PREPARATION OF PURE IRON AND IRON-CARBON ALLOYS

Ind. Eng. Chem. , 1916, 8 (3), pp 217–224. DOI: 10.1021/i500003a003. Publication Date: March 1916. Cite this:Ind. ... Published online 1 May 2002. P...
0 downloads 0 Views 3MB Size
Download PDF
hlar., 1916

T H E JOCRiliAL OF I N D C S T R I A L A N D ELVGINEERING CHEMISTRY

PREPARATION OF PURE IRON AND IRON-CARBON ALLOYS‘ By J , R . CAIN, E.

SCHRAMM AND

H.

E. CLEAVES

Received January 24, 1916 I K T R 0 D I2 CT I 0 N

The fundamental importance of the iron-carbon thermal equilibrium diagram in t h e scientific metallurgy of iron and steel and its utility t o practical workers have long been realized, a n d accordingly this subject has received attention from many points of view and from many investigators during t h e past two decades. I n view of this fact, it niight seem superfluous t o a d d t o the existing literature except for t h e following considerations. Earlier workers have for t h e most part confined their attention t o special portions of t h e diagram or t o disputed questions of theory. Their thermal studies have not been carried out with the degree of accuracy now attainable. They have practically, without exception, employed commercial materials of varying degrees of purity. I n two papers2 published in 1 9 1 3 Prof. H. 11.Horn-e, as a result of a thorough examination of the literature, has fixed t h e most probable position of the equilibrium lines. His net conclusions are t h a t “this calculated line is not entitled t o great; weight because of the weakness of t h e TABLEI-PERCENTAGE COMPOSITION OF IRON-CARBON ALLOYSUSED BY VARIOUSINVESTIGATORS IN DETERMINATIOXS on CRITICALPOIXTS AUTIIORITY Date C Si Mn P S CarpenterandKeeling(n) 1904 0 . 3 8 0 . 0 6 Trace 0 . 0 3 0.01 Carpenter a n d Keeling. . 1904 1 . 8 5 0 , 0 9 ,, ... , . . Carpenter a n d Keeling. . 1904 3 . 9 8 . . ... Carpenter and Keeling.. 1904 4 . 5 0 0 : 12 .., ... . . . . . Carpenter and Keeling.. 1904 2 , 6 3 ... ... Carpenter and Keeling. , 1904 2 . 8 5 ,,, , , ... ... Carpenter and Keeling. , 1904 2 . 8 5 , . , Trace? ,., .,. 7 Heyn(b). . . . . . . . . . . . . . . 1904 0 . 3 9 0 . 0 4 0 . 0 3 ... ... 8 Heyn . . . . . . . . . . . . . . . . . . 1904 0 . 9 5 0 . 0 4 0.06 ... ... 9 Rosenhain(c) 11910 1 . 1 4 0 . 0 9 0 . 4 0 0.014 0.018 10 on Brayshal 1908 1 . O O ,, , 0.25 ,,, ... . . . . . 12 Charpy and Grenet(e . . . 1903 0 . 6 4 ... ... 1903 0 . 6 4 ,.. , , , . . ... 1903 0 . 9 3 ,.. , , , . . ... 15 Charpy and G r e n e t , . , . . 1903 0 . 9 3 .., ., ... ... 16 Charpy and G r e n e t . . . . . 1903 1 . 5 0 17 Brayshaw(f)No.\V4 . . . . 1910 1 . 1 5 o . i i o . o i i + o.oii+ 18 Brayshaw No. W z . . . . . . 1910 1 . 1 6 0 . 1 0 0 . 3 7 0.014 0.023 19 Brayshaw No. A * . . . . . . . 1910 1 . 1 4 0 . 0 9 0 . 4 0 0.014 0.018 . . . . . . . . . . . . . . 1913 0 . 2 3 0 , 0 3 9 0 . 0 5 0 . 0 1 3 0.010 . . . . . . . . . . . . . 1912 0 . 9 2 0 . 1 4 0 . 1 2 3 0 . 0 0 9 0.011 0.005 0.024 Levy(h) . . . . . . 1913 0.027 . . . 0 . 2 6 0.015 0.028 L e v y . , . . . . . . 1913 0.105 0 , 0 1 3 0 . 2 4 0.013 0,010 24 Howe a n d Levy 1913 0 . 2 1 4 0 . 0 3 9 0 . 0 5 25 Howe a n d Levy 1913 0 . 2 2 7 0.039 0 . 0 5 0.013 0,010 1913 0.235 0 , 0 3 9 0 . 0 5 0.013 0,010 27 Howe a n d L e v y . . . . . . . . 1913 0 . 2 4 4 0 , 0 5 0 Nil Nil 0.007 2 8 Howe and L e v y . . . . . . . . 1913 0 , 3 8 2 0.027 0 . 2 2 0.004 . . . . 1913 0 . 4 0 0.103 0 . 1 6 o : o i 4 0.012 . . . . 1913 0 . 5 6 3 0 . 1 8 0 , l j 0 , 0 1 3 0 . 0 1 3 0.018 0,013 31 IIowe and Levy , , . , , , . 1913 0 . 5 9 0.144 0 . 1 7 0,012 0.019 32 Howe and L e v y . , , , , . , . 1913 0 . 7 3 0.141 0 . 0 7 33 Howe and L e v y . . . . . . . . 1913 0 . 9 2 0 . 1 4 0,123 0.009 0.011 ( a ) Carpenter and Keeling, Jour. Ivon and Steel I n s t , 66 (1904), 244; Collected Resear;ches of t h e Nat’l Phys. Lab., 1, p. 227. ( b ) Heyn. Verh. des Vereins zur Befarderung des Gewerbfleisses,” KO.

