SOV., 1912
T H E JOL-RSAL OF Id\7\-nL-STRIAL AlVD E S G I I V E E R I S G C H E M I S T R Y .
THE DETERMINATION OF OXYGEN I N IRON AND STEEL BY REDUCTION I N AN ELECTRIC VACUUM FURNACE.' By \V%r H
\\-ALKER
AXD
WALTER-4.PATRICK.
At the fifth meeting of this Congress, held in Berlin in 1903, F. Lurman presented a paper on the determination of oxygen in iron and steel in which he pointed out the necessity of a new method which would measure the total oxygen in the sample, and especially t h a t portion which is combined with mangahese, aluminium, silicon, and the other oxides not reducible with hydrogen. Since that time no advance has been made in this important field of metallurgical work. The important role which the oxides of iron and the other metals play in determining the physical and chemical properties of iron and steel was first clearly brought out by Prof. Ledebur in 1882. At this time Ledebur proposed the analytical method for determining oxygen nhich bears his name and which is based upon the reduction of the oxide a t a red heat by hydrogen, and weighing the water thus formed. Owing to the fact t h a t at a high temperature hydrogen reduces only the oxides of iron, leaving unattacked t o a large extent the oxides of manganese, aluminium, silicon, titanium, etc., it is obvious that the method is not a satisfactory one, although up t o the present the only one available for the purpose. Other methods have been proposed for determining the oxygen in iron and steel based upon the solution of the constituents not oxides. As examples, may be mentioned the method of dissolving the sample in ethereal solutions of iodine and bromine; or that based upon the volatilization of the iron in a stream of chlorine gas a t a high temperature; but none of these methods have proven of any value. Although the detrimental effect of combined oxygen in iron and steel has been known since the time of Ledebur's first paper, i t is only within the last few years that the full significance of these effects has been appreciated. That iron oxide can exist in iron and steel in more than one form seems certain, and that the oxides of the different metals associated with the iron should have characteristic effects is surely to be expected. But in the absence of any satisfactory analytical method for determining the oxide present, the relation between cause and effect has been of necessity a most imperfect one. I t is hoped that the method of determining oxygen now to be described will be an aid in establishing the connection between the oxygen content and the physical and chemical properties of iron and steel. I n the now well known reaction wherein the oxides of the elements are converted into the corresponding carbides by heating them with a n excess of carbon t o a high temperature in a n electric furnace, the oxygen is given off quantitatively as carbon monoxide. The quantitative formation of carbon monoxide is aided by a n excess of carbon at high temperatures, and by a low pressure; also, as has been pointed out by Moissan, the refractory oxides are more readily reduced in the presence of metallic iron. Paper presented :it the Eighth International ConbTess of Applied Chemistry, New York, September, 1912
799
The details of the method as a t present worked out are as follows: A vacuum furnace of the Arsem type as supplied by the General Electric Co. is employed, and is shown in Fig. I . The gun metal chamber A of the furnace of 2 0 cu. in. capacity rests inside the water E
II
II
5
FIG.1.
jacket R. The cover B is fastened to the chamber by means of 18 cap screws, D , and the joint is made tight by a rigid lead gasket. The tube - J , through which the air is exhausted, is soldered into the cover. The window tube G is fastened to the cover by six cap screws, the joint made tight by another lead washer. The mica window E is placed in the top of the window tube. Current is led in through the electrodes W , which are brass tubes containing running water. The graphite heater L is fastened to the electrodes by means of the clamps G. The crucible is supported by the stand as shown in the figure and is thus placed in the hottest part of the furnace. Twenty t o twenty-five grams of the sample of metal in which the oxygen is to be determined are placed in a small graphite crucible and about four or five grams of finely powdered graphite added. The crucible is then placed in the furnace, the cover bolted down, and by means of a small rotary oil pump operated in series with a Geryk pump a vacuum of 0 ,o I of a mm. is obtained in the furnace in less than fifteen minutes. After thoroughly exhausting the furnace, the cooling water is turned on and the crucible and contents heated to about 500°-6000 with the pump still running. This is necessary in order to pull away as completely as possible the oxygen absorbed by the heater and crucible. As carbon does not begin t o reduce the oxides below 900' there is no danger in heating the crucible up to 500'. After fifteen minutes the furnace is allowed to cool, which i t does very rapidly. Nitrogen that has been dried over sulphuric acid and phosphorus pentoxide is now allowed to enter the furnace until the latter is about half full. This is then pumped out and the crucible
800
T H E J O U R N A L OF I N D U S T R I A L AND E N G I N E E R I N G C H E M I S T R Y .
