EQUILIBRIUM OF IRON-CARBON ALLOYSWITH GASES
July, 1946
1163
scribed column of about fifteen theoretical plates. AnalyAcknowledgment.-The authors are grateful sis by distillation indicated that, in addition to the com- to Dr. Earl W. Balis of the Research Laboratory pounds reported above, appreciable amounts of si1:icon tetrabromide were present in the crude products. The for the carbon and hydrogen analyses reported in following table lists the results of five successive runs ((the this paper. values of ?I refer to the formula (CHJ.SiBrc-,). Summary Product. ml.
CHrBr.
Time,
B.
tu.
g.
1180 1390 1250 1170 1710
118 119 118 96 114
847 1021 962 586 976
492 530 493 249 488
Composition (vol. %) n = 3 n = 2 n - !
18 7 7 3 4
60 35 38 11 37
[Ct3NTRIDCrIONrROM THE
20 32 35 33 33
RESEARCH
Methyltribromosilane, dimethyldibroniosilane and trimethylbromosilane, isolated as products of the reaction of methyl bromide with sintered copper-silicon powder a t 300°,have been purified and characterized. SCHENECTADY,
LABORA~OKY, 'IT.\ITCD
K.Y.
RECEIVED JANUARY 28, 19-16
s1 ATCS s1 EEL C O R P O R A T I U Y ]
Equilibrium of Iron-Carbon Alloys with Mixtures of CO-CO, and CH,-H, BY
RODNEY P. SMITH
B.'
As the consistency and range of the available d a t a on the equilibrium between carbon dissolved in iron and definite mixtures of either carbon monoxide with carbon dioxide, or methane with hydrogen, leaves something to be desired, and as accurate information on this matter has considerable practical significance, it was decided to reinvestigate these systems. Accordingly measurements were made of the concentration of carbon, induced in iron a t a number of temperatures ranging from 750 to 1200', by contact with each of a series of ratios of one or other of these pairs of gases. The equilibria involved may be represented by the equations
The gas mixture in equilibrium with ferrite of composition B is also in equilibrium with austenite of composition C, and the gross composition of the solid phase may have any value from B to C depending on the relative amount of the two solid phases. A further increase in carburizing power of the gas causes the ferrite phase to transform completely to austenite, and the gas ratio fixes its composition in the range C to D. The line S'E' represents the equilibrium of austenite with graphite; thus the gas ratio, r , in equilibrium with austenite of composition D is also in equilibrium with graphite, and its value (r') a t this point gives the equilibrium constant K B or Kd for the reactions
2H2
2Hn
CO,
+ C (dissolved in Fe) = CH,,
+ C (graphite) = CH,; PCH,/P&%= r: = Ks
(3)
P C H lP&,a = r, ' a = K , (1)
COX = C (graphite) = 2CO; I J ~ ~ o / P c=o 27;
P',,IPCo,a =
738 and 910' except that, as temperature increases, the range of stability of ferrite decreases
+ C (dissolved in Fe) = 2'20,
rl/a =
K, (2)
where P is measured in atmospheres and a is t'he activity of dissolved carbon. For either system in equilibrium a t constant temperature and pressure, a given value of the partial pressure ratio r determines the composition of the solid phase, or vice versa, so long as only one solid phase is present. I n the temperature range considered the solid phase may be ferrite, a solid solution of carbon in a-iron (body-centered cubic), or austenite, a solid solution of carbon in y-iron (face-centered cubic), the transition temperature depending upon the carbon content; accordingly interpretation of the results to be presented is facilitated by reference to the pertinent portion of the ironcarbon equilibrium diagram represented in Fig. 1, in which, for clarity, the scale of abscissa for the line G P is ten times as large as for the other lines. If we consider the isothermal line A , B , C, D it is evident that for a gas mixture of low carburizing power the solid phase is ferrite, and as carburizing power increases each gas ratio produces a solid phase of definite composition in the range A 1.0
=
li, (1)
The case is similar for all temperatures between and t h a t of austenite increases. For the temperatures 1000 and 1200' and the carbon range 0 to 1.553, the only solid phases encountered are :iustenite and graphite.' If graphite is chosen as the standard state for carbon, the constants K1, K , of equations (1) and ( 2 ) become identical with the constants K3, K4 of equations ( 3 ) and (4). Equation (1) and ( 2 ) may then be written 12,
=
0,
=
Pcl,,:P&K3 = rI l K 3 P;o!Pco*K, = 72,'Ki
;I:%: (?a)
where a, is the activity of carbon in iron relative to graphite, and rl or r2 is the ratio a t equilibrium with carbon dissolved in iron a t il Concentration corresponding to the activity ace Thus nieasurements of .these ratios aiid the corresponding carbon ( 1 ) I n the case of CO-CO? mixtures l o w in cdrboo monoxide, the system is complicated by the presence of au iron oxide phase. The above qtatemeiits epply only t i > gas m r t u r e s which will not oxidize irou a t these temperatures.
