Prof. V. Ya. Mostovich.

Oxygen in Copper-Smelting Mattes.

From the Metallurgical Laboratory of the Tomsk Technological Institute.

(With 2 figures, drawings, and diagrams).

In the quantitative analysis of copper mattes, one is usually satisfied with determining the content in them of the principal components, namely: copper, iron, and sulfur; less often lead, zinc, arsenic, and others are determined. In analysis, so-called rapid methods are used, which are not distinguished by great accuracy. Thus, for example, copper is determined by Parkes’s method, by titrating an ammoniacal solution of copper with potassium cyanide with an empirically established titre; the accuracy of this method does not exceed 0.5–1%[^1]. Iron is determined by Reinhardt’s method, sulfur is oxidized with a solution of KClO₃ in HNO₃ (1.4), and so on. If, however, for the analysis of mattes, precise analytical methods are used and all elements in the mattes are determined according to the indications of their qualitative analysis, then the sum of these elements amounts to 93–98%. Analyses in which this sum equals 100% deserve little confidence, since factory analyses, as indicated above, are not distinguished by great accuracy. The latter is not important for factory purposes, for mattes are intermediate products, and for determining their further course in production it is sufficient to know only approximately the content of copper in them. For the study of the structure of mattes and their rational analysis, it is important to establish the complete quantitative composition of mattes and to ascertain what is represented, in this composition, by the 2–7% mentioned above that are lacking to reach 100%. In the literature on copper metallurgy there are indications that mattes sometimes contain magnetic iron oxide, to which a place is assigned as an accidental impurity. Thus, Keller[^2] reports analyses of mattes containing from 1.13 to 12.6% Fe₃O₄, or from 0.3 to 2.9% oxygen. Indications of magnetic iron oxide as an impurity in mattes are found in E. Peters, E. Larisson,[^2] and other authors, to whose works we shall return.

Preliminary investigations showed that all factory mattes without exception contain oxygen in greater or lesser quantity, and all of them are ferromagnetic, whereas mattes of analogous composition prepared in the laboratory are not magnetic and contain no oxygen. This suggested that oxygen is not an accidental but an essential component in the composition of factory mattes, determining their magnetism. Therefore the question of oxygen in copper mattes seemed worthy of more detailed study, especially since in the special literature little attention is devoted to it.

The subject of the present investigation comprised the following tasks:

1. the precise determination of the amount of oxygen in factory mattes of various types of copper smelting,

2. clarification of the question of the oxygen compounds in mattes,

3. clarification of the causes determining the presence of oxygen in the composition of copper mattes.

The solution of these tasks promised, in addition, to make it possible to establish the rational composition of mattes and the dependence of the oxygen content in mattes on their origin.

Quantitative Determination of Oxygen in Copper Mattes

This determination can be carried out by several methods.

Thus, E. Keller considered that oxygen occurs in copper mattes exclusively in the form of magnetite. On the basis of the insolubility of Fe₃O₄ in HNO₃, Keller ³) used the following method for separating Fe₃O₄ from mattes. A matte sample of 10–20 g is gently heated with a large excess of HNO₃ of sp. gr. 1.42 until the separated sulfur becomes light yellow. The sulfur is then oxidized with KClO₃ while boiling. The beaker with the sample is placed on a horseshoe magnet in such a way that, when the solution is decanted, the poles of the magnet are situated perpendicular to the “spout” of the beaker. In this way the magnet retains at the bottom of the beaker every particle of Fe₃O₄. Impurities in the beaker, which may consist of Ag(?), Pb and Sb, are removed with ammonia, (NH₄)CH₃COO and tartaric acid. The insoluble matter (SiO₂) is determined from a separate sample. The residue is washed with alcohol, dried, and weighed. In qualitative analysis of residues from various mattes, Keller found no sulfur in them; under the action of HCl the residues did not evolve gases, which indicated the absence in them of metallic iron and sulfides. Quantitative analysis of the residues gave the following results:

Fe total — 70.75%; in Fe₃O₄, by theory, Fe total 72–37%

Fe″ — 23.56 » » Fe″ — 24.13%.

According to Keller, these data leave no doubt that the insoluble residues contain only Fe₃O₄.

Theoretically FeO contains 22.27% O, Fe₃O₄—27.63% O, and Fe₂O₃—30.06% O. Consequently, small differences in the content of oxygen or iron in the residues from the treatment of mattes substantially alter the composition of the iron oxide in the mattes. Therefore only direct determinations of the quantities of oxygen and of the iron associated with it can give indications both as to the composition of the iron oxides in mattes and as to the identity of this composition in all copper mattes. For the solution of these questions Keller’s method cannot be considered sufficiently accurate.

Proceeding from the assumption that oxygen is present in mattes in the form of iron oxides, I used two methods for determining oxygen, based on the reduction of these oxides to metallic iron by carbon monoxide and hydrogen.

Reduction by means of carbon monoxide at first sight had major advantages. First, the sulfides and metals entering into the composition of mattes do not change when heated in an atmosphere of CO even at the melting temperature of the mattes and, consequently, during reduction of the latter no evolution of sulfur is to be expected; therefore the entire determination could be reduced to determining the loss in weight of the matte after ignition of it in a stream of CO at a definite temperature to constant weight. This method was used by J. West ⁴) for determining oxygen in copper alloys. The second advantage of the method was the assumption that, if it were possible to determine the amount of free metallic iron in mattes before and after reduction, then by this very means it would be possible to establish the form of its oxygen compound in the mattes. In view of this, a series of determinations of oxygen in mattes by means of CO was carried out. The material for the work consisted of matte samples taken at the following plants of the Urals and Siberia: Karabash, Kalata, Bogoslovsk, Polevsk, Pyshminsko-Klyuchevsk, Ulensk, and Yulia.

At the Karabash plant the samples were taken over the course of a month, three times a week, three samples per day during tapping from the forehearth. At the Bogoslovsk plant the samples were taken over the course of three days every two...

...from the forehearth. At the other plants, samples were taken from several individual taps during normal operation of the furnaces. From the mattes of the corresponding plants, average samples were taken by successive reduction. In doing so, the layer of oxides formed during cooling of the samples was first removed from the matte cakes (spoon samples) on a carborundum wheel. The mattes were then ground to pass a sieve of 120 openings per linear inch, and the matte powders thus prepared were stored in a desiccator. A preliminary qualitative analysis of the mattes was made, and, on the basis of it, a quantitative analysis. In repeated determinations a discrepancy of not more than 0.1% was allowed. Copper was determined electrolytically after preliminary separation of it in the form of Cu₂S; iron, by Mohr’s method, by titration with a chameleon solution of ferrous-sulfate matte reduced with chemically pure zinc; sulfur, by Fresenius’ method, for which weighed portions of matte of 0.5 g were fused with a mixture of Na₂CO₃ and NaNO₃ (4:1) in a porcelain crucible in a Heraeus electric crucible furnace. Zinc was determined by titration according to Galetti’s method, and lead by titration according to Alexander’s method. The results of the analyses are given in Table 1.

Table 1. Analyses of copper-smelting mattes.

Elements	Karabash I	Karabash II	Kalatinsky	Polevskoi	Bogoslovsky	Nyaminsk-Klyuchevskoi	Ulen	Dolin
% Cu	27.60	37.10	14.55	42.94	38.57	31.98	41.05	44.90
Fe	40.15	32.58	53.40	29.49	32.89	38.31	30.66	27.67
S	24.80	25.00	23.58	24.47	24.74	22.60	24.65	24.22
Pb	0.50	0.20	—	—	—	—	—	—
Zn	1.00	0.60	—	—	0.70	—	—	—
Sum	94.05	95.48	91.53	96.90	96.90	92.89	96.36	96.79

Determination of oxygen in mattes by means of carbon monoxide.

This determination was carried out by heating a weighed portion of matte in a stream of CO in a Heraeus electric furnace with a heating tube 40 cm long and 30 mm in diameter. In the furnace there was a quartz tube 65 cm long and with an internal diameter of 20 mm; in the latter lay a Le Chatelier thermocouple in a thin-walled insulating quartz tube. Carbon monoxide was obtained by the action of concentrated H₂SO₄ on formic acid upon heating

\[ \mathrm{HCOOH + H_2SO_4 = CO + H_2O + H_2SO_4}. \]

The gas was collected in a gasometer over water and, before entering the furnace, was washed and dried by passing through a strong KOH solution and a series of drying apparatuses with H₂SO₄, CaCl₂, and P₂O₅. A weighed portion of finely ground matte, 300–500 mg, was placed as a thin layer in a porcelain unglazed boat, which was in an outer boat of larger dimensions. The boats were introduced into the front end of the tube, which was water-cooled. After the air had been displaced by carbon monoxide and the tube had reached 500–600°, the boats were introduced into the tube until they came into contact with the end of the pyrometer, by means of a wire passing through a mercury seal closing the tube. The same apparatus was used as in the reduction experiments...

lead sulfide salt[^5]). Then the temperature in the furnace was raised to the required limit. By preliminary experiments, for each matte a certain maximum reduction temperature was established, which had to lie below the melting temperature of the mattes and was different for different mattes. The sample was held at a constant temperature for two hours, after which the boat was moved to the cold end of the tube, where it was rapidly cooled to room temperature in a strong stream of CO. Then the CO was displaced from the furnace by dry and pure carbon dioxide, the mercury seal was removed, the boat was placed in a desiccator and weighed. In order to establish whether, under the experimental conditions, sulfur was evolved from the mattes, the gases leaving the furnace were passed through a control tube with a solution of BaCl₂, acidified with hydrobromic acid. In none of the experiments was sulfur evolution observed. As a check, sulfur was determined in some mattes after reduction. Reduced to the initial weight of the unreduced mattes, these determinations gave the results presented in Table 2.

Table 2. Sulfur content in mattes before and after reduction by means of CO.

Mattes	Karabash I	Karabash III	Kalata	Polevskoy	Pyshma-Klyuch.
% sulfur content before reduction	24,80	25,00	23,58	24,47	22,60
% sulfur content after reduction	24,87	25,10	23,80	24,50	22,55

Consequently, no evolution of sulfur took place, and the losses in weight of the mattes after reduction represented oxygen and moisture contained in the mattes. To determine moisture, samples of the mattes were heated in a stream of dry and pure nitrogen at 150°. The moisture content proved to be 0,05—0,06%, a value which was not taken into account in calculating the oxygen in the mattes. With the system previously filled with CO, the consumption of the latter was about 4 liters per determination. It was necessary to ensure a sufficiently rapid stream of gas in order to prevent noticeable decomposition of CO according to the reaction 2CO ⟶ CO₂ + C in parts of the furnace where favorable conditions existed for this. Accumulation of CO₂ in the gaseous atmosphere could have led to a CO : CO₂ ratio at which the iron oxides in the mattes would not have been reduced. In addition, experiments on heating iron and copper sulfides in a stream of CO₂ showed that the latter, already at temperatures of 500—600°, oxidizes these sulfides, and the gas leaving the furnace contained SO₂.

As for the reduction temperature, its limit was the sintering temperature of the mattes, since reduction could be carried out completely only when the mattes were in the powdery state. As soon as the matte powder sintered, the contact surface between CO and the matte particles decreased and the reduction process slowed down or ceased altogether, which led only to partial reduction of the matte. On average, the reduction temperature ranged within 820—830°, which was not particularly favorable, since the reduction of iron oxides proceeds very slowly at this temperature. The results of the determination of oxygen in mattes by means of CO are given in Table 3.

Table 3. Oxygen content in mattes.

Sample No.	Matte of plant	Charge, mg	Duration of heating, h	t, °C	Loss in weight, mg	Oxygen content, %	Average oxygen content, %
1	Kalyatinsky	407,80	2	820	25,20	6,20	—
1a	Kalyatinsky	repeat	1	830	0	0	—
2	Kalyatinsky	366,40	1.50	850	17,60	4,80	Sample fused.
3	Kalyatinsky	271,20	2	830	17,10	6,31	—
4	Kalyatinsky	209,50	2	820	13,30	6,34	—
4a	Kalyatinsky	repeat	3/4	820	0	0	Average 3—8
5	Kalyatinsky	282,80	1 3/4	815	17,30	6,12	6,21
5a	Kalyatinsky	repeat	1	830	0	0	—
6	Kalyatinsky	281,80	2	820	17,20	6,11	—
7	Kalyatinsky	402,20	2	820	24,80	6,17	—
8	Kalyatinsky	426,40	1 3/4	815	26,40	6,19	—
9	Karabashsky I	400	2.5	770	18,20	4,55	4,67
10	Karabashsky I	400	2.5	805	19,10	4,78	—
11	Karabashsky III	400	2.5	800	13,30	3,33	—
12	Karabashsky III	400	2.5	810	13,00	3,28	3,29
13	Karabashsky III	400	2.5	800	13,10	3,25	—
14	Pyshm.-Klyuchevsk.	362,20	1.5	810	18,40	5,08	—
15	Pyshm.-Klyuchevsk.	355,60	2	810	17,80	5,01	5,04
16	Pyshm.-Klyuchevsk.	385,60	2	810	19,40	5,03	—
17	Polevskoy	319,40	2	810	9,0	2,82	—
17a	Polevskoy	repeat	1	830	0	0	—
18	Polevskoy	293,0	2	830	8,30	2,84	2,84
19	Polevskoy	319,0	2	820	9,20	2,88	—
20	Polevskoy	361,60	2	810	10,0	2,77	—
21	Polevskoy	404,90	2	820	11,70	2,88	—
22	Bogoslovsky	422,10	2.5	825	8,90	2,11	—
22a	Bogoslovsky	repeat	1	825	0	0	2,07
23	Bogoslovsky	382,30	2.5	825	7,80	2,04	—
24	Uleab	400	2.5	800	7,50	1,88	—
25	Uleab	400	same	same	7,30	1,80	—
26	Uleab	460,10	same	825	7,90	1,72	1,78
27	Uleab	402,80	same	same	6,90	1,71	—
28	Yuliya	393,40	2.5	825	3,30	0,84	—
29	Yuliya	278,60	same	same	2,20	0,79	0,82
30	Yuliya	390,40	same	same	3,20	0,82	—

As can be seen from the table, the determination of oxygen by means of CO gives, at first glance, satisfactory results in the sense of agreement. After these experiments, the determination of the amount of metallic iron in mattes before and after their reduction was undertaken. This was intended, first, to prove that the oxygen in the mattes is in fact bound to iron, and second, to clarify the form of the oxygen compound of iron.

In choosing a method for determining metallic iron, I settled on the so-called Swedish test, used in assaying practice for determining copper in ores. The method consists in precipitating copper from a sulfuric- or hydrochloric-acid solution of the latter by means of metallic iron, and with pure solutions of copper salts the method gives accurate results. In this case I proceeded from the assumption that free iron, not chemically bound to other elements in the mattes, would precipitate from a neutral solution of $\mathrm{CuSO_4}$ an equivalent amount of copper according to the reaction $\mathrm{CuSO_4 + Fe = Cu + FeSO_4}$. From the amount of copper precipitated, it would then be possible to determine the amount of metallic iron that had reacted. It was necessary to work with a neutral solution of $\mathrm{CuSO_4}$, since even dilute acids, such as HCl and $\mathrm{H_2SO_4}$, decompose mattes and, moreover, to avoid oxidation of the formed $\mathrm{FeSO_4}$ to $\mathrm{Fe_2(SO_4)_3}$; the latter, in neutral solutions, hydrolyzes and is capable of dissolving appreciable amounts of Cu, $\mathrm{Cu_2S}$, and CuS, which was undesirable, since the amount of precipitated Cu was determined from the difference between its content in the $\mathrm{CuSO_4}$ solution before and after reaction with the matte.

The $\mathrm{CuSO_4}$ solution was prepared from $\mathrm{CuSO_4 \cdot 5H_2O}$ salt of the “Kahlbaum p. Garantieschein” grade and contained 7.857 g of salt in 500 cu. cm of solution, corresponding to 2 g of copper. A 1 g sample of matte was placed in 50 cu. cm of $\mathrm{CuSO_4}$ solution contained in a flask through which nitrogen was passed for half an hour. The flask was then filled with nitrogen, tightly stoppered, and shaken from time to time for 20–24 hours. After this the solution was filtered through a dense filter, and the amount of remaining copper was determined in it. From the decrease in the Cu content in the solution, the equivalent amount of dissolved iron was calculated. The experiments showed that the mattes did indeed precipitate copper from the $\mathrm{CuSO_4}$ solution, since the Cu content in the solution decreased, and in the filtered precipitates of the mattes, washed with alcohol and ether, plates of freshly precipitated copper were clearly visible, the amounts of which differed for different mattes. But after the very first experiments it turned out that the solutions that had stood in an atmosphere of nitrogen contained ferrous and ferric iron, which indicated the occurrence of two reactions:

$$ 1.\ \mathrm{CuSO_4 + Fe = FeSO_4 + Cu} $$

$$ 2.\ \mathrm{3CuSO_4 + 2Fe = Fe_2(SO_4)_3 + 3Cu.} $$

Since it was impossible to take into account how much metallic iron participates in each of the above reactions, the method for determining the amount of free Fe that had reacted with $\mathrm{CuSO_4}$ was modified in the sense that, in the solutions after treatment of the mattes, the iron that had passed into solution was determined directly. For this, the solution separated from the matte precipitate was acidified with $\mathrm{H_2SO_4}$, reduced with zinc, and titrated with $\mathrm{KMnO_4}$. For each matte, three determinations were carried out. The results are given in Table 4.

By the method described above, metallic Fe in the mattes was determined after their reduction in CO. For this, the entire reduced sample, or part of it, was treated with $\mathrm{CuSO_4}$ in an atmosphere of nitrogen, and the iron was determined in the solutions.

The results of these determinations are given in Table 5.

Table 4. Metallic iron in mattes.

Mattes	\multicolumn? On titration of iron, cu. cm. KMnO₄: 1	2	3	Average cu. cm. KMnO₄	Content of metallic iron in mattes, in 1 g mg	Content of metallic iron in mattes, %
Karabash I	1.25	1.20	1.30	1.25	6.86	0.69
Karabash III	1.58	1.55	1.58	1.57	8.62	0.86
Kalata	1.00	1.00	1.00	1.00	5.49	0.55
Polevskoi	2.05	2.00	2.10	2.05	11.25	1.13
Bogoslovsky	4.7	4.6	4.7	4.67	25.64	2.56
Pyshm.-Klyuchevskoi	0.50	0.50	0.50	0.50	2.74	0.27
Ulen	6.75	6.70	6.80	6.75	37.06	3.71
Yuliya	9.70	10.00	9.50	9.70	53.25	5.33

Table 5. Metallic iron in mattes after reduction.

From samples of Table 3	Matte charge, mg	Oxygen in 1 g matte, mg	In CuSO₄ solution passed to Fe, mg	Content of metallic Fe in 1 g matte, % after reduction	Content of metallic Fe in 1 g matte, % before reduction
Kalata No. 7	358.50 938.00	62	74.34 194.50	19.45	0.55
Kalata Nos. 5, 6, 7	415.50 938.00	62	80.99 182.90	18.30 18.88	0.55
Polevsk. No. 20	348.00 971.60	28.40	26.45 73.86	7.39	1.13
Polevsk. No. 21	390.50 971.60	28.40	30.03 74.72	7.17 7.34	1.13
Polevsk. No. 18	262.70 971.60	28.40	19.67 72.75	7.28	1.13
Pyshmin. No. 14	302.50 949.60	50.40	48.21 151.30	15.13	0.27
Pyshmin. No. 15	319 949.60	50.40	52.90 157.50	15.75 15.44	0.27

These results showed that the amount of metallic Fe in all mattes increased after reduction, but not in proportion to the oxygen content in the same mattes, as is seen from comparison of the data for the Kalata, Pyshminsky, and Polevskoi mattes:

the figures for the increase in metallic Fe are as
18.33 : 15.17 : 6.21 = 3 : 2.44 : 1

“ for oxygen . . . “ “
6.21 : 5.04 : 2.84 = 2.2 : 1.77 : 1.

Direct proportionality between these figures would have occurred if the oxygen compounds of iron in the investigated mattes were identical. If, in accordance with the statements in the literature, it is assumed that all the oxygen in the mattes is bound with Fe in the form of $\mathrm{Fe_3O_4}$, then the quantities of free iron found in the mattes should have corresponded to the following quantities of oxygen:

for the Kalata matte, 183.3 mg was found, corresponding to 70.02 O.
» the Polevskoi » » 62.1 » » » 23.71 »
» the Pyshminskii » » 151.7 » » » 57.92 »

Reduction by CO gave, for oxygen:

for the Kalata matte . . . . . . . . . . . . 62 mg.
» the Polevskoi » . . . . . . . . . . . . 28.4 »
» the Pyshminskii » . . . . . . . . . . . . 50.4 »

These figures show that some mattes contain iron oxides richer, and others poorer, in iron than magnetite; or else the indicated discrepancies between the quantities of O and Fe are caused by the following circumstances:

1. At the temperature of reduction of mattes in CO, i.e., at 820–830°, processes occur which in some mattes cause the liberation of metallic Fe from its compounds, while in others they cause the transition of part of the reduced iron into a bound state.

2. A solution of $\mathrm{CuSO_4}$ cannot serve as a reagent for the quantitative determination of free iron in its alloys with the sulfides of Fe and Cu that form part of mattes.

3. In some cases the reduction of mattes by means of CO does not proceed to completion, and the method is not suitable for the quantitative determination of oxygen in mattes.

All these suppositions required verification, and such verification was carried out.

The first supposition was clarified by examining the mattes under the microscope before and after reduction. The appearance of the mattes changed greatly after reduction. Thus, in the Kalata and Karabash mattes, before reduction, inclusions of metallic copper were visible in the form of globules, dendrites, etc. They were especially abundant in the Kalata matte, the fresh fracture of which was entirely studded with inclusions of copper, giving the matte a characteristic dark-bronze color. With increasing copper content in the mattes, the number of inclusions decreased, and in the mattes of the Polevskoi, Ulen’, and Yuliya works the copper was almost imperceptible. After reduction, however, the color of the Kalata and Karabash mattes became steel-gray, and the copper inclusions disappeared. Conversely, in rich mattes (Ulen’ and Yuliya) the amount of metallic Cu increased noticeably. All this shows that, when heated in a stream of CO, reactions occur in the mattes which, in poor mattes, lead to the displacement of iron from its sulfide by copper, and, in rich mattes, to the displacement of Cu from its sulfide by iron released from combination with oxygen. These reactions are known and are expressed in the form of the reversible process:

\[ \mathrm{Fe + Cu_2S \rightleftarrows FeS + 2Cu,} \]

studied by Baikov, Yushkevich, and Bornemann.

As Yushkevich showed$^{6)}$, the indicated process in a neutral medium begins already at 400–450° and proceeds according to the schemes:

\[ \text{1. } 4\mathrm{Cu} + 3\mathrm{FeS} = (\mathrm{Cu_2S})_2\,\mathrm{FeS} + 2\mathrm{Fe}. \]

\[ \text{2. } 3\mathrm{Cu_2S} + \mathrm{Fe} = (\mathrm{Cu_2S})_2\,\mathrm{FeS} + 2\mathrm{Cu}. \]

In poor mattes, such as the Kalata and Karabash mattes, in which there are considerable inclusions of metallic Cu, the process proceeds according to scheme 1; in rich mattes, according to scheme 2, with the iron participating in the reaction,

is obtained upon reduction of its oxides by means of CO. These processes among the components of mattes may explain both the changes in the external appearance of the mattes and the discrepancies between the amounts of O and metallic Fe extracted by cuprous sulfate.

1. The question of the suitability of a neutral solution of CuSO₄ for the quantitative determination of metallic Fe in its alloys with sulfides was specially investigated.

Since copper sulfides begin to react with metallic Fe at a temperature lying considerably below the melting temperature of mattes, and Fe binds part of the sulfur from the copper sulfides, it is evident that iron may be present in mattes either only in the free state, or in alloy with FeS. Therefore, the action of a CuSO₄ solution was first determined on iron sulfides intermediate in composition between FeS₂ and FeS, and then on FeS:Fe alloys.

For the experiments Karabash pyrite was taken, previously ground to pass a 60-mesh I.M.M. sieve and washed with hydrochloric acid and water. Composition of the pyrite: Fe—46.35%, S—51.58%, SiO₂—1.1% and Cu—0.2%. During grinding of the pyrite in an agate mortar for subsequent experiments, the odor of SO₂ was clearly perceptible, which indicated oxidation of the pyrite and the possibility that some quantity of oxidized iron might pass into aqueous solution. Thus, E. Allen and J. Johnston⁷) state that pyrite, on fine grinding, is partially oxidized and that the percentage of oxidized pyrite increases with increasing duration and degree of grinding. Therefore, for an accurate determination of sulfur in pyrite, the latter should be ground only coarsely, approximately to a 20-mesh sieve. According to the investigation of the same authors, oxidation of pyrite proceeds according to the reaction: FeS₂ + 3O₂ = FeSO₄ + SO₂.

Samples of 1 g of finely and coarsely ground pyrite were heated on a water bath successively with water, a CuSO₄ solution, and diluted HCl (1:5). In the filtrates from the tests with water and CuSO₄, Fe was determined, and in the hydrochloric-acid solution, Fe and S. The following results were obtained:

Solvent.	Finely ground pyrite.	Coarsely ground pyrite.
water.	4.4 mg.	none.
CuSO₄	4.1 »	none.
HCl	4.2 »	none.
HCl	0.6 »

1 g of pyrite was ground in an agate mortar for 5 minutes, and the powder was extracted with cold CuSO₄ solution for 24 hours in an atmosphere of nitrogen. 2.3 mg Fe passed into solution; when 1 g of pyrite, but unground, was treated under the same conditions, no iron at all was found in the solution. Consequently, CuSO₄ solution does not act on pure pyrite if the latter has not been ground too finely and, as a result, has not oxidized. To determine whether CuSO₄ solution acts on iron sulfides intermediate in composition between FeS₂ and FeS, the following experiments were carried out. Samples of 1 g of coarsely ground pyrite were heated in a stream of pure and dry nitrogen in an electric furnace at temperatures from 600 to 1000° for one hour. The sulfur content in the dissociation products, determined by analysis of the latter, agreed almost exactly with that which could be calculated from the loss in weight of the pyrite upon heating. The results of the experiments are given in Table 6.

Table 6. Effect of CuSO₄ on the dissociation products of pyrite.

Pyrite charge, mg	Content S, mg	t° C.	Loss in weight, mg a	Sulfur in matte, mg b	a + b, mg	Iron passed into CuSO₄ solution, mg
1000	515,80	600	35,60	479,00	514,60	0
1000	515,80	800	238,60	274,00	512,60	0
1000	515,80	1000	256,60	256,70	513,30	0

Upon secondary heating of sample No. 3 at 1000°, no further loss occurred. Thus, the final product of the dissociation of pyrite at 1000° is FeS, with the pyrite losing exactly 1/2 of the sulfur contained in it. Each of the dissociation products of pyrite was treated with a CuSO₄ solution for 20—24 hours in an atmosphere of nitrogen. No iron was found in the solutions. Consequently, pyrite and the products of its dissociation up to and including FeS do not react with CuSO₄.

To obtain iron sulfides richer in iron than FeS, i.e., systems FeS:Fe, two methods may be used.

The first method consists in fusing FeS with various amounts of metallic FeS; the second is based on the following. If a mixture of pyrite and metallic iron is heated in a neutral atmosphere until complete dissociation of the pyrite, then the dissociating sulfur binds an equivalent amount of iron. By introducing various amounts of iron into the mixture, products with any content of excess iron relative to the compound FeS can be obtained.

Preparation of FeS:Fe alloys by the first method. Metallic Fe of Kahlbaum grade was ignited in a stream of pure and dry hydrogen to constant weight. The product contained 100% Fe. Ferrous sulfide was prepared as follows.

700 g of sorted Karabash pyrite were ground to pass 60 openings and washed in a stream of water in three concentrically arranged glass cylinders nested one within another; in this way three products were obtained. In the inner cylinder I—pure coarse-grained pyrite in the amount of 330 g; in cylinder II—an intermediate product with inclusions of quartz; in cylinder III—fine material or material with a large amount of quartz.

100 g of product I were separated by means of Klein’s solution (cadmium borotungstate with sp. gr. 3,28) to remove the remaining quartz, washed with water, boiled again with 20% hydrochloric acid until the reaction for iron disappeared, then thoroughly washed with water, dried on a porous clay plate, and stored in a desiccator. This pure preparation had the composition: Fe—46,46%, S—53,20%, and served for experiments on dissociation, preparation of mattes, etc. 125 g of pyrite I were fused in a graphite crucible under a layer of calcined charcoal powder in a furnace at a temperature of about 1250°. 92,1 g of FeS were obtained. If it is assumed that pyrite I contained 53,2% S, then the theoretical yield of FeS should have been 91,62 g. Analysis of the FeS obtained: Fe—63,00%, S—36,86%; theoretically Fe—63,52% and S—36,48%. The ground FeS was fused with metallic iron powder under a layer of coal in a carbon crucible in a kryptol furnace to obtain alloys with 1, 3, 5, 7, 10, 15, 20, and 25% metallic Fe. All the alloys prepared proved to be very brittle and in air gave off a faint odor of H₂S, which indicated their slow decomposition under the action of CO₂ and moisture of the air.

