Nov.,
1920
T H E J O U R N A L OF 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
t h e absence of a copper deposit. It may be noted also t h a t the bright surface here shown is of approximately t h e same width as the ferrite ring shown for this same specimen in Fig. 14. I n addition t o t h e experiments above described, a few metals and alloys used as inserts in heat-treated steels have failed t o give any striking results. Whether this has been due t o ( a ) failure t o heat t o the necessary temperature, ( b ) extremely low migration velocities of t h e special elements or compounds, or (c) absence of a n y effect of diffused material, is not now clear. These materials and others will be used for additional experiments. SUMMARY
A considerable number of alloys and special steels have been turned into small rods and driven into holes in carbon steels, t h e whole assembled specimen then being heated t o temperatures above the transformation range for the steel. The section of t h e slowly cooled piece then shows, in nearly all cases, ferrite segregation around the insert. T h e hypothesis t h a t has been advanced t o explain this phenomenon is t o the effect t h a t the special element, elements, or compounds have diffused into t h e surrounding steel, and t h a t they have there exerted a n influence toward throwing ferrite o u t of the aiwtenitic solution when t h e latter cools into t h e transformation range. These experiments are cited as possibly throwing some light upon the character of the influence of nonmetallic inclusions upon ferrite segregation. The inclusion must have a certain slight solubility in austenite, and the concentration of the dissolved matter is therefore greatest in the immediate vicinity of the inclusion, There must also be equilibrium products of reactions occurring between the material of the inclusion and the steel, These products also must have a slight solubility, and they also are localized about the inclusion. Such a condition of localized dissolved impurities might have the effect of starting ferrite crystallization first about the inclusion, thus breaking down t h e state of ferrite supersaturation t h a t always occurs in hypoeutectoid steels between t h e temperatures As and Ar3. If these hypotheses are correct i t is easy t o see t h e importance of uniform distribution of all elements of carbon steel, as well as of alloy steels, as otherwise i t is not possible t o have uniform carbon distribution, and the finished piece cannot be brought into its best condition by any ordinary heat treatment. The effect of phosphorus upon carbon distribution is then but one illustration of the general law. Practically all of the elements t h a t can enter steel can exert a similar influence. The importance of this fact in eonnection with t h e subject of nonmetallic inclusions comes from the fact t h a t the inclusion furnishes a continuous supply of the dissolved impurity, and t h a t therefore no amount of heat treatment can cure its evil effects, while the opposite is true with the dissolved element or compound t h a t is not associated with discrete particles of inclusions.
kb95
AN ELECTROMETRIC METHOD FOR DETECTING SEGREGATION OF DISSOLVED IMPURITIES IN STEEL' By E. G. Mahin and R. E. Brewer DEPARTMGNT OF CHEWISTRY, PURDUE UNIVERSITY,LAPAYETTE, INDIANA Received July 23, 1920
As a result of work described in two earlier papers% i t was concluded t h a t segregation of dissolved material derived from nonmetallic inclusions, as well as dissolved phosphides, silicides, sulfides, and elements of alloy steels, causes carbon segregation in steel, iron carbide usually leaving t h e region of higher concentration of impurity as the steel cools from above Ara. I n such a case t h e ferrite grains of the regions so contaminated should possess a solution tension dif ferent from t h a t of purer grains in other parts of t h e steel body. An electrolytic cell in which two of such grains form the electrodes against a common electrolyte, as for example impure ferrite-KC1-pure ferrite, should generate a n electromotive force which could be measured b y use of a suitable potentiometer system. It is manifestly impossible t o isolate two ferrite grains of a given piece of steel in such a way as t o make a n e. m. f . measurement of such a cell practicable. On the other hand, i t should be possible, a t least in principle, t o make successive measurements of the e. m. f . of each of two systems, in each of which a standard calomel or hydrogen electrode is connected against t h e grain in question, the same electrolyte being used in t h e two cases. I n this way a comparison could be made of the electrode potentials of various grains of apparently pure ferrite and differences in degree of purity detected.