1164
RODNEY P. SMITH
Vol. 68
Iron Samples.-Most were of electrolytic iriiii?; tlie others of carbonyl iron.2 They \vcre i i i the form of strips about 0 . 2 tnni. thick, 1 cm. wide arid 6-15 cm. long, wound iiito a loose coil. Each was provided with a i i iiitcgral hook for convenience of suspenyioii; thi.; hook was removed before the v m i l ) l e \vas aiialyzetl. This form of saniplc 1135 thr atlvantagcs of coinpactncss with lai gv siirfacc, arid that it is easily prepared for aiialysis without danger of contamiliatioii h y carboiiaceous material. Preparation of Constant Gas Mixtures.-The mcthods of preparation, purification mid utilization of flowmeters t o produce CO -PO? mixtures of controlled cornposition Lvc'rc exactly as described by Darken arid ( ; L I I T ~ , ~ except that thc inisture was further tlricti hy passing it through a tube surrouiidetl by a Dry Ice-acetone mixture. The CO 'jC0, ratio (r2) as determined by analysis of a number of representative saniplei checked that calculated from the caliliratiori of the flowmeters within a few parts i i i a thousand. The CH4-HH2mixtures were made from comiiitwial hydrogen and either methane from Uuttonwillow field, California or commercial methane'; the total impurities in either were less than 1% and the purified iiiixtures from either source gave the same result. Pure methane is not stable a t ternpc"ratures required in the purification train ; therefore primary mixtures containing about 10, 21; and 60% methane were made in hyclrogcii cylinders which were allowed to staiitl by a radiator for some months t o ensure thorough mixing. These mixtures were il.0 0.5 1.o 1.5 analyzed by combustion over copper oxide Carbon content, weight ?{-, with a reriroducibilitv of about * 0 . 5 ' 7 of methane present. Subsequet1t analysis Fig. I ,--Iportion of the iron carbon equilibrium diagram : 0 Smith, those f,,r graphite SolubiIity; 0 ,l.Iehl a n d Wells; Wells; 0 Gurry, that these nlixtures uliiform. These primary mixtures were purified by cquilibrium with graphite; @ Gurry, equilibrium with carbon from passillg sllccessivc,ly tllrough ascaritc, ac. toluene The abscissa for the line G P is tcn times that for the other lines. tivated alumina, comer gauze at 450". ascaritc, a c t i v a t h iiumi;a, anti finally content permit calculation of the activity of car- phosphorus pentoxide; hydrogen was purified by a similar The final mixture5 iverc made from these gases by bon and of various other thermodynamic quantities train. a flowmeter system similar to that uwd foi the CO-CO? pertaining to these s y s t e m . Equations ( l a ) and mixtures. From the flo\vtneter system the gas passed (?a) also give a convenient means of comparing through a mising chamber, eiitered the furnace t u b e a t its the results obtained with the two sets of gases, pro- lower end through a mercury seal and left a t atniospheric through its upper extension beyond the furnace. vided consistent values of K 3 and K4are available. pressure The flowmeters For the CH4-H:: mixtures were calibrated The range of composition covered was from by measuring the time to deliver a fixed volume; the small ten1pcra ture correct ioiis were calculated E roni the kn o\vn 0.OOOSto carbon in the iron phase, 0 to 3% methane in mixture with hydrogen, and 0.3 to viscosity5 of the two gases. At coristarit tetn~)ernturct h e 27y0c x b o n dioxide mixed with carbon monoxide ; floivinetcr constant Ii' is givcii by the eqnatioii ~
/ I
l.sf7
the temperatures were mainly SO0 to 1000' but a few iieasurements a t 730 and 1200' are included.
