200 The Crystalline Structure of Iron at High .
By Walter Rosenhain , B.A. , D.Sc .
, and J. C. W. Humfrey , B.A. , M.Sc .
, of the National Physical Laboratory .
( Communicated by R. T. Glazebrook , F.R.S. Received November 10 , \#151 ; Read November 25 , 1909 .
) [ Plate 2 .
] In a paper on " The Crystalline Structure of Metals , " * * * S Prof. J. A. Ewing , F.R.S. , and one of the present authors described observations which led to the conclusion that the internal structure of metals is truly crystalline and that plastic deformation takes place either by mechanical twinning or by means of a series of minute slips which occur on the gliding or cleavage planes of the metallic crystals .
This conclusion , with slight modifications , has since been supported by one of the present authors in papers on " The Plastic Yielding of Iron and Steel , " f and on " Deformation and Fracture in Iron and Steel .
In another direction the generalisation quoted above has received further support from the explanation which it has afforded for the failure of metals under alternating stresses and fatigue .
S In these investigations , however , strain was applied to the metal at the ordinary temperature ; the present authors desired to extend the experiments to the study of the modes of deformation of metals at high temperatures .
The behaviour of metals at such temperatures is of interest because for many practical purposes permanent changes of shape are applied to heated metals , as for instance , in such familiar operations as forging and hot-rolling .
Further , iron and steel are known to undergo marked changes of constitution when heated above certain temperatures known as their " critical " points , and the study of the behaviour of these metals under strain at high temperatures promised to throw fresh light upon their true constitution .
Finally , special interest attaches to the manner in which iron behaves under strain at high temperatures , owing to the peculiar character of the slip-bands which appear on a previously polished specimen of iron or very mild steel when subjected to plastic strain at the ordinary temperature ; while the slip-bands seen in such circumstances in copper , lead , and most ductile metals are * Ewing and Rosenhain , * Phil. Trans. , ' A , 1900 ( vol. 193 ) .
+ Rosenhain , ' Journ. Iron and Steel Inst. , ' 1904 , I. J Rosenhain , ' Journ. Iron and Steel Inst. , ' 1906 , II .
S Ewing and Humfrey , " The Fracture of Metals under Repeated Alternations of Stress , " ' Phil. Trans. , ' A , 1903 ( vol. 200 ) .
The Crystalline Structure of Iron at High Temperatures .
201 rectilinear and regular in character , in iron they are usually more or less curved and irregular .
It is thus a matter of interest to determine whether , if deformation by slip occurs at high temperatures , the slip-bands then formed would retain the irregular nature of those formed in the cold , or whether the heated iron would behave in a manner similar to other metals .
Experiments made with a view to ascertaining the structure and constitution of iron and steel at high temperatures have been undertaken , among others , by Saniter , * Osmond and Cartaud , f Baykoff , | and Oberhoffer .
S The method of applying strain to a specimen of metal while at a high temperature has , however , dnly been employed by Osmond and Cartaud , and their work does not appear to cover the ground contemplated in the present research .
The experiments described in the present paper are principally concerned with one variety of iron ; they are of a preliminary character and have been made by means of apparatus which the authors regard as inadequate for the full study of the subject ; the construction of more satisfactory appliances and their use for a fuller study of the phenomena will , however , occupy so much time that the authors think it desirable to publish the present preliminary account of the results already obtained , particularly because they believe that some of their observations constitute novel facts which have a direct bearing on certain important and somewhat doubtful points of theory .
The material which has been employed for most of the experiments in the present research is a pure form of commercial iron or rather mild steel of very low carbon-content .
A chemical analysis of the impurities present in this material gave the following results:\#151 ; Per cent. Carbon ... ... ... ... ... 0*029 Silicon ... ... ... . .
0*118 Sulphur ... ... ... ... .
0*012 Phosphorus ... ... ... .
0*044 Manganese ... ... ... . .
0*014 Arsenic ... ... ... ... . .
0*016 Apart from the presence of one-tenth of 1 per cent , of silicon , this material constitutes a close approach to pure iron .
In its original form ( transformer sheet ) this material probably contained a notable proportion of oxygen or of * Saniter , " Carbon and Iron , " ' Journ. Iron and Steel Inst. , ' 1897 , III .
+ Osmond and Cartaud , " Crystallography of Iron , " 'Journ .
Iron and Steel Inst. , 1906 , III .
+ Baykoff , " Sir la Structure des Aciers aux Temperatures Elevees , " ' Revue de Metallurgies July , 1909 .
