Effect of non-metallic inclusions on cleavage fracture

The effect of non-metallic inclusions on cleavage in wrought steel has been studied using a resulphurized mild steel. The cleavage fracture stress was measured at –196°C and found to exhibit an orientation dependence which is attributed to the morphology of the inclusions. In the short transverse and transverse testing orientations, cleavage facets were seen to initiate at the edges of the elongated inclusions. It is proposed that the inclusions do not themselves act as cleavage crack nuclei, but are internal sites of stress concentration that facilitate carbide-initiated cleavage in the adjacent matrix. In the longitudinal testing orientation, inclusions are not associated with sites of cleavage nucleation. It is proposed that in this orientation decohesion of the inclusion/matrix interface causes a reduction in stress triaxiality which inhibits cleavage nucleation. Measurements of fracture toughness K1c at –196°C show no effect of testing orientation. This is attributed to the low probability of encountering inclusions in the crack tip process zone under the conditions of testing. MST/365

-196°C and found to exhibit an orientation dependence which is attributed to the morphology of the inclusions.In the short transverse and transverse testing orientations, cleavage facets were seen to initiate at the edges of the elongated inclusions.It is proposed that the inclusions do not themselves act as cleavage crack nuclei, but are internal sites of stress concentration that facilitate carbideinitiated cleavage in the adjacent matrix.In the longitudinal testing orientation, inclusions are not associated with sites of cleavage nucleation.It is proposed that in this orientation decohesion of the inclusion/matrix interface causes a reduction in stress triaxiality which inhibits cleavage nucleation.Measurements of fracture toughness K1c at -196°C show no effect of testing orientation.This is attributed to the low probability of encountering inclusi01Js in the crack tip process zone under the conditions of testing.MSTj365 ©

Experimental
The steel selected for the investigation was a resulphurized mild steel, En 7A, the chemical composition of which is given in Table 1.The steel was cast as a full size ingot and then hot rolled and forged to a billet having a cross-section of 65 x 120 mm.The dimensions of the billet enabled mechanical test specimens of appropriate size to be extracted in the three principal testing directions.
The appearance of the inclusions on transverse and longitudinal sections is shown in Figs.1a and b, respectively.On the transverse plane, the average aspect ratio of the inclusions was 2•5; on the longitudinal plane the aspect ratio was 12•1.The majority of the inclusions Venkatasubramanian and Baker. 6For the conditions studied, the fracture path was intergranular rather than transgranular; nevertheless, it was a completely brittle mode of fracture with no contribution from internal gas pressurization.Large effects of inclusion volume fraction and orientation on the crack growth rate and threshold stress intensity were observed.Maximum fracture resistance was obtained in the longitudinal testing orientations and minimum resistance in the short transverse orientations.More recently, a limited series of measurements on a bainitic steel by Bowen and Knott 7 has suggested that the microscopic cleavage fracture stress G'f may be lower in the short transverse orientation than in the transverse and longitudinal orientations .. The series of experiments described in this paper wa-s undertaken to determine whether the cleavage fracture behaviour of a wrought C-Mn steel is orientation dependent and, if so, whether this is related to the presence and geometry of the non-metallic inclusions.To facilitate identification of the role of the inclusions, .asteel containing a deliberately high sulphur content was employed.To avoid the difficulties associated with defining the fracture transition temperature, the cleavage resistance was characterized in terms of the cleavage fracture stress G'f and the plane strain fracture toughness KIc'