The pulverized alloys were analyzed: Fe was determined by Mohr’s method, and sulfur by the following modified Schulte method. 0.3–0.4 g of alloy were dissolved, while boiling, in a Corliss flask in HCl (1:5) in a stream of hydrogen; boiling was continued until the hydrogen sulfide in the exit gases had completely disappeared, which took about 3 hours. The hydrogen sulfide was absorbed by a KOH solution in two absorption bottles. Usually all the H₂S was absorbed in the first bottle; its contents were poured into a flask with 100 cu. cm of iodine solution, the titer of which had been established by K₂Cr₂O₇. To the mixture were added 10 cu. cm of conc. HCl, and the excess iodine was titrated with thiosulfate. The analyses of the alloys are given in Table 7, where the alloy number indicates the percentage content of free metallic Fe introduced by calculation.

Table 7. Analyses of alloys of iron with ferrous sulfide.

Alloy No.	Fe, %	S, %	Fe + S, %	FeS by sulfur	Fe
1	63,96	35,82	99,78	98,20	1,58
3	64,33	35,54	99,87	97,44	2,43
5	64,82	34,81	99,63	95,43	4,20
7	65,49	34,35	99,84	94,17	5,67
10	66,49	33,13	99,62	90,83	8,79
15	68,08	31,44	99,52	86,19	13,33
20	69,81	29,48	99,29	80,82	18,47
25	71,48	27,82	99,30	76,27	23,03

Samples of 500 mg of each alloy were treated with a CuSO₄ solution in an atmosphere of nitrogen for 24 hours, and the metallic Fe that had passed into solution was determined in the filtrates.

The results of the experiments are given in Table 8, the amount of FeS being calculated from sulfur (S × 2.742 = FeS), while the remainder of Fe is taken as free iron.

Table 8. Action of a CuSO₄ solution on FeS:Fe alloys.

Alloy No.	Fe + S, %, by analysis	Metallic Fe, by sulfur a	Metallic Fe, by calculation Fe + S = 100 b	Fe extracted by CuSO₄, %	Difference, %, for a	Difference, %, for b
1	99,78	1,58	1,80	1,84	+0,25	+0,04
3	99,87	2,43	2,56	2,33	−0,10	−0,23
5	99,63	4,20	4,57	3,86	−0,34	−0,71
7	99,84	5,67	5,83	7,33	+1,66	+1,50
10	99,62	8,79	9,17	13,66	+4,87	+4,49
15	99,52	13,33	13,81	19,31	+5,98	+5,50
20	99,29	18,47	19,18	26,24	+7,77	+7,06
25	99,30	23,03	23,73	29,00	+5,98	+5,28

From the data of the table it is evident that, from alloys containing more than four percent metallic iron, the $\mathrm{CuSO_4}$ solution extracts considerably more iron than was introduced into the alloy with $\mathrm{FeS}$. In determining sulfur in the alloys, the separation of amorphous carbon in the form of an insoluble residue was observed. This indicated that, when $\mathrm{FeS}$ and $\mathrm{Fe}$ were melted in carbon crucibles under a layer of charcoal, the alloys dissolved carbon, and consequently we were dealing not with binary alloys $\mathrm{Fe:S}$, but with ternary alloys $\mathrm{Fe:S:C}$. In some alloys, namely Nos. 5, 10, 15, 20, and 25, the carbon content was determined by the Sarnström method$^{3}$. Portions of the alloys of 2–2.5 g were dissolved in a mixture of chromic and sulfuric acids. Dissolution began very vigorously and was accompanied by abundant evolution of $\mathrm{H_2S}$, which was absorbed by a $\mathrm{KMnO_4}$ solution acidified with $\mathrm{H_2SO_4}$; then the gaseous reaction products were dried with concentrated $\mathrm{H_2SO_4}$, passed through heated copper oxide, then through concentrated $\mathrm{H_2SO_4}$, and absorbed in two U-shaped tubes with soda lime. One determination required $2 \frac{1}{2}$–3 hours. Determination of carbon in the alloys gave the following results:

Alloy No.	% C
5	0.05
10	0.094
15	0.033
20	0.086
25	0.243

Comparing the carbon content in the alloys with the diagram of their states given by Prof. Friedrich$^{4}$—see Fig. 1—we see that the solubility of carbon reaches a minimum at the eutectic alloy and increases in alloys lying on both sides of the eutectic, which have a higher melting temperature; thus, the solubility of carbon is the greater, the higher the temperature at the beginning of melting of the alloy.

Pure $\mathrm{FeS}$, obtained by the dissociation of pyrite, does not decompose in air and does not act on a $\mathrm{CuSO_4}$ solution; $\mathrm{FeS}$ remelted in a carbon crucible decomposes in air and acts on a $\mathrm{CuSO_4}$ solution.

Thus, when 500 mg of $\mathrm{FeS}$ was treated with a $\mathrm{CuSO_4}$ solution, 1.01 mg of $\mathrm{Fe}$ passed into solution, which amounts to 0.3% of the $\mathrm{Fe}$ content in $\mathrm{FeS}$. Consequently, the cause of such a considerable discrepancy between the quantities of iron introduced into the alloys and extracted by $\mathrm{CuSO_4}$ is the presence in the investigated alloys of the system $\mathrm{Fe:C:S}$, which, evidently, dissolves readily in $\mathrm{CuSO_4}$ and makes the alloys rather unstable. In view of this, alloys of $\mathrm{FeS}$ with $\mathrm{Fe}$ were prepared under conditions excluding their absorption of carbon.

Carbon-free alloys of $\mathrm{FeS}$ with $\mathrm{Fe}$ were prepared from chemically pure iron and ferrous sulfide obtained by heating purified Karabash pyrite in a porcelain crucible in nitrogen in an electric furnace at $970^\circ$. Metallic iron of the grade “Kahlbaum, reduziert,” free from carbon and containing 95.22% $\mathrm{Fe}$, and therefore somewhat oxidized, was heated in a stream of hydrogen at $950^\circ$ to constant weight. The preparation was found to contain 4.788% oxygen, and after reduction, 100.1% $\mathrm{Fe}$. The ferrous sulfide had the composition: $\mathrm{Fe}$—61.56% and $\mathrm{S}$—38.44%, i.e., it contained 2% excess sulfur as against theory (according to theory, $\mathrm{FeS}$ contains 36.48% $\mathrm{S}$).

The ferrous sulfide was fused with calculated quantities of metallic iron in order to obtain $\mathrm{FeS}$ and alloys of $\mathrm{FeS}$ with 5 and 10% metallic iron. The fusion was carried out in porcelain tubes in Friedrich’s kryptol furnace in an atmosphere of nitrogen, the stream of which was introduced into the crucible through an opening in the lid until the alloys had cooled completely. Analysis of the preparations obtained gave the following results:

No. 1 . . . FeS: \(\left\{\begin{array}{l}\mathrm{Fe}—63.67\%\\ \mathrm{S}—36.33\%\end{array}\right.\) according to theory \(\left\{\begin{array}{l}\mathrm{Fe}—63.52\%\\ \mathrm{S}—36.48\%\end{array}\right.\)

No. 2 . . . . . . . . . 95.35% FeS + 4.65% Fe

No. 3 . . . . . . . . . 90.50% » + 9.50% »

Portions of 1 g of the finely ground preparations were treated with a solution of copper sulfate for 24 hours, as described above.

In the filtered solutions there were found: from No. 1—1.67 mg Fe, from No. 2—51.38 mg Fe, and from No. 3—124.0 mg Fe. Under the condition of complete extraction of the metallic iron into the solutions, there should have passed from No. 1—0 mg Fe, from No. 2—46.5 mg Fe, and from No. 3—95 mg Fe. Consequently, also in the case of alloys FeS:Fe not containing carbon, copper sulfate extracts from them more metallic iron than is contained in the alloy. This can be explained only by the fact that the ferric oxide sulfate formed in solution by the reaction:

\[ 3\mathrm{CuSO}_{4} + 2\mathrm{Fe} = \mathrm{Fe}_{2}(\mathrm{SO}_{4})_{3} + 3\mathrm{Cu} \]

acts upon FeS, converting it into sulfate according to the reaction: \(\mathrm{FeS} + \mathrm{Fe}_{2}(\mathrm{SO}_{4})_{3} = 3\mathrm{FeSO}_{4} + \mathrm{S}\), which is analogous to the action of the same reagent on pyrite and marcasite*).

The investigation carried out shows that copper sulfate cannot serve as a reagent for the precise quantitative determination of metallic iron in its alloys with ferrous sulfide.

Determination of Oxygen in Mattes by the Hampe Method.

After the experiments on the determination of oxygen in mattes with the aid of CO, which did not give satisfactory results, the Hampe method was applied for the same purpose, consisting in heating mattes in a stream of hydrogen. Hampe\(^{10}\) determined oxygen in lead mattes in this way. He heated finely ground matte in a spherical tube of refractory glass. Since, under the action of hydrogen on sulfides, the latter decompose with liberation of sulfur, Hampe passed the gaseous reduction products issuing from the tube through an alkaline solution of litharge, which absorbed \(\mathrm{H}_{2}\mathrm{S}\). The loss in weight of the matte consisted of oxygen and volatilized sulfur. The latter was determined from the weight of the precipitated PbS and subtracted from the total loss in weight of the matte. The difference then constituted the oxygen in the matte under investigation.

For the determination of O by Hampe I used hydrogen which, before entering the glass tube where the reduction was carried out, passed successively through wash bottles with a solution of litharge in KOH for absorption of \(\mathrm{H}_{2}\mathrm{S}\), a solution of \(\mathrm{AgNO}_{3}\) for absorption of arsine and stibine, two wash bottles with conc. \(\mathrm{H}_{2}\mathrm{SO}_{4}\), a tower with \(\mathrm{CaCl}_{2}\), a tube with metallic copper heated in a gas furnace, and then through a series of drying towers with \(\mathrm{CaCl}_{2}\), \(\mathrm{H}_{2}\mathrm{SO}_{4}\), and \(\mathrm{P}_{2}\mathrm{O}_{5}\). Pure and dry hydrogen entered a tube of refractory Jena glass, which was heated in a gas furnace. The temperature of heating was measured with a Le-Chatelier pyrometer. A porcelain boat with a charge of matte was introduced into the cold tube until it touched the end of the pyrometer; the free end of the tube was closed with a rubber stopper with a tube, from which the gases were led through two bottles with a solution of cadmium acetate, acidified with acetic acid. The tube was filled with hydrogen, then slowly heated with 4–6 gas burners; in this way it was possible to maintain \(t\)-ru in the tube,

*) To distinguish pyrite from marcasite the method of oxidation of H. N. Stokes is used. The method is based on the different behavior of both minerals toward a solution of ferric sulfate. Sulfur, present in excess, reduces ferric sulfate at the expense of ferrous sulfate; the mineral is oxidized to \(\mathrm{FeSO}_{4}\), \(\mathrm{H}_{2}\mathrm{SO}_{4}\), and \(\mathrm{S}\): \(\mathrm{FeS}_{2} + \mathrm{Fe}_{2}(\mathrm{SO}_{4})_{3} = 3\mathrm{FeSO}_{4} + 2\mathrm{S}\). See Amer. Journ. of Science. XXXIII, 1912, 175.

to bring the maximum up to 730°. With prolonged heating the tube bent and, on cooling, it burst. Since a temperature of 730° was insufficient for reduction and the work involved great inconveniences, the gas furnace was replaced by an electric one with a quartz heating tube. This, understandably, precluded the possibility of using a mercury seal and nickel wire for introducing the sample into the furnace, which considerably lengthened the procedure, since the sample had to be introduced into a cold furnace before contact with the pyrometer, the air had to be displaced from the heating tube and, after the time required for reduction had elapsed, the sample had to be cooled in the furnace in an atmosphere of hydrogen to 40–50°, when it was possible to remove the sample from the furnace without danger of oxidation of the reduced matte. Therefore, carrying out one determination required no less than 8 hours. The cold end of the quartz tube was heated during the experiments with a gas burner to prevent condensation in this part of the tube of water vapors and their absorption of \(H_2S\). A precipitate of CdS formed only in the first absorption flask; in the second, the solution remained completely unchanged. At the end of the experiment, 5 cu. cm. of acidified \(H_2SO_4\) solution of \(CuSO_4\) was added to the contents of the first flask, the mixture was heated, the precipitated \(CuS\) was rapidly filtered off, washed with hot water, and, together with the filter, was ignited and calcined in a porcelain crucible. From the weight of \(CuO\), the amount of sulfur that had split off from the matte was calculated (\(CuO \times 0.4041 = S\)); the difference—the total loss in weight of the matte minus the sulfur—gave the oxygen content in the taken charge of matte. The duration of reduction at constant temperature was 2 hours. The time required for complete reduction was determined for each matte by reheating the already reduced sample in hydrogen for one hour at a temperature 20–30° above the initial one. In this case, usually, the new loss in weight of the sample corresponded exactly to the amount of sulfur evolved from the matte.

Preliminarily, under the same conditions, several experiments were carried out on the reduction of Kahlbaum’s magnetite, containing, by calculation, 26.5% oxygen. The results obtained are placed in Table 9.

Table 9. Reduction of magnetic iron oxide by hydrogen.

Charge of magnetic iron oxide, mg	Duration of reduction, hours	°C	Loss in weight, mg	Oxygen content, %
273.4	2	850	70.20	25.70
251.0	»	»	63.0	25.10
repeated	1	900	0	—

Since, upon repeated reduction, no new losses in weight were observed, the deviation in the O content from the calculated value can be explained only by insufficient homogeneity of the natural magnetite used in these experiments.

All results of the determination of oxygen in mattes by Hampe’s method are given in Table 10.

Table 10. Determination of oxygen in mattes by Hempel’s method.

Mattes	No.	Matte sample, mg	Reduction, °C	Reduction, hours	Loss of S + O, mg	Sulfur evolved, mg	Oxygen content, mg	Oxygen content, %	Oxygen content, % average
Kalata	1	305.80	830	2	30	7.11	22.89	7.48	—
"	2	325.80	"	"	31.80	8.24	23.56	7.23	7.47
"	2a	repeated	"	1	6.50	5.41	1.09	7.50	—
"	3	300	840	2	31.40	9.13	22.27	7.42	—
Pyshma	4	300.60	830	2	29.0	10.50	18.50	6.15	—
"	4a	repeated	"	1	4.80	4.50	0.30	6.27	6.20
"	5	333.60	"	2	30.60	10.18	20.42	6.12	—
Polevskoi	6	315.20	830	2	20.80	10.91	9.89	3.14	—
"	7	387	"	"	22.20	10.02	12.18	3.15	3.15
"	7a	repeated	"	1.5	6.20	6.10	0.10	—	—
Karabash III	8	301.40	830	2.5	26.40	11.23	15.17	5.03	5.06
"	9	379	840	2	33.00	13.74	19.26	5.08	—
Karabash I	10	321.80	840	2.5	31.20	10.02	21.18	6.58	6.52
"	11	324.20	"	2	31.60	10.71	20.89	6.45	—
Bogoslovsk	12	357	840	2	18.60	8.48	10.12	2.84	2.87
"	13	301.50	"	"	18.80	10.02	8.78	2.91	—
Uleeb	14	360.60	840	2	19.0	10.34	8.64	2.88	2.84
"	15	349	"	"	21.40	11.63	9.77	2.80	—
Yulia	16	367.20	830	2	15.80	7.88	7.92	2.16	2.13
"	17	405.40	"	"	16.80	8.32	8.48	2.09	—

In addition to those investigated, two samples of mattes from shaft-furnace smelting at the Pyshma plant were analyzed. The samples were taken from the stream of matte before it entered the forehearth. Further, one more sample of matte from the forehearth was investigated before the matte entered the converter. All samples were taken on the same day, October 7, 1916: one at 10 a.m., the second at 10 p.m., and the third at 6 p.m. The results of the analysis of these mattes are given below:

	Cu, %	Fe	S	O	Total
Matte, 10 a.m.	30.0	40.0	23.64	5.27	98.91
" 10 p.m.	38.15	33.1	23.68	4.02	98.95
converter matte (6 p.m.)	31.25	38.13	22.93	6.04	98.35

The oxygen content in mattes varies, which may be explained by changes in the composition of the charge, the amount of air blast, etc. In the forehearth, the composition of the matte becomes evened out, and the amount of oxygen in it reaches the value that was found in the average sample of matte from the furnace for the preceding period of smelting.

The mattes investigated belong to all possible types of shaft-furnace smelting, from purely pyritic (Kalata-Karabash) to reducing (Yuliya, Ulen), and in all mattes oxygen was found as an essential constituent part of them. It was interesting and important to determine oxygen in mattes of another type of smelting, namely, reverberatory smelting. I had at my disposal collections of mattes selected specially for this investigation at three plants where reverberatory smelting to matte is conducted on raw and roasted cupriferous ores, namely, the Karabash, Tavaly, and Dzansul plants. In the investigation of these mattes the same procedures were applied as for the mattes of shaft-furnace smelting.

The results of analyses of the mattes from reverberatory smelting are given in Table 11.

Table 11. Analyses of mattes from reverberatory smelting.

Mattes	Cu, %	Fe, %	S, %	Pb, %	Zn, %	O, %	Total, %
Karabash plant, matte 23. I	12.5	51.12	26.32	—	—	7.42	97.36
Karabash plant, matte 2. II	12.5	51.5	25.91	—	—	7.71	97.61
Tavaly plant, matte 12. X	12.25	46.75	25.67	2.99	3.45	7.5	100.37
Tavaly plant, matte 17. X	12.75	48.05	25.89	3.29	3.10	7.63	100.71
Dzansul plant, matte 4. VII	38.26	31.57	25.92	—	—	3.50	99.25
Dzansul plant, matte 11. VII	37.76	31.87	26.06	—	—	3.80	99.49

Consequently, in the mattes from reverberatory smelting as well we encounter a considerable oxygen content.

If the losses of sulfur that occurred during the reduction of mattes in a stream of hydrogen are compared with one another, and these losses are expressed in percent of the initial sulfur content in the same mattes, it turns out that, irrespective of the copper richness of the mattes, these sulfur losses for the mattes investigated are very close in magnitude, as is seen from Table 12.

Table 12. Losses of sulfur during reduction of mattes by hydrogen.

Mattes	Charge, mg	Sulfur content in the charge, mg	Sulfur loss, mg	Sulfur loss, %
Karabash I	646	160.21	20.73	12.80
Karabash III	680.40	170.10	24.97	14.68
Bogoslovsk	658.50	162.9	18.50	11.36
Ulen	649.60	160.1	21.97	13.72
Yuliya	772.60	187.1	16.20	8.66
Kalata	631.60	148.9	15.35	10.30
Pyshma	634.20	143.4	20.68	14.42
Polevskoy	702.20	171.8	22.07	12.82

The figures in Table 12 show that, under the conditions of the experiments, hydrogen decomposes chiefly FeS and, to a lesser degree, the copper sulfides contained in the mattes. To clarify this, pure preparations of FeS and Cu₂S were heated in a stream of hydrogen at 820–830°, and the amount of sulfur split off in the process was determined. For both sulfides, appreciable decomposition began at 550°, which is somewhat below the temperature indicated by Wegscheider¹¹).

At 830°, 501.6 mg of Cu₂S in the course of two hours lost 7.8 mg of S, or 7.43% of the initial sulfur content in Cu₂S (21%). Under the same conditions, 502.6 mg of FeS lost 26.6 mg of sulfur, or 14.36% of its content in FeS (S = 36.86%). Similar data are reported by Schütz¹²), who found that hydrogen reduces Cu₂S, Ni₂S, and FeS at 610°. Thus, when dry hydrogen was passed over the sulfides at 600–610° for 3 hours,

\[ \begin{aligned} &\text{from Cu}_{2}\text{S there was liberated } 0.75\% \text{ sulfur}\\ &\text{from Ni}_{2}\text{S } \qquad\quad 2.64\% \text{ sulfur}\\ &\text{from FeS } \qquad\quad 3.46\% \text{ sulfur.} \end{aligned} \]

Thus, FeS decomposes much more vigorously than Cu₂S, which confirms the supposition stated above regarding the behavior of FeS during the reduction of mattes in a stream of hydrogen.

From a comparison of the results of determining oxygen in mattes by the CO method and by the Hampe method, given in Table 13, it is evident that the latter method gives much higher values for oxygen than the CO method. This is explained by the impossibility of complete reduction of iron oxides in the mattes by means of CO under the conditions of the experiments.

Table 13. Results of the determination of oxygen in mattes by means of carbon monoxide and by the Hampe method.

Mattes	Oxygen content, %, by the CO method	Oxygen content, %, by Hampe	Difference, %
Kalata	6.21	7.47	1.25
Karabash I	4.67	6.52	1.85
Karabash III	3.29	5.06	1.77
Pyshma	5.04	6.20	1.16
Polevskoy	2.84	3.15	0.31
Bogoslovsk	2.07	2.87	0.80
Ulen	1.78	2.84	1.06
Yulia	0.82	2.13	1.31

If oxygen is entered into the table of analyses of the mattes investigated, then, taking it into account, the sum of the elements in the mattes will amount to 99–100%, as is evident from the following Table 14.

Table 14. Analyses of mattes.

	Kalata	Karabash I	Karabash III	Pyshminsk	Polevskoy	Bogoslovsk	Ulen	Yulia	12.X. Tanalyk	17.X. Tanalyk	4.VII. Dzhezkazgan	11.VII. Dzhezkazgan
	percent	percent	percent	percent	percent	percent	percent	percent	percent	percent	percent	percent
Cu	14,55	27,60	37,10	31,98	42,94	38,57	41,05	44,90	12,25	12,75	38,26	37,76
Fe	53,40	40,15	32,58	38,31	29,49	32,89	30,66	27,67	46,75	48,05	31,57	31,87
S	23,58	24,80	25,00	22,60	24,17	24,74	24,65	24,22	25,67	25,89	25,92	26,06
Pb	—	0,56	0,2	—	—	—	—	—	2,99	3,29	—	—
Zn	—	1,00	0,60	—	—	0,70	—	—	3,45	3,10	—	—
O	7,47	6,52	5,06	6,20	3,15	2,87	2,84	2,13	7,75	7,63	3,50	3,80
Sum	99,00	100,63	100,54	99,09	100,05	99,77	99,20	98,92	100,37	100,71	99,25	99,49

Taking into account the circumstance indicated, as well as the sufficiently satisfactory agreement of the results of oxygen determination in individual samples, we may acknowledge that the reduction of the mattes by hydrogen proceeded to completion, and that the figures obtained for oxygen represent its actual content in the mattes.

In the literature¹³ the opinion has become established that oxygen is contained in mattes exclusively in the form of magnetite. Indirectly this is proved by the observed relationship between the amount of magnetite in mattes and their specific gravities. The amount of FeS O₄ in mattes is the greater, the closer the specific gravity of the matte is to the specific gravity of magnetite, i.e., 5.17. To verify this proposition, the specific gravities of some of the investigated mattes were determined. The determination was carried out in water with the aid of a pycnometer, the volume of which at 15° was 50 cu. cm. The mattes were finely ground, and, in order to obtain more accurate results, weighed portions of 4—4.5 g of matte were taken. After the weighed portion was introduced into the pycnometer, the latter was filled to two-thirds with boiled and cooled distilled water and was evacuated in a vacuum desiccator for two hours. During this, when the desiccator was shaken, a considerable evolution of air bubbles from the mattes was observed. When this phenomenon ceased, the pycnometer was filled with water, kept on the balance for ten minutes, and weighed. All specific gravities were referred to 15° and calculated by the well-known formula:

\[ D=\frac{C-A}{(B-A)-(E-C)+(B-A)[Q_1-Q+3\alpha(t_1-t)]}, \]

where

\[ \begin{aligned} D&=\text{specific gravity of the matte at }15^\circ,\\ A&=\text{weight of the pycnometer},\\ B&=\text{weight of the pycnometer filled with water at }15^\circ,\\ C&=\text{weight of the pycnometer with matte},\\ E&=\text{weight of the pycnometer with water and matte},\\ Q_1&=\text{specific gravity of water at the temperature of the experiment},\\ Q&=\text{specific gravity of water at }15^\circ,\\ \alpha&=\text{coefficient of expansion of glass }=0.000025,\\ t_1&=\text{temperature during the experiment},\\ t&=15^\circ. \end{aligned} \]

For each matte, 2–3 determinations were made, and when the difference did not exceed 0.05, the mean values were taken.

The results of the determinations of specific gravities are given in Table 15. In it, the first vertical column shows the copper content in the mattes, the second—the oxygen content, and the third—\( \mathrm{Fe_3O_4} \), calculated from the oxygen in the corresponding mattes.

Table 15. Specific gravities of mattes.

Mattes	Cu	O	\( \mathrm{Fe_3O_4} \)	Specific gravity at 15° C
Kalata	14.55	7.47	27.03	5.36
Karabash I	27.60	6.52	23.59	5.13
Karabash III	37.10	5.06	18.32	5.23
Pyshma	31.98	6.20	22.44	5.25
Polevskoi	42.94	3.15	11.40	5.23
Bogoslovsk	38.57	2.87	10.39	5.23
Ufaley	41.05	2.84	10.26	5.33
Yulia	44.90	2.13	7.71	5.26

From the data in the table we see that there is no dependence between the amounts of \( \mathrm{Fe_3O_4} \) and the specific gravities of the mattes. It is necessary to note the considerable difference between the specific gravities of mattes determined above and those given in the literature.

Thus, according to Keller\(^{14}\), the specific gravities of mattes are the greater, the richer the mattes are in copper, for example:

% Cu in mattes	13.62	43.00	60.22	80.00
Specific gravity of mattes	4.80	5.18	5.42	5.55

Unfortunately, the authors who give the specific gravities of mattes do not in most cases indicate what method they used in the determinations, to what temperature the specific gravities were referred, etc.

Determinations of the specific gravities of mattes without evacuating the pycnometer and without introducing corrections for temperature showed that in such a case, for one and the same matte, a difference of 0.5 and more units is possible. Thus, for example, for

Kalata matte

specific gravity without evacuating the pycnometer at 20° = 4.96 and 4.89

the same after evacuation and at 15° = 5.34 and 5.38;

Bogoslovsk matte

specific gravity without evacuating the pycnometer at 20° = 4.76 and 4.74

» » with » » 15° = 5.20, 5.26.

The foregoing permits us, for the time being, to draw the following conclusions:

1. oxygen is not an accidental component, but a quantitatively very substantial component of mattes of all types of matte smelting;

2. there is no dependence between the oxygen content in mattes and their specific gravity.

Let us now turn to consideration of the question of the form of the oxygen compounds of iron in mattes.

Most metallurgists consider that the sole representative of these compounds in mattes is magnetite, or metferrite of ferrous oxide, $\mathrm{FeO \cdot Fe_2O_3}$. There are also other opinions. Thus, according to Borchers’s opinion¹, based on Schreyer’s work², both in pyrite smelting and in bessemerizing mattes in basic converters, only ferrites can be formed, the composition of which varies between meta- and orthoferrite of ferrous oxide, i.e. $\mathrm{FeO \cdot Fe_2O_3}$ and $2\mathrm{FeO \cdot Fe_2O_3}$. The latter compound is very stable and is readily fused with the sulfides of Fe and Cu without oxidizing them. Borchers points out that, whereas $\mathrm{Fe_2O_3}$ is a strong oxidizing agent with respect to the sulfides of Fe and Cu, the oxidizing action of ferrites ceases when one molecule of $\mathrm{Fe_2O_3}$ is combined with two molecules of $\mathrm{FeO}$.

To clarify the question of the composition of the iron oxides dissolved in mattes, and of their identification with one or another of the compounds indicated in the literature, several approaches were outlined.

The first approach consisted in finding a solvent by means of which it would be possible to extract completely the metallic iron from mattes. For this purpose, the mattes were reduced with carbon monoxide and extracted with a $\mathrm{CuSO_4}$ solution before and after reduction. But this approach did not give positive results for the following reasons:

1) the reduction of iron oxides in mattes by carbon monoxide cannot be brought to completion,

2) the metallic copper present in mattes, and the metallic iron formed during their reduction, enter into reactions with the sulfides of the same metals at the reduction temperature.