Experimental The iron sample was suspended from a platinum wire in a x-ertical porcelain tube, held a t constant temperature, through which passed upward a streain of the fixed gas mixture, and kept there until equilibrium had been attained. In Some cases equilibrium was approached by carburization, in others by partial decarburization of a previously carburized sample. When equilibrium was attained with the particular gas mixture, the sample was quenched bv raising i t into the upper glass extension of the porcelain furnace tube; its carbon content was then determined.
Itigation. I 10) bp\tr.in, " h l l ( ~ ) u uf C a r b o n a n d I r o n , " McGraw-kiill Book Ctt , Iuc , > : r w \':irk. N Y , 193b.
i l l ) J H. Whitely. J. Iron C- Slvel I n s l , 116, 293 (1927). (12) R . S c h e n k , P. Kurzen a n d H. Wessiuck, Z. a n u r p . allpem. Chem., '203, 1 % (1932). (13) R o b e r t F. hlehl a n d Cyril Wells, A m . I n s f . .M:vz. M e t . Engrs., 116, 429 (1937).
EQUILIBRIUM OF IRON-CARBON ALLOYSWITH GASES
July, 194G
TABLEI11 EXPERIMENTAL MEASUREMENTS OF EQUILIBRIUM OF COC 0 2MIXTURES WITH AUSTENITE:Y, = P&,/Pco,, WHERE
0.04
P IS IN ATMOSPHERES
;i
;i
-2
zj, 0.02
1
0.0 0.0
0.2 0.4 0.6 0.8 Carbon content, weight %. Fig. 3.--Carbnn content of austenite a t 800" in relation point ~ ; C is .lower limit of stable aust o rI = Y c H ~ / P ~ H tenite a t 800 '.
composition appreciably. A few measurements were made with CHI-HZ mixtures at 1200' but they were not continued because the silicon pickup proved to be about ten times greater than at 10oOO. Table 111 gives the experitnental results obtained with electrolytic iron and CO-CO:, mixtures for the temperatures 800, 1000, and 1200'. A typical example of the reproducibility of these ineasurenients is illustrated in Fig. 4 where r2 is plotted against carbon content a t 1000°. Figure 4 also provides a comparison of these results with those of the later previous investigation^.^^'^^^^^^^^^^ The data of Diinwald and Wagnerg are in good agreement, the others in poor agreement, with the results of this investigation. The accord with the data of Diinwald and Wagner is illustrated better in Fig. 5, where log r2 for a number of carbon concentrations is plotted against l/T, the rcciprocal of the absolute temperature, for our results a t 800,1000,1200° and for theirs a t 940, 1000 and 1070°; the values of r2 were read from a large scale graph similar to Fig. 1. In general their results, for a given carbon concentration and teniperature, give a value of r2, slightly higher than this investigation; the difference could be a(:counted for by an uncertainty of about 1 P in mewured temperature. Since the variation i i i teiii( 1 1 ) Arthur Brninlry nod H a r r y Denuis L u r d , J . Ciiem. Sui. , 16-11 (1932). (15) A. Johansson a n d It. Von S e t h , J , IroH S l e e l Insl.. 411, 2!l6
(1926). ( 1 6 ) M . I,. Becker. ibid.,111, 3 3 i (1030). (17) G. T a k s h a s h i , .Science Rcporls, Tohoku I m p . C n i s . . 16, l i i 7
2.25 2.48 2.65 2.86 3.11 3.12 3.63 1.21 4.50 1.87 ,511 5.54 d.64 ti 07 6 , 5;i f i 75
0.313 ,356 ,377 ,405 ,443 ,453 ,522 ,568 , 008 ,647 .tifil
.72fi ,726 , 7ii5 ,815 ,831 ,838" ..MR ,875 ,878 .89%
(j.81
6 89 7.24 7.31 7.32 7.52 7,7,5
-
Final carbon content,
Final carbon content, %,by weight
rz
0.03
12000
1000~
800'
(192ti).
