Visual Examination and Light Microscopy

Home , Fractography, Stress concentration

George F. Vander Voort, Carpenter Technology Corporation

THE VISUAL EXAMINATION of fractures by light microscopy. Interesting features can be sections, to determine the origin of the failure, is deeply rooted in the history of metals pro- marked with a scribe, microhardness indents, and to separate the fractures according to the duction and usage, as discussed in the article or a felt-tip pen and then examined by SEM time sequence of failure, that is, which frac- "History of Fractography" in this Volume. and other procedures, such as energy- tures existed before the event versus which ones This important subject, referred to as macro- dispersive x-ray analysis, as required. occurred during the event. This article will fractography, or the examination of fracture The techniques and procedures for the visual assume that such work has already been accom- surfaces with the unaided human eye or at low and light microscopic examination of fracture plished and will concentrate on fracture exam- magnifications (--<50), is the cornerstone of surfaces will be described and illustrated in this ination and interpretation. Other related topics failure analysis. In addition, a number of qual- article. Results will also be compared and specific to failure analyses are discussed in ity control procedures rely on visual fracture contrasted with those produced by electron Volume 11 of the 9th Edition of Metals Hand- examinations. For failure analysis, visual in- metallographic methods, primarily SEM. Vi- book. spection is performed to gain an overall under- sual, light microscopic, and electron micro- standing of the fracture, to determine the frac- scopic methods are complementary; each has ture sequence, to locate the fracture origin or particular advantages and disadvantages. Opti- Macroscopic Examination origins, and to detect any macroscopic features mum results are obtained when the appropriate Locating the fracture origin is a primary goal relevant to fracture initiation or propagation. techniques are systematically applied. of fractography and is vital to successful failure For quality control purposes, the fracture fea- In addition to examination of the gross frac- analyses. The fracture markings formed during tures are correlated to processing variables. In ture face, it is often useful to examine second- the event are like a road map that the analyst this article, examples of visual fracture exami- ary cracks, when present. In many cases, the uses to evaluate the fracture. Fracture initiation nation will be given to illustrate the procedure secondary cracks exhibit less damage than the and propagation produce certain characteristic as it applies to failure analysis and quality gross fracture, which may often be damaged by marks on the fracture face, such as river marks, determination. rubbing, handling, postfracture events (such as radial lines, chevrons, or beach marks, that Although the light (optical) microscope can repair attempts), or fire or corrosion. These fine indicate the direction of crack growth. The be used to examine fracture surfaces, most cracks can be opened by careful sectioning and analyst traces these features backward to find fracture examinations at magnifications above breaking, or they can be examined on polished the origin or origins. The appearance of these 50x (microfractography) are conducted with cross sections. Also, it is often informative to marks on the fracture face is a function of the the scanning electron microscope, as described examine the microstructure ahead of the sec- type of loading, for example, tension, shear, in the article "Scanning Electron Microscopy" ondary cracks, or adjacent to the primary frac- bending, fatigue, or torsion; the nature of the in this Volume. However, if a scanning elec- ture, to detect cracking in constituents or mi- stress system; its magnitude and orientation; the tron microscope is not available, light micros- crovoids, either pre-existing or produced by the presence of stress concentrators; environmental copy can be applied and usually provides sat- deformation associated with crack formation. factors; and material factors. Examples of these isfactory results. In this article, details will also be presented fracture patterns are illustrated in the section Regardless of the equipment available to the concerning the characteristic macro- and micro- "Interpretation of Fractures" in this article and analyst, it is still very useful to examine the scopic features associated with different frac- in the article "Modes of Fracture" in this fracture profile on a section perpendicular to ture mechanisms. These features are detected Volume. the fracture origin. In this way, the origin of and used to characterize the nature of both After a service failure has occurred or after a the fracture can be examined to determine if crack initiation and propagation. Examples will crack has been observed in a component, cer- important microstructural abnormalities are be presented, using a variety of techniques, to tain steps must be taken to ensure that the present that either caused or contributed to illustrate the procedure for classifying frac- fracture features are not obliterated (Ref 1). In fracture initiation. It is also possible to tures. some cases, the fracture face is destroyed in the determine if the fracture path at the initiation incident; alternatively, postfracture events can site is transgranular or intergranular and to occur that drastically alter the fracture face and determine if the fracture path is specific to any Techniques material condition. Such problems can often phase or constituent present. Although such make a conclusive fracture interpretation and specimens can be examined by scanning All fracture examinations should begin with failure analysis difficult, if not impossible. In electron microscopy (SEM), light microscopy visual inspection, perhaps aided by the use of a some instances, however, satisfactory results is more efficient for such work, and certain simple hand lens (up to 10 x ). In failure anal- can be obtained (see the article "Failures of information, such as the color or polarization yses, it may be necessary to examine the entire Locomotive Axles" in Volume 11 of the 9th response of constituents, can be assessed only component or structure to find the broken Edition of Metals Handbook). 92 / Visual Examination and Light Microscopy

The fractured sections must be protected compression, shear, bending, and so on), cracks for examination. The damage to second- from further damage after the incident. It is the relative stress level (high, medium, or ary cracks is usually less than that of the main often necessary to section the failed component low), and the stress orientation. fracture. It is important to remember that the or structure, sometimes at the failure site, so Examine areas selected by macroscopic ex- crack mechanism for propagation may be dif- that it can be studied more extensively in the amination at higher magnifications by light ferent from that at the initiation site. Damage laboratory. Sectioning must be carried out in microscopy, SEM, or replica transmission done to the main fracture may prohibit success- such a manner that the fracture and adjacent electron microscopy (TEM) to determine the ful microfractography, bUt the gross macro- material is not altered. If burning is used, it fracture mode, to confirm the fracture mech- scopic features may still be visible and amena- should be conducted well away from the fail- anism (observation of cleavage facets, duc- ble to interpretation. ure. Similarly, band saw cutting or abrasive tile dimples, fatigue striations, and so on), Mechanical damage to the fracture surface wheel cutting must be conducted well away and to detect features at the fracture origin may occur during crack propagation, for exam- from the fracture. It is generally necessary to Examine metallographic cross sections con- protect the fracture during such work. This taining the origin to detect any microstruc- subject is treated at length in the article "Prep- tural features that promoted or caused frac- aration and Preservation of Fracture Speci- ture initiation, and determine if crack mens" in this Volume. propagation favors any microstructural con- Macroscopic examination is the first step in stituent fracture interpretation. In most cases, the origin must be determined in order to obtain conclusive results. Careful macroscopic examination Visual macroscopic examination is the most should always precede any microscopic exam- efficient procedure for fracture evaluation. This ination. Macroscopic examination will gener- should be followed by stereoscopic examina- ally permit determination of the manner of tion of the fracture features using magnifica- loading, the relative level of applied stress, the tions up to about 50 ×. Before the fracture is mechanisms involved, the sequence of events, sectioned, relevant details should be recorded and the relative ductility or brittleness of the photographically (see the article "Photography material. Other details can be revealed by gross of Fractured Parts and Fracture Surfaces" in (al fracture examination--for example, the pres- this Volume). Sketches are also very useful. ence of hardened cases; apparent grain size or Dimensions should be recorded before cutting variations of grain size; material imperfections, begins. such as segregation, gross inclusions, or hydro- During this work, it is imperative that the gen flakes; and fabrication or machining imper- fracture face be handled carefully. Excessive fections that influenced failure. handling and touching of the fracture face The usual sequence for the examination of should be avoided, and fractures should not be fractured components is as follows (Ref 1): remated. Field fractures often exhibit debris on the fracture face that may be relevant to the • Visually survey the entire component to diagnosis, particularly for corrosion-related obtain an overall understanding of the com- failures. This contamination should not be re- ponent and the significance of the fractured moved without serious consideration. The de- area bris should be analyzed in situ or after removal. • Classify the fracture from a macroscopic Cleaning procedures, which are also discussed viewpoint as ductile, brittle, fatigue, tor- in the article "Preparation and Preservation of sion, and so forth Fracture Specimens" in this Volume, should be (b) • Determine the origin of failure by tracing the performed only after the relevance of such fracture back to its starting point or points debris has been determined. Fig, I Comparison of dark-field light microscope fractograph (a) and an SEM secondary elec- • Based on the observed fracture features, If damage to the gross fracture face is exten- tron image (b) of the same area in an iron-chromium- estimate the manner of loading (tension, sive, it is often helpful to open secondary aluminum alloy. Both 50 ×

(a) (b) (c) Fig. 2 Camparison of light microscope and SFM fractographs of the same area of on iron-chramium-aluminum alloy. (a) Bright-field light fractograph. (b) Dark-field light fractograph. (c) SFM secondary electron image. All 50 × Visual Examination and Light Microscopy / 93

pie, because of crack closure during fatigue crack growth. Although such damage may severely impair macro- and microscopic fractography, it does provide information about the manner of stressing. Light Fractography Several early researchers attempted to examine fracture surfaces with the light microscope, but the limitations of these early instruments precluded successful application. Little work of this type was performed until C.A. Zapffe demonstrated the usefulness of such examinations in the 1940s (Ref 2-5). Before this, fractures were examined only at high magnifi- (a) (b) cations with the light microscope by using plated sections normal to the fracture surface. Although this technique remains very useful, Zapffe recognized that direct observation of the fracture surface would offer advantages and that the restricted depth of field of the light microscope would limit direct examination to brittle flat fractures (Ref 2). Consequently, Zapffe initially defined the new field of fractography as the micrographic study of cleavage facets on fractured metal specimens (Ref 2). Several years later, he generalized the definition of fractography, defining it as the study of detail on fracture surfaces (Ref 5). Zapffe's technique consisted of obtaining a coarse brittle fracture, using low-power observation to find suitable cleavage facets, using a special sample holder to orient the facet perpen- Ic) Id) dicular to the optical axis, and focusing at the Fig. 3 Comparison of light microscope (a and b) and SEM (c and d) fractographs of cleavage facets in a desired examination magnification. Zapffe and coarse-grain Fe-2.5Si alloy broken at -195 °C (-320 °F). (a) Bright-field illumination.(b) Dark-field his co-workers developed this method to a fine illumination. (c) Secondary electron image. (d) Everhart-Thornleybackscattered electron image. All 60 × art. However, the inherently small depth of field, which decreases markedly with increasing magnification and numerical aperture, lim- such devices), good results can still be ob- as those of the light micrograph. Figure 2 ited the technique. Despite the interesting retained. For example, the fracture can be shows bright-field and dark-field light frac- suits shown by Zapffe, relatively few optical searched at low magnification for suitably ori- tographs of similar steplike cleavage fractures fractographs have been published by others, ented facets, or the fracture can be tilted in this iron-chromium-aluminum sample and and microfractography did not gain general slightly. In many cases, dark-field illumination the same area using secondary electron imag- acceptance until the development of TEM rep- can provide better images than bright-field ing. It is apparent that the dark-field image is lication methods in the 1950s. With the com- illumination. The dark-field objective gathers superior to the bright-field image, but neither mercial introduction of the scanning electron the light that is scattered at angles away from reveals the stepped nature as well as the SEM microscope in 1965, the field of microfractog- the optical axis and often provides better image image, although the observer could see this raphy gained popular applicability. contrast with less glare. In some of the exam- effect by focusing in and out. Although microfractography by SEM is a pies to be shown using both bright-field and Figures 3 to 5 show coarse cleavage facets.on much simpler technique that produces equiva- dark-field illumination, certain features are best a fractured Fe-2.5Si sample that was broken lent images with far greater depth of field, the visible in one mode or the other. The photo- after cooling in liquid nitrogen. The SEM views use of the light microscope in fractography graphs, however, cannot record the details that are not of the same areas as the light frac- should not be discarded. Lack of access to a the observer can see as the focus point is moved tographs. Figure 3 shows bright-field and dark- scanning electron microscope should not pre- up and down. field images of a coarse facet. Certain areas are vent the innovative metallographer from pursu- Figures 1 and 2 show two areas observed on best observed on one or the other image being fractography beyond macroscopic tech- the fracture of an iron-chromium-aluminum cause of the different light collection proce- niques. Direct optical examination of the alloy. Figure 1 shows the same area using dures. The two SEM views show a different fracture and examination of the fracture profile dark-field light microscopy and by SEM using area observed using secondary electrons and by with cross sections will often provide the re- secondary electrons. The dark-field image re- backscattered electrons with the Everhart- quired information. veals several parallel steplike flat features sur- Thornley detector. The latter procedure, be- Zapffe's work involved bright-field illumina- rounded by fine cleavage facets. The bright- cause of its sensitivity to the specimen-detector tion, which is perfectly suitable for examination field image was very poor and is not shown. orientation, provides greater height contrast of cleavage facets perpendicular to the optical The SEM image shows the same area more compared to secondary electron images. Fea- axis. However, if a special device is not avail- clearly because of the greater depth of field. tures that are not in a direct line with the able for orienting cleavage facets perpendicular The SEM negative was printed upside down so detector are dark or are in shadow. Comparison to the optical axis (few metallographers have that the details are oriented in the same manner of the secondary electron and Everhart-Thorn- 94 / Visual Examination and Light Microscopy

of the ductile region of the test fracture at the same magnification. Microvoid coalescence (dimples) can be observed in all of the fractographs, but the limited depth of field of the light fractographs is obvious. These examples demonstrate that the microscopic aspects of fractures can be assessed with the light microscope. Although the examination is easier and the results are better with SEM, light microscopy results are adequate in many cases. Such examination is easiest to accom- plish when the fracture is relatively flat. For rougher, more irregular surfaces, SEM is far superior. As a further note on the use of the light (a) Ib) microscope to examine fractures, Fig. 12 shows a cleavage fracture in a low-carbon martensitic steel examined by using three direct and three replica procedures. Figure 12(a) shows an example of the examination of the fracture profile after nickel plating the surface. The flat, angular nature of the fracture surface is apparent. Figures 12(b) and (d) show a light microscope direct view of the cleavage surface and a light microscope view of a replica. Figures 12(c) and (e) show a direct SEM view of the fracture and a view of a replica of the fracture by SEM, respectively. Lastly, Fig. 12(f) shows a TEM replica of the fracture, but at a much higher magnification. The transmission electron microscope cannot be used at magnifications below about 2500 x. (c) (dl

Fig. 4 Comparison of light microscope (a and b) and SEM (c and d) fractographs of cleavage facets in a Replicas for Light Microscopy coarse-grain Fe-2.5Si alloy impact specimen broken at -195 °C (-320 °F). (a) Bright-field illumination. (b) Dark-field illumination. (c) Secondary electron image. (d) Everhart-Thornley backscattered electron image. All In some situations, primarily in failure anal- 60 x ysis, the fracture face cannot be sectioned, generally for legal reasons, so that it can fit ley backscattered electron images shows that tions in the precracked region is barely visible within the chamber of the scanning electron the latter often reveals detail that is not as in the light microscope images compared to the microscope. In such cases, the fractographer obvious in the secondary electron image. The SEM images. The test fracture region is inter- can use replication procedures with examina- secondary electron image is fully illuminated. granular, but this is not obvious in the light tion by light microscopy, SEM, or TEM. The Therefore, again, a combined presentation of fractographs. Figure 7 shows bright-field and replication procedures for light microscopy are both images reveals additional information. dark-field fractographs of the fatigue-pre- similar to those traditionally used for TEM Figure 4 shows similar bright-field and cracked region, in which the striations are more fractography (Ref 6-8). dark-field light fractographs of cleavage facets easily observed than in Fig. 6. Figure 8 shows In general, the 0.25-mm (0.01-in.) thick in the Fe-2.5Si alloy as well as secondary secondary electron SEM fractographs of the cellulose acetate tape used for light microscopy electron and Everhart-Thornley backscattered striations in the precrack region at the same is thicker than that used for TEM. The tape is electron images of another area. The images magnifications as in Fig. 7. Figure 9 shows moistened on one side with acetone, and this shown in Fig. 4 are similar to those shown in bright-field and dark-field light fractographs of side is then pressed onto the fracture surface Fig. 3. Figure 5 shows higher-magnification the intergranular region. The intergranular na- and held tightly in place, without motion, for 1 bright-field light fractographs of coarse cleav- ture of this zone is more obvious in Fig. 9 than to 2 min. When thoroughly dry, the tape is age facets in the Fe-2.5Si specimen and SEM in Fig. 6. Figure 9 also shows corresponding carefully stripped from the fracture surface. An secondary electron images at the same magni- secondary electron and Everhart-Thornley alternate procedure consists of, first, preparing fications. All four fractographs are of different backscattered electron fractographs of the in- a viscous solution of cellulose acetate tape areas. tergranular test fracture. dissolved in acetone and applying a thin coating Figure 6 shows the interface between the For comparison, Fig. 10 shows a similar to the fracture. Then, a piece of cellulose fatigue precrack and test fracture of an X-750 X-750 rising-load specimen tested in air in acetate tape is placed on top of this layer, nickel-base superalloy subsize Charpy rising- which the test fracture is ductile. Figure 10 pressed into the fracture, and held in place for 1 load test specimen after testing in pure water at shows bright-field and dark-field light fracto- to 2 min. After drying, it is stripped from the 95 °C (200 °F). The interface region is shown graphs of the interface between the fatigue fracture. by bright-field and dark-field light microscopy precrack and the ductile test fracture. A second- The stripped tape is a negative replica of the (same areas) and by secondary electron and ary electron fractograph is also included for fracture and can be viewed as stripped from the Everhart-Thornley backscattered electron im- comparison. Figure 11 shows high-magnifica- fracture, and it can be photographed to record ages (different location, same areas). At the tion bright-field and dark-field light fracto- macroscopic fracture features (Fig. 13). This magnification used, evidence of fatigue stria- graphs and a secondary electron fractograph tape is a permanent record of the fracture for Visual Examination and Light Microscopy / 95

though some metals cannot be plated in this manner (Ref 9) and require other plating procedures. Figure 15 demonstrates the excellent edge retention that can be obtained with electroless nickel. A number of other edge retention procedures can also be used (Ref 9). Examination of fracture profiles yields considerable information about the fracture mode and mechanism and about the influence of microstructure on crack initiation and propagation. This is accomplished by examining partially fractured (Ref 10-15) or completely fractured (Ref 16-26) specimens. Quan- titative fractography makes extensive use of fracture profiles (additional information is (a) (bl available in the article "Quantitative Fractog- raphy" in this Volume). One interesting approach defines a fracture path preference index to describe the probability of a particular microstructural constituent being associated with a particular fracture mode to assess the relationship between fracture characteristics and microstructure (Ref 24). In general, it is easier to assess the relationship between crack path and microstructure by using secondary cracks because both sides of the fracture can be examined. On a completely broken specimen, only one side can be examined; this makes the analysis more difficult. Although most light micrographs are taken with bright-field illumination, the analyst should also try other illumination modes. Frac- (c) (d) tures should be initially examined on cross sections in the unetched condition and then Fig. 5 Comparison of light microscope (a and b) and SEM (c and d) fractographs of cleavage facets in a should be examined after etching. Naturally, a coarse-grain Fe-2.5Si alloy impact specimen broken at 195 °C (-320 °F). (a) Bright-field illumination. high-quality polish is required, and the effort (b) Dark-field illumination. (c) Secondary electron image. (d) Secondary electron image. (a) and (c) 120 x . (b) and extended in achieving a high-quality surface is (d) 240 x always rewarded with improved results and ease of correct interpretation. Errors in inter- future examination even if the fracture is examination of a fracture replica by light mi- pretation are made when specimens are not sectioned. croscopy using high magnification. This replica properly prepared. Additional contrast can be obtained by shad- was shadowed with gold-palladium. As-polished samples should be examined owing the replica with either carbon or a heavy first with bright-field illumination and then with metal, such as chromium, molybdenum, gold, dark-field illumination, differential interference or gold-palladium, as is normally done in TEM Fracture Profile Sections contrast (DIC) or oblique light, and polarized fractography. Some fractographers also coat the Despite the progress made in direct exami- light, if the specimen will respond to such back side of the replica with a reflective metal, nation of fracture surfaces, examination of illumination. Dark-field illumination is very such as aluminum, for reflected-light examina- sections perpendicular to the fracture, particu- useful and is highly suited to the examination of tion, or they tape the replica to a mirror surface. larly those containing the initiation site, is a cracks and voids. Photography in dark field is With an inverted microscope, some fractogra- very powerful tool of the fractographer and is more difficult, but not impossible if an auto- phers place the replica over the stage plate and virtually indispensable to the failure analyst. If matic exposure device is available. Oblique then place a polished, unetched specimen the origin of the fracture can be found, the light and DIC are very useful for revealing against the tape to hold it flat and to reflect the failure analyst must examine the origin site by topographic (relief) effects. For example, Fig. light. Others prefer to examine the tape with using metallographic cross sections. This is the 16 shows a fatigue crack in an aluminum alloy transmitted light, but not all metallographers only practical method for characterizing the viewed with bright-field illumination and DIC have access to a microscope with transmitted- microstructure at the origin and for assessing where the specimen was not etched. Although light capability. Figure 14 illustrates low-power the role that the microstructure may have had in the second-phase precipitates can be seen in examination procedures for examination of rep- causing or promoting the fracture. both views, they are more clearly revealed with licas by light microscopy. Figure 14(a) shows The safest procedure is to cut the sample to DIC (compare these views with the bright-field the replica photographed with oblique illumina- one side of the origin, but only after all prior etched micrograph of this specimen shown in tion from a point source lamp, and Fig. 14(b) nondestructive examinations have been com- Fig. 71). shows the same area using transmitted light. pleted. Cutting must be done in such a manner Another example of the examination of frac- Carbon was then vapor deposited onto the that damage is not produced. A water-cooled ture profiles is shown in Fig. 17, which illus- replica, and it was photographed again using abrasive cutoff machine or a low-speed dia- trates an impact fracture in an austenitic weld oblique light from a point source (Fig. 14c). mond saw is typically used. For optimum edge containing ~r phase. This sample is in the Figure 14(c) exhibits the best overall contrast retention, it is recommended that the surface be as-polished condition and is shown examined and sharpest detail. Figure 12(d) illustrates plated, generally with electroless nickel, al- with bright-field, DIC, and dark-field illumina- 96 / Visual Examination and Light Microscopy

from the objective of the microscope and is at the focal plane. During photographic exposure, the specimen is moved at a constant rate up through the light beam. Only the illuminated portions of the specimen are recorded photographically, and all of the illuminated portions are in focus; therefore, the resultant photograph is in focus. The use of the deep-field microscope and the problems encountered in obtaining good fractographs are discussed in Ref 39. Interpretation of Fractures The study of fractures has been approached (a) (b) in several ways. One procedure is to categorize fractures on the basis of macro- or microscopic features, that is, by macro- or microfractography. The fracture path may be classified as transgranular or intergranular. Another approach is to classify all fractures as either ductile or brittle, with all others, such as fatigue, being special cases of one or the other. In general, all fractures can be grouped into four categories: ductile, brittle, fatigue, or creep. After these broad groupings, the fractures can be further classified on the basis of environmental influences, stress situations, or embrittlement mechanisms. In this section, the macro- and microscopic characteristics of fractures produced by the more common mechanisms will be described and illustrated, with emphasis on visual and light microscopy exam- (c) Id) ination. Detailed information on this subject is also available in the article "Modes of Frac- Fig. 6 Comparison of light microscope (a and b) and SEM (c and d) fractographs of the interface between the fatigue-precracked region and the test fracture in an X-750 nickel-base superalloy rising-load test ture" in this Volume. specimen. The test was performed in pure water at 95 °C (200 °F). Note the intergranular nature of the fracture. (a) Bright-field illumination. (b) Dark-field illumination. (c) Secondary electron image. (d) Everhart-Thornley Ductile Fractures backscattered electron image. All 60 x Ductile fractures have not received the same attention as other fracture mechanisms because tion. It is clear that full use of the light to the fracture edge. The magnification factor is their occurrence results from overloading under microscope can provide a better description of defined by the cosecant of the sectioning angle; predictable conditions. From the standpoint of the relationship of the crack path to the micro- an angle of 5 ° 43' gives a tenfold magnifica- failure analysis, ductile failures are relatively structure. tion. uncommon, because their prevention through Examination of properly polished specimens proper design is reasonably straightforward. without etching often presents a clearer picture Etching Fractures Ductile failures, however, are of considerable of the extent of fracture because etched micro- Zapffe (Ref 2) and others (Ref 32-37) have interest in metal-forming operations and in structural detail does not obscure the crack etched fracture surfaces in order to gain addi- quality control studies, such as materials eval- detail. Etching presents other dark linear fea- tional information. In general, etching is used uation. tures, such as grain boundaries, that may be to reveal the microstructure associated with the Ductile failures occur through a mechanism confused with the crack details. Therefore, it is fracture surface (Ref 2, 36) or to produce etch known as microvoid coalescence (Ref 41-43). always advisable to examine the specimens pit attack to reveal the dislocation density and The microvoids are nucleated at any disconti- unetched first. Also, inclusions and other hard the crystallographic orientation of the fracture nuity where a strain discontinuity exists--for precipitates are more visible in unetched than in surfaces. Figure 18 shows an example of frac- example, grain or subgrain boundaries, second- etched specimens. After careful examination of ture surface etching of a cleavage fracture in a phase particles, and inclusions. These mi- the as-polished specimen, the sample should be carbon steel sample. Although the etched frac- crovoids, shown in Fig. 10 and 11, are referred etched and the examination procedure repeated. tures can be examined by light microscopy, to as dimples. Because of the roughness of the Examples of fracture profile sections in the SEM is simpler and produces better results. dimples, they are best observed with the scan- etched condition will be shown later in this ning electron microscope. On a typical ductile article. Deep-Field Microscopy fracture, fine precipitates, generally inclusions, The deep-field microscope provides greater can usually be observed in nearly half of the Taper Sections depth of field for optical examination and dimples (there are two halves of the fracture; Taper sections are often used to study frac- photography of fractures (Ref 38-40). Its theo- therefore, on any fracture face, only half of the tures (Ref 9, 27-31). In this method, the spec- retical depth of field is 6 mm at 38 x and 600 dimples, or fewer, will contain precipitates). imen is sectioned at a slight angle to the fracture I~m at 250 x (Ref 39). The instrument uses a Ductile fracture is sometimes referred to as surface. Polishing of this plane produces a very thin beam of light to illuminate the speci- dimple rupture; such is the case in the article magnified view of the structure perpendicular men. The light beam is at a constant distance "Modes of Fracture" in this Volume. Visual Examination and Light Microscopy / 97

(a) (b) (a)

(cl (d)

Fig. 7 Light microscope fractographs of the fatigue-precracked region of an alloy X-750 rising-load test specimen. (a) Bright-field image. (b) Dark-field image. (c) Bright-field image. (d) Dark-field image. (b) (a) and (b) 60 X. (c) and (d) 240 ×. Fig• 8 Secondary electron images of the fatigue- Ductile fractures exhibit certain characteris- This vessel was designed to contain gas at 31 precracked region of an alloy X-750 test specimen. (a) 65 x. (b) 260 x tic microscopic features: MPa (4475 psig). Therefore, the failures occurred by overloading. • A relatively large amount of plastic defor- From a microscopic viewpoint, a ductile common for microvoids to form at interfaces mation precedes the fracture fracture exhibits microvoid coalescence and between the matrix phase and inclusions or hard • Shear lips are usually observed at the frac- transgranular fracture. The dimple orientation precipitates. Cementite in steel, either spheroid- ture termination areas will vary with stress orientation. Dimple shapes al or lamellar, can act as a nucleation site. For • The fracture surface may appear to be fi- and orientations on mating fractures as influ- example, ductile fractures were studied in an brous or may have a matte or silky texture, enced by the manner of load application are iron-chromium alloy, and large dimples were depending on the material summarized in Ref 44. Dimple shape is a observed at sulfide inclusions that subsequently • The cross section at the fracture is usually function of stress orientation, with equiaxed merged with much finer voids at chromium reduced by necking dimples being formed under uniaxial tensile carbides (Ref 45). • Crack growth is slow loading conditions. Shear stresses develop Ductile fractures in very pure low-strength The macroscopic appearance of a ductile elongated, parabolically shaped dimples that metals can occur by shear or necking, and they fracture is shown by the controlled laboratory point in opposite directions on the mating can rupture without any evidence of microvoid fracture of a spherical steel pressure vessel fracture surfaces. If tearing occurs, the elon- formation. Such fractures can also be observed measuring 187 mm (73/8 in.) outside diameter gated dimples point in the same direction on the in sheet specimens in which triaxial stress and 3.2 mm (1/8 in.) in wall thickness that was mating fracture surfaces. development is negligible. Reference 46 dis- made from quenched-and-tempered AISI 1030 The number of dimples per unit area on the cusses the problem of designing an experiment aluminum-killed fine-grain steel with an impact fracture surface depends on the number of to prove conclusively whether or not ductile transition temperature below -45 °C (-50 °F). nucleation sites and the plasticity of the mate- fracture can occur by void formation in the At room temperature, ductile rupture occurred rial. If many nucleation sites are present, void absence of hard particles. Results of the most when the vessel was pressurized to 59 MPa growth is limited, because the dimples will carefully controlled experiments indicate that (8500 psig) (Fig. 19). Figure 20 shows an intersect and link up. This produces many particles are required for void formation. Stud- identical vessel pressurized to failure at -45 °C small, shallow dimples. If very few nucleation ies of high-purity metals have shown that duc- (-50 °F) at 62 MPa (9000 psig). Greater sites are present, there will be a small number tile fracture occurs by rupture without void pressure was required because the strength of of large dimples on the fracture surface. formation. the steel was greater at -45 °C (-50 °F) than at The interface between two ductile phases can Markings can be observed within dimples room temperature. Both failures are ductile. act as a nucleation site, but it is much more because considerable plastic deformation is oc- 98 / Visual Examination and Light Microscopy

(a) (bl (a)

Fig. 9 Comparison of light microscope (a and b) and SEM (c and d) fractographs of the test fracture in an alloy X~750 rising-load test specimen. Test was performed in pure water at 95 °C (200 °F). Note the (b) intergranular appearance of the fracture. (a) Bright-field image. (b) Dark-field image. (c) Secondary electron image. (d) Everhart-Thornley backscattered electron image. All 60 ×

curring. Such marks have been referred to as ment. Nevertheless, the relative amount of serpentine glide, ripples, or stretching. ductility exhibited by tensile specimens varies Although light microscopy has rather limited considerably. Although the percent elongation value for the examination of ductile dimples (as and percent reduction of area (% RA) are not shown in Fig. 10 and 11), dimples can be easily intrinsic mechanical properties like the yield observed by using metallographic cross sec- and tensile strength, they do provide useful tions. Figure 21 compares the appearance of comparative data. ductile and brittle fractures of a quenched-and- The ideal tensile fracture is the classic cup- tempered low-alloy steel using nickel-plated and-cone type (Fig. 22a). This type of fracture cross sections. The ductile fracture consists of a occurs in highly ductile materials. Pronounced series of connected curved surfaces, and it is necking is evident in Fig. 22(a). In comparison, easy to envision how this surface would appear in the brittle tensile fracture shown in Fig. if viewed directly. Also, near the fracture 22(b), no necking has occurred, and the percent surface, small spherical voids, particularly at elongation and % RA values are nearly zero. In inclusions, are often seen that have opened this type of tensile fracture, the yield and tensile during the fracture process. These should not be strengths are essentially identical. (c) interpreted as pre-existing voids. The brittle In one investigation, the classic cup-and- Fig. 1 0 Comparison of light microscope (a and b) fracture, by comparison, is much more angular, cone tensile fracture was studied by optical and SEM (c) images of the interface be- and crack intrusions into the matrix on cleavage examination and by etch pitting (Ref 47). tween the fatigue-precrack area (left) and the test planes can be observed (Fig. 21). The problems Shear-type fracture was found to be present all fracture region (right) of an alloy X-750 rising-load associated with the measurement of dimples are across the specimen, even in the flat central test specimen broken in air. The test fracture is ductile. (a) Bright-field image. (b) Dark-field image. (c) Second- discussed in the article "Quantitative Fractog- portion of the fracture. The approximate mag- ary electron image. All 68 × raphy" in this Volume. nitude and distributions of longitudinal, radial, and circumferential stresses were determined in fracture in cup-and-cone fractures begins at the Tensile-Test Fractures unfractured tensile bars strained at various interior of the specimen (Fig. 23) and pro- Tensile specimens are tested under condi- amounts. The triaxial stresses in the necked gresses to the surface. tions that favor ductile fracture, that is, room section develop the highest shear stress at the When a smooth tensile specimen is tested at temperature, low strain rate, and a dry environ- center, not at the surface of the specimen. Shear room temperature, plastic deformation is ini- Visual Examination and Light Microscopy / 99

(a) (b) (c)

,,u="-', 1 1 Comparison of light microscope (a and b) and SEM (c) images of a ductile fracture in an alloy X-750 rising-load test specimen broken in air. (a) Bright-field image. (b) Dark-field image. (c) Secondary electron image. All 240 ×

(a) (b) (c)

(d) (e) (f)

Fig. ! 2 Examples of three direct (a to c) and three replication procedures (d to f) for examination of a cleavage fracture in a low-carbon martensitic steel. (a i Light microscope cross section with nickel plating at top. (b) Direct light fractograph. (c) Direct SEM fractograph. (d) Light fractograph of replica. (e) SEM fractograph of replica. (f) TEM fractograph of replica

tially characterized by uniform elongation. At bands of high shear strain. These bands are this stage, voids begin to form randomly at oriented at an angle of 50 to 60 ° from the large inclusions and precipitates. With further transverse plane of the test specimen. Sheets of deformation, plastic instability arises, produc- voids are nucleated in the shear bands, and ing localized deformation, or necking, and a these voids grow and link up, producing a shift from a uniaxial to a triaxial stress state. serrated pattern as the central crack expands This results in void nucleation, growth, and radially. The cup-and-cone walls are formed Fig. 13 Transparent tape replica of a fracture surface. See the article "Transmission coalescence at the center of the necked region, when the crack grows to such an extent that the Electron Microscopy" in this Volume for more informa- forming a central crack. Continued deformation void sheets propagate in one large step to the tion on replication techniques. is concentrated at the crack tips; this produces surface. 100 / Visual Examination and Light Microscopy

Comparison of replica fractographs of a fatigue fracture in an induction-hardened 15B28 steel shaft. Fracture was initiated at the large inclusion in the center of Fig, 14 the views during rotating bending. (a) Oblique illumination from a point source lamp. (b) Same area as (a), photographed using transmitted light. (c) Replica shadowed with a vapor-deposited coating and photographed using oblique illumination from a point source lamp. All 30 ×

..=~=;". ! 5 Example of the use of electroless nickel plating to provide edge retention. The (a) (b) micrograph shows wear damage at the surface of a forged alloy steel Medart roll. Etched with 2% nital. 285 x Fig, ! 6 Comparison of bright-field (o) and Nomarski DIC (b) illumination for examination of a fatigue crack in an as-polished aluminum alloy. See also Fig. 71. Both 600 ×

25um - " m (a) (b) (c)

--'~l~;r= ! 7 Comparison of bright-field (a), DIC (b), and dark-field (c), illumination for viewing a partially fractured (by impact) specimen of AISI type 312 weld metal containing substantial (~ phase. All 240 x