The second approach was the extraction of iron oxides from mattes by smelting them with borax. Borax is a good solvent for metallic oxides and readily converts the latter into metaborates, which are quite stable at high temperatures. Thus, the oxides of copper and iron give metaborates according to the reaction:

\[ \mathrm{CuO\ (FeO) + Na_2B_4O_7 = Cu(Fe)(BO_2)_2 + 2NaBO_2.} \]

It was assumed that, when mattes containing iron oxides were smelted with borax, the oxides would dissolve in the borax. By separating the borates formed from the mattes, it would be possible to determine in them the quantity of iron that had been bound in the mattes with oxygen. Thus, Guess and Lathe³ state that, after remelting plant mattes under a layer of borax, they succeeded in obtaining mattes which, in sulfur content and melting temperature, were very similar to laboratory mattes obtained by fusing $\mathrm{FeS}$ and $\mathrm{Cu_2S}$. Here are several analyses of mattes before and after remelting under a layer of borax:

	Cu %	S %	Smelting temp.
plant matte	10.3	24.2	970°
the same after remelting under a layer of borax	13.6	29.9	890°
plant matte	13.7	26.4	912°
the same after remelting under a layer of borax	16.34	31.3	930°

Experiments in smelting mattes under a layer of molten borax did not give satisfactory results. The mattes mixed with the borax, a dark-brown glass was obtained, and separation of the matte from the borax did not succeed. During smelting there occurred a gradual oxidation of the sulfides of Cu and Fe, and their oxides dissolved in the borax. As Lutz and Chizhev showed\(^{18}\) in their study “On a method of indirect determination of CO₂ in salts by fusing them with borax”—the method with borax is not suitable in those cases when a base combined with CO₂, on smelting, readily passes to higher degrees of oxidation, absorbing oxygen from the air, as occurs, for example, with FeCO₃. Consequently, borax does not prevent oxidation of the substance, for which reason mattes are oxidized when fused with borax. Thus, this method of direct determination of iron oxides in mattes proved unsuitable.

An attempt was made to determine the composition of the iron oxides in mattes by calculation from the elementary composition of the mattes. A. Baikov\(^{19}\) used this procedure for his calculations of the rational composition of works mattes. Assuming that all metals, except copper, are present in mattes in the form of their simplest sulfides, Baikov calculated for a series of mattes their rational composition and the amount of metallic copper contained in them.

We may neglect the small amount of metallic iron in mattes and assume that in molten mattes all metals, including copper, are present in the form of their sulfides stable at high temperatures, i.e. Cu₂S, FeS, PbS, ZnS, etc. This assumption is not without foundation, since all changes in mattes—for example, the formation of compounds between sulfides and the separation in them of metallic copper—occur only during solidification of the mattes, and metallographic study of polished sections fixes the structure of already solidified mattes.

Having made these assumptions, let us calculate for some of the mattes studied their rational composition.

Kalata matte.

\[ \begin{aligned} \mathrm{Cu} — 14.55\% &\text{ require } 3.67\,\mathrm{S} \text{ for the formation of } \mathrm{Cu}_2\mathrm{S},\\ \mathrm{S} — 23.58\% \\ \mathrm{Fe} — 53.40\% \\ \mathrm{O}_2 — 7.47\% \end{aligned} \left\{ \begin{aligned} &\text{for the formation of FeS there remain}\\ &23.58 - 3.67 = 19.91\%\,\mathrm{S}, \text{ which will bind}\\ &34.67\%\,\mathrm{Fe}. \end{aligned} \right. \]

For the formation of ferrite, \((\mathrm{FeO}) \times (\mathrm{Fe}_2\mathrm{O}_3)y\), there remain \(53.4 - 34.67 = 18.73\%\) Fe and \(7.47\%\) O, which gives a ferrite of composition \(71.49\%\) Fe and \(27.51\%\) O. Since FeO contains \(22.27\%\) O, and Fe₂O₃—\(30.06\%\) O, it follows that the calculated ferrite, in its composition, lies between FeO and Fe₂O₃. The quantitative ratio FeO : Fe₂O₃ in the ferrite containing \(28.51\%\) O can be determined from the following two equations.

Denoting FeO = \(x\), Fe₂O₃ = \(y\), we have

\[ 1)\quad \frac{\mathrm{O}}{\mathrm{FeO}}\cdot x + \frac{3\cdot \mathrm{O}}{\mathrm{Fe}_2\mathrm{O}_3}\cdot y = 7.47 \]

\[ 2)\quad \frac{\mathrm{Fe}}{\mathrm{FeO}}\cdot x + \frac{2\,\mathrm{Fe}}{\mathrm{Fe}_2\mathrm{O}_3}\cdot y = 18.73. \]

The solution of the equations gives, for \(x=\mathrm{FeO}=5.17\) and \(y=\mathrm{Fe}_2\mathrm{O}_3=21.05\); the percentage composition of the ferrite will therefore be: \(19.73\%\) FeO and \(80.17\%\) Fe₂O₃, and the molecular composition \(\mathrm{FeO}_{27.45}\ \mathrm{Fe}_2\mathrm{O}_{3\,50.18}\).

\[ \begin{aligned} \text{Magnetite, } \mathrm{Fe}_3\mathrm{O}_4, &\quad \text{contains } 31.03\% \mathrm{FeO}.\\ \text{Orthoferrite, } 2\mathrm{FeO}\cdot \mathrm{Fe}_2\mathrm{O}_3 &\quad \text{contains } 47.84\% \mathrm{FeO}.\\ \text{Calculated ferrite} &\quad \text{contains } 19.73\% \mathrm{FeO}. \end{aligned} \]

Consequently, the ferrite present in the Kalata matte does not belong to any of the indicated types.

A calculation carried out in an analogous manner for the Karabash I, Pyshma, and Polevskoi mattes gave the following results:

ferrite in the Karabash I matte should contain 38.65% O
    »   »  Pyshma          »        »      32.29 » »
    »   »  Polevskoi      »        »      35.60 » »

Similar results are given by the calculation of the composition of ferrites for the other mattes investigated, i.e., ferrites are obtained with a higher oxygen content than the highest iron oxide, $\mathrm{Fe_2O_3}$, which contains 30.06% O. The only satisfactory result for the Kalata matte is accidental.

Since the formation and existence in mattes of iron oxides richer in oxygen than $\mathrm{Fe_2O_3}$ is difficult to assume at high temperatures, we see that the method of calculating the formula of ferrites from the elemental composition of mattes does not lead to the goal. This also indicates that the assumption underlying the calculations—namely, that all metals are present in molten mattes in the form of their stable sulfides at high temperatures—is incorrect. The supposition arises that the principal components of mattes, Cu and Fe, may occur in mattes in the form of sulfides richer in sulfur than $\mathrm{Cu_2S}$ and $\mathrm{FeS}$.

As regards iron sulfides, as Allen and Crenshaw showed$^{20)}$, sulfides richer in sulfur than $\mathrm{FeS}$ are extremely unstable at high temperatures. Being solid solutions of sulfur and $\mathrm{FeS}$, these sulfides have a variable composition depending on the temperature and the vapor pressure of sulfur under which the sulfide is heated. On heating in a neutral atmosphere these sulfides, like pyrite, dissociate very readily, and the final product of their dissociation at 1000–1400° is $\mathrm{FeS}$, which is stable up to 1300°, as shown by Yushkevich$^{21)}$, who melted synthetic $\mathrm{FeS}$ in an atmosphere of nitrogen. In view of the foregoing, the possibility of the existence in molten mattes of iron sulfides richer in sulfur than $\mathrm{FeS}$ can be excluded.

The second dominant component in mattes, copper, may be contained in them in the form of $\mathrm{Cu_2S}$ and, in part, $\mathrm{CuS}$. This is confirmed by the investigation of copper sulfides carried out by Posnjak, Allen, and Merwin$^{22)}$, who proved that $\mathrm{Cu_2S}$ can retain, even above its melting temperature, a considerable amount of $\mathrm{CuS}$ in solid solution. In products obtained by melting copper with sulfur, there is always some excess of sulfur, the magnitude of which depends on the melting temperature, and even by repeated remelting at 1300° it is not possible to remove the excess sulfur from them, i.e., to bring the process

\[ 4\mathrm{CuS} \rightleftharpoons 2\mathrm{Cu_2S} + 2\mathrm{S} \]

to completion. Thus, $\mathrm{Cu_2S}$ obtained by Shad and Bornemann$^{23)}$ by melting Cu with S in a graphite crucible had the composition Cu—78.81% and S—20.95%, i.e., it contained an excess of sulfur of 1.08% compared with theory.

A preparation obtained by me by the same method, after remelting at 1200°, contained 78.30% Cu and 21.7% S, i.e., the excess of sulfur compared with theory was 1.96%. A preparation remelted twice at 1300° had the composition: Cu—78.84% and S—21.16%, i.e., it contained an excess of sulfur of 1.29%.

Only when $\mathrm{Cu_2S}$ is melted in a vacuum furnace is it possible to obtain a preparation of theoretical composition. G. Urazov$^{24)}$, on the basis of studying the phase diagram of the Cu : S system, came to the conclusion that $\mathrm{Cu_2S}$ is capable of dissolving in the liquid state, under ordinary conditions, about 1.12% S; but the comprehensive study of copper sulfides carried out by Posnjak and co-workers led to the establishment of the fact of the existence of solid solutions $\mathrm{CuS} + \mathrm{Cu_2S}$ of variable composition.

The possibility of the existence of CuS together with Cu$_2$S in copper-poor mattes, in the free state or in combination with FeS, is also noted by Neumann and Bauer$^{25}$). In addition, in all copper mattes, both plant and synthetic, obtained by fusing pure preparations of Cu$_2$S with FeS, a macroscopic separation of metallic copper is observed. Without dwelling on the causes of this phenomenon, we shall point out that the separation of metallic copper is observed only in alloys of sulfides containing, along with Cu$_2$S, sulfurous iron.

Thus, in the composition of solidified mattes we have the following components: Cu, CuS, Cu$_2$S, FeS, metallic Fe, and Fe$_x$O$_y$, the sulfides being present in mattes both in the free state and in the form of chemical compounds with one another. The amount of metallic Fe in mattes is, in general, very small, as was shown by the extraction of mattes with a solution of CuSO$_4$; if it were possible, by means of a rational analysis of mattes, to determine the quantities of the copper constituents—Cu, CuS, and Cu$_2$S—this would make it possible to ascertain the composition of the ferrites of interest to us.

Among investigations concerning methods of rational analysis of copper mattes, special attention is deserved by the works of C. van Besneveld and E. Leaver$^{26}$), A. Gibb and K. Philp$^{27}$), and S. Yushkevich$^{28}$). A. Gibb and K. Philp used, for the rational analysis of both laboratory and plant mattes, a solution of AgNO$_3$, which reacts with the components of mattes as follows:

\[ \mathrm{Cu} + 2\mathrm{AgNO}_3 = \mathrm{Cu(NO}_3)_2 + 2\mathrm{Ag} \]

\[ \mathrm{Cu}_2\mathrm{S} + 4\mathrm{AgNO}_3 = 2\mathrm{Cu(NO}_3)_2 + \mathrm{Ag}_2\mathrm{S} + 2\mathrm{Ag} \]

\[ \mathrm{FeS} + \mathrm{AgNO}_3 \text{— does not react.} \]

If neither CuS nor metallic Fe in mattes is taken into account, then a solution of AgNO$_3$ can serve as a good reagent for the quantitative determination of Cu and Cu$_2$S in mattes and, consequently, for calculating their rational composition. Since the investigation of Gibb and Philp has entered the specialized literature on copper metallurgy and is cited by many authors in their works concerning the structure of mattes, it seemed important to verify some of the authors’ observations and to apply their method to the establishment of the rational composition of a series of laboratory and plant mattes. The authors do not give sufficient directions regarding the execution of the method itself; therefore these details had to be established by experiments, to the description of which we now turn.

It was first established how a solution of AgNO$_3$ acts on metallic copper, FeS, and Cu$_2$S; after this, the method of separating Ag from Ag$_2$S was investigated, and then the treatment of mattes was undertaken. For the experiments a neutral solution of AgNO$_3$ containing 32 g of salt per liter was used. The silver content in the solutions before and after treatment of the corresponding substances with them was determined by Volhard titration. The substances studied, in the form of fine powders, were agitated with the AgNO$_3$ solution by means of a glass stirrer in closed brown reagent bottles at room temperature.

1. Action of AgNO$_3$ on metallic copper. The copper was prepared by reducing chemically pure cupric oxide with hydrogen. 180 mg of Cu were agitated with 50 cu. cm of AgNO$_3$ for 4 hours and precipitated 607 mg of metallic silver; 179 mg of Cu were found in solution. According to theory, 610.8 mg should have separated out and 180 mg of Cu should have passed into solution. Consequently, practically all the copper passed into solution and precipitated an almost theoretical quantity of silver.

2. Action of AgNO$_3$ solution on FeS. Ferrous sulfide was obtained by calcining pure pyrite in an atmosphere of nitrogen at 1000°; this FeS did not react with AgNO$_3$.

3. Action of Cu$_2$S on AgNO$_3$. Three preparations of Cu$_2$S served for the experiments, prepared by fusing Cu with S in graphite test tubes in a kryptol furnace.

The preparations had the following composition:

Preparation	% Cu	% S
1st preparation	78.84	21.16
2nd »	78.70	21.30
3rd »	78.35	21.65

Samples of 0.5 g of sulfide were agitated with 100 cc of an \(AgNO_3\) solution for 4–30 hours. In the solutions separated from the precipitates, the amounts of Ag and Cu were determined; in the precipitates—Ag and \(Ag_2S\). To separate Ag from \(Ag_2S\), the thoroughly washed precipitates, together with the filters, were agitated with a 10% solution of \(Fe(NO_3)_3\), which dissolves silver and does not act on \(Ag_2S\), as was established by experiments. Thus, 436 mg of powdered cement silver, when agitated with 50 cc of \(Fe(NO_3)_3\) solution, dissolved completely after 30 minutes, and 436.18 mg Ag was found in the solution. When 500 mg \(Ag_2S\) was treated under the same conditions, no silver appeared in the solution. The results of the experiments with \(Cu_2S\) are given in Table 16.

Table 16. Action of an \(AgNO_3\) solution on \(Cu_2S\). Preparation No. 1 \(Cu_2S\) contained 78.84% Cu and 21.16% S; charge 0.5 g; results recalculated to 1 g \(Cu_2S\).

Experiment No.	Duration of treatment, hr	Total Ag found in precipitate, mg	Metallic Ag found in precipitate, mg	Ag in the form of \(Ag_2S\) found in precipitate, mg	Copper passed into solution, %, calculated from silver	Copper passed into solution, %, by analysis of solutions
1	4	2310	869.40	1440.60	86.32	85.78
2	4	2318	891.0	1427.0	86.62	86.20
3	10	2424.80	1040.0	1384.80	90.62	90.65
4	12	2626.34	1168.0	1458.34	98.13	99.50
5	24	2625.0	1261.0	1364.9	98.10	98.90

Preparation No. 3 contained 78.35% Cu and 21.65% S.

Experiment No.	Duration of treatment, hr	Total Ag found in precipitate, mg	Metallic Ag found in precipitate, mg	Ag in the form of \(Ag_2S\) found in precipitate, mg	Copper passed into solution, mg	Copper passed into solution, mg	Copper passed into solution, %
6	30	2559.20	776.60	1782.60	754	736	93.94
7	30	2559.20	771.70	1787.50	—	—	—
8	27.5	2500.60	965.80	1534.80	736.80	740	94.45

From the analysis of \(Cu_2S\), the theoretical ratio \(Cu_2S : CuS\) in the preparation was calculated; from the amounts of Ag and \(Ag_2S\) in the precipitates from the treatment of \(Cu_2S\) with \(AgNO_3\) solution, \(CuS\) and \(Cu_2S\) were determined. Two examples of such calculations are given below.

For experiment No. 6.

According to the reactions:

\[ Cu_2S + 4AgNO_3 = 2Cu(NO_3)_2 + Ag_2S + 2Ag \]

\[ CuS + 2AgNO_3 = Cu(NO_3)_2 + Ag_2S \]

The compound \(Cu_2S\) corresponds to \(776.6 \times 2 = 1553.2\) mg Ag, 457.6 mg, or Cu in the form of \(Cu_2S\).

The compound CuS corresponds to \(2559.2 - 1553.2 = 1006.0\) mg Ag, or 296.4 mg Cu in the form of CuS.

From the amount of precipitated silver \((\mathrm{Ag} + \mathrm{Ag}_2\mathrm{S})\), 754.0 mg Cu should have passed into solution; analysis of the solution gave 736.0 mg Cu.

Composition of the dissolved copper sulfide:

\[ \begin{array}{rclcc} & & & \% & \text{Theoretical composition} \\ \dfrac{\mathrm{Cu}_2\mathrm{S}}{2\mathrm{Cu}}\cdot 457.6 &=& 572.9\ \text{mg } \mathrm{Cu}_2\mathrm{S} &= 56.24 & 88.72\% \\ \dfrac{\mathrm{CuS}}{\mathrm{Cu}}\cdot 296.4 &=& 445.8\ \text{mg } \mathrm{CuS} &= 43.76 & 11.28\% \\ \hline \text{Sum} && 1018.7 && 100.00 \quad 100.00 \end{array} \]

For experiment No. 8.

The compound \(\mathrm{Cu}_2\mathrm{S}\) corresponds to \(965.8 \times 2 = 1931.6\) mg Ag, or 569.1 mg Cu in the form of \(\mathrm{Cu}_2\mathrm{S}\); the compound CuS corresponds to \(2500.6 - 1931.6 = 569.0\) mg Ag, or 167.7 mg Cu in the form of CuS.

From the amount of precipitated silver, \(569.1 + 167.7 = 736.8\) mg Cu should have passed into solution; by analysis of the solution, 740.0 mg Cu was found.

Composition of the dissolved copper sulfide:

\[ \begin{array}{rclcc} & & & \% & \text{Theoretical composition} \\ \dfrac{\mathrm{Cu}_2\mathrm{S}}{2\mathrm{Cu}}\cdot 569.1 &=& 712.6\ \text{mg } \mathrm{Cu}_2\mathrm{S} &= 73.85 & 88.72\% \\ \dfrac{\mathrm{CuS}}{\mathrm{Cu}}\cdot 167.7 &=& 252.2\ \text{mg } \mathrm{CuS} &= 26.15 & 11.28\% \\ \hline && 964.8\ \text{mg} && 100 \quad - \quad 100 \quad - \end{array} \]

The experiments performed showed that only in individual cases does a solution of \(\mathrm{AgNO}_3\) give more or less satisfactory results, and not all the copper is extracted from \(\mathrm{Cu}_2\mathrm{S}\); the ratio \(\mathrm{Cu}_2\mathrm{S} : \mathrm{CuS}\) according to theory (on the basis of the analysis of \(\mathrm{Cu}_2\mathrm{S}\)) does not at all coincide with that obtained from the precipitated amounts of Ag and \(\mathrm{Ag}_2\mathrm{S}\).

The method with a solution of \(\mathrm{AgNO}_3\) was tested on the following materials: natural copper pyrites, the same pyrites after remelting, and a series of laboratory and works mattes.

Treatment of copper pyrites with a solution of \(\mathrm{AgNO}_3\).

The pyrites had the composition:

\[ \begin{array}{rcl} \text{Pyrites had composition:}\ \ldots\ldots & \mathrm{Cu} &= 32.75\% \\ \text{” ” ”}\ \ldots\ldots & \mathrm{Fe} &= 31.29\% \\ \text{” ” ”}\ \ldots\ldots & \mathrm{S} &= 34.34\% \\ \cline{3-3} && 98.38\% \end{array} \]

The pyrites was ground to pass a 150-mesh sieve, and several 1 g samples were agitated with 50 cc of \(\mathrm{AgNO}_3\) solution. The results of the experiments are given in Table 17 (experiments 1–3).

Table 17. Action of an \(AgNO_3\) solution on copper pyrite.

Experiment No.	Duration of agitation, hours	Found in the precipitate: total Ag, mg	Found in the precipitate: metallic Ag, mg	Found in the precipitate: Ag as \(Ag_2S\), mg	Copper passed into solution: by analysis, mg	Copper passed into solution: calculated from Ag, mg	In % of Cu in the preparation
1	28	142.90	—	—	31.20	—	9.53
2	28	142.90	—	—	31.20	—	9.53
3	27.5	141.60	28.70	112.90	28.94	41.71	8.84
Remelted pyrite.	Remelted pyrite.	Remelted pyrite.	Remelted pyrite.	Remelted pyrite.	Remelted pyrite.	Remelted pyrite.	Remelted pyrite.
4	27	490.80	203.91	286.90	143.47	144.65	40.85
5	27	490.80	203.91	286.90	143.47	144.65	40.85

According to experiment No. 3.

By the action of \(Cu_2S\), \(28.7 \times 2 = 57.4\) mg of Ag was precipitated, which corresponds to 16.91 mg of Cu in the form of \(Cu_2S\); by the action of \(CuS\), \(141.6 - 57.4 = 84.2\) mg of Ag was precipitated, which corresponds to 21.81 mg of Cu in the form of \(CuS\); consequently, in all, 41.71 mg of copper was extracted according to Ag; analysis of the separate solution gave 28.94 mg of Cu.

Copper pyrite was melted in a graphite test tube in a cryptol furnace and ground to pass 150 openings. The preparation had the composition:

\[ \begin{array}{rl} Cu &— 35.13\%\\ Fa &— 33.86\%\\ S &— 28.86\%\\ \hline &98.85\% \end{array} \left\{ \begin{array}{l} \text{In the product there was a noticeable separation}\\ \text{of metallic copper.} \end{array} \right. \]

Two 1-g samples were agitated with 100 cubic cm of \(AgNO_3\) solution. The results of the experiments are given in Table 17, experiments Nos. 4 and 5.

By the action of \(Cu_2S\), \(203.91 \times 2 = 407.82\) mg of Ag was precipitated, which corresponds to 120.2 mg of Cu in the form of \(Cu_2S\); by the action of \(CuS\), \(490.8 - 407.82 = 82.98\) mg of Ag was precipitated, which corresponds to 24.45 mg of Cu in the form of \(CuS\). Consequently, in all, 144.65 mg of copper was extracted (according to Ag); analysis of the solution gave 143.47 mg of Cu. From the experiments given, we see that \(AgNO_3\) solution acts very little on natural copper pyrite, which is a sulfosalt of composition \(Cu_2S \cdot Fe_2S_3\), and vigorously decomposes remelted pyrite, which is an alloy \(Cu_2S + FeS\). Even in this case, however, only 40.85% of the copper contained in it is extracted from the alloy.

Action of \(AgNO_3\) solution on laboratory mattes.

Several mattes were prepared by melting \(Cu_2S\) with pure pyrite and with \(FeS\) obtained by remelting pyrite in a graphite test tube.

Mattes from \(Cu_2S + FeS_2\).

Composition of the mattes:

	1	2	3
Cu	19.64%	29.00%	38.66%
Fe	46.51 »	39.39 »	31.10 »
S	33.85 »	31.61 »	30.24 »
	100%	100%	100%

Mattes of Cu₂S + FeS

Composition of the mattes:

	4	5	6
Cu	20,70%	20,25%	30,24%
Fe	47,94 »	47,25 »	39,27 »
S	31,36 »	32,50 »	30,49 »
	100%	100%	100%

Characterization of the mattes

In the mattes of Cu₂S + FeS₂ no separation of metallic copper was noticeable; the mattes were very brittle and were easily ground into a dark-green powder. In the mattes of Cu₂S + FeS abundant separation of metallic copper was observed, both on the surface and throughout the fracture of the mattes. The mattes were very hard, tough, somewhat malleable, and difficult to grind. The color of the matte powder was dark bronze with a red tint.

The mattes of Cu₂S + FeS₂, when heated with a saturated solution of KClO₃ in HNO₃ (1.4), liberated elemental sulfur; the mattes of Cu₂S and FeS dissolved completely in this operation. All the mattes were ground to pass 150 openings. Samples of 1 g of matte were agitated in brown reagent bottles with 100 cu. cm. of AgNO₃ solution for a definite time. In the solutions the amounts of Ag and Cu were determined, and in the precipitates—Ag and Ag₂S. The results of the experiments are given in Table 18.

Table 18. Action of AgNO₃ solution on laboratory mattes.

Matte No.	Cu, %	Duration of treatment, hours	Found in precipitate: total Ag, mg	Found in precipitate: metallic Ag, mg	Found in precipitate: sulfur in the form of Ag₂S, mg	Passed into solution: calculated from sulfur, mg	Passed into solution: according to solution, mg	Difference from Cu in matte, %
Mattes of Cu₂S + FeS₂
1	19,64	7	212,44	115,26	97,18	62,58	60,8	30,96
2	29,00	7	463,30	214,70	248,60	136,52	135,6	46,76
3	38,66	27	750,60	228,60	522,00	221,15	218,3	56,46
3	38,66	27	750,60	228,60	522,00	221,15	218,3	56,46
Mattes of Cu₂S + FeS
4	20,70	28	646,80	255,90	390,90	190,58	188,04	92,08
4	20,70	28	646,80	255,90	390,90	190,58	188,04	92,08
5	20,25	7	805,50	223,60	581,90	237,40	172,40	85,14
6	30,24	7	829,40	370,10	459,30	244,40	244,40	80,82

In matte No. 1 of Cu₂S + FeS₂ there was no metallic copper, and all of it could be present in the form of Cu₂S and Cu₂S. Under this assumption, 60.8 mg Cu,

extracted from matte No. 1 would have precipitated, according to the reaction: $\mathrm{Cu_2S + 4AgNO_3 = 2Cu(NO_3)_2 + 2Ag + Ag_2S}$, 206.4 mg Ag; the experiment gave 212.44 mg Ag. For matte No. 2, under the same assumption, 135.6 mg Cu in the form of $\mathrm{Cu_2S}$ would have precipitated 460.2 mg Ag; 463.3 mg Ag was obtained. The results of treating mattes Nos. 1 and 2 show that in the extracted portion of matte No. 1 all the Cu is present as $\mathrm{Cu_2S}$; in the extracted portion of matte No. 2 the amount of CuS is equivalent to $463.3 - 2 \times 214.7 = 33.9$ mg Ag and amounts to 9.99 mg Cu in the form of CuS. In addition, the method with an $\mathrm{AgNO_3}$ solution was applied to several laboratory mattes made of $\mathrm{Cu_2S + FeS}$, containing, by analysis, 20.23, 29.78, and 39.16% Cu. To prevent the separation of metallic copper during cooling of the mattes, some of them were quenched below and above the melting temperature. Even so, a slight separation of metallic copper was still noticeable in the mattes. Agitation of the mattes with an $\mathrm{AgNO_3}$ solution was carried out also at 50–60°, but this did not improve the results. All the results obtained are given in Table 19.

Table 19. Action of an $\mathrm{AgNO_3}$ solution on laboratory mattes.

Matte No.	Copper per 1 g matte, mg	Found in precipitate: total Ag, mg	Found in precipitate: metallic Ag, mg	Found in precipitate: sulfur as $\mathrm{Ag_2S}$, mg	Copper passed into solution, mg	Copper passed into solution, % of Cu in matte	NOTE.
1	931.60	1232.90	375.10	857.80	364.90	93.18	Agitation 5 hr. 15° C
2	391.60	1219.40	358.40	861.00	361.00	92.18	Agitation 5 hr. 50° C
3	390.20	1176.10	307.10	869.00	343.50	88.03	Quenched at 1000° C
4	297.80	813.60	320.00	493.60	251.60	84.49	Agitation 5 hr. 15° C
5	303.60	927.70	297.90	629.80	264.50	87.12	Quenched at 600° C
6	202.30	119.90	Not determined	Not determined	167.00	82.55	Agitation 4 hours.
7	202.30	119.90	Not determined	Not determined	163.20	86.67	Agitation 4 hours.

From the data of Table 19 we see that, first, an $\mathrm{AgNO_3}$ solution does not extract all the copper from laboratory mattes, but only part of it; the amount of unreacted copper increases with an increase in the percentage of Fe in the mattes. This shows that part of the copper sulfides in the mattes is bound with FeS in the form of stable alloys or chemical compounds that are not acted upon by an $\mathrm{AgNO_3}$ solution; second, from the ratio of Ag to $\mathrm{Ag_2S}$ in the precipitates from the treatment of mattes with $\mathrm{AgNO_3}$, it is evident that almost all mattes contain a greater or lesser amount of free CuS; it is not possible to determine the latter accurately, since, in the precipitation of metallic silver from $\mathrm{AgNO_3}$, besides $\mathrm{Cu_2S}$, the metallic copper present in the mattes also participates, and its amount cannot yet be determined.

To establish the suitability of an $\mathrm{AgNO_3}$ solution for the rational analysis of mattes, the reagent was tested on several works mattes. The experiments were conducted in the same way as in the case of the laboratory mattes, and gave the results placed in Table 20.

Table 20. Action of an AgNO₃ solution on plant mattes.

