Inter-crystalline Fracture of Metals under Prolonged Stress .
55 the auroral zone , it is the distance from that region , not from the centre of the zone , that is likely to be the all-important thing .
It will make all the difference to a particular station whether the concentration of current is on the near or the remote side of the zone .
From this point of view , the most promising way of extending our knowledge is to compare individual records from a series of stations lying roughly on a north-south line , and so not differing much in local time .
A station in the extreme north of Scotland or in the Orkneys or Shetlands would seem likely to add much to our knowledge .
On the Inter-crystalline Fracture of Metals under Prolonged Application of Stress [ Preliminary Paper ) .
By Walter Rosenhain , B.A. , D.Sc .
, F.R.S. , and Sydney L. Archbutt , F.I.C. ( both of the National Physical Laboratory ) .
( Received March 13 , 1919 .
) [ Plates 1-3 .
] The authors have recently made a series of observations on some cases of inter-crystalline fracture in various metals , occurring as the result of the prolonged application of stress .
In explanation of these phenomena they have formulated an hypothesis which appears to afford a satisfactory account of the present observations and to correlate them with other well-known phenomena whose exact nature has , however , hitherto remained obscure .
In putting their observations and hypothesis on record at the present stage , the authors are well aware that much fuller experimental investigation of the whole subject is required , and they hope to carry this forward .
The evidence now available , however , appears to them to justify preliminary publication , especially in view of the fundamental interest and great practical importance of the subject .
The present paper relates to a group of phenomena some of which have long been known , in the case of brass , as " season cracking .
" Brass articles which have been manufactured by a process of alternate cold-working and annealing\#151 ; such , for instance , as cartridge-cases and other articles made by operations of cupping and drawing\#151 ; sometimes exhibit a tendency , after a period which may vary from a few hours to several years , to undergo spontaneous cracking .
The occurrence of this type of failure of brass has Dr. W. B/ Osenhain and Mr. S. L. Archbutt .
been a serious manufacturing difficulty and much study has been given to the subject ; as a result , modifications of both the annealing and the drawing processes have been made , which , to a considerable extent , eliminate the trouble .
A satisfactory explanation of " season cracking " has not , however , been put forward so far as the authors are aware .
From the present point of view , perhaps the most interesting fact which has been obser\ ed about \#171 ; season cracking " in brass is that the fracture\#151 ; in those cases where its path can be traced clearly among the micro-constituents of the metal\#151 ; markedly follows the inter-crystalline boundaries .
An example of such a crack is illustrated , under a magnification of 100 diameters , in fig , 1 , where the intercrystalline character of the fracture can be clearly traced ( Plate 1 ) .
The other phenomena , recently observed by the authors , which have led up to the present study of the subject , have been met with primarily in an alloy of aluminium containing both zinc and copper .
Further , fractures observed in such widely different metals as lead and mild steel ( boiler plate ) show features which appear to be closely related tQ the same subject .
In the aluminium alloy referred to it has been found that when the material , after being rolled into sheets , has been subjected to a certain degree of annealing\#151 ; which the authors describe as " over-annealing " \#151 ; it displays a tendency to undergo a peculiar type of fracture which bears a striking resemblance to " season cracking .
" If a strip of this aluminium alloy in the " over-annealed " condition is tested in a tejisile testing-machine in the ordinary way , it shows no specially abnormal properties , being merely a little harder and less ductile than when correctly annealed .
If a strip of the alloy is bent to a curve of moderate radius it springs back when released although it does not become quite straight again .
If such a strip is held in the bent position by wrapping a piece of wire around it , the tension on the wire is at first fully maintained .
After a period of time , however , which may vary , according to the degree of " over-annealing " to which the specimen has previously been subjected , from a few minutes to four days or longer , such a bent strip breaks at or near the sharpest part of the bend , the fracture being generally accompanied by many fine ramifying cracks .
Microscopic examination shows quite clearly that this fracture and the accompanying cracks are strictly inter-crystalline .
A typical example is illustrated in fig. 2 , under a magnification of 150 diameters .
The close analogy between the fracture of such a bent strip of aluminium alloy and that of an article of cold-drawn brass is readily understood when it is borne in mind that a cold-drawn article necessarily exists in a state of great internal stress .
