide than the latter. These two pitches are, however, not absolutely fixed; they vary with the practice of the individual refiner and with thickness of the cake or bar that is to be cast:

The thicker the piece, the more oxygen will have to remain in the metal, if a flat surface is to be obtained. A third degree of pitch aimed at is that required by very thin castings, such as electrodes 0.5 in. in thickness. As this pitch lies beyond that of wire-bar copper and differs from it more than to permit its being designated merely a shading, it may be called “ plate-pitch.”

Little was known of the structural relations that existed between copper and cuprous oxide until 1900, when Heyn published the results of his investigations on ” copper and oxygen.” He took pure copper wire, cut it into small pieces, made up charges of 450 grams each, added to them varying quantities of pure cuprous oxide, melted the mixtures (excepting sample No. 1) in graphite crucibles lined with porcelain, inserted into the fused charges the protected couple of a Le Chatelier thermo-electric pyrometer and made the cooling-curves. Sample No. 1, copper wire alone, was fused in a graphite crucible without a lining in order that the graphite might have a reducing effect upon the small quantity of oxide present, and the resulting fused copper be as free from it as was possible. The results of his experiments are given in Table I. and are represented graphically by the freezing-point curve in the subjoined diagram (page 677).

It will be seen from Table I. that upon cooling, only the tests Nos. 1, 4 and 5 show a single fall of temperature, which means that they do not pass through a pasty stage, but freeze suddenly. With the other tests there is a gradual freezing from the beginning to the end of the solidification. With all samples, excepting test No. 1, the cooling is accompanied by surfusion, the depressed temperature rising in every instance to the fixed point of 1084° C. This temperature, corresponding to the freezing-points of tests Nos. 4 and 5 containing 3.4 and 3.5 per cent Cu20 respectively, which have only one distinct point of solidification, must be the temperature of the eutectic. The V-shaped freezing-point curve of the diagram, plotted from the data in Table I., and the photomicrographs, Figs. 1 to 8 inclusive, show this fact clearly. Figs. 4 and 5, the eutectic of copper and cuprous oxide with 3.4 and 3.5 per cent Cu2O respectively, show the characteristic structure, i.e., a conglomerate of the two components arranged more or less in alternate plates which do not cross one another. With alloys of diminishing percentages of cuprous oxide, shown in Figs. 3, 2 and 1, the photo-micrographs give the dark net-work of eutectic and the light meshwork of copper; with alloys having percentages of cuprous oxide greater than 3.4 or 3.5, the photomicrographs, Figs. 6 and 7, show patches of cuprous oxide increasing in size embedded in eutectic mixture.results-of-heyns-experiments

The alloys of copper and cuprous oxide, when in the molten state, form homogeneous solutions similar to salt solutions. Upon the solidification of copper-alloys containing less than 3.45 per cent Cu2O, the cuprous oxide falls out completely and does not form a solid solution with copper. This is seen clearly in Fig. 1, where the net-work of the eutectic is still visible, when the copper contains only 0.08 per cent Cu2O. Should any cuprous oxide form a solid solution, its quantity would have to be less than 0.08 per cent. Whether in an alloy with more than 3.45 per cent Cu2O, the patches of cuprous oxide (Figs. 6 and 7) are pure cuprous oxide, is not settled, but in all probability this is the case.

Finally, Fig. 8 represents the same sample as shown in Fig. 2, viz., copper with 1.16 per cent Cu2O. In the first case, the specimen has been heated to 1100° C. and then quenched in water of 10° C.; in the second, it has been allowed to cool slowly. The disconnected points in the photo-micrograph of the quenched specimen show how the free development of the inner structure has been arrested; in contrast, Fig. 2 illustrates clearly how the eutectic, if given sufficient time, forms a connected network.