Matte	% Cu	Duration of agitation, hours	Found in the precipitate: total Ag, mg	Found in the precipitate: metallic Ag, mg	Found in the precipitate: Ag as Ag₂S, mg	Copper passed into solution, mg	Copper passed into solution, % of the copper in the matte
Pyshminsk., 10 p.m.	38.50	7	1053.16	313.58	739.58	309.20	80.31
Kalatinsk	14.55	32	490.42	227.36	263.06	138.40	95.12
Ulen	41.05	32	1301.76	522.63	779.13	378.8	92.28

As in the investigation of laboratory mattes, increasing the duration of treatment of plant mattes even to 32 hours does not achieve extraction from them of all the copper. The experiments prove the existence in all these mattes, alongside Cu₂S and CuS, of a form that gives an excess of Ag₂S in the precipitate Ag + Ag₂S. Consequently, in all the mattes investigated, part of the copper sulfides is present with FeS in a bound state, and this part is not subject to the action of AgNO₃. Here one observes a general rule according to which a reagent suitable as a solvent for a given metal and its simplest, but free, compounds loses this property if the metal and its compounds are constituents of more complex formations. Thus, a solution of Fe₂(SO₄)₃ readily dissolves metallic copper, its oxides, and sulfides; the same reagent does not dissolve Cu₂S, CuO, and Cu₂O if the latter are constituents of sulfo-salts (e.g., copper pyrite, bornite) or of ferrites of the oxide and suboxide of copper²⁹). Dilute H₂SO₄ is an excellent solvent that has found wide application for extracting copper oxide and zinc oxide from oxidized or roasted ores; but dilute H₂SO₄ does not extract CuO and ZnO from the corresponding ferrites. The relation of the mattes investigated to AgNO₃ solution serves as an illustration of the same general rule.

In copper pyrite, copper is a constituent of the sulfo-salt Cu₂S · Fe₂S₃. When pyrite is treated with AgNO₃ solution, the copper theoretically should not pass into solution; in the experiment, however, there was nevertheless some, although very small, extraction. Thus, after agitation of 1 g of pyrite for 7 hours with AgNO₃ solution, only 23 mg of Cu, or 70.2% of its content in the pyrite, passed into solution, and 106.22 mg of Ag were precipitated instead of 78.07 mg, which corresponds to 23 mg of copper. This may be explained by the presence in the natural pyrite of Cu₂S or CuS.

It remained to clarify one more question, namely, the suitability of AgNO₃ solution for the rational analysis of alloys in which Cu₂S forms neither chemical compounds nor solid solutions with the other components of the alloys. Such a system is represented by the alloys Cu₂S : PbS. According to Friedrich³⁰), these two sulfides give only a simple eutectic consisting of 51% Cu₂S and 49% PbS. For the experiments two alloys were prepared, one before, the other beyond the eutectic, by fusing the preparations Cu₂S : PbS in a charcoal crucible. Analysis of the alloys obtained gave the following results:

	Pb	Cu	S
hypo-eutectic alloy	50.67	32	17.33
hyper-eutectic alloy	33.87	47.55	18.58

There was no separation of metallic copper in the alloys. Treatment of the finely ground alloys with AgNO₃ solution for 14 hours gave the results presented in Table 21.

Table 21. Action of an AgNO₃ solution on Cu₂S : PbS alloys.

Alloy composition	Cu content in 1000 mg, mg	Pb content in 1000 mg, mg	Found in precipitate: total Ag, mg	Found in precipitate: metallic Ag, mg	Found in precipitate: Ag as Ag₂S, mg	Cu extracted, mg	Cu extracted, %	Pb extracted, mg	Pb extracted, %
40 Cu₂S : 60 PbS	320.0	506.70	1066	359.40	706.60	292.80	91.50	69.32	13.70
60 Cu₂S : 40 PbS	475.50	338.70	1521	539.80	981.20	439.60	92.45	12.60	3.72
100 PbS (700 mg)	—	606.06	102.06	—	102.06	—	—	93	15.70

Since lead was found in the separate solutions, an experiment was set up in which 700 mg of pure PbS was treated with an AgNO₃ solution under the same conditions. The results of this experiment are included in Table 21.

If the precipitation of Ag proceeded according to the reaction:

\[ \mathrm{PbS} + 2\mathrm{AgNO}_3 = \mathrm{Pb}(\mathrm{NO}_3)_2 + \mathrm{Ag}_2\mathrm{S}, \]

then the 93 mg of Pb that passed into solution would have precipitated 97.02 mg in the form of Ag₂S; the experiment gave 102.06 mg Ag; so small a difference speaks in favor of the view that the reaction between PbS and the AgNO₃ solution proceeds according to the scheme indicated above. Let us compare the amount of Ag which, according to the known reactions, should have been liberated from AgNO₃ during treatment of the investigated alloys with that obtained in the experiments.

For the alloy 40 Cu₂S : 60 PbS:

\[ \begin{aligned} 69.32\ \text{mg Pb} &\text{ should precipitate } 72.32\ \text{mg Ag}\\ 292.8\ \text{mg Cu} &\text{ should precipitate } 993.8\ \text{mg Ag}\\ \hline &\text{total } 1066.12\ \text{mg.} \end{aligned} \]

Obtained in the experiment . . . . . 1066.00 mg.

For the alloy 60 Cu₂S : 50 PbS:

\[ \begin{aligned} 12.6\ \text{mg Pb} &\text{ should precipitate } 13.14\ \text{mg Ag}\\ 439.6\ \text{mg Cu} &\text{ should precipitate } 1492.00\ \text{mg Ag}\\ \hline &\text{total } 1505.14\ \text{mg Ag} \end{aligned} \]

Obtained in the experiment . . . . . 1521.00 mg Ag.

Consequently, in the alloys we were dealing with free components Cu₂S and PbS. The experiments performed show that even from such alloys an AgNO₃ solution does not extract all the copper, and the almost complete absence of copper in the silver precipitates (Ag + Ag₂S) obtained by Gibb and Philp in their investigation appears hardly understandable. The authors treated 11 mattes containing from 49.7 to 80.1% Cu with an AgNO₃ solution, and only in the precipitates from mattes with 72.1% Cu and higher did they find in all only 1.6 to 4.67 mg Cu, whereas in experiments with laboratory and works mattes we succeeded in extracting no more than 85–98% of the Cu contained in the mattes. Experiments on the application of an AgNO₃ solution for the rational analysis of mattes lead to the following conclusions.

1. A solution of $AgNO_3$ readily reacts with individual free copper sulfides and with metallic copper.

2. In alloys $Cu_2S : CuS$, with the aid of this reagent it is possible to determine the quantities of both components only very approximately.

3. From more complex alloys, e.g., mattes, an $AgNO_3$ solution extracts only part of the copper, irrespective of the duration of treatment of these mattes. The extraction of copper is the smaller, the more iron there is in the mattes.

4. In copper mattes, along with $Cu_2S$ and $CuS$, there is a greater or lesser quantity of metallic copper, which also precipitates $Ag$ and the amount of which cannot yet be determined.

In view of all these circumstances, an $AgNO_3$ solution cannot serve as a reagent for the rational analysis of copper-smelting mattes.

The only valuable result yielded by the use of this reagent is the establishment of the presence in mattes, both laboratory and works mattes, of a considerable amount of $CuS$ along with $Cu_2S$.

Potassium cyanide as a reagent for the rational analysis of mattes.

S. Yushkevich$^{31}$) was engaged in the rational analysis of pure alloys $Cu_2S : FeS$. As the reagent for analysis he used a 7% solution of KCN, which readily dissolves $Cu$ and $Cu_2S$ and does not act at all on $FeS$. Analyzing the solutions obtained from treating $Cu_2S : FeS$ alloys with KCN solution, Yushkevich came to the following conclusion. On heating powdered mixtures $Cu_2S + FeS$, $Cu_2S + Fe$, and $Cu + FeS$, already at a temperature of 200—400° the compound $(Cu_2S)_2FeS$ is formed, soluble in KCN. The existence of the same compound in mattes is confirmed by Bornemann and Schreyer$^{32}$) on the basis of studying the phase diagram of $Cu_2S : FeS$ alloys.

It was therefore decided to use the method of rational analysis of mattes developed by S. Yushkevich.

In view of the ready solubility of $Cu$ and $Cu_2S$ in KCN, it was to be expected that in the residue from treating mattes with this reagent there would be insoluble compounds contained in the mattes, namely $FeS$ and iron oxides. From the contents of $S$ and $Fe$ in the residue it would therefore be possible to determine the formula of the oxygen compounds of iron. But experiments with an $AgNO_3$ solution made it necessary to test the suitability of KCN as a reagent for investigating $Cu_2S : FeS$ alloys.

First, the action of KCN was investigated on the individual components of mattes in the free state, then on $Cu_2S : FeS$ alloys, and finally on works mattes.

Action of a 7% KCN solution on metallic copper, copper and iron sulfides, and iron oxides.

Weighed portions of the indicated substances were treated with KCN solution in an atmosphere of hydrogen for 20—24 hours. After this the solution was filtered from the insoluble residue in an atmosphere of hydrogen, and in separate portions of the solution the amounts of $Cu$, $Fe$, and $S$ were determined. For this purpose a definite portion of the solution was evaporated with $HNO_3$, then with $H_2SO_4$ to white fumes of $SO_3$, in order to destroy the complex and convert $Cu$ and $Fe$ into their sulfates, in which $Cu$ and $Fe$ were determined by the usual methods. To 50 cubic cm of filtrate intended for the determination of sulfur were added 25 cubic cm of a 5% solution of $Hg(CN)_2$ and several cubic cm of ammonium acetate; the mixture was heated to boiling; the precipitated $HgS$ was filtered through a Gooch crucible, washed with hot water, dried at 110°, and weighed. Thus, from the weight of $HgS$ the amount of sulfur that had passed into solution from a definite portion of the substance was determined.

Pure copper was obtained by reducing Cu₂O with hydrogen; Cu₂S had the composition: 79.7% Cu and 20.3% S; among the iron sulfides, pyrite, pyrrhotite, FeS, and copper pyrites were tested; among the iron oxides—ferric oxide, natural magnetite, and ferrous oxide. The composition of these substances is given below.

The pyrite was boiled in 20% HCl; from the natural pyrrhotite the purest crystals were selected and washed with water acidified with H₂SO₄ until the reaction for Fe disappeared; ferrous sulfide was obtained by fusing pyrite in a carbon crucible. All sulfides were ground to pass a 60-mesh sieve.

Composition of the substances

Substance	% Fe	S	Cu
Pyrrhotite	59.74	38.47	—
Pyrite	46.46	53.20	—
FeS	63.08	36.86	—
Copper pyrites	31.29	34.34	32.75
Ferric oxide	69.94	—	100% Fe₂O₃
Magnetite	69.46	and O = 26.53
FeO “Kahlbaum”	34.15% Fe₂O₃ + 65.85% FeO

Charges of 1.25 g of the above-mentioned sulfides were treated with 125 cc of a 6% KCN solution for 24 hours; for the iron oxides the duration of treatment was 6 hours. The samples were shaken periodically; Fe and Cu were determined in 100 cc of the solution, and sulfur in 50 cc of the solution from a parallel sample.

In the solutions from pyrite and FeS neither Fe nor S was found; there was no sulfur in the solution from pyrrhotite, but 2.52 mg of Fe was found; in view of the absence of sulfide sulfur in the solution, it must be assumed that the iron that passed into solution was present in the pyrrhotite in the form of sulfate, which could not be washed out beforehand. These experiments show that a 6% KCN solution does not act on iron sulfides from FeS₂ to FeS inclusive, i.e., those containing from 53.20% to 36.46% S.

Copper pyrites is a sulfosalt, Cu₂S : Fe₂S₃; since Fe₂S₃, in sulfur content, lies between the iron sulfides investigated, it may definitely be said that KCN solution will not act on sulfide of the composition Fe₂S₃ in the free state either.

In the solutions obtained by treatment of the iron oxides (Fe₃O₄, FeO, and Fe₂O₃), no iron was found.

Action of a 6% KCN solution on Cu and Cu₂S

From 625 mg of metallic copper, 535.5 mg, or 85.68% of the copper taken, passed into solution in 24 hours. Subsequently, for mattes, charges of 1.25 g were taken. If it is assumed that all the copper in them is present in metallic form, then, according to the experiment, it should have dissolved completely during 24 hours of treatment.

1.25 g of Cu₂S dissolved completely in KCN within 2 hours. In 50 cc of solution, corresponding to 500 mg of Cu₂S, 398 mg of Cu and 99.25 mg of S were found, instead of 398.5 mg of Cu and 101.5 mg of S, according to the analysis of Cu₂S.

Action of a 6% KCN solution on natural copper pyrites

Selected pure crystals of copper pyrites were ground, thoroughly washed with dilute H₂SO₄ and water, then with alcohol and ether. Three experiments were set up; for each experiment 1.25 g of pyrites, 125 cc of KCN solution were taken, and the treatment lasted 24 hours.

The results of these experiments are given below.

From 1000 mg of pyrite there passed into solution:

	Experiment I	Experiment II	Experiment III
Cu	14.8 mg	12.8 mg	14.4 mg
Fe	5.59 mg	5.04 mg	4.76 mg
S	4.25 mg	2.87 mg	4.20 mg

Consequently, in all, from 1 to 1.5% Fe, contained in 1 g of pyrite, dissolved; the ratio Cu : S in the solution is very close to that in the compound Cu₂S; thus, for example:

	Required S	Found S
14.8 mg Cu in the form of Cu₂S	3.70 mg S	4.25 mg S
12.8 mg Cu in the form of Cu₂S	3.2 mg S	2.87 mg S
14.4 mg Cu in the form of Cu₂S	3.6 mg S	4.2 mg S

On the average, 14 mg of Cu in the form of Cu₂S require 3.5 mg of sulfur; 3.8 mg S was found in the solution.

Since all the sulfur in the solution proved to be combined with Cu, the iron that dissolved must be attributed to Fe sulfates present in the finely ground pyrite, which could not be washed out. The very slight extraction of copper from pyrite should be attributed not to the copper entering into the composition of the pyrite itself, but most probably to an admixture of copper glance in the pyrite, since both minerals very often accompany one another in copper-pyrite deposits. The experiments speak in favor of the fact that a KCN solution does not act upon copper pyrite in a neutral atmosphere; and if, as is accepted by some authors³³), who have studied the structure of mattes, formation of copper pyrite takes place in the latter during cooling, then it should be expected that a KCN solution will not extract from Cu₂S : FeS alloys that portion of copper which enters into the composition of copper pyrite.

Action of a 6% KCN solution on laboratory mattes obtained by fusing Cu₂S with FeS in carbon crucibles

Composition of the matte: Cu—20.25%, Fe—47.25%, and S—32.05%. 1.25 g of ground matte were treated with 125 cc of KCN solution for 24 hours. In 50 cc of filtrate = 0.5 g of matte, Fe and Cu were determined; in the other 50 cc, sulfur.

The results are given below.

	In 1000 mg of matte, mg	Passed into solution, mg	In residue, mg	Extracted, %
Cu	202.5	185.2	17.3	91.46
Fe	472.5	35.3	437.2	.
S	320.5	59.2	261.3

After treatment for 48 hours.

	In 1000 mg of matte, mg	Passed into solution, mg	In residue, mg	Extracted, %
Cu	202.5	187.2	15.3	92.44
Fe	472.5	43.4	429.1
S	320.5	62.5	258.0

Since FeS does not dissolve in KCN, the passage of Fe into solution may be explained by the fact that, when Cu₂S is fused with FeS, a certain compound soluble in KCN is formed. S. Yushkevich assumes that this compound corresponds to the type (Cu₂S)m.(FeS)n. Since in his experiments in the dissolved-

of the matte the ratio of sulfur bound with copper to sulfur bound with iron was close to two, Yushkevich concludes that this compound has the composition \((\mathrm{Cu}_2\mathrm{S})_2 \mathrm{FeS}\). This conclusion of Yushkevich is all the more probable because Bornemann and Schreyer, on the basis of thermal analysis of the system \(\mathrm{Cu}_2\mathrm{S} : \mathrm{FeS}\), likewise established that the sole stable compound in mattes at ordinary temperature is \((\mathrm{Cu}_2\mathrm{S})_2 \mathrm{FeS}\). Calculation of the composition of the portion of the matte dissolved in KCN for experiment I (24 hours of treatment): 35.3 mg Fe bind 20.3 mg S; for Cu there remain 38.9 mg S; the ratio

\[ \frac{S_{\mathrm{Cu}}}{S_{\mathrm{Fe}}} = \frac{38.9}{20.3} = 1.91; \]

for experiment II (48 hours of treatment) an analogous calculation gives \(S_{\mathrm{Cu}} : S_{\mathrm{Fe}} = 1.51\).

An experiment was carried out with laboratory matte of the following composition:
Cu—30.24%, Fe—39.27%, and S—30.49%. The duration of treatment with KCN solution was 6 hours. The following results were obtained:

	In 1000 mg of matte, mg	Passed into solution, mg	In the residue, mg	% extraction
Cu	302.4	275.6	26.8	91.14
Fe	392.7	55.2	337.5
S	304.9	92.5	212.4

For this experiment \(S_{\mathrm{Cu}} : S_{\mathrm{Fe}} = 1.91\).

The experiments performed show that the magnitude of the ratio \(S_{\mathrm{Cu}} : S_{\mathrm{Fe}}\) varies depending on the duration of the action of the KCN solution on the matte and only by chance may turn out to be equal to 2. On the other hand, there are not sufficient grounds for asserting that it is precisely a compound of the type \((\mathrm{Cu}_2\mathrm{S})_m(\mathrm{FeS})_n\), and not of the type \((\mathrm{CuS})_m(\mathrm{FeS})_n\), that passes into the KCN solution. In his calculations of the composition of the dissolving compound, on the sole basis that \(S_{\mathrm{Cu}} : S_{\mathrm{Fe}} = 2\), Yushkevich could with equal right have stopped at the compound \(2\mathrm{CuS}\cdot\mathrm{FeS}\), where the ratio \(S_{\mathrm{Cu}} : S_{\mathrm{Fe}}\) is also equal to 2. Treatment of mattes with \(\mathrm{AgNO}_3\) solution showed that in mattes at ordinary temperature both \(\mathrm{Cu}_2\mathrm{S}\) and \(\mathrm{CuS}\) are present.

Having made an assumption regarding the composition of the dissolving compound, Yushkevich calculates the quantity of metallic copper in the mattes. But here, too, different figures may be obtained depending on which copper sulfide is assumed to be in solution, \(\mathrm{Cu}_2\mathrm{S}\) or \(\mathrm{CuS}\), since the amount of metallic copper is calculated from the difference between the total Cu content in the matte and the copper bound in the form of sulfide.

If we calculate, for the three experiments cited above, the composition of the undissolved portion and assume that in it iron is bound with sulfur in the form of \(\mathrm{FeS}\), while the remainder of the sulfur is bound with copper, we obtain the following results.

For experiment I: 437.2 mg Fe require 250.8 mg S for the formation of \(\mathrm{FeS}\); there remain 10.5 mg S and 17.3 mg Cu.

For experiment II: 429.1 mg Fe require 246.4 mg S for the formation of \(\mathrm{FeS}\); there remain 11.6 mg S and 15.3 mg Cu.

For experiment III: 337.5 mg Fe require 193.8 mg S for the formation of \(\mathrm{FeS}\); there remain 18.6 mg S and 26.8 mg Cu.

With respect to the individual free components of the matte, namely Cu, \(\mathrm{CuS}\), \(\mathrm{Cu}_2\mathrm{S}\), and \(\mathrm{FeS}\), and to KCN solution, we may assume that in the residue from treatment of mattes with KCN solution there are Cu, Fe, and S in combinations not subject to the action of KCN, such as \(\mathrm{FeS}\) and copper pyrite. On this basis let us calculate the rational composition of the residues from experiments I, II, and III.

The amount of copper pyrite is determined from the copper.
The theoretical composition of pyrite: Cu—34.62%, Fe—30.44%, and S—34.94%.

Experiment I. For the residue, composition: Cu—17.3 mg, Fe—437.2 mg, and S—261.3 mg.

34.62 Cu require 34.94 S for formation of CuFeS₂

17.3 ” ” \(x\)

\[ x = 17.47\ \text{mg S}; \]

\[ \begin{aligned} 34.62\ \text{Cu} - 30.44\ \text{Fe}\\ 17.3\quad\ \ ”\quad\ \ y \end{aligned} \left\} \ y = 15.22\ \text{Fe}. \right. \]

After subtracting Cu, Fe, S entering into the composition of the copper pyrite, from the corresponding amounts of these elements in the residue we obtain:

	Residue, mg	CuFeS₂, mg	FeₓSᵧ, mg
Cu	17.3	17.3	—
Fe	437.2	15.22	422.0
S	261.3	17.47	243.83

For the formation of FeS, 422 mg Fe require 242.1 mg S; 243.83 is present. Consequently, the residue from treatment of the matte actually consists of CuFeS₂ and FeS.

In analogous fashion we calculate the rational composition of the residues from experiments II and III.

Residue from experiment II.

	Residue, mg	CuFeS₂, mg	FeₓSᵧ, mg
Cu	15.3	15.3	—
Fe	429.1	13.15	415.95
S	258.0	15.44	242.56

For the formation of FeS, 415.95 mg Fe require 238.7 mg S; 242.56 mg S is present.

For experiment No. 3.

	Residue, mg	CuFeS₂, mg	FeₓSᵧ, mg
Cu	26.8	26.8	—
Fe	337.5	23.6	313.9
S	212.4	27.05	185.35

313.9 mg Fe for the formation of FeS require 180.1 mg S; 185.35 mg S is present in the residue.

The calculations made for the three experiments give, in general, concordant results, showing that the residues from treatment of the mattes with KCN solution consist of CuFeS₂ and FeS, as not reacting with KCN. In view of this, experiments were carried out on treating plant mattes with KCN solution, which gave the following results.

Shaft-furnace mattes.

[[unclear: placzminski/plant]] matte, 10 o’clock in the evening. Duration of treatment 24 hours; 1.25 g matte + 125 cu. cm 5% KCN solution.

	In 1000 mg matte, mg	Passed into solution, mg	Residue, mg	By calculation: CuFeS₂, mg	By calculation: Fe, S, O, mg
Cu	385.0	337.6	47.4	47.4	—
Fe	331.0	89.5	241.5	41.68	199.8
S	236.8	115.2	121.6	47.84	73.76
O	40.2	—	40.2	—	40.2

The residue contains FeS and FeₓOᵧ.

We calculate FeS from the amount of sulfur in the residue:

73.76 S bind 128.6 Fe; there remain 71.2 mgr. Fe and 40.2 mgr. O, which gives ferrite with 36.09% O, i.e., richer in oxygen than Fe₂O₃ (31.0% O). The second sample gave the same results.

Bogoslovsk matte. The treatment conditions are the same.

	In 1000 mgr. of matte, mgr.	Passed into solution, mgr.	Residue, mgr.	By calculation: CuFeS₂, mgr.	By calculation: Fe, S, O, mgr.
Cu	385.7	347.6	38.1	38.1	—
Fe	328.9	113.1	215.8	33.5	182.3
S	247.4	117.1	130.3	38.45	88.42
Zn	7.0	—	7.0	—	—
O	28.7	—	28.7	—	28.7

From the total amount of S in the residue we subtract the sulfur bound with 7 mgr. of zinc, i.e., 3.43 mgr. S; 88.42 mgr. S remain, which bind 154.7 mgr. Fe in the form of FeS. The composition of the ferrite is 28.6 Fe + 28.70 O; consequently, the ferrite contains 50% O.

Kalatin matte. The treatment conditions are the same.

	In 1000 mgr. of matte, mgr.	Passed into solution, mgr.	Residue, mgr.	By calculation: CuFeS₂, mgr.	By calculation: Fe, S, O, mgr.
Cu	145.5	134.8	10.7	10.7	—
Fe	534.0	38.6	495.4	9.41	485.99
S	235.8	44.3	191.5	10.8	180.7
O	74.7	—	74.7	—	74.7

180.7 mgr. S in the residue require 314.9 mgr. Fe for the formation of FeS; the composition of the ferrite is 171.09 mgr. Fe and 74.7 mgr. O; consequently, the ferrite contains 30.39% O. The same results were obtained for the second sample.

Matte of the Ulen plant. The treatment conditions are the same.

	In 1000 mgr. of matte, mgr.	Passed into solution, mgr.	Residue, mgr.	By calculation: CuFeS₂, mgr.	By calculation: Fe, S, O, mgr.
Cu	410.5	379.6	30.9	30.9	—
Fe	306.6	105.0	201.6	27.17	174.4
S	246.5	129.5	117.0	31.19	85.8
O	28.4	—	28.4	—	28.4

85.8 mgr. S in the residue bind 149.5 mgr. Fe (FeS); the composition of the ferrite is: 24.9 mgr. Fe and 28.4 mgr. O; consequently, the ferrite contains more than 50% O.

Mattes from reverberatory smelting.

Dzansul matte. The treatment conditions are the same.

	In 1000 mgr. of matte, mgr.	Passed into solution, mgr.	Residue, mgr.	By calculation: CuFeS₂, mgr.	By calculation: Fe, S, O, mgr.
Cu	377.6	282.0	95.6	95.6	—
Fe	318.7	78.4	240.3	84.06	156.24
S	260.6	95.5	165.1	96.5	68.60
O	38.0	—	38.0	—	38.0

68.6 mg of S in the residue bind 119.6 mg of Fe; composition of the ferrite: 156.24 — 119.6 = 36.64 mg Fe and 38.0 mg O; the ferrite contains about 51% O.

Matte of the Tanalyk Plant No. 12.

	In 1000 mg of matte, mg	Passed into solution, mg	Residue, mg	By calculation, Cu Fe S₂	By calculation, Fe, S, O
Cu	122.5	88.8	33.7	33.7	—
Fe	482.5	30.8	451.7	29.63	422.07
S	256.7	30.07	226.63	34.01	171.06
Pb	30.0	—	30.0	—	—
Zn	34.5	—	34.5	—	—
O	77.5	—	77.5	—	77.5

Pb and Zn require, for the formation of PbS and ZnS, a total of 21.56 mg S; for the formation of FeS there remain 226.63 — (34.01 + 21.56) = 171.06 mg S, which bind 298.2 mg Fe; composition of the ferrite: 123.87 mg Fe + 77.5 mg O, which gives a ferrite with 38.5%.

Thus, calculation of the composition of ferrites in the residues from treatment of mattes with KCN solution leads to ferrites containing more oxygen than is contained in iron oxides; we know of no iron oxides richer in oxygen than Fe₂O₃, and the calculations made compel us to suppose that, upon cooling, iron sulfides richer in sulfur than FeS are formed in mattes. Such a supposition has been expressed more than once in the literature. The composition and quantity of these sulfides in mattes remain an unsolved problem, for a reagent for the identification and separation of iron sulfides in their alloys has not yet been found.

The cause of the formation of such iron sulfides may be, for example, the reaction:

\[ 2 \ \mathrm{Cu_2S} + 2 \ \mathrm{FeS} \rightleftarrows \mathrm{Cu_2S} . \ \mathrm{Fe_2S_3} + 2 \ \mathrm{Cu}. \]

This reaction leads, in mattes during their cooling, to the formation of a stable compound—copper pyrite—and to the separation, at the same time, of metallic copper. Studying the microstructure of Cu₂S:FeS alloys, Baikov and Trutnev came to the conclusion that the brass-yellow segregations observed in solidified mattes correspond in their color and hardness to copper pyrite, and that in alloys containing 40% Cu₂S and 60% FeS, i.e., close in composition to copper pyrite, the amount of the brass-yellow mass reaches a maximum. In the opinion of the authors mentioned, the process of formation of copper pyrite in mattes may be expressed by the reaction:

\[ \mathrm{Cu_2S} + \mathrm{FeS} \rightleftarrows \mathrm{CuS} . \ \mathrm{FeS} + \mathrm{Cu}. \]

The formation of copper pyrite during the freezing of molten mattes is confirmed, among other things, by the following observation from the practice of smelting roasted ores in reverberatory furnaces. In Anaconda, in breakouts of the hearths of reverberatory furnaces that had been in operation for a long time, large accumulations of copper pyrite of secondary origin were found, i.e., separated from matte that had penetrated into cracks of the hearth and solidified there.

A rational analysis by means of KCN solution would make it possible to calculate the composition of ferrites in mattes if this solution were able to extract from mattes all the Cu, both metallic and bound in the form of Cu₂S and CuS, and if it were possible to establish the formula of the iron sulfide present in the residues. But the insoluble residues contain from 10 to 25% of the Cu that was in the matte. Under the assumption that all copper in the residues is present in the form of CuFeS₂, and all oxygen is bound with iron in the form of Fe₃O₄, calculation of the composition of the residues shows that they contain compounds richer in sulfur than FeS. Thus, under these assumptions, in the residue

of the Pyshminsk matte the iron sulfide should contain 43.83% S, in the residue of the Bogoslovsk matte—45.22% S, and similar results are given by calculations for the remaining mattes.