1167
Final carbon content.
70, by
It
1 2 3 4
98 49 12 21 7 29 13 8 27 4 43 4 50 2 70 8 81 1 99 1 113 3 130 2 131 7 132 4
weight
0.0360 ,0487 ,0563 ,0740 ,133 ,242 .455 ,655 ,810
,963 1 1 1 1 1 1
081
206 321 402 466 471b
% bu
r?
weight
3.75 3.80 5.83 7.14 7.23
0.0148 .0141 .0217 ,0252 ,0273 ,0450 . 109 ,215 ,416 ,413 ,738 ,942
12.-l(i
30. 3 01.4 122.5 123.1 243.0 352.2
,911
,929 ,916
-77
I . i f
Carbonyl iroii. b Sample ten times thicker and one half weight of other samples. IC
I
I
I
I
(1. I 0.8 1.2 1. t i C\3rlmi coiitcnt, weight I'ig. l.--Carboi1 coiitcnt of austeiiite a t 1 W ) "in relation to rr = P 2 c o / P c o s : 0 Smith; 0 Diinwald and Wagner: i) Dramley and Lord; P Decker; 0 Johanssoii and Von Seth; t Tahahaslii 1)
:('.
1168
RODNEY P. SMITH
Vol. 68 thermocouple was calibrated from the melting point of potassium chloride, such a temperature difference is not unreasonable. For samples equilibrated with COCOz mixtures at 800°, the austeniteferrite transition occurred a t a value of rz between 2.18 and 2.25, which corresponds t o 0.323y0 carbon a t point C of Fig. 1. This is slightly greater than the value (0.315%) for the CH4-Hz mixtures, also given by Mehl and Wells.13 Equilibrium with Graphite; the Constants K 3 and &.-The constants Ka and Kd at 800 and 1000' were determined by direct measurement of the equilibrium of graphite with each pair of gases, and indirectly from the austenite equilibria and graphite solubility by the method outlined in the introduction. For the direct measurement, the gas composition at equilibrium was determined graphically from the gain or loss of weight in unit time, for a number of gas mixtures a t fixed total rate of flow, of a block of Acheson graphite impregnated with a small amount of iron to serve as catalyst. A typical series for each pair of gases a t 1000° is plotted in Fig. 6. There is a slight scatter, as is to be expected in such a measurement, but the partial pressure of COz or CH4at which the weight remained unchanged can be determined by interpolation with a precision of 1 0 . 2 a t 1000" and *0.47& at 500'. The indirect determination requires
2.5
2.0
1.5
0,' 5. V
cr, M
- 1.0
0.5
8
9 10$/T. austenite i; equilibrium with graphite Fig. 5.--Flot of lag r, against 10'lTfor even increments of carbon content: and a short extrapolation (always less 0 Smith; 0 Diinwald and Wagner. than 0.06% C)of our austenite ecruiI
perature of their furnace was
* 7 O ,
and their
TABLE 11' VALUESOF TIIE BQUILIURIUM CONSTANT K , A N D K4,W H E R E P IS I N ATMOSPHERES 2
1
1, O C .