S Oberhoffer , " Metallographische Beobachtungen im Luftleeren Raume bei Hoheren Temperaturen , " ' Zeitschr .
f. Elektrochemie , ' No. 19 , 1909 .
202 Dr. Rosenhain and Mr. Humfrey .
Crystalline [ Nov. 10 .
iron oxide , and this proved detrimental to the authors ' experiments .
The plan was therefore adopted of first heating the metal to redness ( about 900 ' C. ) in a small electric tube furnace in a current of pure dry hydrogen ; in some cases this treatment was followed by heating to redness vacuo and allowing the metal to cool in the vacuum , but observation proved that this latter treatment produced no sensible change in the subsequent phenomena In studying the process of deformation of metals by slip at the ordinary temperature , the mode of experiment consists in first providing a suitably shaped specimen with a surface carefully polished for microscopic examination and subsequently subjecting the specimen to strain without allowing the polished surface to be damaged by either oxidation or external mechanical injury .
For the purpose of extending these observations to high temperatures it becomes necessary to comply with the same conditions , and a serious difficulty arises from the great sensitiveness of heated metal to oxidising or otherwise chemically active gases .
Some previous experimenters have employed an atmosphere of hydrogen or nitrogen for the purpose of preserving the polished surfaces of their specimens from chemical action , but the earlier experiments by one of the authors* with iron heated in an atmosphere of hydrogen , and the fact that iron and nitrogen are said to combine at high temperatures , led the authors to carry out the heating of the metal in the present experiments in a high vacuum .
Preliminary tests showed that with a vacuum of the order of one-hundredth of a millimetre of mercury , polished specimens of iron could be heated to a bright red heat without undergoing sensible tarnishing .
: .
For the purpose of applying the requisite stress to the specimens of metal , and of heating them electrically while in the exhausted vessel , the apparatus shown in fig. 1 was employed .
The steel stem A carries a small rectangular Fig. 1 .
frame open on one side ; the bar D is hinged to the bottom bar of this frame and the specimen P to be heated and strained\#151 ; in the form of a strip of sheet * Ewing and Rosenhain , " The Crystalline Structure of Metals , Second Paper , " * Phil-Trans ' .
, ' A , 1901 ( vol. 195 ) .
Structure of Iron at High Temperatures .
203 ; 1909 .
] metal\#151 ; is attached to the upper ends of the fixed bar C and the hinged bar D. Between the fixed and hinged bars a strong spiral spring S is placed and this spring is held in compression by a binding-wire of steel W which relieves the specimen of all stress .
The specimen P is electrically insulated from the bar D , while the wire W is completely insulated from the frame by the interposition of slips of mica .
By means of suitable electric connections it is thus possible to heat the specimen to any desired extent by an electric current , while the stress from the compressed spring can be applied to it whenever desired by passing through the wire W a current sufficient either to soften it very much or to fuse it entirely .
The whole of this apparatus is inserted into a conical filter-flask , the stem A and the requisite electrical connections passing out through an india-rubber stopper .
While this apparatus is obviously of a preliminary character , on account of the difficulty of either heating the whole of a specimen to a uniform temperature or of accurately measuring the temperature attained , the circumstance that different parts in the length of a specimen attained widely different temperatures has specially facilitated some of the observations which the authors regard as particularly interesting .
When a strip of iron of the quality described above has been heated and strained in vacuo by means of this apparatus , the surface , which had previously been polished , presents a number of special features ; by observations made on specimens which had been heated without straining , the authors have been enabled to distinguish between those features which arise as a result of heating alone and those which follow upon the application of strain .
A specimen which has been heated in such a manner that the temperature-of its hottest ( central ) part exceeded 1000 ' C. , after cooling down in the exhausted vessel , shows under the microscope an appearance which varies-along thq length of the specimen according to the temperature which had been reached at each point .
At and near the ends , where the metal had not reached a visible red heat , the surface retains unaltered its original bright and smooth appearance .
As the region is approached where the temperature-reached visible redness ( 700 ' to 800 ' C. ) a disturbance of the surface becomes apparent , a system of interlacing lines ( appearing as fine black lines under normal illumination ) are seen , and these show the characteristics of two superposed sets of crystal boundaries such as would arise from minute differences of level between adjacent crystals .
Since there has been no etching-action on the surface in the present experiments , such differences of level may probably be ascribed to some relative displacement of the crystals among themselves .