1 2 3 4 5 5A 6

:

o:ii

1904. D 371. ~~~-

(cy Rosenhain, Proc. I n s f . Mech. Eng., 1910, p. 688. ( d ) Benedicks, Jour. Iron and Steel Inst., 77 (1908), 218. ( e ) Charpy a n d Grenet, Bull. Sac. d’Encouragement pouv l’lndustvie NafionalP, 1903, p. 480. (f) Brayshaw, Pvoc. I n s t . M e r h . Eng., 1910, p p , 525, 537, 656, 670. ( 9 ) Howe, Bull. A m . Inst. Mining Eng., 1913,p . 1068. ( h ) Howe and Levy, Ibid., 1913, p. 1076.

evidence,” a n d t h a t “much better d a t a are needed, reached with pure materials and with t h e many causes of error reduced t o a minimum.” I n Table I , compiled from Prof. Home’s papers, are given analyses 1

Published with the permission of the Director of t h e Bureau of Stand-

ards.

’

“Ael, t h e Equilibrium Temperature for AI in Carbon Steel,” Bull. A m . Inst. Mining Eng., 1913, p , 1066; “A Discussion of the Existing D a t a 0 s t o t h e Position of Aea,” Ibid.. 1913, p. 1099.

of the materials used in some of the more important recent investigations. A glance a t the table will show the justice of the conclusions quoted above. The analyses are incomplete even for t h e impurities ordinarily. determined, and entirely ignore the possible presence of other impurities, such as Cu and S i ; b u t so far as they go, they indicate a very appreciable degree of contamination b y S, Si, Mn, elements which we know exert a marked effect on the critical ranges. I t may be said t h a t t h e iron-carbon diagram has never been worked out with pure iron-carbon alloys. T h e present paper describes the preparation of a series of high degree of purity t o be used as t h e basis of a more accurate study of t h e equilibrium diagram t h a n has heretofore been attempted. For the production of pure iron on a fairly large scale, t h e electrolytic refining method was obviously most suitable and was therefore adopted in this work. Pure carbon was made b y calcining in a Dixon graphite crucible t h e pure sugar used as stock for Bureau of Standards analyzed sample Yo. 17. The latter contains, as the only impurity of importance for present purposes, 0.003 per cent ash, and the carbon obtained from it has a n ash content of 0.17 per cent. 1IAKIXG THE ELECTROLYTIC IROh-

The electrolytic method, using soluble anodes, has been frequently employed in similar investigations1 and its essentials are well known, so t h a t we give details only for the sake of completeness and because, in one respect, our method deviates from t h a t usually followed, namely, in the use of porous anode compy-tments. The first iron was made on a small scale. The essential details of the b a t h are as follows: Two cylindrical anodes of ingot iron2 about z in. in diameter by j in. long, contained in porous clay cups; three cathodes of sheet iron, each 4 in. square; electrolyte, 2 s t o 3 0 per cent FeC12 solution (made b y dissolving the ingot iron in chemically pure hydrochloric acid), prepared as nearly neutral as possible; current density about 0.5 t o 0 . 7 ampere per square decimeter; temperature during electrolysis approximately t h a t of t h e room. No a t t e m p t was made t o determine the yield or t o secure high current efficiency. Good adherent deposits were obtained, t h e greatest thickness being about 0 . 5 cm. Owing t o t h e unfavorable current distribution when working with anodes and cathodes of such unequal sizes, the thickness of deposit was not uniform all over t h e plates. Qualitative tests of t h e sludge f r o m t h e anode cells showed t h a t there

’

Lenz, J . pvakt. Chem., 108 (18691, 438; Cailletet, Campi. rend., 80 (1875), 319; Roberts-Austen. J o u r . Ivon and Steel Inst., 1 (1887), 71 ; Arnold and Hadfield, Slahl u . Eisen, 1894, p. 526; Hicks and O’Shea, Eleclvician, 1896, p. 843; Winteler, Z . Elehlrochem., 4 (1898), 338; Haber, Ibid., 4 (1898), 410; Abegg, Slahl u . Eisen, 1901, 11, p. 736; Skrabal, 2 . Elektrochem., 10 (19041, 749; C. F. Burgess and Hambuechen, Traizs. A m . Electvochem. Sac., 5 (19041, 201; Maximowitsch, Z.Elektrochem., 11 (1905), 52; Ryss and Bogomolny, I b i d . , 12 (1906), 697; Coxper-Cowles, C. A , , 3, 1371; 5, 3013; 8 , 3648; C. F. Burgess and Watts, Trans. A m . Electrochem. Soc., 9 (19061, 229; a m b e r g , Z . Eleklrochem., 14 (1908), 326; Kern, Trans. A m . Electrochem. Soc., 3 (1908), 103; Muller, Melaliurgie, 6 (1909), 145 (contains bibliography); Pfnff, Z. Eleklrochem., 16 (19101, 217; Tucker and Schramm, THISJOURNAL,2 (1910), 237; Watts and Li, Trans. A m . Electrochem. Soc., 26 (1914), 529; Yensen, University of Illinois, Bull. 72, 1914. ?Analysis a s follows: C, 0.016; S, 0.022; Mn, 0.029: P , 0.001: Si, 0.002; and Cu, 0.15.

218

T H E J O U R N A L OF I N D U S T R I A L A N D ENGINEERING C I I E M I S l R Y

was an accumulation of manganese a n d copper derived from anodic impurities. T h e porous cups therefore seemed t o be of service in preventing anode impurities f r o m migrating t o t h e cathodes, a n d t h e y were accordingly used in one of t h e larger baths t o be described later. I n another similar t a n k t h e cups were omitted. Table I1 shows t h a t t h e cathode deposits from t h e b a t h without t h e cups were a little higher i n copper t h a n t h e others b u t were otherwise of similar quality. I t was found t h a t there was much oxidation of t h e surface layers of t h e electrolyte as t h e electrolysis went on, resulting in t h e production of basic salts of

Vol, X, No, 3

The large electrolytic bath is shown in Fig. I. The anode cups were molded from a mixture of equal p a r t s b y volume of alundum cement a n d of clean white Ottawa s a n d ; after careful drying in an oven t h e cups were burned at a temperature of 1000' t o IIOOO a n d were t h e n found t o have a satisfactory degree of porosity. The current density during electrolysis varied from 0.3 t o 0.4 ampere per sq. dm. The electrolyte contained 23.3 per cent FeCI2 (made from t h e ingot iron previously described) a n d 10.3 per cent NaCI, a n d was nearly neutral. Analyses of electrolyte from t h e anode a n d cathode compartments made after a week's run wcre as follows: P E P CEKT IRON