3
From spectroscopic and Direct Extrapolation calorimetric measurement of r data
Kt 800
0 0453
1000 AH',
0 00902 -21,000
800 1000 AH'
7 33 138 6 39,900
Pcn,/P&, 0 04.56 0 04fB U 00950 0 00949 -21,330 -21,700
4 As given by Chipman
=
K4 = Pao/P,.oz 7 08 7 28 151 13'7 2 39,970 40,300
0 051 0 0102 -2l,%W G 22 117 39,830
librium data to the carbon content corresponding to saturation with graphite. The solubility of graphite in austenite a t SO0 and IO0O0 determined by interpolation of the (lata given by WellsIs at Z3S and 8-10', and by Gurrylg at temperatures near 950 and 1100" was O.SS()~i,and l.j0%, respectively. Table I V brings together values of the constants K Q arid IC4 for equilibrium with graphite a t 800 arid 1000" derived: (I) from direct measurements of r i or Y: i n equilibrium with graphite, ( 2 ) by extrapolation of the experimental curve for r , or r2 to the carbon content corresponding to saturation with graphite, ( 3 ) by interpolation of the recently published constants derived from (18) Cyril Wells. Z'viliis. A m . S a . d f r l u l s . 2 6 , '281)(1038). ( I D ) R. W. Curry, Trans. A m . I?zsl. M i ! t i n g d i e l . E ! i g r s . , 160, 147
(1942).
EQUILIBRIUM 01; IKON-CARBON ALLOYSWITH G,w:s
July, 1940
1 lG9
-
cc;
' .? 0 . i j
I
iYi
I
I
3 2 0 hi
i i
'6
.-
0.4 0.2
i
4
0
0.02 0.04 ( I o/i Aitoiiifractioii carboii. Fig. ;.-Activity (a,) of carbon relative L O grapllite (nican from two sets of mcasurenicnts) a t 1000 O in rclntion to atom fraction carbon in iron.
0
made with CO-CO:! mixtures a t SOUo, but the results indicated that the system tends to follow the metastable austenite equilibriuni. For -0.2 -0.1 0 0.1 0.2 instance, the samples (Table 111) with O.91G arid Gain in weight, mg. per hour. 0.929% carbon contain considerably more carbon Fig. (i.--R'eight change of a block of Acheson graphite ' 1 1 than corresponds to the graphite solubility a t this relation ,to the partial pressure of either COz or CHd: temperature, yet these points are in accord with 0 CO-CO, mixtures, 0 CH4-H2 mixtures. the austenite equilibrium. Samples with 0.95 calorimetric and spectroscopic clata,2o (4) from to 0.97570 carbon (analysis by gain in weight) Chipman's summary of the data then available. 2 1 were in accord with the austenite equilibrium The agreement between our direct and indirect after being in the gas mixture for 2 1 hours; howmeasurements (columns 1 and 2 ) is excellent; ever, after being quenched and returned to the but it may be noted that this is not an entirely furnace the carbon increased as would be exindependent check, since the same gas mixtures pected if the graphite phase were then present in were used in both and any error in gas compos:[- the sample. The Activity of Carbon Relative to Graphite,tion would affect both equally. This table also includes values of the heat of reaction AH',j, T h e activity of carbon, a t a given concentration, MIf4 as calculated froiri each pair of values of h-3 relative t o graphite, may be calculated from the or k',a t 800 arid 1000°,respectively. The heats measurenieiits with eithcr gas iiiixture by equain column 3 are i n better agreement with those cf tion ( l a ) or (2.) and the data in Tables TI, I11 and columns 1 or 2 th.an are the equilibrium constants IV. The results of such calculations, based on values of rl and r2 interpolated from the experiupon which they are based. An attempt to make an indepcndeiit determina- mental curve for rcgular increments of carbon tion of this graphite solubility a t 1000° by passin:; content, and of K;jor IC4 given i i i column 2, Table CO-C02 mixtures, of coinpositions near that for IV, are presented in Table 1:. I t is to be rioted equilibrium with graphite, over