Thus one need only imagine the bent strip of alloy to be bent into the form of a complete circle and the two ends united in Inter-crystalline Fracture of Metals under Prolonged Stress .
57 that position , to arrive at a shape produced by cold-working and holding large internal stresses .
Similarly , a cup drawn from a flat sheet or from a shallower cup , may also be regarded as made up of a number of bent radial strips held together in that position , each strip being heavily " sprung .
" The analogy is completed by the observation that cup-shaped spinnings made of the aluminium alloy in the " over annealed " condition undergo the same type of spontaneous inter-cystalline fracture as that described above .
In the case of lead , observations have been made on the sheathing of a small electric cable which failed after several years ' service , by Assuring of the lead sheath.* Examination shows that parts of the lead sheathing are badly cracked , the cracks clearly following the inter-crystalline boundaries .
Their appearance is shown in fig. 3 , under a magnification of 20 diameters .
The similarity between this case and the two previously mentioned resides principally in the unusual type of fracture\#151 ; since inter-crystalline fissures are particularly remarkable in so ductile a metal as lead .
The evidence for the presence of stress is not so clear in this instance , except that the separation of various parts of the material indicates that there must have been forces at work tending to pull the material apart .
The appearance of the lead crystals , however , shows that the forces at work must have been small , since the crystals themselves show no signs of having been deformed .
It is quite probable that a small amount of stress was put on this cable in the process of laying , or subsequently by the effects of thermal expansion .
Finally , in a certain number of cases of failure of mild steel boiler plates , abnormal types of fracture , following the crystal boundaries have been observed.f An example of a fracture of this kind is illustrated in fig. 4 under a magnification of 500 diameters .
In this case the material , forming part of a large boiler has been exposed to stress for a long time .
Like the three other materials already referred to , this steel when broken in an ordinary test shows the usual fracture running across the crystals and avoiding the crystal boundaries in a systematic manner .
Having recognised the essential similarity between these various examples of metal which have failed by inter-crystalline fracture occurring after the * The sample of cracked lead cable sheathing to which the authors ' observations refer was brought to their notice by Mr. L. Archbutt , F.I.C. , of Derby , who drew their attention to the inter-crystalline character of the fissures .
They have since found other examples of this type of failure , and are informed that it is not uncommon in lead-covered cable , where this is allowed to hang in catenaries between supports .
t A good example is described in Departmental Paper No. 231 , 1910 , on a " Report by the National Physical Laboratory on Some Unusual Defects in the Plates of Two Combustion Chambers on Board a Foreign-going Passenger Steamship , " submitted by the Marine Department , Board of Trade .
Dr. W. Rosenhain and Mr. S. L. Archbutt .
lapse of time , the authors now suggest an explanation which , they believe , furnishes a rational account of the mechanism of this type of fracture and correlates the phenomena with one another and with other known facts .
They have not yet had time to carry out more than a few of the test experiments which their view suggests , but so far as they have gone , the evidence appears to be entirely consistent with the explanation put forward .
The view has been held for some time by a number of metallurgists* and has been particularly advocated by one of the authorsf that a thin layer of amorphous metal , in a state corresponding to that of a greatly under-cooled liquid , exists between adjacent crystals in any crystalline aggregate , and especially in metals .
A considerable amount of evidence has been advanced in support of that hypothesis , including , particularly a series of experiments made in 1912-13 by one of the authors and D. Ewen.| These are of special interest in the present connection .
The manner in which fracture under an externally applied stress will occur in a crystalline aggregate held together by an amorphous intercrystalline cement must depend on the relative physical properties of the-crystals and the cement .
If the cement is of the nature of an under-cooled liquid far below its freezing point such " liquid " will be in an extremely viscous condition , in that respect resembling such a substance as glass .
Towards loads applied at the rates employed even in very slow tensile testing such a substance will behave like a solid of apparently high elastic * The conception that metal can be partially converted into an amorphous or vitreous condition by straining or polishing is due to Beilby , whose papers include : " Surface Flow in Crystalline Solids , '3 ' Proc. Roy .
