the preparation of metallic cobalt by reduction of the oxide

Feb., 1914 T H E JOUR-TTTdL O F I , V D T S T R I A L

divested of many complicating assumptions, and greatly simplified.

I should acknowledge my obligations t o N r . James Graham for assistance in t h e experimental work of this paper.

ChsE SCHOOL O F APPLIDD SCIQNCE CLEVEL.4X.D. OHIO

THE PREPARATION OF METALLIC COBALT BY REDUC- TION OF THE OXIDE' By HERBERT T. KALMUS

I n connection with the work on cobalt i t has been necessary t o prepare considerable quantities of t h e metal in as pure a s ta te as possible. Nearly 1000 pounds of conimercial black cobalt oxide have been given t o this laboratory for these researches by t h e Deloro Mining and Reduction Co. of Deloro, Ontario, t o whom me t ake this opportunity of expressing our thanks.

The writer wishes t o acknowledge t h e x o r k of Messrs. C. Harper, IT. L. Savell, C. T V . Day and R . Wilcox, who, in the capacity of research assistants at these laboratories, have done most of t h e actual experiment- ing. T o Professor S. F. Kirkpatrick of t he Depart- ment of Metallurgy, Queen's University, thanks are due for many valuable suggestions.

The process for t h e preparation of fairly pure cobalt oxide has been very completely worked out, and has been practised on a large scale a t several Canadian smelters. For th i s reason t h e oxide was chosen as a raw material from which t o prepare the metal. As t h e work progressed, i t became more a n d more ap- parent t h a t some of t h e uses for cobalt which were being demonstrated a t these laboratories and elsewhere, would lead t o t h e preparation of t h e metal in large quantities. Hence, i t became of increasing im- portance t h a t t h e metallurgy of t h e preparation of t h e metal from the oxide be studied, a n d this has been done with greater care than was necessary merely for t h e production of t h e quantities required for experimental purposes. There are four impor tan t reducing agents for obtaining metal!ic cobalt in reason- ably pure form from commercial cobalt oxide. They are: I , Carbon; 11, Hydrogen; 111, Carbon Monoxide; IT', Aluminum,

The Co30a used2 for these experiments was made f rom cobalt hydra te , precipitated by bleach from a cobalt chloride solution. This hydrate, i n contact with the atmosphere, is greenish black in color. It was calcined a t 750 ' C., yielding a black oxide of approximately the composition Co304. This is shown by the following analyses, made a t widely different times, which are typical of a large number:

Author's abstract of report under the above title to the Canadian Department of Mines. Published by permission of the Director of Mines, Ottawa, Canada. The general investigation of the metal cobalt and its alloys, with reference t o finding increased commerc;al usages for them is being conducted a t the School of Mining, Queen's University, Kingston, Ontario. for the Mines Branch, Canada Department of Mines.

For a consideration of the yarious oxides of cobalt, including the proof t h a t the black oxide used for these reductions was larselv COJO~. see the

A &IT D E iVGI L\~E E RI N G C H E M I S T R Y 1 0 7

? ANALYSES O F PURIFIED COB.4LT OXIDE ( P E R C C S T A G E S )

June, 1912 November, 1912 April, 1913

co . . . . . . . . . . . . . . . . . . . 71.99 7 1 . 5 2 T7.3 Fe . . . . . . . . . . . . . . . . . . . . 0.11 0 . 2 ; 0 10 S i . . . . . . . . . . . . . . . . . . . . 0.040 0 . 0 2 0 Trace s . . . . . . . . . . . . . . . . . . . . . 0.020 T r x e n. o j i C a . . . . . . . . . . . . . . . . . . . 0.030 . . . 0 .15 SiOz . . . . . . . . . . . . . . . . . . 0.19 n. 1s 0 . i')

The oxides corresponding with the theoretical for- mulas would have'cobalt content a s follows:

Formula Percentage, cobalt Co2O3, . . . . . . . . . . 7 1 . 1 CoaOi.. . . . . . . . . . . . . . . . i . 3 .4 CoaO;. . . . . . . . . . . . . . . . . 7 6 . 0 COO . . . . . . . . . . . . . . . . . . i 8 . 8

It is obvious then, when we t ake into account t he portion of the sample which is no t cobalt oxide, t h a t t he oxide itself is largely C0304. It is not necessary for t he purpose of our calculations to assume t h a t this oxide alone is present, for we shall base our computations upon t h e actual analyses as we have found them. However, i n writing t h e reactions throughout this paper, we shall, for simplicity, consider t he oxide t o be Co304.

P U R I F I C A T I O K O F COBALT OXIDE

Cobalt oxide as we obtained it from the smelters, and as sold on the market, analyzed approximately as follows:

Barrel 1 Percentages Co . . . . . . . . . . . . . . . 70.36 Ni . . . . . . . . . . . . . . . 1.12 Fe . . . . . . . . . . . . . . . 0 . 8 2 s. . . . . . . . . . . . . . . . 0 . 4 5 As . . . . . . . . . . . . . . . 0.10 S i 0 . . . . . . . . . . . . . . 0.20 Ca . . . . . . . . . . . . . . 0.50

Barrels 3 and 4 Percentages Co . . . . . . . . . . . . . . 6 9 . 2 Ni.. . . . . . . . . . . . . 1 . 4 Fe . . . . . . . . . . . . . . 0 . 5 0 CaO . . . . . . . . . . . . . 0.37 s . . . . . . . . . . . . . . . 0.54 Insoluble. . . . . . . . 1 . 4 6 Ag. . . . . . . . . . . . . . Trace

Analyses, of course, vary considerably from one shipment t o another ; t h e above samples are high in Fe , S and Ca, and would be considered by most smelters as No. 2 grade.

Metal produced from oxide analyzing as above, by the method t o be described. is of sufficient pur i ty for most purposes. This is especially t rue if lime be added t o t h e melt t o slag off t h e sulfur. HoTverer, for other purposes metal is required in which the impurities, nickel, irbn, sulfur, arsenic and silica, are reduced t o very small percentages. I n this case it is best t o remove these impurities from the oxide before reduction Starting with a crude cobalt oxide, these impurities may be reduced as far as is desired by t h e following procedure:

sILIca-Dissolve the crude oxide in hydrochloric acid according t o the reaction:

CoaOl + 8HC1 = 3CoC12 + qHzO + C12 This may be done best by heating a n d agitating with steam. If silica is present, i t will not dissolve, and may be removed by filtration or decantation. The same is t rue of silicates which are not decomposed by this t rea tment . Decomposable silicates would send a certain amount of silica into solution. which would be thrown out during the next step.