again gently heated. Only by such a treatment is i t possible t o reduce the oxygen left in the furnace t o a reasonably small value. It seems impossible t o pump the absorbed oxygen entirely out of the furnace, but by washing with dry nitrogen i t can be almost eliminated. The stopcock leading t o the pump is now turned off and seventy volts applied t o the electrodes which causes a current of about 2 0 0 amperes to flow through the heater. A high temperature is reached very rapidly, the metal melting in three or four minutes. Just as the metal melts, violent ebullition very often occurs. This is prevented b y opening the circuit for a short time and allowing the charge t o cool. The metal soon becomes quiet and the heating is continued for twenty minutes. The furnace is then allowed t o cool thoroughly, after which fresh air, or better nitrogen, t h a t has been dried over sulphuric acid and phosphorus pentoxide is allowed t o completely fill the furnace. The gas now in the furnace is analyzed for carbon monoxide in the following manner: The vessel C in Fig. 2 is exhausted by means of a
Toepler pump, and connections made with the furnace as shown in the diagram. The stopcock from the furnace is opened and if the gas in the furnace is under a pressure other than atmospheric, the difference is shown b y the differential gage E . Correcting the barometric pressure b y this amount gives the pressure of the gas inside the furnace. Stopcock F is now opened and the vessel and the pump filled with gas. A decrease in pressure corresponding t o the volume of gas taken from the furnace is shown by the gage
Nov., 1 9 1 2
E , and we are thus able to calculate the fraction of gas taken from the furnace. Bv this method we are practically free from a n y errors due t o temperature variation. . The gas is now slowly forced over iodine pentoxide which is heated to 130' C. and here the carbon monoxide is oxidized to carbon dioxide, liberating a n equivalent amount of iodine. The latter is absorbed in a I O per cent. solution of potassium iodine and subsequently titrated with N / I O Osodium thiosulphate.1 A much more simple method of withdrawing a n aliquot part of the gas for analysis, if a large quantity of mercury is available, is t o use a mercury aspirator instead of the Toepler pump. By replacing the reservoir C by a liter bottle which may be filled with mercury, a known fraction of the contents of the furnace may be withdrawn, and forced directly through the iodine pentoxide tube. The first point to be definitely determined was, how free from oxygen can the furnace and its heating element be made; in other words, using a sample of iron known t o be free from oxygen, what amount of oxygen will be shown in a blank experiment? It
was found that the oxygen indicated by the iodine liberated from the iodine pentoxide was due t o three factors: first, a certain practically constant amount of iodine set free from the iodine pentoxide when air or nitrogen free from carbon monoxide was led over i t ; second, the actual oxygen absorbed on the walls and on the heater of the furnace; third, the moisture in the furnace. That iodine pentoxide when heated t o 1 5 o O in a current of dry air or nitrogen would give up small amounts of iodine has been recognized by previous investigators (L. A. Levy, SOC.Chem. Indust., 30, 1 4 3 7 ) . We have found t h a t by reducing the temperature t o 130' the oxidation of the carbon monoxide is complete, while the decomposition of the pentoxide is reduced t o a minimum. By withdrawing from the furnace at each analysis a constant volume of gas and drying the same over phosphorus pentoxide, a uniform blank amounting t o 0.006 gram oxygen for the furnace contents was obtained. The weight of oxygen absorbed on the walls of the furnace and in the heater when a vacuum of 0.01 mm. was maintained was found in blank analysis using iron free from oxygen t o be 0 ~ 0 2gram. If the air used t o dilute the furnace contents previous t o withdrawing the sample of gas for the carbon monoxide determination be not dried over phosphorus pentoxide, the amount of oxygen 1
Nicloux and Gautier, Comfit. rend., 126, 746; Kinnicut and Sanford,
J . Ant. Chem. Soc., 22,
14.