Soc. , ' A , vol. 72,1903 ; " The Hard and Soft State in Metals , " ' Faraday Soc. , ' 1904 , and ' Phil. Mag. , ' August , 1904 ; " The Hard and Soft State in Ductile Metals , '3 ' Proc. Roy .
Soc. , ' A , vol. 79 , 1907 ; and " The Hard and Soft State in Metals , " ' Journ. Inst. Metals,3 No. 2 , 1911 , vol. 6 .
The further conception that amorphous layers exist between the crystals of unstrained metal first occurs in the following papers : F. Osmond , Discussion on " Transformations of Steel within Limits of Temperature Employed in Heat-Treatment,33 ' Journ. Iron and Steel Inst.,3 No. 2 , pp. 61-3 ( 1911 ) ; G. D. Bengough , " A Study of the Properties of Alloys at High Temperatures , " Journ. Inst. Metals , No. 1 , vol. 7 .
pp. 176-7 ( 1912 ) ; W. Rosenhain , Discussion on above ; Rosenhain and Ewen , " Inter-crystalline Cohesion in Metals , " ' J ourn .
Inst. Metals/ No. 2 , vol. 8 ( 1912 ) ; ibid. , vol. 10 ( 1913 ) ; Rosenhain and Humfrey , " The Tenacity , .
Deformation , and Fracture of Soft Steel at High Temperatures , " ' Journ. Iron and Steel Inst. , ^ No. 1 ( 1913 ) ; J. C. W. Humfrey , " The Inter-crystalline Fracture of Iron and oteel , lion and Steel Inst. , Carnegie Scholarship Memoirs/ 1911 ; " Influence of Inter-crystalline Cohesion upon the Mechanical Properties of Metals , " ' Iron and Steel Inst. , Carnegie Scholarship Memoirs , ' vol. 5 ( 1913 ) .
+ Rosenhain .
See papers cited above , also " Metals , Crystalline and Amorphous.3 ' Paper before Section B , British Association , 1913 , 'Engineering,3 September , 1913 .
J Rosenhain and Ewen .
See papers cited in note * above .
Inter-crystalline Fracture of Metals under Prolonged Stress .
59 limit .
The crystals , on the other hand , are capable of undergoing deformation by slip under relatively light stresses .
Deformation and fracture will therefore occur by rupture of the crystals themselves and not by any tearing away of adjacent crystals from one another .
This is in fact the type of fracture met with under ordinary conditions in all metals in a normal condition , i.e. , in the absence of impurities which are known to exist in the form of weak and brittle membranes along the crystal boundaries .
On the other hand , an amorphous cement of the nature of an under-cooled liquid will become rapidly less viscous\#151 ; and consequently less resistant to displacement and rupture\#151 ; with rising temperature .
By the time that the freezing point of the metal is reached , this liquid will no longer be under-cooled , and will have the full mobility of the ordinary molten metal .
From this condition the viscosity of the liquid will increase continuously , and over a certain range very rapidly , until extreme viscosity is reached at very 1owt temperatures .
In other words , the under-cooled liquid may be expected to exhibit a large temperature-coefficient of viscosity .
This is , of course , well known to be so in the case of glass .
Now , there is no reason to believe that the physical properties of solid crystals undergo any very rapid change with rising temperature ; even at their own freezing point they possess a considerable degree of strength and hardness .
It follows from these considerations , as has been pointed out by Rosenhain and Ewen { loc. cit. ) , that at some temperature between the ordinary temperature and that of the freezing point of the metal , the relative behaviour of the crystals and the cement in a crystalline aggregate exposed to stress will be reversed , so that at any temperature above that point the cement will offer less resistance to displacement than the crystals , so that if the metal is fractured under these conditions , a purely inter-crystalline fracture will result .
This inference has been very fully verified by the experiments of Rosenhain and Ewen already referred to .
They produced entirely inter-crystalline fractures , unaccompanied by any deformation of the crystals , in a number of very pure metals\#151 ; including lead , gold , aluminium , tin , and others\#151 ; at tem-peratures some degrees below their melting points .
For the present purpose , however , it is important to bear in mind that the authors named worked with light loads and did not very fully consider the time factor in such expeiiments .