Tensile fractures of oxygen-free high-con- resulting in a cup-and-cone fracture or by an cone fractures (Ref 48, 49). The fracture mode ductivity (OFHC) copper were studied by de- alternate slip method producing a double-cup changes from chisel-point to double-cup to forming the specimens in tension until neck- fracture. Cup-and-cone fractures are observed cup-and-cone as the precipitate density and ing and then halting the test (Ref 10). After in ductile iron-base alloys, brass, and Dur- alloy content increase. radiographic examination, the specimens were alumin; double-cup fractures are seen in face- One investigation studied the influence of sectioned and examined. This work also centered cubic (fcc) metals, such as copper, particle density and spacing and solute content showed that ductile tensile fractures begin by nickel, aluminum, gold, and silver. on tensile fracture of aluminum alloys (Ref 50). void formation, with the voids linking up to Three types of tensile fractures have been As the particle density increases, there is a large form a central crack. The fracture can be observed in tests of fcc metals: chisel-point initial decrease in ductility, followed by a completed either by continued void formation fractures, double-cup fractures, and cup-and- gradual loss in ductility with further increases Visual Examination and Light Microscopy / 101

in particle density. Other studies demonstrated that tensile ductility improves as the volume fraction of second-phase particles decreases (Ref 51, 52). Particle size and mean interparticle spacing also influence ductility, but the volume fraction is of greater importance. Dim- ple size has been shown to increase with increasing particle size or increasing interparticle spacing, but dimple density increases with increasing particle density (Ref 53). Tensile ductility in wrought materials is higher in longitudinally oriented specimens (fracture perpendicular to the fiber axis) than in transversely oriented specimens (fracture parallel to the fiber (a) (b) axis) because of the elongation of inclusions, and some precipitates, during hot working. The tensile test is widely used for quality control and material acceptance purposes, and its value is well known to metallurgists and mechanical engineers. Tensile ductility measurements, although qualitative, are strongly influenced by microstructure. Transversely oriented tensile specimens have been widely used to assess material quality by evaluation of the transverse reduction of area (RAT). Numerous studies have demonstrated the structural sensitivity of RAT values (Ref 54-60). The macroscopic appearance of tensile fractures is a result of the relative ductility or brittleness of the material being tested. Conse- Ic) (d) quently, interpretation of macroscopic tensile fracture features is an important skill for the Fig. 18 Example of the use of etching to produce etch pits (arrows) on a cleavage fracture. (a) As-fractured. 320 x. (b) Etched 60 s with nital. 320 ×. (c) Etched 360 s with nital. 320 x. (d) Etched 360 s with metallurgist. In addition to the nature of the nital. 1280x material, other factors can influence the mac-

Fig, 19 Macrograph showing ductile overload fracture of a high-pressure steel Fig. 20 Macrograph showing ductile overload fracture of a high-pressure steel vessel tested at room temperature. See also Fig. 20. vessel tested at -45 °C (-50 °F). See also Fig. 19. 102 / Visual Examination and Light Microscopy

(al Ibl

Fig. 21 Brittle (a) and ductile (b) crack paths in a fractured quenched-and-tempered low-alloy steel. Both etched with 2% nital. 800 ×

(al (b) (c) (d)

Fig. 22 Macroscopic oppearonce of ductile (a) and brittle (b) tensile fractures

roscopic tensile fracture appearance--for ex- the latter is illustrated in Fig. 24. These ridges either the edge of the fibrous zone or from the ample, the size and shape of the test specimen are normal to the direction of crack propagation origin itself. In the latter case, it is easy to trace and the product form from which it came, the from the origin to the surface of the specimen. the radial marks backward to the origin. The test temperature and environment, and the man- The presence of such ridges indicates stable, radial marks may be fine or coarse, depending ner of loading. subcritical crack growth that requires high enon the material being tested and the test The classic cup-and-cone tensile fracture ex- ergy (Ref 61). The fracture origin is usually temperature. The radial marks on tensile hibits three zones: the inner flat fibrous zone located at or near the center of the fracture and specimens of high-strength tempered mar- where the fracture begins, an intermediate ra- on the tensile axis; it can often be observed to tensite steels are usually rather fine. Tempering dial zone, and the outer shear-lip zone where initiate at a hard second-phase constituent, such of such samples to lower strengths results in the fracture terminates. Figure 22(a) shows as an inclusion or a cluster of inclusions. coarser radial marks. Low tensile-test temper- each of these zones; the fiat brittle fracture Inspection of tensile fractures at low magnifi- atures result in finer radial marks than those shown in Fig. 22(b) has no shear-lip zone. cation-for example, with a stereomicro- produced with room-temperature tests. In a The fibrous zone is a region of slow crack scope--will frequently reveal the initiating mi- study of AISI 4340 steel, for example, fine growth at the fracture origin that is usually at or crostructural constituents (Ref 62). radial marks were not produced by a shear very close to the tensile axis. The fibrous zone The radial zone results when the crack mechanism but by quasi-cleavage, inter- has either a random fibrous appearance or may growth rate becomes rapid or unstable. These granular fracture, or both (Ref 61). Coarse exhibit a series of fine circumferential ridges; marks trace the crack growth direction from radial marks on steel specimens are due to Visual Examination and Light Microscopy/ 103

Fig. 24 Tensile fracture of a 4340 steel specimen tested at 120 °C (250 °F). The fracture contains a fibrous zone and a shear-lip zone. The steel Fig. 25 Macrograph showing a granular brittle microstructure consisted of tempered martensite; hard- fracture in a cast iron tensile bar. Note ness was 46 HRC. The fracture started at the center of Fig. 23 Initiation of fracture in a tensile-test speci- the large cleavage facets. 2 x men. Note that the fracture initiated at the fibrous zone, which shows circumferential ridges. The the center of the specimen. 4.75 x outer ring is the shear-lip zone. About 11 x shear, and longitudinally oriented splits can be observed along the ridges or peaks. Radial marks on tensile fractures are usually straight, but a special form of tensile fracture exhibiting coarse curved radial marks (the star or rosette pattern) can also occur, as discussed below. If the origin of the tensile fracture is off axis and if the fibrous zone is very small or absent, some curvature of the radial marks will be observed. The appearance of the radial marks is partly the result of the ductility of the material. When tensile ductility is low, radial marks are fine with little relief. If a material is quite brittle with a coarse grain size, the amount of tensile ductility is extremely low, and the crack path will follow the planes of weakness in directions associated with each grain. Thus, cracking will be by cleavage, will be intergranular, or a combination of these. Figure 25 illustrates such a fracture in a cast iron tensile bar. The outer shear-lip zone is a smooth, annular area adjacent to the free surface of the specimen. The size of the shear-lip zone depends on the stress state and the properties of the material tested. Changing the diameter of the test spec- Fig. 26 Side views of four types of ASTM A490 high-strength steel bolt tensile specimens, See also Fig. 27. Left imen will change the stress state and alter the to right: bolts 1, 4, 6, and 7 nature of the shear lip. In many cases, the shear-lip width will be the same, but the per- had a major portion of the shank turned down to exhibits a flat brittle fracture with a small split; centage of the fracture it covers will change. a diameter of 9 mm (0.357 in.). Bolt 6 had a bolt 7 has a slanted brittle fracture and a large However, exceptions to this have been noted major portion of the shank turned down to 13 split. (Ref 63). mm (0.505 in.) in diameter and was then The presence of voids, such as microshrink- An example of the influence of changes in notched (60 °) to 9 mm (0.357 in.) in diameter age cavities, can alter the fracture appearance. the section size, the presence of notches, and in the center of the turned section. Bolt 7 was a Figure 28 shows a tensile fracture from a the manner of loading on the appearance of full-size bolt tested with a 10 ° wedge under the carbon steel casting with a slant fracture due to fractures of round tensile specimens is provided head. Bolts 1, 4, and 6 were axially loaded. the voids present. by tests of 25-mm (1-in.) diam, ll4-mm (4l/2 Figure 26 shows side views of the four bolts Because the shape and size of the tensile in.) long ASTM A490 high-strength bolts (Fig. after testing, and Fig. 27 shows their fracture specimen influence the stress state, fracture 26 and 27). Bolt 1 was a full-size bolt with a surfaces. The fractures for bolts 1 and 4 exhibit zones will be different for square or rectangular portion of the shank turned to a diameter just the rosette star-type pattern, which is more fully sections compared to those with the round cross smaller than the thread root diameter. Bolt 4 developed in bolt 1. Bolt 6, which is notched, sections discussed previously. For an un- 104 / Visual Examination and Light Microscopy

Fig. 27 Macrographs of fracture surfaces of ASTM A490 high-strength bolt tensile specimens shown in Fig. 26. Top left to right: bolts 1 and 4; bottom left to right: bolts 6 and 7 notched, rectangular test specimen, for exam- Rosette star-type tensile fractures are ob- hot rolled appears to be an important crite- ple, the fibrous zone may be elliptical in shape, served only in tensile bars taken parallel to the rion for the formation of such fractures (Ref with the major axis parallel to the longer side of hot-working direction of round bar stock (Ref 65). the rectangle. Figure 29 shows a schematic of 65). The radial zone is the zone most Splitting also has been observed in ordinary such a test specimen, as well as two actual characteristic of such fractures, and it exhibits cup-and-cone tensile fractures of specimens fracture faces. The radial zone of the test longitudinally oriented cracks. The surfaces of machined from plates (Ref 72). As with the fracture is substantially altered by the shape of these cracks exhibit quasi-cleavage, which is rosette star-type tensile fracture, cup-and-cone the specimen, particularly for the example in formed before final rupture. Rosette, fractures fractures have been observed in quenched- Fig. 29(c). As shown by the schematic illustra- have frequently been observed in temper- and-tempered (205 to 650 °C, or 400 to 1200 tions in Fig. 30, as the section thickness deembrittled steels (Ref 64), but are also seen in °F) alloy steels. One study showed that the creases, the radial zone is suppressed in favor nonembrittled steels. occurrence of splitting and the tensile fracture of a larger shear-lip zone (Ref 72). For very Certain tensile strength ranges appear to appearance varied with test temperature (Ref thin specimens (plane-stress conditions), there favor rosette, fracture formation in specimens 72). Tests at 65 °C (150 °F) produced the is no radial zone. machined from round bars. To illustrate this cup-and-cone fracture, but lower test tempera- Tensile fractures of specimens machined effect, Fig. 31 shows tensile fractures and test tures resulted in one or more longitudinally with a transverse or short-transverse orienta- data for seven tensile specimens of heat-treated oriented splits. The splits were perpendicular tion from materials containing aligned second- AISI 4142 alloy steel machined from a to the plate surface and ran in the hot-working phase constituents--for example, sulfide in- 28.6-mm (ll/s-in.)diam bar. Specimens 1 and direction. The crack surfaces exhibited quasi- clusions, slag stringers (wrought iron), or 2 were oil quenched and tempered at cleavage. When the same material was rolled segregation---often exhibit a woody fracture 205 and 315 °C (400 and 600 °F) and exhibit to produce round bars, the tensile fracture appearance (Ref 13). In such fractures, the classic cup-and-cone fractures. Specimens 3 exhibited rosette star-type fractures. Therefore, aligned second phase controls fracture initia- to 7, which were tempered at 455, 510, 565, the split, layered cup-and-cone fracture was tion and propagation, and ductility is usually and 675 °C (850, 950, 1050, and 1250 °F), concluded to be a two-dimensional (plate) low or nonexistent. respectively, all exhibit rosette star-type ten- analog of the rosette, star fracture observed in Another unique macroscopic tensile fracture sile fractures; specimens 4 and 7 exhibit the heat-treated tensiles from round bars. appearance is the star or rosette fracture, which best examples of such fractures. Specimens 3 Splitting of fracture surfaces of both tensile exhibits a central fibrous zone, an intermediate to 7 exhibit longitudinal splitting. Shear lips specimens and Charpy V-notch impact speci- region of radial shear, and an outer circumfer- are well developed on specimen 3, but are mens has been frequently observed in speci- ential shear-lip zone (Ref 64-71). The nature poorly developed on specimens 4 and 7 and mens machined from controlled-rolled plate and size of these zones can be altered by heat are essentially absent on specimens 5 and (Ref 73-85) and line pipe steels (Ref 86-90). treatment, tensile size, and test temperature. 6. The testing temperature can also influ- The splits, which are also referred to as de- Figure 27 shows two examples of this type of ence formation of the rosette star-type fracture. laminations, are fissures that propagate in the tensile fracture. The texture produced when round bars are hot-rolling direction. Visual Examination and Light Microscopy/ 105

Section thickness ~l~_t~'\\ ?.r~~k\ ~ r~ g\ in\ x....\\ \\\ \ ..,Zx ~ \ \ \Fibrous \ ×4xR.ad~axlzo zone n\\; ~ c~ r°antkh io~ iksies ~,., ~ Shzeanil iP/F~F~ b i°lUzSoZ7 e

[ Arrows indicate Shear-lip zone direction of crack propagation

(e) Section thickness Shear-lip Fibrous zone / Crackorigin--~z°n~ ~'~ Radialzone

Section thickness Crack origin J Fibrous zone

Shear-lip zone Arrows indicate direction of crack propagation

(a) ''~glEln 30 Effect of section thickness on the fracture surface markings of rectangular tensile specimens. Schematics show the change in size of the radial zone of specimens of progressively decreasing section thickness. The thinnest of the three examples has a small fibrous zone surrounding the origin and a shear-lip zone, but no radial zone. Source: Ref 63

longitudinal specimens when the plates were finish rolled at 480 °C (895 °F) and below Fig. 29 Appearance of fracture surfaces in rectangular steel tensile specimens. (a) Sche- and at 370 °C (700 °F) and below, respec- matic of tensile fracture features in a rectangular spec- tively. The splits were in the hot-rolling imen. (b) light fractograph with fracture features direction, and the frequency of splitting conforming to those of the schematic. (c) Light increased as the finishing temperature was fractograph of a fracture similar to (b) but having a much narrower shear-lip zone. Source: Ref 63 lowered below the temperatures mentioned previously. This study revealed that splitting followed Various factors have been suggested as the ferrite grain boundaries (Ref 83). The causes for these delaminations (Ref 82): aspect ratio of the deformed ferrite grains was found to be related to the occurrence of • Elongated ferrite grains produced by low splitting. Material that was susceptible to finishing temperatures that promote grain- splitting continued to exhibit splitting after boundary decohesion annealing until the ferrite grains were almost • Residual stress concentration (b) completely recrystallized. • Grain-boundary segregation As a final note on tensile fractures, numerous • Grain-boundary carbides studies have used metallographic cross sections • Ferrite/pearlite banding to assess the influence of second-phase constit- • Nonmetallic inclusions uents on fracture initiation and tensile proper- • Cleavage on (100) planes ties. Many studies have shown void formation • Mechanical fibering at the interface between hard constituents (car- • Duplex ferrite grain size bides, intermetallics, and inclusions) and the • Increased amounts of deformed ferrite matrix (Ref 20, 91-95); other studies have • Prior-austenite grain boundaries demonstrated quantitative relationships between inclusion parameters and tensile ductility In a study of a relatively pure Fe-lMn alloy (Ref 51, 96-102). The use of light microscopy that was essentially free of carbides and has been of great importance in such studies. inclusions, delaminations were observed along grain boundaries; splitting occurred when the Brittle Fractures ferrite grains were deformed beyond a certain Brittle fractures can occur in body-centered degree by controlled rolling (Ref 83). Exam- cubic (bcc) and hexagonal close-packed (hcp) (c) ples of splitting in 12.7-mm 2 (0.02-in. 2) metals but not in fcc metals (except in certain tensile specimens from this study are shown specific cases). Brittle fractures are promoted in Fig. 32, which illustrates splitting on by low service temperatures, high strain rates, Fig. 28 Three views of an unusual tensile fracture from a carbon steel casting. (a) Macro- longitudinal and transverse tensile specimens the presence of stress concentrators, and certain graph of the fracture surface. (b) SEM view of voids on where the hot-rolling finishing temperatures environmental conditions. The ductile-to-brittle the fracture surface. (c) Light micrograph showing were 315 and 150 °C (600 and 300 °F). transition over a range of temperatures is a shrinkage cavities. (c) Etched with 2% nital Splitting was observed in transverse and well-known characteristic behavior of steels 106 / Visual Examination and Light Microscopy

Tensile 0.2% yield [-"'- Temper --'I r-- strength -'-I r-- strength -7 Specimen *C *F MPa ksi MPa ksi Elongation, % %RA 1...... 205 400 1970 285 1690 245 10 39 2 ...... 315 600 1730 251 1550 225 10 43 3 ...... 455 850 1410 204 1310 190 12.5 47 4 ...... 510 950 1250 181 1170 169 15 54 5 ...... 565 1050 1130 164 1030 150 16 58 6 ...... 620 1150 945 137 850 123 20 63 7 ...... 675 1250 770 112 670 97 24.5 66

Macrographs of quenched-and-tempered AISI 4142 steel tensile specimens showing splitting parallel to the hot-working axis in specimens tempered at 455 °C Fig. 31 (850 °F) or higher

and is influenced by such factors as strain rate, of plate in the longitudinal direction (Fig. 35). bides, intermetallics, and nonmetallics) at the stress state, composition, microstructure, grain Therefore, splitting is always associated with grain boundaries will facilitate crack nucle- size, and specimen size. Macroscopic examina- planes of weakness in the rolling direction. ation. Because these particles are brittle, they tion and light microscopy, as well as electron Figure 36 shows Charpy V-notch absorbed- inhibit relaxation at the tip of blocked slip metallographic procedures, have played an im- energy curves for Fe-1Mn steel that was finish bands, thus reducing the energy required for portant role in gaining an understanding of rolled at temperatures from 960 to 316 °C (1760 crack nucleation. brittle fracture, and these analysis techniques to 600 °F) from the study discussed in Ref 83. Another important feature is the relationship are basic failure analysis tools. As the finishing temperature decreased, the between the cleavage planes in these panicles Most metals, except fcc metals, exhibit a absorbed-energy transition temperature (tem- and in the neighboring grains. If the cleavage temperature-dependent brittleness behavior that perature for a certain level of absorbed energy, planes in the grain and in the panicle are has been studied by using a wide variety of for example, 20 or 34 J, or 15 or 25 ft • lb) favorably oriented, little or no energy will be impact-type tests. The Charpy V-notch impact decreased, but the upper shelf energy also expended when the crack crosses the particle/ test has had the greatest overall usage, and decreased. Figure 37 shows the fracture appear- matrix interface. If the cleavage planes are macroscopic examination of the fracture sur- ance of Charpy V-notch specimens used to misoriented, considerable energy is required, faces is used to assess the percentages of ductile produce the curves in Fig. 36. This demon- and crack nucleation is more difficult. For a and brittle fracture on the specimens as a strates that the occurrence of splitting increased given stress level, the probability for crack function of test temperature. Figure 33 shows as the finishing temperature decreased. In gen- nucleation increases with the number of grain- fractures of six Charpy V-notch impact speci- eral, lowering the test temperature for a given boundary panicles. Panicle shape and size are mens of a low-carbon steel tested between -18 finishing temperature increased the number of also very important, as is the degree of segre- and 95 °C (0 and 200 °F), along with the test splits unless the test temperature was low gation of the panicles to the grain boundaries. data (absorbed energy, lateral expansion, and enough to produce a completely brittle cleavage The strength of the interface and the presence of percent ductile, or fibrous, fracture). Figure 33 fracture, for example, the -73-°C (-100-°F) pre-existing voids at the particles also influence also shows SEM views of the fractures resulting samples of the plates finish rolled at 707 and crack nucleation. from testing at -18 and 95 °C (0 and 200 °F) 538 °C (1305 and 1000 °F). An interesting study of cleavage crack initi- that illustrate cleavage and microvoid coales- Figure 38 shows the microstructural appear- ation in polycrystalline iron was conducted in cence, respectively. ances of the splitting in two Charpy V-notch which two vacuum-melted ferritic irons with Another example of the macroscopic appear- specimens in plate that was finish rolled at 315 low carbon contents (0.035 and 0.005%) were ance of Charpy V-notch impact specimens is °C (600 °F). The regions between the splits tested using tensile specimens broken between given in Fig. 34, which shows four specimens resemble a cup-and-cone tensile fracture. room temperature and -196 °C (-321 °F) (Ref of heat-treated AISI 4340 tested between - 196 Although brittle fractures are characterized 103, 104). For a given test temperature, micro- and 40 °C (-321 and 104 °F), as well as plots by a lack of gross deformation, there is always cracks were more frequent in the higher-carbon of the test data (Ref 63). The test specimen at some minor degree of plastic deformation pre- heat, and nearly all of the microcracks origi- -80 °C (-112 °F), which is near the ductile- ceding crack initiation and during crack nated at cracked carbides. The microcracks to-brittle transition temperature, shows a well- growth. However, the amounts are quite low, were most often arrested at pre-existing twins defined ductile zone surrounding an inner brit- and no macroscopically detectable deformation or grain boundaries, but were also arrested by tle zone. Such clear delineation between the occurs. the initiation of twins and slip bands at the ductile and brittle zones is not always obtained, Light microscopy has been used to study the advancing crack tip. as was the case with the samples shown in Fig. initiation of brittle fracture. In polycrystalline Grain size also influences both initiation and 33. bcc metals, grain boundaries can block the propagation of brittle fractures. The quantita- Splitting has been observed on Charpy V- motion of slip or twinning, resulting in high tive description of this relationship is known as notch specimens, as well as on tensile speci- tensile stresses and crack nucleation. If plastic the Hall-Petch equation (Ref 105, 106): mens, as discussed previously. One investiga- deformation is not able to relax these tensile tion (Ref 83) has shown that the orientation of stresses, unstable crack growth results. The the splits is always parallel to the rolling plane presence of hard second-phase particles (car- crc = % + kd -1/2 (Eq 1) Visual Examination and Light Microscopy / 107

where crc is the cleavage strength, d is the grain diameter, and ~o and k are constants. For single-phase metals, a finer grain size produces higher strength. Many studies of a very wide range of single-phase specimens (tensile) have demonstrated the validity of this relationship using several experimentally determined strengths (yield point, yield stress, or flow stress at certain values of strain). Equations similar to Eq 1 have been used to relate grain size to other properties, such as toughness, fatigue limit, creep rate, hardness, and fracture strength in stress-corrosion cracking (SCC). Macroscopically, brittle fractures are characterized by the following:

• Little or no visible plastic deformation precedes the fracture • The fracture is generally fiat and perpendicular to the surface of the component • The fracture may appear granular or crystalline and is often highly reflective to light. (al (b) Facets may also be observed, particular/y in coarse-grain steels • Herringbone (chevron) patterns may be present • Cracks grow rapidly, often accompanied by a loud noise

Figure 39 shows an example of a well- developed chevron fracture pattern in a railroad rail. This type of brittle fracture pattern is frequently observed in low-strength steels, such as structural steels. The origin of the chevron fracture is easy to find because the apexes of the chevrons point back to the origin. The origin is not on the sections shown in Fig. 39. Such fractures can propagate for some distance before being arrested. The wavy nature of the fracture edge is evident in Fig. 39. In general, the more highly developed the chevron pattern, the greater the edge waviness. Such patterns are not observed on more brittle materials. The chevron fracture results from a discontinuous growth pattern. It proceeds first by initiation or initiations, followed by union of (c) (d) the initiation centers to form the fracture surface. One investigation of chevron-type brittle Fig. 32 Macrographs of fractured longitudinal and transverse tensile specimens from plates finish-rolled at two temperatures. (a) Longitudinal specimen finished at 315 °C (600 °F). (b) Transverse specimen fractures, demonstrated that they are caused by finished at the same temperature. (c) Longitudinal specimen finished at 150 °C (300 °F). (d) Transverse specimen discontinuous regions of cleavage fracture finished at the same temperature. Source: Ref 83 joined by regions of shear and that the chevron features are the ridges between the cleavage and between chevron formation and cup-and-cone Fracture begins along the centerline and spreads shear zones (Ref 107). When such fractures type tensile fracture. This model predicts that as a disk-shaped crack toward the surfaces. occur, they tend to extend into the metal and the angle 0 (Fig. 40b) will be 72 ° and that this However, before reaching the surface, the frac- produce cracks below the main fracture surface, angle is independent of material thickness and ture mode changes, and rupture occurs on the particularly in the crack growth direction. The properties. To construct this model, 161 mea- conical surfaces of maximum shear, approxi- spreading of the crack toward the surfaces is not surements were made on 20 fracture surfaces, mately 45 ° from the tensile axis (dotted lines in always equal, and this leads to asymmetrical and a mean angle of 69 ° was observed, which A, B, and C, Fig. 40c). chevron patterns. The front of the propagating varied somewhat depending on the presence As necking progresses, plastic deformation crack tends to become curved, and traces of the and size of shear lips. Because plastic deforma- (shaded area, Fig. 40c) occurs ahead of the curved front can often be observed. tion is required for formation of chevron pat- crack front, with lines of maximum shear strain The growing crack front in chevron fractures terns, such patterns are not observed on more approximately 45 ° to the plate axis. If this can be considered a parabolic envelope enclos- brittle materials. plastic zone is large enough, fracture will ing a number of individually initiated cracks In Fig. 40(c), locations A to E demonstrate progress by shear only. However, if necking is that spread radially (Ref 108). Figure 40 shows the close resemblance between chevron frac- limited, thus limiting the extent of plastic de- a model for chevron formation and the analogy tures and the cup-and-cone tensile fracture. formation, many small shear fractures of dif- 108 / Visual Examination and Light Microscopy

Temperature, °C (*F) 25 (75) 65 (150) 95 (200) sions, and the nature of the fluid within the Energy, J (ft. Ib) 34 (25) 134 (99) 152 (112) pipe. Lateral expansion, mm (in.) 0.81 (0.032) 1.85 (0.073) 1.85 (0.073) % fibrous 65 95 100 Full-scale testing of line pipe, as shown in Fig. 43 to 46, revealed a good correlation between the crack speed and the 50% shear area transition temperature as determined by the Battelle drop-weight tear test (DWTT). The fracture appearance of the pipe reflects the crack speed, for example:

• Less than 10% shear area corresponds to D brittle fracture with a fast-running crack • Greater than 40% shear area corresponds to a ductile fracture with a crack speed below 275 m/s (900 ft/s) B~ Figures 43 to 46 show full-scale test results of American Petroleum Institute (API) grade X-60 line pipes in which the test temperature was varied with respect to the 50% shear area Temperature, *C (*F) -18 (0) -4 (25) 10 (50) DWTT transition temperature. In each test, the Energy, J (ft • Ib) 5.5 (4) 13.5 (10) 23 (17) Lateral expansion, mm (in.) 0.15 (0.006) 0.35 (0.014) 0.53 (0.021) pipe was loaded to 40% of its yield strength, % fibrous 15 20 40 and a 30-grain explosive charge was detonated beneath a 460-mm (18-in.) long notch cut in the pipe while the entire pipe was maintained at the desired test temperature. Figure 43 shows a completely ductile fracture in a line pipe tested at 5 °C (8 °F) above the 50% shear area DWTT transition temperature; the resultant crack speed was 85 m/s (279 ft/s). From the 460-mm (18-in.) notch, the crack propagated 840 mm (33 in.) in full shear and then 460 mm (18 in.) in tearing shear before stopping. Figure 44 shows the result of testing at 1 °C (2 °F) below the 50% shear area DWTT transition temperature; the average crack speed was 172 m/s (566 ft/s). The crack began by brittle cleavage with a 15% shear area and progressed 200 mm (8 in.) in this manner before changing to 100% shear. The shear fracture ran 940 mm (37 in.) before stopping. The macrograph shows the initial fracture area. Figure 45 shows the result of testing a line Fig. 33 SEM froctographs of ductile (D) and brittle (B) fractures in Chorpy V-notch impact specimens shown at pipe at 6 °C (10 °F) below the 50% shear area top. Both 400 × DWTT transition temperature; the resultant crack speed was 470 m/s (1550 ft/s). From the notch, the fracture propagated by cleavage with fering orientation will form, resulting in a Chevron patterns are observed in the samples a 15 to 18% shear area, with some small rough or serrated appearance. tested at -3 and 16 °C (27 and 60 °F) and are patches having shear areas as high as 70% (see This description reveals that chevron frac- most clearly developed in the latter. Testing at macrograph of fracture, Fig. 45). The crack tures are intermediate between completely brit- higher temperatures resulted in 100% shear propagated straight along the top of the pipe for tle and completely ductile (shear) fractures. fractures. 685 mm (27 in.) and then developed a wave Figure 41 shows five ship steel fractures pro- Figures 43 to 46, which are photographs of pattern. Only a half wave was completed before duced by drop-weight testing at temperatures controlled laboratory fractures of full-size line the crack changed to 100% shear and then tore from -30 to -70 °C (-25 to -96 °F). The pipes, illustrate several important macroscopic circumferentially for 840 mm (33 in.) before mating halves of each fracture are shown with features of ductile and brittle fractures (Ref stopping. Although the actual crack speed was the fracture starting at the notches at the top of 109). When a line pipe fails catastrophically, 470 m/s (1550 ft/s), the straight line crack each specimen. The specimen broken at -45 failure occurs with rapid lengthening of an speed was 380 m/s (1250 ft/s). °C (-50 °F) has the best definition of the initially small crack in these large, continuously Figure 46 shows the result of testing a pipe at chevron pattern. Testing at higher temperatures welded structures. To prevent such failures, a 22 °C (40 °F) below the 50% shear area DWTT produced shear fractures, but testing at -70 °C weldable low-strength steel that resists crack transition temperature; the crack speed was 675 (-96 °F) produced a nearly flat brittle fracture propagation is needed. The potential for a m/s (2215 ft/s) (maximum straight line crack with only a hint of a chevron pattern. This same catastrophic failure in a line pipe depends on speed was 535 m/s, or 1760 if/s). From the trend is also demonstrated in Fig. 42, which the toughness of the material (both transition notch, the fracture traveled in a wave pattern shows sections of line pipe steel tested at four temperature and upper shelf energy), the mag- for a full wave by cleavage with less than 10% temperatures from -3 to 40 °C (27 to 104 °F). nitude of the operating stresses, pipe dimen- shear areas present and then changed to full Visual Examination and Light Microscopy/ 109

Tested at -196 °C (-321 °F) Tested at -120 °C (-184 °F) Tested at -80 °C (-112 °F) Tested at 40 °C (104 °F) Test temperature, °F - 320 - 240 160 80 0 80 140 i i i i i Test temperature, °F 320 - 240 160 - 80 0 80 90 I0 11 f I I I i i 120 10 0.40 80 2O 100 9 70 3o E 8 Fibrous • E 0.30 = 60 I 80 40 ~_ 7 ' "----"~'-, I m o~ 50 6 ~o I 60 ff 5 / 0.20 o a~ 40 a0 o 4 Shear lip c ~ 30 40 70 x~ L~ hZ 3 0.10 20 80 2 ) 20 10 Fibrousness 90 1 \ Radial 0 0 0 of fracture 100 0 200 -160 -120 -80 40 0 40 - 200 160 120 - 80 - 40 0 40 (a) (b) Test temperature, °C Test temperature, °C Fig. 34 Transition curves for fracture appearance and impact energy versus test temperature for specimens of 4340 steel. Light fractographs at top show impact specimens tested at various temperatures Linear measurements were made para lel to the notch for the shear-lip zones, perpendicular to the notch far the fibrous zone, and perpendicular to the notch for the radial zone. The measurements yielded the three curves shown in (a). The curve of percentage of fibrousness of fracture (b) was constructed from visual estimates of the fibrous-plus-shear-lip zones. This curve, together with the impact energy curve shown in (b), shows that the transition temperature for fracture appearance is essentially the same as for impact energy. Source: Ref 63 shear and tore circumferentially for 480 mm (19 speed in such cases is quite high, the vibration 47 shows an additional example of a cleavage in.) before stopping. frequency generating the wave pattern must be fracture in a low-carbon steel with a ferrite- The formation of sinusoidal fracture paths substantial. pearlite microstructure using a light microscope in line pipes has been analyzed (Ref 110). Microscopically, brittle fractures have the cross section. The orientation of the cleavage In a homogeneous (or reasonably homoge- following characteristics: planes can be seen to vary in the different ferrite neous) material, the direction of crack propa- grains. gation is perpendicular to the plane of • Transgranular cleavage or quasi-cleavage Brittle fractures can also occur with an in- maximum stress. Therefore, if the fracture path • Intergranular separation tergranular fracture pattern. Examples of light changes direction, it must do so as a result of • Features on transgranular facets, such as microscope and SEM views of intergranular the imposition of additional stresses on the river marks, herringbone patterns, or brittle fractures are shown in Fig. 6 and 9. In circumferential stresses in the pipe due to the tongues general, it is rare for a fracture to be completely internal pressure of the contained gas or fluid. intergranular. Rather, a small amount of trans- The sinusoidal fracture paths, as illustrated in Engineering alloys are polycrystalline with ran- granular fracture can also be found, and it is Fig. 45 and 46, occur only in steels with domly oriented grains (or some form of pre- also possible, in certain cases, for a small substantial ductility. Also, the crack length ferred texture). Because cleavage occurs along amount of microvoid coalescence to accom- must be considerable, as in these examples. As well-defined crystallographic planes within pany intergranular fractures, although this is the crack lengthens, the gases or fluids each grain, a cleavage fracture will change not commonly observed. escaping from the rupture cause the free directions when it crosses grain or subgrain In one study, intergranular fractures were surfaces to vibrate. This produces a tearing boundaries. Engineering materials contain observed in Fe-0.005C material (coarse-grain) force normal to the crack propagation direc- second-phase constituents; therefore, true fea- that had been reheated to 705 °C (1300 °F), tion. As the crack grows, the vibrations tureless cleavage is difficult to obtain, even quenched, and tested at -195 °C (-320 °F) produce maximum tensile stresses alternating within a single grain. Examples of cleavage (Ref 104). The intergranular fracture may have between the outside and inside of the pipe. If fractures, as examined with the light micro- been due to strengthening of the grain bound- the tearing stresses are sufficiently large, the scope and SEM, are shown in Fig. 1 to 5, 12, aries by equilibrium carbon segregation, or they crack can continue to tear circumferentially and and 18, while light microscope views on cross may have been embrittled by the segregation of halt further crack growth. Because the crack sections are shown in Fig. 12 and 21(a). Figure some other element, such as oxygen. Inter- 1 10 / Visual Examination and Light Microscopy

Test temperature Room temperature -18 *C (0 *F) -73 °C (-100 *F)

Fig. 35 Macrograph illustrating the influence of specimen orientation (with respect to the hot-working direction) on splitting observed in Charpy V-notch impact specimens. A, notch parallel to the plate surface; B, notch perpendicular to the plate surface; C, notch 45 ° to the plate surface. Source: Ref 83

Test temperature, °F - 300 200 - 10~ 0 100 200 240 , ' I ' I ' ' ' I I 160 Hot rolled 200 • (finish 960 °C) ¢"J | ---~

160 I ~7~ 07~ :C 120 "~

"~° 120 / 80 a~ oc o BO / J= < 4O ~

0 0 200 150 - 100 - 50 0 50 100 Test temperature, °C Fig. 36 Plot of absorbed-energy Charpy V-notch test data for Fe-lMn steels finished at different temperatures (indicated on graph). Source: Ref 83 granular fractures have also been observed at -196 °C (- 321 °F) in wet hydrogen decarburized low-carbon steel. If such samples are recarburized, the embrittlement disappears (Ref 111). Intergranular fractures can be separated into three categories: those in which a brittle second-phase film in the grain boundaries causes separation; those in which no visible film is present, with embrittlement occurring because of segregation of impurity atoms in the grain boundaries; and those caused by a particular environment, as in SCC, in which neither grain-boundary films nor segregates are present. In the case of brittle grain-boundary films, it is not necessary for the film to cover Fig. 37 Macrographs showing Charpy V-notch impact specimens from Fe-lMn steels finish-rolled at four the grain boundaries completely; discontinuous different temperatures and tested at three different temperatures. Top row, finished at 960 °C films are sufficient. Some common examples of (1760 °F); second row, finished at 705 °C (1300 °F); third row, finished at 540 °C (1000 °F); bottom row, finished intergranular embrittlement by films or seg- at 315 °C (600 °F). Source: Ref 83 regants include:

• Grain-boundary carbide films in steels • Iron nitride grain-boundary films in nitrided • Grain-boundary carbide precipitation in aus- Intergranular fractures are generally quite steels tenitic stainless steels (sensitization) smooth and flat unless the grain size is rather • Temper embrittlement of alloy steels by • Embrittlement of molybdenum by oxygen, coarse. In the latter case, a rock-candy fracture segregation of phosphorus, antimony, ar- nitrogen, or carbon appearance is observed. In a classic experi- senic, or tin • Embrittlement of copper by antimony ment, the microhardness was shown to increase Visual Examination and Light Microscopy / 1 1 1

going from the center of the grains toward the component may eventually fail by a process cycling, the crack will grow in length in a grain boundaries (Ref 112). Grain-boundary known as fatigue. Although fatigue fractures direction perpendicular to the applied tensile strengthening is characteristic of intergranular are best known/in metal components, other stress. After the crack has progressed a certain fractures caused by embrittlement. materials, such as polymers, can also fail by distance, the remaining cross section can no fatigue. longer support the loads, and final rupture Fatigue Fractures With cyclic loading at tensile stresses below occurs. In general, fatigue failures proceed as If a component is subjected to cyclic loading the yield strength, a crack will begin to form at follows: involving tensile stresses below the statically the region of greatest stress concentration after • Cyclic plastic deformation before crack ini- determined yield strength of the material, the some critical number of cycles. With continued tiation • Initiation of microcrack(s) • Propagation of microcrack(s) (Stage I) • Propagation of macrocrack(s) (Stage II) • Final rupture (overload) Macroscopically, fatigue fractures exhibit many of the same features as brittle fractures in that they are fiat and perpendicular to the stress axis with the absence of necking. However, the fracture features are quite different; part of the fracture face is cyclically grown, but the re- mainder occurs by overloading, that is, one- step fracture. The apparent ductility or brittleness of the overload portion varies, depending on the strength, ductility, and toughness of the material, as well as the temperature and environment. The most distinct characteristic of fatigue failures in the field are the beach or clam shell markings on the cyclically grown portion of the fracture. It should be mentioned that similar marks on fractures can be produced under certain conditions by other fracture mechanisms that involve cyclic crack growth without cyclic (a) i loading. Also, such marks may not be visible on all materials that fail by fatigue; for exam- Fig. 38 The microstructure of Fe-lMn steel finish rolled at 315 °C (600 °F) and impact tested at two ple, many cast irons do not develop beach temperatures. A, tested at -18 °C (0 °F); B, tested at -135 °C (-210 °F). Source: Ref 83 marks. Laboratory fatigue test specimens also