Introduction
The role of non-metallic inclusions in the microvoid coalescence mechanism of fracture is well known and it is recognized that elongated non-metallic inclusions are the major cause of the orientation dependence of fibrous fracture toughness in wrought steels.! Whether inclusions can have a significant influence on cleavage fracture behaviour is.unclear.For steels having ferrite-pearlite microstructures, it is generally acknowledged that the microstructural feature controlling the susceptibility to cleavage fracture is the thickness of iron carbide particles. 2owever, recent studies by Tweed and Knott 3 and McRobie and Knott 4 have suggested that, in C-Mn weld metals, cleavage fracture can nucleate at brittle nonmetallic inclusions.The fracto graphic evidence presented indicates clearly the presence of small spherical inclusions at the origin of radiating riverlines on cleavage facets.The authors 3 ,4 suggest that in the weld metal, the micromechanism of cleavage is essentially similar to that in wrought steel, but with fine inclusions rather than iron carbides acting as nucleating sites.
If inclusions can act as cleavage nuclei in weld metal, the question that arises is whether the inclusions in wrought steels can also act as nucleating sites for cleavage.It is known that sulphide and oxide inclusions can crack when steel undergoes plastic deformation. 5Since the inclusions are usually thicker than the carbide particles, it might be expected that they would act as more effective cleavage nuclei.Also, in view of the elongated form of most nonmetallic inclusions in wrought steel, it might be expected that inclusion-nucleated cleavage fracture would show a pronounced orientation dependence.
The effect of sulphur content on the ductile-brittle transition temperature in a wrought C-Mn steel has been studied by Pickering. 5It was found that in both the longitudinal and transverse directions, an increase in the MnS volume fraction first increased the transition temperature and then caused it to decrease at higher volume fractions.It was also reported that for sulphur contents greater than about 0•1%, the transition temperature in the transverse orientation was higher than that in the longitudinal direction.The latter observation is quite common and finds expression in many steel specifications where the transverse orientation is specified for the determination of Charpy impact properties.
The effect of inclusions on hydrogen-induced brittle fracture in a high-strength steel has been studied by a transverse section; b longitudinal section Inclusions in steel were type I MnS.In addition, there were significant numbers of oxide inclusions.These were two-phase mixed oxides which were usually larger and more highly elongated than the sulphide inclusions.A typical example of an oxide inclusion is shown in Fig. 2. Such inclusions are not an unusual feature of general engineering steels.
The microstructure of the steel in the as-received condition is shown in Fig. 2. The ferrite grains were equiaxed, but banding of the pearlite is evident.This type of microstructure is typical of hot-worked structural steels, and for this reason one series of tests was undertaken with the steel in the as-received condition.It was recognized that the banding of the pearlite could itself give rise to anisotropy in cleavage behaviour.To explore this 2 Banded pearlite in as-received steel (longitudinal section) 3 Heat treated microstructure possibility, the steel was austenitized and rapidly cooled to produce the Widmanstatten ferrite.-bainitemicrostruct~re illustrated in Fig. 3.The underlYIng soluble segregatIon was of course not removed by the heat treatment; however, the alignment of the pearlite, which was of main concern with regard to cleavage, was eliminated.A second series of tests was undertaken with the steel in the heat treated condition.
Mechanical test specimens• were obtained from the three principal orientations.For the notched and precracked testpieces, the orientations selected were L-T, T-L, and S-T' these are illustrated in Fig. 4. A limited number of Cha~py impact tests were performed over a range of tes~ing temperatures to illustrate the general form of the ducttlebrittle transition and to demonstrate the degree of anisotropy in the upper shelf energy.The main part of the testing programme was carried out in liquid nitrogen at -196°C and involved measurements of cleavage fracture stress uniaxial yield stress, and plane strain fracture toughness.The cleavage fracture stress was determined from notched four-point bend specimens using the specimen geometry and finite element analysis of Griffi!hs and Owen. 8The uniaxial yield stress was measured uSIng Hounsfield no. 13 tensile specimens which were extracted from the broken halves of the four-point bend specimens.F or the plane strain fracture toughness determinations, 8 mm thick fatigue precracked single edge notched (SEN) specimens were employed.All tests gave linear loaddisplacement plots and the testing procedures and specimen dimensions satisfied the requirements of BS 5447 (1977) for valid K 1c determinations.