When these factors are taken into consideration , the conclusion is indicated that if higher stresses had been used , and if they had been allowed to act for a longer time , similar inter-crystalline fracture would have taken place at considerably lower temperatures .
A case in which this conclusion has been fully verified is provided by the experiments of Rosenhain Dr. W. Bosenhain and Mr. S. L. Archbutt .
and Humfrey , * in which the deformation and fracture of a very mild steel has been investigated , both by physical measurements and by microscopic observation when gradually broken in tension at temperatures above the upper critical point\#151 ; i.e. , above 900 ' C. There , it has been shown that , where crystals of y-iron are held together by an amorphous cement at a high temperature , and therefore probably of relatively low viscosity , both deformation and fracture are essentially inter-crystalline , while the process of deformation is found to follow a law which at once suggests the presence of viscous flow .
When it is remembered that y-iron just above Ac3 is still some 600 ' C. below its melting point , it will be realised that there need be nothing surprising in meeting with similar phenomena in brass or in an aluminium alloy at the ordinary temperature .
The explanation of the inter-crystalline fracture of certain metals resulting from prolonged exposure to stress which the authors desire to put forward for consideration may , then , be stated as follows:\#151 ; That the constituent crystals of a metal are held together by an amorphous inter-crystalline cement , which , in some metals at all events , is sufficiently mobile to yield to the tendency of any applied stress to produce flow .
In such cases it would follow that inter-crystalline rupture must occur , after the lapse of sufficient time , as the result of the- prolonged action even of small stresses which may arise either from externally applied forces or from elastic deformations and consequent internal stresses .
It would seem , at first sight , that such an explanation covers too much , since it would suggest the probability that inter-crystalline failure would ultimately occur in all metals as the result of prolonged exposure to stresses far below their ordinary breaking stresses .
Experience clearly indicates that this is not the case , since metal structures of various kinds are known to have withstood considerable and continued stresses for very long periods .
Cases of " season cracking " are in fact rather exceptional and isolated .
This consideration shows that the explanation , as stated above , is incomplete , and the authors believe that the completion is to.be found in a consideration of the conditions existing at a crystal boundary where a tendency exists for gradual flow of the " cement " under prolonged application of stress .
It is at once obvious that flow sufficient to cause rupture cannot occur at a crystal boundary unless the shape of that boundary and of other adjacent boundaries is favourable to movement .
If , for instance , the crystals were thoroughly dovetailed into one -another , rupture along the boundaries could not occur at all , even if there were no adhesion whatever between adjacent crystals .
Although such a dovetailed , truly " interlocked , " structure is not * Rosenhain and Humfrey , loc. cit. , see note ( * ) , p. 58 .
Inter-crystalline Fracture of Metals under Prolonged Stress .
61 met with in metals , there are wide differences in the character of the crystal boundaries in various cases , and most notably between the crystals of the same metal in different conditions of thermal and mechanical treatment .
One of the authors has recently pointed out* that , in a metal which is undergoing more or less rapid recrystallisation or crystal growth , the crystal boundaries are notably irregular\#151 ; sometimes even jagged or serrated .
When crystal growth has come to an end under the conditions to which the metal is exposed , on the other hand , it is found that the crystal boundaries become smooth and tend towards rectilinear , polygonal forms .
That such a process of filling up and smoothing-out of irregularities should occur is in conformity with what would be anticipated from the point of view of the energy-content of the boundaries .
Whether such energy-content be ascribed to the presence of an amorphous film , or merely regarded as " surface energy , " diminution of the area of these inter-crystal surfaces will diminish the amount of energy stored in boundaries .
Nor need it be supposed that it is only the coarser irregularities of the boundaries which come into play in the present connection .
Since crystal-growth in solid metals is known in many cases to take place by a process in which the growing crystal thrusts out arms or projections into its neighbours , it follows that the surface of a growing crystal is likely to be covered with minute irregularities\#151 ; small projections thrusting forward towards or into the neighbouring crystals , but always , according to the present hypothesis , with a film of amorphous material interposed .