Fig. 39 Classic appearance of chevrons on a brittle fracture of a steel railroad rail. The fracture origin is not on the section shown. 1 1 2 / Visual Examination and Light Microscopy do not exhibit beach marks, regardless of the / material, unless the test is deliberately con- /%e trolled to do so, for example, by using load blocks at widely varying loads. Beach marks '11//I//////~ I document the position of the crack front at various arrest points during its growth and can // /*" // /~ //" // .// //" .~k reflect changes in loading that either retard or accentuate crack growth plus the influence of £;_/_~_ "_ i__ J_-_2~_/5_ the environment on the fracture face. In a " X ~ .. x . x. \ laboratory test conducted at constant cyclic loading in a dry environment, there is no opportunity for beach mark formation. \ k ,, \ \/ The presence of beach marks is fortuitous, at least for the investigator, because beach marks Direction of propagation Direction of propagation permit the origin to be easily determined and provide the analyst with other information con- ceming the manner of loading, the relative (a) (b) Stress magnitude of the stresses, and the importance .~ + + + + ~++ + ++ ++ ++ +~ + of stress concentration. Guides for interpreting fatigue fracture markings have been shown End of End of crack at schematically (Ref 113, 114) and have been Starting visible crack mid-thickness discussed by others (Ref 63, 115-122). The most comprehensive schematics for interpreting L/point_ __ at surface ~I ___ _4 / Internalcrack fatigue fractures are shown in Fig. 48 for round } Direction of propagation [ I cross sections and in Fig. 49 for rectangular

cross sections. Each consists of examples of , i high and low nominal stress with three degrees I ' i of stress concentration and five types of load- Stress [ I Internal ing: tension-tension or tension-compression, Chevron Main ~ crack unidirectional bending, reversed bending, rotat- Border vI curves front ! I; I forming ing bending, and torsion. .... ~-" - -~...... "-T~-~/--~ In most cases, beach marks are concave to ~"2,72-/ nV.~/, zT- ~-;,~' /72~ //7,. • t the failure origin. However, the notch sensitiv- -- I.------i-~...... ~- -.x-- ~x~ T~------( ., -T-thickness ity of the material and the residual stress pat-

terns can influence crack propagation to pro- ..... -- X, duce beach marks in notch-sensitive materials F(at E D C B A that are convex to the origin (Ref 117), al- portion J -JJ / L though such cases are rare. It is also possible for a fatigue fracture to exhibit a pattern of /- .I- /~ /~ / / i x x fanlike marks similar to the radial fracture Borders ~ j- /- / / X i / i,~'t" " // t Lines of \ markings present on chevron fractures. In this case, many small fatigue cracks have joined together at shear steps (Ref 123). This is an example of multiple fatigue cracks initiating from a common location. Fatigue cracks can be initiated at a wide [ SectionE [ [ SectionD SectionC [ SectionB [ r SectionA \] variety of features, such as scratches, abrupt I I I I changes in cross section, tool marks, corrosion ] ComPletee [[[ She!rinaS earing CriticalI | Crack, // Interna,, crack / separation of borders crack width spreading forming pits, inclusions, precipitates, identification marks, and weld configuration defects. In some (c) cases, microcracks may be present before loading begins--for example, grinding cracks, ''~1~=='~ "grvAl't Schematicsshowing a model for chevron crack formation. (o) Crock front propagation model. (b) quench cracks, or hot or cold cracks from Crack front at any given time. (c) Analogy between chevron formation in plate and cup-and-cone fracture. Source: Ref 108 welding. All these problems increase the like- lihood of early failure by fatigue, assuming the presence of alternating stresses of suffi- Another macroscopic feature observed on small compared to the overall cross section; this cient magnitude. The absence of surface stress certain fatigue fractures, particularly shafts and indicates that the loads were rather low. Com- raisers, smoothly polished surfaces, and a very leaf springs, is called ratchet marks. These parison of Fig. 50 to the schematics in Fig. 48 low inclusion content will not prevent fatigue marks are seen where there are multiple fatigue suggests that loading was tension-tension or if the alternating stresses are of sufficient crack origins that grow and link up. The ratchet tension-compression with a mild stress concen- magnitude and are applied long enough, but marks are the steplike junctions between adja- tration at low nominal applied stresses. these factors are all desirable features for cent fatigue cracks. Figure 51 shows the fracture of a drive shaft long life. In the macroscopic examination of Figure 50 shows a bolt that failed by fatigue. that failed by fatigue. The fracture started at the fatigue fractures, the analyst must carefully The arrow indicates the origin, which was the end of the keyway (B), which was not filletted; examine the surface at and near the origin for root of the first thread, the location of greatest that is, the corners were sharp. Final rupture any contributing factors, such as those men- stress concentration. The overload portion of occurred at C, a very small zone. The slight tioned above. the fracture (side opposite the arrow) is rather curvature of the beach marks becomes substan- Visual Examination and Light Microscopy / 1 13

Fig. 41 Mating drop-weight tear test fractures in ship steel showing the influence of test temperature on fracture appearance. Note that chevrons are most clearly developed at -45 °C (-50 °F). The fractures were located by the notch at the top of each specimen. tial near the final rupture zone. Comparison of of the uniformity of the load cycles, the result- circular spall and the line spall. Circular spalls Fig. 51 to the schematics shown in Fig. 48 ant beach marks are extremely uniform in exhibit subsurface fatigue marks in a circular, indicates that the failure occurred by rotating- appearance. The pattern increases in spacing semicircular, or elliptical pattern. They are bending stresses, which would be expected in a and clarity as the crack grows because, as the generally confined to a particular body area. A drive shaft, with the direction of rotation clock- crack grows, the stresses are increasing even line spall has a narrow width of subsurface wise looking at the fracture and with a mild though the applied loads are the same. The two fatigue that extends circumferentially around stress concentration and low nominal stresses. macrographs (b and c) of the regions shown in the body of the roll. Most line spalls originate at Figure 52 shows two fractures in AISI 9310 (a) illustrate the progression of the fracture at or beneath the surface in the outer hardened coupling pins that failed by fatigue. Both were the lower loads, that is, for the very fine marks zone. subjected to reversed-bending alternating in (a) rather than the high-load coarse marks. Figure 55 shows a classic example of a stresses. These pins actually failed by low- Although many fatigue failures begin at a circular spall that formed beneath the surface in cycle fatigue (< 10 000 cycles). High nominal free surface, some important types of fatigue a hardened steel roll. The very clear develop- stresses were applied, and stress concentration failures begin at both surface and subsurface ment of the fatigue marks out to the overload was either absent or mild. Multiple origins are origins. These all involve rolling-contact fa- zone is evident. Radial marks can be observed frequent (arrows). tigue loading and occur in forged hardened steel all around the overload portion starting from the Figure 53 shows a torsion failure of a railroad rolls (Ref 125-130), bearings (Ref 131-135), last fatigue mark. spring that began at a small fatigue crack (right railroad rails (Ref 136-139), and wheels, for Figure 56 shows a circular spall that formed arrow) that was initiated at an abraded region example, crane wheels (Ref 140, 141) and at the body shoulder of a forged hardened steel (left arrow) at the top of the spring. The railroad wheels. In each of these products, roll. Two small thumbnail-shaped fatigue abrasion probably resulted from rubbing of the fatigue failures (spalling) are common modes of cracks (see enlarged view) served as the origin. coil spring when it was fully compressed. In failure. These two origins were connected by subsur- this case, the fatigue crack is small, and the Spalling is the primary cause of premature face cracks (not shown). The overall view of overload portion is quite large, suggesting high failures of forged hardened steel rolls. Spalls the spall shows that the region around the two nominal stresses. are sections that have broken from the surface origins is relatively featureless. However, a Figure 54 shows the fatigue fracture of a of the roll. In nearly all cases, they are observed faint ring around the two origins suggests that plate of DrAC high-strength steel. The part was in the outer hardened zone of the body surface, the crack propagated outward from the two fractured in the laboratory under block loading, and they generally exhibit well-defined fatigue origins, stopped, and then fractured by over- which produced a beach mark pattern. Because beach marks. The most common spalls are the load. 1 14 / Visual Examination and Light Microscopy

Figure 57 shows another circular spall--the Figure 64 shows another example of a line mating fracture on the roll body rather than the spall on a hardened steel roll body. In this case, spalled portion. The well-defined fatigue origin the spall is rather small and did not progress far (arrow) and beach marks are visible. The before the section spalled off. The enlarged overload portion of the fracture exhibits much view shows a well-defined fatigue origin (ar- more detail than the previous failure. The row). The ridge marks emanating from the overload portion exhibits a high density of fatigue area clearly reveal the crack growth Tested at -3 °C (27 °F) 0.96x ridge marks, and it appears to have grown direction in the overload portion. cyclically in that there are a number of arrest Figure 65 shows another example of a line points in its growth. spall fracture from a hardened steel roll. Fine Figure 58 shows a circular spall from a fatigue marks can be seen in the finger-shaped forged hardened steel roll with well-defined portion of the line spall fracture at the bottom of fatigue origins (areas A and B) and other the macrograph. The marks indicate that the regions, such as areas C and D, away from the crack growth direction was from the arrow at origin that exhibit beach marks. Enlargements the bottom of the picture and toward the top. of areas A to D are shown in Fig. 59 and 60. It After the linear portion in the center of the spall Tested at 16 °C (60 °F) 0.92 x is clear that the spall originated from the mul- formed, final rupture occurred from the linear tiple fatigue origins in the light-colored area (A portion outward, as indicated by the ridge and B) near the center. However, it appears that marks in the overload zone and their origin at other fatigue origins, such as at areas C and D, the boundary of the linear portion of the spall. were also growing when the overload fracture Figure 66 shows the outer roll surface of the occurred. spall, that is, the reverse side shown in Fig. 65. Figures 61 and 62 show two views of a line Macroetching of the surface revealed two inter- spall fracture in a 545-mm (211/2-in.) diam secting craze crack patterns indicative of abu- forged hardened steel roll. The term line spall sive service conditions. Tested at 27 °C (80 °F) 1.16x comes from the linear shape of the fatigue Most spalls are caused by local overloading, portion of the fracture that initiated rupture. In that is, surface or near-surface damage due to a line spall, the linear fatigue portion often runs either mechanical or thermal abuse in service. part way around the periphery; the overload A common cause of roll failures is the failure to portion and the overall shape of the spalled remove completely the damage by grinding section need not be linear. The arrows in Fig. from previous abusive service experience. Op- 61 and 62 indicate the portion of the line spall tical examination of replicas of roll surfaces has that extended under the unbroken portion of the been implemented to study spall nucleation Tested at 40 °C (104 °F) 1.25x roll body. This area was opened and is shown in (Ref 142). Fig. 63, along with an enlarged view of the The literature regarding the microstructural Fig. 42 Transition of fracture appearance with fatigue marks at the origin of the line spall. aspects of spalling fatigue in bearings is exten- change in test temperature. Light frac- Figures 61 and 62 show that the spalled portion sive and relies heavily on light microscopy and tographs show four specimens cut from a single length of low-carbon steel pipe that were burst by hydraulic broke off as two pieces. The radial marks electron microscopy (Ref 143-156). Spalling pressure at the temperatures indicated. Note the in- around the two spalled regions (Fig. 61 and 62) failures in bearings are caused by factors dif- creasing size of the shear-lip zone as test temperature show how the overload portion of the fracture ferent from those in hardened steel rolls, be- increases. formed around the line spall. cause the service conditions are much different.

TEST TEMPi +4B°F CRACK SPEED

Fig. 43 Fracture of API grade X-60 line pipe tested 5 °C (8 °F) above the 50% shear area drop-weight tear test transition temperature. See also Fig. 44 to 46 and text for details. Visual Examination and Light Microscopy / 1 1 5

Fig° 44 Fracture of API grade X-60 line pipe tested 1 °C (2 °F) below the 50% shear area drop-weight tear test transition temperature. See also Fig. 43, 45, and 46.

Although spalls in rolls can occur after a quite Detailed microstructural examinations were are not always preceded by butterfly formation limited service life, spalls in bearings generally conducted on these white-etching regions in (Ref 148). arise after considerable service time. The most 52100 bearing steel (Ref 146). These areas The wheel spinning that occurs during the common origin for spalls in bearings is oxide formed in the subsurface region at a depth of start up of movement by a locomotive can inclusions. Hard carbides, nitrides, and about 0.3 mm (0.01 in.) and were most prom- generate frictional heat and stresses in rails carbonitrides are other possible sources for inent at a depth of 0.47 to 0.55 mm (0.019 to when traction is poor. These conditions can initiation. The initiation sources for contact 0.22 in.). The number and size of these areas lead to spalling, or shelling, on the rail head fatigue in bearings, in addition to inclusions, increased with bearing service life, while thin surface. are geometric stress concentrations, point sur- lenticular carbides formed at the edges of these An example of such a problem is shown in face origins, peeling (general superficial pit- areas and thickened with time. Temper carbides Fig. 68. The particular steel rail was a 67-kg/m ting), and subcase fatigue (carburized compo- were not observed in these white-etching areas, (136-1b/yd) rail with 0.74% C and 0.90% Mn. nents) (Ref 148). and the amount of proeutectoid carbide de- Top and side views of the spalled regions are Although loaded in compression, bearings creased. Tempering of specimens containing shown. X-ray diffraction analysis of the spalled will fail by rolling-contact fatigue after long white-etching zones between 450 and 750 °C area revealed the presence of iron oxides life even when the inclusion content is (840 and 1380 °F) caused precipitation of car- (Fe203 and Fe304) and sand (sand is sometimes extremely low. The high applied loads alter bides in these zones. Electron microscopy (thin used to improve traction). Spinning locomotive the surface microstructure, generally by local- foil) examination of white-etching carbide-free wheels can generate enough frictional heat to ized strain hardening around a hard oxide areas revealed a fine cellular structure with a reaustenitize the surface of a rail head. The inclusion, producing a white-etching region cell size of about 0.1 i~m. mass of the rail acts as a heat sink to provide referred to as a butterfly. Figure 67 shows an Another investigation showed that the lenti- cooling rates high enough to form as-quenched example of such a white-etching feature in a cular carbides bordering white-etching areas in martensite. Microhardness testing of the white- hardened steel roll. These features never form 52100 steel formed by carbon migration from etching layer found in the spalled region (Fig. around sulfide inclusions, which are softer the white-etching region (Ref 150). The white- 68) revealed a hardness of 60 HRC. The inte- and more ductile than oxides. Butterflies etching areas form by solutioning of the car- rior hardness was 28 HRC, typical of an as- nucleated at spherical primary carbides in M50 bides in these regions. Once the butterfly is rolled fully pearlitic rail. bearing steel have also been observed (Ref well developed, microcracking occurs at the The maximum depth of the white-etching 155). In some cases, lenticular carbides form at edges of the butterfly where the lenticular layer was 0.4 mm (0.0175 in.). Surface regions the interface between the butterfly and the carbides are observed. These cracks grow until away from the spall (Fig. 68) exhibited a matrix, and they can also act as nucleation they reach the surface, producing a spall (Ref heat-checked crack pattern and scorch marks. sites. The hardness of the white-etching 151). In some cases, butterflies have been The white-etching layer consisted of two zones, butterflies is considerably greater than that of observed that did not initiate at oxide inclu- the outer, featureless zone (S) and an inner zone the matrix. sions. Fatigue cracks in rolling-contact loading with a ferrite network (F). These micrographs 1 16 / Visual Examination and Light Microscopy

TEST TEMP. DWTT CRACK SPEED + 3°F +I3°F 1550 fps

Fig. 45 Fracture of API grade X-b0 line pipe tested 6 °C (10 °F) below the 50% sheer area drop-weight teor test transition temperature. ,See also Fig. 43, 44, and 46.

TEST TEMP. DWTT CRACK sPEED - 15 ° ~25 ° 22t5 fps Fig. 46 Fracture of APi grode X-60 line pipe tested 22 °C (40 °F) below the 50% shear area drop-weight tear test transition temperature. Temperature given in °F. See also Fig. 42 to 4,5. Visual Examination and Light Microscopy / 1 1 7

Most of the slip bands were easily removed by electropolishing, but some required more effort. Upon retesting, slip bands formed again at these locations; therefore, the term persistent slip bands was coined to describe these regions of intense plastic deformation. In another study, two-beam interferometry (white light) was used to measure the surface distortion associated with persistent slip bands (Ref 159). Typical bands were about 25 Ixm wide and about 0.3 txm above the surface, with a number of sharp hills and valleys with heights and depths up to about 5 Ixm. Whether a fatigue crack is initiated at a slip band, a second-phase particle, or a stress riser depends on which source is the easiest for nucleation. In a study of a quenched-and- tempered medium-carbon alloy steel, cracks were observed within and around alumina (A1203) inclusions that initiated fatigue cracks (Ref 160). If the Young's modulus of the particle was greater than that of the matrix, tensile stress concentrated in the particle and caused cracking. Cracking was not observed at 2% sulfide inclusions or cementite particles, because their Young's modulus was less than that of the matrix. Therefore, certain inclusions or Fig. 47 Cleavagecrack path in a ferritic-pearlitic low-carbon steel. Note the subsurfacecracks (arrows). One precipitates act as fatigue crack initiators, but crack has been partially filled by nickel plating. Etched with 2% nital. 1000 × others do not. Once nucleated, microcracks must grow to a size that can be detected. Precisely when a were taken in the heat-checked region near the erated Service Testing (FAST). The fracture microcrack becomes a macrocrack is a matter spall. Beneath the S and F zones is the normal face shows well-defined fatigue marks growing of definition and of the resolving power of the pearlitic rail microstructure. Tempering of the from the detail fracture. observation method. The number of micro- specimen at 540 °C (1000 °F) precipitated Fatigue fractures are usually transgranular. cracks that form is a function of the stress or spheroidized carbides in the white-etching In the absence of a stress-raiser, fatigue crack plastic-strain amplitude. At low stress, that is, layer. The distortion of the ferrite in the F zone nucleation involves slip-plane fracture due to near the endurance limit, the growth of a single near the rail head field side (outside) occurred repetitive reversal of the operative slip systems. microcrack to a macrocrack occurs; at higher before formation of the white-etching layer Studies have demonstrated that slip lines form loads, numerous microcracks form and link up, because of deformation of the microstructure by as in static loading and that some of these will producing one or more macrocracks. heavy wheel loads. This deformed structure broaden into intense bands with further cycling. Macrocrack propagation has been widely was then austenitized by the spinning locomo- Cracks will eventually be observed at these studied and shown to be a function of the stress tive wheels. Martensite subsequently formed intense slip bands. With high stress levels, the intensity range, AK, as defined by linear- with cooling. Later, spalling in the white- density of such bands is greater, and cracks elastic fracture mechanics. Macrocrack growth etching layer occurred because of subsequent begin earlier. It has been shown that the critical has been subdivided into three regions. In the train traffic. shear stress law was followed with cyclic load- initial growth stage, there is a critical stress The shell-type fracture on the rail head forms ing and that a stress exists below which crack- intensity range, AKo, required for crack on the plane of maximum residual tensile stress free slip bands could be formed (Ref 157). growth. Once this threshold value has been and may be aided by the presence of large A surface roughening peculiar to fatigue exceeded, the crack growth rate da/dN in- inclusions or clusters of inclusions at this loca- loading has been observed after cyclic loading. creases rapidly with increasing AK until a tion. Another type of rail fracture, called a Thin slivers of metal are seen to protrude from steady-state condition is obtained. In this detail fracture, can form as a perturbation from the surface at some of the dense, cyclically second stage of crack growth, the Paris relation the shell crack under cyclic loading. The detail formed slip bands. Close examination reveals is followed (Ref 161). Above this region, the crack is constrained and exists as an internal that there are intrusions as well as extrusions maximum stress intensity Kma x approaches the flaw during the early stages of its growth and that fatigue cracks begin from the intru- critical stress intensity for fracture Kc, and final because it is impeded at the gage side (inside sions and propagate into the slip band. This fracture occurs. edge) of the rail and at the rail head by phenomenon appears to be a general character- Considerable use has been made of special longitudinal compressive stresses. The detail istic of fatigue crack initiation. Indeed, studies straining stages for in situ direct observation of crack forms perpendicular to the shell fracture have demonstrated that fatigue life can be fatigue crack propagation within the SEM and is connected to the shell fracture (Fig. 69). dramatically improved if the metal surface is chamber (Ref 162-166). Electron channeling Such defects are relatively rare; only about periodically removed (only a minor amount of contrast is used to reveal the plastic-strain 3 (2% of 30 000 of all defects were detail cracks metal need be removed); however, such a distribution at the crack tip in conductive met- (Ref 137). However, they can grow with cyclic practice is not easy to apply commercially (Ref als. Such observations would be much more loading and can lead to gross fracture of the rail 158). difficult with light microscopy. (Fig. 70). Figure 70 shows the fracture face of An examination of the polished surfaces of Microscopic examination of fatigue fracture a rail with a detail crack that was used under cyclically deformed copper and nickel speci- surfaces began with a study by Zapffe and controlled conditions at the Facility for Accel- mens revealed slip-band formation (Ref 158). Worden using light microscopy (Ref 167). The 1 1 8 / Visual Examination and Light Microscopy

( High nominal stress ~ r Low nominal stress -~

No stress Mild stress Severe stress No stress Mild stress Severe stress F concentration ~' (" concentration "~ ( concentration--'~ (~concentration "~ (~concentretion ~ (~ concentration ~

Tension-tension or tension-compression )

' Unidirectional bending --'

' Reversed bending '

Torsion J Fast-fracture zone ~ Stress-concentration notch Fig. 48 Schematic representation of fatigue fracture surface marks produced on smooth and notched components with circular cross sections under various loading conditions

restricted depth of field of the light microscope lower stress levels, however, rather high mag- tures (Ref 168, 169), but others have detected limited such examinations until development of nifications (1000 to 20 000 ×, for example) are them (Ref 170, 171). replication procedures using TEM. With the required to resolve these fine marks. Programmed fatigue loading of fcc metals, subsequent development of the scanning elec- Striations have been observed on the surfaces such as aluminum, copper, and austenitic stain- tron microscope, observation of fatigue frac- of many fatigue fractures of metals and poly- less steel, has demonstrated that each striation tures has become simpler. Many fractogra- mers. They are more easily observed on the represents crack extension from each load cy- phers, however, still prefer to examine fatigue surfaces of more ductile metals, especially fcc cle. In service conditions, loads are variable in fractures with replicas because of the excellent metals, than on steels. Striations on low- magnitude, and not all will be of sufficient image contrast. strength steels tend to be wavy in nature and magnitude to cause crack propagation. Also, The most distinguishing microscopic feature exist in patches rather than over the entire after an abrupt change from a high load level to of a fatigue fracture is its striated surface surface. Fatigue striations may be quite difficult a low load level, there will be a brief period appearance. If these striations are coarse to observe on high-strength steel fatigue frac- during which the crack does not grow. enough, as in low-cycle fatigue fractures (high tures, depending on the manner of loading, the The fracture surface during microcrack for- stress levels), they can be observed by light environment, and so on. Some researchers have mation (Stage I) is generally featureless, al- microscopy as demonstrated in Ref 167. At been unable to observe striations on such frac- though rub marks may be observed. The tran- Visual Examination and Light Microscopy / 1 19

High nominal stress r Low nominal stress

No stress Mild stress Severe stress ~concentration No stress Mild stress Severe stress (--concentration r concentration "~ (--concentration "~ r concentration ~ (~ concentration "~

..g.

\ i i..g. i Tension-tension or tension-compression

* *

i i i Unidirectional bending

t Reversed bending J Fast-fracture zone Stress*concentration notch ~Asterisk indicates crack origin at a corner. Corner initiation isfavored bythe presenceof machining burrs.

Fig. 49 Schematic representation of fatigue fracture surface marks produced in square and rectangular components and in thick plates under various loading conditions

sition from microcrack formation (Stage 1) to Although Stage I microcracks are highly (Ref 115, 168, 174, 175), slip traces (Ref 115, macrocrack propagation (Stage 1I) often occurs sensitive to crystallographic and microstruc- 168, 176), fractured pearlite lamellae (Ref 41, at a grain boundary or a triple point. Fatigue tural features, Stage II macrocrack formation is 115, 168), and rub marks (for example, tire- striations are observed during Stage II crack less dependent on these features. Second-phase tracks) (Ref 41, 115, 168, 175). Striationlike growth, but their absence does not prove that particles, such as inclusions, have a complex marks have also been observed on certain the failure was not due to fatigue. Also, other influence on striations and on the crack growth metals not tested under cyclic loading condi- fracture features resembling striations have rate. The mechanism by which inclusions can tions. These problems are discussed in the been mistakenly interpreted as striations. either inhibit or promote crack propagation has articles "Modes of Fracture" and "Scanning Two types of fatigue striations, referred to as been schematically illustrated (Ref 173). Dif- Electron Microscopy" in this Volume. Fatigue ductile or brittle striations, have been classified ferent inclusion types and morphologies have striations are not observed at all crack propaga- (Ref 172). Ductile striations are the type most been shown to influence the fatigue crack tion rates (Ref 177). At very high crack prop- commonly observed on fatigue fractures, while growth rate by using different plate steels tested agation rates (high stress intensities), the fa- brittle striations have been observed on fatigue with six possible orientations (Ref 174). tigue fracture exhibits microvoid coalescence or fractures in corrosion fatigue failures and in Fracture features that have been mistakenly cleavage, depending on the alloy. As the fa- hydrogen-embrittled steels. identified as striations include Wallner lines tigue crack propagation rate decreases, stria- 120 / Visual Examination and Light Microscopy

Fig° 50 Fatigue failure of a bolt due to unidirectional cycling bending loads. The failure started at the thread root (arrow) and progressed across most of the cross section before final fast fracture. Actual size

tions are observed. At very low crack growth rates (for example, 2 to 20 × l0 -6 ram, or 0.07 to 0.7 p~in., per cycle), the fracture exhibits an appearance similar to cleavage, even in fcc metals. Facetlike fatigue fractures were Fig° 5 1 Macrograph of the fracture face of an AIS14320 drive shaft that failed by fatigue. The failure began observed in aluminum, copper, titanium, and at the end of a keyway that was machined without fillets (B) and progressed to final rupture at (C). The austenitic stainless steels at crack growth rates final rupture zone is small, indicating that loads were low. below about 10 -2 mm, or 0.3 p.in., per cycle (Ref 177). Testing of specimens in a vacuum has revealed conflicting results; some observers have found striations, but others have not. Fatigue testing at high temperatures has revealed many fractures that do not exhibit striations. Many studies have demonstrated that the striation spacing is a function of the applied stress; that is, the spacing increases with load and crack length. In most cases, light micros~ copy is unsuitable for such measurements, and electron metallographic techniques are required (see the articles "Scanning Electron Micros~ copy" and "Transmission Electron Micros- copy" in this Volume). Laboratory studies have shown that the crack growth rate determined from striation spacings is within a factor Fig. 52 Two examples of fatigue fractures in AISI 9310 quenched-and-tempered coupling pins caused by of two of the macroscopic growth rate (Ref reversed cyclic bending loads. Arrows indicate the fracture origins. Actual size 178-183). In many cases, at low crack growth rates, the macroscopic growth rate was less the specimen surfaces, the time to initiation and roughnesses on the fracture surface. A color than the microscopic growth rate; at high crack the crack growth rate measured macroscopi- chart correlating known crack propagation rates growth rates, the macroscopic rate exceeds the cally or from striation spacings will usually be with known stresses has been prepared (Ref microscopic growth rate. These studies have somewhat different. A technique has been de- 186). Very low crack propagation rates produce shown that there is a relationship between the veloped for measuring the crack propagation black coloration, intermediate rates produce striation spacing and the stress intensity factor rate based on the observation that fracture color gray colors, and high rates produce uniform range AK (Ref 182). and roughness vary with the crack propagation coloration. Such a procedure is qualitative at Macroscopic crack growth rates are com- rate (Ref 184, 185). The technique is usually best and is limited in application, especially for monly measured using an optical telescope applied to test samples using programmed fa- service failures. focused on the growing crack front. Because tigue loading with regularly stepped variable The crack propagation rate and stress inten- the crack front is usually not perpendicular to loads. The steps appear in different colors and sity can be estimated through measurement of Visual Examination and Light Microscopy/121

fracture mechanisms, electron metallographic procedures complement these techniques and are necessary tools. The simplest measure of high-temperature properties is provided by the use of short-time tensile tests. Hot hardness tests are also quite useful. Figure 74 shows fractures of short-term high-temperature tensile specimens of type 316 stainless steel tested at 760 to 980 °C (1400 to 1800 °F). The reduction of area of the tensile fractures shown increases with temperature. Cup-and-cone fractures are not observed, however. The microstructures shown illustrate the formation of voids at the grain boundaries, which is typical of high-temperature tests. Pre- cipitation of grain-boundary carbides (sensitization) can be clearly seen in the high-magnification view of the specimen tested at 815 °C (1500 °F). This is an example of a change in microstructure caused by high-temperature exposure. Because metals expand with high-temperature exposure, creep tests are widely performed on alloys used at high temperature to assess the Fig. 53 Torsion failure of an AISI 51B60 railroad spring. The failure began by fatigue at the abraded area at the top (arrows). magnitude of the length change as a function of load and temperature. The creep rate will increase with load and temperature. Tests con- the striation spacing. Obviously, this is limited specimen was completely broken, and only one ducted at relatively high loads until fracture to those fatigue fractures exhibiting striations. side of the fracture can be observed. Compari- occurs are referred to as stress rupture tests. In !If the crack propagates partly by another son of Fig. 71(a) and (b) clearly shows that the this test, the time to rupture at a given load and ,"fracture mechanism, the microscopically deter- nature of the crack pattern is much easier to temperature is assessed, rather than the creep mined growth rate will be lower than the observe in a partially broken specimen where rate. Creep tests are conducted at relatively low macroscopically determined rate. The striation both sides of the fracture can be observed. loads at which fracture may not occur for many measurement can provide an estimate of the Figure 72 shows a fatigue crack in a years. number of stress cycles strong enough to grow ferrite-pearlite carbon steel after etching in The fracture path for metals and alloys the crack during Stage II, but can provide no which both sides of the crack can be observed. tested at high temperatures changes from trans- information about the number of cycles The irregular path of the crack through the granular to intergranular with increasing tem- required to initiate and propagate a microcrack. ferrite phase is evident in Fig. 72 as compared perature (Ref 191). Transgranular fracture oc- The problem of fatigue striation measure- to the nature of cleavage cracks in ferrite curs at low temperatures, at which the slip ment is discussed in the article "Quantitative shown in Fig. 47. planes are weaker than the grain boundaries. At Fractography" in this Volume and in Ref 187. As a final note on the microscopic aspects of high temperatures, the grain boundaries are The finest reported striation spacing was about fatigue, the use of DIC illumination to study the weaker, and fracture is intergranular. Such 10 nm (Ref 188). The striation spacing in progression of slip during cyclic loading will be observations have led to the introduction of the contiguous locations, even for specimens sub- briefly discussed. The sensitivity of DIC for equicohesive temperature concept to define the jected to reasonably uniform loading, can vary revealing the progression of slip and micro- temperature at which the grains and grain by as much as a factor of five (Ref 189). crack formation has been demonstrated by us- boundaries exhibit equal strength; that is, the Consequently, at each location from the crack ing polished specimens of Nickel 200 examined temperature at which the fracture mode changes origin, it is important to measure a large num- after cyclic loading (Ref 190). When the Wol- from transgranular to intergranular (Ref 192, ber of striations to ensure that the average value laston prism is fully inserted to produce a dark 193). The equicohesive temperature is not a is representative. blue-green image, the surface deformation from fixed temperature, but varies with the stress and Metallographic observation of either com- slip is vividly revealed. Slip can also be ob- strain rate for a given composition. Above the pletely broken or partially broken fatigued sam- served by using bright-field illumination (Fig. equicohesive temperature, coarse-grain speci- ples is a common procedure, particularly in 73), but DIC is more sensitive, particularly in mens exhibit greater strength than fine-grain failure analysis. After macrofractographic and the early stages of slip. specimens because of the lower grain-boundary microfractographic examination of the fracture surface area. Transgranular fractures have been face has been completed, it is common practice High-Temperature Fractures observed in high-purity metals at rather high to section the specimen so that the microstruc- As service or test temperatures are increased, temperatures. ture at the fatigue origin can be examined to the strength of metals and alloys decreases Tests at temperatures above the equicohesive determine if any microstructural anomaly is while ductility increases, although various em- temperature have revealed two types of inter- present. brittlement phenomena may be encountered. At granular fracture. When grain-boundary sliding Figure 71(a) shows the same fatigue crack in elevated temperatures, the strength will vary occurs, wedge-shaped cracks may form at an aluminum alloy as shown in Fig. 16, but with strain rate and exposure time. Also, mi- grain-boundary triple points (as observed on a after etching. The roughness of the crack crostructures will change with high-temperature plane-of-polish) if the tensile stresses normal to pattern and the interaction with the intermetal- exposure. Visual and light microscope exami- the boundaries exceed the boundary cohesive lic particles are evident. Figure 71(b) shows a nation are very useful for examining high- strength. In the literature, these cracks are similar fatigue crack in the same aluminum temperature fractures and for observing referred to as wedge or w-type, or as triple alloy after etching; but in this case, the changes in microstructure. As with the other point or grain-coruer, cracks. High stresses 122 / Visual Examination and Light Microscopy

either parallel or perpendicular to the stress axis Interferometry is used to measure the vertical displacement produced during sliding; alternatively, profile measurements can be made on cross sections The change in grain shape of the grains in the specimen interior is statistically determined by using quantitative metallographic procedures