In view of the foregoing, attempts to elucidate the composition of ferrites of iron oxide in mattes by means of their rational analysis, as not promising positive results, were abandoned.

The experiments carried out show that a KCN solution also cannot serve as a reagent for the indicated purpose, and the assumptions made by S. Yushkevich are not sufficiently substantiated.

All plant mattes are, to a greater or lesser degree, magnetic, even in the weak field of an ordinary magnet. This property of mattes is explained by the presence in them of magnetic components. The principal of these components, namely, Cu, Cu₂S, CuS and FeS, are not themselves magnetic.

The action of mattes on a CuSO₄ solution showed that mattes contain metallic Fe in the form of a free component or a member of the system FeS : Fe. Although CuSO₄ solution, as the investigation showed, cannot serve as a reagent for the exact quantitative determination of iron in its alloys with sulfides of iron and copper, nevertheless, without making an error, it may be accepted that the amount of metallic Fe in mattes is, in general, not large and does not exceed, in mattes of pyritic smelting, 0.5–0.8%, in mattes of semi-pyritic smelting 1–2.6%, and in mattes of reducing smelting 3.7–5.3%. This iron, being magnetic, may account for the magnetism of mattes. Another component of mattes is the iron oxide ferrites dissolved in them, which are strongly magnetic. Further, according to Bornemann⁽³⁴⁾, in mattes at ordinary temperature the compounds \((\mathrm{Cu}_2\mathrm{S})_2 \mathrm{FeS}\) and \((\mathrm{Cu}_2\mathrm{S})_3 (\mathrm{FeS})_2\) are present. Finally, according to Borchers⁽³⁵⁾, the separation of hair-like copper in mattes may be the result of the following process:

\[ 2 \left[ (\mathrm{Cu}_2\mathrm{S})_2 \mathrm{FeS} \right] \Longrightarrow (\mathrm{Cu}_2\mathrm{S})_3 \mathrm{Fe}_2\mathrm{S}_3 + 2\mathrm{Cu}. \]

Consequently, a new component, \(\mathrm{Fe}_2\mathrm{S}_3\), or, in general, iron sulfides richer in sulfur than FeS, in the free state or in combination with copper sulfides, is assumed in the composition of mattes. As established by Allen—Crenshaw, these iron sulfides are solid solutions of S in FeS; unlike FeS, they are strongly magnetic and form a series of so-called pyrrhotites. The possibility is therefore not excluded that pyrrhotites, being present in mattes, determine the magnetism of the latter.

Thus, the magnetism of mattes may be due to the presence in them of: a) metallic Fe or the system FeS : Fe, b) compounds \((\mathrm{Cu}_2\mathrm{S})_2 \mathrm{FeS}\) or \((\mathrm{Cu}_2\mathrm{S})_3 (\mathrm{FeS})_2\), c) iron sulfides richer in sulfur than FeS, e.g. \(\mathrm{Fe}_2\mathrm{S}_3\), in the free state or in combination with Cu₂S, and, finally, d) ferrites of iron oxide \((\mathrm{FeO}) \times (\mathrm{Fe}_2\mathrm{O}_3)y\). To clarify the question, two series of mattes were investigated, one with an excess of sulfur relative to the amount necessary to bind Fe and Cu as FeS and Cu₂S, the other with the theoretical amount of sulfur.

The starting materials for obtaining mattes of the first series were purified Karabash pyrite, containing 46.11% Fe and 52.35% S, and Cu₂S of composition Cu—78.30% and S—21.70%. FeS₂ and Cu₂S sulfides, calculated for mattes of a definite composition, were fused in a charcoal crucible under charcoal, stirred, and slowly cooled in the crucible. In none of the mattes obtained was separation of metallic copper observed.

The mattes had the following composition.

No.	Cu, %	Fe, %	S, %	Theoretical S content from $\mathrm{Cu_2S + FeS}$, %	Excess sulfur, %
1	14.84	50.00	35.16	32.45	2.71
2	19.64	46.51	33.85	31.63	2.22
3	29.00	39.00	32.00	29.69	2.31
4	39.16	30.97	29.87	27.68	2.19

From the analyses it is evident that all the mattes contained an excess of sulfur relative to the theory. This excess may be attributed to $\mathrm{Cu_2S}$, in accordance with the data of Posnjak and Allen, and it may be assumed that part of the $\mathrm{Cu_2S}$ passed into CuS. This is indicated, among other things, by the results of rational analysis of analogous mattes by means of an $\mathrm{AgNO_3}$ solution. It is also possible to assume that part of the excess sulfur is dissolved in FeS, forming with the latter solid solutions of the pyrrhotite type. And this assumption is not without basis, since the melting temperature of the mattes did not exceed 900–950°, and it is quite possible that under these conditions the dissociation of $\mathrm{FeS_2}$ did not proceed as far as FeS; moreover, the dissociation pressure of the sulfides $\mathrm{Fe_nS_{n+1}}$ in their alloys with other sulfides must be considerably lower than in the free state. In obtaining mattes with an excess of sulfur, conditions existed that were favorable for the formation of iron sulfides richer in sulfur than FeS; nevertheless, all mattes of this series proved to be completely nonmagnetic. The question arose whether the mattes of this series could contain, in the free state, magnetic iron sulfides which would account for the magnetism of the mattes themselves. In this connection, of course, it was not overlooked that the magnetism of substances is not an additive property, but a constitutive one, as is known from the work of Heusler^36) on non-iron magnetic alloys of nonmagnetic metals, Kohlmeyer^37) and others on ferrites of calcium, lead, etc. Since in the literature I could not find data on the dependence of the magnetism of iron sulfides on the sulfur content in them, an investigation was carried out whose aim was to establish the limiting composition of iron sulfides at which their magnetism appears or disappears.

Pyrite and pure FeS are nonmagnetic. Iron sulfides intermediate in sulfur content could be prepared by the dissociation of pyrite at various temperatures.

Purified Karabash pyrite (46.46% Fe and 53.20% S) was heated in an electric furnace in a stream of dry and pure nitrogen. The purity of the nitrogen was checked by heating metallic copper in it at 980–1000°; 1.494 g of copper shavings, when ignited for one hour, showed an increase of 0.3 mg.

Preliminary experiments established that pyrite, when heated in a stream of nitrogen to 500°, does not lose weight and does not change. The first noticeable decomposition is observed at 500°; 1911.4 mg of pyrite, after heating for one hour, lost 5.5 mg of S. The product did not change in appearance and was completely nonmagnetic. The results of the experiments are summarized in Table 22^38).

From the results of the experiments the following conclusions may be drawn. Dissociation of pure pyrite in a stream of pure nitrogen begins practically at 500°; the onset of magnetism in the products of pyrite dissociation is observed at 600°, after the pyrite has lost about 5.9% of the initial sulfur content in it, i.e., the sulfide containing 48.33% Fe and 51.67% S is magnetic. Thereafter, as the dissociation temperature increases, the dissociation also increases, and with it the magnetism also increases. At 700° the sulfide acquires an almost constant composition. In six experiments out of seven, the composition of the sulfide at 700° was 61.3% Fe and 38.70% S. At this point the sulfide is still magnetic. The magnetism of the sulfide disappears when the sulfur content in it reaches 38.5%, and the iron content 61.5%; such a composition must be recognized as limiting in the sense of magnetism.

Table 22. Magnetism of iron sulfides.

Test No.	Charge FeS₂, mg	Sulfur content, mg	Heating, min	Heating, °C	Sulfur loss, mg	Sulfur loss, %	Residue composition, % Fe	Residue composition, % S	Magnetism of the dissociation product
1	1911.4	1017	60	500	5.50	0.54	46.96	53.04	Non-magnetic.
2	1500	798	35	600	47.20	5.92	48.33	51.67	Magnetic.
3	1308	693	60	600	66.20	9.55	49.30	50.70	Same.
4	1304.20	693.72	60	600	68.40	9.86	49.41	50.59	Same.
5	”	”	30	650	266	38.35	58.80	41.21	Strongly magnetic.
6	”	”	”	700	307.60	44.34	61.26	38.74	Same.
7	”	”	”	”	309.80	44.66	61.40	38.60	Same.
8	”	”	40	700	309.20	44.57	61.36	38.64	Same.
9	”	”	60	”	308.20	44.43	61.30	38.70	Same.
10	1400	744.80	60	700	328.0	44.04	61.12	38.88	Same.
11	”	”	80	”	331.20	44.47	61.30	38.70	Same.
12	”	”	210	”	339.80	45.62	61.80	38.20	Non-magnetic.
13	1304.20	693.72	120	740	311.5	44.90	61.50	38.50	Non-magnetic.
14	1400	744.80	30	750	334.0	44.85	61.47	38.53	Same.
15	1304.20	693.72	30	800	318	45.84	61.90	38.10	Same.
16	”	”	”	900	326.6	47.08	62.45	37.55	Same.
17	”	”	60	1100	336.20	48.47	63.07	36.93	Same.
18	”	”	”	1100	336.60	48.52	63.09	36.91	Same.
			Theoretical composition			FeS	63.55	36.45

With this composition the formula of the sulfide is close to the expression \(\mathrm{Fe}_{11.06}\mathrm{S}_{12.01}\). Consequently, in the Fe:S system all sulfides containing from 38.5 to 51.7% S are magnetic; outside these limits magnetism disappears*). For comparison with the dissociation products of pyrite, the purest crystals of strongly magnetic pyrrhotite from Olkhovka (Minusinsk district) were investigated; they had the following composition:

\[ \begin{aligned} \mathrm{Fe} &— 59.74\% \\ \mathrm{S} &— 38.47\% \\ \mathrm{SiO_2, Cu} &— 1.79\% \end{aligned} \qquad \begin{array}{c} \text{when recalculated} \\ \text{to } \mathrm{Fe}+\mathrm{S}=100 \end{array} \qquad \begin{aligned} \text{we obtain}\quad \mathrm{Fe} &— 60.83\% \\ \mathrm{S} &— 39.17\% \end{aligned} \]

According to the analysis, the formula of pyrrhotite is: \(\mathrm{Fe}_{10.9}\mathrm{S}_{12.2}\).

*) Although natural pyrrhotites are magnetic, non-magnetic ones are also encountered, though rarely. See Mining and Metallurgy, June 1923; 283.

Comparing the composition of pyrrhotite and of the products of dissociation of pyrite at \(700^\circ\), we see that both substances have a very similar composition:

FeS\(_2\) + \(700^\circ\)	Pyrrhotite
Fe — 61.3%	60.83%
S — 38.7 »	39.17 »
100—	100—

Samples of pyrrhotite were heated in a stream of nitrogen at \(700^\circ\)—\(750^\circ\) for 30 minutes; the experiments gave the results presented in Table 23.

Table 23. Magnetic character of the products of dissociation of pyrrhotite.

Sample weight, mg	Sulfur content, mg	Heating, min	Heating, °C	Sulfur loss, mg	Sulfur loss, %	Composition of residue, % Fe	Composition of residue, % S	Magnetic character
1400	538.60	30	700	23.20	4.31	61.85	38.15	Magnetic.
1400	538.60	30	750	29.0	5.38	62.14	37.86	Non-magnetic.

The magnetic character of pyrrhotite disappears when its sulfur content is reduced to 38%. Consequently, the following dissociation products are non-magnetic:

	pyrite	pyrrhotite
at content	S—38.5%	38.0%
	Fe—61.6 »	62.0 »

Thus, the limiting composition of non-magnetic sulfides in the Fe:S system is obtained as one and the same, whichever of the natural sulfides—pyrite or pyrrhotite—is subjected to dissociation. Both sulfides lose their magnetic character within the temperature range \(700\)—\(750^\circ\) and acquire an almost identical composition: Fe—61.5—62% and S—38—38.5%, to which the formula Fe\(_{11}\)S\(_{12}\) corresponds closely.

For pyrite, pyrrhotite, and the products of their dissociation, the relative magnetic permeability was determined by the method described below in the chapter on the magnetic permeability of mattes. For the present, we give the data obtained in these measurements. Charge of substance 0.95 g, current strength in the circuit 0.99 amp. Resistance in the external circuit 2000 ohms.

Substance	Temperature, °C	Sample no.	Deflection of the ballistic galvanometer
Pyrite	—	0	0
»	600°	3	0.5
»	650	5	3.0
»	700	6	2.5
»	750	14	0
»	800	15	0
»	1100	17	0
Pyrrhotite	—	—	5.0
»	700	—	1.0
»	750	—	0.0

Consequently, the product of dissociation of pyrite at 650° possesses the greatest magnetic permeability.

Let us now turn to the mattes of series I and calculate the composition of the iron sulfide contained in them, assuming that all the Cu in them is present in the form of Cu₂S and CuS.

Copper in the form of Cu₂S.

Matte composition	Bound S	Fe + S	Composition Feₓ + Sᵧ, %
Cu . . 14.84%	3.74	—	—
Fe . . 50.0 »	—	50.0	61.41
S . . 35.16 »	—	31.42	38.59

Copper in the form of CuS; under this assumption, for the iron sulfide sought we obtain the composition:

\[ \begin{array}{rccc} \mathrm{Fe}\ .\ .\ . & 50 & & 64.37\% \\ \mathrm{S}\ .\ .\ . & 27.68 & & 35.63\,» \end{array} \]
\[ \overline{\phantom{000000}} \quad 100— \]

By its sulfur content, the calculated iron sulfide lies in the region of nonmagnetic sulfides.

Calculation for matte with 19.64% Cu.

All the copper is in the form of Cu₂S.

Matte composition	Bound S	Fe + S	Composition Feₓ + Sᵧ
Cu . . 19.64%	4.95	—	—
Fe . . 46.51 »	—	46.51	61.68%
S . . 33.85 »	—	28.9	38.32 »

Assuming that all the copper in the matte is bound in the form of CuS, for iron sulfide we obtain the composition:

\[ \begin{array}{rcc} \mathrm{Fe} — 46.51 — & 65.96\% \\ \mathrm{S} — 24.00 — & 34.04\,» \end{array} \]
\[ \overline{\phantom{000000}} \quad 100— \]

Calculations carried out in analogous fashion for the remaining mattes of series I show that the iron sulfides contained in them, by sulfur content, lie in the region of nonmagnetic iron sulfides, since all of them, even under the assumption that all the copper in the mattes is present in the form of Cu₂S, contain 38.59 and less % S.

Then a second series of mattes was prepared from Cu₂S and FeS; FeS was obtained by remelting pyrite in a carbon crucible under coal at a temperature of about 1200°. The product contained 63.08% Fe and 36.0% S, instead of the theoretical 63.77% Fe and 36.23% S; the Cu₂S preparation was the same as for the mattes of series I.

Below are given the analyses of the mattes of series II.

	% Cu	% Fe	% S (calculated)	% S (by analysis)
1.	14.85	51.86	33.49	33.17
2.	20.25	47.25	32.20	32.05
3.	30.00	39.65	30.30	30.05
4.	39.65	31.54	28.09	28.17

All mattes contained the theoretical amount of sulfur required to bind Cu in the form of Cu₂S and Fe in the form of FeS. Nevertheless, in all mattes a considerable separation of metallic copper was observed, in the form of globules on the surface and of copper moss on the fracture of the mattes. The separation of copper was the more abundant the poorer in copper the corresponding matte was. During grinding of the mattes the copper globules were partly lost, partly flattened out and did not pass through the sieve, owing to which the difference in copper content for parallel samples reached 0.5%. Thus, for example, for matte No. 1 Cu = 14.55 and 15.10%, for matte No. 2 Cu = 19.95 and 20.20%.

In preparing mattes of the second series, all conditions were present that were favorable for the formation in them of chemical compounds between Cu₂S and FeS, as indicated by Bornemann and others, but all the mattes proved to be completely nonmagnetic. Thus, investigation of mattes of both series leads us to the following conclusions: first, the formation in mattes of magnetic iron sulfides as free components is impossible, since on fusion of Cu₂S with FeS₂ the pyrite dissociates and acquires a composition lying in the region of nonmagnetic iron sulfides; second, neither the individual components of mattes in the free state, such as Cu₂S, CuS, FeS, nor their alloys, which might correspond to chemical compounds, are magnetic and are not the cause of the magnetism of mattes.

Let us now proceed to clarify the role of metallic iron as an element capable of determining the magnetism of mattes.

As measurements of the relative magnetic permeability of FeS : Fe alloys showed, all of them proved to be magnetic, and the degree of their magnetism was almost proportional to the amount of free metallic iron contained in the alloys.

In laboratory mattes of the first and second series the content of metallic iron was determined by means of a CuSO₄ solution. The determinations gave the following results:

No. of matte	Mattes of series I, % metallic iron	Mattes of series II, % metallic iron
1	not determined	0.61
2	0.61	0.63
3	0.71	1.04
4	1.06	1.67

As is evident from these data, the amount of extractable metallic iron from the mattes increases with increasing copper content in them. If it is assumed that the CuSO₄ solution extracts not all, but only part, of the free metallic iron, it still turns out that the amount of the latter varies within the same limits as in plant mattes treated by the same method.

Since measurements of magnetic permeability showed that the laboratory mattes are completely nonmagnetic, it may be concluded from this that metallic iron, in the quantities in which it is present in the plant mattes investigated, cannot account for the magnetism of the latter.

Here it is appropriate to ask what the causes are for the formation, in laboratory mattes, of metallic iron extractable by CuSO₄ solution. To explain this phenomenon the following assumptions are possible. First, additional dissociation of FeS; second, assimilation by Cu₂S : FeS alloys of carbon, forming with FeS the ternary alloy Fe : S : C, decomposed by CuSO₄; and third, certain processes taking place during smelting and during freezing of the Cu₂S : FeS alloys. Additional dissociation of FeS is impossible within the t-r melting limits of the Cu₂S : FeS systems lying below 1000°; moreover, in mattes of the first series

there is an excess of sulfur beyond that required to bind all the Cu into Cu₂S and all the Fe into FeS and, despite this, a certain amount of metallic Fe separated out. The assimilation of carbon by mattes cannot be recognized as the cause of the phenomenon under consideration, since, first, the solubility of carbon even in FeS is negligible, and iron sulfide remelted at 1250–1300° gave up to a solution of CuSO₄ only 0.2% Fe; and, second, in mattes poor in copper and rich in iron there should have been more free metallic iron than in mattes rich in copper; in other words, relations opposite to those obtained should have been observed.

Finally, speaking of the processes between the sulfides of Fe and Cu during their melting and solidification, we have in mind the following possible reactions:

1) During the solidification of mattes the reaction

\[ \mathrm{Cu_2S + FeS \rightleftarrows 2CuS + Fe} \]

is possible; during melting the reaction will take the reverse direction, in view of the ease with which CuS dissociates:

\[ \mathrm{4CuS \rightleftarrows 2Cu_2S + S_2} \]

\[ \mathrm{S_2 + 2Fe \rightleftarrows 2FeS}. \]

According to the investigations of Preuner\(^{39}\) and Brockmöller, the total vapor pressure of sulfur during the dissociation of CuS reaches 250 mm already at 475°.

The following process is also possible:

\[ \mathrm{Cu_2S \cdot FeS \rightleftarrows Cu \cdot CuS \cdot FeS} \]

\[ \mathrm{2Cu + FeS \rightleftarrows Cu_2S + Fe}. \]

The tendency of Cu₂S to pass into CuS, even at ordinary temperature, is very strong, as is evident from the interesting observations of I. Frankel\(^{40}\) on the formation of artificial covellite from chalcocite when the latter is in contact with sulfur-containing substances, such as, for example, rubber tape or elemental sulfur.

From all that has been set forth it follows that the substances imparting magnetism to mattes are iron oxides dissolved in them. According to the opinion prevailing in the metallurgical literature, such oxides can only be magnetite, while in the opinion of Borchers\(^{41}\) they are ferrites, the composition of which may vary between meta- and orthoferrite of ferrous oxide, i.e. FeO·Fe₂O₃ and 2FeO·Fe₂O₃.

If the ferrites in all mattes had the same composition, then the relative magnetic permeability of mattes possessing, moreover, a very similar specific gravity would be proportional to the quantity of active magnetic substance, or, what is the same, to the quantity of oxygen in the mattes. Therefore it seemed that measurement of the magnetic permeability of mattes could give indications at least for deciding the question whether the ferrites in the mattes have the same composition or whether they belong to the series of magnetic alloys FeO:Fe₂O₃. These alloys were studied by Kohlmeyer, Hilpert, and Sosman-Hostetter.

In what follows a brief review is given of the conclusions reached by the investigators mentioned. Kohlmeyer\(^{42}\) melted pure iron oxide in a platinum crucible in an atmosphere of air and, in doing so, took curves of melting and solidification of Fe₂O₃. On the basis of the latter, and also of analyses of the products of melting, Kohlmeyer assumes that FeO, formed during the dissociation of Fe₂O₃, gives with the latter the following series of ferrites, which are chemical compounds unstable on melting. At 1370° FeO·3Fe₂O₃ is formed; at 1470°—3FeO·5Fe₂O₃; at 1520°—3Fe₂O₃·4Fe₂O₃; above 1520° all ferrites acquire the composition of magnetite, FeO·Fe₂O₃. A ferrite with 10% FeO (20 mol.% FeO + 80 mol.% Fe₂O₃) was still strongly magnetic.

Hilpert\(^{43}\) prepared a series of ferrites of ferrous oxide by subjecting pure Fe₂O₃ to reduction by means of mixtures of hydrogen and water vapor. By varying

temperature and composition of the gas mixture, one can obtain a mixture in which the partial pressure of oxygen will correspond to any stage of iron oxidation from FeO to $\mathrm{Fe_3O_4}$. Studying the magnetic properties of the iron oxides obtained in this way, Hilpert came to the following conclusions.

1. Since FeO and $\mathrm{Fe_2O_3}$ are not magnetic, the strong magnetism of the intermediate iron oxides is explained by the formation of ferrous ferrites, and only those oxides are ferromagnetic in which ferric oxide acts as the acid radical.

2. The magnetic permeability of ferrites depends on the amount of FeO in them. The maximum magnetic permeability is observed at 31.03% FeO, which corresponds to the composition of magnetite, $\mathrm{FeO \cdot Fe_2O_3}$. Up to 66% FeO the magnetism remains constant; thereafter, with an increase in % FeO, it decreases and at 82% FeO disappears. The phenomena observed in the reduction of ferric oxide, and magnetic measurements of oxide powders, make it possible to conclude that solid solutions exist among the iron oxides.

Sosman and Hostetter^44), on the basis of measuring the dissociation pressure of $\mathrm{Fe_2O_3}$, came to the conclusion that the dissociation of $\mathrm{Fe_2O_3}$ leads to the formation of a series of solid solutions $\mathrm{Fe_3O_4 : Fe_2O_3}$. The dissociation pressure of $\mathrm{Fe_2O_3}$ decreases continuously from $\mathrm{Fe_2O_3}$ to $\mathrm{Fe_3O_4}$. In a subsequent investigation the same authors^45), having determined FeO and measured the magnetism of natural and artificial iron oxides, arrived at the following conclusions.

1. Artificial iron oxides obtained at 1100 and 1200° consist of a solid solution of $\mathrm{Fe_3O_4}$ in $\mathrm{Fe_2O_3}$. Their relative magnetic permeability is approximately proportional to the percentage of FeO from $\mathrm{Fe_2O_3}$ to $\mathrm{Fe_3O_4}$. This proportionality is preserved only in the case where FeO is in solid solution, and not in some other form.

From a comparison of the results of the investigations of Kohlmeyer, Hilpert, Sosman, and Hostetter, we can draw the following conclusions. Kohlmeyer accepts the existence of ferrites of the type $(\mathrm{FeO})x(\mathrm{Fe_2O_3})y$, allowing the possibility of the formation of several chemical compounds between FeO and $\mathrm{Fe_2O_3}$. Analogous compounds were described by Kaufmann^46) $(2\mathrm{FeO}\cdot 3\mathrm{Fe_2O_3})$, Gregory $(2\mathrm{FeO}\cdot \mathrm{Fe_2O_3})$, Döbereiner $(3\mathrm{FeO}\cdot \mathrm{Fe_2O_3})$, Hauser $(4\mathrm{FeO}\cdot \mathrm{Fe_2O_3})$, and others^47). Hilpert believes that solid solutions of FeO in $\mathrm{Fe_2O_3}$ predominate among the ferrites; Sosman and Hostetter assert that in the dissociation of $\mathrm{Fe_2O_3}$ only a continuous series of solid solutions of $\mathrm{Fe_3O_4}$ in $\mathrm{Fe_2O_3}$ is formed, and deny the existence of chemical compounds of FeO with $\mathrm{Fe_2O_3}$. Ferrites containing from 10 to 80% FeO are strongly magnetic, and only those of them are ferromagnetic in which $\mathrm{Fe_2O_3}$ acts as the acid radical or in which FeO is present in solid solution. In this case the magnetic permeability is proportional to the content of ferrous oxide.

Measurement of the magnetic permeability of mattes.

As Hilpert and Beyer^48) have pointed out, in measuring magnetic permeability one can speak only of relative, and not absolute, permeability, since for ferromagnetic powdered substances the determination of the degree of magnetization and of magnetic permeability in absolute values is theoretically impossible. Even at the highest pressures these powders cannot be compressed so that their permeability would even approximately reach the magnetic permeability of a massive body. Air interlayers always remain in the substance, causing scattering of the lines of force, and the substance remains magnetically inhomogeneous. For paramagnetic substances, such as Mohr’s salt, ferric oxide, etc., the weight and form of the substance and its distribution in the magnetic field have very little effect on the magnetic attraction, as was shown by the investigations of Sosman and Hostetter^49).

But as the permeability of a substance increases, its attraction is influenced, besides the amount of magnetic component contained in it, by a whole series of other factors, such as the shape of the particles, their size, the degree of their proximity to one another, their distribution in the magnetic field, etc.

The magnetic permeability of mattes was measured with the apparatus proposed by Kohlmeyer^50), which was also used by Selivanov^51). The apparatus consists of a simple electromagnet with separate pole pieces, between which, on a thin Pt wire (about 0.02 mm), a thin-walled glass tube 4 mm in diameter and 10 mm long is suspended on a soldered hook. Into the tube was placed 380–399 mg of matte ground to 100 mesh, and the powder was lightly tamped with a glass pestle. The tube was suspended between the pole pieces of the electromagnet, set with their flat sides parallel to one another at a distance of 40 mm. This distance was kept unchanged in all experiments. At a certain strength of current passing through the winding of the electromagnet and measured with an ammeter with divisions of 0.1 amp., the tube is attracted to one of the poles. By regulating the current strength, it is possible to mark fairly accurately the moment at which the attraction begins and the moment at which the tube falls away from the pole piece. Obviously, the relative magnetic permeability will be inversely proportional to the current strength in the winding of the electromagnet required for the beginning of attraction of the tube to the pole piece of the electromagnet. In this way the figures given in Table 24 were obtained, as averages of several determinations, characterizing the relative magnetic permeability of mattes.

Table 24. Magnetic permeability of plant mattes.

Matte	Beginning of attraction at amp.	Oxygen, %	Metallic iron, %
1. Kalata	5.4	7.45	0.55
2. Karabash I	6.7	6.52	0.69
3. Karabash III	7.5	5.06	0.86
4. Pyshma	5.7	6.2	0.27
5. Polevskoi	6.9	3.15	1.13
6. Bogoslovsk	14.0	2.87	2.56
7. Uley	12.80	2.84	3.70
8. Yulia	11.2	2.13	5.33

The table also gives the content of oxygen and metallic iron in the mattes. From the data of the table it is evident that there is no proportionality between the magnetic character of the mattes and their oxygen content. Thus, the magnetic permeability of the Karabash I and Polevskoi mattes is almost the same, while the oxygen contents in them are in the ratio 6.52 : 3.15. With almost identical magnetic permeability of the Kalata and Pyshma mattes, the amounts of oxygen in these mattes are in the ratio 7.45 : 6.2. Metallic iron in mattes has some influence on magnetic permeability, but this

the influence is very small, as may be concluded from the data characterizing the magnetic permeability of FeS : Fe alloys, for which measurements were made under the same conditions as in the case of mattes. The results of these measurements are given in Table 25.

Table 25. Magnetic permeability of FeS : Fe alloys.

Alloy composition, % Fe	Alloy composition, % FeS	Beginning of attraction, amp.	Alloy composition, % Fe	Alloy composition, % FeS	Beginning of attraction, amp.
1.58	98.20	30	8.80	90.83	6.2
2.43	97.44	21	13.33	86.19	5.0
4.37	95.16	13.5	18.47	80.82	4.2
5.30	94.55	—	23.05	76.27	3.8

Consequently, up to 4.37% metallic iron does not exert a noticeable influence on the magnetic permeability of FeS : Fe alloys, and therefore of mattes as well. According to magnetic permeability, the mattes investigated may be divided into two groups: strongly magnetic Nos. 1—5 and weakly magnetic Nos. 6—8. The mattes of the first group are products of pyrite and semipyrite smelting, i.e., of a strongly oxidizing process; the mattes of the second group are products of reducing smelting. This difference in the origin of the mattes accounts for the difference in the composition and degree of magnetism of the magnetic component in the mattes.