Such projections can only maintain themselves so long as the crystal is growing at a rate which is relatively rapid in comparison with the tendency for irregularities of crystal outline to become smoothed out , so that\#151 ; as the crystals cease to grow\#151 ; we may expect their outlines to become both microscopically and\#151 ; probably\#151 ; ultra-microscopically smoother .
The part which even minute surface irregularities in the crystal boundaries may play in hindering , or even in entirely preventing movement by viscous flow of the inter-crystalline film , may , perhaps , be better understood by the aid of a rough analogy .
If we have , as in fig. 5a , two smooth metal plates lying one upon the other , with a film of thick oil between them , a very slight pull will cause a displacement of one over the other , since here the viscosity of the oil is unaided , so far as resistance to motion is concerned , by any other factor .
This is the case corresponding , in principle , to the very smooth crystal boundary with its interposed amorphous film .
On the other hand , if the metal plates have on their surfaces even minute projections , such as are indicated in the second diagram ( fig. 5b ) , then the resistance to relative * Rosenhain , Discussion on paper by Z. Jeffries on " Grain Growth , " ' Journ. Inst. Metals , ' No. 2 , vol. 20 ( 1918 ) .
Q2 Dr. W. Rosenhain and Mr. S. L. Archbutt .
motion of the plates over one another must necessarily be very much greater .
If the projections are large enough to come into actual contact by mutual Fig. 5 .
sliding of the two surfaces , then any further movement by viscous flow alone will become quite impossible , and can only occur by the actual shearing of these projections .
But , as these would be numerous , this would imply the application of a stress large enough to bring about serious deformation of the crystals , and would at once lead to an entirely different type of fracture from that under discussion .
The inference that\#151 ; if the authors ' view of the nature of inter-crystalline fracture under prolonged loading is correct\#151 ; such fracture should only occur where the crystal boundaries have assumed smooth and regular outlines , is strongly supported by the authors ' observations .
In the first place , the phenomena of inter-crystalline fracture in the aluminium alloy only occur when the metal has previously undergone relatively drastic annealing .
Thus it is found that , when this alloy is softened by annealing at 250 ' C , or lower , the material shows no tendency to " season cracking .
" After annealing at higher temperatures the phenomenon makes itself apparent , the time required to bring about inter-crystalline fracture under bending-stress decreasing from four days after annealing at 300 ' C. down to an hour or less after annealing at 450 ' C.\#151 ; an exposure of one hour to the temperatures named being given in each case .
Microscopic examination of the material after each of these treatments serves to confirm the inference drawn above .
After annealing at 250 ' C. the crystals show no considerable amount of growth , and retain the irregular elongated outlines given to them by the previous rolling operation .
On the other hand , after annealing at 450 ' C. , when the " season cracking " phenomenon is most fully developed , the crystals have attained considerable size , no longer show signs of their former elongated shapes , and exhibit extremely smooth and regular boundaries .
Inter-crystalline Fracture of Metals under Prolonged Stress .
63 The typical character of the crystal boundaries in this alloy in the two \#166 ; conditions described is illustrated in figs. 6 and 7 under a magnification of 150 diameters ( Plate 3 ) .
In this connection the authors have made a further observation of some interest .
They find that , by adding to this aluminium alloy a small quantity of manganese ( not more than 0*5 per cent , is required ) , the entire phenomenon of " season cracking , " even after annealing for six hours at 450 C. , disappears .
Microscopic examination at once serves to correlate this fact with the inference discussed above , since it is found that the alloy containing manganese does not undergo recrystallisation at anything like the late met with in the alloy free from manganese .
The size and shape of the crystals , and the irregularity of boundary which is seen in the " pure " alloy when annealed at low temperatures ( 250 ' C. and below ) , is still found in the alloy containing manganese even after annealing at 450 C. Ihis observation , therefore , serves to establish the connection between irregularity and incomplete growth of crystal boundaries and resistance to inter-crystalline fracture , which the authors ' hypothesis suggests .
Evidence of the same kind is obtained by observations made on the examples of inter-crystalline fracture found in the case of lead and mild steel .
In the case of the lead , already referred to above , examination shows that cracking has only occurred in patches , the remainder of the lead being perfectly sound .