samples of elec- that for a given carbon content, the activity of trolytic iron packed in graphite containing :i carbon is only half as great a t 1000 as a t SOUo, small amount of ferric oxide, was not successful; corresponding to the lower saturation concentrathe carbon content of the sample changed with tion of graphite in austenite :it the lower tciiiperachange in gas ratio although to a less extent than ture; also that a t both temperatures the activity when no graphite was present. However, the in- as calculated froiii r1 is greater than that froiii r 2 . tersection of the lines obtained by plotting r2 'l'hese two calculated activities would be itlenticul against carbon content for these results and thosc: only if (1) the ratio of the selected values of K j for the austenite equilibrium should give the and K4 is exact, (2) the measurement of r l , r2 and graphite solubility; the value thus derived is the carbon contcnt a t equilibrium arc equally 1.49iy0 carbon in good agreement with that accurate in both cases, ( 3 ) thc final iron ])haw is il.50%) determined from the direct measure- identical and contains only carbon or, at most, identical sinall ainounts of other elements, (4)the ments of U'ells arid Gurry. It should be possible to determine graphite solu- gases are equally ideal, an assumptioil which bility, and equilibrium constant, by a method seems unlikely to introduce an appreciable error. similar to that used to fix the carbon concentra- The extent to which these conditions were iultion a t which ferrite bcgins to transforin to aus- filled is discussed below. (1) T h a t the differeiicc betwccii the two sets tenite. A few measurements of this type were of activities is not entirely due to mutually incoii(20)Wagman, Kilpatrick, Taqlrir, Pitzer a n d R o s i i n i , D u i sisteiit values of the constants K3, K4 is indicated Slardurds J . R c s i u ~ c l i 34, , 1-13 (1943). by the fact that the ratio rs,lrl, whicli would be i z i ) joi,n c i l i p n l a t l , i i r i ~ ' J i ec/tc,ttl., . a d , 1013 ( 1 9 2 2 1 . I
I
RODNEY I-'. SMITH
1liO
1-01. A8
TABLE V
ACTIVITYOF Carhon content weight %
c 1 .2 .3 .4
.5 .6 .7 .8 .88 .9
C A R n O N IN , 4 U S l E N I T E
RELATIVE TO GRAPHITE AT 800 A N D
AT 100fl" Temperature 1000'
___n_l__----------
'rernprrnture 800°
--
-I
rz
ri
12
2.04 2.80 3.61 4.47 5 , 40 fl .42 7.28
0,0132 ,0181 ,0232 ,0287 ,0344 ,0405 ,0456
0 390
0 280
155
,397 509 .G30 753 888 1 . 000
,385 .39G ,614 , 712 ,882
153 1iic, 151; 157 158 1W
1.000
5 .6 11.1 17.4 23.7 30.9 38.7 4lj. 8 55.7 I . .
--
65.0
1.0 1.1 1 2 1 3
li3.0
85.8 97. 2
109.7 122. 8 137.2
1.4
1.50
the equilibrium constant, K ,for the reaction CO CHI = 2CO 2H2 shows a regular trend with carbon content covering a range of about 3Yc a t 800' and lsPh a t 1000'. In each case this K is lower than that computed from spectroscopic and calorimetric data but approaches it as the carbon content increases. The agreement between the two sets of activities is not improved by using the constants given in column 3 , Table IV, instead of those of column 2 . ( 2 ) The values of y I and 72 should be consistent within a few per cent. The value of r2 may have been mnewhat more accurate than that of 71 since the preparation of the CC)-CO2 mixtures was riore direct. The difference between our two sets of measuretnents is shown better in the very sensitivc plots (Fig. Y and 9) of log ( 7 n l / n 2 ) against nzlnl,where nd i i l is the ratio of the number
+
+
I
I
I
K x 10-2 = r d r i X 10 - 2
n/K3
rdK4
0.050 .097 ,144 195 ,247 ,301 ,362 ,424
0.0ll ,081 ,127 . 173 ,225 ,282 ,341 ,406
...
...
...
,447
,896
139 140 141 142 143 144
1,000
144
. 4 01 . 513 ,639 ,720 ,808 ,899 1.000
119
121 137 128 132 13