This displacement the authors regard as arising from a volume change undergone by the iron in passing through the critical point 204 Dr. Rosenhain and Mr. Humfrey .
Crystalline [ Nov. 10 , which occurs in this material at a temperature of 770 ' C. on heating .
Since this change is reversed on cooling at 763 ' C. , a second series of mutual displacements would occur , thus producing the second system of lines referred to .
When the central portions of one of these specimens , which had been heated but not strained , is observed under the microscope , particularly if the vacuum employed has not been quite high enough to prevent a minute amount of oxidation , a certain shading or tinting of the surface is found ; this shading is in its nature the same as the colouring of ordinary metal sections by the oxidation which occurs in the well-known process of heat-tinting , and it may be regarded as indicating\#151 ; in the manner of a very faint etching\#151 ; the structure of the metal while at the highest temperature .
The structure thus outlined differs very materially from that indicated by the black boundaries already described , upon which it is in fact superposed ; the crystals of the new system are larger , possess much more regular outlines , and show numerous and well-defined examples of twinning .
In this respect the structure closely resembles that observed in " 7-iron , " as seen in nickel and other alloy steel at the ordinary temperature , and as observed by Baykoff* in specimens etched with hydrochloric acid gas at 1120 ' C. The photomicrograph fig. 2 ( Plate 2 ) shows , under a magnification of 500 diameters , both the double system of black boundaries ( which would be seen alone in a cooler part of the specimen ) and the structure existing at the highest temperatures as indicated by the faint heat-tinting .
When a specimen which has been strained while its central portion was at a bright red heat ( exceeding 1000 ' C. ) is examined under the microscope in a similar manner , viz. by commencing at one of the ends of the specimen where the metal had not been seriously heated and proceeding towards the central parts of the specimen where the highest temperature had been reached , a corresponding series of observations is obtained , the effects of strain now to be described being superposed on the effects of heating alone as described in the previous paragraph .
At the extreme end of the specimen the original smooth polished surface is found unaltered , thus indicating that the stress applied had not been sufficient to produce plastic extension in the cold or nearly cold metal .
As the region of higher temperatures is approached the effects of strain make their appearance in the same manner as if the specimen had been strained in the cold:\#151 ; the boundaries of the ordinary ferrite crystals make their appearance and the surfaces of the crystals become cross-hatched by slip-bands .
These effects increase in intensity up to a certain point , but then a sudden change is found .
* Baykoff , paper already cited .
1909.1 Structure of Iron at High .205 Along a well-defined line running across the specimen , the appearance of slip-bands ceases , the surface of the metal beyond this line is simply covered by the double system of boundaries already referred to , but no signs of plastic deformation are visible , at all events in the region lying close to the dividing line .
The microscopic appearance of this line of demarcation is a striking one and is illustrated in fig. 3 under a magnification of 100 diameters ; the photograph is taken in such a way as to represent increasing temperature of the specimen from left to right .
The large amount of slip , as well as the dark areas due to the mutual displacement of the crystals which are visible on the left of the dividing line , indicate the severity of the deformation , undergone by this part of the material .
This line of demarcation is not only a striking microscopic feature of specimens treated in this way , but it is also readily visible with the unaided eye ; such a specimen exhibits two roughened areas near each end , followed\#151 ; after a sharp line of demarcation\#151 ; by a smooth area , while the centre of the specimen again shows the roughening due to plastic deformation .
When the microscopic examination of the surface of such a specimen is carried further towards the central ( and hottest ) parts , the appearance of the surface undergoes a further change .
At a little distance from the dividing line described above , minute slip-bands can again be noticed ; these increase in number and intensity , and on proceeding still further towards the centre of the specimen the slip-bands become well defined and very regular in character , and they frequently assume the stepped character of twinned crystals .
This appearance is illustrated in fig. 4 under a magnification of 150 diameters .
These twin crystals as revealed by slip-bands correspond exactly with the twin crystals as indicated by the heat-tinting by the residual gases as described above .
A specimen treated in the manner described thus shows three distinct regions , corresponding to three definite ranges of temperature , and the observations already described point to the existence of very considerable differences between the mechanical properties of the metal corresponding to these three regions .
According to the theory of Osmond* and Roberts-Austen , f iron is capable of existing in three allotropic modifications , known as \#171 ; , / ?
, and 7 , the transition from each of these to the next being marked by an evolution of heat during cooling and an absorption of heat during heating .