IROS ASD ARSESIC-TO the cobalt chloride solution - . following article, page 115. formed by dissolving t h e oxide in hydrochloric acid,

I08 T H E J O U R N A L O F I N D C S T R I 4 L

gradually add finely divided CaC03 or pure marble, until no further precipitate is formed. The heavy brown mud precipitated contains the iron and arsenic content of t he original oxide.

NICKEL-For most purposes it will not be necessary t o separate the small amount of nickel from the cobalt, bu t i t may be done as follows: The cobalt chloride solution, containing a certain amount,of nickel chloride, is of an intense red or claret color. Add a solution of bleach t o the solution until i t has almost completely lost i ts color. The bleach solution differentially pre- cipitates hydrates of nickel and cobalt, so t h a t t he nickel is not appreciably brought down until t he cobalt has been almost entirely precipitated.

The bleach will precipitate a black, hydrated oxide of cobalt, and the diminishing redness of t he solution will indicate the end point. If all of t h e steps above outlined have been applied t o t he original oxide. this final black precipitate may be calcined a t about 7 5 0 " C., t o yield black Co304.

SULFUR-Any sulfur present in the original oxide and carried through t o t h e final product, or introduced with the bleach, may be removed by boiling the final dried oxide with sodium carbonate and dilute hydrochloric acid. The reaction is:

Cas04 + NaZC03 = NaZS04 + CaC03

The soluble sodium sulfate formed is washed out with water. A further washing is given with dilute hydrochloric acid, which decomposes the calcium carbonate in to soluble calcium chloride and COZ gas. The CaC12 is washed out with water. This method is, of course, applicable only for t h e removal of t h e small percentages of Ca and S found in the oxides in question.

A shipment of oxide from the smelter was analyzed before and after t rea tment by the above method, with t h e following results:

Percentages Before C o . . . . . . . . . . . . 70.36 Ni . . . . . . . . . . . . . 1.12 F e . . . . . . . . . . . . . 0 .82 s . . . . . . . . . . . . . . 0.45 Ca . . . . . . . . . . . . 0.50 As . . . . . . . . . . . . . 0.10 SiO?. . . . . . . . . . . 0 . 2 0

After 71.99 0.041 0 . 1 1 0.020 0.021 None None

There are other obvious methods of purifying the Co304. For example, t he bleach soiution may be freed of i ts SO4 content with BaC12, and the Ca and excess Ba precipitated with Na2C03, thus yielding a fairly pure solution of soda bleach. The so4 content of t he CoC12 solution may be precipitated with BaClz and the differential precipitation of cobalt and nickel accom- plished with the purified solution of soda bleach.

I-REDUCTIOK O F C O B A L T OXIDE WITH C A R B O N

METHOD OF EXPERIMENT-These experiments all consisted in intimately mixing definite amounts of finely divided carbon in various forms with Co304, and heating the mixture t o constant temperature for a measured time. The charges employed varied in size from a few grams t o I O lbs., and were heated in lined and unlined graphite crucibles, and in porcelain crucibles.

FURNACES-The reduction took place either in a n

A N D E N G I N E E R I N G C H E M I S T R Y Vol. 6 , S o . 2

oil-fired "Steele-Harvey" furnace of 60 pounds, metal capacity, No. 2 0 crucible, which could be controlled a t any temperature up t o 1550' C., or in a modified Hoskins electric resistor furnace. This latter has a heating chamber, 8 inches cube, which can be maintained constant t o within about 1 0 - 2 0 ' C. , a t any temperature up t o 16 jo' C. Some of the small charges were run in porcelain crucibles heated within a n electric resistor furnace.

The reactions for t he reduction of cobalt oxide with carbon are:

(1) c0304 + 4 c = 3 c 0 + 4 0 , and ( 2 ) c0304 f 4 C O = 3Co + 4'202;

or, combining (3) zCo304 + 4C = 6Co + 4CO2. If all t he oxygen for t he oxidation of t he carbon be

supplied by the cobalt oxide, and if all the carbon be burned t o C o t then the reaction goes according t o the last equation.

I n practice, neither of these conditions is strictly obtained, b u t with proper design of furnace they may be closely approximated.

THE RUN-In each case the charge was made up by intimately mixing a weighed amount of finely divided oxide with a weighed amount of finely ground carbon. This mixture was placed in the crucible, which, with i ts charge, was placed either in the Steele-Harvey oil furnace or in the electric furnace. The mixture was frequently stirred with a n iron rod during the reduction.

THE CARBON-The form of carbon chosen for the reduction, whether powdered charcoal, coke, coal, etc., depends somewhat upon the impurities f rom which i t is desirable t o keep the resulting metal free, bu t also this choice greatly influences the speed of t h e reduction.

Three sets of experiments were made with powdered anthracite coal, while further runs were made with lampblack or with powdered charcoal. The carbon was in all cases powdered t o a n extremely fine flour.

ings were made a t frequent intervals with a platinum platinum-rhodium thermo-element, with a Wanner optical pyrometer, or with a FCry radiation pyrometer, and the furnace adjusted t o keep the temperature constant t o within about 20'.

The charge was pu t into the crucible which was within the furnace, both crucible and furnace being a t a temperature somewhat higher than the intended temperature of t h e run. Some of t he smallest charges were inserted with containing crucible. I t was learned, by experience, for the different sizes of charge and qualities of crucible, a t about what temperature t o maintain the furnace prior to inserting the charge, in order t h a t t he charge might come t o the desired equi- librium temperature, with proper furnace adjustment , in about ten minutes. There is, therefore, a period of about ten minutes, a t t he beginning of each run , during which the average temperature of the charge is not as high as t h a t noted with the Wanner optical pyrometer, which observes the surface of t he charge. We satisfied ourselves t h a t the center of the charge was a t t he same temperature as t he surface, within 20 or 30' C., after t he first t en minutes, by exploring the center with a thermo-element, and noting simul-

TEMPERATURE uEASUREMENTS-TemperatUre read-

Feb . , 1914 T H E J O C R N A L O F I N D U S T R I A L AiVD E N G I A V E E R I N G C H E M I S T R Y

taneously i t s readings and those of another thermo- element near t he surface. and of the Wanner optical pyrometer.

I n t h e following runs we have not a t tempted t o make a correction for t h e lag in coming t o temperature during these first t en minutes. This lag would be considerably less t h a n t en minutes for t h e smallest crucibles used, about t e n minutes for t h e four-pound charges. and possibly as long as twenty m h u t e s in the worst cases. with t h e ten-pound charges.