Aiiode c o m ~ r r t m e m

Original . . . . . . . . . . . . . . . .9 . 5 2 Alter one r e e k ' s rtin.. ... 8.82

Pm. I-Tmr

*OR P ~ S P A R I N EG ~rcrao~vrI i cs m

iron, which floated i n t h e bath a n d which migrated t o some extent t o t h e cathodes. With t h e intention of avoiding or minimizing this oxidation, t h e small b a t h was provided with a hydraulically sealed cover having windows for observation a n d conduits f o r t h e current leads. T h e air in t h e space over t h e electrolyte was displaced with purified carbon dioxide a n d t h e electrolysis conducted as before. This method led t o no marked improvement a n d i t was abandoned, especially as there seemed to be a tendency toward higher percentages of carbon in t h e cathodic deposits. Although t h e greater part of t h e sediment settled t o t h e bottom, t h e b a t h was never quite free from turbidity caused b y these basic salts, a n d t h e r e is no doubt t h a t TABLB

II-rmzNmCn

C O ~ P O S ~ O N OP ~ V A R I O U S SOURCES

Caulode ~ o m ~ a r t m ~ o t 7.8R 7.53

The character of t h e deposits is shown b y Figs. I 1 a n d 111. Deposits of j t o i mm. thickness were obtained; t h e characteristics of these were a b o u t t h e same as of those made in the smaller b a t h . They were very hard a n d brittle in consequence of contamination by hydrogen. On removal from t h e b a t h a n d after washing with distilled water they corroded rapidly in t h e air. No account was taken of this surface oxidation, inasmuch as t h e iron was t o be used f o r making iron-carbon alloys in such a way t h a t t h e oxidized compounds of iron would be reduced t o metal a n d t h e hydrogen expelled. About 1 2 t o 1 5 kg. of iron were made f o r use in preparing t h e alloys. I n Table I1 are given analyses of this iron made with a n d without t h e use of porous anode compartments, a n d for comparison analyses of electrolytic iron from other sourccs. MELTING T H E ELECTROLYTIC IRON

The next step for t h e further utilization of t h e electro-deposited iron is t o melt it into ingots. The material as taken from thc bath is brittle

~ . r ~ iaoN r ~ o ~ ~ r ~ ~

Total AnalN i a n d im- ysir cu co D U I i l Y 1~, 1" C S P M n Si I(o. b l . . . . . 0.004 0.002 'hcc n . m 0.005 0.011 0.026 B. S. Z l r , b ) . .... 0.on4 0.003 T~~~~~ r a n.oo6 ~ e T C n.oii ~ 0.024 ~ ~ B.S. 3 ( d . b ) . . . . . n.nm 0.006 T~~~~Trace 0,008 0.006 0.011 o . 0 ~B.S. 4(CI ........ 0 . no4 n, on4 T~~~~T~~~~n . o m 0.008 0.011 0.03s m.s. 51Jn... ..... 0.063 0 . 0 0 2 0.005 0.009 0.005 . . . . . . 0.084 Mllller 6(gI.. . . . . . 0.008 T~~~~n . m 0.009 0 . 0 1 4 0 . 0 3 3 Stead 7 ( h ) . . . . . . . o m 8 0.004 T~~~~T~~~~ 0.006 o i l 0.029m.s. 8(iI ........ 0.009 0.003 TI^^^ T~~~~0.006 0.006 ... 0.024 B. s. ~~

. ~ ~ ~ ~ ,

:::

FIG.

t h e deposits were contaminated b y t h e m i n slight degree. This was of n o consequence, inasmuch as these impurities were either volatilized or were reduced t o iron in subsequent melting operations. T h e analysis of t h e iron stripped from cathodes of t h e small electrolytic b a t h is given in Table 11. About 2 or 3 kg. were made i n this bath.

II-CAY,,,,,,,:

I~,ii.OS,l

a n d is contaminated n,ith occludcd chlorides from t h e electrolyte, with hydrogen, a n d with oxidized compounds of iron. Annealing at 600" t o gooo suffices t o remove most of t h e hydrogen a n d some of t h e chlorides; this renders t h e iron more malleable a n d resistant to corrosion, b u t t o a t t a i n a still higher degree of purity a n d t o simplify subsequent operations t h e iron must be fused, preferably in a reducing atmosphere, a n d held in a s t a t e of fusion for some

Mar., 1916

T H E 1 0 L : R N A L OF I N D L ' S T R I A L I S D E . V C I S E E R I N G CIII,I11 I' CU nrisill;li irml not n l e i t r d . . . n.on4 n.nn1 0.004 'hce. . 111 krgpfoi fllrllace filled ~ i t i ACIXSOU l L . W ~ n . o u n.ni n . n m :rmcc 'rrace . . I,, ~ i ~iurnrcr ~ .... i . ,. ~. . . ~ 0.014 ~ 0~. 0 2 4~ l'r:lce . . . in . A iurLmce.. ~ ~ . . . ..~. . . ~ ~ m ~o o 7 n.nw 'rrnce l'rrre 11.012 I n l l c l i , e r ! w iurnacc am1 reII,Ciced i,, . A iarnnrr. ~ ~ n.nn? ~ n.nn4 ~ ~ n.nns ~ nnce l'rnre I,, iurnncc.. . . . . . . . . . 0.012 n.nm n . w

'rmce

... .

n.nm

rlaCe

.