When the lead is etched with dilute nitric acid , it becomes perfectly evident that , in the regions where inter-crystalline fracture has taken place , the crystals have assumed a smooth regular formation , such as is characteristic of crystals which have approached an equilibrium condition\#151 ; i.e. , in which rapid growth has ceased .
Elsewhere in this same specimen , the crystals show the ragged , almost feathery , edges typical of lead crystals in process of active growth.* The difference in type is clearly illustrated by comparing the photograph ( fig. 2 ) with that shown in fig. 8 , which shows , under a magnification of 20 diameters , a region found in the same specimen having very irregular crystal boundaries .
The authors suggest that this difference of character , coupled with the fact that , where the crystals have irregular boundaries , there are no signs of failure , provides a considerable degree of support for their views .
In mild steel , examples of inter-crystalline failure are very rare , but they have been met with in several cases in steel boiler-plates which have been found badly cracked near their edges about the rivet holes .
On examination , these cracks were found to be typically inter-crystalline .
An example has already been illustrated ( fig. 4 ) .
Careful study of several cases has shown * Ewing and Rosenhain , ' Phil. Trans. , ' A , vol. 195 , pp. 279-301 ( 1900 ) .
Dr. W. Rosenhain and Mr. S. L. Archbutt .
that failure of this type is associated with a particular kind of heat-treatment of the steel , viz. , prolonged annealing or very slow cooling through the Fig. 8 .
temperature range just below the lowest critical point of the steel ( Ari ) .
This treatment results in the aggregation of the eementite or iron-carbide of the steel into the form of solid globules or membranes , instead of leaving it distributed in the relatively diffuse form of pearlite more usually found in steel .
This distribution of the carbide is , in itself , undesirable , and causes weakness , more particularly under impact .
In the form of pearlite also , with , its minutely irregular outline , the carbide would serve as a much better " key " to hinder relative motion of the crystals than it could do in the form of smooth plates or membranes .
Beyond this , however , it is important to note that the heat-treatment which produces this peculiar distribution of the carbide also allows the iron or " ferrite " crystals to approach an equilibrium condition in which further growth is very slow , and which is characterised by very smooth and regular crystal outlines .
We thus see the characteristic feature required by the authors ' explanation reproduced in a striking manner in an aluminium alloy , lead , and mild steel , in samples all of which exhibit the phenomenon of inter-crystalline fracture following upon prolonged exposure to stress .
Another fact of interest in this connection is the effect which even a small rise of temperature appears to exert on the phenomenon of " season cracking , " ' Inter-crystalline Fracture of Metals under Prolonged Stress .
65 an effect which , in the authors ' view , would be anticipated as a result of the rapid diminution of the viscosity of an under-cooled liquid with rise of temperature .
In the case of brass , it is well known that failure from this cause is more likely to occur iu a hot climate than in a cold one ; it has been found that the cracking of cartridge cases , for example , is more liable to occur when goods are stored in India than with those kept in England , while the brass-work of gasoliers , exposed to the heated air in the upper part of a gas-lighted room , is particularly liable to this type of failure .
In the case of lead , the ordinary temperature is so near its melting point that the metal undergoes spontaneous annealing ( i.e. , recrystallisation ) at the ordinary temperature .
For such a metal , the ordinary temperature may well be regarded as " high .
" In the case of mild steel , although the temperature is still very far away from its melting point or even its softening point , it is at least suggestive that the cases of inter-crystalline fracture observed by the authors have in every instance occurred in the plates of steam boilers working under a high pressure and , therefore , at a correspondingly high temperature , rising in some cases up to 200 ' C. In the case of the aluminium alloy , the effect of a rise of temperature has not yet been investigated .
The alloy , however , is one which rapidly becomes soft and weak with rising temperature\#151 ; a softening which is appreciably lessened by the addition of manganese , whose powerful effect on the " season cracking " phenomenon has already been described .
The effects of temperature , as a whole , however , so far as they are known , appear to be entirely consistent with the authors ' view , that this type of inter-crystalline fracture results from a gradual yielding or " flow ' ' of the viscous inter-crystalline amorphous cement , since the viscosity of such a cement would rapidly diminish with rising temperature .
In the case of glass , it has recently been shown* that at a temperature where slight softening becomes noticable , every rise of 8 ' C. reduces the viscosity by one half .