In spite of numerous experiments directed to that end , it has not hitherto been possible\#151 ; so far as the authors are aware\#151 ; to establish any convincing * Osmond , * Journ. Iron and Steel Inst. , ' 1890,1 .
t Roberts-Austen , " Reports to the Alloys Research Committee , " ' Proc. Inst. Mech. Eng. , ' 1897 and 1899 .
.206 Dr. Rosenhain and Mr. Humfrey .
Crystalline [ Nov. 10 , difference between the mechanical properties of iron just above and just below these critical temperatures .
The observations of the present authors , however , appear to furnish some evidence of the existence of very profound differences between iron as it exists in the three ranges of temperature described above* For this reason they have endeavoured to establish a connection between the critical changes occurring in their experiments and the critical points of iron .
The first step in this direction consisted in obtaining the " cooling curve " of the actual material employed in these experiments ; this was done by means of a modification of Roberts-Austen 's " differential " method , the result of both heating and cooling observations being plotted in fig. 5 in the form of AieW 4^\#151 ; -\#151 ; 900* 1 L J. | i\ L tn Irfto* * % \ Ni 1 \ 0 tt ; I a/ W _ l \ !
t j I ' \lt ; \#151 ; SO\amp ; !
Fig. 5 .
the " derived differential curve."* In this curve , as would be anticipated from the very low carbon-content shown by chemical analysis , the critical point due to the carbon-transformation at or near 700 ' C. is absent , while the positions of the two higher arrest-points correspond accurately with those found by other observers for similar material.f The next step consisted in endeavouring to determine the temperature attained during the heating and straining process by various points in the length of the specimens .
The method adopted was in principle that of Joly 's * Rosenhain , " Observations on Recalescence Curves , " 'Proc .
Phys. Soc. , ' vol. 21 , 1908 .
+ Carpenter and Keeling , " The Range of Solidification and the Critical Ranges of Iron Carbon Alloys , " ' Journ. Iron and Steel Inst. , ' 1904 , I. 1909 .
] Structure of Iron at H Temperatures .
207 Melclometer.* Minute particles of various salts possessing well-defined and accurately ascertained melting points were placed upon the rear surface of the specimen and were observed during the heating and straining process by means of a microscope of moderate magnifying power .
The exact instant when a given particle of salt melted could thus be ascertained , and by making the rise of temperature very gradual the desired maximum temperature could be obtained .
On removing the specimen from the apparatus an examination of the rear surface at once disclosed which particles of salt had become fused and which had remained unaltered .
The substances found most useful in the present experiments were pure fused sodium chloride ( melting-point 801 ' C. ) and pure potassium sulphate ( melting-point 1070 ' C. ) .
As a result of a number of observations made in this way the authors find that in those parts of the specimens where sodium chloride and potassium sulphate had both been melted the microscopic features remained those of the central or hottest region described above ; this result agrees with the view that this region is that of the existence of 7-iron above 902 ' C. In that region where potassium sulphate remained solid , but sodium chloride had melted , the features observed were those of the intermediate zone corresponding with the range of / 8-iron , from 760 ' C. to 902 ' C. The marked transition line was found to lie nearly midway between two particles of sodium chloride , one of which had fused while the other remained unaltered ; this agrees with the identification of the transition line with the critical points on the cooling and heating curves representing the change from / 3- to a-iron .
Although the means at the disposal of the authors in the present apparatus did not enable them to carry this identification to any closer degree of approximation , there appears to them to be a sufficient degree of correspondence between the three allotropic forms of iron as defined by the cooling-curve method , and the three different regions observed in their experiments to justify them in referring to the latter as the / 3- , and 7-conditions of iron .
The general fact of this identification is further supported by experiments which the authors have made with two metals which show no critical points in their heating and cooling curves , namely platinum and copper .
Thin strips of these metals were heated and strained in a manner exactly resembling that described above for the iron specimens .
The resulting surface appearances are somewhat complicated , but the authors ' observations establish the fact that the surface appearances from the coldest to the hottest part of the specimens show no sign of any discontinuity or change other than the gradual change due to the increasing temperature and consequently greater softness of the metal .
The conclusion is therefore justified that the discontinuities * Joly , ' Proc. Roy .
Irish Acad. , ' 3rd Series , vol. 2 , p. 38 .
208 Dr. Rosenhain and Mr. Humfrey .
Crystalline [ Nov. 10 observed in iron are directly associated with the critical points , since metals not possessing critical points exhibit no such discontinuities .
The general bearing of the authors ' observations and inferences on the questions outlined at the beginning of the present paper may now be considered .