FRACTOG RAPHY
The fractographic examination concentrated on the notched four-point bend specimens, particular attention being paid to the area up to O' 5 mm ahead of the notch root where the maximum tensile stress would have developed. 8Examination of the cleavage facets in the SEM revealed a region ahead of the notch root from which river markings radiated both back towards the notch surface and forwards through the remainder of the section.A dominant nucleating site could not usually be identified within this region.However, many potential nucleating sites were examined, and particular attention was paid to the participation of non-metallic inclusions.
In the longitudinal testing orientation, no cleavage nucleation sites that could be attributed to the presence of non-metallic inclusions were identified.Part of a cleavage facet which contains several MnS inclusions is shown in Fig. 7. From the river markings, it is clear that the cleavage facet shown had spread from the left hand side of the photograph and happened to have intercepted the inclusions.It is of note that the inclusions are separated from the matrix by an obvious gap.It is possible that there could have been some separation between the inclusions and the matrix before loading commenced.However, void expansion would be expected under the influence of the triaxial state of stress accompanying cleavage.The orientation.The highest values are found in the longitudinal orientation and the lowest in the short transverse orientation (Fig. 6).The difference between the L-T and T-L orientations is greater than that between the T-L and S-T orientations.As with the anisotropy in upper shelf values, this is consistent with the inclusion morphology.The results from the heat treated condition show the same type of behaviour as the as-received condition.This demonstrates that the anisotropy in cleavage susceptibility is not associated with banding of the pearlite.Only one cleavage stress value is available for the short transverse orientation of the heat treated condition, and it is of note that this is lower than the corresponding value for the asreceived condition.Examination of the fracture surface in the scanning electron microscope (SEM) showed that the low value was associated with the presence of a large oxide inclusion which penetrated the notch root.
In view of the similar behaviour shown by the asreceived and heat treated structures in the cleavage stress determinations, fracture toughness measurements were carried out only on the as-received steel.These results are also given in Table 2.The surprising feature, in view of the cleavage stress results, is that there is no significant effect of orientation.

MECHANICAL TESTS
The Charpy transition curves for the three orientations of the steel in the as-received condition are shown in Fig. 5.
As expected from the high volume fraction of inclusions, there is a pronounced effect of orientation on the upper shelf energy.Compared to the L-T orientation, there is a relatively small difference between the T-L and S-T orientations.This is consistent with the dimensions of the forged billet and the small aspect ratio of the inclusions on the transverse plane.
It is difficult to draw any conclusions concerning the orientation dependence of cleavage susceptibility from the Charpy curves.If the transition is defined in terms of a particular energy level, say, 20 J, then the transition temperature is obviously much higher in the transverse and short transverse orientations than in the longitudinal orientation.On the other hand, if the transition is defined as the temperature corresponding to 50% of the upper shelf energy, the differences between the three orientations are small.Neither of these transition temperatures is meaningful as far as the pure cleavage mode of fracture is concerned, because within the transition region the fracture necessarily involves a mixture of cleavage and fibrous mechanisms, the latter of which is known to be very sensitive to orientation.
The results of the fracture tests carried out at -196°C are given in Table 2.The yield strength shows little orientation dependence in either the as-received or heat treated condition.By comparison, the measurements of cleavage fracture stress show a pronounced effect of cleavage fracture surface of a longitudinal specimen are shown in Fig. 8.In Fig. 8a, it can be seen that the interface between the inclusion and the matrix has separated in a direction at right angles to the fracture surface and the projecting part of the inclusion has pulled out from the mating face.In Fig. 8b, a sulphide inclusion appears to have caused a branching cleavage crack to arrest.On the fracture surfaces from the T-L testing orientation, several cleavage initiation sites were associated with elongated inclusions.Typical examples are illustrated in Fig. 9.Both of the inclusions shown are fragmented oxides and it was found that these were more commonly associated with cleavage nucleation than were the sulphide inclusions.Another interesting feature of the nucleation sites is that they occurred most commonly along the sides of the inclusions.A nickel-plated section through the fracture surface of the T-L specimen is shown (1) 10 Nickel plated section through fracture surface of T-L orientation showing cleavage crack (arrowed) associated with sulphide inclusion in Fig. 10.Beneath the main fracture surface, an arrested cleavage crack can be seen which appears to have nucleated at a sulphide inclusion.The void expansion is seen to be similar to the opening of the crack.This implies that the development of the void around the inclusion must have followed the formation of the cleavage crack.
In the short transverse testing orientation, inclusions were frequently observed to act as nucleating sites for cleavage.Typical examples are shown in Fig. 11.Fragmented oxide stringers (as in Fig. 11a) were again the preferred sites for cleavage nucleation, but sulphides were also observed to act in a similar manner.It is of note that the cleavage facet in Fig. 11b has nucleated a short distance away from.the edge of the sulphide inclusion.