The displacement measurements are used to calculate the contribution of grain-boundary sliding to the overall extension during creep. Cavity formation, density, and orientations have also been quantified by using a number of procedures (Ref 212-221). These studies have demonstrated that the cavities form preferen- tially on grain boundaries oriented approximately perpendicular to the applied stress. Creep cavity size, shape, and density can be measured at various stages during creep testing by using standard quantitative metallographic methods. The average cavity diameter increases with time, temperature, and strain rate. Measurements have also been made of the angular distribution of cavitated grain boundaries with respect to the stress axis, showing that the cavitated grain boundaries are oriented between 60 and 90 ° to the stress axis. At low strain rates, the most frequently observed cavities are on boundaries perpendicular to the stress axis. With increasing strain rates, the most frequently observed cavitated grain boundaries shift toward a 45 ° angle to the stress axis. These measurements are usually made on longitudinal sections cut from creep test specimens. The angles between the cavitated grain boundary and the stress axis on the sectioning plane are measured, and the distribution is Fig. 54 Three views of a fatigue fracture in D6AC steel plate, showing beach marks. (a) Plate subjected to a series of varied loading cycles in the laboratory. The crack origin, at the bottom center, was at a statistically analyzed by using the procedure starter notch formed by electrical discharge machining. (b) Area in lower square in (a), just above beach mark discussed in Ref 221. Metallographic studies of registering the first change in loading. (c) Area in upper square in (a), just ahead of the zone of final fast fracture. high-temperature test specimens and fractures Both (b) and (c) contain faintly resolved, fine beach marks. (a) About 5 X. (b) and (c) 60 x . Source: Ref 124 have used macrophotography, optical microscopy, x-ray techniques, microradiography, TEM, and SEM, as demonstrated in the cited promote this type of crack formation. One of Qualitative observations of grain-boundary references and in Ref 222-232. the ways in which such cracks can form is sliding during high-temperature tests were first Figure 76 illustrates r-type cavities at grain shown in Fig. 75. made in 1913 (Ref 191). Since then, similar boundaries in type 316 stainless steel tested Under low-stress conditions, intergranular observations have been made for many metals under creep conditions. The SEM fractograph fractures occur by void formation at the grain under creep conditions. Quantitative measure- shows the r-type cavities on the fracture sur- boundaries (Ref 195, 196). These cavities form ments have been made using bicrystals and face, while the etched cross sections reveal the along grain edges rather than at grain corners. polycrystals with a variety of methods (Ref cavities at and behind the fracture face and their Because they appear to be round or spherical on 198-211). These measurements, primarily by relationship to the grain boundaries. Figure 77 metallographic cross sections, these voids are light microscopy, have been useful in determin- shows w-type cracks in type 316L stainless sometimes referred to as r-type cavities. Sub- ing the contribution that sliding makes in the steel tested under creep conditions. The w-type sequent studies have shown that voids of more overall creep extension and in understanding cracks and the orientation to the stress axis complex shapes, such as polyhedra, can be creep mechanisms. (vertical in the micrograph) are easily observed formed during cavitation. Inclusions and pre- Measurements of grain-boundary sliding are in the light micrograph, but are not obvious in cipitates on grain boundaries can act as sites for made on grain boundaries that intersect the the SEM fractograph. Figure 78 shows grain- void nucleation. Cavity nucleation occurs dur- specimen surface or on boundaries in the spec- boundary triple-point cracking and extensive ing the initial (primary) stage of creep (Ref imen interior, using either bicrystals or poly- deformation in type 316 stainless steel tested at 197). These cavities grow during the second crystalline specimens. Most of the procedures a slightly lower temperature. It is interesting to stage (steady-state) of creep. The third (tertiary) fall into one of the following three types: compare the three SEM fractographs in Fig. 76 stage of creep begins when these cavities grow to 78. Figure 79 shows crack nucleation in a to such an extent that their size and spacing are • Lines or grids are scribed on the specimen type 316 stainless steel specimen with heavy approximately the same as the grain size. surface, and the displacements are measured carbide precipitation. Cracking has occurred Visual Examination and Light Microscopy/ 123

boundaries (Ref 269). These grooves were not observed after either mechanical or electrolytic polishing, but were visible after etching. In another study, copper containing up to 4.68% Bi was tested, and the results were similar to those discussed in Ref 269; however, in alloys with high bismuth contents, either continuous grain-boundary films or discrete particles of bismuth with a lenticular shape were observed. Studies of the embrittlement of copper by antimony revealed results similar to that of the low-bismuth alloy (Ref 271, 272); that is, grain-boundary grooves, rather than discrete films, were observed after etching. The embrittled specimens fractured intergranularly. The influence of impurity elements on the hot workability of metals is well known. Cop- per will be embrittled during hot working in the presence of bismuth, lead, sulfur, selenium, tellurium, or antimony (Ref 273). Lead and bismuth also degrade the hot workability of brass (Ref 273). The hot workability of steels is degraded by sulfur (Ref 274-280) and by residual copper and Fig. 55 Circular spal[ that began at a large subsurfaceinclusion in a hardened steel roll tin (Ref 281-284). Sulfides have also caused intergranular cracking in alloy steel castings along the grain-boundary carbides, and the of oxygen on the intergranular brittleness of (Ref 285). Poor hot workability is also a prob- SEM fractography reveals a heavy concentra- iron has produced the most conflicting test lem with free-machining steels containing lead tion of carbides on the intergranular fracture results. For example, in one investigation a and tellurium (Ref 286). Residuals such as surface. series of iron-oxygen alloys with up to 0.27% O lead, tin, bismuth, and tellurium can cause hot Because of the economic importance of creep was tested, and intergranular fractures were cracking during hot working of stainless steels in high-temperature service, particularly in observed in all but the lowest (0.001%) oxygen (Ref 287, 288), and residual elements such as power generation equipment, considerable em- sample (Ref 251). On the other hand, in a study sulfur, phosphorus, bismuth, lead, tellurium, phasis has been placed on predicting the re- of high-purity iron and electrolytic iron, no selenium, and thallium are detrimental to maining life of components (Ref 233-238). influence of oxygen content (up to 2000 ppm) nickel-base superalloys (Ref 289-291). Exces- This work has involved metallographic exami- was observed on the ductile-to-brittle transition sive precipitation of aluminum nitride can cause nation of the creep damage, including field temperature (Ref 256). Increasing the carbon cracking in steel castings and during hot work- metallographic procedures (Ref 239-243). Such content to about 40 ppm decreased the ductile- ing (Ref 292-303). predictions must also take into consideration to-brittle transition temperature and decreased Certain materials are inherently brittle be- the changes in microstructure that occur during the intergranular brittleness, irrespective of ox- cause of their crystal structure, microstructure, the extended high-temperature exposure of met- ygen content. or both. For example, gray cast iron is an als and alloys (Ref 244-249). Other bcc metals, such as molybdenum, inherently brittle material because of the weak- chromium, and tungsten, are embrittled by ness of the nearly continuous graphite phase. lmbrittlement oxygen, nitrogen, and carbon (Ref 250, 258, However, if the graphite exists in isolated, Phenomena 259). When embrittled, the fractures of these spherical particles, as in nodular cast iron, metals are intergranular. Face-centered cubic excellent ductility can be obtained. Grain- The expected deformation and fracture pro- metals may also be embrittled by oxygen (Ref boundary cementite films in high-carbon or cesses can be altered by various embrittlement 260,261) and sulfur (Ref 262-265). For exam- carburized steels produce extreme brittleness, phenomena. These problems can arise as a ple, in a study of the grain-boundary embrittle- but if the same amount of cementite exists as result of impurity elements (gaseous, metallic, ment of intermetallics with a stoichiometric discrete spheroidized particles, ductility is or nonmetallic), temperature, irradiation, con- excess of active metal component, the extreme good. As-quenched high-carbon martensite is tact with liquids, or combinations of these or brittleness of these materials was shown to be quite brittle, but tempering improves the duc- other factors. Metals can become embrittled due to grain-boundary hardening through ab- tility, although at a sacrifice in strength. The during fabrication, heat treatment, or service. If sorption of gaseous impurities (oxygen and/or normally ductile austenitic stainless steels can the degree of embrittlement is severe enough nitrogen) segregated to the grain-boundary ar- be embrittled by the formation of hcp for the particular service conditions, premature eas (Ref 266). e-martensite during service (Ref 304-306). failures will result. Some of these problems Metallography and fractography have played Numerous types of embrittlement phenom- introduce rather distinctive features that may be important roles in developing an understanding ena can occur in certain metals and alloys or observed by macro- or microscopic fracto- of embrittlement mechanisms. For example, under certain environmental conditions. These graphic methods, and the ability to categorize early work on the embrittlement of copper by problems can be traced to compositional or these problems properly is imperative for deter- bismuth attributed the embrittlement to the manufacturing problems and/or service condi- mining cause and for selecting the proper cor- presence of thin grain-boundary films of ele- tions. The more familiar embrittlement prob- rective action. mental bismuth (Ref 267,268). However, care- lems and their fractographic characteristics are It is well recognized that many metals, such ful metallographic preparation and examination summarized below. as iron (Ref 250-257), are embrittled by high of copper containing low amounts of bismuth Creep-Rupture Embrittlement. Under levels of oxygen, nitrogen, phosphorus, sulfur, (up to 0.015%) showed that the apparent films creep conditions, embrittlement can occur and and hydrogen. Of these elements, the influence were actually steplike grooves at the gram result in abnormally low rupture ductility. This 124 / Visual Examination and Light Microscopy

Graphitization. In the early 1940s, several failures of welded joints in high-pressure steam lines occurred because of graphite formation in the region of the weld heat-affected zone (HAZ) that had been heated during welding to the critical temperature of the steel (Ref 316-320). Extensive surveys of carbon and carbon-molybdenum steel samples removed from various types of petroleum-refining equipment revealed graphite in about one-third of the 554 samples tested (Ref 316, 319). Generally, graphite formation did not occur until about 40 000 h or longer at temperatures from 455 to 595 °C (850 to 1100 °F). Aluminum-killed carbon steels were susceptible, but silicon- killed or low-aluminum killed carbon steels were immune to graphitization. The C-0.5Mo steels were more resistant to graphitization than the carbon steels, but were similarly influenced by the manner of deoxidation. Chromium additions and stress relieving at 650 °C (1200 °F) both retarded graphitization. Hydrogen Embrittlement. Hydrogen is (a} known to cause various problems in many metals, most notably in steels, aluminum, nickel, and titanium alloys (Ref 321-332). Var- ious forms of hydrogen-related problems have been observed.

• Blistering, porosity, or cracking during processing due to the lack of solubility during cooling of supersaturated material, or by cathodic charging, or other processes that form high-pressure gas bubbles • Adsorption or absorption of hydrogen at the surface of metals in a hydrogen-rich environment producing embrittlement or cracking • Embrittlement due to hydride formation • Embrittlement due to the interaction of hydrogen with impurities or alloying elements

The problem of hydrogen effects in steels has been thoroughly studied. Hydrogen embrittlement is most noticeable at low strain rates and at ambient temperatures. A unique aspect of hydrogen embrittlement is the delayed nature of the failures; that is, after a specimen is charged with hydrogen, fracture does not occur instantly (b) but only after the passage of a certain amount of Fig. 56 Circular spall from the shoulder of a forged, hardened alloy steel mill roll with two small fatigue time. Therefore, some researchers have used origins. A subsurface crack connected the two fatigue zones. Note that the spall surface is relatively the term static fatigue to describe the phenom- featureless. (a) About actual size. (b) 5.5 x enon. However, this term is misleading. Ten- sile and bend tests have historically been used to detect and quantify the degree of embrittle- problem has been encountered in aluminum ductility than fine-grain weldments (Ref 312). ment. For example, in tensile testing, it is (Ref 307) and steels (Ref 308-315). Iron, in Impurities such as phosphorus, sulfur, copper, common practice to compare the normal tensile amounts above the solubility limit in alumi- arsenic, antimony and tin have been shown to ductility--the %RA--with the %RA in the num, has been shown to cause creep-rupture reduce rupture ductility, although rupture life presence of hydrogen in order to calculate an embrittlement by development of intergranular increases. This behavior appears to be due to embrittlement index E showing the loss in cracking (Ref 307). the grain-boundary segregants blocking grain- reduction of area: The creep embrittlement of chromium- boundary diffusion, which reduces the cavity molybdenum steels has been extensively stud- growth rate. High impurity contents increase E = (%RA)u - (%RA)c (Eq 2) ied. Matrix precipitation strengthening has been the density of the cavities. Substantial inter- (%RA)u shown to cause creep embrittlement (Ref 308). granular cracking is observed in high-impurity Also, coarse-grain areas in 2.25Cr-lMo welds material and is absent in low-impurity heats where u and c indicate unchanged and have been found to exhibit much lower creep (Ref 313). changed, respectively. Visual Examination and Light Microscopy/ 125

duced cracking was shown to occur by cleavage at very high stress intensities (Ref 344). At high impurity levels (grain-boundary impurities), the fracture path was intergranular, and the stress intensity required for crack growth decreased. In another study, tempering between 350 and 450 °C (660 and 840 °F) produced entirely intergranular fractures (Ref 345). Therefore, the fracture mode will be influenced by the cooperative action of temper embrittlement and hydrogen embrittlement. A common fracture feature in hydrogen-embrittled low-strength steels is fiat fracture regions (100 to 200 Ixm in size) that are circular or elongated and centered around an inclusion or a cluster of inclusions (Ref 348). These zones are transgranular. An investigation of crack nucleation and growth in a low-carbon ferrite-pearlite steel tested in both hydrogen and oxygen environments revealed that crack growth rates were faster in hydrogen than in oxygen and were faster when the crack growth direction was in the hot-working direction (Ref 346). Crack initiation in hydrogen was not sensitive to orientation. The specimens tested in hydrogen exhibited cleavagelike fractures with a small amount of ductile microvoids. Plastic deformation occurred near the crack tip before crack growth in hydrogen, producing voids at inclusions ahead of the crack. When the crack propagated, these voids were linked to the fracture by cracking of the matrix perpendicular to the maximum normal stress. The fracture path was transgranular with no preference, or aversion, for any microstructural feature. Hydrogen also influences ductile fracture (Ref 349, 350). For example, a study of spheroidized carbon steels (0.16 and 0.79% C) found F |g. 5 7 Circular spall on the surface of a forged, hardened alloy steel mill roll. The arrow indicates the fracture no significant influence of hydrogen on the origin. Note the fatigue marks showing the growth away from the origin, followed by brittle fracture. 0,68 x initiation of voids at carbides or on the early eutectoid growth of the voids before linking (Ref 349). In the eutectoid steel, hydrogen Hydrogen flaking is a well-known problem Rail producers have found that the cooling exposure increased the dimple size and assisted in the processing of high-carbon and alloy cycles used in the past to prevent flaking are void growth during linking. In the low-carbon steels (Ref 333-339). Whether or not flaking inadequate for this purpose when inclusion steel, flat quasi-cleavagelike facets were ob- occurs depends on several factors, such as steel contents are very low. The flakes in such steels served, and the void size decreased because of composition, hydrogen content, strength, and do not exhibit the classic appearance shown in hydrogen exposure. An investigation of low- thickness. Steels prone to flaking are made Fig. 80. and medium-carbon spheroidized steels con- either by vacuum degassing to reduce hydrogen In addition to flaking, blisters can be pro- cluded that hydrogen charging promoted void to a safe level or by using controlled cooling duced by excess hydrogen (Fig. 81). Hydrogen nucleation at carbides and accelerated void cycles. High-carbon steels are particularly sus- can also be introduced into steels during other growth, particularly for carbides at grain or ceptible to flaking. Therefore, rail steel is processes, such as welding, pickling, bluing, subgrain boundaries (Ref 350). Void growth control cooled slowly after rolling to prevent enameling, or electroplating. Consequently, it acceleration was greatest in the latter stage of flakes (Ref 340). When the hydrogen content is is necessary to bake the material after such void growth. Quasi-cleavage facets were ob- not properly controlled, flakes result (Fig. 80). processes if the material is prone to hydrogen served around inclusions in steels with high Flaking can occur in a wide variety of steels. damage. inclusion contents. Low-carbon steels appear to be relatively im- The influence of hydrogen on fracture ap- The influence of inclusions, particularly sul- mune to flaking, but alloy steels, particularly pearance is complex (Ref 341-348). Studies fides, on hydrogen embrittlement has been those containing substantial nickel, chromium, have shown that cracks propagate discontinu- demonstrated (Ref 351-355). In one investiga- or molybdenum, are quite susceptible. In gen- ously, suggesting that the crack growth rate is tion of hydrogen-embrittled ultrahigh strength eral, as the strength of the material increases, controlled by the diffusion of hydrogen to the steels, for example, cleavage areas were ob- less hydrogen can be tolerated. The number of triaxially stressed region ahead of the crack tip. served around nonmetallic inclusions (Ref flakes formed has been shown to be related to The fracture appearance is influenced by the 353). Increasing the sulfur content reduced the cooling rate after hot working (Ref 338). strength of the material. As the strength level hydrogen embrittlement under certain test con- Manufacturers have observed that if the inclu- increases, fractures are more intergranular. Im- ditions. Sample size also influenced results. sion content is quite low, flaking can occur at purities also influence fracture mode. For ex- Oxides with low coefficients of thermal expan- hydrogen levels normally considered to be safe. ample, at low impurity levels, hydrogen-in- sion, such as silicates, are detrimental under 126 / Visual Examination and Light Microscopy

microvoids in the matrix beside and ahead of the crack. These often nucleate at large carbides. In the completely broken specimen, shown by SEM and with a cross section, there is little evidence for intergranular fracture in the traditional form. There are some tendencies, but the fracture surface is covered with fine dimples that form around the carbides. Liquid-metal embrittlement (LME) is a phenomenon in which the ductility or fracture stress of a solid metal is reduced by exposure of the surface to a particular liquid metal (Ref 371-376). The phenomenon was first observed in 1914 in experiments where 13-brass disinte- grated intergranularly in liquid mercury. Sepa- ration is so complete that various investigators have used this process to study the three- dimension characteristics of grains. The study of LME is complicated by the existence of at least four forms of the phenomenon: • Instantaneous fracture of a particular metal under applied or residual tensile stresses when in contact with specific liquid metals • Delayed fracture of a particular metal in contact with a specific liquid metal after a time interval at a static load below the tensile strength of the metal • Grain-boundary penetration of a particular solid metal by a specific liquid metal; stress does not appear to be required in all instances • High-temperature corrosion of a solid metal by a liquid metal causing embrittlement The first type is the classic, most common form of LME; the second type is observed in steels. Many metals and alloys are known to fail by Fig. 58 Spalled section from a forged, hardened alloy steel mill roll showing three regions with fatigue beach LME when in contact with some particular marks. The large, shiny region appears to be the origin. See Fig. 59 and 60 for close-up views of areas A, B, C, and D. 0.73 × liquid metal. If the solid metal being embrittled is not hydrogen charging conditions; sulfides, with 1500 °F), either isothermally or by slow cooling notch-sensitive, as in fcc metals, the crack will high coefficients of thermal expansion, were through this range, precipitates M23C 6 carbides propagate only when the liquid metal feeds the harmless or beneficial (Ref 354). Microcracks in the grain boundaries, thus sensitizing the crack. However, in a notch-sensitive metal, as were commonly observed at oxides but not at alloy to intergranular corrosion. Ferritic stain- in bcc metals, the crack, once nucleated, can sulfides. Another study found that the critical less steels are also susceptible (Ref 362, 363). become unstable and propagate ahead of the concentration of hydrogen for cracking in- Corrosion occurs in the matrix adjacent to the liquid metal. Crack propagation rates during creased with increasing sulfur content and that sensitized grain boundary because of depletion LME can be quite fast. Generally, the cracks embrittlement increased with increasing oxygen of chromium, as demonstrated by analytical are intergranular, although a few cases of trans- content, that is, oxide inclusions (Ref 355). electron microscopy (Ref 364-366). The granular fractures have been reported. Intergranular Corrosion. Typically, me- metallographic examination of specimens elec- Figure 84 shows a metallographic cross sec- tallic corrosion occurs uniformly; however, un- trolytically etched with a 10% aqueous oxalic tion of eutectoid rail steel embrittled by liquid der certain conditions, the attack is localized at acid solution, as defined by Practice A of copper and an SEM view of the fracture sur- the grain boundaries, producing intergranular ASTM A 262, is widely used as a screening test face. Tensile specimens were heated to 2110 °F corrosion. for sensitization (Ref 367-370). (1100 °C) and loaded at 12.5 to 50% of the High-strength precipitation-hardened alumi- Figure 82 shows two views of an austenitic normal tensile strength at this temperature. num alloys are susceptible to intergranular stainless steel exhibiting intergranular corro- Liquid copper was present at the base of a corrosion. Aluminum-copper and aluminum- sion. Figure 82(a) shows the surface of the V-notch machined into the specimen. Fracture magnesium alloys are susceptible to intergranu- sample. The grain structure is visible due to the occurred in a few seconds at 50% of the tensile lar corrosion in certain environments. Die-cast attack, and some grains have fallen out (re- strength (SEM view of this sample), and the zinc-aluminum alloys can fail by intergranular ferred to as grain dropping). The cross-sec- time to fracture increased with decreasing load. corrosion in steam or salt water. tional view (Fig. 82b) shows the depth of pene- Hot shortness in steels can be caused by Austenitic stainless steels are known to be tration of the attack along the grain boundaries. copper segregation in steels alloyed with cop- susceptible to intergranular corrosion in envi- Figure 83 shows the fracture of sensitized per. Figure 85 shows an example of hot short- ronments that are normally harmless if the type 304 austenitic stainless steel broken in a ness in a structural steel section in a copper- material has been subjected to a sensitization noncorrosive environment. The micrograph of containing grade. At the rolling temperature, treatment (Ref 356-361). Exposure to temper- the partially broken specimen shows that the the segregated, elemental copper was molten. atures in the range of 480 to 815 °C (900 to fracture is not totally intergranular. There are When the section was rolled, it broke up Visual Examination and Light Microscopy / 127

(a) lb)

Fig. 59 Close-ups of the primary initiation site of the spall shown in Fig. 58. (a) Area A. 4×. (b) Area B. 8×. See also Fig. 60. because the liquid copper wetted the austenite with grain-boundary fracture processes, partic- be some loss in tensile ductility. The cooling grain boundaries. ularly for deformation at temperature above 550 rate after overheating influences the critical Neutron irradiation of nuclear reactor °C (1020 °F). In one study, neutron irradiation overheating temperature. Low-sulfur steels are components causes a significant increase in the of annealed aluminum alloy l l00 at high flu- more susceptible to overheating than high- ductile-to-brittle transition temperature in fer- ences at about 323 K caused a large increase in sulfur steels. ritic alloys (Ref 377-385). The degree of strength, a large decrease in ductility, and Fracture tests have been widely used to irradiation-induced embrittlement depends on intergranular fracture at 478 K (Ref 385). reveal facets indicative of overheating (Ref the neutron dose, neutron spectrum, irradiation Overheating occurs when steels are heated 386). The heat treatment used after overheating temperature, steel composition, and heat treat- at excessively high temperatures prior to hot has a pronounced influence on the ability to ment. Tempered martensite is less susceptible working (Ref 386-396). Overheated steels may reveal the facets. Quench-and-temper treat- to embrittlement than tempered bainite or exhibit reduced toughness and ductility as well ments are required. It has been shown that ferrite-pearlite microstructures (Ref 379). Im- as intergranular fractures. The faceted grain facets are best revealed when the sample is purity elements in steels can influence embrit- boundaries exhibit fine ductile dimples because quenched and tempered to a hardness of 302 to tlement; for example, phosphorus levels above of reprecipitation of fine manganese sulfides at 341 HB (Ref 386). The size of the plastic zone 0.015% and copper levels above 0.05% are the austenite grain boundaries present during at the crack tip during fracturing of the testpiece detrimental. the high-temperature exposure. Heating at tem- is influenced by the yield strength of the spec- Radiation produces swelling and void forma- peratures above 1150 °C (2100 °F) causes imen, which in turn controls the nature of the tion following a power law dependent on flu- sulfides to dissolve, with the amount dissolved fracture surface in an overheated specimen (Ref ence. Void density decreases as irradiation increasing as the temperature increases above 392). Progressively higher tempering tempera- temperature increases, but the average void size 1150 °C (2100 °F). tures increase the size of the plastic zone, which increases. Examination of radiation-induced Burning occurs at higher temperatures (gen- enhances the ability of the crack to follow the voids requires thin-foil TEM. Examination of erally above about 1370 °C, or 2500 °F) at prior-austenite grain boundaries. However, if fractures of irradiated ferritic materials tested at which incipient melting occurs at the grain the sample is highly tempered, it is more low temperatures reveals a change from cleav- boundaries. Hot working after burning will not difficult to fracture. Therefore, the optimum age fracture to a mixture of cleavage and repair the damage. In the case of overheating, tempering temperature is one that permits the intergranular fractures. Fractures of specimens facet formation can be suppressed if adequate crack to follow the austenite grain boundaries to tested at higher temperatures reveal a change in hot reduction, usually greater than 25% reduc- reveal faceting while still permitting the speci- dimple size and depth. Irradiation embrittle- tion, is performed. In such cases, there will be men to be broken easily. Facets are more easily ment in austenitic stainless steels is associated little or no change in toughness, but there may observed on impact specimen fractures than on 128 / Visual Examination and Light Microscopy

la) (b)

Flgo 60 Close-ups of two regions of the spall surface shown in Fig. 58 that exhibited fatigue beach marks. (a) Area C. 10×. (b) Area D. 9×. See also Fig. 59. tensile fractures. For impact specimens, facets ative of overheating (Fig. 86). Figure 87, which dimples in the 1370-°C (2500-°F) specimen are are best observed when the test temperature is shows another example of facets, illustrates the clearly visible, but those in the 1205-°C (2200- above the brittle-to-ductile transition tempera- fracture of a vanadium-niobium plate steel slab °F) specimen are extremely fine. This differ- ture. Faceting is generally easier to observe due to overheating. The accompanying micro- ence arises from the greater dissolution of the when the fracture plane is transverse to the graph shows that substantial carbon segregation original sulfides at the higher temperature. rolling direction rather than parallel to it. and grain growth were present in the overheated Because more of the sulfides are dissolved at Metallographers have made considerable use region. higher temperatures, more are available for of special etchants to reveal overheating in Another investigation showed interesting reprecipitation in the austenite grain boundaries suspected samples (Ref 9,387,393). One study features of fractures of overheated ASTM A508 upon cooling. documents the evaluation of over 300 different class II forging steel (Ref 390). The number of The number of facets in these specimens etchants in an effort to develop this technique facets per square inch of test fracture depended varied from 4/cm 2 (24/in. 2) at 1205 °C (2200 (Ref 387). Several etchants have been found to on both the overheating temperature and the °F) to 121/cm 2 (783/in. 2) at 1370 °C (2500 °F). produce different etch responses between the tempering temperature for the fracture test (also The facet size also increased with temperature matrix and the grain-boundary area in over- shown in Ref 392). For a given soak tempera- because of grain growth at the soaking temper- heated specimens. These procedures work rea- ture, the number of facets per square inch ature. The facet fracture appearance has been sonably well for severely overheated speci- decreased as the tempered hardness decreased. referred to as intergranular microvoid coales- mens, but are not as sensitive as the fracture test For the same tempered hardness, the facet cence due to the presence of dimples on the when the degree of overheating is minor, par- density increased as the soaking temperature intergranular surfaces. Although some early ticularly in the case of low-sulfur steels. increased. Figure 88 shows a series of test researchers tried to study the facets using light Figure 86 shows a section through the center fractures of ASTM A508 class II material fractography, subsequent investigators have of an alloy steel compressor disk that cracked soaked at 1205 to 1370 °C (2200 to 2500 °F) used electron metallographic techniques. during forging because of overheating. A spec- and then quenched and tempered to a hardness Overheating will not occur if the sulfur imen was cut from the disk and normalized, of 37 to 39 HRC. Figure 89 shows SEM views content is below 0.001%. Increasing the sulfur quenched and tempered (to a hardness of 321 to of typical facets in samples soaked at 1205 and content raises the temperature at which over- 341 HB), and fractured, revealing facets indic- 1370 °C (2200 and 2500 °F). The sulfides in the heating begins. Low-sulfur steels are more Visual Examination and Light Microscopy / 129

Fig. 61 Example of a line spall in a forged, hardened alloy steel mill roll. Note that the portion that spoiled off broke as two sections, with the fractures propagating from the two circular fatigue regions growing from the line spall. The arrow at the top indicates that the spoil continued under the unbroken roll surface. Ultrasonic examination showed that it progressed another 137 mm (53/e in.) under the surface at a maximum depth of 11 mm (7/% in.). The subsurface, unspoiled portion of the roll is shown in Fig. 63; Fig. 62 is another view of the line spoil. prone to overheating problems than high-sulfur in o~ iron. Nitride precipitation can also ature increases, the time for maximum embrit- steels (Ref 393-396). Rare-earth additions raise occur, but the amount of nitrogen present is tlement decreases. Certain coating treatments, the overheating temperature by reducing the generally too low for substantial hardening. such as hot-dip galvanizing, can produce a high solubility of the sulfides. Transmission electron microscopy is the pre- degree of embrittlement in areas that were cold Quench Aging and Strain Aging. If a ferred technique for studying the aging phe- worked the critical amount, leading to brittle low-carbon steel is heated to temperatures nomenon. fractures. This can be prevented if the material immediately below the lower critical tempera- Strain aging occurs in low-carbon steels de- is annealed before coating. Additions of ele- ture and then quenched, it becomes harder and formed certain amounts and then aged, produc- ments that will tie up nitrogen, such as alumi- stronger, but is less ductile. This problem is ing an increase in strength and hardness and a num, titanium, vanadium, or boron, also help referred to as quench aging (Ref 397-402). loss of ductility (Ref 399-406). The amount of prevent strain aging. Brittleness increases with aging time at room cold work is critical; about 15% reduction Strain aging can also lead to stretcher-strain temperature, reaching a maximum in about 2 to provides the maximum effect. The resulting formation (Ltiders bands) on low-carbon sheet 4 weeks. The steels most prone to quench brittleness varies with aging temperature and steels. These marks are cosmetic defects rather aging are those with carbon contents from time. Room-temperature aging is very slow, than cracks, but the formed parts are unaccept- about 0.04 to 0.12%. Quench aging is caused and several months are generally required for able (Fig. 90). During tensile loading, such a by precipitation of carbide from solid solution maximum embrittlement. As the aging temper- sheet steel exhibits nonuniform yielding, fol- 130 / Visual Examination and Light Microscopy

Fig. 62 Anotherview of the line spall shown in Fig. 61. The arrow indicates the location of the line spall under the unfractured surface. See also Fig. 61 and 63. 0.6x lowed by uniform deformation. The elongation steels are known to be susceptible to cracking Control of the austenitizing temperature at maximum load and the total elongation are during or slightly after quenching. This is a is very important in tool steels. Excessive reduced, lessening cold formability. In non- relatively common problem for tool steels (Ref retained austenite and coarse grain structures aluminum-killed sheet steels, a small amount of 407), particularly those quenched in liquids. both promote quench cracking. Quench unifor- deformation, about 1%, will suppress the yield- Many factors can contribute to quench- mity is important, particularly for liquid point phenomenon for several months. If the cracking susceptibility: carbon content, harden- quenchants. Because as-quenched tool steels material is not formed within this safe period, ability, M S temperature (the temperature at are in a highly stressed condition, tempering the discontinuous yielding problem will even- which martensite starts to form), part design, must be done promptly after quenching to tually return and impair formability. surface quality, furnace atmosphere, and heat minimize quench cracking. Surface quality is Strain aging occurs due to ordering of carbon treatment practice (Ref 407-413). As the carbon also important because seams, laps, tool and nitrogen atoms at dislocations. Strain aging content increases, the M S and Mf (temperatures marks, stamp marks, and so on, can locate results because the dislocations are pinned by at which martensite formation starts and fin- and enhance cracking. These and other the solute atmospheres or by precipitates. Dur- ishes) temperatures decrease, and the volu- problems are reviewed and illustrated in Ref ing tensile deformation, most dislocations re- metric expansion and transformation stresses 407. main pinned. New dislocations appear in areas accompanying martensite formation increase. Quench cracking has been shown to be a of stress concentration and must intersect and Steels with less than 0.35% C are generally free statistical problem that occasionally defies cut through the pinned dislocations, resulting in from quench-cracking problems. The higher M S prediction and is frequently difficult to diag- higher flow stresses. Transmission electron mi- and M e temperatures permit some stress relief to nose. Heat treaters often experience short croscopy is required to study these effects. occur during the quench, and transformation periods in which cracking problems are Quench Cracking. Production of marten- stresses are less severe. Alloy steels with ideal frequent. An occasional heat of steel may show sitic microstructures in steels requires a heat critical diameters of 4 or higher are more an abnormally high incidence of quench treatment cycle that incorporates a quench after susceptible to quench cracking than lower- cracking for no apparent reason. Instances have austenitization. The part size, the hardenability hardenability alloy steels. In general, quench also been documented in which extensive of the steel, and the desired depth of hardening crack sensitivity increases with the severity of cracking has been associated with material dictate the choice of quench media. Certain the quench medium. from the bottom of ingots (Ref 411). Visual Examination and Light Microscopy/131

Fracture beneath unspalled roll OD

Fig. 63 The line spall fracture surface beneath the unspalled roll surface (see Fig. 61 and 62), revealed by sectioning and fracturing. The smooth, white fracture region formed when the spall was opened. The arrow at the top indicates the origin, shown at the right at higher magnification, which revealed beach marks.