.

sulfur in t h e amorphous carbon resistors is due to t h e use of pctroleum coke in their manufacture. Annlyses of t h e best grades of pctrolcum coke obtnincd hy t h c Geological Survey showed pcrcentages of sulfur ranging from 0.63 t o 1 . 3 7 per ccnt. Aftcr having discovcrcd these defects i n amorphous carbon w e discontinued its use, employing granular Achcson graphite of t h c best gradc for t h e kryptol furnace, a n d a graphite

T H E J O U R N A L OF I N D U S T R I A L A N D ENGINEERING CHEMISTRY

220

spiral for t h e Arsem furnace. A resistor from t h e Arsem furnace contained 0.02 per cent sulfur a n d 0.03 per cent silica. T h e results were satisfactory. It may be concluded t h a t if carbon resistance furnaces are intended for making melts with minimum contamination b y volatile substances from t h e heating element, t h e best material available a t present is first-quality graphite; a n d t h a t the resistors should be carefully analyzed t o insure against impurity before

Vol. 8, KO.3

these for melting pure iron-carbon alloys. T h e clays used for making t h e crucibles which we tried were not sufficiently refractory a n d were badly corroded b y t h e iron oxide which coats t h e surface of all melts made in gas-fired furnaces. This did not occur i n t h e electric furnaces, b u t here t h e reducing atmosphere, which prevented oxidation, also caused t h e introduction of relatively large amounts of silicon from t h e clay. Alundum crucibles were tried, b u t gave t h e same trouble as those of clay. Crucibles made of electrically fused or sintered magnesia from two different sources were given a trial i n t h e various types of furnaces, b u t with these also there was more contamination of t h e melts b y silicon t h a n seemed desirable (Table IV). TABLEIV-PERCENTAGE COMPOSITION OF ALLOYSMADE I N CRUCIBLES COMMERCIALLY PURE MAGNESIA C Gas and vacuum furnaces., , . . , , , . . . . , 0.584 Gas and vacuum furnaces. , . , , . . . . . . . . 0 , 0 2 2 Gas and vacuum furnaces.. , . . . . . . . . , 0.367 Single melt in Helberger furnace. . . . . . 0.886 Melted twice in Helberger furnace.. , , 0.688 Helberger and vacuumfurnaces.. , .. . . , 0.210 Helhereer and vacuum furnaces.. , . . . . . 0.252 Helbcrger and vacuum furnaces. ........ , 0.094 Hclberger and vacuum furnaces. ........ , 0.146 Helberger and vacuum furnaces. ....... , , 0.088 Helberger and vacuum iurnaces , ,,.... , , 0.765 Helherger and vacuum furnaces. ,...... , . 0.058 0.927 Melted twice in kryptol furnace. .... . OF

NO.

F I G IV-KRYPTOL FURNACE WITH CRUCIBLE I N POSITION

installing t h e m in t h e furnace. It is further evident t h a t if t h e refractory walls of t h e furnace are i n immediate contact with t h e heater, t h e former should be made of material not likely t o react with t h e heated carbon. Such reaction would cause not only possible contamination of t h e melt, b u t also irregularities i n t h e working of t h e furnace. For this reason ordinary fire clay or silica bricks, or a n y others containing silica in considerable amount, are excluded. I n our own kryptol furnace commercial magnesite bricks were used, a n d while t h e y were not all t h a t could be desired they proved serviceable, provided t h e resistor was renewed occasionally. G A S FCRXACES-TWO types of gas furnace, shown in Figs. V I a n d V I I , were used a n d no difficulty was experienced in maintaining t h e necessary temperatures. The furnace shown in Fig. V I was a standard t y p e except i n respect to t h e blowpipe, which is similar t o t h a t used i n tool forges, or for brazing purposes. A preheater raised t h e temperature of t h e necessary I n some experiments volume of air t o about 350'. made with this furnace pure platinum wire was melted, indicating t h a t a temperature in excess of 1750' h a d been attained. The refractory lining supplied b y t h e manufacturers was found t o have a very short life under these conditions, so t h a t we later covered it with alundum cement, or with a mixture of alundum cement a n d clay, which prolonged t h e life. The surface combustion furnace (Fig. V I I ) proved t o be very suitable. This furnace has already been illust r a t e d a n d briefly described.' T h e same trouble was experienced with t h e original refractory lining, a n d repairs were again made with alundum a n d clay. T h e highest temperature reached in this furnace, 1670°, was obtained when burning about 180 cubic feet of city gas per hour. I n using this t y p e of furnace for pure melts, t h e crucible must be protected from t h e large amount of very fine dust blown out of t h e contact material during operation. CRUCIBLES

A few preliminary experiments with clay crucibles showed t h a t i t would be out of t h e question t o use 1 Lucke,

THISJOURNAL, 5 (1913).

801.

P 27 P 28 P 29 P 41 P 39 P 31 P 32 P 35 P 35 P 37 P 40

P 42

P 36

M E T H O D OF -hfELTING

.. . . ~.. .. . .. .. . .

...

Si

S

0.056 0.029 0.015 0.024 0.054 0.032 0.022 0.041

0.004 0.030 0 029 0.013 0.026 0.020 0.010 0.024 0.050 0.029 0 . 0 5 0 0.039 0.033 0.015 0.070 0.016 0'.045 0.019

As i t h a d become evident t h a t t h e desired results could not be secured with a n y kind of crucible on t h e market, we began t h e experiment of making our crucibles of various grades of chemically pure magnesia

FIG.V-ARSEM FUR~XACE WITH CRCCIBLE AND PROTECTING TUBEI N PLACE

calcined in t h e electric furnace a t 1600' t o 1800'. Although our product contained usually not over 0.05 t o 0.10per cent silica t h e alloys melted in crucibles made from such magnesia still carried too much silicon (see Table IV). Because of t h e difficulty of securing from chemical dealers magnesia suffi-

M a r . , 1916

T H E S O U R N A L OF I N D U S T R I A L AND' ENGINEERING C H E M I S T R Y

ciently low i n silica, t h e high cost of a good grade of this material, a n d t h e need of large quantities for several contemplated investigations, we decided t o prepare our own material. An endeavor was made t o develop a method free from t o o m a n y complicated manipulations. As raw material we used t w o or three grades of pharmaceutical magnesium carbonate