The leading statements of Heyn’s paper, including eight of his nine photo-micrographs, have been repeated here, for the reason that they have an important bearing upon the work to be described; in fact, the suggestion in his conclusions that observation of the microstructure of refined copper might be substituted for the lengthy determination of oxygen was the cause of undertaking the present investigation. This embodies the examination of fractures of samples taken in different stages of refining, the determination of the oxygen-content, the preparation of photomicrographs, and, lastly, the planimetric measurement of enlarged photomicrographs, with calculation of the percentage of oxygen.

Description of Samples

Button-samples for this investigation were kindly furnished by Mr. W. T. Burns, of the Boston & Montana Consolidated Copper and Silver Mining Co., Great Falls, Mont.; by Mr. M. B. Patch, of the Buffalo Smelting Works, of the Calumet & Hecla Mining Co., Buffalo, N. Y.; and by Mr. G. M. Luther, of the Nichols Chemical Co., Laurel Hill, N. Y.

Sample No. 1 of the Boston & Montana Co. represents cathode copper after it has been melted down in the reverberatory furnace and skimmed, but not rabbled; No. 2 was taken after the rabbling had been completed and the stage of set-copper reached; No. 8 is the sample after the poling has been finished, and the copper is ready to be ladled into wire-bars. The tests made at the works give: silver, 0.8 oz. per ton; arsenic and antimony, 0.0035 per cent; conductivity (hard-drawn) 97.5 per cent; tensile strength, 64,200 lb. per sq. in.; elongation, 1 per cent; torsion-twists in 6 in., 89.

Sample No. 0 of the Buffalo Smelting Works represents set- copper; the remaining six of the set, Nos. 1, 2, 3, 4, 5 and 6, were taken at intervals of 15 minutes during the poling-period: No. 1 was cast after the poles had been in the furnace for 15 minutes; No. 6 is finished refined copper brought to a pitch at which ingots or cakes are cast. Samples Nos. A and B, from another charge, represent copper brought to ingot-pitch and wire-bar pitch respectively, special care having been taken to allow the samples to cool slowly.

The sample of the Nichols Chemical Co. represents plate- pitch, that is, the pitch desired for casting thin electrodes.

Fractures of Samples

The fractures reproduced in Figs. 9 to 18, inclusive, were prepared in the usual way. An incision about 0.125 in. in depth was made across the convex side of a button with a cold-chisel, the button then clamped in a vise with the incision just protruding above the jaws, and given one or more shearing blows with a heavy short-handle hammer. With set-copper, one blow was sufficient to break the specimen in two; the nearer the sample approached tough-pitch copper, the larger was the number of blows required to obtain a fracture.

Several experiments were made to find the best light and time of exposure necessary, the position of the specimen with regard to the light, and the proper magnification to bring out the details in a photograph. A reversible-back Premo camera, 4 by 5 in., was used with sunlight, the specimen having been placed on a white background. Exposures of 40, 80 and 160 seconds were tried, using a No. 32 diaphragm. With negatives half the size, the same size, one and one-half times and twice the size of the original, exposures of from 40 to 80 seconds gave good results. Trying back-light and side-light, it was found that the latter brought out the structure more satisfactorily than the former. A magnification of one and one-half was necessary to show clearly the details in the photograph.

As it is not easy to have constant conditions with sunlight, electric light from a 16-candle-power incandescent lamp was substituted and an enlarging camera of E. & N. T. Anthony used. With the light placed at about 5 in. from the specimen, so as to make an angle of 45° with the face, and using the middle diaphragm, an exposure of 6 minutes gave the best negative. With finely-granular fractures, filtering the light through ground glass was an improvement, but the time of exposure had to be prolonged to 10 minutes; with the coarser structures, better results were obtained without the ground glass.