The fracture surfaces of quench cracks are lists temperatures at which different temper treated in an automated, high production rate always intergranular. Macroscopic examples of colors occur for a carbon tool steel and a furnace system. Quench cracks in bolts usually quench cracks are shown in Ref 407. Quench stainless steel. The microstructure adjacent to occur longitudinally, often due to the presence crack surfaces are easiest to observe using SEM the crack will not be decarburized unless a of seams. These cracks, however, were circum- (Fig. 91). In quenched-and-tempered steels, specimen with an undetected quench crack is ferential, running part way around the head of proof of quench cracking is often obtained by rehardened. Quench cracks always begin at the the bolts, as shown in Fig. 93 (magnetic parti- opening the crack and looking (visually) for part surface and grow inward and are most cles show the cracks). The furnace had not been temper color typical for the tempering temper- commonly oriented longitudinally or radially used for about 2 years, and no cracking had ature used (Ref 414). Figure 92 shows a guide unless located by a change in section size. occurred in the past. During the time the for predicting temper colors as a function of Figure 93 shows an interesting example of furnace was not in use, two factors had oc- temperature and time for carbon steels. Table 1 quench cracking on ASTM A325 bolts heat curred that influenced the problem. First, a 132 / Visual Examination and Light Microscopy

Table 1 Temper colors observed on steels at various temperatures

Temperature(a) for development I of temper color I Carbon tool steel, Type 410, I--" ground(b) "--7 1"--- pelished(c) -"1 Color °C °F *C OF

First straw ...... 195 380 230 450 Straw ...... 215 420 Brown (bronze) ..... 235 460 430 8'i0 Purple ...... 260 500 Cobalt blue ...... 305 580 6~ 1275 Pale blue (gray) ..... 350 660 ...... (a) Samples held 60 min at heat. (b) Source: Ref 415. (c) Source: Ref 416

occurred had a reducing atmosphere, while the others had oxidizing atmospheres. This differ- ence in atmospheres is known to reduce the cooling rate during quenching. When the furnace atmosphere was made oxidizing, cracking stopped. Higher-hardenability bolts treated in the furnace did not crack even when the reduc- 1a1 ing atmosphere was used. Sigma-Phase Embrittlement. Sigma is a hard, brittle intermetallic phase that was dis- covered by Bain and Griffiths in 1927. Subse- quent studies have identified ~r-type compounds in over 50 different transition element alloys. Because of the influence of cr phase on the properties of stainless steels and superalloys, many studies have been performed. A few selected reviews are given in Ref 417 to 423. In austenitic stainless steels, ~r precipitates at grain and twin boundaries at temperatures between about 595 and 900 °C (1100 and 1650 °F). Sigma precipitation occurs fastest at about 845 °C (1550 °F). Cold working prior to heating in this range accelerates initiation of cr precipitation. Sigma is not coherent with the matrix. Embrittlement from ~r phase is most pronounced at temperatures below 260 °C (500 °F). Therefore, cr-embrittled components present serious maintenance problems. Sigma phase can be formed in iron- chromium alloys with chromium contents between 25 and 76%. Additions of silicon, molybdenum, nickel, and manganese permit ~r to form at lower chromium levels. Carbon additions retard ~r formation. Sigma forms more (b) readily from ferrite in stainless steels than from Fig. 64 Example of a line spall in a forged, hardened steel roll. (a) Section containing the spall cut from the austenite. This presents problems in the weld- roll. The arrow indicates the origin of the fracture, about 6 mm (0.25 in.) below the roll surface. (b) ing of austenitic stainless steels because a small The fracture origin at 6.5 x. Fatigue beach marks originate at the arrow; gross fracture marks can be seen radiating amount, about 5%, of ~-ferrite is introduced to from the fatigue zone. prevent hot cracking. Sigma will slightly increase bulk hardness, cooling tower was installed so that the quench Metallographic examination (Fig. 93) but the loss in toughness and ductility is sub- water was recirculated rather than used once showed that the bolt heads were not uniformly stantial. Sigma does provide increased high- and discharged. Therefore, the quench water hardened. The outer surface of the bolts from temperature strength, which may prove to be was typically about 10 to 35 °C (20 to 60 °F) the wrench fiats inward were hardened, but the beneficial if the reduced ductility is not a warmer, depending on the time of the year. middle of the top surface was not. Bolts from problem. Sigma also improves the wear resis- Second, basic oxygen furnace (BOF) steel the same heats treated in other furnaces were tance, and some applications have made use of (AISI 1040) was now being used rather than uniformly case hardened across the heads and this. Sigma does reduce creep resistance and electric furnace (EF) steel. Basic oxygen fur- did not crack. The quench cracks were ob- has a minor influence on corrosion resistance. nace steel has a lower residual alloy content served to form in the hardened region, near the The magnitude of these effects depends on the than electric furnace steel. Both of these factors interface of the unhardened surface zone. The amount of ~r present and its size and distribu- reduced the hardenability. furnace atmosphere in the line where cracking tion. Visual Examination and Light Microscopy/ 133

Figure 94 shows part of a broken hook used to hold a heat treatment basket during austenitization and quenching. The hook was made from cast 25Cr-12Ni heat-resisting steel; but the composition was not properly balanced, and a higher-than-normal 8-ferrite content was present in the hook. The 8-ferrite transformed to o~ during the periods that the hook was in the austenitizing furnace (temperatures from 815 to 900 °C, or 1500 to 1650 °F, generally). The micrograph shows a very heavy, nearly continuous grain-boundary cr network. Figure 95 shows three views of impact- formed cracks in type 312 stainless steel weld metal that had been heated at 815 °C (1500 °F) for 160 h before breaking. The micrographs show both partially broken and completely broken fractures to illustrate the nature of the crack path. This sample has a rather high a content. The SEM fractograph shows the rather brittle fracture appearance, a mixture of fine dimples around the ~r and quasi-cleavage. The impact energy was only 7% of that of a similar specimen without a present. Stress-corrosion cracking of metals and alloys occurs from the combined effects of tensile stress and a corrosive environment (Ref 424-434). Many different metals and alloys can fail by SCC under certain specific circum- stances. These failures may be catastrophic, may be due solely to SCC, or an SCC-nucleated crack may be propagated by another fracture mechanism--for example, by fatigue. Stress-corrosion cracks may propagate trans- granularly or intergranularly. In most cases, there is little evidence of the influence of corrosion, but energy-dispersive x-ray analysis can usually detect the presence of the corrosive agent on the fracture surface. Certain forms of Fig. 65 Macrograph showinga line spall that broke off a forged, hardened alloy steel roll. The central region SCC have been given other identifying names, of the fracture shows evidenceof crack growth by fatigue. Final rupture occurred by brittle fracture. such as season cracking of brass or caustic See also Fig. 66. 0.65 x embrittlement of riveted carbon steel structures. In some cases, it is difficult to determine amounts of impurity elements will make them intergranular fracture. Transgranular cracks of- whether the failure was due to hydrogen embrit- susceptible. In some alloys, the heat treatment ten follow specific crystallographic planes. tlement or SCC, or to their combined effects. condition is very important. Figure 96 shows a macrograph of an ASTM There are many close parallels between these Most commonly used alloys can fail by SCC A325 bolt that fractured in a bridge. The two mechanisms. under certain conditions. The best known case fracture surface is covered by rust, but it is Stress-corrosion fractures exhibit many of is SCC failures of austenitic stainless steels due apparent that the fracture began at the root of the characteristics of brittle fractures in that to chloride ions. Many aluminum alloys will the threads in the region near the arrow. Exam- little or no deformation accompanies the frac- fail by SCC in chloride environments. Copper ination of the microstructure in this region ture and the fracture is macroscopically flat. alloys fail by SCC in ammonia-containing en- revealed intergranular secondary cracks, as The speed at which a stress-corrosion crack vironments. Carbon steels can fail by SCC in shown. Due to a heat treatment error, the bolt propagates, however, is slow compared to a environments containing sodium hydroxide or was not tempered, and the hardened surface brittle fracture, and crack propagation may be other caustics. was in the as-quenched condition (53 to 57 discontinuous. At the fracture initiation site, the The stress needed to produce SCC failures is HRC). This made the bolt susceptible to SCC. metal must be stressed in tension. If the tensile generally low. Therefore, many studies have Figure 97 shows the fracture and microstruc- stress is relieved, the crack will stop propagat- determined threshold stress levels below which ture of a type 304 stainless steel wire (solution ing by stress corrosion. Cases of SCC under cracking does not occur by using a fracture annealed) that failed by SCC in boiling MgC12. compressive loading under certain circum- mechanics approach. In certain metals, a minor The wire was bent around a 13-mm (0.5-in.) stances have been reported, but these are not amount of deformation may cause a change diam pin before being placed in the solution. common. from transgranular to intergranular fracture. Therefore, the region where cracking occurred There are some interesting features of SCC. Corrosion products have been shown to aid was cold worked. The fracture is predominantly For example, such failures often occur under crack propagation by becoming trapped in the intergranular. relatively mild conditions of stress and corro- crack and exerting a wedging action. Figure 98 shows a micrograph of a predom- sive environment. In many cases, residual The crack path can be influenced by the inantly transgranular stress-corrosion crack in a stresses alone are adequate. Pure metals are microstructure. Grain-boundary precipitates or manganese-chromium austenitic drill collar al- immune to SCC, but the presence of minor grain-boundary denuded regions will promote loy. The crack occurred at the inner diameter of 134 / Visual Examination and Light Microscopy

Table 2 Properties of step-cooled embrittled AISI 4140 alloy steel

Tensile 50% Hardness, [~ strength, "--] [~ FATT(a), Phosphorus, % HRC MPa ksi °C °F

0.004 ...... 33 1031 149.5 -70 95 0.013 ...... 33.5 1071 155.4 39 -38 0.022 ...... 35.5 1099 159.4 30 85 (a) FATT, fracture appearance transition temperature based on a temperature for a 50% ductile, 50% brittle fracture appearance

rapid cooling. Carbon steels are not susceptible to temper embrittlement. Selected reviews on temper embrittlement and its fractographic aspects are given in Ref 435 to 441. The phenomenon of temper embrittlement has been known since 1883. Numerous service failures have been attributed to, or have been influenced by, temper embrittlement. Prior to development of electron fractographic techniques, the degree of embrittlement was identified by macroscopic fracture examination, degradation of mechanical properties, and use of grain-boundary etchants. The Charpy V- notch impact test has been widely used to assess the shift in transition temperature between embrittled and nonembrittled conditions. Elec- tron fractography has added another tool for ~R~IC|~" vv/"/" The roll surface of the spall shown in Fig. 65 after macroetching with 10% aqueous HNO 3. Etching assessment of the degree of embrittlement by revealed a craze crack pattern similar to heat checks, caused by abusive service conditions. About determination of the area fraction of in- 0.5× tergranular facets on the fracture surface. The amount of grain-boundary fracture depends on the degree of impurity segregation to the grain boundaries, but is also influenced by the matrix hardness, the prior-austenite grain size, and the test temperature. Also, the amount of intergranular fracture will vary as a function of the distance from the root of the V-notch. Phosphorus will segregate to the austenite grain boundaries during austenitization and during tempering in the critical range. Other em- brittlers, such as antimony, segregate to the grain boundaries only during tempering. In embrittled martensitic steels, the fracture follows the prior-austenite grain boundaries. In nonmartensitic steels embrittled by antimony, the fracture is intergranular when tested below (a) (b) the transition temperature, and the crack path Fig. 67 Two views of white-etchlng deformation bands, termed butterflies, that formed at a growing fatigue follows ferrite and upper bainite boundaries crack in a forged, hardened alloy steel mill roll. The crack would have led to spalling in later service. rather than the prior-austenite grain boundaries These white-etching zones form from the original tempered martensite due to cyclic service stresses and are often (Ref 440). When phosphorus is segregated, observed at the initiation sites of spalls. fractures follow the prior-austenite grain boundaries. the drill collar where residual stresses were the impurity content and the time within the The first etchant deliberately formulated to high. The cracking was caused by the chloride critical tempering range. Embrittlement occurs reveal temper embrittlement was developed in ion concentration in the drilling fluid. fastest at about 455 to 480 °C (850 to 900 °F). 1947 (Ref 442). Other etchants have also been Temper Embrittlement. Alloy steels con- It is most easily observed by using toughness developed (Ref 443, 444). Use of these etch- taining certain impurities (phosphorus, anti- tests, but tensile properties will be affected by ants is reviewed in Ref 9. To demonstrate the mony, tin, and arsenic) will become embrittled severe embrittlement. The alloy content also use of these etchants, three laboratory ingots of during tempering in the range of 350 and 570 influences embrittlement. Nickel-chromium AISI 4140 alloy steel were prepared from the °C (660 and 1060 °F) or during slow cooling steels are particularly prone to temper embrit- same melt, with the amount of phosphorus through this region. Embrittlement occurs be- tlement, but molybdenum additions reduce the varied in the three ingots. Wrought samples cause of the segregation of phosphorus, anti- susceptibility. Fortunately, the embrittlement is were heat treated and then subjected to a mony, arsenic, and/or tin to the grain bound- reversible and can be removed by retempering stepwise embrittlement cycle; the results are aries. The degree of embrittlement depends on above the critical tempering range, followed by given in Table 2. Visual Examination and Light Microscopy / 135

Light etching zone

Spalled region

(a) (b)

Fig. 68 Spalling of the hard surface region of a rail head was caused by rolling-contact fatigue in service. The hard area was formed by localized overheating, probably by spinning locomotive wheels. (a) Section through rail head, with field side at top and gage side at bottom. (b) Rail surface. 1.2 x. (c) Rail surface. 9 x. (d) Area adjacent to spall showing featureless surface zone (S), featureless zone with ferrite (F), and pearlite matrix (P). 7 x. (e) Section from near field side. Etched with 4% picral. 80 x

Scanning electron microscopy examination in samples containing 0.004% P, a few were ature will also increase with tempering in this revealed no intergranular fracture in the visible in samples with 0.013% P, and most range. samples containing 0.004% P, considerable were visible in the samples containing 0.022% Depending on the test temperature, TME intergranular fracture in the samples with P. This etchant is known to be very sensitive to produces a change in the fracture mode from 0.013% P, and predominantly intergranular segregated phosphorus, but will not reveal either predominantly transgranular cleavage or fracture in the samples containing 0.022% P. prior-austenite grain boundaries in temper- microvoid coalescence to intergranular fracture Etching of all of the sample types with the embrittled steels free of phosphorus, that is, if along the prior-austenite grain boundaries. In reagents described in Ref 442 and 444 revealed embrittlement is due only to segregated room-temperature tests, the fracture may prior-austenite grain boundaries in all samples. antimony, tin, or arsenic. change from microvoid coalescence to a mix- Etching of the samples with the saturated Tempered Martensite Embrittlement ture of quasi-cleavage and intergranular frac- aqueous picric acid reagent described in Ref (TME). Ultrahigh strength steels with marten- ture. 443, using sodium tridecylbenzine sulfonate as sitic microstructures are susceptible to embrit- Many studies have been conducted to deter- the wetting agent, also revealed the grain tlement when tempered between about 205 and mine the cause of TME, but the results are not boundaries in all samples. Micrographs of 400 °C (400 and 750 °F). Tempered martensite as clear-cut as for temper embrittlement (Ref these samples are shown in Fig. 99. All embrittlement is also referred to as 350-°C or 445 to 453). Martensitic microstructures are of samples were then deembrittled by tempering 500-°F embrittlement or one-step temper em- course susceptible to TME; but mixtures of at 620 °C (1150 °F), followed by water brittlement. Embrittlement can be assessed with martensite and lower bainite also suffer a loss in quenching. Next, they were repolished and a variety of mechanical tests, but the room- toughness while fully bainitic and pearlitic etched with each of the three reagents. The temperature Charpy V-notch absorbed energy microstructures are less affected or unaffected. etheral-picral etchants described in Ref 442 plotted against the tempering temperature is the Tempered martensite embrittlement occurs in and 444 did not reveal any grain boundaries. most common procedure. These data reveal a the tempering range in which e-carbide changes However, with the saturated aqueous picric decrease in impact energy in the embrittlement to cementite. Early studies concluded that TME acid etchant, grain boundaries were not visible range. The ductile-to-brittle transition temper- was due to precipitation of thin platelets of 136 / Visual Examination and Light Microscopy

fractured because of thermal embrittlement. The light micrographs of the fracture profile and a secondary crack reveal the intergranular nature of the fracture. This is more easily observed by direct SEM examination of the fracture, but the TiC and Ti(C,N) on the intergranular fracture is more clearly revealed by the extraction fractograph. Analysis of the ex- tracted grain-boundary precipitates is not hin- dered by detection of the matrix around the precipitate, as might occur with SEM-EDS analysis. 885-°F (475-°C) Embrittlement. Ferritic stainless steels containing more than about 13% Cr become embrittled withextended exposure to temperatures between about 400 and 510 °C (750 and 950 °F), with the maximum embrittlement at about 475 °C (885 °F). Therefore, this problem is referred to as 885-°F or 475-°C embrittlement (Ref 460-469). Aging at 475 °C (885 °F) increases strength and hardness, decreases ductility and toughness, and changes electrical and magnetic properties and corrosion resistance. The time at the aging temperature intensifies these changes. Embrittlement produces microstructural changes, notably a wid- ening of the etched grain boundaries followed by a darkening of the ferrite grains. Early x-ray studies suggested that a chromium-rich precipitate formed during embrittlement; this precipitate was incorrectly identified as ~r phase. In one study, TEM extraction replicas and selected-area diffraction patterns were used, followed by x-ray diffraction and fluorescence, to show that the precip- Fig. 69 Macrograph of broken rail showingthe intercannectionbetween the shellfracture (across the top) and itate had a bcc crystal structure and contained the detail fracture (on the transverseplane). 3.5 ×. (R. Rungta, Battelle Columbus Laboratories) about 80% Cr (Ref 460). These precipitates were coherent with the matrix and were extremely small. Aging of a 27% Cr alloy for cementite at the grain boundaries. However, increases with decreasing cooling rate through 1000 h at 480 °C (900 °F) produced particles TME has also been found to occur in very low this range. smaller than 5 nm in diameter, but aging for carbon steels (Ref 447, 448), and residual Increases in the concentration of carbon and 34 000 h produced particles about 22.5 nm in impurity elements have also been shown to be nitrogen render maraging steels more suscepti- diameter. This work showed that 885-°F em- essential factors in TME (Ref 445). The decom- ble to thermal embrittlement. Also, as the brittlement produces iron-rich a and chromium- position of interlath retained austenite into ce- titanium level increases, thermal embrittlement rich a' ferrites. Chromium nitrides also precip- mentite films with tempering in the range of problems become more difficult to control. itate during aging and are observed at grain 250 to 400 °C (480 to 750 °F) has also been Auger electron spectroscopy (AES) has shown boundaries, dislocations, and inclusions. found to be a factor in TME (Ref 449, 452). It that embrittlement begins with the diffusion of The formation of a' can occur in two ways: appears that TME results from the combined titanium, carbon, and nitrogen to the grain by nucleation and growth or by spinodal de- effects of cementite precipitation on prior- boundaries. Precipitation of TiC or Ti(C,N) on composition. The latter mechanism appears to austenite grain boundaries or at interlath bound- the grain boundaries represents an advanced be operative in higher chromium content alloys aries and the segregation of impurities, such as stage in the embrittlement. and has been the most commonly observed phosphorus and sulfur, at the prior-austenite Light microscopy can be used to reveal the formation process. The reaction is reversible; grain boundaries. nature of the fracture path in severely em- heating above the embrittlement range dis- Thermal Embrittlement. Maraging steels brittled specimens; but the chance of observing solves a'. In duplex stainless steels, the embrit- fracture intergranularly when the toughness has precipitates along the fracture path is low, and tlement temperature range appears to be been severely degraded because of improper pre-precipitation segregation is not detectable. broader, with additional phases precipitating in processing after hot working. This problem, Scanning electron microscopy examination of the upper portion of the range. called thermal embrittlement, occurs upon the fracture face is helpful, but the preferred As an example of 885-°F embrittlement, heating above 1095 °C (2000 °F), followed by approach is the use of extraction replica frac- Figure 101 shows the fracture of a duplex alloy slow cooling or by interrupted cooling with tography (Ref 459). This procedure reveals the similar in composition to type 329 stainless holding in the range of 815 to 980 °C (1500 to precipitates with strong contrast, and they can steel. This material was aged for 1000 h at 370 1800 °F) (Ref 454-458). Embrittlement has be easily identified with energy-dispersive °C (700 °F), producing a substantial hardness been attributed to precipitation of TiC and spectroscopy (EDS) and electron diffraction increase and a dramatic loss in toughness. Ti(C,N) on the austenite grain boundaries procedures. Room-temperature half-size Charpy V-notch during cooling through the critical temperature Figure 100 shows the fracture of a cobalt-free impact tests revealed 1.36 J (1 ft • lb) of range. The severity of the embrittlement high-titanium maraging steel specimen that absorbed energy compared to 64 J (47 ft • lb) Visual Examination and Light Microscopy / 137

Fig. 70 Macrograph of a rail that contained a detail fracture (upper left, beneath the rail head surface) that was placed in service in the FAST test track. Note the fatigue fracture that grew from the detail (a) fracture. 0.7,5 ×. (R. Rungta, Battelle Columbus Laboratories) for the same alloy in the hot rolled and intergranular fracture paths have been ob- annealed condition. The fracture is by cleav- served. age with some dark-etching ferrite grain Three factors combine to produce cold boundaries in the embrittled specimen, but the cracks: stress, hydrogen, and a susceptible same etch (Vilella's reagent) does not reveal microstructure. Stresses may be due to thermal the ferrite grain boundaries in the annealed expansion and contraction, or they may be due nonembrittled specimen. to a phase transformation, such as the austenite-to-martensite transformation. High- Weld Cracking carbon (plate) martensite microstructures are most susceptible to cold cracking. Low-carbon Cold cracking generally occurs at temper- (lath) martensite microstructures are less sus- atures below 205 °C (400 °F); however, in some ceptible, while ferritic and bainitic micro- (b) cases, these cracks are initiated at higher tem- structures are least susceptible. Hydrogen peratures and then grow to a detectable size at must be present above some critical level for Fig. 71 Fatigue cracks in an unbroken (a) and a completely broken (b) aluminum alloy. lower temperatures (Ref 470-479). Cracking cracking to occur. Figure 103 shows a (a) Etched with Kroll's reagent. 680 × (b) Etched with may occur during cooling to room temperature metallographic cross section of a cold crack in Keller's reagent. 510 × or a short time after reaching room temperature, a plate steel implant specimen welded with a or it may occur after a considerable time delay. high-hydrogen content electrode and an SEM Studies of cold cracking in plate steel weld- Cold cracks may be found in the weld metal or view of the fracture after sectioning to open the ments demonstrated that at high hydrogen lev- in the HAZ (Fig. 102). Both transgranular and crack. els cracking depends primarily on weld metal 138 / Visual Examination and Light Microscopy

Fig. 73 Fatigue crack appearance in an austenitic stainless steel specimen polished before fatigue loading, revealing slip lines on surfaces associated with crack formation and growth. Arrows indicate the direction of crack growth. 52 ×

microvoid coalescence to cleavage to inter- considered to occur in four stages: primary granular patterns. dendrite formation, dendrite interlocking, Hat cracking, also referred to as solidifica- grain-boundary development, and final solidi- tion cracking, is thought to occur at or above fication of the remaining interdendritic liquid. the solidus temperature of the last portion of the The third phase is critical, because cracks metal to freeze due to the influence of stresses formed at this time will not be healed. The from shrinkage (Ref 481-491). Austenitic stain- concept of the relative interfacial energy and less steels are known to be susceptible to hot dihedral angle has been used to predict cracking in the fusion zone and in the base hot-cracking tendencies (Ref 483). This ap- metal adjacent to the fusion zone. Cracking has proach demonstrates the critical influence of been attributed to low melting point films at sulfur and copper impurities on hot-cracking grain and subgrain boundaries. Analysis of susceptibility. Cracking is most likely when these intergranular cracks has shown that the nearly continuous liquid films are present in low melting point films are enriched in sulfur, the fourth stage of solidification. phosphorus, silicon, and manganese. There- Hot cracking can also occur in the base fore, one approach to reducing hot cracking is metal, and three problems have been suggested to restrict the sulfur and phosphorus levels to to produce liquid films in the base metal: less than 0.002% each and the silicon content to incipient grain-boundary melting, melting of less than 0.10%. Another approach, which is segregates, and absorption of liquid from the well established, is to introduce a small molten weld pool. The base metal along the Fig. 72 Fatigue crack (arrows)in a ferrite-pearlite amount, generally about 5 to 8%, of ~-ferrite to fusion line is stressed in tension initially during microstructure in a carbon steel. Etched with 2% nital. 800 × the austenitic weld metal. There are also reports welding. As the weld nugget cools, the HAZ of hot cracking at temperatures about 100 to becomes stressed in tension, producing separa- 300 °C (180 to 540 °F) below the equilibrium tion of regions containing liquid films. hardness (Ref 480). The risk of cold cracking solidus temperature. Light microscopy examination of cross sec- increased with increasing hardness in the range Several theories have been proposed to tions is a very important aspect of diagnosing of about 200 to 330 HV. At lower hydrogen account for weld metal hot cracking. The hot cracking. Figure 104 illustrates the use of levels, the resistance to cracking depends pri- shrinkage-brittleness theory states that cracking light microscopy in the study of extensive hot marily upon the microstructure. Acicular ferrite occurs during the final stage of solidification cracking in an electron beam weld that was provided good resistance to cracking, while because the partially solidified dendritic struc- contaminated by liquid copper by accidental ferritic structures with aligned second phases ture is unable to tolerate the contraction strain melting of the copper back-up plate. The cracks produced poor resistance to cold cracking. At or external restraint. The strain theory states are clearly intergranular, and elemental copper hardnesses below about 270 HV, the crack path the thin liquid films present during the final can be found throughout the region in the was not preferential to grain-boundary ferrite; stage of solidification lower the strength and prior-austenite grain boundaries. Figure 105 at higher hardnesses, there was an increased ductility of the solidifying metal. Thermal shows two examples of fine hot tears in the tendency for the cracks to follow the ferrite gradients present during solidification create HAZ of a welded HY-80 plate steel. grain boundaries. In the absence of grain- stresses that tend to pull the solid apart, Lamellar tearing occurs in the base metal boundary ferrite, intergranular fracture oc- creating tears. These concepts have been because of high through-thickness strains intro- curred. Fracture surfaces in plate steels can refined in a generalized theory of super-solidus duced by weld metal shrinkage under condi- exhibit a wide range of morphologies, from cracking (Ref 483). Weld solidification was tions of high joint restraint (Ref 492-502). Visual Examination and Light Microscopy / 139

F j g. 74 Macrographs (top) and microstructures (bottom) of short-time type 316 stainless steel tensile specimens tested at various temperatures. Top, from left: specimens tested at 760 °C (1400 °F), 815 °C (1500 °F), 870 °C (1600 °F), 925 °C (1700 °F), and 980 °C (1800 °F). Bottom, from left: Specimen tested at 760 °C (1400 °F). 35 x. Specimen tested at 980 °C (1800 °F). 35 x. Specimen tested at 815 °C (1500 °F). 350 X. Specimens at bottom etched with mixture of HCI, HNO s, and H20.

Tearing occurs by decohesion and linking along shearing (Fig. 106). These shear walls produce very beneficial in providing resistance to lamel- the rolling direction of the plate in the base the characteristic steplike appearance of lamellar tearing. metal beneath the HAZ of multipass welds lar tears (Fig. 107). Stress-relief cracking, also called post- approximately parallel to the fusion line. The Lamellar-tearing susceptibility is influenced weld heat treatment cracking, stress rupture cracks usually have a steplike appearance indic- by joint design factors that increase weld metal cracking, and reheat cracking, usually occurs in ative of discontinuous growth. The fracture shrinkage strains in the through-thickness direc- the HAZ, and sometimes in the weld metal, surfaces appear fibrous or woody. tion, that is, joint designs that orient the fusion during postweld heat treatments or during high- Steels susceptible to lamellar tearing exhibit line parallel to the hot-working direction (tee temperature service (Ref 503-512). This type of low through-thickness ductility and toughness. and comer joints); excessive joint restraint cracking occurs when crack-free weldments of The welding process must produce a weld- (larger welds than needed, full-penetration a susceptible alloy composition are subjected to fusion boundary approximately parallel to the welds rather than fillet welds, and so on); a thermal stress-relief heat treatment to reduce plate surface. For cracking to occur, the joint excessively high weld metal strength, which residual stresses and improve toughness. After design must produce high through-thickness concentrates the strain in the base metal; and such a treatment, cracking may be found that strains. When these conditions exist, lamellar high levels of component restraint. Material can be substantial in some instances. Cracking tearing is likely, depending on the inclusion properties are also very important. Through- usually occurs at stress raisers, for example, at types and content in the steel. thickness ductility and toughness must be max- geometrical discontinuities. Cracking begins at the inclusion/matrix inter- imized to inhibit lamellar tearing. Anisotropy Stress-relief cracking only occurs in metals faces, as shown schematically in Fig. 106. The of mechanical properties is enhanced by the that exhibit precipitation hardening during the cracked interfaces that lie in the same plane banding of microstructural constituents, by the high-temperature heat treatment or exposure. grow and link together to form terraces. The type and amount of microstructural phases and Cracking results when the creep ductility is high strain level, due to thermal contraction and constituents present, and by the number, size, inadequate to accommodate strains that accom- joint restraint, produces ductile tearing of the shape, and types of the inclusions present (Ref pany the relief of applied or residual stresses. matrix between the inclusion/matrix interfacial 502). Elongated sulfide and silicate stringers Stress-relief cracks are intergranular and are cracks. Further straining connects terrace are very detrimental. Low sulfur content and generally observed in the coarse-grain region of cracks on different parallel planes by ductile inclusion shape control have been shown to be the HAZ. They usually initiate at a stress 140 / Visual Examination and Light Microscopy

A B m (a) c

B A

(a) (b)

Fig. 75 One mechanism of intergranular cracking. (a) Schematic showing cracking due to grain-boundary sliding. Arrows along a grain boundary indicate that this boundary underwent sliding. (b) Cracks and (b) voids in AI-5.1Mg that was stress rupture tested at 260 °C (500 °F). Electrolytically polished. 60 x. Source: Ref 194 concentrator, such as a weld toe or an area of The Fracture Test. The examination of test incomplete penetration or fusion. The inter- sample fractures is a well recognized, simple granular crack path may exhibit considerable procedure for quality evaluation. The break- branching. The problem is most commonly ening of testpieces can be a very crude operation, countered in thick welded sections in which re- or it may be carefully controlled in test straint can produce very high residual stresses. machines. The simplest procedure is to support Steel microstructures that promote stress- the sample on its ends and strike the center relief cracking usually contain fine carbides that with a sledge hammer. In the fracturing of have precipitated within the grains in the HAZ; hardened steel disks, a mold can be designed to the grain boundaries are frequently denuded of support the specimen edges while a top cover carbides. This concentrates the creep strain at is used to locate a chisel over the center of (c) I I the weakened grain boundaries, producing ex- the specimen. The chisel is then struck with a 10 Fm tensive deformation, which is often accommo- sledge hammer. The closed mold prevents Fig. 76 Microstructure and fracture appearance dated by grain-boundary sliding, and inter- pieces from striking people in the area. The of type 316 stainless steel tested in creep granular cracking. Carbides may also lie along use of such a device to break cast coupons to fracture in air at 800 °C (1470 °F) at a load of 103 the grain boundaries, which promotes cracking. for inspection is discussed in Ref 513. In MPa (15 ksi). Time to rupture: 808 h. Light micrographs Alloy steels containing carbide-forming ele- some cases, it may be necessary to nick the (a and b) illustrate r-type cavities caused by vacancy condensation on boundaries perpendicular to the stress ments, such as chromium, molybdenum, and specimen in order to locate the fracture in axis. Both 90 × and electrolytically etched with oxalic vanadium, are most susceptible. A metallo- the desired area. A fracture press is ideal for acid. The SEM fractograph (c) illustrates the formation graphic example of the nature of stress-relief such work. A less satisfactory procedure is to of the r-type cavities at the grain boundaries. 1260 x. cracks is shown in Fig. 108. place one end of the specimen in a sturdy vise (W.E. White, Petro-Canada Ltd.) in a cantilever fashion and strike it on the free Quality Control end. Hardened high-hardness tool steels and Some of these applications of the fracture test Applications high-carbon steels can usually be broken at are reviewed in Ref 514. The fracture appear- room temperature, but more ductile alloys ance can be used to classify broadly the com- The examination of fractures often provides often require refrigeration in liquid nitrogen position of certain steels and cast irons in field information about microstructure or quality, to facilitate breaking. Face-centered cubic work (Ref 515). and fracture tests have been used for such materials are more difficult to fracture, partic- Inclusion stringers of macroscopic size can purposes since antiquity and are invaluable in ularly when section thicknesses are apprecia- be readily detected on a fractured specimen after failure analysis work. Fractures can often pro- ble. heat tinting in the blue heat range (Ref vide the metallurgist with interesting informa- Some of the uses for test fracture examina- 516-518). Several ASTM standards (A 295, A tion. For example, Figure 109 shows a fraction include: 485, A 535, and A 711) include test limitations tured ingot of an iron-chromium-aluminum Details about composition for macroscopic inclusion stringers detected in alloy in which the solidification structure is Detection of inclusion stringers this type of test. In most work with this test, the clearly revealed. A related example is shown in Assessment of degree of graphitization macroetch billet disks are hardened, fractured, Fig. 110, which illustrates a fatigue fracture Grain size measurement and blued on a hot plate or in a laboratory located by a shrinkage cavity. The dendritic Depth of hardening furnace. Heating to 260 to 370 °C (500 to nature of solidification is observed in the Detection of overheating 700 °F) is usually adequate. The fracture face shrinkage cavity. Detection of segregation should be a longitudinal plane. The lengths of Visual Examination and Light Microscopy / 141

(a)

(a) (al

(b) I I 4 iLm Fig. 79 Microstructure and fracture appearance of type 316 stainless steel tested in creep I I I I to fracture at 770 °C (1420 °F) using a 62-MPa (b) 40 itm (b) 40 ~tm (8.95-ksi) load. Time to rupture: 808 h. (a) Optical micrograph showing crack nucleation and growth by 77 Microstructure and fracture appearance 78 Microstructure fracture appearance Fig. Fig. and decohesion along the carbide/matrix interfaces. Etched of type 316L stainless steel tested in creep of type 316 stainless steel tested in creep to fracture in air at 800 °C (1470 °F), using a 53-MPa to fracture in air at 685 °C (1265 °F) at a load of 123 with dilute aqua regia. 440×. (b) SEM fractograph illustrating carbide morphology at the fracture surface. (7.7-ksi) load. Time to rupture: 839 h. The light micro- MPa (17.9-ksi). Time to rupture: 710 h. The light graph (a) illustrates w-crack coalescence by slow shear- micrograph (a) shows triple boundary cracking with 3150 ×. (W. E. White, Petro-Canada Ltd.) ing along grain boundaries. The SEM fractograph extensive bulk deformation and grain elongation. The (b) shows the fracture appearance characterized by SEM fractograph (b) illustrates classic shear dimple poorly formed shear dimples and particle dragging. topography. (a) Electrolytically etched with oxalic (a) Electrolytically etched with oxalic acid. 84x. acid. 110×. (b)380×. (W.E. White, Petro-Canada (b) 420×. (W.E. White, Petro-Canada Ltd.) Ltd.) any visible stringers, which appear white steels that should contain graphite, such as AISI against the blue fracture, are measured. Figure 06, or in high-carbon tool steels that should be 111 shows an example of inclusion stringers free of graphite. Figure 113 shows four frac- detected by this procedure. tured test disks of a high-carbon tool steel that Chill and wedge tests have been extensively contained considerable undesired graphite due Fig. 80 Example of the macroscop,c appearance of hydrogen flakes in plate steel. 1.6 x used in the manufacture of gray and white irons to accidental addition of aluminum. to reveal carbide stability or the tendency of an The prior-austenite grain size of high-carbon iron to solidify as white iron rather than gray martensitic steels, such as tool steels, can be etched, as shown. The sample is then hardness iron (Ref 519-521). These tests show the com- assessed by comparing a fractured specimen to tested with the Rockwell superficial test to bined effect of melting practice and composi- a series of graded fracture specimens (Ref 9, determine the depth to 55 HRC, that is, the tion on carbide stability, but are not a substitute 522, 523). This method is fast, simple, and location for 50% martensite. This depth corre- for chemical analysis. The chill test uses a accurate, but can be applied only to high- lates well with the case/core transition depth on small rectangular casting in which one face is hardness relatively brittle martensitic (retained the test fracture. a chill plate, while the wedge test employs a austenite may also be present) steels. Complete As discussed earlier, fracture tests are com- wedge-shaped mold in which the cooling rate details on the use of this test and its limitations monly used to detect overheating. A test spec- varies with the wedge thickness. After a casting are provided in Ref 9 and 523. imen of appropriate size for subsequent fractur- is made with either type of mold, it is broken The depth of hardening in case-hardened ing is normalized, austenitized, quenched and and the fracture is examined. Figure 112 shows steels, such as carbon tool steels, can be easily tempered to about 321 HB, and fractured, an example of a wedge test fracture. In the assessed by using fractured specimens. The P-F usually at room temperature. The appearance of wedge test, the distance from the wedge tip to test is commonly used for this purpose (Ref coarse-grain facets on the fracture, as shown in the end of the white fracture is measured; in the 522). Figure 114 shows P-F test specimens. A Figures 86 to 88, is taken as proof of overheat- chill test, the depth of the white iron chill is 19-mm (3/4-in.) diam, 100-mm (4-in.) long ing. measured. specimen is austenitized at the recommended Fractures of transverse or longitudinal sec- In the production of tool steels, the fracture temperature and brine quenched. After fractur- tions from wrought products or from cast- test can be used to detect graphitization in tool ing, one fracture face is ground and macro- ings have been used for many years to evalu- 142 / Visual Examination and Light Microscopy

(a) (b) Fig. 8:2 Planar (al and cross sectional (b) v~ews of ~ntergronulor corrosion (gra~n dropping)~n a sensit~z~ austenitic stainless steel. As-polished. (o) 50 x. (b) 100 x

Fig. 81 Hydrogen blister in the web of a structural steel section. About 0.3 × (b)

201~m ,;N|rk~l PlntJno ate metal quality (Ref 514, 524, 525). The fracture test can reveal texture, flaking, graphitization, slag, blow holes, pipe, inclusions, and segregation. The general appearance of the test fracture is often classified as being coarse or fine, woody, fibrous, ductile, or brittle. Fibrous and woody fractures (longitudinal) result from microstructural anisotropy due to banding, segregation, or excessive inclusion stringers. Woody fractures usually result from grosser conditions compared to fibrous fractures. w " L~ Figure 115 shows an example of the transverse fracture of an AISI 1070 steel bar that broke during heat treatment due to gross chemical segregation and overheating. The trans- (a) (c) verse fracture reveals a fine-grain appearance at the hardened outer periphery of the bar. The Fig. 83 Three views of a fractured sensitized specimen of type 304 stainless steel. (a) Partially broken coarse intergranular appearance of the fracture specimen. Etched with mixed acids. (b) SEM view of fracture; specimen broken at - 195 °C (-320 °F). (c) Cross section of fracture. Etched with acetic glyceregia in the interior is evident in Fig. 115. Radial marks can be seen emanating from the coarse- As noted previously, the appearance of index of quality. If the tensile fracture is not grain central region of the bar. tensile-test fractures has long been used as an affected by gross defects, its general appear- Visual Examination and Light Microscopy/ 143