In the case of the aluminium alloy , the authors have made some preliminary measurements of the time required to produce rupture under various intensities of loading .
The results , examples of which are given in the subjoined Table , indicate clearly that the time required to produce rupture 1 ensile stress per square inch .
Time taken to produce rupture .
tons , secs .
* Twyman , 'Journal Soc. Glass Technology , ' vol. 1 , No. 1 , May , 1917 .
VOL. XCVI.\#151 ; A. F Dr. W. Kosenhain and Mr. S. L. Archbutt .
increases very rapidly with decreasing intensity of stress , in the manner which would be anticipated in a case of viscous flow .
More numerous observations are , however , required to establish any definite relationship .
In these tests there is little or no elongation of the specimen .
If a similar test-piece of the same material is tested in the ordinary way , it shows a breaking stress of 24 tons per square inch and an elongation of 8 per cent , on 2 inches .
Having put forward the evidence so far available in support of their explanation of the mechanism of inter-crystalline fracture , the authors wish to consider briefly some of the main objections to their views which have presented themselves .
The first of these difficulties relates to the fundamental question whether an amorphous inter-crystalline cement can or does exist .
The evidence for and against this has been fully discussed elsewhere ; * here , it need only be said that until other methods of investigation become available , any direct proof of the existence of the cement can hardly be hoped for .
On the other hand , the increasing number and range of phenomena which the hypothesis of an amorphous inter-crystalline cement serves to explain and correlate furnishes important indirect evidence in favour of the validity of the hypothesis .
From that point of view , the case appears to be considerably further strengthened by the observations described in the present paper .
Perhaps the most serious difficulty which confronts the authors ' views is that of accepting the somewhat novel idea that a sufficient degree of mobility can exist in the under-cooled liquid forming the inter-crystalline cement to allow of viscous flow resulting in fracture at temperatures not very much above the ordinary .
On this point it may be said that we have no data as to the viscosity of under-cooled metal , and can only judge by analogy with other substances , such as glass .
But while glass is very readily under-cooled , this is not the case with metal , which has not been experimentally undercooled in any considerable quantity sufficiently far below the freezing point to save it from spontaneous crystallisation .
This difference may well be due to a difference in regard to viscosity\#151 ; that while glass when under-cooled at once increases so rapidly in viscosity that crystallisation soon becomes impossible , in metals the rapid rate of increase of viscosity may not begin until a much lower range of temperatures is reached , with the consequence that a degree of under-cooling which could easily be achieved experimentally is not sufficient to retain the material in a " vitreous " condition .
If this is the case then at temperatures not far above that of the air , the inter-crystal cement might be sufficiently mobile to yield very slowly to the prolonged effects of stress .
* Rosenhain , see papers cited in note , p. 58 .
Inter-crystalline Fracture of Metals under Prolonged Stress .
67 It may further be pointed out that in the case of 7-iron , Rosenhain and Humfrey ( loc. cit. ) have shown definite evidence of viscous flow taking place in the crystal boundaries .
Yet 7-iron at a temperature just above 900 ' C. is still 600 ' C. below its melting point .
As compared with this , brass , lead and the authors ' aluminium alloy at the ordinary temperature are approximately some 900 ' C. , 300 ' C. , and 600 ' C. below their respective melting points .
Their relative conditions , therefore , need not be regarded as essentially different ; if viscous flow in the amorphous cement can be accepted in the one case , there need be no great difficulty in extending the conception to the others .
The mild steel referred to above in connection with inter-crystalline failure in boiler plates is , however , in a somewhat different position , since at the ordinary temperature this is some 1500 ' C. below the melting point of iron , so that even a rise of some 200 ' C. in a high-pressure boiler does not bring it into the same range as aluminium or brass .
The explanation may , however , lie in the allotropic transformations which iron undergoes .
The melting point near 1500 ' C. is that of 7-iron , or possibly of some other modification stable at those very high temperatures .
The iron which undergoes intercrystalline fracture , however , is a-iron .
If this variety could be prevented from passing into the other modifications on heating , it seems probable that its melting point might lie not very far above the transformation temperature .