Figs. 9, 10 and 11 give the fractures of the Boston & Montana samples Nos. 1, 2 and 3 in one and one-half times their natural sizes. Fig. 9, cathode copper after melting and skimming, but before rabbling, has a fracture radiated and columnar, luster is absent, the color a dark red. Fig. 10 is set-copper, the fracture has lost its radial character and has become coarse- columnar to coarse-cubical, it remains dull, the color has changed to a brick-red; in the apex of the depressed surface there has appeared the characteristic single bubble. Finally, Fig. 11 represents refined copper brought to wire-bar pitch; the fracture is finely-granular and fibrous, the luster is very silky, and the color roseate.

freezing-point-curve-of-copper-cuprous-oxide-alloys

Samples Nos. 0 to 6 of the Calumet & Hecla Co., shown in Figs. 12 to 18, begin with set-copper and end with ingot-copper. The fractures, starting from coarse-columnar and cubical (Fig. 12), lose their columnar character, remaining coarse and cubical (Fig. 13), they become coarsely radiated (Fig. 14), then the

photo micrographs of copper-cuprous oxide of alloys

photo micrographs of copper-cuprous oxide of alloys-1

fracture-of-copper

sample-no.-2-set-of-copper

sample-no.-3-wire-bar

sample-no.-0-calumet-and-hecla-copper

sample-no.-1-calumet-and-hecla-copper-after-poling

sample-no.-2-calumet-and-hecla-copper-after-poling

sample-no.-3-calumet-and-hecla-copper-after-poling

sample-no.-4-calumet-and-hecla-copper-after-poling

sample-no.-5-calumet-and-hecla-copper-after-poling

sample-no.-6-calumet-and-hecla-copper-after-poling

copper-stages

sample-no.-3-wire-bar-copper

sample-hecla-set

sample-poling-copper

sample-poling-copper-2

ingot-copper-cooled-quickly

photo-micrographs-of-copper

radiation assumes finer forms and granulation puts in an appearance (Fig. 15), granulation predominates over radiation (Fig. 16), both become finer (Fig. 17), until with Fig. 18 radiation has been entirely replaced by granulation. In a similar manner the luster, from being absent with Fig. 12, becomes at first slightly silky; then silkiness increases until full silkiness is reached with Fig. 18. The dark brick-red color of Fig. 12 becomes lighter (Figs. 13, 14 and 15), rose-color begins to be seen (Figs. 15 and 16) until full rose-color is reached with Fig. 18.

The observations on the fractures are brought together in Table II. (on p. 686).

Determination of Oxygen in Samples

The oxygen of the different samples was determined by means of Hampe’s method, which consists in reducing the oxide of finely-divided copper (brought to a bright red heat) in a current of hydrogen, the loss in weight giving a measure for the oxygen-content. Hampe and later Heyn, give evidence that the reduction is complete. The apparatus used and recommended by Hampe was somewhat modified, partly along the lines suggested by Archbutt and partly by changes which suggested themselves during the work. On account of the smallness of the samples, it was necessary to use less material for the analyses than did Hampe, viz., from 10 to 13 grams. The apparatus used consisted of a 16.5-in. Kipp gas-generator (charged with hydrochloric acid and feather zinc), a gas-washing bottle filled two-thirds full with a saturated solution of caustic soda, a drying-tower with sticks of caustic soda, a U-tube filled with calcium chloride, a bulb-tube filled with copper borings and a second U-tube filled with calcium chloride.

The bulb-tube, of Bohemian glass, 3/16 in. thick, was 8 in. long and had a bulb 3 in. long and 1.25 in. in diameter. It was supported by a frame (5 in. long by 3 in. wide by 2.5 in. high) of “ uralite ” (an asbestos boarding) having slots 1.25 in.

observations-on-fractures-of-sample-buttons

deep in the ends to receive the cylindrical ends of the bulb. The frame was placed on a thin sheet of asbestos-paper resting upon a ring-stand. Under this was placed a Tirrel burner with a flame spreader. The whole was enclosed by a frame of heavy asbestos matting (7 in. long by 6 in. wide by 13 in. high) with slots in the sides to receive the protruding cylindrical ends of the bulb.