FIG.VI-SMALL GASFURXACE WITH PREHEATER

221

lasted about two hours. As a result of this treatment t h e magnesia h a d shrunk t o about one-third i t s original volume, all t h e carbon had burned out, a n d t h e silica content was very slightly increased. We found i t important t o blow air through t h e furnace for several minutes after shutting off t h e gas, in order t o remove t h e last traces of products of combustion carrying sulfur. If these are allowed t o remain in t h e furnace during t h e cooling period, t h e magnesia takes u p some sulfur; our best material contained less than. 0.01 per cent of this element. Calcining in t h e gas furnace at t h e temperature named gives a product which still shrinks a little when used in crucibles heated t o higher temperatures, b u t we h a d no serious trouble with crucible failures on this account. Two sizes of crucibles were used which, together with t h e molds employed for making t h e m , are shown in Fig. VIII. T h e calcined material was mixed with about IO per cent b y weight of uncalcined magnesia a n d t h e whole wet with water until i t formed a pasty '

TABLE V-PERCENTAGE OF SILICA IN MAGNESIA FROM DIFFERENT SOURCES

carrying 0.1or 0.2 per cent silica, a n d later a calcined Grecian magnesite with about 3.5 per cent silica. Attempts t o prepare silica-free magnesia from these sources b y dissolving t h e m in hydrochloric acid, evaporating t h e solutions t o dryness a n d baking, followed b y solution of t h e MgC12, filtration a n d precipitation of magnesium carbonate b y ammonium carbonate were not very successful on t h e scale tried, for t h e technique was difficult a n d t h e product unsatisfactory, as well as expensive. After trials of several other methods we developed t h e following procedure, which has produced a magnesium oxide carrying usually not over 0.01 per cent silica, a n d a t low cost: T h e Grecian magnesite was dissolved in commercial acetic acid ( 7 0 per cent), using a slight excess of t h e l a t t e r ; more of t h e magnesite was t h e n added until t h e solution was alkaline t o litmus paper, after which t h e solution was diluted with about twice its volume of water, t h e whole thoroughly stirred a n d allowed t o s t a n d in barrels for a d a y or t w o ; at t h e end of this time t h e clear solution was siphoned off into a large wroughtiron basin a n d rapidly evaporated over a large Fletcher burner, adding fresh liquid a t intervals until a suffi-t cient a m o u n t of t h e magnesium acetate h a d separated. T h e evaporation was t h e n carried t o t h e point where t h e solution solidified on cooling, after which t h e decomposition of t h e acetate into oxide was effected b y directing t h e flame from a large Teclu burner over t h e surface of t h e separated salts. T h e magnesium oxide so obtained is contaminated b y carbon, undecomposed acetate, a n d a l i t t l e iron, b u t after calcining is quite suitable for use in making crucibles. The calcining was done i n large gas furnaces which gave a temperat u r e of f r o m ~ j j o ' t o 16jo'. T h e magnesia, as taken from t h e iron basin, was moistened slightly with water a n d made u p into large balls, which were placed inside a n ordinary No. 2 0 plumbago crucible coated on t h e inside with alundum cement, or lined with a n Acheson graphite crucible. T h e crucible was carefully covered during t h e calcining operation, which

SOURCE

Calcined Grecian magnesite.. ...................... University of Illinois electrically calcined MgO. . . . . . Magnesium aluminate t u b e . . ..................... MgO reagent (uncalcined) A , . .................... MgO reagent (uncalcined) B . . Bureau of Standards MgO pre Bureau of Standards MgO calc' Crmible made from Bureau MgO . . . . . . . . . . . . . . . . . . . .

...

Si03 3 . 1 5 and 4.61 1 .99 5.39 0.03 0.013 and 0.017

mass. T h e t h i n brass cups used for lining t h e molds were p u t i n place a n d there was introduced into t h e mold a sufficient amount of t h e material (ascertained b y preliminary trials); a pressure of 5,000 t o 10,000 pounds per square inch was applied t o the head of t h e plunger a n d kept there a minute or two after t h e

FZ.VII-SURFACE

COMBUSTION

CRUCIBLE

FURNACE

brass cup carrying t h e crucible removed, a n d t h e plunger taken out. The crucibles were dried a t 100' t o 1 2 0 ' for a d a y a n d t h e brass cups stripped off b y melting tke solder from t h e joints. They were t h e n either used directly for melts, or after heating t o 1200' t o 1600' in a gas or electric furnace. The

222

T H E JOCRN.4L OF I N D U S T R I A L A N D EJVGINEERINC C H E M I S T R Y

latter procedure is necessary for all crucibles t h a t are t o be used in vacuum furnaces. PROCEDURE I S B I A K I S G ALLOYS

We first attempted t o make ingots i n t h e following manner: T h c iron as stripped from t h e cathodes was dried, brokcn into small pieces, a n d placed in one of t h e larger magnesia crucil,lcs (Fig. T'III), together with a suitable quantity of carbon. These were brought gradually t o tcmpcraturcs above t h e melting point of iron a n d left in t h e furnace for I O or 1 5 minutes. After coolinE,

PIG.