(a) (b) (a) Fig. 84 Eutectoid carbon steel specimen embrittled by liquid copper at 1100 °C (2010 °F). (o) Microgroph of partially broken specimen;arrows point to grain-boundary copper penetration. Etched with 4% picral. (b) SEM fractograph of completely broken specimen ance is classified as irregular, angular, flat, color to the fracture. The texture of cast bronze partially cupped, or a full cup-and-cone. The test fractures varies with composition and grain latter type is generally regarded as the optimum size (Ref 531). Increased tin or phosphorus fracture appearance. An extensive study of contents generally produce finer, silkier frac- tensile fracture appearance and its relationship tures, while lead additions produce a granular to test results was reported in 1950 (Ref 58). appearance. Fibrous textures are often observed For a given type of steel and heat treatment in low-tin bronzes. condition, there is an approximately linear relationship between tensile strength and tensile REFERENCES ductility. The data scatter for the relationship between tensile strength and percent elongation 1. G.F. Vander Voort, Conducting The Fail- or %RA was shown to be much less for full ure Examination, Met. Eng. Q., Vol 15, cup-and-cone test fractures than for partial cup- May 1975, p 31-36 and-cone fracture specimens (Ref 58). 2. C.A. Zapffe and M. Clogg, Jr., Fractog- The fracture test is widely used in the pro- raphy--A New Tool for Metallurgical duction of castings. Test fractures reveal the Research, Trans. ASM, Vol 34, 1945, p influence of casting temperature on the primary 77-107 macrostructure. The change in fracture appear- 3. C.A. Zapffe et al., Fractography: The (b) ance of hypereutectic white irons with increas- Study of Fractures at High Magnification, ing casting temperature is illustrated in Ref Iron Age, Vol 161, April 1948, p 76-82 526. The same trend was not observed for tests 4. C.A. Zapffe and C.O. Worden, Fracto- of hypoeutectic white irons. One investigation graphic Registrations of Fatigue, Trans. found that the fracture appearance of tensile ASM, Vol 43, 1951, p 958-969 specimens of white irons varied with tensile 5. C.A. Zapffe et al., Fractography as a strength (Ref 527). A specimen with a high Mineralogical Technique, Am. Miner- tensile strength (490 MPa, or 71 ksi) had a fine alog., Vol 36 (No. 3 and 4), 1951, p fracture, while one with a low tensile strength 202-232 (280 MPa, or 40.6 ksi) had a coarse columnar 6. K. Kornfeld, Celluloid Replicas Aid appearance. Study of Metal Fractures, Met. Prog., The fracture test has been widely used in the Vol 77, Jan 1960, p 131-132 production of copper castings (Ref 528-531). 7. P.J.E. Forsyth and D.A. Ryder, Some Such tests are often performed by the removal Results of the Examination of Aluminum of the gating or feeding portions of the casting Alloy Specimen Fracture Surfaces, Met- by fracture. In copper-base castings, the color allurgia, March 1961, p 117-124 of the fracture provides considerable informa- 8. K.R.L. Thompson and A.J. Sedriks, The tion because it is influenced by alloy composi- Examination of Replicas of Fracture Sur- tion and the presence of various phases or faces by Transmitted Light, J. Austral. constituents (Ref 531). For example, the oL Inst. Met., Vol 9, Nov t964, p 269-271 solid-solution phase in tin bronze has a reddish- 9. G.F. Vander Voort, Metallography: brown fracture color, the tin-rich 8 phase has a Principles and Practice, McGraw-Hill, bluish-white fracture color, copper phosphide 1984 (c) appears pale blue-gray, and lead appears dark 10. H.C. Rogers, The Tensile Fracture of blue-gray. Bronzes with low tin and phospho- Ductile Metals, Trans. A1ME, Vol 218, Fig. 85 not shortness in a structural steel caused rus contents are reddish-brown, and the fracture June 1960, p 498-506 during rolling by internal LME due to color changes to gray-brown and then to gray as 11. C. Laird and G.C. Smith, Crack Propa- copper segregation. (a) Macrograph of section from toe of flange. 0.4 ×. (b) and (c) Micrographs showing the tin or phosphorus contents are increased. gation in High Stress Fatigue, Philos. grain-boundary copper films that were molten during Zinc additions to copper impart a brassy yellow Mag., Vol 7, 1962, p 847-857 rolling. (b) and (c) Etched with 2% nital. Both 70 X 144 / Visual Examination and Light Microscopy

and Metallographic Morphology of Fa- tigue Initiation Sites, in Fractography in Failure Analysis, STP 645, American Society for Testing and Materials, 1978, p 235-248 22. W.R. Kerr et al., On the Correlation of Specific Fracture Surface and Metallo- graphic Features by Precision Sectioning in Titanium Alloys, Metall. Trans., Vol 7A, Sept 1976, p 1477-1480 23. D.E. Passoja and D.J. Amborski, Frac- ture Profile Analysis by Fourier Trans- form Methods in Microstructural Sci- ence, Vol 6, Elsevier, 1978, p 143-158 24. W.T. Shieh, The Relation of Microstruc- ture and Fracture Properties of Electron Beam Melted, Modified SAE 4620 Steels, Metall. Trans., Vol 5, May 1974, p 1069-1085 25. J.R. Pickens and J. Gurland, Metallo- graphic Characterization of Fracture Sur- face Profiles on Sectioning Planes, in Proceedings of the 4th International Con- gressfor Stereology, NBS 431, National Bureau of Standards, 1976, p 269-272 26. S.M. E1-Soudani, Profilometric Analysis of Fractures, Metallography, Vol 11, July 1978, p 247-336 27. E. Rabinowicz, Taper Sectioning. A Method for the Examination of Metal Surfaces, Met. Ind., Vol 76, 3 Feb 1950, p 83-86 28. L.E. Samuels, An Improved Method for the Taper Sectioning of Metallographic Specimens, Metallurgia, Vol 51, March 1955, p 161-162 29. W.A. Wood, Formation of Fatigue Cracks, Philos. Mag., Vol 3 (No. 31), Series 8, July 1958, p 692-699 30. W.A. Wood and H.M. Bendler, Effect of Superimposed Static Tension on the Fa- tigue Process in Copper Subjected to (a) (b) Alternating Torsion, Trans. AIME, Vol 224, Feb 1962, p 18-26 Fig. 86 Alloy steel compressor disk that cracked from overheating during forging. (a) Macrograph of disk (cracking at arrow). 0.4 x. (b) Fracture surface of a specimen from the disk that was normalized, 31. W.A. Wood and H.M. Bendler, The quenched, and tempered to 321 to 341 HB. The treatment revealed facets indicative of overheating. 5 x Fatigue Process in Copper as Studied by Electron Metallography, Trans. AIME, Vol 224, Feb 1962, p 180-186 12. C. Laird, The Influence of Metallurgical 5-29 32. P.J.E. Forsyth, Some Metallographic Ob- Structure on the Mechanisms of Fatigue 16. R.W. Staehle et al., Mechanism of Stress servations on the Fatigue of Metals, J. Crack Propagation, in Fatigue Crack Corrosion of Austenitic Stainless Steels in Inst. Met., Vol 80, 1951-1952, p 181-186 Propagation, STP 415, American Society Chloride Waters, Corrosion, Vol 15, July 33. K. Kitajima and K. Futagami, Fracto- for Testing and Materials, 1967, p 1959, p 51-59 (373t-381t) graphic Studies on the Cleavage Fracture 131-180 17. K.W. Burns and F.B. Pickering, Defor- of Single Crystals of Iron, in Electron 13. D.P. Clausing, The Development of Fi- mation and Fracture of Ferrite-Pearlite Microfractography, STP 453, American brous Fracture in a Mild Steel, Trans. Structures, J. Iron Steel Inst., Vol 202, Society for Testing and Materials, 1969, ASM, Vol 60, 1967, p 504-515 Nov 1964, p 899-906 p 33-59 14. I.E. French and P.F. Weinrich, The 18. J.H. Bucher et al., Tensile Fracture of 34. R.M.N. Pelloux, Mechanisms of Forma- Shear Mode of Ductile Fracture in a Three Ultra-High-Strength Steels, Trans. tion of Ductile Fatigue Striations, Trans. Spheroidized Steel, Metall. Trans., Vol A1ME, Vol 233, May 1965, p 884-889 ASM, Vol 62, 1969, p 281-285 10A, March 1979, p 297-304 19. U. Lindborg, Morphology of Fracture in 35. W.D. Hepfer et al., A Method for Deter- 15. R.H. Van Stone and T.B. Cox, Use Pearlite, Trans. ASM, Vol 61, 1968, p mining Fracture Planes in Beryllium of Fractography and Sectioning Tech- 500-504 Sheet by the Use of Etch Pits, Trans. niques to Study Fracture Mechanisms, 20. C.T. Liu and J. Gurland, The Fracture ASM, Vol 62, 1969, p 926-929 in Fractographic-Microscopic Cracking Behavior of Spheroidized Carbon Steels, 36. A. Inckle, Etching of Fracture Surfaces, Processes, STP 600, American Socie- Trans. ASM, Vol 61, 1968, p 156--167 J. Mater. Sci., Vol 5 (No. 1), 1970, p ty for Testing and Materials, 1976, p 21. D. Eylon and W.R. Kerr, Fractographic 86-88 Visual Examination and Light Microscopy / 145

(a) (b|

Fig. 87 Fracture (a) of a slab that was rolled from a vanadium-nlobium plate steel. The coarse fracture facets indicate overheating before hot rolling, 1.25 ×. The etched section (b, with fracture olong the top edge) shows carbon segregation in the center, as evidenced by the greater amount of pearllte. Note also the coarseness of the pearlite in the center. Etched with 2% nital. 3.5 ×

37. Y. Mukai et al., Fractographic Observa- 47. E.R. Parker et al., A Study of the Ten- 57. C. Wells et al., Effect of Composition on tion of Stress-Corrosion Cracking of AISI sion Test, Proc. ASTM, Vol 46, 1946, p Transverse Mechanical Properties of 304 Stainless Steel in Boiling 42 Percent 1159-1174 Steel, Trans. ASM, Vol 46, 1954, p Magnesium-Chloride Solution, in Frac- 48. I.E. French and P.F. Weinrich, The Ten- 129-156 tography in Failure Analysis, STP 645, sile Fracture Mechanisms of F.C.C. Met- 58. H.H. Johnson and G.A. Fisher, Steel American Society for Testing and Mate- als and Alloys--A Review of the Influ- Quality as Related to Test Bar Fractures, rials, 1978, p 164-175 ence of Pressure, J. Austral. Inst. Met., Trans. AFS, Vol 58, 1950, p 537-549 38. D. McLachlan, Jr., Extreme Focal Depth Vol 22 (No. 1), March 1977, p 40-50 59. J. Welchner and W.G. Hildorf, Relation- in Microscopy, Appl. Opt., Vol 3 (No. 49. H.C. Rogers, The Effect of Material ship of Inclusion Content and Transverse 9), 1964, p 1009-1013 Variables on Ductility, in Ductility, Ductility of a Chromium-Nickel-Molyb- 39. M.D. Kelly and J.E. Selle, The Opera- American Society for Metals, 1968, p denum Gun Steel, Trans. ASM, Vol tion and Modification of the Deep Field 31-56 42, 1950, p 455-485 Microscope, in Proceedings of the 2nd 50. L.D. Kenney et al., Effect of Particles on 60. H.D. Shephard and E.A. Loria, The Na- Annual Technical Meeting, International the Tensile Fracture of Aluminum Alloys, ture of Inclusions in Tensile Fractures of Metallographic Society, 1969, p 171-177 in Microstructural Science, Vol 8, Elsev- Forging Steels, Trans. ASM, Vol 41, 40. J.H. Waddell, Deep-Field Low-Power ier, 1980, p 153-156 1949, p 375-395 Microscope, Res. Dev., Vol 19, May 51. B.I. Edelson and W.W. Baldwin, Jr., 61. F.L. Carr et al., Correlation of Micro- 1968, p 34 The Effect of Second Phases on the Me- fractography and Macrofractography of 41. A. Phillips et al., Electron Fractography chanical Properties of Alloys, Trans. AISI 4340 Steel, in Application of Handbook, AFML-TDR-64-416, Air ASM, Vol 55, 1962, p 230-250 Electron Microfractography to Materials Force Materials Laboratory, 31 Jan 1965 52. J. Gurland and J. Plateau, The Mecha- Research, STP 493, American Socie- 42. A. Phillips et al., Electron Fractography nism of Ductile Rupture of Metals Con- ty for Testing and Materials, 1971, p Handbook, MCIC-HB-08, Air Force Ma- taining Inclusions, Trans. ASM, Vol 56, 36-54 terials Laboratory and the Metals and 1963, p 442-454 62. J.A. Kies et al., Interpretation of Fracture Ceramics Information Center, June 1976 53. T. Inoue and S. Kinoshita, Mechanism of Markings, J. Appl. Phys., Vol 21, July 43. B.V. Whiteson et al., Electron Fracto- Void Initiation in the Ductile Fracture 1950, p 716-720 graphic Techniques, in Techniques of Process of a Spheroidized Carbon Steel, 63. J. Nunes et al., Macrofractographic Metals Research, Vol II, Pt. I, Intersci- in Microstructure and Design of Alloys, Techniques, in Techniques of Metals Re- ence, 1968, p 445-497 Vol 1, Institute of Metals and the Iron and search, Vol 2, Pt. 1, John Wiley & Sons, 44. C.D. Beachem, The Effects of Crack Tip Steel Institute, 1973, p 159-163 1968, p 379-444 Plastic Flow Directions Upon Micro- 54. C. Wells and R.F. Mehl, Transverse 64. J.H. Hollomon, Temper Brittleness, scopic Dimple Shapes, Metall. Trans., Mechanical Properties in Heat Treated Trans. ASM, Vol 36, 1946, p 473-541 Vol 6A, Feb 1975, p 377-383 Wrought Steel Products, Trans. ASM, 65. W.R. Clough et al., The Rosette, Star, 45. B.J. Brindley, The Mechanism of Ductile Vol 41, 1949, p 715-818 Tensile Fracture, J. Basic Eng. (Trans. Fracture in an Fe-21% Cr-0.5% C Alloy, 55. E.A. Loria, Transverse Ductility Varia- ASME), Vol 90, March 1968, p 21-27 Acta Metall., Vol 16, April 1968, p tions in Large Steel Forgings, Trans. 66. A.S. Shneiderman, Star-Type Fracture in 587-595 ASM, Vol 42, 1950, p 486-498 the Tensile Testing of Testpieces, Ind. 46. A.W. Thompson and P.F. Weihrauch, 56. E.G. Olds and C. Wells, Statistical Meth- Lab., Vol 41, July 1974, p 1061-1062 Ductile Fracture: Nucleation at Inclu- ods for Evaluating the Quality of Certain 67. F.R. Larson and F.L. Cart, Tensile Frac- sions, Scr. Metall., Vol 10, Feb 1976, p Wrought Steel Products, Trans. ASM, ture Surface Configurations of a Heat- 205-210 Vol 42, 1950, p 845-899 Treated Steel as Affected by Tempera- 146 / Visual Examination and Light Microscopy

Three Ultra-High-Strength Steels, Trans. A1ME, Vol 233, May 1965, p 884-889 71. F.L. Carr et al., Mechanical Properties 1370 °C and Fracture Surface Topography of a (2500 °F) Thermally Embrittled Steel, in Temper Embrittlement in Steel, STP 407, Ameri- can Society for Testing and Materials, 1968, p 203-236 72. R.J. Hrubec et al., The Split, Layered, Cup-and-Cone Tensile Fracture, J. Basic Eng. (Trans. ASME), Vol 90, March 1968, p 8-12 1345 *C (2450 °F) 73. B.M. Kopadia et al., Influence of Me- chanical Fibering on Brittle Fracture in Hot Rolled Steel Plate, Trans. ASM, Vol 55, 1962, p 289-398 74. E.A. Almond et al., Fracture in Lami- nated Materials, in Interfaces in Compos- ites, STP 452, American Society for Test- ing and Materials, 1969, p 107-129 1315 °C (24O0 °F) 75. E.A. Almond, Delamination in Banded Steels, Metall. Trans., Vol 1, July 1970, p 2038-2041 76. D.F. Lentz, Factors Contributing to Split Fractures, in Mechanical Working and Steel Processing, Meeting XII, American Institute of Mining, Metallur- gical, and Petroleum Engineers, 1974, p 1285 *C (2350 OF) 397-411 77. D.M. Fegredo, The Effect of Rolling at Different Temperatures on the Fracture Toughness Anisotropy of a C-Mn Struc- tural Steel, Can. Metall. Q., Vol 14 (No. 3), 1975, p 243-255 78. H. Hero et al., The Occurrence of De- lamination in a Control Rolled HSLA 1260 °C Steel, Can. Metall. Q., Vol 14 (No. 2), (2300 OF) 1975, p 117-122 79. P. Brozzo and G. Buzzichelli, Effect of Plastic Anisotropy on the Occurrence of "Separations" on Fracture Surfaces of Hot-Rolled Steel Specimens, Scr. Metall., Vol 10, 1976, p 233-240 80. D.N. Hawkins, Cleavage Separations in 1230 °C Warm-Rolled Low-Carbon Steels, Met. (2250 °F) Technol., Sept 1976, p 417-421 81. A.J. DeArdo, On Investigation of the Mechanism of Splitting Which Occurs in Tensile Specimens of High Strength Low Alloy Steels, Metall. Trans. A, Vol 8, March 1977, p 473-486 82. B.L. Bramfitt and A.R. Marder, Splitting 1205 *C Behavior in Plate Steels, in Toughness (2200 °F) Characterization and Specifications for HSLA and Structural Steels, American Institute of Mining, Metallurgical, and Petroleum Engineers, 1977, p 236-256 --'~I~1~|~° vv~R Macrographs of specimens of ASTM A508 class II steel heated to the indicated temperatures, 83. B.L. Bramfitt and A.R. Marder, A Study normalized, quenched, tempered to about 37 to 39 HRC, and fractured. Overheating facets are of the Delamination Behavior of a Very observed in all samples, but are not excessively large until the steel is heated to 1260 °C (2300 °F) or above. Source: Ref 390 Low-Carbon Steel, Metall. Trans., Vol 8A, Aug 1977, p 1263-1273 ture, Trans. ASM, Vol 55, 1962, p 663-682 84. J.E. Ryall and J.G. Williams, Fracture 599-612 69. F.L. Carr and F.R. Larson, Fracture Sur- Surface Separations in the Charpy V- 68. F.R. Larson and J. Nunes, Low Temper- face Topography and Toughness of AISI Notch Test, BHP Tech. Bull., Vol 22, ature Plastic Flow and Fracture Tension 4340 Steel, J. Mater., Vol 4, Dec 1969, Nov 1978, p 38-45 Properties of Heat-Treated SAE 4340 p 865-875 85. B. Engl and A. Fuchs, Macroscopic and Steel, Trans. ASM, Vol 53, 1961, p 70. J.H. Bucher et al., Tensile Fracture of Microscopic Features of Separations in Visual Examination and Light Microscopy/147

Phase Morphology on the Tensile Frac- ture Characteristics of Carbon Steels, Eng. Fract. Mech., Vol 7, Sept 1975, p 411-417 96. W.A. Spitzig, Effect of Sulfides and Sulfide Morphology on Anistropy of Ten- sile Ductility and Toughness of Hot- Rolled C-Mn Steels, Metall. Trans., Vol 14A, March 1983, p 471-484 97. J.E. Croll, Factors Influencing the Through-Thickness Ductility of Struc- tural Steels, BHP Tech. Bull., Vol 20, April 1976, p 24-29 98. I.D. Simpson et al., Effect of the Shape and Size of Inclusions on Through- Thickness Properties, BHP Tech. Bull., Vol 20, April 1976, p 30-36 99. I.D. Simpson et al., The Effect of Non- Metallic Inclusions on Mechanical Prop- (a) (b) erties, Met. Forum, Vol 2 (No. 2), 1979, p 108-117 100. H. Takada et al., Effect of the Amount and Shape of Inclusions on the Direction- ality of Ductility in Carbon-Manganese Steels, in Fractography in Failure Anal- ysis, STP 645, American Society for Testing and Materials, 1978, p 335-350 10l. W.A. Spitzig and R.J. Sober, Influence of Sulfide Inclusions and Pearlite Content on the Mechanical Properties of Hot- Rolled Carbon Steels, Metall. Trans., Vol 12A, Feb 1981, p 281-291 102. W.C. Leslie, Inclusions and Mechanical Properties, in Mechanical Working & Steel Processing, Meeting XX, American Institute of Mining, Metallurgical, and Petroleum Engineers, 1983, p 3-50 103. C.J. McMahon, Jr. and M. Cohen, Initi- ation of Cleavage in Polycrystalline Iron, Acta Metall., Vol 13, June 1965, p (c) Id) 591-604 104. C.J. McMahon, Jr. and M. Cohen, The Fig. 89 S EM fractographs of the appearance of facets in specimens heated to 1205 and 1370 °C (2200 and 2500 °F). (a) and (b) Specimens heated to 1370 °C (2500 °F). (c) and (d) Specimens heated to 1205 Fracture of Polycrystalline Iron, in Pro- °C (2200 °F). (a) 135 x. (b) 680 x. (c) 135 x. (d) 680 x. Source: Ref 390 ceedings of the First International Con- ference on Fracture, Sendai, Japan, 12-17 Sept 1965, p 779-812 Structural Steels, Prakt. Metallogr., Vol 91. A. Gangulee and J. Gurland, On the 105. E.O. Hall, The Deformation and Aging 17, Jan 1980, p 3-13 Fracture of Silicon Particles in Alu- of Mild Steel: III Discussion of Results, 86. G.D. Fearnehough Fracture Propagation minum-SiliconAlloys, Trans. A1ME, Vol Proc. Phys. Soc. (London) B, Vol 64, 1 Control in Gas Pipelines: A Survey of 239, Feb 1967, p 269-272 Sept 1951, p 747-753 Relevant Studies, Int. J. Pressure Vessels 92. C.J. McMahon, Jr., The Microstructural 106. N.J. Fetch, The Cleavage Strength of Piping, Vol 2, 1974, p 257-282 Aspects of Tensile Fracture, in Funda- Polycrystals, J. Iron Steel Inst., Vol 173, 87. R. Schofield et al., "Arrowhead" Frac- mental Phenomena in the Material Sci- May 1953, p 25-28 tures in Controlled-Rolled Pipeline ences, Vol 4, Plenum Press, 1967, p 107. C.F. Tipper, The Study of Fracture Sur- Steels, Met. Technol., July 1974, p 247-284 face Markings, J. Iron Steel Inst., Vol 325-331 93. J.C.W. Van De Kasteele and D. Broek, 185, Jan 1957, p 4-9 88. T. Yamaguchi et al., Study of Mecha- The Failure of Large Second Phase Parti- 108. G.M. Boyd, The Propagation of Frac- nism of Separation Occurring on Frac- cles in a Cracking Aluminum Alloy, Eng. tures in Mild Steel Plates, Engineering, tured Surface of High Grade Line Pipe Fract. Mech., Vol 9 (No. 3), 1977, p Vol 175, 16 Jan 1953, p 65-69; 23 Jan Steels, Nippon Kokan Tech. Rep. (Over- 625-635 1953, p 100-102 seas), Dec 1974, p 41-53 94. A.S. Argon and J. Im, Separation of 109. J.B. Cornish and J.E. Scott, Fracture 89. D.S. Dabkowski et al., "Splitting-Type" Second Phase Particles in Spheroidized Study of Gas Transmission Line Pipe, in Fractures in High-Strength Line-Pipe 1045 Steel, Cu-0.6 Pct Cr Alloy, and Mechanical Working & Steel Processing, Steels, Met. Eng. Q., Feb 1976, p 22-32 Maraging Steel in Plastic Straining, Vol II, American Institute of Mining, 90. M. Iino et al., On Delamination in Line- Metall. Trans. A, Vol 6, April 1975, p Metallurgical, and Petroleum Engineers, pipe Steels, Trans. ISIJ, Vol 17, 1977, p 839-851 1969, p 222-239 450-458 95. T. Kunio et al., An Effect of the Second 110. D.E. Babcock, Brittle Fracture: An Inter- 148 / Visual Examination and Light Microscopy

Fig. 91 SEM fractograph of a quench crack surface in AISI 5160 alloy steel showing a nearly complete intergranular fracture path. 680 x

Temperature, ~'C 400 350 300 250 200 175 05 I i I i I

, / / i rl l / ///I oc Zl/ g . 1/ 1. ~.~

1[ l I I [ I 1 I I 370 33~ ~ 280 ) 24s ~ 20~ Fig. 90 Stretcher-strain marks (tLiders bands) on the surface of a range component after forming 0.25 x. 310 270 225 i''C J pretation of Its Mechanism, in AISI Year- 118. D.J. Wulpi, Understanding How Compo- 1.4 1.6 1.8 20 2.2 24 book, American Iron and Steel Institute, nents Fail, American Society for Metals, 1000/T, K 1968, p 255-278 1985 Fig. 92 Temper colors as a function of time at heat 111. J.R. Low, Jr. and R.G. Feustel, "Inter- 119. J. Mogul, Metallographic Characteriza- for AISI 1035 steel. Source: Ref 414 Crystalline Fracture and Twinning of Iron tion of Fatigue Failure Origin Areas, in at Low Temperature, Acta Metall., Vol 1, Metallography in Failure Analysis, Ple- March 1953, p 185-192 num Press, 1978, p 97-120 Materials, lron Steel Eng., Vol 42, Oct 112. J.H. Westbrook and D.L. Wood, Embrit- 120. R.D. Barer and B.F. Peters, Why Metals 1965, p 141-154 tlement of Grain Boundaries by Equilib- Fail, Gordon & Breach, 1970 126. J.D. Keller, Effect of Roll Wear on Spall- rium Segregation, Nature, Vol 192, 30 121. J.A. Bennett and J.A. Quick, "Mechan- ing, Iron Steel Eng., Vol 37, Dec 1960, p Dec 1961, p 1280-1281 ical Failures of Metals in Service," NBS 171-178 113. C. Lipson, Basic Course in Failure Anal- Circular 550, National Bureau of Stan- 127. F.K. Naumann and F. Spies, Working ysis, Penton (reprinted from Mach. Des.) dards, 27 Sept 1954 Roll With Shell-Shaped Fractures, Pract. 114. G. Jacoby, Fractographic Methods in Fa- 122. V.J. Colangelo and F.A. Heiser, Analysis Metallogr., Vo] 13, 1976, p 440-443 tigue Research, Exp. Mech., March of Metallurgical Failures, John Wiley & 128. J.M. Chilton and M.J. Roberts, Factors 1965, p 65-82 Sons, 1974 Influencing the Performance of Forged 115. D.A. Ryder, The Elements of Fractogra- 123. D. Mclntyre, Fractographic Analysis of Hardened Steel Rolls, in AISE Yearbook, phy, NATO Report AGARD-AG-155-71, Fatigue Failures, J. Eng. Mater. Technol. Association of Iron and Steel Engineers, NTIS AD-734-619, National Technical (Trans. ASME), July 1975, p 194-205 1981, p 85-90 Information Service, Nov 1971 124. C.E. Feddersen, "Fatigue Crack Propa- 129. M. Nakagawa et al., Causes and Coun- 116. G.F. Vander Voort, Macroscopic Exam- gation in D6AC Steel Plate for Several termeasures of Spalling of Cold Mill ination Procedures for Failure Analysis, Flight Load Profiles in Dry Air and JP-4 Work Rolls, in A1SE Yearbook, Associa- in Metallography in Failure Analysis, Fuel Environments," AFML-TR-72-20, tion of Iron and Steel Engineers, 1981, p Plenum Press, 1978, p 33-63 Battelle Memorial Institute, Jan 1972 134-139 117. D.J. Wulpi, How Components Fail, 125. M.K. Chakko and K.N. Tong, Evalua- 130. S.J. Manganello and D.R. Churba, Roll American Society for Metals, 1966 tion of Resistance to Spalling of Roll Failures and What to Do When They Visual Examination and Light Microscopy / 149

Bearing Failures, in Iron and Steel Engi- 146. J.A. Martin et al., Microstructural Alter- neering Yearbook, Association of Iron ations of Rolling Bearing Steel Undergo- and Steel Engineers, 1955, p 717-724 ing Cyclic Stressing, J. Basic Eng. 132. W.E. Duckworth and G.H. Walter, Fa- (Trans. ASME) D, Vol 88, Sept 1966, p tigue in Plain Bearings, in Proceedings of 555-567 the International Conference on Fatigue 147. J.L. O'Brien and A.H. King, Electron of Metals, The Institute of Mechanical Microscopy of Stress-Induced Structural Engineering and The American Society of Alterations Near Inclusions in Bearing Mechanical Engineers, 1956, p 585-592 Steels, J. Basic Eng. (Trans. ASME) D, (a) 133. W.R. Good and A.J. Gunst, Bearing Vol 88, Sept 1966, p 568-572 Failures and Their Causes, Iron Steel 148. W.E. Littmann and R.L. Widner, Propa- Eng., Vol 43, Aug 1966, p 83-93 gation of Contact Fatigue From Surface 134. R.L. Widner and J.O. Wolfe, Valuable and Subsurface Origins, J. Basic Eng. Results From Bearing Damage Analysis, (Trans. ASME) D, Vol 88, Sept 1966, p Met. Prog., Vol 93, April 1968, p 79-86 624-636 135. S. Borgese, An Electron Fractographic 149. J.A. Martin and A.D. Eberhardt, Identi- Study of Spalls Formed in Rolling Con- fication of Potential Failure Nuclei in tact, J. Basic Eng. (Trans. ASME) D, Vol Rolling Contact Fatigue, J. Basic Eng. 89, Dec 1967, p 943-948 (Trans. ASME) D, Vol 89, Dec 1967, p 136. R.J. Henry, The Cause of White Etching 932-942 Material Outlining Shell-Type Cracks in 150. J. Buchwald and R.W. Heckel, An Anal- Rail-Heads, J. Basic Eng. (Trans. ASME) ysis of Microstructural Changes in 52100 D, Vol 91, Sept 1969, p 549-551 Steel Bearings During Cyclic Stressing, 137. R. Rungta et al., An Investigation of Trans. ASM, Vol 61, 1968, p 750-756 (b) Shell and Detail Cracking in Railroad 151. R. Tricot et al., How Microstructural Rails, in Corrosion, Microstructure & Alterations Affect Fatigue Properties of Metallography, Vol 12, Microstructural 52100 Steel, Met. Eng. Q., Vol 12, May Science, American Society for Metals and 1972, p 39-47 the International Metallographic Society, 152. H. Swahn et al., Martensite Decay Dur- 1985, p 383-406 ing Roiling Contact Fatigue in Ball Bear- 138. C.G. Chipperfield and A.S. Blicblau, ings, Metall. Trans., Vol 7A, Aug 1976, Modelling Rolling Contact Fatigue in p 1099-1110 Rails, Railw. Gaz. Int., Jan 1984, p 153. R. C)sterlund and O. Vingsbo, Phase 25-31 Changes in Fatigued Ball Bearings, 139. H. Masumoto et al., Some Features and Metall. Trans., Vol 11A, May 1980, p Metallurgical Considerations of Surface 701-707 (c) Defects in Rail Due to Contact Fatigue, in 154. P.C. Becker, Microstructural Changes Rail Steels--Developments, Processing, Around Non-Metallic Inclusions Caused and Use, STP 644, American Society for by Rolling-Contact Fatigue of Ball- Testing and Materials, 1976, p 233-255 Bearing Steels, Met. Technol., Vol 8, 140. T. Mitsuda and F.G. Bauling, Research June 1981, p 234-243 on Shelling of Crane Wheels, in Iron and 155. R. Osterlund et al., Butterflies in Fa- Steel Engineering Yearbook, Association tigued Ball Bearings--Formation Mecha- of Iron and Steel Engineers, 1966, p nisms and Structure, Scand. J. Metall., 272-281 Vol 11 (No. 1), 1982, p 23-32 141. S. Neumann and L.E. Arnold, Prediction 156. K. Tsubota and A. Koyanagi, Formation and Analysis of Crane Wheel Service of Platelike Carbides during Rolling Con- Life, in Iron and Steel Engineering Year- tact Fatigue in High-Carbon Chromium book, Association of Iron and Steel Engi- Bearing Steel, Trans. 1SIJ, Vol 25, p neers, 1971, p 102-110 496-504 142. L.E. Arnold, Replicas Enable New Look 157. H.J. Gough, Crystalline Structure in Re- at Roll Surfaces, Iron Steel Eng., Vol 43, lation to Failure of Metals--Especially by (d) Aug 1966, p 129-133 Fatigue, Proc. ASTM, Vol 33, Pt. II, 143. J.J. Bush et al., Microstructural and Re- 1933, p 3-114 Fig. 93 Example of quench cracks on the head of sidual Stress Changes in Hardened Steel 158. N. Thompson et al., The Origin of Fa- AIS11040 steel bolts. Cracks were caused Due to Rolling Contact, Trans. ASM, Vol tigue Fracture in Copper, Philos. Mag., by incompletedevelopment of the case. (a) Bolt heads at 54, 1961, p 390-412 Series 8, Vol 1, 1959, p 113-126 0.72 x ; cracks accentuated using magnetic particles. (b) Quench crack near a corner. Etched with 2% nital. 144. A.J. Gentile et al., Phase Transforma- 159. J.M. Finney and C. Laird, Strain Local- 54 x. (c) Opened quench crack, with arrows indicating tions in High-Carbon, High-Hardness ization in Cyclic Deformation of Copper temper color. 1.5 X. (d) Macrograph showing lack of Steels Under Contact Loads, Trans. Single Crystals, Philos. Mag., Series 8, complete case hardening around head. Actual size A1ME, Vol 233, June 1965, p 1085-1093 Vol 31 (No. 2), Feb 1975, p 339-366 145. A.H. King and J.L. O'Brien, 160. N.M.A. Eid and P.F. Thomason, The Occur, in Mechanical Working & Steel Microstructural Alterations in Rolling Nucleation of Fatigue Cracks in a Low- Processing, Meeting XVIII, American Contact Fatigue, in Advances in Electron Alloy Steel Under High-Cycle Fatigue Institute of Mining, Metallurgical, and Metallography, STP 396, American So- Conditions and Uniaxial Loading, Acta Petroleum Engineers, 1980, p 204-230 ciety for Testing and Materials, 1966, p Metall., Vol 27, July 1979, p 1239-1249 131. A.F. Kaminskas, Antidotes for Sleeve 74-88 161. P.C. Paris, The Fracture Mechanics Ap- 150 / Visual Examination and Light Microscopy