All magnetic measurements were made with mattes crushed to pass through 100—200 mesh, since, owing to their brittleness, it was difficult to prepare from them solid and whole specimens of definite size and shape. In ferromagnetic powdered substances, such as the investigated matte specimens were, attraction in a given magnetic field is affected, in addition to the amount of magnetic component contained in them, by the size and shape of the grains, their mutual spacing, their distribution in the given field, and other factors. Nevertheless, under identical conditions, well comparable results can be obtained in measurements. And if one takes into account that mattes are alloys and not mechanical mixtures of magnetic and nonmagnetic substances, and that the magnetic components in mattes crushed to a definite grain size are distributed almost uniformly throughout the whole mass of the matte substance, then, provided the magnetic component is identical, mattes with the same content of the latter should have the same magnetic permeability. Since the magnetic components in mattes are iron oxides, the amount of which is determined by the oxygen content in the mattes, then, with the same oxygen content, different mattes should have the same or nearly the same magnetic permeability. The measurements made show that, at the same oxygen content, mattes possess very different magnetic permeability; and this is possible if the composition of the magnetic component in the mattes is not the same. Consequently, the carriers of the magnetic properties of mattes are not exclusively magnetite, but iron ferrites, possessing different composition and magnetic permeability depending on the origin of the mattes.

In determining the relative magnetic permeability by means of an electromagnet, it was observed that, in repeated measurements, the current strength required for the onset of attraction of the tube with one and the same substance yielded varying values—sometimes larger, sometimes smaller—depending on the length of time the substance remained in the field of the magnet. This may be explained by the particles of the substance becoming permanently magnetized, which changed the current strength required for the onset of attraction. Thus, the force of attraction in a given field depended on the preceding magnetic treatment of the substance. Although the order of magnitude for a number of the mattes studied remained more or less constant, for the reason stated the method must be regarded as very approximate and not very reliable.

In view of this, another method was used for determining the relative magnetic permeability, based on measuring the magnetic induction produced by a ferromagnetic substance in a given magnetic field. Fig. 2 shows the arrangement of the apparatus for the measurements. The primary coil 7 was connected into the circuit through switch 6, two rheostats 3, 4, and ammeter 5 for one ampere, with divisions of 0.01 ampere. The source of current consisted of four storage cells 1 connected in series. The primary coil was 60 mm in diameter and 42 mm long. Inside it was the secondary coil 8, 55 mm long and 16 mm in diameter, which was connected with a ballistic mirror galvanometer 12 through a plug rheostat 11 of 10,000 ohms. In this external circuit, during the measurements, 2,000 ohms were included. When the current was closed by means of switch 2, and when the secondary coil was connected, an induction current arose in the latter, the strength of which was recorded by the ballistic galvanometer; the readings of the galvanometer were read off on the scale with the aid of telescope 13.

If a ferromagnetic substance is introduced as a core into the secondary coil, the induction current arising in the coil will receive a certain increase in intensity, proportional to the magnetism of the substance under investigation. Subtracting from the galvanometer reading obtained the magnitude of its deflection for air, we obtain the increment of deflection caused by the given substance and proportional to the magnetism of the latter.

Finely ground matte powder was placed in a glass tube 9, sealed at one end, and tamped with a glass pestle. The length of the tube was equal to the length of the secondary coil. For each series of measurements, one and the same quantity of matte was taken. The reading was made three and four times in both directions of the scale, and mean figures were taken if the deflections differed by no more than one division. Since, with prolonged passage of current through the external coil, the latter gradually heated up, thereby increasing the resistance in the external circuit and, consequently, changing the strength of the induction current in the secondary circuit, the measurements were made only after the entire system had reached a state of thermal equilibrium. Special attention was paid to the constancy of the current strength in the primary circuit. Two rheostats served for regulating the current, of which 4 was for fine adjustment. The results of the measurements are given in Table 26.

Series—magnetite: quartz.

Since for pure magnetite the intensity of the external field was too great, and the deflections of the galvanometer mirror went beyond the limits of the scale, the deflections of the galvanometer were determined for mixtures of magnetite and pure quartz, ground to 100 mesh and carefully triturated in an agate mortar. The weight of the mixture for each measurement was 2.83 g. The results of the measurements are given in Table 27.

Table 26. Relative magnetic permeability of iron and copper sulfides.
Current in the external circuit: 0.99 amp.
Charge of substance: 2.83 g.

Substance	Galvanometer deflection	Per 1% metallic Fe
FeS from FeS₂	0
Mattes from Cu₂S + FeS:
15% Cu	0
30% Cu	0
Mattes from Cu₂S + FeS₂:
19.64% Cu	0
29.0% Cu	0
39.16% Cu	0
Alloys FeS + Fe:
% metallic Fe
1.58	3.2	2.02
2.43	3.7	1.52
4.37	7.0	1.60
5.30	9.5	1.80
8.80	15.0	1.70
13.33	26.0	1.95
18.47	33.3	1.80
23.03	45.0	1.95

Substance	% O	Galvanometer deflection
Mattes of shaft smelting:
Karabash I	6.52	9.3
» III	5.06	8.3
Kalata	7.17	14.0
Pyshma	6.20	12.5
Pyshma, 10 o’clock a.m.	5.27	10.8
» 10 o’clock p.m.	4.02	11.7
» from the front furnace	6.04	16.0
Bogoslovsk	2.87	4.5
Polevskoi	3.15	11.0
Ulen	2.84	1.7
Yulia	2.13	5.6
Mattes of reverberatory smelting:
Tanalyk 12 X	7.75	10.0
» 14 X	not determined	10.0
» 17 X	7.63	8.3
Karabash 23 I	7.42	12.2
» 2 II	7.71	11.0
» 4 II	not determined	9.3
Dzansul 4 VII	3.50	6.0
» 11 VII	3.80	9.0

Table 27. Magnetic permeability of Fe₃O₄ : SiO₂ mixtures.

Mixture composition, % Fe₃O₄	% O	Galvanometer deflection, total	Galvanometer deflection, per 1% Fe₃O₄	Galvanometer deflection, per 1% O
5	1.38	3	0.6	2.18
10	2.76	9	0.9	3.26
20	5.52	18.2	0.91	3.30
25	6.90	27	1.08	4.0
40	11.04	49.5	1.24	4.48
50	13.80	89.0	1.78	6.45
60	16.58	93.5	1.56	5.64

From a comparison of the results of measurements of the relative magnetic permeability of mattes, FeS : Fe alloys, and magnetite–quartz mixtures, the following conclusions may be drawn.

1. The complete absence of magnetism in mattes obtained by smelting Cu₂S with FeS, Cu₂S with FeS₂, and containing both the theoretical amount of sulfur and an excess of it, shows that the components responsible for the magnetic properties of plant mattes are iron oxides dissolved in them and only partly metallic iron. The role of the latter in this case is, in general, not great, since in mattes from the oxidizing shaft process and from reverberatory smelting the content of metallic iron does not exceed tenths of a percent, as was shown by extraction of the mattes with a CuSO₄ solution. Only in mattes from reducing smelting, as, for example, in the mattes from the Uleab and Julia plants, does the content of metallic free iron reach 3.77–5.33%. But even in this case the influence of metallic iron on the relative magnetic permeability of the mattes is insignificant, as is seen from a comparison of the magnetic permeability of FeS : Fe alloys and of mattes containing the same amount of iron extractable by CuSO₄ solution. Thus, whereas for an FeS : Fe alloy with 5.3% metallic Fe the magnetic permeability is expressed by a galvanometer deflection of 9.5, for the Julia matte with 5.33% metallic iron this deflection is only 5.6, for the Uleab matte with 3.7% metallic Fe it is 1.7, while for the corresponding FeS : Fe alloy with 3.7% metallic Fe it is about 5.4. It follows from this that either the CuSO₄ solution extracts more iron from the mattes than is present in the matte in the metallic state, or the magnetic permeability of iron is abnormally strongly lowered if it is part of more complex sulfides of copper and iron, or, finally, the distribution of metallic iron in the mattes is nonuniform and there are local concentrations of it in separate parts of the matte.

2. Comparing the magnetic permeability of mattes with the amount of oxygen in them, we see that the magnetic properties of mattes do not change proportionally to the content of the magnetic component, as would be the case if the composition of the latter were one and the same in all mattes. This gives us grounds for asserting that the magnetic component in mattes is not only magnetite, but also ferrous ferrites, which in mattes may have the most varied composition. As follows from the work of Kohmeyer, Hilpert, Sosman, and Hostetter, magnetism in the FeO : Fe₂O₃ system is observed over wide limits of FeO concentration, namely from tenths of a percent to 82% FeO.

Evidence that the absence of proportionality between the magnetic permeability of mattes and the oxygen content in them is due to a difference in the composition of the magnetic component itself in the mattes, and not to its nonuniform distribution or to its great dilution by nonmagnetic components in the mattes, can be provided by the results of measuring the magnetic permeability of FeS : Fe alloys. If the numerical values of these measurements are represented graphically as a function of % Fe in FeS : Fe alloys, the change in magnetism is expressed by an almost straight line. This indicates the existence of direct proportionality between the magnetic properties and the content of the magnetic component in alloys in which we are dealing not with a mechanical mixture of magnetic and nonmagnetic components, but with their close contact and uniform distribution throughout the entire mass of the substance. In mechanical mixtures of magnetic and nonmagnetic substances there is no proportionality between the magnetic properties and the concentration of the magnetic component in the mixture, as is shown by measurements of the magnetic permeability of magnetite–quartz. The same observations were made by Benedicks⁵²), Sosman and Hostetter⁵³). The experiments of the latter showed that dilution of magnetite

a nonmagnetic substance, e.g., quartz, greatly reduces the magnetic permeability of the mixture. This is explained by the fact that the quartz particles in the magnetite–quartz mixture separate the magnetite particles, whereby the mutual induction of the latter decreases, and this decrease grows with increasing dilution of the magnetite, as is seen from Table 28.

Table 28. Magnetism of mixtures: \(Fe_3O_4 +\) quartz according to Sosman–Hostetter.

	50	40	25	20	10	5
% \(Fe_3O_4\) in the mixture	50	40	25	20	10	5
Observed deflection, galvan. I	89	49.5	27.0	18.2	9.0	3
Deflection in the case of direct proportionality to % \(Fe_3O_4\)—II	89	72	44.5	36	18	8.9
Ratio II : I	1.0	1.45	1.65	2	2	3

Sosman and Hostetter determined the relative magnetic permeability of ferromagnetic natural and artificial iron oxides by weighing the force of attraction exerted by a nonuniform magnetic field on a definite weight of the given substance. In their experiments the force of attraction is expressed as the number of mg needed to attract one gram of substance. The influence of the weight of the sample on the attraction proved negligible; thus, with a fivefold increase in the amount of magnetite, the increase in attraction per 1 g of mixture was only 3%, as is seen from the following figures:

magnetite, mg	attraction, mg per 1 g of magnetite
500	9.283
350	9.395
100	9.549

For quartz–magnetite mixtures the authors obtained the following values for attraction, in mg per one gram of mixture, with a current strength in the electromagnet of 500 milliamperes.

	100	50	10	1.0
% magnetite in the mixture	100	50	10	1.0
attraction in the case of direct proportionality to the percentage of \(Fe_3O_4\)	8718	4359	872	87
the same, but with correction for the changing amount of \(Fe_3O_4\)	8718	4441	900	90.3
obtained in the experiments	8718	3896	735	69.7

These experiments show that dilution of magnetite decreases the induction by an amount the greater, the greater this dilution; nevertheless the relative magnetic permeability of the mixture increases, although not in direct proportion to the increase in the concentration of magnetic substance in this mixture.

If the iron oxides in all mattes were identical in composition with magnetite, then the amounts of the latter, calculated from the oxygen content in the mattes, would be as follows—see Table 29.

Table 29. Relative magnetic permeability of mattes.

Matte	% O	% Fe₃O₄ calculated from oxygen	Calculated galvanometer deflection	Observed galvanometer deflection
Kalatinskii	7,47	27,03	30	14,9
Karabashskii I	6,52	22,65	22,9	9,5
Karabashskii III	5,06	17,57	15,8	8,5
Pyshminskii	6,02	21,50	20,8	12,5
Pyshminskii, 10 a.m.	5,27	19,08	17,1	10,8
Pyshminskii, 10 p.m.	4,02	14,55	13,1	11,7
Pyshminskii, from the front furnace	6,04	21,86	21,50	16,0
Bogoslovskii	2,87	9,98	9,0	4,5
Polevskoi	3,15	10,94	9,9	11,0
Ulen	2,84	9,9	8,9	1,7
Yuliya	2,13	7,39	5,9	5,6
Tanalyk, 12 X.	7,75	28,05	31,5	10,0
Tanalyk, 17 X.	7,63	27,62	30,9	8,3
Karabash, 23 I.	7,42	26,86	29,9	12,5
Karabash, 2 II.	7,41	26,86	29,9	11,6
Dvansulsk, 4 VII.	3,50	12,67	11,5	6,0
Dvansulsk, 11 VII.	3,80	13,76	12,5	9,0

The table gives the observed galvanometer deflections and those calculated under the assumption that the magnetic permeability is proportional to the percentage of Fe₃O₄ in the mattes.

There is nothing in common between the calculated and the observed values. This may serve as proof of the proposition stated above, that the iron oxides dissolved in mattes have different chemical compositions and belong to a series of ferrous ferrites. It is not excluded, of course, that in a particular case the ferrite will correspond to magnetite, as is probably the case in the mattes of the Polevskoi and Yuliya works, if one judges from the close values for the magnetic permeability of these mattes and the corresponding quartz–magnetite mixtures.

It was of interest to determine the degree of reducibility of the ferrites dissolved in mattes, and the accompanying change in their composition and magnetic properties. For this purpose two series of melts of certain oxygen-rich mattes were carried out in a charcoal crucible in an electric furnace. In the first series of melts, weighed portions of mattes were kept in the molten state for five minutes at a temperature of about 1100°; in the second, for 20 minutes, the mattes being stirred with a charcoal stirrer, which remained in the melt the whole time, whereby it was intended to achieve a more complete ...

reduction. In the mattes, before and after smelting, the quantity of metallic Fe was determined by means of a solution of $\mathrm{CuSO_4}$; in addition, measurements of the magnetic permeability of the mattes were made. The results are given in Table 30.

Table 30. Reducing smelting of mattes.

Mattes	% metallic Fe: Before reduction	% metallic Fe: After I smelting	% metallic Fe: After II smelting	Relative magnetic permeability: Before reduction	Relative magnetic permeability: After I smelting	Relative magnetic permeability: After II smelting
Pyshminsky	0.22	5.38	8.51	8.1	3.7	6.2
Karabashsky III	1.01	3.89	7.85	4.3	0	4.0
Kalatinsky	0.84	2.59	6.18	9.2	1.3	3.3
Karabashsky I	0.84	4.19	not smelted	9.3	4.0	—
Polevskoy	1.10	4.04	not smelted	11.0	4.3	—

To obtain comparable results in measurements for one and the same matte, identical charges were taken. Thus, the charges for

Matte	Charge
Pyshminsky matte	1.12 g
Kalatinsky matte	1.42 g
Karabashsky III matte	1.42 g
Karabashsky I matte	2.83 g
Polevskoy matte	2.83 g

The data of Table 30 show that, first, the ferrites of iron oxides dissolved in the mattes are very stable compounds, requiring for their complete reduction prolonged smelting in a reducing medium at high temperature. Charges of mattes of 6 g had to be smelted for 20–25 minutes without, judging by the loss in weight of the mattes, their complete reduction taking place; secondly, during reducing smelting the composition of the ferrites changes, which entails a change in the magnetic properties of the mattes. Since the magnetism of ferrites depends on the ratio in them $\mathrm{FeO} : \mathrm{Fe_2O_3}$, it may be assumed that during the first smelting, owing only to incomplete reduction, part of the $\mathrm{Fe_2O_3}$ was oxidized to $\mathrm{FeO}$, or part of the $\mathrm{FeO}$ was reduced to metallic Fe. Judging from the increase in the quantity of the latter extracted by a solution of $\mathrm{CuSO_4}$, it was precisely the latter of the indicated processes that occurred, and this entailed a decrease in magnetic permeability, which in the case of Karabashsky III reached zero. With more prolonged and stronger reduction, the ferrites in the mattes pass into metallic iron, which is connected with the formation of the system $\mathrm{FeS} : \mathrm{Fe}$ and an intensification of the magnetism of the reduced mattes. But in this case the magnetic permeability does not reach the original value for unreduced mattes. This may be explained by the fact that part of the metallic iron reduced from the ferrites enters into reaction with copper sulfides and passes into $\mathrm{FeS}$, with simultaneous separation of metallic copper. Indeed, in all mattes after smelting an abundant separation of metallic copper was observed.

Now let us turn to consideration of the question of the origin of oxygen or of magnetic ferrites of iron oxides in copper-smelting mattes. In the opinion

For the majority of metallurgists, oxygen is sometimes found in copper mattes, and then exclusively in the form of magnetite, which accounts for the magnetic character of mattes. This magnetite is assigned a place as a secondary or accidental impurity in the composition of mattes. But, as the present investigation has shown, we find oxygen in plant mattes of all types of copper smelting, from purely pyritic shaft smelting to the smelting of roasted ores in reverberatory furnaces. Even if we allow that the sole representative of iron oxides dissolved in mattes is magnetite, its amount, calculated from the oxygen in the mattes, varies from 7.39% to 28.05%, and in some mattes even exceeds the amount of copper contained in them (for example, Kalatinskii, Karabashskii, Tanalykskii). Consequently, iron oxides are not an accidental impurity, but an essential constituent part of plant mattes.

There is little definite information in the specialized literature concerning the causes of the appearance of iron oxides in mattes and their origin. This is explained, first, by the fact that the majority of metallurgists, regarding iron oxides as merely an accidental impurity in mattes, were interested neither in the role of this impurity, nor in its amount, nor, finally, in the conditions of its formation; and, second, by the fact that in ordinary matte smelting of low-copper ores the formation of any considerable amount of ferrous zinc ferrites was not observed, and the latter did not cause any special difficulties in the process. Only recently, in connection with the spread of reverberatory ore smelting and the bessemerizing of poor copper mattes in basic converters, which has entailed the necessity of de-copperizing large quantities of bessemer slags, have questions both concerning the conditions of formation of magnetite and concerning its prevention at various stages of copper smelting begun to occupy the attention of metallurgists, since in some processes magnetite causes great difficulties and increases copper losses, while in others it is a highly desirable product of the smelting. From the literature we may cite the following indications, the authors of all the articles cited below speaking only of magnetite.

A. Gibb and Philp ⁵⁴), who studied the structure of mattes, altogether deny the presence of magnetite in mattes. In their opinion, magnetite is formed when mattes are treated in the laboratory with oxidizing reagents, such, for example, as a mixture of concentrated \(HNO_3\) and potassium chlorate. In this case, upon dissolving the mattes, a black precipitate remains, the composition of which corresponds to the compound \(Fe_3O_4\), and the amount of which increases with intensification of the oxidizing action of the solvent.

E. Keller ⁵⁵) refutes the proposition of Gibb and Philp and says the following about magnetite in mattes. In order to obtain magnetite in matte or in slag, two conditions are necessary: magnetite must be present in the charge, and the reduction in the furnace must not be too strong to convert \(Fe_3O_4\) into \(FeO\), in which form the iron is slagged; or else the oxidation in the furnace must be sufficiently strong for \(Fe_3O_4\) to be able to form from iron sulfides during the smelting itself. In ore smelting the second condition undoubtedly predominates, although some amount of \(Fe_3O_4\) may enter the charge from roasted ore. The reason that we find larger amounts of \(Fe_3O_4\) in poorer copper mattes is the difference in the specific gravities of \(Fe_3O_4\) and mattes. So long as the matte has a specific gravity less than the specific gravity of \(Fe_3O_4\), the latter, naturally, sinks in the matte. Later the same Keller ⁵⁶) points out that \(Fe_3O_4\) is found in every charge of shaft ore smelting; in the latter, magnetite is very difficult to reduce to \(FeO\), which could be slagged by silica.

Peters ⁵⁷) at first saw the cause of the formation of magnetite in shaft smelting in an excessively high blast pressure. In this case, in Peters’s opinion, the iron sulfides passing by the tuyeres are overoxidized to higher degrees

...of oxidation. Later, on this same question, Peters[^58] expresses another view. He says that magnetite can form under conditions not yet clarified, both in shaft and in reverberatory slag, and in part originates from roasted ore and, during smelting, is not reduced to ferrous oxide. Peters’s opinion on the role of a high blast pressure in the formation of magnetite from iron sulfides in the region of the furnace tuyeres is also expressed in the more recent literature. Thus, Stören[^59], from his observations in pyritic smelting, comes to the conclusion that molten FeS is oxidized at the tuyeres to magnetite, which dissolves in the slag and matte.

Larisson[^60] explains the presence of magnetite in mattes by its passage from the shaft furnace. He reaches this conclusion on the basis of the results of smelting two grades of pyritic ore at the Ducktown Sulphur Co. plant. The ores contained different amounts of magnetite, and the mattes obtained from them also differed in their content of \(\mathrm{Fe_3O_4}\), the percentage of \(\mathrm{Fe_3O_4}\) in the matte being greater the more magnetite was present in the ore. The formation of magnetite is associated by some authors with problematic reactions occurring during smelting. Thus, for example, Walter[^61] explains the formation of magnetite during smelting by the peroxidation of strongly basic ferrous silicates, for example, \(4\mathrm{FeO}\cdot\mathrm{SiO_2}\). Wright[^62] believes that magnetite is the result of the reaction \(\mathrm{FeS} + 10\mathrm{Fe_2O_3} = 7\mathrm{Fe_3O_4} + \mathrm{SO_2}\), etc.

The well-known authority in the field of pyritic smelting, R. Sticht[^63], explains the formation of \(\mathrm{Fe_3O_4}\) in pyritic smelting as follows. In the upper part of the focus of a pyritic furnace, ferrous sulfide is oxidized exclusively to \(\mathrm{FeO}\), and the latter immediately forms with \(\mathrm{SiO_2}\) a silicate of a higher degree of acidity, for example a bisilicate or one still more acid. Further, this silicate is gradually enriched in \(\mathrm{FeO}\) at the expense of the oxidizing \(\mathrm{FeS}\) and passes into mono- and subsilicate; at the same time, owing to the peroxidation of \(\mathrm{FeO}\), much \(\mathrm{Fe_2O_3}\) and \(\mathrm{Fe_3O_4}\) is formed. With highly basic slags and a deficiency of \(\mathrm{SiO_2}\), so much magnetite is inevitably formed that the latter crystallizes out of the slags, collects on the hearth, and clogs the furnace. Thus, Sticht ascribes the principal role in the formation of magnetite in pyritic smelting to the basic silicates of ferrous oxide.

The most important observations explaining the method and conditions of the formation of magnetite during the oxidation of iron sulfides were made by Wheeler, Krejci[^64], and Howard[^65], and were published by them in a series of articles on the Bessemerizing of matte in basic-lined converters. In the article by the first two of these authors, entitled “Magnetite Lining for Basic Copper Converters,” the authors describe a method of forming a magnetite accretion on the magnesite walls of a converter. The method consists, in brief, in blowing 30–35% matte in the converter in the absence of, or with a deficiency of, quartz. To prevent too strong overheating obtained during this operation, the temperature in the converter is lowered by additions of cold matte. The operation is repeated until the matte is enriched to white metal, which is poured out of the converter. After this operation the converter walls prove to be covered with an accretion of magnetite. Then a new portion of matte is poured into the converter and quartz is added in an amount smaller than in normal operation. The charge is blown as usual, and the procedure is repeated until the magnetite accretion reaches such a thickness that it covers all the joints in the lining of the converter walls by 3–4 inches. The thickness of this accretion is regulated by the temperature in the converter and by the \(\mathrm{SiO_2}\) content in the slag. In a converter thus prepared, the Bessemerizing of matte is then carried out with the normal addition of quartz. Little by little the accretion is eaten away and it has to be renewed by the method described above. This operation is not complicated, requires little time, and is carried out without interrupting the normal operation of the converter. The patented auto-

The method of building up a magnetite lining on the walls of basic converters with ore soon found wide application, since the magnetite lining considerably increased the service life of converter walls and reduced the cost of their repair. The point is that magnesite, though a refractory and neutral material, nevertheless poorly withstands the sharp temperature fluctuations that occur when a converter is periodically tilted for pouring out blister copper. When in an incandescent state, magnesite is mechanically destroyed by the solid additions charged into the converter, such as ore, quartz, etc. A magnetite lining, on the other hand, is the most stable material both physically and chemically, since it is highly refractory, plastic at high temperature, and chemically inert with respect to the products of bessemerizing. In addition, a major drawback of magnesite is its considerable thermal conductivity, which makes it impossible to bessemerize rich mattes in small basic converters, since heat losses through the thermal conductivity of the converter walls are very large, while the heat supplied by the bessemerizing process is insufficient. The use of a magnetite lining as a good heat insulator made it possible to bessemerize rich mattes in small basic converters as well. The introduction of a magnetite lining as a protective layer for the magnesite walls of the converter is the implementation of the most correct principle in selecting materials for the walls of melting apparatus. In general terms, this principle consists in the fact that, when a given material is being melted, the strongest walls of the melting apparatus, from the physical and chemical standpoints, are walls made of the same material as that which is being melted. We encounter examples of this proposition throughout metallurgy. Such, for example, is the lining in water-jacketed shaft furnaces, consisting of molten and then solidified charge that is being smelted in the furnace. In modern reverberatory furnaces for ore smelting, the charge is introduced through openings in the long side walls of the furnace, and the ore is smelted essentially not among the refractory walls of the furnace, which are subject to the chemical action of the charge and slags, but in a bath whose walls are formed by that same charge. In refining werkblei, an arsenical and antimonial skimming forms on the surface of the metal bath, strongly corroding the furnace walls, especially in places where lead, air, and the walls come into contact. This belt, which suffers the most in furnaces, is lined with magnesite brick and cooled with water. But in practice corrosion of the walls is also reduced by not removing the skimming completely, leaving at the walls of the bath a crust of solidified skimming, which then protects the furnace walls it covers during further refining. If the metallurgist cannot find a suitable resistant material for the walls of a given melting apparatus, nature itself, as it were, comes to meet him. Similarly, for example, in pyritic shaft smelting the process taking place in the furnace creates within it the profile most suitable for its successful progress; the same process also often creates the most suitable wall in the melting apparatus.

Thus, for example, during the distillation of roasted zinc ore in muffles, the walls of the latter, over prolonged service, gradually change; from the original material of the walls—silicates of alumina and quartz—under the action of high temperature and the molten oxides of zinc and its vapors, completely new materials are formed, such as willemite, zinc spinel, and tridymite. The complete transformation of the clay mass of the muffle not only does not reduce its stability; on the contrary, this transformation determines the refractoriness and chemical strength of the muffle. Thus, the strength of the muffle increases during its operation, and the muffle would last much longer than 20–30 days if it were not gradually corroded by slag-forming constituents of the charge⁶⁶.

Similarly to this, the bessemerizing process of matte creates in the converter, under suitable conditions, a lining of the material most stable for the process.

Krejci \(^{67}\) explains in this way the formation of magnetite during the bessemerizing of matte. Experience has shown that magnetite is readily obtained only in the case when the temperature in the converter is too low for the formation of iron protoxide silicate and the amount of silica is small. Then any excess iron over that which is necessary for binding the available \(\mathrm{SiO_2}\) is oxidized to magnetite. The melting temperature of magnetite is \(1527\text{--}1538^\circ\), and since the temperature in the converter does not exceed \(1200^\circ\), the magnetite formed separates in a doughy form and adheres to the walls of the converter, gradually covering them with a magnetite accretion. Thus, the role of the magnesite lining of the converter has now been reduced to protecting the metal shell of the converter during the period until a sufficiently thick layer of accretion from magnetite has formed; the latter assumes the role of the true lining, in which the bessemerizing is carried out. Howard \(^{68}\) gives the following analyses of slags which were obtained during the build-up of magnetite accretion in the converter and remained in the liquid state.