Nov.,
1912
T H E J O U R N A L OF I A ; D U S T R I A L A iVD EA'GINEERI*YG C H E M I S T R Y .
indicated may rise to 0.035 gram. By first exhausting the air from the furnace, and then filling with nitrogen and re-exhausting the amount of oxygen shown by a blank, analysis was found to be between 0 . 0 1 2 0 and 0.0129 gram. The accuracy of the method, so far as the ability t o carry out the operations without introducing errors not corrected for as above, was shown b y the fact t h a t a sample of iron or steel containing oxygen may be heated, and the oxygen determined; upon reheating the same charge with additional carbon no further formation of carbon monoxide is obtained. To determine the accuracy with which the oxygen of a sample of iron or steel will be converted to carbon monoxide, a number of analyses were made using a n iron of very low oxygen content, and adding known quantities of the various oxides in the pure form. I n addition to iron oxide, only the very refractory oxides of aluminium and' silicon were used; other easily reducible oxides such as manganese and copper will introduce no difficulties. The weighed amount of the pure oxide together with a known weight of iron was placed in the graphite crucible and covered with powdered graphite and heated as described. From the oxygen as determined was subtracted t h a t due t o the blank and the iron and the following results were obtained: Weight of Oxygen oxide. calculated. 0.1695 Ferric oxide., . . . . . . . . . . . . . . . . . 0.5650 Ferric oxide.. . . . . . . . . . . . . . . . . . . 0.6525 0.1960 Ferric oxide.. . . . . . . . . . . . . . . . . . . 0.8400 0.2520 Ferric oxide.. . . . . . . . . . . . . . . . . . . 0,1516 0.0455 Aluminium oxide., . . . . . . . . . . . . . . 0.4925 0.2310 Aluminium oxide . . . . . . . . 0.1860 0,0875 Aluminium oxide. . . . . . . . . . . . . . . . 0.3065 0.1440 Silica. . . . . . . . . . . . . . . . . . . . . . . . . . 0,0673 0.0360 Silica . . . . . . . . . . . . . . . . . . . . . . . . . . 0,0320 0.0181
Oxygen found. 0,1700 0.1940 0.2440 0.0480 0.2040 0.0810 0.1260 0.0342 0.0171
While the results as obtained are not so accurate a s is desired, the reduction being somewhat incomplete, and therefore the results uniformly low, the method has served to explain the discrepancies noted between the oxygen content as determined by the Ledebur method in certain samples and their physical and chemical properties. As examples of such analyses may be mentioned the following: Sample No. 1 is a special heat of an open hearth steel to which iron ore and an excess of manganese was added in the ladle. The finished steel gave every evidence of having a high oxygen content, although, by the Ledebur method, but 0.006 per cent. oxygen was obtained. When determined by the vacuum furnace reduction method the real oxygen content of the steel proved to be 0.20 per cent. Sample No. 2 is a high grade of open hearth steel in which the Ledebur method detected no oxygen. The vacuum method showed that there was 0.09 per cent. present. KO.3, open hearth steel. KO.4, same as No. 3 but from ingot to which ore was added. Samples KO.5 and 6 represent two ingots from the same heat of open hearth steel of high quality. T o No. 5 was added some iron oxide as the ingot was poured. The oxygen content of the ingot giving satisfactory results was 0.065 per cent. n-hile that to which the ore was added and which was highly unsatisfactory proved to he 0.31 per cent. h-o. 7, open hearth steel. No. 8, open hearth steel. h-0. 9, open hearth iron. No. 10, open hearth iron. Xos. 1 la and 1 1 b are duplicate analyses uf an ingot iron of early manufacture. The vacuum furnace method indicates about three times as much oxygen as the Ledebur method. Nos. 12a and 126 are ingot iron of later make and show an oxygen content of but 0.10 per cent. The full analyses are given in the following table:
801 0.