Discussion
The present series of experiments has demonstrated that in a steel containing a high volume fraction of inclusions there is a significant orientation dependence of the cleavage fracture stress.The similarity in the behaviour of the asreceived and heat treated conditions indicates that the anisotropy is attributable to the presence of the inclusions rather than to any other microstructural feature.This is supported by the fracto graphic evidence.
The fracto graphic and metallographic observations show that the inclusions can crack in a brittle manner.It could be argued, therefore, that inclusions in wrought steel could act in a similar manner to grain boundary carbides, which are generally regarded as the dominant sites of cleavage nucleation in mild steels.This is the suggestion that has been put forward by Tweed and Knott 3 to account for the participation of fine inclusions in the cleavage fracture of weld metals.
An essential requirement for cleavage fracture in steel is that there must be prior plastic deformation to develop the very high local stresses which are necessary for the formation of the crack nucleus. 9lastic deformation usually occurs by a slip mechanism.
However, at very low testing temperatures and at high rates of loading, deformation can occur by twinning.Both deformation mechanisms are effective in nucleating cleavage cracks.In the present series of experiments, the mechanical tests were carried out at -196°C.
Twins could be seen in micro sections adjacent to the fracture surface (Fig. 8b), and could be detected on the fracture surface (Fig. 11a).
Twinning is commonly seen to be associated with cleavage fractures, even with those that develop at normal ambient temperatures, because of the very high strain rates which accompany cleavage crack propagation.
In the present case, however, it is possible that twinning may also have been responsible for crack nucleation.
For slip-initiated cleavage, it has been suggested 2 that the cleavage fracture stress can be related to the size of the crack nucleus by the following Griffith equations: for a penny-shaped crack of radius a, and for a through-thickness crack in a plate-like particle of thickness 2a, where E is Young's modulus, yp is the effective surface energy of ferrite, and v is Poisson's ratio.
If cleavage is twin initiated, similar equations may be derived for the cleavage fracture stress, but it is necessary to include a term which contains the shear yield stress Ly and the grain size d (Ref.10).
For the plate-like particle Involvement of the shear yield stress causes the twininitiated cleavage fracture stress to show a temperature dependence which is not usually observed for the slipinitiated mode.In the present series of experiments, all tests were carried out at the same temperature and there was no significant anisotropy in yield strength.Consequently, as far as the anisotropy of the cleavage fracture stress is concerned, it does not matter whether nucleation is slip or twin induced.
If an inclusion were to act as a cleavage nucleus, then the relevant dimension in the short transverse -testing orientation would be the intermediate axis; the average value of this was 7•3 Jlm.Inserting this in equation (1)   together with values of E = 210 MN m -2, Y p = 14 J m -2, and v = O• 3 gives a calculated value of (Jf "'"' S30 MN m -2.For the longitudinal testing orientation, the relevant inclusion dimension is the minor axis (2'9 Jlm) and the corresponding value of (Jf is 11S0 MN m -2. The calculated values of (Jf are lower than the measured values and indicate a greater influence of testing orientation than that observed.Nevertheless, it could be argued that the agreement between theory and experiment is comparable to that observed in other cleavage studies and as such is not inconsistent with a direct inclusion nucleation mechanism.Although this model is superficially attractive, there are a number of reasons suggesting that it may not be relevant to the steel under investigation.First, if a particle is to act as an effective cleavage nucleus, it must be more brittle than the ferrite matrix.While this is the case for cementite and some oxide inclusions, it is not necessarily so for sulphide inclusions.Some of the fractographic observations suggest that the inclusions crack as a consequence of the matrix separation (Figs. 7 and Sa) and in some cases cleavage cracks appear to have been arrested by inclusions (Fig. Sb).A second requirement for an effective cleavage nucleus is that the particle should be well bonded to the matrix in order that a continuous crack can propagate from the particle into the adjacent ferrite.For manganese sulphide inclusions, the interface with the matrix is extremely weak.The fractographic observations suggest that there is complete decohesion between the sulphide and the matrix before the onset of cleavage (Figs. 7 and Sa).Some oxides may be more strongly bonded, but this is unlikely to be the case for the duplex oxide stringers in the present steel.These oxides had undergone internal fragmentation during hot working and exhibited an irregular interface with the matrix which separated readily during testing.It is concluded that it is unlikely that the inclusions could have acted directly as cleavage nuclei.
An alternative way in which inclusions may influence cleavage behaviour is by modifying the stresses ahead of the notch root.In the absence of inclusions, cleavage initiates at some distance ahead of the notch root when the critical stress for the unstable propagation of a carbidenucleated crack is attained.This cleavage fracture stress is usually significantly larger than the uniaxial yield stress and is attained by a combination of work hardening and triaxial constraint.If an inclusion is located within the relevant region ahead of the notch root, it may modify the development of plasticity on a local scale and hence facilitate, or even discourage, the normal carbide-nucleated cleavage mechanism.