\(\%\) \(\mathrm{SiO_2}\)	\(\mathrm{FeO}\)	or \(\mathrm{Fe}\)	\(\mathrm{CaO}\)	\(\mathrm{Al_2O_3}\)	\(\mathrm{Cu}\)
4.2	83.2	64.68	0.2	0.8	6.0
10.0	81.1	63.05	0.3	0.9	4.2
11.0	79.2	61.57	0.3	1.1	4.1

Normal slag in bessemerizing had the composition \(\mathrm{SiO_2}\)—22.7%, \(\mathrm{Fe}\)—37.6%, \(\mathrm{CaO}\)—1.0% and \(\mathrm{Al_2O_3}\)—2.9%. Taking into consideration that part of the iron in slags is bound with \(\mathrm{SiO_2}\) in the form of silicate, we see that the total \(\mathrm{Fe}\) content in the slags reported by Howard is below 65%, whereas in magnetite we have 72.37% Fe. Hence it follows that the slags of the period of formation of magnetite in the converter consist not of magnetite, but of an alloy of \(\mathrm{FeO}\) and \(\mathrm{Fe_2O_3}\), with a predominance of \(\mathrm{Fe_2O_3}\), i.e., they are ferrites of iron protoxide. Let us also cite the views expressed by H. O. Hofman and W. Borchers \(^{69}\) in their manuals on the metallurgy of copper, which are the most up-to-date.

H. Hofman assumes that in the slags and mattes of copper smelting only magnetite is present. The latter owes its origin to magnetite in the ore itself, to incomplete reduction of iron oxide during smelting, or, finally, to oxidation of metallic iron, but not of \(\mathrm{FeS}\) from the mattes themselves. In normal pyrite smelting \(\mathrm{FeS}\) is oxidized to \(\mathrm{FeO}\), which is slagged by silica. In an abnormal course of smelting, for example, with a deficiency of \(\mathrm{SiO_2}\) in the charge, magnetite may form, which is distributed between the matte and the slag.

Borchers says that in roasted ores smelted for matte there is present not only free iron oxide, but also its compounds with basic oxides, i.e., ferrites, the amount of which in the ores is the greater, the higher the roasting temperature was. These ferrites interfere with the process of slagging iron, and during smelting one must see to their preliminary reduction to \(\mathrm{FeO}\). Of the ferrites of iron protoxide, only meta- and orthoferrite can exist in slags. Further, Borchers says that when matte is blown in the converter without the addition of quartz, silicates cannot form, but only ferrites, varying in composition within the limits of meta- and orthoferrite.

The opinions expressed by various authors on the question of the conditions of formation and the composition of ferrites may be reduced to the following propositions.

1. The majority accept that the only ferrite encountered in slags and mattes of copper smelting is magnetite; the latter is found in mattes rather as an accidental impurity.

1. The sources of origin of magnetite are:

a) magnetite contained in the ore itself and passing without change into the slag and partly into the matte,

b) the over-oxidation of highly basic iron silicates, the oxidation of metallic iron in mattes or of ferrous iron when there is a deficiency in the charge of silica, as compared with the amount required for the formation of a slag normal at the given temperature,

c) the composition of the iron oxides obtained under these conditions varies within the limits of meta- and orthoferrite.

As will be clarified below, all the causes of magnetite formation indicated by various authors have their basis; but to this it must be added that, first, in copper-smelting processes ferrites of ferrous oxide of the most varied composition may form, and, in the particular case, magnetite as one member of the series FeO : Fe₂O₃; and, secondly, the quantity of ferrites formed—indeed the quantity of ferrites taken up by the matte—as well as the composition of the ferrites are determined by the nature of the process, the physical condition of the ore, and its chemical and mineralogical composition. The propositions stated may be illustrated by an analysis of the conditions under which the mattes investigated in the present work were obtained, or by their metallurgical characteristics. These mattes may be divided, according to the nature of the smelting by which they were obtained, into four groups:

1. Mattes from pyritic smelting: the Kalata and Karabash plants.
2. Mattes from semi-pyritic smelting: the Pyshma, Bogoslovsk, and Polevskoi plants.
3. Mattes from reducing smelting at the Yulia and Ulen plants.
4. Mattes from ore smelting in reverberatory furnaces.

Mattes from pyritic smelting. At the Kalata plant, by the method of pyritic smelting, copper pyrite is smelted, having the following average composition*): Cu—2.3%; Fe—43.5%; S—50% and SiO₂—1.9%. A characteristic feature of the Kalata pyrite ore is its massive character. The fluxes in ore smelting are quartz with 93% SiO₂ and limestone with 51% CaO. The coke consumption in smelting is 1.7% by weight of the charge, or 2.5% by weight of the ore. The sulfur content in the charge is 34%. Desulfurization, or the percentage of sulfur burned off, amounts to 89%, and the matte produced contains 12–14% Cu. Smelting is conducted on basic slags typical of the pyritic process, containing 31% SiO₂, 44.14% Fe (or 56.78% FeO), and 6.3% CaO. Since under Kalata conditions the oxidation and slagging of iron in the converter is cheaper than in the shaft furnace, the matte is directly bessemerized to blister copper in Peirce-Smith basic vertical converters. The bessemer slag, containing 20–22% SiO₂, 54% Fe (=69.46 FeO), and 1.7–2% Cu, is returned as a circulating product to the charge for ore smelting.

At the Karabash plant, by the method of pyritic smelting, copper pyrite is smelted, having the following average composition: % Cu—2.7, % Fe—36.38; % S—43; % SiO₂—6; Al₂O₃—2%; BaSO₄—5% and Zn—1.2%. The fluxes are quartz with 93–97% SiO₂ and limestone with 52–53% CaO. In the charge there is 33% S; the coke consumption is 1.7% by weight of the charge; desulfurization is 88–90%. Smelting is carried out on basic slags of the following composition: SiO₂—31%; FeO—53–54%; CaO—4.5% and Al₂O₃—3.4%. Matte with 30–37% Cu is bessemerized in a large basic Peirce-Smith converter. The bessemer slag with 25% SiO₂, 65% FeO, and 1–1.5% Cu is poured into the forehearth of the shaft furnace.

Thus, at Kalata and Karabash we have pyritic smelting in its pure form, and the conditions that cause the formation of iron ferrites during smelting, as well as their passage into the matte, must in both cases be identical.

) These data refer to 1916–17; they were collected partly during personal visits to the plants, and partly taken from Min. Sc. Press* 114, 1917, 309 and 115, 1917, 818.

Let us briefly examine the scheme of the processes occurring in pyrite smelting in a shaft furnace. This smelting is a strongly oxidizing process. Along the height of the shaft we distinguish three zones in the furnace: the upper—preparatory; the middle—the zone of oxidation and slagging, called the focus of the furnace; and the lower, which begins somewhat above the tuyeres and ends at the furnace hearth. In the preparatory zone there occur processes associated with the absorption of heat, namely, the thermal preparation of the charge, i.e. its heating by the ascending stream of hot neutral gases consisting of $N_2$, $SO_2$, and small amounts of $CO_2$ and $CO$. In the same zone there occurs the dissociation of sulfides, leading to the conversion of pyrite into $FeS$, and of copper glance into $Cu_2S$, $FeS$. Sticht^70) believes that the dissociation of pyrite at the end of the preparatory zone, owing to the high temperature prevailing there, proceeds further than $FeS$, and that what flows down into the focus of the furnace is not $FeS$, but sulfides poorer in sulfur, representing a solution of $Fe$ in $FeS$, for example, $4FeS.Fe$ or $3FeS.Fe$. This additional dissociation is, in Sticht’s opinion, harmful, since compounds with a lower calorific capacity than $FeS$ enter the focus, which diminishes the heat input in the furnace. One cannot agree with Sticht’s view of the additional dissociation of $FeS$, often expressed by other metallurgists as well, if one takes into account that in a neutral medium $FeS$ melts without decomposition and does not change in composition even after prolonged superheating to $1400^\circ$, as shown by the experiments of Yushkevich and Bornemann. The sulfides that have undergone dissociation flow down into the focus of the furnace, which begins above the tuyeres and does not reach the last ones. The position of the focus, or middle zone, is unstable; it rises or falls depending on the pressure and amount of blast entering the furnace. In the focus, liquid $FeS$ is rapidly oxidized and is immediately slagged by the white-hot quartz present here in the form of a porous aggregate. Both processes, oxidation and slagging, proceed very rapidly within a limited region and supply all the heat by means of which the smelting is carried out and all the items of heat expenditure of the pyrite furnace are covered. The iron silicate that is formed dissolves the remaining components of the charge and flows downward in the form of finished slag. The residue of unoxidized $FeS$ forms with $Cu_2S$ the second product of smelting—matte. Below the tuyere zone we have in the furnace a channel along which matte and slag flow down and where no chemical processes occur.

In the focus of the furnace, which is the region of maximum chemical activity, all reactions occur between $FeS$, $SiO_2$, and the oxygen of the air. For the proper and successful course of the process, the reacting substances must be present in a definite quantitative ratio, expressed by the reaction:

$$ 2FeS + SiO_2 + 3O_2 = 2FeO \cdot SiO_2 + 2SO_2; $$

any change in the quantity of one of the substances participating in the reaction must be accompanied by a simultaneous proportional change in the quantities of the two other substances; otherwise profound changes in the process are inevitable, leading to one or another difficulty in smelting. Thus, for example, if the volume of air blown into the furnace is changed without a corresponding change in the amounts of $FeS$ and $SiO_2$, part of the unused oxygen will penetrate into the preparatory zone, produce partial oxidation of the sulfides, whereby the calorific capacity of the descending charge will decrease, which entails a lowering of the temperature in the focus and, consequently, a cold running of the furnace. If the amount of $FeS$ is increased beyond that which is necessary for the principal reaction, the excess $FeS$ will not be oxidized but will pass into the matte. As a result, the yield of matte will increase, the copper content in it will decrease, which is economically irrational, for it is more advantageous to transfer iron to the slag than to the matte, which is subject to concentration or bessemerizing; bessemerizing of mattes is advantageous only at a certain minimum content of copper in them. Finally, if there is not enough quartz in the focus according to the requirements of the process, then part of the $FeS$, not having been oxidized, will pass into ...

matte, while the greater part of the FeS will be over-oxidized to ferric oxide. In the absence of SiO₂, it is precisely Fe₂O₃, and not FeO, that will be obtained, since the latter is unstable in the strongly oxidizing medium prevailing in the focus. Ferric oxide at high temperatures is even a stronger acidic anhydride with respect to bases than SiO₂. In the interaction of Fe₂O₃ and metallic oxides, stable and refractory ferrites are obtained. The oxidation of FeS in the presence of incandescent Fe₂O₃ will furnish the necessary base in the form of FeO, which will dissolve in the ferric oxide with the formation of ferrites of the type \((\mathrm{FeO})x.(\mathrm{Fe}_2\mathrm{O}_3)y\). The composition of the latter will be determined by the whole complex of conditions existing in the focus. The ratio \(\mathrm{FeO}:\mathrm{Fe}_2\mathrm{O}_3\) in the ferrites, or the degree of saturation of the ferric oxide by its ferrous oxide, will depend chiefly on the temperature of the medium. In the formation of ferrites, much more heat will be evolved than in the formation of iron silicates, as is evident from a comparison of the heats of formation of at least FeO and Fe₂O₃ from FeS:

\[ 2\mathrm{FeS} + 3\mathrm{O}_2 = \mathrm{FeO} + 2\mathrm{SO}_2 + 220520\ \text{cal.} \]

\[ \text{Per 1 mol. FeS there is evolved—}110460\ \text{cal.} \]

\[ \text{Per 1 mol. S}_2 \quad » \quad —\ 73506\ » \]

\[ 4\mathrm{FeS} + 7\mathrm{O}_2 = 2\mathrm{Fe}_2\mathrm{O}_3 + 4\mathrm{SO}_2 + 668240\ \text{cal.} \]

\[ \text{Per 1 mol. FeS there is evolved }167060\ \text{cal.} \]

\[ » \quad » \quad \mathrm{O}_2 \quad » \quad 95463\ » \]

In this, the heats of formation of orthosilicate, \(2\mathrm{FeO}.\mathrm{SiO}_2\), and of the corresponding orthoferrite, \(2\mathrm{FeO}.\mathrm{Fe}_2\mathrm{O}_3\), have not been taken into account, since for the former this heat is known and amounts to 22236 cal., while for the latter there are as yet no data⁷¹). In any case, we may suppose that the ferritization of ferrous oxide will also be an exothermic process, and, even if this heat is not taken into account, the exothermic effect of ferrite formation proves to be 50% higher than that of silicate formation. This conclusion is confirmed by the following observation from the practice of bessemerizing in a basic converter. Bessemerizing in a basic converter represents pyrite smelting in pure form, and the converter focus is the focus of a pyrite furnace. Therefore observations made during bessemerizing make it possible to explain many facts from the practice of pyrite smelting. Thus, for example, with a decrease in the addition of quartz, the temperature in the converter during blowing of matte rises sharply, and frequent cold additions (matte, quartz, etc.) are necessary so that, as a result of overheating, the walls of the converter are not destroyed. It is noteworthy that, with one and the same amount of blast and quartz, during the bessemerizing of poor mattes much more heat is evolved than during the bessemerizing of rich mattes. This is understandable, since in the case of a poor matte, owing to a deficiency of quartz, many ferrites are formed; in other words, ferrite formation predominates over slag formation, which is connected with an increased supply of heat in the converter and, consequently, with an increase in the temperature of the process. The ferrites formed in the focus of the furnace will dissolve in the slag and matte in proportion to their solubility in these substances. Since ferrites are distinguished by high viscosity and their melting temperature, regardless of the FeO content in them, is close to the melting temperature of pure ferric oxide, i.e., 1500–1600°⁷²), while the temperature in the furnace does not exceed 1240–1300°⁷³), the entry into the slags of such a refractory component will cause an increase in the viscosity of the slags, a sluggish running of the furnace, and an apparent lowering of the temperature in the furnace. In reality the temperature in the furnace will be considerably higher than in the formation of normal iron silicate, but the degree of viscosity and fluidity of the slags, which serves as the external sign for determining the temperature in the furnace, will in the case of ferrite formation indicate a cold running of the furnace. In addition, analogous to the separation of spinels of the type \(\mathrm{ZnO}.\mathrm{Al}_2\mathrm{O}_3\) and \(\mathrm{MgO}.\mathrm{Al}_2\mathrm{O}_3\) from slags in the smelting of lead and copper ores containing simultaneously

ZnO, MgO, and Al₂O₃, then in pyritic smelting we shall observe the separation of ferrites either in the furnace itself in the form of accretions on its walls and hearth, which may lead to clogging of the furnace, or in the forehearth, where this separation of ferrites will be especially favored by the gradual lowering of the temperature of the molten masses. The consequences of the phenomena indicated above will be an irregular furnace operation, a decrease in its productivity, an increase in the viscosity and specific gravity of the slags, poor separation of matte, and an increase in the copper content of the slags, owing partly to a decrease in the free volume of the forehearth as a result of build-up on its walls and bottom. To remove both wall and hearth accretions, so-called warming charges are charged into the furnace. The best results in this respect are given by charges consisting of a mixture of coke fines with limestone. Such a mixture produces, in the furnace focus, ferrites of calcium oxide, which are considerably more fusible than ferrous-oxide ferrites. Thus, according to Kohlmeyer’s investigations⁷⁴, the melting temperature of these ferrites is as follows:

\[ \begin{array}{rcl} \text{at }20\text{–}40\ \text{mol. }\%\,\mathrm{CaO} &\text{—}& \text{about }1230^\circ,\\ \text{” }40\text{–}50\ \text{” ” ”} & & 1200^\circ,\\ \text{” }60\text{–}75\ \text{” ” ”} & & 1300\text{–}1400^\circ. \end{array} \]

The readily fusible calcium-oxide ferrites gradually dissolve the refractory ferrous-oxide ferrites. When the temperature of slags and matte is lowered in the forehearth, iron ferrites separate out of the slags and, being only slightly soluble in the matte and having almost the same specific gravity as the mattes, are arranged in the form of an intermediate viscous layer between the matte and the slag. In practice this layer is called foam or paste (magnetic mush). In the forehearth we have a kind of differentiation of the matte. Dissolving in the matte, the ferrous-oxide ferrites raise the temperature of its free flow, or fusibility, as was shown by measurements of the temperature of free flow, or fusibility, of laboratory and plant mattes of identical chemical composition, carried out by Guess and Lathe⁷⁵.

Here are some results of their measurements:

	% Cu	% S	Temperature of free flow
1) Plant matte	10.3	24.2	970°
2) The same matte after remelting under charcoal	11.8	25.5	850°
3) Plant matte	13.7	26.4	912°
4) The same matte after remelting under charcoal	15.25	27.2	900°

From what has been set forth it follows that the principal cause of the formation of ferrous-oxide ferrites in the focus of a pyritic furnace is a local deficiency of free silica, as compared with the amount required to bind the FeO formed in the focus in the form of the normal slag of pyritic smelting. Such slag is considered to be ferrous monosilicate, as may be illustrated by numerous examples from the practice of pyritic smelting. Below are examples of such slags:

	SiO₂	FeO	CaO, Al₂O₃
Mount—Lyell⁷⁶	32.5	52.1	12
Kalata	31	57	6.3
Karabash	31	54	8
Knudsen process⁷⁷ (period of pyritic smelting)	28–31	52–60	—

Pure ferrous monosilicate has the composition: 29.2% SiO₂—70.8% FeO. Consequently, the composition of normal slags from pyritic smelting at various plants is almost identical. To obtain a normal slag, the amounts of pyrite and quartz in making up the charge are proportioned accordingly. The quantity of air at most plants at the present time

also regulate correctly, measuring not the pressure but the volume of air blown into the furnace per unit of time or per ton of charge. In spite of all this, in a pyrite furnace there always occurs the formation of ferrites of ferrous oxide. Without pausing for the moment over the question of how normal and necessary this ferrite formation is in pyrite smelting, we shall try to examine the principal causes of this phenomenon.

One of the causes, as indicated above, is a local deficiency of quartz in the focus of the furnace, which may arise for the following reasons. First, the charging of the materials entering the furnace does not yet ensure their correct proportion in the focus of the furnace, where the proper balancing of the components of the charge plays the chief role. In this connection, the physical character and mineralogical composition of the materials entering into the charge are of great importance. If quartz occurs in pyrite in the form of uniformly distributed grains, then it may be assumed that this distribution of pyrite and quartz will be only slightly disturbed by the time the charge reaches the focus of the furnace, even when the pyrite decrepitates. With such an ore fewer ferrites will form than with a charge consisting of large pieces of massive pyrite and separate pieces of free quartz. The Karabash and Kalata pyrites illustrate what has been said. The Karabash pyrite is penetrated by fine grains of pure quartz; the Kalata pyrite is a pure, massive, fine-crystalline pyrite. The content of SiO₂ in separate pieces of Karabash pyrite reaches 7%; in the Kalata pyrite it rarely exceeds 1–2%. At Kalata, owing to the more frequent formation of accretions, the furnace campaign is shorter than at Karabash. At Kalata it has been observed that ore from the Northern stope, containing up to 5% SiO₂, smelts more easily and causes fewer difficulties than ore from other stopes that is poorer in quartz. Another cause of a local deficiency of quartz in the furnace focus is the sorting of the charge as it descends in the furnace, and the associated non-simultaneous arrival of the charge components in the focus. This is caused by the great nonuniformity in the size of the pieces of pyrite and quartz, the difference in the specific weights of these minerals, the large amount of fines in the charge, and the sorting action of the blast. When the quartz is too coarse in comparison with the pyrite, the latter is sorted out, outstrips the quartz, and an excess of the liquated sulfide enters the focus of the furnace. The result is the formation of ferrites and a lowering of the degree of concentration of the matte. From the practice of the Blagodat plant one may cite the following fact. While, in the concentration smelting of crude matte by the method of pyrite smelting, coarse-lump quartz was charged, only a slight concentration of the matte was obtained. Matters immediately improved when lump quartz was replaced by quartz tailings (5–10 mm) from the concentrating mill, which, understandably, ensured a more correct proportion of FeS (matte) and SiO₂ in the focus of the furnace.

The next cause of ferrite formation is the friability of pyrite and its tendency to decrepitate strongly on heating, which leads to the formation in the furnace of a large quantity of pyritic fines. The great elasticity of the blast still further promotes sorting of the ore mixture; the pyrite outstrips the quartz, and in the focus of the furnace conditions are obtained that favor the formation of ferrites. In this respect the assertions of Peters, Stören, and others—that a high elasticity of the blast promotes the formation of magnetite during smelting—are entirely correct. Peters, for example, says that in cases known to him a lowering of the elasticity of the blast at once diminished and even eliminated the formation of this undesirable oxide.

As indicated above, ferrites dissolve in slag and matte in amounts depending on the temperature. At the same time, ferrites dissolved in the matte are very stable compounds; they do not change their composition and do not react with the components of mattes, as is proved by the constancy of the composition of mattes observed when they are remelted in a neutral medium. Mattes re-

carry out of the active zone of the furnace ferrites, so to speak, in statu nascendi, with the same composition in which they originated during the process. Therefore, if it were possible to establish the rational quantitative composition of the ferrites in the mattes and the heat of formation of these ferrites, then it would thereby be possible to take into account the role of ferrite formation in smelting and the participation of this process in the heat balance of pyrite smelting.

Consideration of the causes of ferrite formation leads us to the following conclusion. The realization of ideal conditions that would exclude the formation of ferrites in the furnace focus is in practice unattainable, and therefore ferrite formation in pyrite smelting is an inevitable process. If the high thermal effect of this process is taken into account, we must acknowledge that the formation of a certain quantity of ferrites in pyrite smelting is even necessary, since in it the oxidation and slagging reactions of FeS are the only incoming item in the heat balance of the smelting, at whose expense all heat expenditures and losses are covered. There is no excess of heat in the focus, and any increase in the amount of the latter makes the process more elastic and the normal operation of the furnace less dependent on accidental causes.

Owing to differences in the composition and character of the ores smelted at the Karabash and Kalata plants, the quantity and composition of the ferrites formed in the furnace focus and assimilated by the mattes at the two plants must also be different; this difference is also reflected in the unequal oxygen content in the mattes and in the difference in their magnetic permeability. In addition, the composition and content of ferrites in the mattes of both plants is influenced, to some extent, also by the difference in the method of bessemerizing the Bessemer slags at these plants. At Kalata, poor matte with 12–13% Cu is bessemerized in six-ton basic converters; at Karabash, 28–34% matte is bessemerized in a large horizontal converter. To obtain white metal with 78% Cu and 2% Fe from Kalata matte, it is necessary to oxidize and slag, out of 100 kg of matte, 61 kg of iron; with Karabash matte, only 41 kg. The small capacity of the converter at Kalata forces the process to be conducted on highly basic, hotter slags with 20–22% SiO₂, in order to obtain a smaller quantity of them; otherwise the heat of the reaction would not suffice to maintain the enriched matte and, in particular, the large quantity of slag in a liquid state at the temperature normal for the process. At Karabash, the greater capacity of the converter and the richer matte make it possible not to restrict the quantity of slags by increasing the % SiO₂ in them, since the exothermic effect of bessemerizing large masses of matte is sufficiently great to maintain the bath temperature at the proper level. The features of bessemerizing at both plants are reflected in the composition of the slags, analyses of which are given below:

	% SiO₂	FeO	Cu	O (by difference)
Kalata Bessemer slag	20	54	1.7	26
Karabash “ ”	28	63	1.5	23

In the plant analyses the form of the iron compounds is not indicated; iron is usually given as FeO, but from comparison of the analyses it is evident that in the Kalata slag iron ferrites rich in Fe₂O₃ predominate, while in the Karabash slag there are ferrous iron silicates mixed with ferrites.

At the Kalata plant, cold Bessemer slag is added in small quantities to the charge of the shaft smelting. In the pyrite furnace there are no conditions that could cause any changes in the composition of the ferrites introduced into the charge with the Bessemer slag. These ferrites reach the furnace focus, increase the quantity of ferrites formed there, and mechanically dissolve in the slag and, partly, in the matte; since these ferrites are richer in oxygen than the ferrites formed during the shaft smelting itself, this also accounts for the higher oxygen content in the Kalata matte.

At Karabash the Bessemer slag is poured into the forehearth of the shaft furnace. Without dwelling on the advisability of such a practice, we may merely point out that the heavy Bessemer slags, rich in ferrites, mix very imperfectly with the lighter and more acid furnace slags. The ferrites from the poured Bessemer slag have less opportunity to dissolve in the slag and, especially, in the matte than occurs when Bessemer slags are smelted together with ore at Kalata. This cause also plays a certain role in the higher oxygen content in the Kalata matte.

In the forehearth, as the temperature falls, the ferrites dissolved in the slag and matte are gradually separated out in the form of a doughy mass. The latter, being heavier in specific gravity than the slag, forms an intermediate layer between the slag and the matte. In addition, these ferrites form accretions on the walls and hearth of the forehearth. A piece of such a hearth accretion from the Karabash plant was examined by me. The accretion was an agglomerate of octahedral black crystals with a metallic luster, cemented by matte and strongly magnetic. In the selected crystals, analysis showed 67.12% Fe and 25.04% O; pure magnetite contains 72.36% Fe and 27.64% O; consequently, the crystals from the accretion are FeO ferrites close to magnetite. The average composition of the ground sample of the accretion proved to be as follows: Fe—58.74%, metallic Fe—1.5% (by extraction with a solution of CuSO₄), S—6.5%, Cu—10.72%, SiO₂—2.50% and O—19.40%. In composition this is a FeO ferrite mixed with matte.

Let us now turn to the consideration of the second group of mattes, i.e., mattes from pyritic smelting at the Bogoslovsk, Polevskoi, and Pyshminsko-Klyuchevskoi plants. At the Bogoslovsk plant the charge for shaft smelting consists of pyritic and oxidized ores, roasted agglomerated fines, and dust from water-jacket furnaces and converters. The average percentage composition of the charge is: Cu—5.8, Fe—23, SiO₂—29, S—10, Al₂O₃—7, CaO—12.6, and MgO—1.5%. The iron is present in the charge in the form of pyrite, pyrrhotite, copper pyrite, iron oxide, and magnetite contained in the Auerbach ore. The coke consumption is 7% of the weight of the charge. Desulfurization is 66%; the matte contains 35–40% Cu; the discard slag has the composition: % SiO₂—38.6, FeO—34, Al₂O₃—12, CaO + MgO—14.3. The matte is Bessemerized in six-ton converters with acid lining; the Bessemer slag contains 30% SiO₂, 47% Fe, 7.5% Al₂O₃ and 2% Cu; it is poured into the forehearth.

The smelting conditions at the Polevskoi plant are entirely analogous, where the smelted charge consists of pyrites, return low-grade matte, and quartz. Owing to the high sulfur content in the charge, the coke consumption is only 4–4.5% of the weight of the latter, and desulfurization is 75%. The matte contains 20% Cu, and the composition of the slag is close to that of the slag of the Bogoslovsk plant: 36% SiO₂, 46% FeO, 10% CaO and 2.7% Al₂O₃. The high desulfurization is achieved by a large volume of blast, which creates an oxidizing atmosphere in the upper part of the furnace, owing to which a considerable portion of the sulfides is oxidized here. Under identical smelting conditions, the presence of oxygen in the mattes of both plants must be due to analogous causes. In pyritic smelting the atmosphere in the furnace is weakly oxidizing, as is characterized by the following examples of the composition of grate gases in this smelting⁷⁸):

Component	Content
CO₂	10.9—6.49% (vol. %)
CO	2.15——
SO₂	2.5—1.27%
O	8—10.18%

Analysis of gases sampled one foot below the charge level at the Pyshminsky plant gave the following results:

Gas
$\mathrm{SO_2}$	2.62	7.23	5.48% (by volume, %)
$\mathrm{CO_2}$	13.3	10.2	13.0%
$\mathrm{CO}$	0.4	0.5	1.4%
$\mathrm{O}$	0.1	1.1	1.0%.

Some samples gave a maximum of 2% $\mathrm{O}$. It is obvious that the difference in the oxygen content of the waste gases in the examples cited above is explained by the difference in the volumes of air entering the furnace, which is determined by the sulfur content of the charge and by the desired degree of desulfurization and, consequently, of matte concentration. At American plants, where semipyritic smelting is practiced, the charge column is incandescent and its temperature is not below 700°; at our Ural plants, with the same type of smelting, the charge column is dark, and the temperature of the charge gases immediately above the charge—for example, at the Pyshminsky plant—varies within the limits of 400–550°.