C. Mn. 0.19 0.92
No. 1 2 3 4 5 6 7
. . . . .
0.17 0.65 0.12 0.17 0.09 0.18 0.14 0.24 0.09 0.33 0.08 0.33 0.01 0.03 0.01 0.04 0.01 trace 0 . 0 1 trace 0 . 0 2 0.03 0.02 0.03
8 9
10 lla llb 12a 126
P.
S.
0,052 0.123
...
...
0.097 0.065 0.061 0.070
0.064 0.088 0.087 0.092 0.068 0.070 0.007 0.008 0,002 0.002
0.065
Cu.
Si.
0. Vacuum Ledebur. furnace.
. . . . . . . . . . . . 0.017
...
0.015 .., 0.019 ... 0.009 ... 0.006 0.17 0,005 0.22 0.003 0.20 0.004 0.19
0.036 0.050 0.015 0.015 . . . . . . 0.015 . . . . . . 0.029 0 . 0 0 4 0.0014 0.043 0.029 0.004 0.0014 0.043
0.006 0.000
... ... ... ...
0.009 0.010 0.037 0,052 0.069 0.076
...
...
0.29 0.09 0.11 0.33 0.31 0.065 0.021 0.039 0.056 0.064 0.23 0.21 0.10
0.11
I t is intended t h a t this paper be considered a preliminary communication, a n d it is hoped that the method will be both simplified and improved by further work. I t is published a t this time in the hope t h a t others interested in the effect of oxygen on steel will find in the idea something of value. MASSACHUSETTSINSTITUTE O F TECHNOLOGY, BOSTON.
THE METHODS OF THE UNITED STATES STEEL CORPORATION FOR THE COMMERCIAL SAMPLING AND ANALYSIS OF PIG IRON.' By
THE
CHEMISTS' COMMITTEEOF
THE
C.S. STEELCORPORATION.
PREFACE.
I n conformity with the design of the officials of the United States Steel Corporation for the standardization of the methods employed in the sampling and analysis of all materials encountered in their various lines of manufacture, the Chemists' Committee pre.sents this compilation of standard methods for the sampling and subsequent analysis of molten pig iron. I n selecting the methods, the committee employed the same line of procedure as in former cases, viz., the careful consideration of all the methods employed in each laboratory of the corporation, evolving therefrom the several methods herein described, the immediate adoption of which is desired. The services of Messrs. W. B. N. Hawk, Wm. Brady and C. H. Rich, a sub-committee appointed for the preparation of this pamphlet, are gratefully acknowledged. INTRODUCTION.
A quite obvious cause of nonconformity of results of comparative analyses in the various laboratories of the Corporation has been a n apparent lack of uniformity of method in the sampling and analysis of molten iron. This condition is particularly apparent with regard to the determination of the sulphur, numerous discrepancies in which would appear t o indicate the absence of, and the necessity for, some uniform method of procedure. The estimation of the sulphur by the gravimetric method and the determination of the remaining constituent elements of the iron, are susceptible of a most satisfactory degree of accuracy. The successful issue of the volumetric determination of the sulphur, however, is largely dependent upon 1 Paper presented a t the Eighth International Congress of Applied Chemistry, New York City, September, 1912. Copyright 1912 by J. M. Camp, Chairman Chemists' Committee, Munshall, Pa.