(a) (hi ,u;,-. 94 Cracked 25Cr-12Ni cast stainless steel quenching fixture. (a) Macrograph of part of the fixture. (b) Microstructure showing substantial ~ phase. Electrolytically F etched with 10 N KOH. 500x

proach to Fatigue, in Fatigue~An Inter- Trans. ASM, Vol 62, 1969, p 651-658 Society for Testing and Materials, 1976, disciplinary Approach, 10th Sagamore 170. I. LeMay and M.W. Lui, Fractographic p 220-234 Army Materials Research Conference, Observations of Fatigue Fracture in High- 178. R. Koterazawa et al., Fractographic Syracuse University Press, 1964, p Strength Steels, Metallography, Vol 8, Study of Fatigue Crack Propagation, J. 107-132 1975, p 249-252 Eng. Mater. Technol. (Trans. ASME), 162. W.L. Morris, Microcrack Closure Phe- 171. M.W. Lui and I. LeMay, Fatigue Frac- Oct 1973, p 202-212 nomena for AI 2219-T851, Metall. ture Surface Features: Fractography and 179. B.V. Whiteson et al., Special Fracto- Trans., Vol 10A, Jan 1979, p 5-11 Mechanisms of Formation, in Micro- graphic Techniques for Failure Analysis, 163. M. Kikukawa et al., Direct Observation structural Science, Vol 8, Elsevier, 1980, in Electron Fractography, STP 436, and Mechanism of Fatigue Crack Propa- p 341-352 American Society for Testing and Mate- gation, in Fatigue Mechanisms, STP 675, 172. P.J.E. Forsyth et al., Cleavage Facets rials, 1968, p 151-178 American Society for Testing and Mate- Observed on Fatigue-Fracture Surfaces in 180. A. Yuen et al., Correlations Between rials, 1979, p 234-253 an Aluminum Alloy, J. Inst. Met., Vol Fracture Surface Appearance and Frac- 164. D.L. Davidson and J. Lankford, Dy- 90, 1961-1962, p 238-239 ture Mechanics Parameters for Stage II namic, Real-Time Fatigue Crack Propa- 173. M. Sumita et al., Fatigue Fracture Sur- Fatigue Crack Propagation in Ti-6A1-4V, gation at High Resolution as Observed in faces of Steels Containing Inclusions, Metall. Trans., Vol 5, Aug 1974, p the Scanning Electron Microscope, in Trans. Natl. Res. Inst. for Metals, Vol 14 1833-1842 Fatigue Mechanisms, STP 675, Ameri- (No. 4), 1972, p 146-154 181. A.J. Brothers and S. Yukawa, Engineer- can Society for Testing and Materials, 174. R.M.N. Pelloux, "The Analysis of Frac- ing Applications of Fractography, in 1979, p 277-284 ture Surfaces by Electron Microscopy," Electron Fractography, STP 436, Amer- 165. D.L. Davidson and J. Lankford, Fatigue Report DI-82-0169-R1, Boeing Scientific ican Society for Testing and Materials, Crack Propagation: New Tools for the Research Laboratory, Dec 1963; see also 1968, p 176-195 Study of an Old Problem, J. Met., Vol Technical Report P19-3-64, American 182. R.C. Bates and W.G. Clark, Jr., Fractog- 31, Nov 1979, p 11-16 Society for Metals, Oct 1964 raphy and Fracture Mechanics, Trans. 166. D.L. Davidson, The Study of Fatigue 175. C.D. Beachem, "Electron Microscope ASM, Vol 62, 1969, p 380-389 Mechanisms With Electron Channeling, Fracture Examination to Characterize and 183. R.C. Bates et al., Correlation of Fracto- in Fatigue Mechanisms, STP 675, Amer- Identify Modes of Fracture," Report graphic Features With Fracture Mechan- ican Society for Testing and Materials, 6293 (AFML-TR-64-408), Naval Re- ics Data, in Electron Microfractography, 1979, p 254-275 search Laboratory, 28 Sept 1965 STP 453, American Society for Testing 167. C.A. Zapffe and C.O. Worden, Fracto- 176. C.D. Beachem and D.A. Meyn, "Illus- and Materials, 1969, p 192-214 graphic Registrations of Fatigue, Trans. trated Glossary of Fractographic Terms," 184. E. Gassner, Fatigue Strength. A Basis for ASM, Vol 43, 1951, p 958-969 NRL Memorandum, Report 1547, Naval Measuring Construction Parts With Ran- 168. W.J. Plumbridge and D.A. Ryder, The Research Laboratory, June 1964 dom Loads Under Actual Usage, Kon- Metallography of Fatigue, Met. Rev., No. 177. R.W. Hertzberg and W.J. Mills, Charac- struction, Vol 6, 1954, p 97-104 136, Aug 1969, p 119-142 ter of Fatigue Fracture Surface Micromor- 185. E. Gassner, Effect of Variable Load and 169. G.A. Miller, Fatigue Fracture Appear- phology in the Ultra-Low Growth Rate Cumulative Damage in Vehicle and Air- ance and the Kinetics of Striation Forma- Regime, in Fractographic-Microscopic plane Structures, in International Confer- tion in Some High-Strength Steels, Cracking Processes, STP 600, American ence on Fatigue of Metals, 1956, p Visual Examination and Light Microscopy/ 151

(al

(bl

(b)

Fig. 96 Broken 25-mm (1-in.) diam AISI 1040 steel bolt. (a) Macrograph of fracture (al (c) surface; corrosion products obscure most of the surface. 2 ×. Intergranulor secondary cracks (b) were observed F ig. 9 5 Three views of a fractured specimen of type 312 weld metal that was exposed to high temperatures to in the region near the surface of the bolt shown by the transform the ~ ferrite to ~ phase. The specimen was subsequently broken by impact at room arrow in (a). The bolt was not tempered (surface temperature. (a) Partially broken specimen. Etched with mixed acids. (b) SEM view of fracture. (c) Cross section of hardness was 53 to 57 HRC) and probably failed by fracture. Etched with acetic glyceregia SCC. (b)Etched with 2% nital. 340×

304-309 STP 436, American Society for Testing 1917, p 300-324 186. G. Quest, Quantitative Determination of and Materials, 1968, p 89-123 193. Z. Jeffries, Effect of Temperature, Defor- the Load and the Number of Cycles from 189. P.J.E. Forsyth and D.A. Fyder, Fatigue mation, and Grain Size on the Mechanical the Surface of Fatigue Fractures, Der Fracture. Some Results Derived From the Properties of Metals, Trans. AIME, Vol Maschineenschaden, Vol 33, 1960, p Microscopic Examination of Crack Sur- 60, 1919, p 474-576 4-12, 33-44 faces, Aircr. Eng., Vol 32, April 1960, p 194. H.C. Chang and N.J. Grant, Mechanisms 187. E.E. Underwood and E.A. Starke, Jr., 96-99 of Intercrystalline Fracture, Trans. Quantitative Stereological Methods for 190. C.E. Price and D. Cox, Observing Fa- AIME, Vol 206, 1956, p 544-551 Analyzing Important Microstructural Fea- tigue With The Nomarski Technique, 195. J.N. Greenwood, Intercrystalline Crack- tures in Fatigue of Metals and Alloys, in Met. Prog., Vol 123, Feb 1983, p 37-39 ing of Metals, J. Iron Steel Inst., Vol Fatigue Mechanisms, STP 675, Ameri- 191. W. Rosenhain and D. Ewen, The Inter- 171, Aug 1952, p 380 can Society for Testing and Materials, crystalline Cohesion of Metals, J. Inst. 196. J.N. Greenwood, Intercrystalline Crack- 1979, p 633-682 Met., Vol 10, 1913, p 119-149 ing of Metals, Bull. Inst. Met., Vol 1, Pt. 188. J.C. McMillan and R.W. Hertzberg, Ap- 192. Z. Jeffries, The Amorphous Metal Hy- 12, Aug 1952, p 104-105; [ntercrystalline plication of Electron Fractography to Fa- pothesis and Equicohesive Temperatures, Cracking of Brass, Bull. Inst. Met., Vol tigue Studies, in Electron Fractography, J. Am. Inst. Met., Vol 11 (No. 3), Dec 1, Pt. 14, Oct 1952, p 120-121 152 / Visual Examination and Light Microscopy

(a) (b)

Fig. 98 Two views of the crack path, which was predominantly tronsgranular, in an austenitic manganese- chromium stainless steel drill collar alloy. The SCC was caused by chlorides in the drilling fluid. Cracking began at the inside bore surface. Etched with acetic glyceregia. Both 65 x

1a1 num-Copper, Aluminum-Magnesiumand 215. P.W. Davies etal., On the Distribution of Aluminum-Zinc Alloys, J. Inst. Met., Cavities During Creep, Philos. Mag., Vol 83, 1954-1955, p 1-10 Vol 18 (No. 151), July 1968, p 197-200 203. D. McLean, Deformation at High Tem- 216. T. Johannesson and A. Tholen, Cavity peratures, Met. Rev., Vol 7 (No. 28), Formation in Copper and in a Steel Dur- 1962, p 481-527 ing Creep, J. Inst. Met., Vol 97, 1969, p 204. R.C. Gifkins and T.G. Langdon, On The 243-247 Question of Low-Temperature Sliding at 217. D.MR. Taplin, A Note on the Distribu- Grain Boundaries, J. Inst. Met., Vol 93, tion of Cavities During Creep, Philos. 1964-1965, p 347-352 Mag., Vol 20 (No. 167), Nov 1967, p 205. C.M. Sellars, Estimation of Slip Strain of 1079-1982 Interior Grains During Creep, J. Inst. 218. V.V.P. Kutumbarao and P. Rama Rao, Met., Vol 93, 1964-1965, p 365-366 On the Determination of the Distribution 206. Y. Ishida et al., Internal Grain Boundary of Creep Cavities, Metallography, Vol 5, Sliding During Creep, Trans. AIME, Vol 1972, p 94-96 (b) 233, Jan 1965, p 204-212 219. B.J. Cane, Creep-Fracture Initiation in 207. F.N. Rhines et al., Grain Boundary 2-1/4% Cr-l%Mo Steel, Met. Sci., Vol Fig. 97 Type 304 stainless steel specimen offer Creep in Aluminum Bicrystals, Trans. 10, Jan 1976, p 29-34 testing in boiling MgCI 2. (o) Cross section ASM, Vol 48, 1956, p 919-951 220. D.A. Miller and R. Pilkington, The Ef- of partially broken specimen. Etched with mixed acids. 208. R.N. Stevens, Grain Boundary Sliding in fect of Temperature and Carbon Content (b) SEM fractograph of completely broken specimen Metals, Met. Rev., Vol 11, Oct 1966, p on the Cavitation Behavior of a 1.5 Pet 129-142 Cr-0.5 Pet V Steel, Metall. Trans., Vol 197. W. Pavinich and R. Raj, Fracture at 209. R.L. Bell and T.G. Langdon, An Inves- 9A, April 1978, p 489-494 Elevated Temperatures, Metall. Trans., tigation of Grain-Boundary Sliding Dur- 221. R.A. Scriven and H.D. Williams, The Vol 8A, Dec 1977, p 1917-1933 ing Creep, J. Mater. Sci., Vol 2, 1967, p Derivation of Angular Distributions of 198. W.A. Rachinger, Relative Grain Transla- 313-323 Planes by Sectioning Methods, Trans. tions in the Plastic Flow of Aluminum, J. 210. R.L. Bell et al., The Contribution of AIME, Vol 233, Aug 1965, p 1593-1602 Inst. Met., Vol 81, 1952-1953, p 33-41 Grain Boundary Sliding to the Overall 222. D.M.R. Taplin and L.J. Barker, A Study 199. J.A. Martin et al., Grain-Boundary Dis- Strain of a Polycrystal, Trans. AIME, Vol of the Mechanism of lntergranular Creep placement Vs. Grain Deformation as 239, Nov 1967, p 1821-1824 Cavitation by Shadowgraphic Electron the Rate-Determining Factor in Creep, 211. T.G. Langdon and R.L. Bell, The Use of Microscopy, Acta Metall., Vol 14, Nov Trans. AIME, Vol 209, Jan 1957, p 78- Grain Strain Measurements in Studies of 1966, p 1527-1531 81 High-Temperature Creep, Trans. AIME, 223. G.J. Cocks and D.M.R. Taplin, An Ap- 200. D. McLean and M.H. Farmer, The Rela- Vol 242, Dec 1968, p 2479-2484 praisal of Certain Metallographic Tech- tion During Creep Between Grain- 212. P.W. Davies and B. Wilshire, An Exper- niques for Studying Cavities, Metal- Boundary Sliding, Sub-Crystal Size, and iment on Void Nucleation During Creep, lurgia, Vol 75 (No. 451), May 1967, p Extension, J. Inst. Met., Vol 85, J. Inst. Met., Vol 90, 1961-1962, p 229-235 1956-1957, p 41-50 470-472 224. D.M.R. Taplin and A.L. Wingrove, 201. H.C. Chang and N.J. Grant, Observa- 213. R.V. Day, Intercrystalline Creep Failure Study of Intergranular Cavitation in Iron tions of Creep of the Grain Boundary in in l%Cr-Mo Steel, J. Iron Steel Inst., by Electron Microscopy of Fracture Sur- High Purity Aluminum, Trans. AIME, Vol 203, March 1965, p 279-284 faces, Acta Metall., Vol 15, July 1967, p Vol 194, June 1952, p 619-625 214. A. Gittins and H.D. Williams, The Effect 1231-1236 202. D. McLean and M.H. Farmer, Grain- of Creep Rate on the Mechanism of Cav- 225. K. Farrell and J.O. Stiegler, Electron Boundary Movement, Slip, and Frag- ity Growth, Philos. Mag., Vol 16 (No. Fractography for Studying Cavities, Met- mentation During Creep of Alumi- 142), Oct 1967, p 849-851 allurgia, Vol 79 (No. 471), Jan 1969, p Visual Examination and Light Microscopy/ 153

Fig. 99 Microstructures of AISI 4140 steel with 0.004% P (left column), 0.013% P (center column), and 0.022% P (right column). Specimens were etched with the etheral-picral etchant described in Ref 442 (top row) and Ref 444 (middle row) and with saturated aqueous picric acid plus a wetting agent (bottom row). All 425 x

35-37 1969, p 789-796 Vol 4, Sept 1970, p 167-170 226. A.L. Wingrove and D.M.R. Taplin, The 227. H.R. Tipler et al., Some Direct Observa- 228. W.E. White and I. LeMay, Metallo- Morphology and Growth of Creep Cavi- tions on the Metallography of Creep- graphic and Fractographic Analyses of ties in c~-Iron, J. Mater. Sci., Vol 4, Sept Cavitated Grain Boundaries, Met. Sci. J., Creep Failure in Stainless Steel Weld 154 / Visual Examination and Light Microscopy

Applied Techniques of In-Place Analysis, in Corrosion, Microstructure, & Metal- Iography, Vol 12, Microstructural Sci- ence, American Society for Metals and the International Metallographic Society, 1985, p 537-549 244. M.C. Murphy and G.D. Branch, Metal- lurgical Changes in 2.25 CrMo Steels During Creep-Rupture Test, J. Iron Steel Inst., Vol 209, July 1971, p 546-561 245. J.M. Leitnaker and J. Bentley, Precipitate Phases in Type 321 Stainless Steel After Aging 17 Years at --600 °C, Metall. Trans., Vol 8A, Oct 1977, p 1605-1613 (a) (bl 246. M. McLean, Microstructural Instabilities in Metallurgical Systems---A Review, Met. Sci., Vol 12, March 1978, p 113-122 247. S. Kihara et al., Morphological Changes of Carbides During Creep and Their Ef- fects on the Creep Properties of Inconel 617 at 1000 °C, Metall. Trans., Vol 11A, June 1980, p 1019-1031 248. S.F. Claeys and J.W. Jones, Role of Microstructural Instability in Long Time Creep Life Prediction, Met. Sci., Vol 18, Sept 1984, p 432-438 249. Y. Minami et al., Microstructural Changes in Austenitic Stainless Steels During Long-Term Aging, Mater. Sci. [cl Id) Technol., Vol 2, Aug 1986, p 795-806 250. J.R. Low, Jr., Impurities, Interfaces and "'~11~|~" 1 O0 Fracture in a thermally embrittled cobalt-free high-titanium moraging steel. (a) Secondary electron Brittle Fracture, Trans. AIME, Vol 245, image of fracture surface. 1300 x. (b) TEM extraction fractograph. 2150 ×. (c) Light micrograph Dec 1969, p 2481-2494 of fracture edge. 260 x. (d) Light micrograph of internal cracks. 260 x. Light micrograph specimens etched with 251. W.P. Rees and B.E. Hopkins, Inter- modified Fry's reagent granular Brittleness in Iron-Oxygen Al- loys, J. Iron Steel Inst., Vol 172, Dec ments, in Microstructural Science, Vol 5, ation and Growth of Creep Cavities in a 1952, p 403-409 Elsevier, 1977, p 145-160 Type 347 Steel, Met. Sci., Vol 14, Feb 252. J.R. Low, Jr. and R.G. Feustel, Inter- 229. V.K. Sikka et al., Twin-Boundary Cavi- 1980, p 64-72 Crystalline Fracture and Twinning of Iron tation During Creep in Aged Type 304 237. Y. Lindblom, Refurbishing Superalloy at Low Temperatures, Acta Metall., Vol Stainless Steel, Metall. Trans., Vol 8A, Components for Gas Turbines, Mater. 1, March 1953, p 185-192 July 1977, p 1117-1129 Sci. Teehnol., Vol 1, Aug 1985, p 636-641 253. B.E. Hopkins and H.R. Tipler, Effect of 230. D.G. Morris and D.R. Harries, Wedge 238. J. Wortmann, Improving Reliability and Heat-Treatment on the Brittleness of Crack Nucleation in Type 316 Stainless Lifetime of Rejuvenated Turbine Blades, High-Purity Iron-Nitrogen Alloys, J. Iron Steel, J. Mater. Sci., Vol 12, Aug 1977, Mater. Sci. Technol., Vol 1, Aug 1985, p Steel Inst., Vol 177, May 1954, p p 1587-1597 644-650 110-117 231. W.M. Stobbs, Electron Microscopical 239. C.J. Bolton et al., Metallographic Meth- 254. B.E. Hopkins and H.R. Tipler, The Ef- Techniques for the Observation of Cavi- ods of Determining Residual Creep Life, fect of Phosphorus on the Tensile and ties, J. Microsc., Vol 116, Pt. 1, May Mater. Sci. Eng., Vol 46, Dec 1980, p Notch-lmpact Properties of High-Purity 1979, p 3-13 231-239 Iron and Iron-Carbon Alloys, J. Iron 232. R.J. Fields and M.F. Ashby, Observation 240. R. Sandstrom and S. Modin, "The Re- Steel Inst., Vol 188, March 1958, p on Wedge Cavities in the SEM, Scr. sidual Lifetime of Creep Deformed Com- 218-237 Metall., Vol 14 (No. 7), July 1980, p ponents. Microstructural Observations for 255. A.R. Troiano, The Role of Hydrogen and 791-796 Mo- and CrMo-Steels," Report IM-1348, Other Interstitials in the Mechanical Be- 233. A.J. Perry, Cavitation in Creep, J. Swedish Institute for Metals Research, havior of Metals, Trans. ASM, Vol 52, Mater. Sci., Vol 9, June 1974, p 1016- 1979 1960, p 54-80 1039 241. C. Bengtsson, "Metallographic Methods 256. C. Pichard et al., The Influence of Oxy- 234. B.F. Dyson and D. McLean, A New for Observation of Creep Cavities in Ser- gen and Sulfur on the Intergranular Brit- Method of Predicting Creep Life, Met. vice Exposed Low-Alloyed Steel," Re- tleness of Iron, Metall. Trans., Vol 7A, Sci. J., Vol 6, 1972, p 220-223 port IM-1636, Swedish Institute for Met- Dec 1976, p 1811-1815 235. B. Walser and A. Rosselet, Determining als Research, March 1982 257. M.C. Inman and H.R. Tipler, Grain- the Remaining Life of Superheater-Steam 242. J.F. Henry and F.V. Ellis, "Plastic Rep- Boundary Segregation of Phosphorus in Tubes Which Have Been in Service by lication Techniques for Damage Assess- an Iron-Phosphorus Alloy and the Ef- Creep Tests and Structural Examinations, ment," Report RP2253-01, Electric fect Upon Mechanical Properties, Acta Sulzer Res., 1978, p 67-72 Power Research Institute, Sept 1983 Metall., Vol 6, Feb 1958, p 73-84 236. N.G. Needham and T. Gladman, Nucle- 243. J.F. Henry, Field Metallography. The 258. G.T. Hahn et al., "The Effects of Solutes Visual Examination and Light Microscopy/155

2 1 3 3

(a)

(a) (b)

(b)

Direction of welding

(c) (d) (c) Fig. 102 Schematics illustrating the nomenclature Fig. 10 ! Example of the fracture appearance and microstructure of a duplex stainless steel (similar to type used to describe weldment cracks. (a) 329) in the embrittled condition after heating at 370 °C (700 °F) for 1000 h. (a) and (b) SEM Weld metal (1), fusion line (2), and HAZ (3) cracks. (b) fractographs at 165 and 650 x, respectively, showing cleavage fracture with some splitting. (c) Light micrograph Underbeod crack. (c) Longitudinal and transverse weld showing dark-etching ferrite-ferrite grain boundaries in a specimen subjected to the embrittling treatment. (d) Light metal cracks. Source: Ref 480 micrograph of a hot-rolled and annealed nonembrittled specimen, which shows only carbides. (c) and (d) Etched with Vilella's reagent. Both at 260 x 1949-1950, p 91-102 on the Ductile-to-Brittle Transition in Re- Sulfur Embrittlement in Nickel Alloys, 270. C.W. Spencer et al., Bismuth in Copper fractory Metals," DMIC Memorandum Scr. Metall., Vol 8, Aug 1974, p 971- Grain Boundaries, Trans. AIME, Vol 155, Battelle Memorial Institute, 28 June 974 209, June 1957, p 793-794 1962 265. C. Loier and J.Y. Boos, The Influence of 271. D. McLean and L. Northcott, Antimonial 259. R.E. Maringer and A.D. Schwope, On Grain Boundary Sulfur Concentration on 70:30 Brass, J. Inst. Met., Vol 72, 1946, the Effects of Oxygen on Molybdenum, the Intergranular Brittleness of Nickel of p 583-616 Trans. A1ME, Vol 200, March 1954, p Different Purities, Metall. Trans., Vol 272. D. McLean, The Embrittlement of Cop- 365-366 12A, July 1981, p 1223-1233 per: Antimony Alloys at Low Tempera- 260. T.G. Nieh and W.D. Nix, Embrittlement 266. J.H. Westbrook and D.L. Wood, A tures, J. Inst. Met., Vol 81, 1952-1953, p of Copper Due to Segregation of Oxygen Source of Grain Boundary Embrittlemeut 121-123 to Grain Boundaries, Metall. Trans., Vol in Intermetallics, J. Inst. Met., Vol 91, 273. R. Carlsson, Hot Embrittlement of Cop- 12A, May 1981, p 893-901 1962-1963, p 174-182 per and Brass Alloys, Scand. J. Metall., 261. R.H. Bricknell and D.A. Woodford, The 267. E. Voce and A.P.C. Hallowes, The Vol 9 (No. 1), 1980, p 25-29 Embrittlement of Nickel Following High Mechanism of the Embrittlement of 274. H.K. Ihrig, The Effect of Various Ele- Temperature Air Exposure, Metall. Deoxidized Copper by Bismuth, J. Inst. ments on the Hot-Workability of Steel, Trans., Vol 12A, March 1981, p 425-433 Met., Vol 73, 1947, p 323-376 Trans. AIME, Vo! 167, 1946, p 749-790 262. K.M. Olsen et al., Embrittlement of High 268. T.H. Schofield and F.W. Cuckow, The 275. J.M. Middletown and H.J. Protheroe, Purity Nickel, Trans. ASM, Vol 53, Microstructure of Wrought Non- The Hot-Tearing of Steel, J. Iron Steel 1961, p 349-358 Arsenical Phosphorus-Deoxidized Cop- Inst., Vol 168, Aug 1951, p 384-400 263. S. Floreen and J.H. Westbrook, Grain per Containing Small Quantities of Bis- 276. C.T. Anderson et al., Effect of Various Boundary Segregation and the Grain Size muth, J. Inst. Met., Vol 73, 1947, p Elements on Hot-Working Characteristics Dependence of Strength of Nickel-Sulfur 377-384 and Physical Properties of Fe-C Alloys, Alloys, Acta Metall., Vo! 17, Sept 1969, 269. L.E. Samuels, The Metallography of J. Met., Vol 5, April 1953, p 525-529 p 1175-1181 Copper Containing Small Amounts of 277. C.T. Anderson et al., Forgeability of 264. W.C. Johnson et al., Confirmation of Bismuth, J. Inst. Met., Vol 76, Steels with Varying Amounts of Manga- 156 / Visual Examination and Light Microscopy

(al (b) ,.=,~:;,-, 103 Cold cracks in on RQC-90 steel plate welded with a high-hydrogen electrode. The sample was an implant specimen loaded to 193 MPo (28 ksi) during solidification. (a) Light micrograph showing cracking. Etched with nital. 80 x. (b) SEM fractograph showing the intergranular nature of the cracks. 200 x. (J.P. Snyder, Bethlehem Steel Corporation)

nese and Sulfur, Trans. AIME, Vol 200, Metall. Trans., Vol l lA, June 1980, p Nitrogen, J. Iron Steel Inst., Vol 200, July 1954, p 835-837 919-934 April 1962, p 299-307 278. D. Smith et al., Effects of Composition 287. R.A. Perkins and W.O. Binder, Improv- 297. N.H. Croft et al., lntergranular Fracture on the Hot Workability of Resulphurized ing Hot-Ductility of 310 Stainless, J. of Steel Castings, in Advances in the Free-Cutting Steels, J. Iron Steel Inst., Met., Vol 9, Feb 1957, p 239-245 Physical Metallurgy and Applications of Vol 210, June 1972, p 412-421 288. L.G. Ljungstr6m, The Influence of Trace Steels, Publication 284, The Metals Soci- 279. W.J. McG. Tegart and A. Gittins, The Elements on the Hot Ductility of Austenit- ety, 1982, p 286-295 Role of Sulfides in the Hot Workability of ic 17Crl3NiMo-Steel, Scand. J. Metall., 298. E. Colombo and B. Cesari, The Study of Steels, in Sulfide Inclusions in Steel, Vol 6, 1977, p 176-184 the Influence of AI and N on the Suscep- American Society for Metals, 1975, p 289. W.B. Kent, Trace-Element Effects in tibility to Crack Formation of Medium 198-211 Vacuum-Melted Superalloys, J. Vac. Sci. Carbon Steel Ingots, Metall. ltal., Vol 59 280. A. Josefsson et al., The Influence of Technol., Vol 11, Nov/Dec 1974, p (No. 2), 1967, p 71-75 Sulphur and Oxygen in Causing Red- 1038-1046 299. S.C. Desai, Longitudinal Panel Cracking Shortness in Steel, J. Iron Steel Inst., Vol 290. R.T. Holt and W. Wallace, Impurities in Ingots, J. Iron Steel Inst., Vol 191, 191, March 1959, p 240-250 and Trace Elements in Nickel-Base March 1959, p 250-256 281. P. Bjornson and H. Nathorst, A Special Superalloys, Int. Met. Rev., Vol 21, 300. R. Sussman et al., Occurrence and Con- Type of Ingot Cracks Caused by Certain March 1976, p 1-24 trol of Panel Cracking in Aluminum Con- Impurities, Jernkontorets Ann., Vol 139, 291. A.R. Knott and C.H. Symonds, Compo- taining Steel Heats, in Mechanical Work- 1955, p 412-438 sitional and Structural Aspects of Pro- ing & Steel Processing, Meeting XVII, 282. W.J.M. Salter, Effect of Mutual Addi- cessing Nickel-Base Alloys, Met. Tech- American Institute of Mining, Metallur- tions of Tin and Nickel on the Solubility nol., Vol 3, Aug 1976, p 370-379 gical, and Petroleum Engineers, 1979, p and Surface Energy of Copper in Mild 292. C.H. Lorig and A.R. Elsea, Occurrence 49-78 Steel, J. Iron Steel Inst., Vol 207, Dec of Intergranular Fracture in Cast Steels, 301. L. Ericson, Cracking in Low Alloy Alu- 1969, p 1619-1623 Trans. AFS, Vol 55, 1947, p 160-174 minum Grain Refined Steels, Scand. J. 283. W.J. Jackson and D.M. Southall, Effect 293. B.C. Woodflne, "First Report on Inter- Metall., Vol 6, 1977, p 116-124 of Copper and Tin Residual Amounts on granular Fracture in Steel Castings," 302. F. Vodopivec, Influence of Precipitation the Mechanical Properties of 1.5Mn-Mo BSCRA Report 38/54/FRP.5, British and Precipitates of Aluminum Nitride on Cast Steel, Met. Technol., Vol 5, Pt. 11, Steel Casting Research Association, Torsional Deformability of Low-Carbon Nov 1978, p 381-390 March 1954 Steel, Met. Technol., Vol 5, April 1978, 284. K. Born, Surface Defects in the Hot 294. B.C. Woodfine, Effect of A1 and N on the p 118-121 Working of Steel, Resulting from Resid- Occurrence of Intergranular Fracture in 303. G.D. Funnell and R.J. Davies, Effect of ual Copper and Tin, Stahl Eisen, Vol 73 Steel Castings, J. Iron Steel Inst., Vol Aluminum Nitride Particles on Hot Duc- (No. 20), BISI 3255, 1953, p 1268-1277 195, Aug 1960, p 409-414 tility of Steel, Met. Technol., Vol 5, May 285. I.S. Brammar et al., The Relation Be- 295. R.F. Harris and G.D. Chandley, High 1978, p 150-153 tween Intergranular Fracture and Sul- Strength Steel Castings. Aluminum Ni- 304. T. Lepist6 and P. Kettunen, Embrittle- phide Precipitation in Cast Alloy Steels, tride Embrittlement, Mod. Cast., March ment Caused by e-Martensite in Stainless in IS1 64, Iron and Steel Institute, 1959, p 1962, p 97-103 Steels, Scand. J. Metall., Vol 7, 1978, p 187-208 296. J.A. Wright and A.G. Quarrell, Effect of 71-76 286. D. Bhattacharya and D.T. Quinto, Chemical Composition on the Occurrence 305. F.W. Schaller and V.F. Zackay, Low Mechanism of Hot-Shortness in Leaded of Intergranular Fracture in Plain Carbon Temperature Embrittlement of Austenitic and Tellurized Free-Machining Steels, Steel Castings Containing Aluminum and Cr-Mn-N-Fe Alloys, Trans. ASM, Vol Visual Examination and Light Microscopy/ 157

(a)

(b) (c) (d)

F i g. 104 Hot cracking of an electron beam weld due to accidental melting of the copper backing plate. Note the extensive intergranu[ar cracking and the grain-boundary copper film. (a) 4 x. (b) 13 x. (c) 68 x. (d) 340 x

51, 1959, p 609-628 311. R.A. Swift and H.C. Rogers, Study of 265-278 306. D. Hennessy et al., Phase Transformation Creep Embrittlement of 2-1/4Cr-lMo 316. H.J. Kerr and F. Eberle, Graphitization of Stainless Steel During Fatigue, Metall. Steel Weld Metal, Weld. J., Vol 55, July of Low-Carbon and Low-Carbon- Trans., Vol 7A, March 1976, p 415-424 1976, p 188s-198s Molybdenum Steels, Trans. ASME, Vol 307. H.H. Bleakney, The Creep-Rupture Em- 312. R.A. Swift, The Mechanism of Creep 67, 1945, p 1-46 brittlement of Metals as Exemplified by Embrittlement in 2-1/4Cr-lMo Steel, in 317. S.L. Hoyt et al., Summary Report on the Aluminum, Can. Metall. Q., Vol 2 (No. 2-1/4 Chrome-1 Molybdenum Steel in Joint E.E.I.-A.E.I.C. Investigation of 3), 1963, p 391-315 Pressure Vessels and Piping, American Graphitization of Piping, Trans. ASME, 308. D.E. Ferrell and A.W. Pense, Creep Society of Mechanical Engineers, 1971 Aug 1946, p 571-580 Embrittlement of 2-1/4% Cr-l% Mo 313. L.K.L. Tu and B.B. Seth, Effect of 318. R.W. Emerson and M. Morrow, Further Steel, in Report to Materials Division, Composition, Strength, and Residual Observations of Graphitization in Alumi- Pressure Vessel Research Council, May Elements on Toughness and Creep Prop- num-Killed Carbon-Molybdenum Steel 1973 erties of Cr-Mo-V Turbine Rotors, Met. Steel Steam Piping, Trans. A1ME, Aug 309. H.R. Tipler, "The Role of Trace Ele- Technol., Vol 5, March 1978, p 79-91 1946, p 597-607 ments in Creep Embrittlement and Cavi- 314. S.H. Chen et al., The Effect of Trace 319. J.G. Wilson, Graphitization of Steel in tation of Cr-Mo-V Steels," National Impurities on the Ductility of a Cr-Mo-V Petroleum Refining Equipment, Weld. Physical Laboratory, 1972 Steel at Elevated Temperatures, Metall. Res. Counc. Bull., No. 32, Jan 1957, p 310. R. Bruscato, Temper Embrittlement and Trans., Vol 14A, April 1983, p 571- 1-10 Creep Embrittlement of 2-1/4Cr-lMo 580 320. A.B. Wilder et al., Stability of AISI Shielded Metal Arc Weld Deposits, 315. M.P. Seah, Impurities, Segregation and Alloy Steels, Trans. A1ME, Vol 209, Oct Weld. J., Vol 49, April 1970, p 148s- Creep Embrittlement, Philos. Trans. R. 1957, p 1176-1181 156s Soc. (London) A, Vol 295, 1980, p 321. I.M. Bernstein and A.W. Thompson, 158 / Visual Examination and Light Microscopy

(a) (b)

Two examples of hot tears in the HAZ of gas-metal arc welded HY-80 steel. Note the crack Fig. 105 associated with the manganese sulfide inclusion (b). Both etched with 1% nital. (a) 370x. (b) 740 ×. Courtesy of C.F. Meitzner, Bethlehem Steel Corporation

Shear wall (located at terraces Principal tensile strength (in z direction) / at different levels) [~ Terrace [. < • < Terrace • (complete fracture through inclusions Elastic stress and ligaments concentration between inclusion) I I i at tips ...... ~-~ ...... : Decohsion Decohesion at inclusion I -- r~ r ...... \ ...... --//~'~'~l~///~x, x Ductile fracture / ~ / Heavy plastic of ligament Cohesion between steel ~ Plastic zone at Heavy plastic shearing between between inclusions and inclusion inclusion tips straining of terraces ligament

Fig. ! 06 Schematic illustrating the formation of lamellar tears. Source: Ref 499

Ed., Hydrogen in Metals, American So- 327. I.M. Bernstein and A.W. Thompson, Ef- ciety for Metals, 1974 fect of Metallurgical Variables on Envi- 322. C.D. Beachem, Ed., Hydrogen Damage, ronmental Fracture of Steels, Int. Met. Fig. ! 07 Macrographs showing typical lamellar American Society for Metals, 1977 Rev., Vol 21, Dec 1976, p 269-287 tears in structural steel 323. R.W. Staehle et al., Ed., Stress Corro- 328. J.P. Hirth, Effects of Hydrogen on the sion Cracking and Hydrogen Embrittle- Properties of Iron and Steel, Metall. ment in Iron Base Alloys, NACE Refer- Trans., Vol 11A, June 1980, p 861-890 ence Book 5, National Association of 329. I.M. Bernstein et al., Effect of Dissolved 332. H.G. Nelson, Hydrogen Embrittlement, Corrosion Engineers, 1977 Hydrogen on Mechanical Behavior of in Embrittlement of Engineering Alloys, 324. A.E. Schuetz and W.D. Robertson, Hy- Metals, in Effect of Hydrogen on Behav- Academic Press, 1983, p 275-359 drogen Absorption, Embrittlement and ior of Materials, American Institute of 333. C.A. Zapffe, Hydrogen Flakes and Shat- Fracture of Steel, Corrosion, Vol 13, July Mining, Metallurgical, and Petroleum ter Cracks, Metals and Alloys, Vol 11, 1957, p 437t-458t Engineers, 1976, p 37-58 May 1940, p 145-151; June 1940, p 325. I.M. Bernstein, The Role of Hydrogen in 330. L.H. Keys, The Effects of Hydrogen on 177-184; Vol 12, July 1940, p 44-51; the Embrittlement of Iron and Steel, the Mechanical Behavior of Metals, Met. Aug 1940, p 145-148 Mater. Sci. Eng., Vo16, July 1970, p 1-19 Forum, Vol 2 (No. 3), 1979, p 164-173 334. C.A. Zapffe, Defects in Cast and 326. M.R. Louthan, Jr. et al., Hydrogen Em- 331. S.P. Lynch, Mechanisms of Hydrogen- Wrought Steel Caused by Hydrogen, brittlement of Metals, Mater. Sci. Eng., Assisted Cracking, Met. Forum, Vol 2 Met. Prog., Vol 42, Dec 1942, p Vol 10, 1972, p 357-368 (No. 3), 1979, p 189-200 1051-1056 Visual Examination and Light Microscopy / 159