vfII-cR"C,eLn

MOLOS

t h e crucibles were removed a n d broken away f r o m t h c ingots. This procedure, while apparently wasteful, was necessary for sereral reasons: ( I ) Magnesia crucibles are extremely fragile at high temperatures a n d a n y a t t e m p t t o handle them results in breakage with resultant loss of melts; ( 2 ) with t h e small mass of metal which we used (100 g.) i t would be impossible t o pour successfully; a n d ( 3 ) even if i t could be accomplished, pouring would be objectionable because of t h c added danger of contamination. hloreover, t h e loss of t h e crucible is not serious, since t h e material can be recovered a n d worked u p i n t o new crucibles repeatedly. The ingots obtained in t h e manner above described were found t o be very unsound a n d full of blowholes; in this respect there was little difference hetween those made in t h e various furnaces. This seemed surprising in view of t h e diRercncc in atmosphere over t h e melt in a gas furuacc, where t h e conditions are oxidizing t o iron, a n d in an electric furnace, where gases carbon dioxide, water vapor, or oxygen-the which would oxidize iron at high temperatures-arc present only in very small amounts. I t would t h u s appear t h a t the blowholes i n iron may be caused by carbon monoxide as well as h y a n y or all of t h e other gases named, and t h n t lhc maintenance of a reducing atmosphere is no guarnntec of freedom from b l o u holes. T h a t the melts made in t h e kryptol furnace were made under reducing conditions is evident from Fig. I V , which shows t h a t l h c crucibles are completely covered with carbon a t all times; this is furthcr evidcnt f r o m an oxygen determination made on drillings from a n ingot melted i n l h e kryptol furnace. T h e pcrccntiige of osygcn in this ingot, notwithstan(ling t h e fact t h a t n o deoxidizer had hccn used, was 0.03.' T h e \wight of thcsc inxots was a b o u t IOO g. 8 \Ye arc indrl,fed to 1. A. .hwimle. m e t a l l u r ~ i r t of the American Rolling Mill Co.. for this andyrix.

Vol. 8, No. 3

Fig. IX(o) is a photograph of a split ingot, showing t h e unsound structure. T h e introduction of a regulated amount of carbon into t h e alloys gave a good deal of trouble. I n t h e gas furnace t h e amount of oxidizing gases was so great, relative t o t h e weight of carbon introduced, t h a t t h e lattcr was all burned o u t before t h e melting operation was completed, even when special precautions were taken t o protect t h e crucibles. These difficulties were finally overcome b y using t h e following procedure u-hich has been fairly satisfactory: The electrolytic iron was first melted down in t h e larger crucibles i n a gas or electric furnace. The ingots of pure iron so made mere sawed into longitudinal strips of convenient size for insertion into the smaller magnesia crucibles (Fig. V I I I ) a n d t h e neccssary amount of carbon was added. T h e crucible was placed in the vacuum furnace with t h e protecting chimney in place as shown in Fig. V. T h e furnace was cvacuated t o 0 . 2 mm., a n d t h e current through t h e heater was increased gradually until t h e iron had mclted and dissolved the carbon; this point was determined by observation through t h e window. As soon as this stage was reached a violent ebullition took place; sometimes t h e contents of the crucible were ejected. We attribute this principally t o t h e expulsion of gases from t h e cavities in t h e ingots. I n I O or 1 5 minutes t h e surface of t h e melt became quiescent a n d t h e operation was ended. After cooling, the ingots were removed by breaking t h e crucibles. Fig. I X ( h ) shows t h e sound structure of thesc ingots, which were usually entirely or nearly free from blowholes. After discarding t h e surface down t o clean metal, t h e ingots were turned down t o t h e size required for t h e thermal test specimens, Fig. I X ( c ) , re-

Pia. IX-SPLIT INGOTSA N D Trisr S P ~ ; C ~ M ~ N

taining t h e chips for analysis. Complete analyses of typical samples are given in Table 1'1. For comparison, Table 1'11 is given, showing t h e results of a t t c m p t s by t w o earlier workers t o make pure ironcarbon alloys. We are nom building a larger vacuum furnace f o r producing large ingots of pure iron a n d iron-carbon alloys, which will be examined either as made or after forging, rolling, a n d application of h e a t t r e a t m e n t