T h e oxides used for t he runs reported in Tables I and I1 analyzed as follows:

Percentages

Runs A-H R I - R V I I I Co . . . . . . . . . . . . . . . . . . . . . . 71.36 69 .2 S i , . . . . . . . . . 1 . 4

0.50 0.54

CaO. . . . . . . . . 0.37 Insoluble. . . . 1 .46

Si02 . . . . . . . . . . . . . . . 0 . 2 0 Ag . . . . . . . . . . Trace

The anthracite coal used was very finely powdered. I n t h e typical runs A-H, No. 12 unlined carbon crucibles mere used, and the charge was stirred every ten minutes during reduct ion . . I n these runs no

TABLE I-REDUCTIOX OF Cos01 WITH, POWDERED ANTHRACITE COAL (4) R u s s A. C, H-HARVEY-STEEL OIL FURNACE G, B, D, E, F-ELECTRIC CRUCIBLE FURNACE

Yield of cobalt . Charge Average Time

h-0, -- tempera- of reduc- Per cent Per cent of Cor04 Coal ture tion of theo- carbon in run Lhs. 0 2 . C. Min. 1.b. Oz. retical metal A , . . , . . 5 8 . 3 1200 90 3 1.5 87 ... C . . . . . . 4 6 .9 1200 30 2 12.5 98 0 .18 H . . . . . . 10 lh.O 1200 105 6 9 92 0.086 G , . . , . . i 6 . 6 900 150 2 13 99 0 . 2 1 B . . . . . . 4 6 . 5 4 1200 60 2 13 99 0 .29 D . . . . . 4 6 . 9 1200 120 2 13.5 100 0 .20

, . 4 6 . 9 1500 60 2 11.75 96 0 . 2 2 , . 4 6 . 6 1500 90 2 12.7 98 0 . 2 3

A-Considerable unreduced oxide slag. Carbon used is approxi-

C-Melt free from unreducible oxide slag. Carbon 10 per cent in

H-3 oz. lime added shortly before pouring. Carbon, theoretical

G A t end of 2.5 hours, charge not completely reduced, but completed

mately the theoretical amount according to reaction (3).

excess of theoretical requirement.

amount.

during subsequent raising to melting point.

TABLE 11-REDUCTION OF Cos04 WITH POWDERED ANTHRACITE COAL (B) HARVEY-STEELE OIL FURNACE

Reduction Charge time to

No. ___7 Average removal Per cent of CoaOd Coal temp. of sample C o i n Reduction

run Lbs. 0 2 . C. Minutes sample@) complete = 100 I.. , . 10 1 7 . 5 601 R2 82 7 3 , 6 Very slight

I1 . . . . 10 16.5 750 R3b 91 74.1 Very slight I11 . . . . IO 16.5 888 R4b 90 73.8 Very slight I V . . . . 10 16.5 1057 R5b 95 80.8 28 VI . . . . I O 16.5 1203 R6a 30 81 .3 30

R6b 49 93 .O 74 VI1 . . . . IO 16.5 1283 R7a 31 91 .1 7 0

R i b 47 93.9 I ,

VI11 . . . . 10 17.4 1502 1 11 76.7 12 2 16 81 .8 32 3 21 91 .O 66 4 26 93.9 7 7 5 31 93 .9 Apparently some 6 36 91 .8 oxidation

(a)These analyses are for cobalt, nickel and iron combined. of which about 97 per cent was cobalt, as may be seen from the analysis of the original oxide. Carbon analyses were of course made, and the percentage of cobalt given in this column takes into account the residual carbon.

- _

a t t empt was made to show the progress of the reduction, b u t a t t h e close of t h e run t h e charge was raised as rapidly as possible t o the melting point and the melt poured into a n iron mould t o be weighed. Considerable reduction must t ake place during the interval of melting the charge after t he close of t he run. The purpose of these particular runs was to s tudy the yields under somewhat the same conditions which must necessari y obtain in practice. I n the runs . I-YIII, KO. 2 0 unlined carbon crucibles were used; no a t t empt was made to obtain a yield. They are intended t o show t h e progress of the,reduction.

It will be noticed in the above runs with powdered anthracite coal t h a t t h e reductions are extremely low. I t was, therefore, thought advisable t o check these runswi th experiments on a very small scale in porcelain crucibles, in such a manner t h a t there could be no doubt as t o the t ime during which the charge was maintained a t t he temperature in question.

A number of such runs ' was made with a thermo- element near t he center and a t t he outside of t he charge. I n the small furnace used, t he crucible with i t s charge came to temperature in a very few minutes, so t h a t t h e outside a n d inside thermo-element agreed t o within 2 0 ' C. A4pproximately this condition was maintained throughout t he run.

The results of the previous runs with powdered anthracite coal were confirmed by these small scale runs, and a satisfactory complete reduction could not be obtained at temperatures much below I Z O O O C.

Fur ther experiments were tried on the reduction of CoaO4 with very finely powdered charcoal. A large number of these gave fairly concordant results, which showed a greater reduction a t all temperatures than the corresponding powdered anthracite coal runs.

Without giving the details of about twenty-five runs, it may be said t h a t complete reduction was obtained with from 2 0 t o 30 per cent excess of powdered charcoal, a t goo0 C. or over, in less t h a n a n hour. ,4t IOOC-1100' C., t he reduction with powdered charcoal was very much more rapid than a t 900' C.. often completing itself in less t han I O minutes. Of course, t h e t ime required depends, t o some extent. upon t h e size of t he furnace and charge.

R E D U C T I O N O F cos04 W'ITH P O W D E R E D CHARCOAL

R E D U C T I O N O F cos04 W I T H LAUPBLACK

Experiments on t h e reduction of C0304 with lampblack were tr ied with results identical with those on the reduction of Co304 with powdered charcoal.

BRIQUETS-Experiments on the reduction of cos04 with powdered charcoal were tried, forming t h e charge into briquets. A small' percentage of molasses was used as a binder. These experiments were made under t h e same furnace and temperature conditions as those on the reduction of cos04 with powdered charcoa in bulk. Seven such runs showed, throughout, t h a t t h e reduction was not very different in i ts velocity from the corresponding runs with powdered charcoal. although t h e difference was uniformly in favor of t h e briquetted charges. A satisfactory reduction

1 The molasses would correspond to the addition of about 1 per cent carbon.

I IO T H E J O U R N A L O F I i V D C S T R I A L A N D E;t'GI,VEERING C H E M I S T R Y Yol. 6 , No. 2

could probably not be made a t temperatures below 8 o c A j o 0 C., even briquetting the charges, as compared with goo0 C. for t h e charges in bulk.

The preparation of metallic cobalt b y reduction in briquets or rondelles offers distinct commercial advantages in t h a t t h e resulting metal is i n a salable form without further melting and casting.