The mode of procedure was as follows: From 20 to 25 grams of borings were taken for a sample, small bits of iron were removed by a magnet, the borings were washed in a beaker four or five times with alcohol and dried to remove the last traces of alcohol, care being taken to avoid any oxidation of the copper. The borings were then divided into approximately equal parts, transferred to the weighed bulb-tubes and weighed. A bulb was placed in the furnace, connected by rubber tubing with the train of hydrogen-apparatus, hydrogen passed through for five minutes at the rate of six bubbles per second, the gas issuing from the second calcium-chloride tube ignited and the bulb slowly brought to a bright-red heat. When water ceased to appear in the glass leading to the second calcium-chloride tube, the gas-current was reduced so that only three bubbles passed the wash-bottle per second, and the bulb kept at a bright-red heat for one-and-a-half hours. At the end of this time, the supply of gas was again increased to six bubbles per second, the lamp removed and the copper allowed to cool. When cool, the bulb was disconnected, air aspirated through it, and the bulb cleaned and weighed.

The results obtained are given in Table III. It will be noticed that the average percentage of cuprous oxide of the Boston & Montana wire-bar copper is higher than that of the Calumet & Hecla cake-copper, although the former had been brought to a higher pitch and should, therefore, contain less oxygen. The discrepancy may be explained by the fact that the Boston & Montana copper contains more impurities than the Calumet & Hecla; and these impurities are present as oxides.

determinations-of-oxygen-in-sample-of-copper

Microscopical Examination of Copper Samples

The pieces of copper used for making micro-sections were sawed out as nearly as possible from the center of a fracture, as it was thought that some segregation might have taken place in the cooling. Later observations, however, showed that this precaution was unnecessary as long as the superficially oxidized surface was excluded. The samples were all finished with the polishing-machine made by the Boston Testing-Laboratories. In using the machine for rough polishing, the emery on the canvas-wheel pitted the surface to such an extent that it became necessary to file the specimen smooth before proceeding any further. In regular work, therefore, the sawed specimens were first treated with a rough, followed by a smooth, file, and then polished with rouge and water on a revolving wooden disk covered with broadcloth. In polishing, considerable difficulty was encountered at first, as the polished surfaces, when examined under the microscope, showed disturbing scratches. Polishing by hand with rouge and water on a smooth board, covered first with sheet-rubber and then with chamois, gave a more lustrous surface than when the wheel was used, but at the same time it intensified the scratches. This pointed to the probable presence of coarse particles in the rouge. In order to remove them, about one volume of rouge was stirred up with two volumes of water in a beaker, allowed to settle for about 30 seconds and the suspended matter applied with a brush to the broadcloth disk. The results were satisfactory, and all samples were treated in this manner. When polished, they were cleaned with alcohol and wiped dry with chamois. By thus applying the rouge, running of water onto the machine could be dispensed with, which made the whole operation cleaner. Experiments with decanting the suspended rouge from that which had settled, filtering and then applying the filter-contents to the disk did not work well.

In examining the polished sections with the microscope, magnifications ranging from 30 to 750 diameters were tried. As a high magnification did not bring out the structure more clearly than did one of a smaller diameter, but only narrowed the field of observation, a comparatively low magnification of 100 diameters was chosen. This gave a magnification of about 230 diameters on the photographic plate. To the eye the contrast between the black cuprous oxide in the eutectic (or the bluish-black excess-cuprous oxide in samples containing over 3.45 per cent Cu2O) with the red-colored copper was clearly visible, but the photographic plate failed to show it. In order to bring out the structure more clearly, various attempts were made to etch with nitric acid, sulphuric acid, silver nitrate and with the electric current, but they did not improve matters. Heat-tinting did some good, but not enough. Yellow and orange-colored screens were then tried; of these the orange- colored glass proved to be the better, especially when a rapid isochromatic plate particularly sensitive to orange and yellow light was used for photographing. The orange light gave the copper a yellowish tint, but had little effect upon the cuprous oxide. The copper alone having an actinic effect upon the photographic plate, it appeared white in the positive, and the cuprous oxide black. The time of exposure giving the best results was found to be two-and-a-half minutes.