Considering first the short transverse orientation, it is expected that the inclusion/matrix interface would have separated before the onset of cleavage.Under these circumstances the inclusion would act as a site of stress concentration.If the inclusion is treated as an oblate ellipsoid, the sites of maximum stress concentration would be at the ends of the transverse semi-axis.For the inclusions in the present steel, the radius of curvature at this position was about 0•6 Jlm.For elastic conditions of loading, the corresponding stress concentration factor would be about 6.In practice, this degree of stress concentration cannot be achieved because yielding precedes cleavage nucleation.There would, however, be a concentration of plastic strain along the edges of the inclusion and this would be reflected in an increase in the value of the maximum principal tensile stress due to the local development of increased constraint and increased work hardening.Under these conditions, cleavage fracture is expected to nucleate at a lower nominal stress than would be the case in the absence of the inclusions.This is reflected in the observed reduced value of the cleavage fracture stress.It would also be expected that cleavage would nucleate preferentially along the edges of the elongated plate-like inclusions.The latter is borne out by the fracto graphic observations.A similar influence of elongated inclusions on brittle fracture nucleation is seen in the hydrogen-assisted cracking of wrought steels in H 2 S-containing environments. 11ecause of the dimensions of the forged billet, the width of the inclusions in the transverse direction is not very different from that in the short transverse direction (Fig. 1).For the T-L testing orientation, the position of maximum stress concentration is again along the edges of the elongated inclusions, but the relevant radius of curvature is about 9 Jlm.The corresponding reduction in the degree of stress concentration at the inclusion makes cleavage nucleation less easy than in the short transverse testing orientation and this is reflected in the observed increase in the measured cleavage fracture stress.
In the longitudinal testing orientation, the inclusions can give rise to no significant stress concentration provided that they do not undergo brittle fracture before the onset of cleavage in the matrix.It does not follow that in this orientation the inclusions have no influence on the cleavage process.The fractography has demonstrated that the inclusion/matrix interface is extremely weak and probably decoheres before cleavage separation of the matrix.As far as the propagating crack is concerned, the metallographic evidence indicates that the separated interfaces can arrest and/or deflect a running cleavage crack (Fig. Sb).As far as the nucleation event is concerned, the separation of the inclusion interfaces which are running at right angles to the fracture plane can lead to a local relaxation of triaxial constraint.The local stress elevation is thereby reduced and cleavage nucleation is hindered.In the notched bend test this is manifested as an increase in the cleavage fracture stress.A similar effect of weak inclusion interfaces leading to enhanced resistance to hydrogen-induced brittle fracture has been reported previously by Venkatasubramanian and Baker. 6 feature of the fractographic studies was that oxide inclusions appeared to act as more effective nucleating sites for cleavage than sulphides.A similar observation has frequently been made in relation to inclusion-nucleated fatigue in high-strength steels. 12The generally accepted explanation for the fatigue phenomenon. is that because of the relative thermal expansion coefficients of the inclusion and matrix, oxide inclusions give rise to a residual tensile stress in the surrounding steel, whereas sulphides tend to contract away from the matrix. 13If present, a residual tensile stress would be expected to facilitate cleavage nucleation.In the steel used in the present experiments, the oxides had fragmented during hot working and were no longer fully dense.Consequently, it is questionable whether they could give rise to significant residual stress.There are, however, two other factors which make the oxides preferred sites for cleavage nucleation.First, the oxides are larger and more highly elongated than the sulphides.As a result they act as more effective local sites of stress concentration.Secondly, the microstructure of the matrix in the vicinity of the oxide inclusions is different from that in the vicinity of the sulphides.When a steel transforms from austenite to ferrite and pearlite, manganese sulphide inclusions act as very effective ferrite nucleating sites.Consequently, as shown in Fig. 2, the sulphides tend to be surrounded by carbide-free ferrite.By comparison, the oxide inclusions are seen to traverse the carbide-rich regions.Since it is assumed that in the presence of the inclusions, cleavage nucleation still occurs by a carbide cracking mechanism, it follows that there is a higher probability of encountering a favourable nucleating site in the vicinity of an oxide inclusion.
There remains to be considered the absence of any significant orientation dependence in the measurements of KIc made at -196°C.This is because of the low probability of encountering an inclusion within the crack tip plastic zone.In the blunt notched tests, the maximum tensile stress would have developed at a position approximately 0•25 mm ahead of the notch root.8 Within this region, very many inclusions are present and cleavage can nucleate at the most favourable site.In the sharp-crack KIc test, the fracture processes occur within a much smaller zone ahead of the crack tip.At -196°C, the extent of the plastic zone is about 80 flm, and within this zone, the position of maximum tensile stress is located about 10 flm ahead of the crack tip.Under these circumstances, there is a low probability that a suitable inclusion would be located within the high-stress region of the crack tip plastic zone.It is more likely that cleavage would nucleate at one of the numerous carbide particles present.As a result, the inclusion influence is not revealed.