-
PRECAUTIONS TO B E OBSERVED
Many precautions must be observed in t h e attempt t o apply this method in practice. Some of these may be enumerated briefly as follows: I-Immersion of the specimen in the electrolyte is, obviously, not practicable. I t then becomes necessary t o pick out certain areas in the polished and etched section, making contact of the electrolyte with these alone. The microscope must be used for observation of the specimen, and a n extremely fine point of some sort must be used for touching the grain under investigation, this point carrying the solution of t h e electrolyte and establishing the necessary connection with t h e remainder of the circuit. 2-In any electrode-potential measurement, contact of electrode with electrolyte must be maintained for a certain period of time, in order t o establish equilibrium conditions. Using the method here discussed, i t then becomes necessary t o maintain a microscopic point or tube in contact with the grain under observation for some time. This presents considerable mechanical difficulties. 3-Evaporation of water from t h e solution a t t h e point changes the concentration of the electrolyte at. the contact surface. Using a molar solution of potassium chloride, the relative change in concentration is 1 Presented a t the 60th hfeeting of the American Chemical Society. Chicago, Ill.,September 6 t o 10, 1920. *THISJOURNAL, 11 (1919),739, 12 (1920),1090.
1096
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 Vol.
not serious except in case measurements of a high degree of precision are desired. 4--The positive component of the dbuble layer over the electrode contains ferrous ions, but oxidation of these ions occurs t o some extent during the time t h a t is required for equilibrium t o be established. To prevent this a current of hydrogen was directed over the specimen during the earlier experiments. I t was soon found t h a t this made little observable difference in the values obtained and its use was discontinued. 5-Polishing produces a layer of amorphous metal over the crystalline structure underneath, according t o the now generally accepted theory of Beilby.1 This amorphous (strained) material is slightly more positive t h a n crystals of the same materiala2 This layer is removed by etching, so t h a t i t should have no effect upon experimentally determined values. 6-Etching the specimen with nitric acid in alcohol might, conceivably, produce a condition of slight passivity of the iron. This was tested by comparing the potential of such a specimen with t h a t of the same specimen etched with alcoholic hydrochloric acid, a less satisfactory etching solution from the standpoint of clearness of the etched surface. There was no consistent difference between the two sets of values. 7-Exposure of the surface t o the air before testing gradually raises the electrode potential by oxidation. This interference was avoided by polishing and etching each specimen just before making the measurements. Greater precautions with regard t o all of these points would have t o be taken if a high degree of precision were required. But even in the purest steel there will be so much variation in either ferrite or pearlite grains, from point t o point, t h a t values will probably have little significance beyond the third, or possibly even the second, decimal place. EXPERIMENTAL
APPARATUS-Fig. I illustrates the manner of assembling the apparatus, the illuminating system and potentiometer being omitted from the drawing. The specimen under investigation had a short copper wire soldered t o its lower surface, this wire dipping into a cup of mercury placed beneath the stage of the microscope. This made a connection with the potentiometer which permitted focusing the stage of the metallographic microscope a t will. The specimen was held by a Sauveur magnetic holder. This, of course, resulted in grounding the specimen through the microscope. Experiments carried out with the specimen insulated from the microscope and with a nonmagnetic holder showed t h a t neither of these factors affected the observations of the e. m. f . in the third decimal. The electrolyte was a molar solution of potassium chloride in boiled distilled water. A solution of ferrous sulfate is not suitable for this purpose because of t h e facility with which hydrolysis and oxidation take place. A clear solution cannot be kept without having a n excess of acid present. 1 Proc. Roy. SOC., T2 (19031, 227. f Lord Kelvin, Phil. Mag., 46 (1898), 82; Hambeuchen, University of Wisconsin, Engineering Bulletin 2 (1901), 235; Jackson, Trans. A m . S o b Mcch. Eng., 22 (1900-1901), 816.
12,
No.
11
Contact with the specimen was made by means of a very fine, capillary, glass tube, G, drawn t o hair-like dimensions. An attempt was first made t o use a porous or tubular fiber, either animal or vegetable, cemented in the end of a fine tube. No success has as yet been attained in this direction. Further experiments along this line are in progress. Fig. z will indicate the appearance of t h e capillary point as applied t o one of the ferrite rings obtained in the course of the work described in the second paper dealing with this subject.l It will be noticed t h a t only the tip of the tube is in focus, so t h a t the grain structure of the steel is visible through the upper part of the tube.
I
I
B-mercury;
FIG.1-ASSEMBLED APFARATUS C-copper wire: S--specimen of steel: G-glass tubing with capillary point: R-rubber tubing: F-molar KCI; E-normal calomel electrode
A normal calomel half-element completed the electromotive system, and a Leeds and Northrup precision potentiometer, with wall-type galvanometer having a sensitivity of 93 megohms, was used for making the measurements by the Poggendorff compensation method. The apparatus was shielded from the interference of stray fields by t h e method of White.2 For the illumination of t h e microscope a n Ediswan “Pointolite” lamp, I O O candle power, was used, the light passing through the usual condensing system t o the 45 O reflector in the microscope. RESULTS-The measurements t h a t have been made t o date would indicate t h a t this method has possibilities for research purposes, although i t is not considered likely t o have any particular value as a routine testing method. Many variable and unsatisfactory results were obtained during t h e early part of the work and changes in the apparatus were made t o meet difficulties t h a t developed. A single set of measurements will be cited t o show the nature of the results. These measurements were made on a piece of carbon steel containing a n insert of aluminium bronze, the specimen having been heated t o 850’ C. for 7 hrs. and cooled in the furnace. The appearance of the polished and etched section is shown in Fig. 3 . I n the table each group of values represents a series of readings of the e. m. f . of the system for a given microscopic spot on the steel section, the values in1
THISJOURNAL, 12 (1920), 1090.