THE METAL-The metal produced by reduction of cobalt oxide with carbon is sufficiently pure for most purposes; it need not contain more t h a n a few ten ths of a per cent of carbon. The following characteristic analyses are taken a t random from a large number t o show t h e na ture of t h e metal:

,

A N A L Y S E S O F I\IETALLIC C O B A L T P R O D U C E D B Y REDUCTIO~; O F COMMERCI.4L

COBALT OXIDE WITH CARBON (PERCESTAGES) 7-12 8-15-12 10-10-12 10-11-12 6-11-13

Co . . . . . . . . . . . . . . . . . . . . . 97.05 98 .50 98.84 98.62 98 .30 1.50 0.65 0.61 0 .50 Trace 1.00 0 .58 0 . 5 6 0.15 1.39

S . . . . . . . . . . . . . . . . . . . . . . . 0 .22 0 .47 0 . 2 1 0 .22 0 .46 C . . . . . . . . . . . . . . . . . . . . . . 0 . 2 0 . 0 . 2 2 0 . 2 4 0.13 . . . Ca . . . . . . . . . . . . . . . . . . . . . 0 .25 0 . 6 0 0 . 2 4 0.27 0 . 5 8 h ln . . . . . . . . . . . . . . . . . . . . . . . . . . . 0.06 . . . . . . As. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0 . 1 1 Si02 . . . . 0 .12 0.13 0.14 0 . 1 1 0 .12

It is obvious t h a t t he oxides taken for reduction with carbon n-ere those directly from t h e smelter, which had not been treated by the method outlined above t o remove the impurities. The iron, nickel, sulfur a n d silica content could have been reduced t o mere traces by purifying the oxide before reduction in accordance with the method given. We have done this repeatedly, where a pure metal was required for experimental purposes. I t is, however, of import- ance t o note t h a t metal with very low carbon content may be made by direct reduction of t he oxide with carbon.

C O X C L U S I O N S

I . Reduction of c0301 with powdered anthracite coal does not t ake place rapidly enough t o make i t commercially interesting, either i n the oil-fired crucible t y p e of furnace or i n t h e electric crucible type of furnace, until a tempera ture in the neighborhood of 1200' C. is reached.

11. I n either t he oil-fired crucible type of furnace or in t h e electric crucible type of furnace, substantially complete yields of metallic cobalt may be obtained by reduction of Co304 with powdered anthracite coal, i n t h e neighborhood of 1 2 0 0 ~ C., for no t more than I hour, with subsequent rapid melting and pouring.

111. With the oil-fired crucible furnace, using unlined graphite crucibles, complete yields are obtained with powdered anthracite coal only when there is a n excess of approximately IO 'per cent of this latter.

IV. With the electric crucible type of furnace used b y us, complete reduction may be obtained, using only t h e theoretical quant i ty of powdered anthracite coal. I n this furnace there is a considerable reduction due t o t h e carbon monoxide atmosphere caused by t h e carbon resistor plates. V. Both in the oil-fired a n d in t h e electric crucible

type of furnace, greater reductions of Co304 are obtained using pohdered charcoal t h a n with powdered anthracite coal, a t corresponding temperatures.

VI . Wi th the oil-fired or electric crucible type of furnace, complete reduction may be obtained with powdered charcoal at goo0 C. or higher. For this reduction a considerable excess of charcoal mas required: under our conditions from 20 t o 30 per cent.

VII . Powdered lampblack shows results in accordance with those for powdered charcoal.

VI I I . Briquetting the charges with a n organic binder tends t o increase t h e r a t e of reduction a t all temperatures. A minimum of about 800' C. may be employed for t he reduction of Co304 with charcoal in the form of briquets as against *gooo C. for t h e same charge in bulk.

I X . With sufficient carbon t o get a complete yield of metal, the final product need contain only about 0 . 2 per cent of carbon.

X. At this laboratory, in a n electric furnace not especially designed for this work, we reduce enough oxide to make j 6 pounds of t h e metal i n a n eight- hour day , with t h e furnace absorbing 1 2 kw. Thus , on a commercial basis, t he power charge for this reduction would be small.

11-PREPARATION O F M E T A L L I C COBALT B Y R E D U C T I O N

O F THE O X I D E WITH H Y D R O G E N

METHOD A N D APPARATUS.-These experiments consisted in placing a n a lundum boat , containing a weighed amount of dried cobalt oxide, in a horizontal tube electric resistor furnace, maintaining i t s temperature therein constant for a definite length of t ime, during which a stream of hydrogen gas was passed through t h e furnace. A schematic sketch of t h e appara tus i s shown in Fig. I.

After purification, t h e hydrogen entered t h e furnace ,l Y

u x w

FIG 1-ARRAKGEMENT OF APPARATUS FOR REDUCTION OF Cos04 BY HYDROGEN

H = Hydrogen Tank CI and Cz = Hot Copper K = KzCrz01 Tower

0 = KOH Tower S = HzSOc Washer J = Carbon Rings

Y =Stop Cock L = Leads t o Ammeter

and Bus Bars

at F a n d the excess was burned at B. During the run t h e exit for t h e gas was through the by-pass PB, t h e end Q being sealed.

The heating element of t h e furnace itself was a series of co-axial carbon ring plates, which could be pressed together more or less t ightly b y suitable screws. The furnace was supplied with alternating current at 2 j volts from a transformer, a n d could be controlled at a n y temperature up t o 1350' C.

The .de ta i l s of t he furnace are shown in Fig. 2. The temperature measurements were made by a

Feb., 1914 T H E J O C R S d L O F I N D L - S T R I A L A S D E S G I S E E RI S G C H E JlI S T R k’ 1 1 1

platinum-rhodium thermo-element T h , and readings were taken a t frequent intervals on a very sensitive millivoltmeter. I n this v:ay, t h e temperature was maintained substantially constant b y hand regula- tion of t h e screws I. All temperature measurements were made with thermo-elements calibrated a t frequent intervals, in the usual way, against k n o x n melting points.

C O S D U C T I N G A Rus-After having heated t h e furnace t o the desired temperature. b y a suitable current through t h e carbon rings, runs were made as follows:

( a ) Xfter closing the cock X , which separates t h e purifying system from t h e furnace, t h e air was exhausted from the hydrogen system by opening cock Y , and operating a pump.

( b ) Gas burners were lighted t o heat copper filings in tubes C1 and C1.

(c) Solutions of potassium bichromate, potassium hydra te a n d sulfuric acid were s ta r ted flowing through the purifying towers K, 0 and S; which were partially filled with glass beads.

FIG 2-DETAILS OF ELECTRIC FURNACE FOR REDUCTION O F COaO4 WITH HYDROGEN

( d ) Cock Y was closed and cock a t outlet of hydrogen t a n k H was partially opened to a l lox a flow of hydrogen into the purifying system, until t he pressure inside the system was a little greater t h a n atmospheric pressure.

( e ) Cock X was now opened t o allow hydrogen t o flow into the hot furnace.