The Boston & Montana sample No. 1 (Fig. 19), taken after melting and skimming the cathodes, is seen to contain a slight excess of cuprous oxide over the eutectic, although the analysis gives only 3 per cent cuprous oxide. The black crystals are small, but easily distinguished from the cuprous oxide of the eutectic. Sample No. 2 (Fig. 20), set-copper, contains a large excess of cuprous oxide over the eutectic; it shows fern-like forms which spring up in relief against the eutectic background. The fern-like forms are very unevenly distributed; the eutectic field in some places was free from them, in others it was entirely covered with them. Sample No. 3 (Fig. 21), wire-bar copper, shows an evenly distributed fine network of eutectic enclosing large meshes of copper.

In the Calumet & Hecla series, sample No. 0 (Fig. 22), set-copper, shows patches of excess-cuprous oxide in the eutectic. Sample No. 1 (Fig. 23), taken 15 minutes after poling had begun, does not differ much from sample No. 0, proving that the reduction had not proceeded very far. On the whole, both samples resemble very much the set-copper sample (No. 2) of the Boston & Montana, although they do not show the fern-like forms so clearly developed. The eutectic, in most cases, is slightly separated from the patches of the excess-cuprous oxide crystals by a narrow band of copper, and the cuprous oxide in the eutectic seems to have separated somewhat from its copper, thus giving the field a spherulitic appearance. In sample No. 2 (Fig. 24), the third taken, reduction has progressed rapidly, but it still contains a slight excess of cuprous oxide over that of the eutectic mixture. It resembles sample No. 1 of the Boston & Montana series. In sample No. 3 (Fig. 25), the eutectic has been passed, and the excess-copper becomes apparent. Sample No. 4 (Fig. 26) shows that little progress was made in the reduction in the 15 minutes that elapsed between the taking of samples No. 3 and No. 4. An explanation for this is that during this period, 45 minutes after poling had begun, the poles were withdrawn and new ones put in their places. In sample No. 5 (Fig. 27), the cuprous oxide is very much diminished, the eutectic forms a thin net-work enclosing copper in its meshes. Finally, sample No. 6 (Fig. 28), represents refined copper brought to ingot-pitch. The net-work of the eutectic is imperfect and broken, and the dark parts of the eutectic are bunched together and larger than expected. The explanation of this peculiar structure may be found, when Figs. 2 and 8 of Heyn’s photo-micrographs are compared, by the supposition that the sample was chilled when still above 1084° C., the melting- point of the eutectic, which prevented the eutectic from separating out in the form of a continuous skeleton. In order to find out whether this idea was a correct one, a new sample (A), taken at the ingot-copper stage from another furnace-charge and cooled slowly was obtained and examined. Fig. 29 brings out clearly the difference between quick and slow cooling. Fig. 30, sample (B), represents the same batch of copper when ready to be ladled into wire-bar, the sample having been also cooled slowly. The skeleton here also is seen to be continuous and not broken as in Fig. 28.

Finally, photo-micrograph Fig. 31 represents a sample of copper from the Nichols Chemical Co. which has been brought to plate-pitch, i.e., the poling has been carried further than is the case with the highest degree of wire-bar pitch of the Calumet & Hecla Co.

Area-Measurements

Measurements of areas which gave Sauveur such interesting facts seemed very promising when applied to samples of copper containing less cuprous oxide than the eutectic mixture. Measuring the copper areas and deducting them from the total area would give the eutectic area, and from this the percentage of cuprous oxide could be readily calculated. It would be useless,, if not impossible, to measure the areas of cuprous oxide in the eutectic; and on account of the unequal distribution of cuprous- oxide in specimens with more cuprous oxide than the eutectic, the data would be misleading.