M a r . , 1916

T H E J O U R N A L OF I N D U S T R I A L A N D ENGINEERING C H E M I S T R Y

223

Acheson graphite as resistor material this difficulty disappeared, as is apparent from a comparison of Tables V a n d V I I . M A N G A N E S E A N D PHOSPHORUS-Both these elements have been reduced t o mere traces (less t h a n 0.001 per cent) in all our alloys. impurity, one of t h e most difficult TABLE VI-PERCENTAGE COMPOSITION OF TYPICAI, BUREAU OF STANDARDS CoPPER-This IRON-CARRON ALLOYS t o guard against, is present i n objectionable quantities C Si S P Mn Cu NiandCo Total in some of our melts a n d t o some extent in all. The Trace 0.020 0.01 1(a) 0.084 0,007 0.009 Trace 0.01 1( a ) Trace 0.005 0.376 0.013 0.009 Trace ingot iron used for anodes contains copper which is not 0.01 1(a) 0.012 Trace 0.395 0.008 0.013 Trace Trace 0 . 0 1 1(a) 0.004 0.597 0 , 0 1 0 0.008 Trace completely removed in t h e electrolytic refining process, 0.008 Trace 0.01 1(a) 0.624 0 , 0 0 4 0.010 Trace 0.008 0,692 0 , 0 0 6 0.011 Trace Trace 0.011(a) so t h a t our deposits contain about 0.005 per cent of 0,007 0.860 0 , 0 0 6 0.006 Trace Trace 0.01 1( a ) 0.013 0.01 1(a) Trace 1.087 0 , 0 0 6 0.006 Trace this element. This copper persists through t h e melt0.018 Trace 0.011 ( a ) 1.797 0 , 0 1 0 0.005 Trace 0.020 0 . 0 1 1( a ) Trace ing operation, a n d unless great care is taken, more is 2.240 n ,005 0.005 Trace 0.014 0.01 1(a) Trace 2.560 0 , 0 0 5 0.005 Trace introduced, owing t o t h e use of copper connections in Trace 0.016 0.011(a) 0.006 0.006 Trace 3.27 (a) T h e figures given for nickel and cobalt do not represent individual furnaces. I n particular, t h e copper blocks used i n t h e determinations for each specimen since t h e amount of sample was insufficient t o allow of these being carried out. Four representative analyses vacuum furnace must be smooth a n d make good on pure iron and iron-carbon alloys having given the result indicated, i t was assumed t h a t these elements were present in t h a t amount in all the contact with t h e graphite heater, since a n y arcing samples. causes t h e vaporization of considerable quantities of TAELEVII-DEGRSE OF CONTAXINATION (PERCENTAGES) OF IRON MELTS MADEBY OTHERS copper with resulting contamination of t h e melt. C Si P S Mn KICKEL A N D CoBALT-Our anode iron contains 0 . 0 2 Muller's electrolytic i r o n . . . . . . . . . 0.0630 0.0053 0.0045 0.0024 0.0090 Muller's electrolytic iron after reper cent (Ni Co); i n t h e electrolytic refining this is melting in vacuo in "pure MgO crucible"(a). . . . . . . . . . . . . . . . . .0.017 0.089 0.028 0.037 0.025 reduced t o 0.01 per cent which persists through t h e C. F. Burgess' electrolytic iron, , . . 0.009 0.006 0.001 0.010 Trace melting operations. C. F. Burgess' electrolytic iron melted with sugar carbon in UAGNESIUM-AS all our melting was done i n mag0.050 0.035 None magnesia crucible b y Howe(b) , , 2.954 0.040 (a) A. Muller Metallurgic 6 (1909) 159. nesia crucibles it was thought desirable t o look for (b) Home, B h . Am. Inst.'Mining kng., 1913, p . 1118. this element as a possible impurity. Several analyses D I S C U S S I O N O F T H E SOURCES O F C O N T A M I N A T I O N of ingots made i n t h e regular way showed t h a t Mg sILIcoN-In our earlier experiments, where we were was not present in a n y determinable quantity. A obliged t o make crucibles of magnesia higher i n silica t h a n rather interesting result, however, was obtained on t h a t produced b y t h e acetate method already described, analyzing some ingots which h a d been melted a t high we noted t h a t occasionally a n alloy of very low silicon temperatures (over 1700'). It was found t h a t these content would result from a melt made i n a crucible contained appreciable amounts of Mg (from 0.00; t o relatively high in silica. T h e use of our purified magnesia 0.01 per cent). Furthermore, t h e samples were so h a d eliminated all trouble from silicon contamination brittle t h a t t h e pieces broke while turning in t h e of melts, b u t we later decided t o make some experi- lathe. It appears at least possible t h a t there may be a ments t o determine t h e relationship between t h e intro- direct connection between t h e two circumstances, duction of silicon a n d temperature of melting. For though our present d a t a are not sufficient t o justify this purpose a series of runs was made in t h e vacuum a definite conclusion t o t h a t effect. furnace varying independently t h e temperature of oxYGEr-Unfortunately, t h e Bureau is not a t melting a n d t h e silica content of crucibles. One per present prepared t o make accurate oxygen determinacent of carbon was added t o all these melts since in t h e tions on this class of material, b u t it is hoped later t o presence of carbon there is additional likelihood of publish analyses of some of t h e alloys for this element, contamination of silica at high temperatures. The if i t is found t o be present. I n our method of prepresults i n Table VIII show t h a t if t h e temperature is aration, starting with a n ingot already low in oxygen not allowed t o rise much above 1600°, crucibles cona n d carburizing in a vacuum furnace where t h e carbon taining as much as 0.9 per cent silica m a y be used monoxide resuIting from interaction of oxides or oxygen safely. This is of importance when a great deal of with carbon would be removed as formed, t h e dework is being done, for i t enables one t o use repeatedly oxidation would tend toward completion, and acold crucible material until t h e silica becomes dangercordingly t h e alloys should contain b u t small residual ously high. amounts of oxides a n d oxygen. TABLE VIII-FACTORS GOVERKIXG CONTAXINATION OF MEI,TSBY SILICON a s t o physical properties, magnetic a n d electrical characteristics, resistance to corrosion, a n d related properties. Reports of progress i n this work will be issued as t h e y are justified. Later we shall deal with t h e effect of alloying elements other t h a n carbon on t h e properties of iron.

+

Si02 in Temperature crucible of melting Per (degrees) cent 1610 . . . . . . . . . . . . 0 . 5 7 1610 . . . . . . . . . . . . 0.67 1610 . . . . . . . . . . . . 0.75 1610 . . . . . . . . . . . . 0 . 9 1 1610 . . . . . . . . . . . . 1 . 1 4 1610 . . . . . . . . . . . . 1 . 6 6 1650 . . . . . . . . . . . . 1 35

SULFUR-AS

may

Si in melt Per cent 0,007

0.007 0,007 0.006 0,015 0.023 n 072 . - ~

Si02 Si in crucible melt Per Per cent cent 1720.. . . . . . . . . . 1.27 0.040 1780 . . . . . . . . . . . 1.20 0.042 1770.. . . . . . . . . . 0 . 1 9 0.007 1720 . . . . . . . . . . . 0.24 0.006 1760.. . . . . . . . . . 0 . 3 1 0.021 1740 . . . . . . . . . . . o.60 Temperature of(degrees) melting

has been shown, 'Ontamination from the use Of gas furnaces

by Or

Of

coke in furnaces' On abandoning t h e use of gas furnaces a n d employing

SPECTROSCOPIC E X A M I N A T I O N

I n order t o confirm t h e results of t h e chemical analyses for small quantities of impurities, t h e arc spectra of a number qf samples of iron a n d iron-carbon were studied b y Dr. K* Of this Bureau, t o whom we are indebted for t h e d a t a given below: MAGNESIUM-The spectra fully confirmed the chemical tests. Line 28j1.1A showed strong in two samples containing 0.007 and 0.010per cent Mg while it was very faint or absent in

ia

1 For an account of the methods used, cf. K. Burns, Bull. Bur. Stand., (1916), 179-196.

2 24

T H E J O U R N A L O F I N D U S T R I A L AiVD ElVGINEERING C H E M I S T R Y

unfused irons and in alloys which failed to give magnesium by the chemical method. sIucoN-Line 2881.58 showed presence of silicon in the alloys, but unfused iron showed no traces, MANGAXESE-Sevei-al manganese lines show in various samples of electrolytic iron that have been melted. These lines are absent from the spectrum of unmelted electrolytic iron. The faintness of the manganese lines as compared with those given by a specimen containing 0.01 per cent of this metal indicate a very low manganese content. CHROMIUM-The statements made in regard to manganese also apply to chromium. coPPER-Lines 3247.7 and 3274.1A are always present though SO faint as to be questionable in unmelted specimens. NICKEL AND COBALT-several nickel and cobalt lines show faintly in the various samples. In the course of the examination of the arc spectrum of pure iron several faulty identifications have been corrected. 2795.j42 ascribed to magnesium is iron; 3369.555 ascribed to nickel is iron and nickel; 3412.347 ascribed to iron is probably cobalt; and 3443.645 ascribed t o iron is no doubt cobalt. No attempt has as yet been made to clear up all the doubtful identifications in the iron spectrum; the above list is given to show the possibilities in this direction which may be realized by means of the use of pure iron. ZIETHODS O F C H E X I C A L A N A L Y S I S