(f) Flow of hydrogen was adjusted by cock at outlet of hydrogen t ank H, until hydrogen burned freely a t outlet end of furnace B. During t h e run the end Q ~ - n s closed, and t h e gas escaped through the by-pass P-B. (9) When adjus tment u) was satisfactory, assuring

a n excess of hydrogen within t h e furnace, t he weighed dried a lundum boat , containing the charge of cobalt oxide, mas placed in t h e hot furnace a t t he position A, a n d t h e t ime noted.

( h ) The run proper h a d now begun, durirfg which observations of t ime, temperature, and power were made, a n d the furnace adjusted t o keep t h e tempera- t u re constant.

(i) After a definite time, t he boat, with i t s contents, was withdrawn from the centre of t he furnace t o the overhanging cool end of t h e furnace core nT, in which i t was allowed t o cool, bu t through which, during the cooling, hydrogen was passed.

( j ) When cool, t h e boat was removed t o a desiccator and weighed.

11-C 0 B A LT 0 X I D E R E D U C T I 0 iX 11-1 T H H Y D R 0 G E S- C 0 0 L- I N G IS -41- A T M O S P H E R E O F H Y D R O G E K

The cobalt oxide used for t h e follon-ing runs, I to I X , analyzed as follows in percentages:

C o . . . . . . . . . . 72.3 h- i , . . . . . . . . . . Trace Fe , . . . . . . . . . . 0 . 10

Ca . . . . . . . . . 0.15 S . . . . . . . . . . . 0.052 S O ? . . . . , . , . . 0 . 3 9

I t will be noticed t h a t this oxide contained 72.4 TABLE 111-REDUCTION OF COBALT OXIDE WITH HYDROGEN

Temperature C. Number I . Time of

Run I

II@)

I11

I V

5‘

VI

VII(e)

17111

IX

Boat Alean I

II(a) I

I1

T

I1

IV(c)

I ( d )

I

I1

I

I1

I

I1

I

I1

583

584

609

597

598

i 2 i

824

965

10i3

Average deviation

1 3

1 0

4.2

2 . 0

1 .7

1 5

4 . 3

1 0

2 . 1

reduction Minutes

5 15 5

15

60 15 30 60

15 30 60

15 30

15 70 15 60 60

120 5

10 30 60 5

10

30

,

30 2 5 5

15 30 60

150 2 5 5

15 30 60

1 5

30 60

1 5

30 60

2 5

30 60

1 5

30 60

Loss in weight

Per ren t 25.5 25.4 25 .8 26.0 26 .6 26.2 26.1 2 6 . 2 26.1 16 .8 22.1 22 .4 25.2 25 .3 25 .4

25 .4 2 5 . 2

25 ..i 25.5 25.8 2 5 . 7 26 .5 2 6 . 5 26 .6 2 6 . 6 26 .6 26 .5 2 6 . 6 26 .6 26 .6 2 6 . 7 26.75 26.75

26.61 26.71 2 6 , i 5 26.75

25.6 26.9

‘ 2 i . I 2 i . l

2 6 . 9 2 6 . 9 27.0 26 ,?8 26.90 2 7 . 0 0 27.05

26.90

2 7 . 10

. .

. .

26.8

2 6 . 80

2 6 . 8

24.38

27.00

Reduction 100 =

complete 94.2 9 4 . 0 9 5 . 6 9 6 . 2

9 7 . 0

97 .0

62 .2

98 .3

9 6 . 8

96.8

81.8 a3 0 9 3 . 0 9 4 . 0 94.2

9 4 . 2 94 .1

91.3 94 .3 95.1 95 .3

. .

. .

9 8 . 1 9 8 . 1 9 8 . 5 98 .5 98.5 98 .1 98 .5 98.5

9 8 . 8

98.5 98 .9

Y8,8 9 9 . 0 98 .4 98.1 ‘18. a 9 8 . 8 99 . 0 9 4 . 8 9 9 . 7

100 .0

99 .1 99 .4 9 9 . 4 99,1

99 .4 9 9 . 7 9 9 . 9 90 .0 99.; 9 9 . 8

100 .0

100.0

96.8

( a ) It >vas noted a t the close of this run tha t there was a slight oxidation

( b ) All the reduced samples were steel-gray. (L) Boatshowed slight reoxidation a t one end whenremoved from furnace. ( d ) This final material analyzed 97.25 and 97.30 per cent cobalt on dupli-

cates. T h e material resulting from this run contained 0.75 per cent of unreducible CaSOI. CaO and SiOz, and 1.4 per cent of oxygen presumably in the form of c o s 0 4 (the stable oxide a t 598O C., see following article, page l l j ) , and 0.10 per cent of nickel and iron. It should, therefore, contain 100 - 2.3 = 97.i per cent of cobalt. This checks with thpvalue determined by analyses, 97.3, to within the accumulative error in the analyses.

( e ) The product from this run seemed to be of a slightly lighter gray shade than tha t from the runs a t lower temperatures.

a t one point in the boat.

*

I 1 2 T H E J O U R N A L OF I N D U S T R I A L

per cent of t he metals cobalt, nickel and iron, in the form of oxides which may be computed without error t o be cobalt oxide. Any sample‘ contains, therefore, 0.75 per cent of unreducible calcium sulfate, calcium oxide and silica, 99.2 per cent of cobalt oxide running 72.4199.2 = 72.9 per cent in cobalt. This oxide, therefore, corresponds very closely t o Co304. The oxygen content of t h e substance which could be reduced by hydrogen, is equal t o 27.1 per cent of 99.2 per cent = 2 7 . 0 per cent. This figure is accurate t o within t h e experimental error of t he runs, and is used as t h e basis of t h e following computations; t h a t is t o say, in t h e column headed “Percentage loss in weight,” 2 7 per cent would represent complete reduction, and the last column headed “Reduct ion where IOO per cent is complete reduction’’ is computed in te rms of 2 7 per cent actual reduction as total . The boats used ranged in weight from 5 t o 6.5 grams and t h e charges of cobalt oxide from 2 . 0 t o 2.1 grams.

The check between the composition of the oxide used for these hydrogen reduction experiments, as determined by analysis and as determined b y the reduction experiments. is entirely satisfactory (see following article, p. 11 5 ) .

A number of t h e early experiments t o reduce cos04 with hydrogen were made allowing the reduced product t o cool in t h e atmosphere. In every case reoxidation took place. These runs were made a t various temperatures from 500’ C. t o 1000’ C., and curiously enough the reoxidation a t t h e higher temperatures was progressively less t h a n a t the lower temperatures.

I. The reduction of Co304 t o metallic cobalt by hydrogen gas takes place very rapidly a t all t em- peratures above 500’ C.

At temperatures between 500’ C. and 700’ C., over 90 per cent of t he reduction of C0304 t o Co takes place in a few minutes, bu t a further reduction takes place very slowly, if a t all.