Conclusions
1.The cleavage fracture stress in a resulphurized wrought mild steel has been shown to be orientation dependent.The orientation dependence of the cleavage fracture stress is attributed to the presence of elongated non-metallic inclusions.
2. In the transverse and short transverse testing orientations, cleavage facets initiate at the edges of elongated inclusions.It is proposed that the inclusions act as internal sites of stress concentration and encourage carbide-initiated cleavage in the adjacent matrix.
3. In the longitudinal testing orientation, decohesion of the inclusion/matrix interface leads to a local reduction in stress triaxiality which inhibits cleavage nucleation.
4. No effect of testing orientation on fracture toughness KIc is observed.This is attributed to the low probability of encountering inclusions in the crack tip process zone.

6
Effect of testing orientation on cleavage fracture stress

a
separation of inclusion/matrix interface; b arrest of branching cleavage crack 8 Nickel plated sections through fracture surface of L-T orientation inclusions have fractured in a brittle manner, but it is of note that they have not fractured in the same plane as that of the matrix cleavage.This is indicative of weak bonding between the inclusion and the matrix at the moment of cleavage propagation.Nickel-plated sections through the 9 Fracture surface (SEMs) from T-L testing orientation showing cleavage nucleation adjacent to oxide inclusions (arrowed)

Materials Science and Technology June 1986 Vol. 2 a
cleavage nucleation at oxide inclusion (arrowed); b cleavage nucleation adjacent to sulphide inclusion (arrowed) 11 Fracture surface from S-T testing orientation (SEMs)

I
VITAL TECHNOCOMMERCIAL BULLETIN REPORTING ON THE RAPIDLY CHANGING WORLD OF ENGINEERED MATERIALS.Each monthly bulletin contains a collection of concise abstracts of recently published items of commercial significance from over 1300 worldwide sources, providing competitive knowledge essential to your business.Coverage includes: World News/Energy Conservation/ NewAlloys/Materials and Processes/International Trade/ Economics/Management Topics/New Plants/Competitive Materials.Current, comprehensive, cost-effective information ...relevant and useable on a daily basis.Imagine what it would cost you or your company to provide a similar service!SUBJECT: The world's nonferrous metals industries, including rare earth, strategic, light and precious metals, from Aluminum to Zirconium CONTENT: 300-500 summaries each month derived from related industry literature worldwide PRICE: UK £140 (£100 to Metals Abstracts subscribers) Send to: Materials Information, The Institute of I Metals, 1 Carlton House Terrace, London SWI Y 5DB, England 1986 The Institute of Metals.Manuscript received 23 September 1985.The authors are in the Department of Metallurgy and Materials Science, Imperial College of Science and Technology, London.
Table 1 Chemical composition of steel En 7A, wt-%

Table 2
Effect of testing orientation on mechanical properties at -196°C