2
J . A m . Ckem. Soc., 36 (1914), 2011
Nov.,
1920
T H E J O U R N A L OF I N D U S T R I A L A N D ENGIATEERING C H E M I S T R Y
creasing as the readings are repeated. I n each group the value marked with an asterisk is the maximum value, the reading remaining constant t o three decimals for 30 sec. The first column contains values, in volts, of the e. m. f . for various points on the ferrite ring X of Fig. 3. I n t h e second and third columns are given values for various points on t h e body Y of t h e steel.
I097
and aluminium have passed into t h e steel mass, thus causing the appearance of t h e ferrite ring, i t is not t o be supposed t h a t the concentration of these metals in the ring is a t all a constant quantity. The electrode potential would then be somewhat variable, and t h e e. m. f . of the system would vary accordingly. Ferrite grains in t h e body of t h e piece, so far removed from the alloy insert as t o preclude the possibility of contamination, are also undoubtedly impure. They probably contain most of t h e phosphorus and silicon of the steel, as well as traces of other impurities, a n d their electrode potentials also will vary. Considering both of these probabilities i t is yet significant t h a t the starred values for the ferrite of t h e body of the steel are distinctly higher than those €or t h e ferrite ring. ELECTRODE POTEXTIAL
FIG.2-GLASS
CAPILLARY POINT APPLIEDTO RING. X 50
A
FERRITE
For the last-named points, an effort was made t o restrict the contact t o ferrite grains. Where the grains were small i t was not always possible t o do this, but the results are not appreciably affected by contact of t h e solution with pearlite grains, since pearlite is a conglomerate of ferrite and cementite, and since ferrite is the more positive of t h e two substances. TABLE I-E. M. F. IN VOLTS, FOR STEEL WITH ALUMINIUM BRONZE INSERT Different Points on Different Points Different Points on on Ferrite Ring Body of Steel Body of Steel (Con$.) 0.438 0.481 0.506 0.456 0.483 0.510 . . ~ ~ . 0.471 0.486 0.514 0.497 n ,519 *0.471 0.560 U0.519 0.456 *0.560 0.467 0.421 0.504 0.471 0.498 0.503 0.483 0.509 0.512 0.485 *O. 509 0.516 *O.485 0.520 0.447 0.521 0.437 0.468 0.456 *0.521 0.475 0.457 *O. 475 *o. 457 0.494 0.510 0.510 0.537 0.379 0.512 0.541 0.435 0.514 0.459 0.543 0.518 ~. *0.543 0.464 0.519 0.466 *0.519 *O,466 0.432 0.452 0.451 0.500 n mi -.-”_ 0.454 0.478 *O.501 *0.478 0.507 0.517 0.523 Average of Maximum 0.533 Average of Maximum Values for Ferrite 0.536 Values for Body of Ring 0.541 Steel 0.552 0.471 *O .552 0.522
Using 0. j60 volt for t h e value for the calomel halfelement, it will be seen t h a t t h e electrode potentials for the metal, as calculated from these results, have negative values. The average for the ferrite ring is 0.471- 0.560 = - 0.089 volt, and for t h e body of the = -0.038 volt, a difference of piece, 0.522-0.560 about 0.05 volt. This is not in agreement with most of the values t h a t have been found by others for iron. Walker and Dill1 found t h e value 0.156 volt for unstrained Swedish iron in a ferrous sulfate solution, whose concentration was not stated. This value {indicates t h a t the solution was positive t o t h e metal, the opposite of which is true in our measurements. Other recorded results vary somewhat from those obtained by Walker and Dill, b u t variations in purity, occluded gases, etc., are largely responsible.2
~
There is considerable variation between the maxima f o r the different groups in a given This, it would seem, might be expected, since we are dealing with a material whose composition must vary from point t o point. Even upon t h e assumption t h a t Copper
FIG. 3-sTEEL ADJOINING AN INSERT OF ALUMINIUM BRONZE, HEATED FOR 7 HRS.AT 8.50’ C. X 100
It was a t first thought t h a t this indicated a n error in the method of reading, but a comparison of the electrode potential of a sample of polished and etched ingot iron- made by t h e point-contact method above described and by immersion of t h e specimen in t h e electrolyte, shows that the immersion method gives Trans. A m . Electrochem. S O L 11 (1907), 153. See also Richards and Behr, “The Electromotive Force of Iron, and the Effect of Occluded Hydrogen,” Pub. Carnegie Inst., 1906. l 2
1098
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 Vol.