T h e analyses recorded in this paper were carried out according t o well-known principles with suitable precautions, and t h e methods are given below merely for reference. The chips obtained when making t h e thermal test specimens were thoroughly mixed before weighing portions for analysis. Chips from high-carbon alloys which contained admixed graphite were finely ground and mixed before weighing. CARBOS was determined by the barium-carbonate titration method devised by one of the authors.‘ The chips ( I to j g.) were burned in purified oxygen, passing the products of combustion into a solution of barium hydroxide; the barium carbonate was filtered and washed in an atmosphere free from carbon dioxide (see the original for details of apparatus used), and the barium carbonate was titrated against standard hydrochloric acid, using methyl orange as indicator. SCLFCR was determined by dissolving 5 g . of the metal, contained in an appropriate evolution apparatus having all groundglass connections, in concentrated hydrochloric acid, the gases given off being passed into an ammoniacal solution of hydrogen peroxide. After complete solution of the metal the contents of the evolution flask were boiled for I O minutes while a slow current of purified hydrogen was passed through the solution. The ammoniacal peroxide solution was transferred to a beaker and boiled a few minutes, then the solution was slightly overneutralized with hydrochloric acid and the sulfur precipitated a t boiling temperature as barium sulfate. After digestion for a sufficient length of time the precipitate was filtered, washed, ignited and weighed, and the percentage of sulfur calculated. srx,rcoN-Five to I O g. of metal were dissolved in an Erlenmeyer flask in hydrochloric acid (equal volumes of water and hydrochloric acid of specific gravity I . 2 0 ) , the solution evaporated to dryness and the flask heated on the hot plate a t about 200’ for an hour. The residue was digested with hydrochloric acid of the same strength as that used for dissolving; the insoluble matter containing the silica was filtered off, washed with dilute : J. R. Cain, “Determination of Carbon in Steel and Iron by t h e Barium Carbonate Titration Method,” Technologic Paper IYo. 8s of t h e Bureau of Standards.

V O ~ 8. , YO.3

hydrochloric acid and water, ignited in a platinum crucible and weighed, after which the silica was volatilized with hydrofluoric acid and its amount determined from the change in weight of the crucible after again igniting. The results were then calculated to silicon. PHOSPHoRus-The usual method of precipitation as ammonium phosphomolybdate was employed, and the phosphorus estimated by comparing the volume of the precipitate with that produced by treating a standard steel in the same way. MANGANESE-The sodium bismuthate method was used. COPPER-Ten to 2 0 g. of metal were dissolved in a slight excess of hydrochloric or sulfuric acid, and hydrogen sulfide passed into the hot solution until all the copper was precipitated. The precipitate was filtered off, and, after washing the paper carrying it, was transferred to a porcelain &ucible, and the whole ignited until all the carbon was burned off. A little potassium bisulfate was added and the copper oxide brought into solution by fusion, following by leaching with water and filtration. The solution was compared with a standard solution colorimetrically, either by an ammonia or ferrocyanide method, or by both. MAGNESIUM-Tell to 2 0 g. of metal were dissolved in aqua regia, the solution evaporated to dryness, and dehydrated. The residue was dissolved in I : I €1‘21 and silica removed by filtration. The iron was extracted by the ether method. After the removal of the iron, hydrogen sulfide was passed through the solution (acidified with acetic acid) to precipitate copper, etc. Manganese and iesidual iron were removed from the filtrate by bromine and ammonia and the magnesium precipitated as magnesium-ammonium phosphate. The accuracy of the above procedure was checked by running duplicates to which sniali amounts of a magnesium salt had been added. NCKBL ASD COBALT-The solution of I O g. of the iron in HNO8 HC1 was evaporated to dryness, dehydrated, taken up with HC1 of I , I specific gravity, filtered, the filtrate evaporated to a small volume, and the iron removed by the ether method. Copper was precipitated with hydrogen sulfide, and the iron and manganese in the filtrate were precipitated by ammonia and bromine, The filtrate was acidified with acetic acid, and nickel and cobalt were precipitated as sulfite from the boiling solution. The two metals were either weighed as oxides or deposited electrolytically from ammoniacal solution, the two methods giving concordant results, The oxides (or metals) were dissolved in hydrochloric acid, the solution was neutralized, and finally made acid with acetic acid and the cobalt precipitated as K3Co(N02)6. After filtering and igniting this precipitate a t a low temperature, the cobalt was dissolved, reprecipitated with hydrogen sulfide, and finally weighed as CoSOa. The nickel was determined by the dimethylglyoxime method in the filtrate from the cobalt. The sum of these determinations checked very closely with the total (Ni Co) found directly.

+

+

SUMNARY

Methods have been developed for producing laborat o r y samples of iron-carbon alloys, of a very high degree of purity; sources of contamination of melts and means of eliminating t h e m have been described; a method for producing magnesia of a satisfactory degree of purity for making crucibles t o be used in work of this kind has been developed; a procedure for making small ingots, which are sound and free from blowholes, without t h e use of deoxidizers has been vorked out. A series of iron-carbon alloys containing 99.96 per cent of the two elements has been prepared, t o serve as a basis for the redetermination of the iron-carbon equilibrium diagram. BUREAUOF S T A N D A R D S WASHIXGTON, D. C.