With a weakly oxidizing atmosphere in the furnace, there are, of course, no conditions for reducing the higher iron oxides introduced into the furnace with the charge. Moreover, in order to achieve the greatest possible desulfurization in semipyritic smelting, efforts are made to ensure that the ascending gas stream is rich in free oxygen and oxidizes the excess $\mathrm{FeS}$ already in the preparatory zone. In this zone partial roasting of sulfides takes place, and since the temperature here is still not high enough for the formation of iron silicates, the $\mathrm{FeS}$, despite the presence of $\mathrm{SiO_2}$, is oxidized chiefly to the higher iron oxides, e.g., $\mathrm{Fe_2O_3}$ or ferrites $\mathrm{FeO}$, as is also observed in roasting pyritic ores in roasting furnaces. The higher iron oxides and ferrites introduced with the charge and formed in the preparatory zone of the furnace enter the smelting zone and there may be deoxidized to the lower degree of oxidation necessary for slag formation, i.e., to ferrous oxide, only at the expense of the incandescent solid carbon present in this zone. But the quantity of coke, which in semipyritic smelting does not exceed 4.5–8% of the weight of the charge, is insufficient to reduce the entire mass of higher iron oxides to $\mathrm{FeO}$, and part of them dissolves unchanged in the slag and matte. These are the causes of the occurrence of oxygen in the mattes of the Pyshminsky, Bogoslovsky, and Polevskoy plants. In semipyritic smelting, coke supplies 50–65% of all the heat necessary for the process 79), and plays the role of the sole reducing agent, since with an excess of air blown into the furnace the concentration of $\mathrm{CO}$ in the gases must be very small, as is evident from the analyses of these gases. To prevent the transition of the higher iron oxides into the slag, which increases their viscosity and the smelting temperature, it is necessary to ensure the fullest possible reduction of these oxides to $\mathrm{FeO}$ and the slagging of the latter. It is therefore very important that a large part of the coke charged into the furnace reach the smelting zone, located in the region of the tuyeres, and not burn in the upper parts of the furnace shaft. At the Pyshminsky plant it has been observed that, in preserving the coke down to the tuyeres, where this coke performs useful work, a major role is played by the amount and fusibility of the coke ash. When the ash content of the coke is too low and its ash is refractory, a considerable part of the coke has time to burn out before reaching the smelting zone, which leads to a lowering of the temperature in this zone, a cold run of the furnace, and viscous slags. If the quantity of ash in the coke is sufficient and it is relatively fusible, then on the lumps and in the pores of the coke there forms a kind of slag glaze, which retards the burning of the coke and preserves it until the smelting zone. At the Pyshminsky plant, in the charge for ore smelting, consisting of pyritic and oxidized ores, the predominant components are

ores of the latter category. In view of the deficiency of sulfur in the charge, and in order to reduce the consumption of coke, from 15 to 16% of Kalata and Belorechensk pyrite, containing up to 50% S, is introduced into the charge. The charge has the following composition: % \( \mathrm{SiO_2} \)—31; Fe—27; Cu—4.9; S—8–10; CaO, MgO—10; \( \mathrm{Al_2O_3} \)—8.9; coke consumption—9% of the weight of the charge; desulfurization—70–74%; the matte contains 24% Cu, and the slag has the following composition: 35% \( \mathrm{SiO_2} \), 28% FeO, 9% \( \mathrm{Al_2O_3} \), and 17% \( \mathrm{CaO + MgO} \). The matte is bessemerized in a vertical basic converter; the weight of the initial charge is 3.5–4 tons. Bessemer slag contains 18–20% \( \mathrm{SiO_2} \) and 55% Fe. In the character of the smelting, that at the Pyshma plant is analogous to that at Bogoslovsk and Polevskoi. Meanwhile, the matte of the Pyshma plant is much richer in oxygen than the matte of analogous plants. This is explained by the fact that at the Pyshma plant bessemerizing is practiced in a basic converter, and the Bessemer slag, rich in ferrites, is poured into the forehearth or partly admitted into the ore smelting. The ferrites from the Bessemer slag dissolve in the furnace slag and matte, which results in a high oxygen content in the latter. The high content of refractory ferrites in the poor mattes of Kalata and Pyshma may be explained, in part, by the longer service life of the small converters at these plants. Ferrites from the mattes promote the formation of that protective “magnetite” accretion, which is specially built up on the converter walls by the method of Wheeler, Krejci, and Howard.

Let us now consider the third group of mattes, namely, the mattes of reducing smelting at the Yuliya and Ulen plants.

At Yuliya, matte is obtained in shaft smelting of ores of mixed character, oxidized and pyritic, with the former predominating. The ores smelted have the following composition: % \( \mathrm{SiO_2} \)—34.92; Fe—4.8; Cu—1.8; S—4; CaO, MgO—20.9 and \( \mathrm{Al_2O_3} \)—16.1. The charge for matte smelting consists of 93% ores of the composition indicated above and 7% return slags from the concentrating smelting of roasted matte to black copper. The smelting is conducted on charcoal, which is consumed in an amount of 17% of the weight of the charge. The matte contains 40–41% Cu, and the slags have the following composition: % \( \mathrm{SiO_2} \)—41.5; Fe—12; CaO—28.4; \( \mathrm{Al_2O_3} \)—12.3 and MgO—3.5. Desulfurization amounts to about 60%. At the Ulen plant, matte is smelted from analogous ores from the Antoninin and Glafirin mines; these ores have the following composition:

%	Cu	Fe	S	\( \mathrm{SiO_2} \)	\( \mathrm{Al_2O_3} \)	\( \mathrm{CaO + MgO} \)
Antoninin ore	2.7	18.2	2.2	30.3	5.4	20.9
Glafirin ore	5.3	17.6	7	29.7	5.4	15.7

From these ores a mixture is made up, containing on average 3.5% Cu and 4% S. The charge consists of 75% ores, 12% return slags from the smelting of roasted matte to black copper, and 13% limestone. The smelting is carried out on charcoal, the consumption of which amounts to 25% of the weight of the charge. A matte with 40% Cu is obtained, and slags of the following composition: % \( \mathrm{SiO_2} \)—38.7, FeO—21, \( \mathrm{Al_2O_3} \)—5 and CaO—28. Desulfurization does not exceed 40%. The presence of oxygen in the mattes of both plants is explained by insufficient reduction of the higher iron oxides contained in the charge and passing into the slag and, in part, into the matte. Measurements of the relative magnetic permeability of the mattes showed that the iron oxides in the mattes of Yuliya and Ulen are only weakly magnetic and probably belong to the extreme member of the series of ferrites rich in \( \mathrm{Fe_2O_3} \).

It remains for us to consider the mattes from smelting ores in reverberatory furnaces at the Karabash, Tanalyk, and Dzhezkazgan plants.

In the regenerative furnace of the Karabash plant, 10–12 thousand poods of charge are smelted per day, consisting of fine cupriferous pyrite, dust, and screenings

from water-jacket furnaces and converters. Thus, the reverberatory furnace serves only for the utilization smelting of fine Cu-bearing materials. In a furnace fired with producer gas, the atmosphere is weakly oxidizing and desulfurization amounts to 25–30%. Analyses of the materials smelted in the furnace and of the smelting products on the days when matte samples were taken for the present investigation are given in Tables 31 and 32.

Table 31. Analyses of materials smelted in the reverberatory furnace of the Karabash plant.

%	Cu	Fe	S	Zn	SiO₂	Al₂O₃	CaO	BaSO₄
Copper pyrite . . .	3.60	39.20	46.40	1	2	2	—	4.60
Dust from water-jacket furnaces . . .	2.70	40	7.10	2.20	14.80	4.90	2.50	7.20
Screenings and cleaning . . .	4.0	37.20	9.60	1.70	31	0.90	2.60	2
Ivanovo SiO₂ flux .	0.25	3.50	1.70	—	73	3.80	1	8

Table 32. Analysis of slags and matte from reverberatory smelting at the Karabash plant.

%	Cu	Fe	S	SiO₂	CaO	Al₂O₃	BaSO₄	O
Slag 23 I . . . . . .	—	36	—	31.30	2.30	4.80	7.20	—
" 2 II . . . . . .	—	37.20	—	35.20	2.10	5.10	7.0	—
Matte 23 I . . . . . .	12.50	51.50	26.30	—	—	—	—	7.42
" 2 II . . . . . .	12.50	51.50	26.90	—	—	—	—	7.71

Daily samples of matte, taken over a period of 10 days, are very similar in composition, which is explained by the fairly constant composition of the charge and by the fact that there is always a large bath of matte in the furnace, and the mattes from the smelting of individual charge loads are continuously mixed.

Analogous smelting conditions exist at the Tanalyk plant. Here smelting is carried out in an open-hearth-type furnace on producer gas of the following composition: CO — 28%, CO₂ — 8–9%, CH₄ — 6.6%, H₂ — 8–9.9% and O — 0.2–0.5%. In the furnace there is a weakly oxidizing atmosphere, as is seen from the analysis of the gas-combustion products: CO₂ — 16–16.2%, O — 2.5–3.5%. Per day the furnace smelts 10–11 thousand poods of charge, consisting of screened ores with grain size up to 12 mm in the raw state and converter breakage with acid lining. Part of the fine ore, too rich in sulfur, is roasted before smelting on grates.

The composition of the ores and smelting products is given in Table 33.

Table 33. Analysis of ores and products of reverberatory smelting at the Tanalyk plant.

%.	Cu.	Zn.	Pb	Fe.	S.	Si O₂	Al₂O₃	Ba SO₄.	O.
Gubinsk ore . .	1	0.70	4.20	13.30	2.10	25.40	9.60	30.80	—
Calcined fine ore	2.90	4.80	1.0	18.20	16.60	27.60	14.80	10	—
Semenov ore .	—	—	—	38.30 Fe O	—	25.70	3.60	3.80	—
Slag . . . . . .	0.32	2.60	2.90	35.60	—	35.90	12.40	6.36	—
Matte . . . . . .	12.25	3.10	3.29	48.05	25.89	—	—	—	7.63—7.75

The mattes from reverberatory smelting at the Karabash and Tanalyk plants are almost identical in composition. The presence of a large quantity of oxygen in the mattes is explained by the fact that at both plants raw sulfide ores enter the smelting charge in considerable quantity. During the oxidation of iron sulfides in the reverberatory furnace there are conditions highly favorable for the formation of ferrous ferrites. The charge, which is a mechanical mixture of sulfides and quartz, lies for a prolonged time on the hearth of the furnace or floats on the surface of the slag until it itself melts and unites with the general bath. In this case, understandably, a local deficiency or absence of SiO₂ will contribute to a considerable part of the ferrous iron being oxidized to ferric oxide; the latter assumes the role of an acid anhydride in relation to the remaining part of the oxidizing FeS and dissolves the FeO thereby formed. The degree of saturation of ferric oxide with ferrous oxide, and consequently the composition of the ferrites obtained, depends mainly on the temperature of the furnace and the degree of fusibility of the charge. The more refractory the latter is, the longer the period of oxidation will be, the more ferrites will be formed, and the richer the latter will be in ferric oxide. Ferrites dissolve in the slag and matte and cause the high oxygen content in the latter. In this respect, owing to the immobility of the charged materials in the reverberatory furnace, the conditions for ferrite formation are even more favorable than in the shaft furnace, where the charge is in continuous downward motion and FeS, having reached the focus, has a greater chance of encountering there incandescent quartz and, upon being oxidized, of forming ferrous silicate.

The formation of viscous and refractory “magnetite” slags during the smelting of ores in reverberatory furnaces is noted by many metallurgists. Thus, Fulton ⁸⁰), who studied these slags in order to establish their mineralogical composition, says that microscopic examination of reverberatory-furnace slags shows that in these slags a considerable quantity of iron is present in the form of magnetite, which is a very widespread and even common component of slags. It is therefore incorrect to regard all the iron in slags as ferrous oxide combined with SiO₂. Reverberatory-furnace slags with 38.5—42.74% SiO₂ belong to the binary augite–olivine system or the ternary augite–olivine–magnetite system. Both ferrous and ferric iron oxides may enter into the composition of augite. The amount of magnetite in some slags investigated by Fulton exceeded 10%. In Fulton’s opinion, the causes

the origin of magnetite in reverberatory-furnace slags has not yet been clarified. It is possible that magnetite is introduced with converter slag, but part of the magnetite is present in roasted concentrates and may form during the smelting itself. Further, Fulton points out that the iron bound in magnetite cannot be regarded as a flux, and that the presence of magnetite is, in general, harmful, since it crystallizes out of slags at a comparatively high temperature and partly passes into the matte; in considerable quantity magnetite makes slags pasty, owing to which they carry off much copper. According to Flynn’s report ⁸¹), at the Arizona Copper Co. works, in the front part of the reverberatory furnace, where there is a large excess of air, pyrite readily gives magnetite-bearing pasty slags (magnetic mush). At the copper works in El—Paso ⁸²), in ore smelting in a reverberatory furnace, a deficiency of quartz in the charge causes the formation of an extremely viscous, copper-rich material located between the matte and the normal slag. In the opinion of the author of this report, magnetite forms in the layer of slag adjacent to the matte. Increasing the addition of quartz eliminates this phenomenon, since quartz binds FeO from the magnetite (?), and then further reduction of ferric oxide occurs. It is scarcely possible to agree with this last remark of the author, if one takes into account the extraordinary stability of zinc ferrites of iron even in a strongly reducing medium, as shown by experiments on remelting mattes in carbon crucibles. Evidence of the great stability of ferrites in slags is furnished by unsuccessful attempts to impoverish basic Bessemer slags in reverberatory furnaces. These slags are rich in ferrites and at most American works they are poured, in the liquid state, into the bath of reverberatory furnaces. In doing so, owing to their high specific gravity, the heavy ferrite slags do not mix with the normal ore-smelting slag, but are situated as an intermediate layer between the matte and the ore slag. If there is too much ferrite slag, it forms a pasty mass which has to be scraped out of the furnace. In view of the absence of methods for the rational analysis of ferrite slags, it is impossible to determine in them the quantity and composition of the ferrites and thus establish their fluxing value for ores of a given composition and, finally, to ascertain whether, in general, the composition of the ferrites of ferrous oxide changes as they pass through the furnace bath ⁸³). Stout ⁸⁴), chief metallurgist of the Copper Queen works, Douglas (Arizona), reports that all attempts to impoverish basic Bessemer slags by pouring them into the bath of a reverberatory furnace have so far been unsuccessful. According to Stout’s observations, Bessemer slag, being heavier than ore slag, does not mix with the latter, and in the bath a settling or differentiation of slags takes place. To eliminate this phenomenon and avoid the copper losses contained in the Bessemer slag, it has been proposed to carry out Bessemerizing in basic converters on slags as acid as possible, which will reduce the amount of “magnetite” in the Bessemer slag and impart to it the property of mixing with the acid slag of the reverberatory furnace. In a 12-foot vertical converter it proved possible to bring the percentage of SiO₂ in the slag up to 40% by continuously blowing powdered quartz into the converter during the first period of Bessemerizing.

It remains for us to consider the matte of the Dzansul works. At this works ore smelting is carried out in a reverberatory furnace 48 ft × 15 ft, fired with oil. The smelting charge consists of roasted flotation concentrates, slag from acid Bessemerizing, and limestone. The roasted concentrates have the composition: Cu—11.3—12%, S—10.9—9.3%, and SiO₂—27.5—26%; the Bessemer slag contains 32% SiO₂.

From the collection of mattes for the week, the mattes of July 4 and 11 were selected, in view of the fact that, in obtaining them, the amount of Bessemer slag in the charge of the reverberatory furnace differed greatly. Thus, the composition of the charge on the indicated days was as follows:

	July 4	July 11
roasted concentrates	39.8 t	35.9 t
Bessemer slag	8.5 »	3.0 »
limestone	0.7 »	4.25 »
matte	0.5 »	0 »

The investigation of the mattes showed that, despite this circumstance, the mattes differed little in composition. This is explained by the fact that the acid Bessemer slag, as containing only few ferrites of ferrous oxide, mixes well with the slag obtained in ore smelting. In this case no separation of ferrites and transfer of the latter into the mattes is observed, whereby their oxygen content is not high. Owing to the strongly oxidizing atmosphere in the furnace, part of the sulfides remaining in the roasted concentrates is oxidized to ferrites, and the latter are assimilated by the matte and slag.

The strongly oxidizing character of the atmosphere in the reverberatory furnace operating on oil is demonstrated by the following composition of the gases in the furnace:

\[ \mathrm{CO_2}—5.9\%;\ \mathrm{CO}—0.4\%,\ \mathrm{SO_2}—0.3\%\ \text{and}\ \mathrm{O}—12.9\%. \]

The metallurgical characterization of the mattes investigated and of the processes by which they were obtained is given in Table 34.

It is evident from the table that the content of oxygen, and consequently also of ferrites, in mattes from shaft smelting is directly proportional to the percentage of sulfur in the charge and to the degree of desulfurization, and inversely proportional to the acidity, or the content of $\mathrm{SiO_2}$, in the slags accompanying the production of the corresponding mattes.

Conclusions.

1. In all mattes of copper ore smelting, irrespective of the character of the ores and the type of smelting, oxygen is found; the latter is bound in the mattes with iron in the form of ferrites of ferrous oxide.

2. Contrary to the opinion established among metallurgists, ferrites are not an accidental impurity, but an essential constituent part of mattes, as is evident from the fact that the amount of oxygen in copper mattes varies within the limits from 2.5 to 8%.

3. Ferrites in copper mattes belong to the members of the system $\mathrm{FeO}:\mathrm{Fe_2O_3}$, possess variable composition, and are the carriers of the ferromagnetic properties of mattes.

4. Both the results of measuring the relative magnetic permeability of mattes and the variety of conditions of ferrite formation speak in favor of differences in the composition of ferrites in mattes. Only in a particular case are these conditions such that they can lead to the formation of magnetite. Therefore the view established in the literature, that oxygen is present in mattes exclusively in the form of magnetite, cannot be recognized as correct.

5. The formation of ferrites of ferrous oxide and their transfer into matte are caused by various reasons, of which the most important are the following:

a) oxidation of iron sulfides in the absence of silica, or with such an amount of it as is insufficient for the formation of ferrous silicates. In this case the iron sulfides are oxidized to ferric oxide, which at high temperatures is a strong acid with respect to bases, in particular to ferrous oxide, and with the latter gives the corresponding ferrites. This process takes place in oxidizing smelting of pyritic ores both in shaft and in reverberatory furnaces, and does not depend on the degree of the oxidizing character of the smelting,

Table 34. Metallurgical characteristics of matte-smelting slags

PLANTS	Composition of smelted ores	Sulfur in slag, %	Fuel	Fuel consumption, % of charge	Type of furnace	Atmosphere in the furnace	Desulfurization, %	SiO₂ in slag, %	Cu, % of matte composition	Fe, % of matte composition	O, % of matte composition
Karabash	Coppery schist	33	Coke	1.7	Pyritic	Strongly oxidizing	90	31	27.6	40.15	6.52
Karabash	Coppery schist	33	Coke	1.7	Pyritic	Strongly oxidizing	90	31	37.10	32.58	5.06
Kalatinskii	Coppery schist	34	Coke	1.7	Pyritic	Strongly oxidizing	89–90	31	14.55	53.40	7.42
Psh.-Kirovskii	Mixture of oxidized and pyritic ores, with predominance of pyritic ores	11	Coke	9.0	Oxidizing	Oxidizing	70–74	35	31.98	38.31	6.20
Psh.-Kirovskii	Mixture of oxidized and pyritic ores, with predominance of pyritic ores	11	Coke	9.0	Oxidizing	Oxidizing	70–74	35	30	40	5.27
Psh.-Kirovskii	Mixture of oxidized and pyritic ores, with predominance of pyritic ores	11	Coke	9.0	Oxidizing	Oxidizing	70–74	35	38.15	33.10	4.02
Psh.-Kirovskii	Mixture of oxidized and pyritic ores, with predominance of pyritic ores	11	Coke	9.0	Oxidizing	Oxidizing	70–74	35	31.25	38.13	6.04
Polevskoi	Same	15	Coke	4.5	Oxidizing	Oxidizing	70–75	36	42.94	29.49	3.15
Bogoslovskii	Same	10	Coke	7.5	Oxidizing	Oxidizing	65	38.60	38.57	32.89	2.87
Yulia	Roasted pyrites	4	Charcoal	17	Reducing	Weakly reducing	40	41.50	41.90	27.67	2.84
Ulen	Roasted pyrites	4	Charcoal	25	Reducing	Weakly reducing	40	38–39	41.95	30.66	2.13
Tanaevskii	Pyritic ore, partly roasted	10–11	Producer gas		Shaft furnace in a refractory lining	Weakly oxidizing	25–30	36	12.25	48.25	7.75
Tanaevskii	Pyritic ore, partly roasted	10–11	Producer gas		Shaft furnace in a refractory lining	Weakly oxidizing	25–30	36	13.10	46.75	7.63
Karabash	Pyritic ores	10	Producer gas		Shaft furnace in a refractory lining	Weakly oxidizing	25–30	32–35	12.50	51.10	7.42
Karabash	Pyritic ores	11	Producer gas		Shaft furnace in a refractory lining	Weakly oxidizing	25–30	32–35	12.50	51.50	7.71
Dzhezkazgan	Roasted concentrates	10	Petroleum		Shaft furnace in a refractory lining	Weakly oxidizing	25–30	42	38.26	31.57	3.50
Dzhezkazgan	Roasted concentrates	9	Petroleum		Shaft furnace in a refractory lining	Weakly oxidizing	25–30	42	37.76	31.87	3.80

c) insufficiently complete reduction of the higher iron oxides, in particular ferrites, present in the charge being smelted, owing to the insufficiently strong reducing character of the atmosphere in the furnace or to a shortage of coke in the melting and slagging zone;

c) ferrites are assimilated by mattes from slags of the first period of basic Bessemerizing, which are smelted with ores or poured into the forehearth of the furnace for the purpose of impoverishing them.

6) The quantity of ferrites assimilated by mattes depends on the smelting conditions. The quantity of ferrites or oxygen in mattes does not depend on the specific gravity of the latter; it is directly proportional to desulfurization during the process and inversely proportional to the acidity of the slags formed in the given smelting process.

7) The composition of ferrites dissolved in mattes cannot be established, in view of the fact that no method has yet been developed for the rational analysis of mattes, which are alloys more complex than follows from the phase diagram of the system \(\mathrm{Cu}_2\mathrm{S}:\mathrm{FeS}\). In plant mattes, alongside \(\mathrm{Cu}_2\mathrm{S}\) and \(\mathrm{FeS}\) and their compounds, there is enclosed a considerable quantity of \(\mathrm{CuS}\), \(\mathrm{Cu}\), a small quantity of metallic iron, and possibly the presence of iron sulfides richer in sulfur than \(\mathrm{FeS}\). Simple reduction of mattes does not lead to the goal, since the metallic iron thereby obtained, as well as the metallic iron and copper enclosed in the mattes, enter into reactions with the sulfides of the same metals according to the scheme:

\[ \mathrm{Fe} + \mathrm{Cu}_2\mathrm{S}\;(\mathrm{CuS}) \rightleftarrows \mathrm{FeS} + 2\mathrm{Cu}. \]

8) Of the methods for the analytical determination of oxygen in copper mattes, the Hampe method is the most accurate and, so far, the only one. When determining oxygen by this method, the matte must be finely ground, and reduction must be carried out in a stream of pure, dry hydrogen at a temperature lying below the sintering temperature of the mattes, but not below \(820\text{—}830^\circ\).

To V. Mostovich’s paper “Oxygen in Copper Smelting Mattes.”

Fig. 1. Phase diagram of FeS : Fe alloys according to Prof. Friedrich (K. Friedrich).

For the work of V. Mostovich, “Oxygen in Copper Smelting Mattes.”

Fig. 2. Diagram of the apparatus for determining relative magnetic permeability.

1. Four storage batteries connected in series.

2. Switch.

3. Coarse rheostat.

4. Fine rheostat.

5. Ammeter for 1 amp., with accuracy to 0.01 amp.

6. Spider.

7. Outer coil, 60 mm × 42 mm.

8. Inner coil, 16 mm × 55 mm.

9. Glass tube with the substance under test.

10. Contacts for switching on and short-circuiting (k).

11. Plug rheostat (10,000 ohms).

12. Hartmann-Braun ballistic galvanometer.

13. Telescope tube with scale (scale length 100 cm).

Literature

1) V. Pazukhin. Journal of the Russian Metallurgical Society, 1914, 591.
2) Eng. Min. Journ. LX, 1895, 465.
3) H. O. Hofman. Metallurgy of Copper (1914), 213.
4) Metall u Erz, X, 1913, 858.
5) V. Mostovich. Journal of the Russian Metallurgical Society, 1915, 303.
6) S. Yushkevich. Ibid., 1911, 490.
7) E. Allen and J. Johnston. Journ. Ind. Eng. Chem. 2, 1910, 196.
8) Ledebur. Leitfaden für Eisenhüttenlaboratorien (1903), 73.
9) Friedrich. Metallurgie. VII, 1910, 257.
10) Hampe. Chem. Ztg. 16, 1892, 458. A. Classen. Ausgewählte Methoden d. analyt. Chemie II, 52.
11) Monatshefte f. Chemie. 14, 1893, 315.
12) Metallurgie, IV, 1907, 694.
13) H. O. Hofman. Metallurgy of Copper (1914), 214. E. D. Peters. Principles of Copper Smelting, 413.
14) T. Rickard. Pyrite Smelting, 263. Mineral Industry, IX, 1901, 240.
15) W. Borchers. Kupfer, 76.
16) Metallurgie. VI, 1909, 190.
17) Bull. A. I. M. E. June 1916, 1067.
18) Journal of the Russian Physico-Chemical Society 36, 1904, 1274.
19) Proceedings of the St. Petersburg Polytechnic Institute. IX, 1908, 595.
20) Amer. Journ. of Sci. 33, 1912, 169.
21) See No. 6.
22) Z. f. anorg. Ch. 94, 1916, 25.
23) Metall u. Erz. XI, 1916, 251.
24) Proceedings of the Petrograd Polytechnic Institute 23, 1915, 593.
25) Metallurgie III, 1906, 73.
26) Eng. Min. Journal, 105, 1918, 552.
27) Trans. A. I. M. E. 36, 1906, 665.
28) See No. 6.
29) Metallurgie I, 1904, 8.
30) Metallurgie IV, 1907, 671.
31) See No. 6.
32) Metallurgie VI, 1909, 619.
33) Trutnev. Proceedings of the St. Petersburg Polytechnic Institute, X, 1908, 309. Baykoff et Trutneff. Rev. de Metallurgie, VI, 1909, 519.
34) See No. 32, p. 619.
35) See No. 15, p. 7.
36) Heussler. Verh. Deutsch. Phys. Ges. V, 1903, 219.
37) Metall u. Erz. X, 1913, 447.
38) Metallurgie IX, 1912, 204. Berg u. Hüttenmän. Jahrb. (Wien) 58, 1910, 97.
39) Z. f. physik. Ch. 81, 1912, 129. Z. f. anorg. Ch. 94, 1916, 134.
40) Eng. Min. Journ. 104, 1917, 252.
41) Borchers. Kupfer, 76, 188.
42) Metall u. Erz. VI, 1909, 323.
43) Ber. d. Deutsch. Chem. Gesell. 44, 1911, 1668.
44) Journ. Amer. Chem. Soc. 38, 1916, 807—833.
45) Bull. A. I. M. E. June, 1917, 907.
46) Z. f. Electrochemie 7, 1900, 733.
47) Journ. Amer. Chem. Soc. 38, 1916, 1188.

48) Ber. d. Deutsch. Chem. Gesell. 44, 1911, 1608.

49) Bull. A. I. M. E. June, 1917, 907.

50) Metall u. Erz. VI, 1910, 193.

51) Journal of the Russian Metallurgical Society, 1915, 349.

52) Journ. of the Iron and Steel Inst. 1, 1914, 407—143.

53) See No. 49.

54) See No. 27.

55) Eng. Min. Journ. 60, 1895, 465.

56) Trans. A. I. M. E. 36, 1906, 837.

57) E. Peters. Modern Copper Smelting, 1907, 229.

58) E. Peters. Principles of Copper Smelting, 413.

59) Metall u. Erz XII, 1915, 200.

60) Eng. Min. Journ. 104, 1917, 669.

61) Ibid. 95, 1913, 213.

62) Ibid. 96, 1913, 825.

63) R. Sticht. Ueber d. Wesen d. Pyritschmelzverfahrens, 1906, 30, 31.

64) Bull. A. I. M. E. 1913, 2769.

65) Ibid. 1914, 1785.

66) Liebig. Zink u. Cadmium (1913), 327.

67) Eng. Min. Journ. 104, 1917, 669.

68) Bull. A. I. M. E. 1914, 1785.

69) See Nos. 2 and 15.

70) Metallurg. Chem. Eng. XIV, 1916, 223.

71) Trans. A. I. M. E. 1910, 557. Journal of the Russian Metallurgical Society, 1911, 412.

72) Metall u. Erz. VI, 1909, 323.

73) Metall. Chem. Eng. 11, 1913, 447.

74) Metallurgie VII, 1910, 193.

75) Bull. A. I. M. E. 1916, 1068.

76) See No. 13, p. 174.

77) Eng. Min. Journ. 77, 1904, 757.

78) See No. 13, p. 201.

79) M. Ortin, Journal of the Russian Metallurgical Society, 1912. H. O. Hofman. Metallurgy of Copper (1914), 210.

80) Trans. A. I. M. E. 44, 1913, 751.

81) Bull. A. I. M. E. 1916, 1583.

82) Min. Sci. Press. 116, 1918, 70.

83) Eng. Min. Journ. 106, 1918, 138.

84) Bull. A. I. M. E. 1916, 1583.