As regards the three allotropic modifications of iron , the results indicate that a-iron becomes weaker and softer with increasing temperature ; in a strip of uniform thickness observation shows that the maximum amount of extension occurs in those parts corresponding with a-iron at the highest temperature at which it can exist .
In certain specimens this observation was rendered still more striking by using an amount of extension which resulted in the fracture of the specimen : this , occurred just on the colder side of the transition line marking the a/ /3-change .
In view of the fact that the central portions of the specimen were in some cases nearly 300 ' C. hotter than the point where fracture took place , this observation points to a very remarkable discontinuity in the tensile strength of the material ; ordinarily\#151 ; and the authors ' experiments on copper and platinum confirm this statement\#151 ; the tensile strength of a metal diminishes steadily with increasing temperature , and a specimen of uniform cross-section breaks at or near the hottest part ; in the case of iron , therefore , we have the striking fact that , as a result of the a/ /8-transformation , the iron becomes stronger and harder in spite of the tendency of increasing temperatures to reverse this relation .
The fact that in the / 3-iron , near the transition temperature , there are Very few signs of slip or disturbance of any kind , further shows that not only has this material a greater tensile strength , but that it is also harder and has undergone practically no plastic deformation at all under a stress sufficient to rupture an equal section of a-iron .
In view of the fact that the allotropic theory , as applied to the explanation of the hardening of steel by quenching , depends upon the existence of a strong and hard / 3-variety of iron , this observation appears to possess special interest ; it bears out the contention that the failure of all attempts to materially harden carbon-free iron by quenching is solely due to the difficulty of suppressing the / 3/ a-change by rapid cooling .
The existence of numerous and regular slip-bands in the region of 7-iron confirms the inferences drawn by previous investigators that the " 7-iron found in certain alloy steels , such as manganese or nickel steel , corresponda to the 7-iron existing in pure iron at high temperatures .
The truly crystalline nature of the hot 7-iron is further clearly indicated by the manner in which the slip-bands follow the steps of the twin crystals .
Aa to the relative hardness of 7-iron , little can be said from the present experiments ; the iron in the 7-range has undoubtedly undergone much .
1909.1 Structure of Iron at High .209 more plastic extension than that in the / 3-range , but it must be remembered that the relatively very high temperatures attained in this region probably cause a material reduction of stiffness and strength .
The passage of the material through the / 3/ 7-transformation is not , however , indicated on the specimens by any well-marked line such as that which is found at the boundary of the a/ /3-ranges.* This may , however , be explained by the fact that in these hotter portions of the specimens the temperature gradient was much smaller , so that the transition would appear more gradual .
As regards the general question of the manner in which metals undergo plastic deformation at high temperatures , the present experiments have not yet been carried far enough to afford a complete answer except in regard to iron ; in that material the authors have found that at high temperatures , up to and above 1000 ' C. , the crystalline character and mode of deformation is retained by the metal , while the characteristic curved and forked slip-bands of ferrite at ordinary temperature are replaced by regular and approximately straight bands , resembling those seen in annealed wrought copper .
In platinum , also , the formation of slip-bands at high temperatures has been observed , but in this metal the surface appearance of specimens strained at high temperatures is of a complicated nature , requiring further study .
The general conclusions , which the authors regard as established by the preliminary experiments described above , may be summarised as follows:\#151 ; ( 1 ) That iron at temperatures up to 1100 ' C. behaves as a crystalline aggregate , and undergoes plastic deformation by a process of slip on the cleavage or gliding planes of its constituent crystals ; this may or may hot be accompanied by mechanical twinning .
( 2 ) That iron between the ordinary temperature and 1000 ' C. exists in three distinct modifications possessing widely different mechanical properties , and that the temperature ranges in which these modifications exist are consistent with the view that they are identical with the a- , / 3- , and 7-forms of Osmond and Roberts-Austen , as indicated by cooling-curves .
( 3 ) That / 3-iron , although existing at a higher temperature , is harder and stronger than a-iron , and that the a/ /3-transf ormation involves a volume change .
( 4 ) That 7-iron , as found in approximately pure iron at high temperatures , possesses the characteristic structure and properties found in the " 7-iron " alloy steels .
The authors desire to express their thanks to Dr. R. T. Glazebrook , F.R.S. , for his interest in the work , and the facilities afforded to them for carrying out the experiments in the Metallurgy Department of the National Physical Laboratory .
vol. lxxxiii.\#151 ; A. p