J. Iron Steel Inst., Vol 191, April 1959, p 353. K. Farrell and A.G. Quarrell, Hydrogen 342-352 Embrittlement of an Ultra-High-Tensile 338. J.M. Hodge et al., Effect of Hydrogen Steel, J. Iron Steel Inst., Vol 202, Dec Content on Susceptibility to Flaking, 1964, p 1002-1011 Trans. A1ME, Vol 230, 1964, p 354. A. Ciszewski et al., Effect of Nonmetal- 1182-1193 lic Inclusions on the Formation of Micro- 339. J.E. Ryall etal., The Effects of Hydrogen cracks in Hydrogen Charged 1% Cr Steel, in Rolled Steel Products, Met. Forum, in Stress Corrosion Cracking and Hydro- Vol 2 (No. 3), 1979, p 174-182 gen Embrittlement of Iron Base Alloys, 340. R.E. Cramer and E.C. Bast, The Preven- NACE Reference Book 5, National Asso- tion of Flakes by Holding Railroad Rails ciation of Corrosion Engineers, 1977, p at Various Constant Temperatures, 671-679 Trans. ASM, Vol 27, 1939, p 923-934 355. G.M. Presouyre and C. Zmudzinski, In- (a) 341. F. Terasaki and S. Okamoto, Fractogra- fluence of Inclusions on Hydrogen Em- phy of Hydrogen-Embrittled Steels, brittlement, in Mechanical Working & Tetsu-to-Hagand (J. Iron Steel Inst. Steel Processing, Meeting XVIII, Amer- Jpn.), Vol 60, 1974, p $576 (HB 9225) ican Institute of Mining, Metallurgical, 342. A.P. Coldren and G. Tither, Metallo- and Petroleum Engineers, 1980, p graphic Study of Hydrogen-Induced 534-553 Cracking in Line Pipe Steel, J. Met., Vol 356. R.L. Cowan II and C.S. Tedmon, Jr., 28, May 1976, p 5-10 Intergranular Corrosion of Iron-Nickel- 343. M.R. Louthan, Jr. and R.P. McNitt, The Chromium Alloys, in Advances in Corro- Role of Test Technique in Evaluating sion Science and Technology, Vol 3, Hydrogen Embrittlement Mechanisms, in Plenum Press, 1973, p 293-400 Effects of Hydrogen on Behavior of Ma- 357. E.M. Mahla and N.A. Nielson, Carbide terials, American Institute of Mining, Precipitation in Type 304 Stainless Metallurgical, and Petroleum Engineers, Steel--An Electron Microscope Study, 1976, p 496-506 Trans. ASM, Vol 43, 1951, p 290-322 344. C.L. Briant et al., Embrittlement of a 5 358. R. Stickler and A. Vinckier, Morphology Pct. Nickel High Strength Steel by Impu- of Grain-Boundary Carbides and its Influ- rities and Their Effects on Hydrogen- ence on Intergranular Corrosion of 304 Induced Cracking, Metall. Trans., Vol Stainless Steel, Trans. ASM, Vol 54, 9A, May 1978, p 625-633 1961, p 362-380 345. D. Hardie and T.I. Murray, Effect of 359. R. Stickler and A. Vinckier, Precipitation Hydrogen on Ductility of a High Strength of Chromium Carbide on Grain Bound- Steel in Hardened and Tempered Condi- aries in a 302 Austenitic Stainless Steel, tions, Met. Technol., Vol 5, May 1978, p Trans. AIME, Vol 224, Oct 1962, p 145-149 1021-1024 346. T.A. Adler et al., Metallographic Studies 360. K.T. Aust et al., Intergranular Corrosion of Hydrogen-Induced Crack Growth in and Electron Microscopic Studies of Aus- A- 106 Steel, in Microstructural Science, tenitic Stainless Steels, Trans. ASM, Vol Vol 8, Elsevier, 1980, p 217-230 60, 1967, p 360-372 347. J.E. Costa and A.W. Thompson, Effect 361. S. Danyluk et al., Intergranular Fracture, of Hydrogen on Fracture Behavior of a Corrosion Susceptibility, and Impurity Quenched and Tempered Medium- Segregation in Sensitized Type 304 Stain- Carbon Steel, Metall. Trans., Vol 12A, less Steel, J. Mater. Energy Syst., Vol 7, May 1981, p 761-771 June 1985, p 6-15 348. B.D. Craig, A Fracture Topographical 362. A.P. Bond and E.A. Lizlovs, Intergran- Feature Characteristic of Hydrogen Em- ular Corrosion of Ferritic Stainless Steels, (b) brittlement, Corrosion, Vol 37, Sept J. Electrochem. Soc., Vol 116, Sept 1981, p 530-532 1969, p 1305-1311 Fig. 1 08 Stress-relief crack in plate steel Lehigh restraint sample. The crack is intergran- 349. R. Garber et al., Hydrogen Assisted Duc- 363. A.P. Bond, Mechanisms of Intergranular ular, but the high restraint level suppressescrack branch- tile Fracture of Spheroidized Carbon Corrosion in Ferritic Stainless Steels, ing. The square marks in the weld HAZ in (a) are Steels, Metall. Trans., Vol 12A, Feb Trans. AIME, Vol 245, Oct 1969, p hardness impressions. (a) 9 x. (b) Etched with 2% nital. 1981, p 225-234 74 × 2127-2134 350. H. Cialone and R.J. Asaro, Hydrogen 364. E.L. Hall and C.L. Briant, Chromium Assisted Ductile Fracture of Spheroidized Depletion in the Vicinity of Carbides in Plain Carbon Steels, Metall. Trans., Vol Sensitized Austenitic Stainless Steels, 335. E.R. Johnson et al., Flaking in Alloy 12A, Aug 1981, p 1373-1387 Metall. Trans., Vol 15A, May 1984, p Steels, in Open Hearth Conference, 351. T. Boniszewski and J. Moreton, Effect of 793-811 1944, p 358-377 Micro-Voids and Manganese Sulphide In- 365. J.B. Lee et al., An Analytical Electron 336. A.W. Dana et al., Relation of Flake clusion in Steel on Hydrogen Evolution Microscope Examination of Sensitized Formation in Steel to Hydrogen, Micro- and Embrittlement, Br. Weld. J., Vol 14, AISI 430 Stainless Steel, Corrosion, Vol structure, and Stress, Trans. AIME, Vol June 1967, p 321-336 41, Feb 1985, p 76-80 203, Aug 1955, p 895-905 352. P.H. Pumphrey, Effect of Sulphide Inclu- 366. E.P. Butler and M.G. Burke, Chromium 337. J.D. Hobson, The Removal of Hydrogen sions on Hydrogen Diffusion in Steels, Depletion and Martensite Formation at by Diffusion from Large Masses of Steel, Met. Sci., Vol 16, Jan 1982, p 41-47 Grain Boundaries in Sensitized Austenitic 160 Visual Examination and Light Microscopy

Fig. 109 Fracture surface of a small ingot of iron-chromium-aluminum alloy. The as-cast solidification pattern is clearly revealed. 0.57 ×

Induced Brittle Fracture in Liquid-Metal Environments, in Fracture, Vol III, Aca- demic Press, 1971, p 589-644 374. M.H. Kamdar, Embrittlement by Liquid Metals, Prog. Mater. Sci., Vol 15, Pt. 4, 1973, p 289-374 375. M.H. Kamdar, Liquid Metal Embrittle- merit, in Embrittlement of Engineering Alloys, Academic Press, 1983, p 361- 459 376. M.H. Kamdar, Ed., Embrittlement by Liquid and Solid Metals, American Insti- tute of Mining, Metallurgical, and Petro- leum Engineers, 1984 (a) (b) 377. M.B. Reynolds, Radiation Effects in Some Engineering Alloys, J. Mater., Vol Fig. 1 1 0 Macrograph (a) of defect in test bar that initiated fatigue failure. 1.5x. (b) SEM fractogroph 1, March 1966, p 127-152 showing that the defect was a shrinkage cavity. Note the dendritic appearance. 375 x 378. K. Ohmae and T.O. Ziebold, The Influ- ence of Impurity Content on the Radiation Stainless Steel, Acta Metall., Vol 34, ing on Sensitization in Surface Layers in Sensitivity of Pressure Vessel Steels: Use March 1986, p 557-570 Type 304 Stainless Steel," Paper 25, of Electron Microprobe for Irregular Sur- 367. M.A. Streicher, Screening Stainless presented at Corrosion/83, National As- faces, J. Nucl. Mater., Vol 43, 1972, p Steels from the 240-Hr Nitric Acid Test sociation of Corrosion Engineers, 1983 245-257 by Electrolytic Etching in Oxalic Acid, 370. J.M. Schluter and J.A. Chivinsky, Sur- 379. J.R. Hawthorne and L.E. Steele, Metal- ASTMBuII., No. 188, Feb 1953, p 35-38; face Preparation Requirements for ASTM lurgical Variables as Possible Factors July 1953, p 58-59 A262, in Laboratory Corrosion Tests and Controlling Irradiation Response of 368. M.A. Streicher, Theory and Application Standards, STP 866, American Society Structural Steels, in The Effects of Radi- of Evaluation Tests for Detecting Suscep- for Testing and Materials, 1985, p ation on Structural Metals, STP 426, tibility to Intergranular Attack in Stainless 455-464 American Society for Testing and Mate- Steels and Related Alloys--Problems and 371. W. Rostoker et al., Embrittlement by rials, 1967, p 534-572 Opportunities, in Intergranular Corro- Liquid Metals, Reinhold, 1960 380. Irradiation Effects on the Microstructure sion of Stainless Alloys, STP 656, Amer- 372. N.S. Stoloff, Liquid Metal Embrittle- and Properties of Metals, STP 611, ican Society for Testing and Materials, p ment, in Surfaces and Interfaces, Pt. II, American Society for Testing and Mate- 3-84 Syracuse University Press, 1968, p rials, 1976 369. A.P. Majidi and M.A. Streicher, "The 157-182 381. J.O. Stiegler and E.E. Bloom, The Ef- Effect of Methods of Cutting and Grind- 373. A.R.C. Westwood et al., Adsorption-- fects of Large Fast-Neutron Fluences on Visual Examination and Light Microscopy/161

Iron, J. Iron Steel Inst., Vol 172, Oct 1952, p 163-171 398. T.C. Lindley and C.E. Richards, The Effect of Quench-Aging on the Cleavage Fracture of a Low-Carbon Steel, Met. Sci. J., Vol 4, May 1970, p 81-84 399. A.S. Keh and W.C. Leslie, Recent Ob- servations on Quench-Aging and Strain Aging of Iron and Steel, in Materials Science Research, Vol 1, Plenum Press, 1963, p 208-250 400. E.R. Morgan and J.F. Enrietto, Aging in Steels, in AISI 1963 Regional Technical Meeting, American Iron and Steel Insti- tute, 1964, p 227-252 401. W.C. Leslie and A.S. Keh, Aging of Flat-Rolled Steel Products as Investigated by Electron Microscopy, in Mechanical Working of Steel, Pt. II, Gordon & Breach, 1965, p 337-377 402. E. Stolte and W. Heller, The State of Knowledge of the Aging of Steels, I. Fundamental Principles, Stahl Eisen, Vol 90 (No. 16), BISI 8920, 1970, p 86•-868 403. L.R. Shoenberger and E.J. Paliwoda, Accelerated Strain Aging of Commercial Fig. ! 1 2 Example of a fractured wedge test Sheet Steels, Trans. ASM, Vol 45, 1953, specimen used to assess the carbide p 344-361 stability in a cast iron melt. Note the change in fracture appearance in the tip (white fracture due to the presence 404. F. Garofalo and G.V. Smith, The Effect of carbide) compared to the base of the wedge (dark of Time and Temperature on Various due to the presence of graphite). Actual size Mechanical Properties During Strain Ag- ing of Normalized Low Carbon Steels, Trans. ASM, Vol 47, 1955, p 957-983 Inst., Vol 164, Jan 1960, p 46-50 405. J.D. Baird, Strain Aging of Steel--A Fig. 1 I 1 Example of inclusion stringers (white 389. B.J. Schulz and C.J. McMahon, Fracture Critical Review, Iron Steel, Vol 36, May streaks against the dark background) on a longitudinal test fracture. The specimen was blued of Alloy Steels by lntergranular Mi- 1963, p 186-192; June 1963, p 326-334; to reveal the stringers. crovoid Coalescence as Influenced by July 1963, p 368-374; Aug 1963, p Composition and Heat Treatment, Metall. 400-405; Sept 1963, p 450-457 Trans., Vol 4, Oct 1973, p 2485-2489 406. J.D. Baird, The Effects of Strain-Aging the Structure of Stainless Steel, J. Nucl. 390. J.A. Disario, "Overheating and Its Effect Due to Interstitial Solutes on the Mechan- Mater., Vol 33, 1969, p 173-185 on the Toughness of ASTM A508 Class II ical Properties of Metals, MetalL Rev., 382. S.D. Harkness and C.Y. Li, A Study of Forgings," MS thesis, Lehigh Univer- Vol 16, Feb 1971, p 1-18 Void Formation in Fast Neutron- sity, 1973 407. G.F. Vander Voort, Failures of Tools and Irradiated Metals, Metall. Trans., Vol 2, 391. T.J. Baker and R. Johnson, Overheating Dies, in Failure Analysis and Prevention, May 1971, p 1457-1470 and Fracture Toughness, J. Iron Steel Vol 11, 9th ed., Metals Handbook, 383. H.R. Brager et al., Irradiation-Produced Inst., Vol 211, Nov 1973, p 783-791 American Society for Metals, 1986, p Defects in Austenitic Stainless Steel, 392. T.J. Baker, Use of Scanning Electron 563-585 Metall. Trans., Vol 2, July 1971, p Microscopy in Studying Sulphide Mor- 408. L.D. Jaffee and J.R. Hollomon, Harden- 1893-1904 phology on Fracture Surfaces, in Sulfide ability and Quench Cracking, Trans. 384. H.R. Brager and J.L. Straalsund, Defect Inclusions in Steel, American Society for AIME, Vol 167, 1946, p 617-626 Development in Neutron Irradiated Type Metals, 1975, p 135-158 409. M.C. Udy and M.K. Barnett, A Labora- 316 Stainless Steel, J. Nucl. Mater., Vol 393. G.E. Hale and J. Nutting, Overheating of tory Study of Quench Cracking in Cast 46, 1973, p 134-158 Low-Alloy Steels, Int. Met. Rev., Vol 29 Alloy Steels, Trans. ASM, Vol 38, 1947, 385. K. Farrell and R.T. King, Radiation- (No. 4), 1984, p 273-298 p 471-487 Induced Strengthening and Embrittlement 394. R.C. Andrew et al., Overheating in Low- 410. J.W. Spretnak and C. Wells, An Engi- in Aluminum, Metall. Trans., Vol 4, Sulphur Steels, J. Austral. Inst. Met., neering Analysis of the Problem of May 1973, p 1223-1231 Vol 21, June-Sept 1976, p 126-131 Quench Cracking in Steel, Trans. ASM, 386. J.R. Strohm and W.E. Jominy, High 395. R.C. Andrew and G.M. Weston, the Vol 42, 1950, p 233-269 Forging Temperatures Revealed by Fac- Effect of Overheating on the Toughness 411. C. Wells, Quench Cracks in Wrought ets in Fracture Tests, Trans. ASM, Vol of Low Sulphur ESR Steels, J. Austral. Steel Tubes, Met. Prog., Vol 65, May 30, 1946, p 543-571 Inst. Met., Vol 22, Sept-Dec 1977, p 1954, p 113-121 387. A. Preece et al., The Overheating and 171-176 412. T. Kunitake and S. Sugisawa, The Quench- Burning of Steel, J. Iron Steel Inst., Vol 396. S. Preston et al., Overheating Behavior Cracking Susceptibility of Steel, Sumito- 153, Pt. 1, 1946, p 237p-254p of a Grain-Refined Low-Sulphur Steel, mo Search, No. 5, May 1971, p 16-25 388. A. Preece and J. Nutting, The Detection Mater. Sci. Technol., Vol 1, March 413. K.E. Thelning, Why Does Steel Crack on of Overheating and Burning in Steel by 1985, p 192-197 Hardening, Hgirt.-Tech.-Mitt., Vol 4, Microscopical Methods, J. Iron Steel 397. A.L. Tsou et al., The Quench-Aging of BISI 12602, 1970, p 271-281 162 / Visual Examination and Light Microscopy

Fig. 1 1 4 Example of specimens used in the Shep- herd P-F test. The 19-mm (3/4-in.) diam, 100-mm (4-in.) long specimen is shown at the bottom. The test consists of fracturing an austenitized and brine-quenched specimen. Case depth can be observed on the fracture. After fracture grain size is rated, the fracture surface is carefully ground off. Macroetching then reveals the depth of hardening. Superficial hardness tests are used to determine the case depth to a particular hardness.

Fig. 1 1 3 Four fractured, hardened etch disks of a high-carbon (1.3% C) tool steel that contained excessive graphite (dark regions) due to an accidental aluminum addition. About 0.5 x American Society for Testing and Mate- rials, 1978, p 164-175 435. B.C. Woodfine, Temper-Brittleness: A 414. P. Gordon, The Temper Colors on Steel, tific Translations and the National Sci- Critical Review of the Literature, J. Iron J. Heat Treat., Vol 1, June 1979, p 93 ence Foundation, 1961 Steel Inst., Vol 173, March 1953, p 415. Modern Steels, 6th ed., Bethlehem Steel 426. H.L. Craig, Jr., Ed., Stress Corrosion-- 229-240 Corporation, 1967, p 149 New Approaches, STP 610, American 436. J.R. Low, Jr., Temper Brittleness--A 416. D.J. McAdam and G.W. Geil, Rate of Society for Testing and Materials, 1976 Review of Recent Work, in Fracture of Oxidation of Steels as Determined From 427. R.W. Staehle et al., Ed., Fundamental Engineering Materials, American Society Interference Colors of Oxide Films, J. Aspects of Stress Corrosion Cracking, for Metals, 1964, p 127-t42 Res. NBS, Vol 23, July 1939, p 63-124 National Association of Corrosion Engi- 437. C.J. McMahon, Jr., Temper Brittle- 417. A.J. Lena, Effect of Sigma Phase on neers, 1969 ness--An Interpretive Review, in Temper Properties of Alloys, Met. Prog., Vol 66, 428. C. Edeleanu, Transgranular Stress Corro- Embrittlement in Steel, STP 407, Ameri- Aug 1954, p 94-99 sion in Chromium-Nickel Stainless can Society for Testing and Materials, 418. J.I. Morley and H.W. Krikby, Sigma- Steels, J. Iron Steel Inst., Vol 173, Feb 1968, p 127-167 Phase Embrittlement in 25Cr-20Ni Heat- 1953, p 140-146 438. I. Olel]ord, Temper Embrittlement, Int. Resisting Steels, J. Iron Steel Inst., Vol 429. N.A. Neilsen, Environmental Effects on Met. Rev., Vol 23 (No. 4), 1978, p 172, Oct 1952, p 129-142 Fracture Morphology, in Electron Frac- 149-163 419. A.M. Talbot and D.E. Furman, Sigma tography, STP 436, American Society for 439. R.G.C. Hill and J.W. Martin, A Fracto- Formation and Its Effect on the Impact Testing and Materials, 1968, p 124-150 graphic Study of Some Temper Brittle Properties of Iron-Nickel-Chromium Al- 430. J.C. Scully, Scanning Electron Micro- Steels, Met. Treat. Drop Forg., Vol 29, loys, Trans. ASM, Vol 45, 1953, p scope Studies of Stress-Corrosion Crack- Aug 1962, p 301-310 429-442 ing, in Scanning Electron Microscopy, 440. H. Ohtani and C.J. McMahon, Jr., 420. E.O. Hall and S.H. Algie, The Sigma liT Research Institute, 1970, p 313-320 Modes of Fracture in Temper Embrittled Phase, Met. Rev., Vol 11, 1966, p 61-88 431. J.C. Scully, Failure Analysis of Stress Steels, Acta Metall., Vol 23, March 421. G. Matern et al., The Formation of Sigma Corrosion Cracking with the Scanning 1975, p 377-386 Phase in Austenitic Ferrite Stainless Electron Microscope, in Scanning Elec- 441. J. Yu and C.J. McMahon, Jr., Variation Steels and Its Influence on Mechanical tron Microscopy, Pt. IV, liT Research of the Fracture Mode in Temper Em- Properties, Mem. Sci. Rev. Met., BISI Institute, 1974, p 867-874 brittled 2.25Cr-lMo Steel, Metall. 13972, 1974, p 841-851 432. B. Poulson, The Fractography of Stress Trans., Vol 16A, July 1985, p 1325-1331 422. W.J. Boesch and J.S. Slaney, Preventing Corrosion Cracking in Carbon Steels, 442. J.B. Cohen et al., A Metallographic Sigma Phase Embrittlement in Nickel Corros. Sci., Vo115, Sept 1975, p 469-477 Etchant to Reveal Temper Brittleness in Base Superalloys, Met. Prog., Vol 86, 433. H. Okada et al., Scanning Electron Mi- Steel, Trans. ASM, Vol 39, 1947, p July 1964, p 109-111 croscope Observation of Fracture Faces 109-138 423. J.R. Mihalisin et al., Sigma--Its Occur- of Austenitic Stainless Steels by Stress 443. D. McLean and L. Northcott, Micro- rence, Effect, and Control in Nickel-Base Corrosion Cracking, Nippon Kinzoku Examination and Electrode-Potential Superalloys, Trans. A1ME, Vol 242, Dec Gakkai-shi, Vol 37, 1973, p 197-203 Measurements of Temper-Brittle Steels, 1968, p 2399-2414 434. Y. Mukai et al., Fractographic Observa- J. Iron Steel Inst., Vol 158, 1948, p 424. H.L. Logan, The Stress Corrosion of tion of Stress-Corrosion Cracking of AISI 169-177 Metals, John Wiley & Sons, 1966 304 Stainless Steel in Boiling 42 Percent 444. J.P. Rucker, "Improved Metallographic 425. V.V. Romanov, Stress-Corrosion Crack- Magnesium-Chloride Solution, in Frac- Technique for Revealing Temper Brittle- ing of Metals, Israel Program for Scien- tography in Failure Analysis, STP 645, ness Network in Ordnance Steels," NPG Visual Examination and Light Microscopy/ 163

Trans., Vol 9A, March 1978, p 439-450 450. R.M. Horn and R.O. Ritchie, Mecha- nisms of Tempered Martensite Embrittle- ment in Low Alloy Steels, Metall. Trans., Vol 9A, Aug 1978, p 1039-1053 451. C.L. Briant and S.K. Banerji, Tempered Martensite Embrittlement in a High Pu- rity Steel, Metall. Trans., Vol 10A, Aug 1979, p 1151-1155 452. M. Sarikaya et al., Retained Austenite and Tempered Martensite Embrittlement in Medium Carbon Steels, Metall. Trans.,Vol 14A, June 1983, p 1121-1133 453. N. Bandyopadhyay and C.J. McMahon, Jr., The Micro-Mechanisms of Tempered Martensite Embrittlement in 4340-Type Steels, Metall Trans., Vol 14A, July 1983, p 1313-1325 454. G.J. Spaeder, Impact Transition Behavior of High-Purity 18Ni Maraging Steel, Metall. Trans., Vol 1, July 1970, p 2011-2014 455. D. Kalish and H.J. Rack, Thermal Em- brittlement of 18Ni (350) Maraging Steel, Metall. Trans., Vol 2, Sept 1971, p 2665-2672 456. W.C. Johnson and D.F. Stein, A Study of Grain Boundary Segregants in Thermally Embrittled Maraging Steel, Metall. Trans., Vol 5, March 1974, p 549-554 457. E. Nes and G. Thomas, Precipitation of TiC in Thermally Embrittled Maraging (a) Steels, Metall. Trans., Vol 7A, July 1976, p 967-975 458. H.J. Rack and P.H. Holloway, Grain Boundary Precipitation in 18Ni-Maraging Steels, Metall. Trans., Vol 8A, Aug 1977, p 1313-1315 459. A.J. Birkle et al., A Metallographic In- vestigation of the Factors Affecting the Notch Toughness of Maraging Steels, Trans. ASM, Vol 58, 1965, p 285-301 460. R.M. Fisher et al., Identification of the Precipitate Accompanying 885 °F Embrit- tlement in Chromium Steels, Trans. AIME, Vol 197, May 1953, p 690-695 461. B. Cina and J.D. Lavender, The 475 °C Hardening Characteristics of Some High- Alloy Steels and Chromium Iron, J. Iron Steel Inst., Vol 174, June 1953, p 97-107 (b) (cl 462. A.J. Lena and M.F. Hawkes, 475 °C (885 Fig. ! ! 5 a~s~ ]070 steel bar that broke during heat treatment due to gross chemical segregation. (a) °F) Embrittlement in Stainless Steels, Macrograph at actual size. (b) Fracture near the outside surface of the bar. 8 x. (c) Fracture near Trans. A1ME, Vol 200, May 1954, the center of the bar. 8 x. Note the coarse grain structure of the center region. p 607-615 463. R.O. Williams and H.W. Paxton, The Nature of Aging of Binary Iron-Chromi- 1555, United States Naval Proving Martensite Brittleness in Extra-Low- um Alloys Around 500 °C, J. Iron Steel Ground, 28 Aug 1957 Carbon Steels, J. Iron Steel Inst., Vol Inst., Vol 185, March 1957, p 358-374 445. J.M. Capus and G. Mayer, The Influence 207, Dec 1969, p 1628-1634 464. R. Lagneborg, Metallography of the of Trace Elements on Embrittlement Phe- 448. G. Delisle and A. Galibois, Microstruc- 475C Embrittlement in an Iron-30% nomena in Low-Alloy Steels, Metal- tural Studies of Tempered Extra-Low Chromium Alloy, Trans. ASM, Vol 60, lurgia, Vol 62, 1960, p 133-138 Carbon Steels and Their Effectiveness in 1967, p 67-78 446. E.B. Kula and A.A. Anctil, Tempered Interpreting Tempered Martensite Brittle- 465. P.J. Grobner, The 885 °F (475 °C) Em- Martensite Embrittlement and Fracture ness, in Microstuctural Science, Vol 1, brittlement of Ferritic Stainless Steels, Toughness in SAE 4340 Steel, J. Mater., Elsevier, 1974, p 91-112 Metall. Trans., Vol 4, Jan 1973, p Vol 4, Dec 1969, p 817-841 449. G. Thomas, Retained Austenite and Tem- 251-260 447. G. Delisle and A. Galibois, Tempered pered Martensite Embrittlement, Metall. 466. H.D. Solomon and E.F. Koch, High 164 / Visual Examination and Light Microscopy

Temperature Precipitation of od in a Mul- 1960, p 508-512 April 1983, p 89s-96s ticomponent Duplex Stainless Steel, Scr. 484. J.C. Borland, Hot Cracking in Welds, Br. 502. J. Heuschkel, Anisotropy and Weldabil- Metall., Vol 13, 1979, p 971-974 Weld. J., Vol 7, Sept 1960, p 558-559 ity, Weld. J., Vol 50, March 1971, p 467. H.D. Solomon and L.M. Levinson, Mrss- 485. J.H. Rogerson and J.C. Borland, Effect l10s-126s bauer Effect Study of "475 °C Embrittle- of the Shapes of Intergranular Liquid on 503. T. Boniszewski and N.F. Eaton, Electron ment" of Duplex and Ferritic Stainless the Hot Cracking of Welds and Castings, Fractography of Weld-Reheat Cracking in Steels, Acta Metall., Vol 26, 1978, p Trans. AIME, Vol 227, Feb 1963, p 2-7 CrMoV Steel, Met. Sci. J., Vol 3, 1969, 429-442 486. W.F. Savage et al., Copper-Contam- p 103-110 468. J. Chance et al., Structure-Property Re- ination Cracking in the Weld Heat- 504. R.A. Swift, The Mechanism of Stress lationships in a 25Cr-7Ni-2Mo Duplex Affected Zone, Weld. J., Vol 57, May Relief Cracking in 2-1/4Cr-lMo Steel, Stainless Steel Casting Alloy, in Duplex 1978, p 145s-152s Weld. J., Vol 50, May 1971, p 195s-200s Stainless Steels, American Society for 487. H. Homma et al., A Mechanism of High 505. D. McKeown, Re-Heat Cracking in High Metals, 1983, p 371-398 Temperature Cracking in Steel Weld Nickel Alloy Heat-Affected Zones, Weld. 469. P. Jacobsson et al., Kinetics and Harden- Metals, Weld. J., Vol 58, Sept 1979, p J., Vol 50, May 1971, p 201s-206s ing Mechanism of the 475 °C Embrittle- 277s-282s 506. A.W. Pense et al., Stress Relief Cracking ment in 18Cr-2Mo Ferritic Steels, Metall. 488. M.J. Cieslak et al., Solidification Crack- in Pressure Vessel Steels, Weld. J., Vol Trans., Vol 6A, Aug 1975, p 1577-1580 ing and Analytical Electron Microscopy 50, Aug 1971, p 374s-378s 470. A.L. Schaeffier et al., Hydrogen in Mild- of Austenitic Stainless Steel Weld Metals, 507. R.A. Swift and H.C. Rogers, Embrittle- Steel Weld Metal, Weld. J., Vol 31, June Weld. J., Vol 61, Jan 1982, p ls-8s ment of 2-1/4Cr-1Mo Steel Weld Metal by 1952, p 283s-309s 489. T. Ogawa and E. Tsunetomi, Hot Crack- Postweld Heat Treatment, Weld. J., Vol 471. H.G. Vaughan and M.E. deMorton, Hy- ing Susceptibility of Austenitic Stainless 52, April 1973, p 145s-153s, 172s drogen Embrittlement of Steel and Its Steels, Weld. J., Vol 61, March 1982, p 508. C.F. Meitzner, Stress-Relief Cracking in Relation to Weld Metal Cracking, Br. 82s-93s Steel Weldments, Weld. Res. Counc. Weld. J., Vol 4, Jan 1957, p 40-61 490. G. Rabensteiner et al., Hot Cracking Bull., No. 211, Nov 1975 472. T.E.M. Jones, Cracking of Low Alloy Problems in Different Fully Austenitic 509. J. Myers, Influence of Alloy and Impurity Steel Weld Metal, Br. Weld. J., Vol 6, Weld Metals, Weld. J., Vol 62, Jan 1983, Content on Stress-Relief Cracking in Cr- July 1959, p 315-323 p 21s-27s Mo-V Steels, Met. Technol., Vol 5, Nov 473. N. Christensen, The Role of Hydrogen in 491. J.A. Brooks et al., A Fundamental Study 1978, p 391-396 Arc Welding With Coated Electrodes, of the Beneficial Effects of Delta Ferrite 510. A. Vinckier and A. Dhooge, Reheat Weld. J., Vol 40, April 1961, p 145s- in Reducing Weld Cracking, Weld. J., Cracking in Welded Structures During 154s Vol 63, March 1984, p 71s-83s Stress Relief Heat Treatments, J. Heat 474. E.P. Beachum et al., Hydrogen and De- 492. J.C.M. Farrar and R.E. Dolby, An In- Treat., Vol 1, June 1979, p 72-80 layed Cracking in Steel Weldments, vestigation Into Lamellar Tearing, Met. 511. C.P. You et al., Stress Relief Cracking Weld. J., Vol 40, April 1961, p 155s- Constr. Br. Weld. J., Vol 1, Feb 1969, p Phenomena in High Strength Structural 159s 32-39 Steel, Met. Sci., Vol 18, Aug 1984, p 475. F. Watkinson et al., Hydrogen Embrittle- 493. D.N. Elliott, Fractographic Examination 387-394 ment in Relation to the Heat-Affected of Lamellar Tearing in Multi-Run Fillet 512. J. Shin and C.J. McMahon, Jr., Compar- Zone Microstructure of Steels, Br. Weld. Welds, Met. Constr. Br. Weld. J., Vol 1, ison of Stress Relief Cracking in A508 2 J., Vol 10, Feb 1963, p 54-62 Feb 1969, p 50-57 and A533B Pressure Vessel Steels, Met. 476. F. Watkinson, Hydrogen Cracking in 494. J.C.M. Farrar et al., Lamellar Tearing in Sci., Vol 18, Aug 1984, p 403-410 High Strength Weld Metals, Weld. J., Welded Structural Steels, Weld. J., Vol 513. A.G. Fuller, Apparatus for Breaking Test Vol 48, Sept 1969, p 417s-424s 48, July 1969, p 274s-282s Castings, BCIRA J., Vol 8, July 1960, p 477. T. Boniszewski and F. Watkinson, Effect 495. D.N. Elliott, Lamellar Tearing in Multi- 586-587 of Weld Microstructures on Hydrogen- Pass Fillet Joints, Weld. J., Vol 48, Sept 514. G.M. Enos, Fractures, in Visual Exami- Induced Cracking in Transformable 1969, p 409s-416s nation of Steel, American Society for Steels, Met. Mater., Vol 7, Feb 1973, p 496. S. Hasebe et al., Factors for Lamellar Metals, 1940 p 37-54 90-96; March 1973, p 145-151 Tearing of Steel Plate, Sumitomo Search, 515. B. Ostrofsky, Materials Identification in 478. J.M.F. Mota and R.L. Apps, "Chevron No. 13, May 1975, p 19-27 the Field, Mater. Eval., Vol 36, Aug Cracking"--A New Form of Hydrogen 497. S. Ganesh and R.D. Stout, Material Vari- 1978, p 33-39, 45 Cracking in Steel Weld Metals, Weld. J., ables Affecting Lamellar Tearing Suscep- 516. Blue-Brittleness Test for Assessing Mac- Vol 61, July 1982, p 222s-228s tibility in Steels, Weld. J., Vol 55, Nov roscopic Inclusion Contents in Steels, 479. P.H.M. Hart, Resistance to Hydrogen 1976, p 341s-355s StahI-Eisen-Prufblatt 1584, BISI 12397, Cracking in Steel Weld Metals, Weld. J., 498. J.C.M. Farrar, The Effect of Hydrogen Dec 1970 Vol 65, Jan 1986, p 14s-22s on the Formation of Lamellar Tearing, 517. "Wrought Steels--Macroscopic Methods 480. P.A. Kammer et al., "Cracking in High- Weld. Res. Int., Vol 7 (No. 2), 1977, p for Assessing the Content of Non- Strength Steel Weldments--A Critical 120-142 Metallic Inclusions," ISO 3763, Interna- Review," DMIC 197, Battelle Memorial 499. J. Sommella, "Significance and Control tional Standards Organization Institute, 7 Feb 1964 of Lamellar Tearing of Steel Plate in the 518. "Determination of Slag Inclusion Con- 481. P.W. Jones, Hot Cracking of Mild Steel Shipbuilding Industry," SSC-290, tent of Steel. Macroscopic Methods," Welds, Br. Weld. J., Vol 6, June 1959, p United States Coast Guard, May 1979 SIS 11111105, Swedish Institute of Stan- 269-281 500. E. Holby and J.F. Smith, Lamellar Tear- dards 482. J.C. Borland, Some Aspects of Cracking ing. The Problem Nobody Seems to Want 519. D.E. Krause, Chill Test and the Metal- in Welded Cr-Ni Austenitic Steels, Br. to Talk About, Weld. J., Vol 59, Feb lurgy of Gray Iron, Trans. AFS, Vol 59, Weld. J., Vol 7, Jan 1960, p 22-59 1980, p 37-44 1951, p 79-91 483. J.C. Borland, Generalized Theory of 501. A.D. Hattangadi and B.B. Seth, Lamellar 520. A.T. Batty, The Wedge Test and Its Use Super-Solidus Cracking in Welds (and Tearing in Fillet Weldments of Pressure in the Ironfoundry, SEAIS1 Q., Vol 4, Castings), Br. Weld. J., Vol 7, Aug Vessel Fabrications, Weld. J., Vol 62, April 1975, p 55-59 Visual Examination and Light Microscopy/165

521. W.H. Moore, Melting Engineering Cast 2, April 1922, p 620-622 Brass, Trans. AFS, Vol 58, 1950, p Iron in the Electric Furnace, Cast. Eng./ 525. J.H. Hruska, Fracture Tests and Ingot 122-132 Foundry World, Vol 16, Summer 1984, p Defects, Blast Furn. Steel Plant, Vol 22, 529. Fracture Test for Determining Melt 40, 43-46 Dec 1934, p 705,707 Quality of 85-5-5-5 Red Brass, Am. 522. B.F. Shepherd, The P-F Characteristic of 526. A.L. Norbury, The Effect of Casting Foundryman, Vol 24, Oct 1953, p 68-69, Steel, Trans. ASM, Vol 22, Dec 1934, p Temperature on the Primary Microstruc- 71 979-1016 ture of Cast Iron, J. Iron Steel Inst., Vol 530. A.R. French, Melt-Quality Tests for 523. G.F. Vander Voort, Grain Size Measure- 140 (No. II), 1939, p 161P-180P Copper-Base Alloys, Foundry Trade J., ment, in Practical Applications of Quan- 527. W.J. Williams, A Relationship of Micro- Vol 98, 10 March 1955, p 253-257; 17 titative Metallography, STP 839, Ameri- structure to the Mechanical Properties of March 1955, p 281-293 can Society for Testing and Materials, White Iron, BCIRA J., Vol 5, Dec 1953, 531. D. Hanson and W.T. Pell-Walpole, 1984, p 85-131 p 132-134 Methods of Assessing the Quality of Cast 524. W.J. Priestly, Fracture Test on Steel To 528. F.M. Baker et al., Melt Quality and Frac- Bronzes, in Chili-Cast Tin Bronzes, Determine its Quality, Trans. ASST, Vol ture Characteristics of 85-5-5-5 Red Edward Arnold, 1957, p 22-55