Between 700’ C. and 1100’ C., t h e amount of reduction of Co3O4 t o Co which takes place during the first few minutes increases very rapidly with rising temperature , and a t the higher temperatures i t is complete.

IV. The hydrogen reduction method is t o be especially recommended for t he production of moderate quantit ies of. very pure carbon-free cobalt for special purposes, just as i t has been used for the production of metallic tungsten.

V. For the production of cobalt from Co304 by hydrogen, t he charge must be completely cooled in an atmosphere of hydrogen.

C O N C L U S I O N S

11.

111.

111-PREPARATION O F M E T A L L I C C O B A L T B Y R E D U C T I O N

O F T H E O X I D E IVITH C A R B O N M O K O X I D E

M E T H O D A N D APPaRATns-These experiments were performed by placing an alundum boat, con ta inhg a weighed amount of dried cobalt oxide, in a horizontal t ube electric resistor furnace, maintaining its temperature therein constant for a definite length of time, during which a s t ream of carbon monoxide was passed through the furnace (see Figs. 3 and 4) .

C A R B O N M O N O X I D E GENERATOR-The carbon mon-

A N D E N G I N E E R I N G C H E M I S T R Y Vol. 6, No. 2

oxide was supplied by passing carbon dioxide over hot wood charcoal, which reduced it according t o the reaction, COz + C = 2CO.

Carbon dioxide, after purification, entered the lower end of t he carbon monoxide generating furnace a t G.

F I G . 3 - - A P P A R A T U S FOR REDUCTION OF coa04 W I T H C A R B O N MONOXIDE

This furnace was of t he electric resistance type , made by winding nichrome wire over a n alundum cylinder, t he two being embedded in magnesite cement, and insulated within a cylindrical iron container. The wire is shown in section a t H and the iron container a t I. T h e entire core of t he furnace was filled with wood charcoal, maintained a t about 1000’ C. by an appropriate current through the heating element. As a result, carbon monoxide gas left t he generator a t J , with a certain amount of moisture which was ab- sorbed by passing through calcium chloride a t K. Thus, substantially pure carbon monoxide entered the reaction furnace proper a t L, passed over t he alundum boat 11, with i ts cobalt content, and the excess burned off a t N .

proper is shown in Fig. 3. It consisted of a central silica tube L N , 2 ft. in length, and I in. in internal

THE REACTION FURNACE-The reaction furnace

FIG. 4

diameter. This was wound with calorite or nichrome wire, of such resistance t h a t i t could be controlled by a suitable rheostat R h on I I O volt direct current mains, t o maintain a temperature constant t o within less t h a n IO’ during a run , a t any temperature up t o I0OO0 c.

Feb., 1914 T H E J O U R N A L O F I N D U S T R I A L A N D E N G I N E E R I N G C H E M I S T R Y 1 I3

T E M P E R A T U R E MEAsvREMEKTs-The temperature measurements were made by a platinum platinum- rhodium thermo-element T h , and readings were taken at frequent intervals on a very sensitive millivoltmeter Mv. The thermo-elements used for t he temperature measurements were calibrated a t frequent intervals.

-The silica tube N L extended beyond the end of t he furnace proper from 0 t o N. The portion 0-N was I f t . i n length and was kept cool by a circulation of water, so t h a t a t t h e close of a run the boat was removed from t h e centre of t h e furnace t o 0-N, where i t cooled t o room temperature in the stream of CO gas.

C O N D U C T I N G A RUN-After having heated t h e reaction furnace, and the CO producer furnace, by suitable currents, t o t he desired temperature, runs were made as follows:

( a ) The cock C was opened t o allow C02 gas t o pass through the purifying system D E F into t h e producer furnace a t G.

The CO gas generated in the producer furnace GJ passed through the reaction furnace L N and was lighted at N.

(c) The weighed dried boat with i ts charge was introduced into the exterior O N of t he reaction furnace.

i d ) Temperature and t ime observations were begun, and when the desired temperature had been reached, t he boat was moved t o M.

The run proper had now begun, during which observations of t ime and temperature were made and the rheostat R h adjusted t o keep t h e temperature constant.

(f) After a definite time the boat, with i ts contents, was withdrawn from M t o O N , where i t was allowed to cool in a current of carbon monoxide gas.

When the boat was cool, t he current of carbon monoxide was gradually diminished by closing the cock C, until it was finally entirely cut off.

( h ) l17hen the boat was cooled t o room temperature, i t was removed to a desiccator and weighed.

C O O L I N G C H A R G E I N C A R B O N M O N O X I D E ATMOSPHERE

( b )

( e )

(g)

C O B A L T O X I D E FOR C A R B O N M O K O X I D E R E D C C T I O X

EXPERIBIESTS

The cobalt oxide for t he CO reduction experiments was identical with t h a t used for t he hydrogen reduction experiments so t h a t the column headed “Reduction where IOO per cent is complete reduction” is computed in terms of 27.0 per cent actual reduction as total. A number of the first runs were made, reducing

c0304 with CO and allowing the reduced product t o cool in the atmosphere before weighing. Under these conditions, reoxidation took place rapidly, so tha t bu t a single pair of typical runs are given. The boats used weighed IO. j+ and 13.3 grams and the charges of cobalt oxide, 2- grams.

The reoxidation of cobalt oxide after reduction

with carbon monoxide takes place with great vigor. If t he boat be withdrawn from the hot furnace directly into t h e atmosphere, i t may be seen t o glow with great brilliancy. If t he content of t he boat, while still warm, be snapped out on the floor, i t will re- oxidize with such vigor t h a t a cracking sound, as of a mild explosion, attends the reaction, i. e . , the reoxidation taking place according t o the reaction 6coo + O2 = zCo301 is extremely exothermic.

I n the runs of Table IV , during the first par t of t he run , and up t o the t ime t h a t i t began t o gain in weight,

TABLE IV-REDUCTION OF Cor04 WITH CO-COOLING IN AIR

Temperature Number - Time of Per cent Reduction

c_*___ Average reduction loss in 100 = Run Boat Mean deviation Minutes weight complete

I 1 602‘ 9 15 10 .6 39 .3 30 1 1 . 1 41 .1 45 1 0 . 8 3 9 . 9 60 12.9 47 .8 75 12 .9 4 7 . 8 82 1 1 . 8 4 3 . 6 92 11 .3 4 1 . 8

107 7 . 8 3 9 . 6 I1 594 12 30 12 .4 4 6 . 0

45 11.: 4 3 . 3 75 1 3 . 2 4 8 . 8 9: 12.3 4 5 . 5

112 11.9 4 4 . 0 142 10 .3 3 8 . 2

t he Co304 in both boat; gradually became a gray color. This gray material is COO. -4t the end of t he run i t was black again.