higher readings, more nearly in accord with the results of others for practically pure iron in ferrous sulfate. It will then be concluded, either ( a ) t h a t the pointcontact method does not give true equilibrium values, contact being made for too short a time, or ( b ) t h a t surface tension or some similar influence consistently lowers the actual potential of the grain under observation. The latter hypothesis seems not unlikely, in view of the fact t h a t in approximately the same time period the e. m. f . values for the system containing the immersed specimen of ingot iron were considerably higher than those obtained by point contact. It is not likely t h a t absolute equilibrium has been reached in any case, but the starred values were not changing t o the extent of 0.001 volt in 30 sec. Longer contact than a few minutes is not practicable, evaporation and oxidation then serving t o render the results undependable.
12,
No.
IB
brated in tenths of a cubic foot. A large valve opened into a vertical cylindrical chamber about I O cm. in diameter and 30 cm. long, with clamps a t the open upper end by which steel orifice plates could be fastened in place. There were six of these orifice plates, 2 mm. in thickness, having circular orifices of j, IO, 15, 2 0 , 2 j , and 3 0 mm. diameter, respectively.
SUMMARY
A method has been developed for measuring the electrode potential of a single grain or microscopic point on a metal specimen. Results so far obtained show lower values t h a n are given when a pure metal is immersed according t o t h e usual scheme. No reason for this has been demonstrated, but i t is shown t h a t fairly consistent values may be obtained on a given kind of material. When the method is applied t o a piece of steel containing a segregated ring of ferrite, produced by heating in contact with aluminium bronze, i t is found t h a t average values for the ferrite ring are o . o j I volt lower than for ferrite in the unaffected body of the steel. This indicates a different degree of purity for ferrite under the two conditions, and sustains the theory already a d v a n c e d t o account for the ferrite ring. It is hoped t h a t the method may be improved so t h a t i t may be applied t o the exploration of the regions adjoining nonmetallic inclusions in ferrite, as well as the segregated masses of ferrite often found in steels but not associated with inclusions. INFLAMMABILITY OF JETS OF HYDROGEN AND INERT GAS' By P. G. Ledig BUREAUOF STANDARDS, WASHINGTON, D. C. Received July 29, 1920
The purpose of the experimental work herein described was t o determine the maximum amount of hydrogen which could be used with helium in balloons without losing the advantage of noninflammability. Preliminary experiments t o determine the inflammability of a jet of hydrogen and inert gas under various conditions were made with hydrogen and nitrogen, in order t o save helium. Later work showed t h a t , as was supposed, there was not a great enough difference in the behavior of the gases t o make i t necessary t o use helium throughout. The gas mixtures were made up in a water-sealed gasometer (a meter prover) with a capacity of somewhat over 5 cu. f t . A scale on the tank was caliPublished with permission of the Director of the Bureau of Standards.
FIG.1-RELATION OF BLOW-OFFVELOCITYTO ORIFICE DIAMETER AND HYDROGEN CONTENT
A gas mixture of approximately the composition desired was allowed t o stand in the gasometer for a period of half a n hour or more t o permit diffusion of the hydrogen and nitrogen. The gas was then analyzed for hydrogen content, w&,h a n accuracy of 0.1 per cent. The orifice chamber was now swept out with this gas and the gas stream was then slowed down sufficiently t o permit ignition or proof t h a t the g a s mixture would not burn in air. A small jet of burning coal-gas was used t o ignite the gas stream. If t h e mixture burned, the valve was slowly opened until the velocity of the gas stream extinguished t h e flame by blowing i t from the orifice. Since t h e hydrogen flame was practically invisible, i t wits usually found most convenient t o judge this blowing-off point b y sound. The flame fluttered more and more loudly as i t approached the blowing-off velocity. After t h e flame had blown off, the valve was left open a t t h e same position, and the gas velocity measured immediately by timing with a stopwatch the fall in t h e gas holder during the discharge of a definite quantity of gas. I n practically all cases there was sufficient: gas in the tank t o make from three to five determinetions with a single orifice.