At all events , the variations of the physical properties of a-iron with rising temperature suggest that they are related rather to its transformation temperature , in the neighbourhood of 900 ' C. , than to the melting point of 7-iron near 1500 ' 0 .
If this view is tenable , the difficulty in the present connection disappears , since the transformation temperature of iron is rather lower than the melting point of brass .
If the explanation of " season cracking " and similar phenomena of intercrystalline fracture which has here been put forward proves , on further investigation , to be well founded , consequences of considerable practical importance may be drawn from it .
It would appear that the risk of intercrystalline fracture following upon the prolonged action of stresses far below the normal " breaking strength " of the material is of very considerable importance , and that our immunity from numerous and serious failures from this cause is due to the fact that in the majority of metals as ordinarily employed the condition of the crystal boundaries is not favourable to viscous flow .
In the case of certain alloys , such as steels containing moderate amounts of carbon , this safety is no doubt due to the presence of a second constituent possessing characteristically irregular outlines .
In other cases , the sale character of the boundaries can only be due to the fact that as a 68 Inter-crystalline Fracture of Metals under Prolonged Stress .
result of manufacturing conditions the crystals are left in a condition of incomplete growth in which they have not been able to develop smooth and regular outlines .
Failures from this cause have , therefore , only arisen in a relatively few special cases where metal has been subjected to somewhat drastic annealing which has allowed the crystals to attain a condition of metastable equilibrium in which their outlines have had an opportunity of becoming smooth and regular .
As has been found by practical trials in the case of brass , such " season cracking " can be prevented by a suitable modification of the annealing operations .
The most serious cases , however , are those arising in steel .
If these have been rightly correlated with the others , the remedy is clear\#151 ; the heat-treatment of the steel must be altered in such a manner as to avoid all risk of the formation of the smooth and regular boundaries which\#151 ; quite apart from any theory\#151 ; are definitely associated with this dangerous type of failure .
Fortunately the well-known operation of " normalising " mild steel by heating it for a very short time to a temperature just above the upper critical point , followed by rapid cooling in air , is sufficient entirely to eliminate all risk of the development of these apparently dangerous structural arrangements .
The authors gratefully acknowledge much valuable help in carrying out some of the experimental work described in this paper which they have received from several of their colleagues , particularly from Mr. F. S. Tritton .
They are very specially indebted to Mr. L. Archbutt of Derby for first calling their attention to the occurrence of inter-crystalline failure in lead .
They also desire to acknowledge the interest which has been taken in this research by Sir Richard Glazebrook , C.B. , F.R.S. , Director of the National Physical Laboratory , where the work has been carried out .
Rosenhain and Arclibutt .
Roy .
Soc. Proc. , A t vol. 96 , PI .
1 .
Fig. 1 .
Inter-crystalline crack in " season-cracked " brass , x 100 .
Fig. 2 .
Rosenhain and Archbutt .
Roy .
Soc. Proc , vo/ .
96 , PI .
2 .
Fig. 3 .
Fig. 4 .
Rosenhain and Archbutt .
Roy .
Soc. Proc. , A , vol. 96 , 3 .
* V* \ V .
i V iAmM \gt ; V " - * fir"* V A wm *\#165 ; ! .
; !4VT : \#166 ; * y \lt ; .
r t I -v l ; sk .
, .
'i \#166 ; ^ -M* 'A \#166 ; \#166 ; " r\* |\#171 ; * .
: \#166 ; .'i'j ( i ; 'v# ' 'k j ; 'v \ i vW A- #\gt ; ' % ' " *k A** .
\#166 ; ?v v : .
* , i \#166 ; '* % \#166 ; 4-\gt ; x *A \lt ; \ x\gt ; \#166 ; A 1 *v^ . .
\'i *^v -v ii.\gt ; .
; . . .
, i ' ( \\\ ' 's ' v \#166 ; \#166 ; \#166 ; Ak'v .
* Jl \#166 ; ' * NN* ' % , , v- ; .
i- ' .
\#171 ; \#166 ; \ .
* ' 'f '\lt ; tt ' V '3\#163 ; lfir$ . .
\#171 ; p - t f Fig. 6 .
Fig. 7 .