On account of t he irregularities of reoxidation, the furnace reaction chamber was lengthened by sub- stituting a silica tube of length L N for the one of length LO as shown in Fig. 3. The overhanging tube O N , about I ft. in length, was cooled by water, and served as a cooling chamber for t he boat while CO gas x a s still passed through it .

The following runs, representative of a large number, shorn the ra te of the reduction of cobalt oxide by CO gas when the cooling was controlled so tha t no reoxidation could take place. The boats used weighed about 8 grams and the charges of oxide about z grams.

I n R u n 111, Table T‘? t he oxide became a greenish gray color a t t he end of the first five minutes, and a uniform steel-gray color a t t he end of fifteen minutes. From then on it began t o gain in weight, due t o a deposit of carbon. A t t he close of many runs a t this temperature, there was an extremely heavy deposit of carbon in the boat.

This run was typical of a number, which showed a reduction from the original black oxide to the green, followed by a change from the green to the gray, and then a gain in weight, due to a deposit of carbon. The only possible source of carbon was from the carbon monoxide gas, so t h a t t he finely divided metallic cobalt, which mas formed during the first stage of the reduction of t he gray oxide, probably acted catalyti- cally t o decompose carbon monoxide gas a t this temperature. This is an extremely interesting decomposition which might well be studied with considerab1.e care.

e

1 I 4 T H E JOL'RA'AL O F I N D C S T R I A L

T A B L E v--REDCCTIOh- O F COBALT OXIDE BY CARBON &?OXOXIDE GAS- COOLING IN CO Gas

Piumber r . Time of Loss in Reduction

Run Boat 3Iedn deviation Minutes Per cent complete

Temperature C.

w Average reduction weight 100 =

I11 34; 3 5 18.2 6 i , 3 15 2 5 . 2 9 3 . 2 45 Deposit of . . . . 60 carbon . . . .

IV I 451 3 5 2 2 , l 81.8

v I 453 3- 5 2 1 , 4 7 9 . 2 I5 1 3 . 2 4 8 , 4 45 Gain , . . . 60 Gain . . . .

VI I 583 5 5 16.2 60 0

35 26 .8 9 8 . 0 50 26.9 9 9 . 6 65 7 7 . 0 100.0

VII I1 596 3 20 26 .5 9 8 . 1 2 6 . 6 9 8 . 4 30

VIII I 600 5 10 21 . 0 , , . 9 IX 11 597 6 5 21.2 7 8 . 3

15 24 .7 91 . 0 X 611 6 49 26.3 97.7

69 26 .7 98.8 86 2 7 . 2 100.0

146 2 7 . 1 100.0 XI I 594 6 5 2 5 . 7 95 .0

15 26 .3 97 .2 45 2 6 . 7 99.0 60 26 .7 9 9 . 0

XI1 I1 601 1 5 2 5 . 6 94 .9 15 26 .5 98.0 45 26 .5 9 8 . 0 60 2 6 . 6 9 8 . 4

XI11 754 3 5 26.7 9 9 . 0 15 2 6 . 8 99.3 45 26 .9 99 .8 60 26 .9 99.8

XIV 752 4 5 26 .8 99 .6 15 26.8 99.6 45 26 .9 9 9 . 8 60 26.9 99.8

X V 749 3 5 24.9 92 .3 15 26 .7 9 9 . 1 45 26 .8 99 .6 60 26 .8 99.6

150 2 6 . 8 9 9 . 6 XVI 751 3 5 26.0 96.2

15 2 6 . 6 9 8 . 8 45 26 .7 99:1 60 26 .7 9 9 . 1

XVII 900 4 5 26.5 98 .3 15 26 .9 99 .8 45 2 7 . 0 100.0 60 2 7 . 0 100.0

XVII I 900 4 5 2 6 . 6 98 .8 15 26 .8 99.7 45 26 .9 9 9 . 8 60 26 .9 99 .8

15 13 .4 , 4 9 . 5

15 26 .4 9 7 . 8

r -

A strong odor of hydrocyanic acid was noticed throughout run IT. After five minutes or so, a deposit of carbon began t o form in the boat, due t o the decomposition of carbon monoxide by finely divided cobalt, as in the runs a t 350 ' C.

Throughout run V a strong odor of H C N was noticed. This is t rue of all t he reductions of cobalt oxide with carbon monoxide in the neighborhood of 4 j o " C.

These two runs are typical of a large number of similar ones. Our observations seem t o show tha t the decomposition of CO by cobalt takes place only through a temperature interval in t h e neighborhood of from 300-450' C.

The check between composition of the oxide used for

AiVD EAITGILVEERIiITG C H E J I I S T R Y 5'01. 6, NO. 2

these CO reduction experiments, as determined b y analysis and as determined by t h e reduction experiments themselves, is entirely satisfactory. (See following article, p. 115.)

C O X C L U S I O N S

I. The reduction of Co303 t o metallic cobalt by carbon monoxide gas takes place very rapidly a t all temperatures above 600' C.

Between 3 j o " C. and 4 j o o C.: carbon monoxide a t first reduces c0304 t o cobalt, b u t after a t ime t h e finely divided cobalt decomposes the CO gas, de- positing carbon.

At temperatures between j o o o C. and i jo" C., over 90 per cent of t he reduction of Co304 t o Co takes place in a few minutes, bu t a further reduction to com- pletion takes place very slowly. IV. Between i j o o C. and 900" C. , t he amount of

reduction of Cos04 t o Co, which takes place during the first few minutes increases very rapidly, a n d a t t h e higher temperatures it is complete.

Where producer gas is available i t should offer a cheap and efficient means of producing large quantities of pure metallic cobalt from t h e oxide.

For the production of cobalt from Co304 by CO, the charge must be completely cooled in a n atmosphere of CO.

11.

111.

V.

VI.

IV-REDUCTIOK O F COBALT O X I D E WITH A L U M I X U M

The heat of formation of a molecular weight i n kilograms of aluminum oxide (A1203) is 392,600 kilogram-calories, and is greater t h a n t h a t of any other metallic oxide. The molecular heat of formation of ferric oxide (FePOs) is correspondingly 19 j ,600 kilogram-calories. It is therefore obvious t h a t if finely divided aluminum be intimately mixed with ferric oxide (Fe203) , t he latter, possibly in t h e form of rolling mill scale, t h a t t he reaction Fe203 + z A l = X1203 + 2Fe will t ake place, provided t h e tempera- tu re be raised at some point in the mixture sufficient t o s ta r t t he reaction. This principle has been used by the Goldschmidt Thermit Co. t o produce molten iron for welding purposes.

It is obvious t h a t for every 160 kilograms of ferric oxide a n d j4 kilograms of metallic aluminum t h a t are mixed together and fired in this way, there are developed 392,600 - 19j,600 = 197,000 kilogram- calories of heat. This is sufficient t o raise the entire mass t o a white hea t , so t h a t t he molten iron readily settles t o the bottom from where i t may be tapped.