For the purpose of measurement, enlargements were made of sample No. 3 (wire-bar copper) from the Boston & Montana Co., of samples Nos. 3, 4, 5 and 6 (ingot-copper, chilled) from the Calumet & Hecla Co.; also of sample A (ingot-copper, cooled slowly) and sample B (wire-bar copper, cooled slowly) from the Calumet & Hecla Co.; and of the sample of plate-copper from the Nichols Chemical Co.

The enlargements measured 16 by 20 in., giving five times the magnification of the photo-micrographs, or about 1,150 diameters. A circle, 12 in. in diameter, was drawn on an enlargement, divided into four quadrants, and measurements made on each. The Amsler planimeter was the instrument employed. It is commonly used for the measurement of indicator-cards, and is accurate to 0.1 per cent. The copper areas on an enlarged photo-micrograph were outlined with a pencil in order to facilitate measurement. All measurements were carried out in duplicate. It took from 1.5 to 2 hours to measure the copper-areas of a photo-micrograph, the time varying with the clearness with which the edges of the eutectic were defined. Thus, samples No. 5 and especially No. 6 of the Calumet & Hecla series were very difficult to measure. In calculating the percentage of cuprous oxide in the eutectic, the figure 3.45 was chosen, being the average of Heyn’s two determinations, 3.4 and 3.5.

The degree of accuracy of the measurements, carried out at least in duplicate, is shown by examples in Table IV., in which A-V represent the copper areas of one sample.

planimeter-measurement-of-copper-areas

The percentages of cuprous oxide resulting from the measurements in the several quadrants and the averages are given in Table V.

planimeter-measurements-of-quadrants

It will be seen that the measurements of the quadrants of a sample show some discrepancies, as the copper-islands are not uniformly distributed in the eutectic net-work. An excess of constituent in one quadrant is, however, balanced by a lack in another, giving on the whole a very satisfactory average.

The results obtained by fracturing and by chemical and microscopical analysis are brought together in Table VI.

The features relating to fractures have already been summarized. Comparing the cuprous-oxide content obtained by chemical analysis and by planimetric measurement, it will be seen that the percentage of cuprous oxide found by analysis in the Boston & Montana sample No. 3, and in the Calumet & Hecla samples Nos. 3 and 4, is somewhat higher than that by measurement. This may be due to the fact that the chemical analysis gives the total oxygen, that of the copper as well as that of the impurities, while measurement gives only the oxygen of the copper. That the oxygen found by analysis in the Calumet & Hecla samples Nos. 5 and 6 is lower than that obtained by measurement is probably due to the segregation of the cuprous oxide in the eutectic, causing the latter to spread somewhat. Taking the results as a whole, they show that area- measurements of enlarged photo-micrographs of pure coppers containing less oxygen than the eutectic give good valuations of the oxygen-content. Further, it seems entirely feasible to make quickly a close estimation of the percentage of cuprous oxide contained in a sample of copper by simply examining a polished surface with the microscope, when once some experience has

physical-characteristics-and-corresponding-content-of-cuprous-oxide

been gained. The mode of operating might be as follows:—To take a button-sample, cool it slowly and quench it when it had solidified, cut out a piece with a circular saw, grind it smooth on a number of revolving wooden disks covered with emery- cloth or on revolving files, polish with rouge and water on a revolving disk covered with broadcloth (a mirror-like surface would not be necessary), and estimate with the microscope the percentage of cuprous oxide present. The whole operation could be done in from 6 to 8 minutes. The poling could then be controlled by the microscope, and the degree of pitch desired for ingot-, large or small cake-, wire-bar- or electrode-copper defined by a readily ascertainable amount of cuprous oxide that should be present.

In conclusion we wish to thank Professors Richards and Fay for many valuable suggestions made during the course of the investigation.