I n a similar manner: metallic cobalt may be prepared by reduction of cobalt oxide with aluminum according t o t he reaction:

3 c o S o 4 + 8Al = 4&03 + 9Co

The molecular heat of formation of Cos04 is 193,400 c'alories.

I t is therefore obvious t h a t for every 723 kilograms Co304 and 216 kilograms of aluminum t h a t are mixed together and fired, there are developed 4 X 392,600 - 3 X 193,400 = 990,200 kilogram- calories of heat. We would, therefore, expect a reaction

1 Tables Annuelks Inlcrnationales des Conslanles, 1, 428 (1910).

Feb., 1914 T H E J 0 L- R S A L 0 F I S D r.5 2' RI .4 L -1 S D E S G I S E E RI AJ- G C H E MI S T R Y 11;

quite as vigorous, if not more vigorous, t h a n the corresponding one with ferric oxide.

Experiments n-ere t.ried, October, 191 2 , using a s tandard Goldschmidt Thermit conical welding furnace. In to this was charged j-IO lbs. of finely divided Cos04 with the theoretical amount of aluminum, according t o the equation 3C0301 + 8A1 = 4.k1203 + 9Co. The reaction was s tar ted b y lighting a fuse of finely divided aluminum and potassium chlorate, rolled in a piece of tissue paper. T h e furnace fired with extreme violence, in every case becoming a n intense white heat . The vigor of the reaction was so great t h a t the lining of the furnace, although the best alundum-magnesite-cement mixture, n-ould s tand u p for only t v o or three charges.

T H E METAL-The metal produced in this manner n-as readily tapped from the bot tom of the furnace into iron or sand moulds. I t frequently contained less t h a n 0.1 per cent of aluminum. and, of course, was carbon-free.

The various metals, chromium, molybdenum. etc. , made by the Goldschmidt Co. b y this method. as they h a r e come t o us, run about 0 . j per cent in aluminum and are carbon-free.

c 0 s c L U S I O s s This aluminum reduction method can obriously

be used with considerable satisfaction where ab- soltitely carbon-free metal is required, and where a somewhat increased cost is not prohibitive. XIore- over, i t affords a method of preparing cobalt-aluminum alloys a t once by adding a n excess of metallic aluminum.

The price of crude aluminum, such as might be used for this purpose, is in the neighborhood of I ; cts. per lb. One pound of a luminum \%-ill reduce and melt . in this way a little over two pounds of metallic cobalt. Therefore, there is a charge of 1 7 cts. in the form of I 11). of metallic aluminum, for the power for reducing a n d melting two pounds of metallic cobalt. There might, of course, be some return for the fused aluminum oside which resulted from the process, b u t even allowing liberally for this, the costs are high as .compared with the carbon and carbon monoxide methods of reduction described elsewhere in this paper.

I t is obvious t h a t the heating costs must be high b y t h e aluminum method, for heat is being supplied a t a temperature greater t h a n 2100' C., t h a t is, a t a temperature far in excess of what is required for t h e reduction of the oxide and the melting of the metal, and with consequent a t tendant increased losses, due t o conduction and radiation.

ELECTROCHEMICAL AND >fETALLURGICAL RESEARCH L A B O R A T O R I E S

S C H O O L OB M I N l i Y G , QUEEN'S UNIX7ERSITY

KISGSTON, ONTARIO

-

OXIDES OF COBALT' By HERBERT T. KALMUS

T h e following oxides of cobalt have been described in various places throughout t h e l i terature: Co20 , COO, COSO,, C O 6 0 7 , CO405, CO304, c O , o ~ o , c0203,

' Published by permission of the Director of Mines, Ottawa, Canada. See footnote to previous article, page 107.

C O ~ ~ O , ~ , C o 3 0 5 , Coo2, and considerable disagreement is t o be found among the statements concerning them.

The existence of many of these compounds is doubt- ful, and there are but three of them which particularly concern the commercial manufacturer of cobalt oxide: Co304, CoeO; and COO. These concern us in the production of metallic cobalt. TTe shall. therefore, describe these three oxides as m-e have observed them in the course of the experiments reported in our previous article, p. 1 0 7 of this issue of THIS J O U R S A L .

C 0 B .4 L T 0 - C 0 B A L TI C 0 X I D E , c 030 .I

The ordinary black commercial cobalt oxidc which has been prepared from the hydrate , b y calcining in the neighborhood of ; joo C.? is a mixture of Co304 and Co607 , but largely the former.

There is a n abundance of proof throughout the previous paper t h a t this black oxide is lai-gely C o 3 0 4 , of which the following may be particularly noted:

( a ) The purified cobalt oxide used for hydrogen reduction experiments. making allowance for the impurities according t o the analyses, w a s computed t o contain 7 2 . 9 per cent cobi l t . The hydrogen reduction experiments, using this same oside. showed, n-herever the reduction xyas complete, a loss of oxygen amounting t o 2 7 . 0 per cent. A s was shown on p . I I I , this checks m-ith the 7 2 . 9 per cent of cobalt, with allowance made for t h e slight impurities. Hencc, this black oxide must be largely Co301> as may bc seen from thc following theoretical percentages:

Per ccrit cobalt Co?Oa.. . . . , . . , , . 7 1 1 CoaOl. . . , . . . . . . , , . 7 3 . 4 C o , 0 7 . , , , . . , . , . , , , , . . i5.9 coo . . . . . . . . . . . . , , , . , , 7 8 . 8

( b ) The purified cobalt oxide used for thc carbon monoxide experiments, making allowance for the impurities according t o the analysis, was computet1 t o contain 72.9 per cent cobalt. The C O reduction experiments, using this same oxide, whererc'r reduction was complete, showed a loss of oxygen amounting t o 2 7 . 0 per cent. As was shown on pp. I I O and 111,

this checks with 7 2 . 9 per cent cobalt, with allow-:ince made for t h e slight impurities. Hence, this black oxide must be largely Co304 according t o the table under ( a ) .

-1s a further proof t h a t the black oside calcined a t a good red heat is C0~0.3, the following experiment was tr ied: A pure black hydrate of cobalt was made from

electrolytic cobalt by the potassium-cobalti-nitrite method. This was calcined t o constant veight a t IO j O C., yielding a chocolate-brown powder, which was uniform under t h e microscope. Several samples of this brown powder were calcined t o constant \\-eight a t 640' C., and in each instance showed a loss of '

water between I I . 5 per cent a n d I I .8 per cent. There- fore, the brown powder corresponds very closely t o

The material resulting from these calcinations n-as a black powder identical in appearance under t h e microscope with the black cobalt oxide of commerce.

(c )

C0203.HaO.

the preparation of metallic cobalt by reduction of the oxide

Documents