Scribd Logo
Upload

Welcome to Scribd!

  • Failure Analysis of Ceramics

    Uploaded by

    rajeshyalmanchili2
    14 views14 pages
    failure analysis of ceramics

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    Report this Document
    failure analysis of ceramics

    Copyright:

    © All Rights Reserved
    Download as PDF or read online from Scribd
    Flag for inappropriate content
    14 views14 pages

    Failure Analysis of Ceramics

    Uploaded by

    rajeshyalmanchili2
    failure analysis of ceramics

    Copyright:

    © All Rights Reserved
    Download as PDF or read online from Scribd
    Flag for inappropriate content
    of 14
    Failure Analysis of Cereaamics* FAILURE ANALYSIS is extremely impor- fant in engineering, especially with ceramics, because it is the only means of isolating the failure-causing problem. In particular, failure analysis helps determine whether failure oF ‘damage occurred due to a design deficiency of ‘mate deticieney. Until ths has been deter mined, efforts cannot be efficiently. directed ‘owand finding a solution, The result is usually 4 shotgun approach that includes alle design analysis, a litle empirical testing, and a litle ‘material evaluation and often ends up nly in & repeat of the test or operating conditions that initially caused failure Much of the shotgun approach can often be avoided hy Factire analyte ractine alysis for fractography is the examination of the frnc- {ured or damaged hardware in an effort to reconstruct the Sequence and cause of fracture ‘The path a crack follows as it propagates through « component provides substantial infor- mation about the stress distribution atthe time ‘of failure, Features on the fracture surfaces provide further information, especially the po- sition at whieh the fracture initiated (fracture ‘origin, the cause of fracture mutation (umpact, tensile overload, thermal shock, material aw tc.) and even the approximate local stess that ‘caused fracture. The primary objective of this fanicle is to acquaint the feader with these fracture surface features and the techniques used to interpret the cause of fracture in ceramic components Location of the Fracture Origin The fist step is to determine where the fracture initiated. Often, simply reconstructing the pieces will pinpoint the fracture origin and may even give useful information about the cause of fracture, Afler assembling the pieces, look for places where a group of cracks come together or where a single crack branches, Presi (Ref 1) hus shown that the angle of forking is an indicator ofthe stress distribution causing fracture. Examples are shown schemat- really im Fig. Ape rr Char 12 Maden Cermic Eginering Dovid W. Richerson, Ceramatec, Ine. Fig. 1 Information available by examining crack direction and crack branching 2)... % oF PY ‘The frequency of erick branching provides qualitative information about the amount of energy available during fracture. To branch, & ‘rack must reach acrtieal speed. For glas, the ctitcal speed is typically a litle greater than half the speed of sound inthe specific glass. At this instant of crack initiation, the erack veloc- ity is zero, but quickly accelerates, The rate of acceleration isa function of the energy’ avail- able ether due tothe stress applied orto energy stored in the part (Such as residual stresses or Prestresses, ain tempered glass). The more fenergy, the more rapidly the crack will reach its titcal branching velocity and the more branch- ing that will occur. A baseball stiking a win dow will cause much more branching than BB, duc tothe larger applied engszy. Tempered lass will brcak into many fragments due 10 release of the high stored energy. On the other hand, a thetmal shock fracture may not branch at al, especially if it initiates from a localized hheat Soures and propagates into a relatively lunstressed or compressvely stressed region of the component. In this ease, the fracture wil tend t0 follow a temperature or stress contour Fig. 2 Thermal shock fracture showing lack of branching yy. typical fracture features that surround the fracture origin (6) eral inition. (b) Srfze inion ‘main ® sutce » and will have a chaauicrisie wavy oF curved appearance, as shown in Fig. 2 fora thermally fractured eeramic setter plate fora furnace. ‘The pattem of branching will often lead the engineer to the vicinity of the facture origin, The engineer will then have « examine the fracture surfaces in this region, often under a low-power optical binocular microscope, 0 Fig. 4 Example of ty1 polycrystalline ceramics rors for high-strength (2) hittin ot surface flowin hopes scan mide, (b) tition ot an ternal low in reton sintered Steen ide o Fig. 5 Examples of fracture surfaces with indistinct fracture features (c) Sneed con carbide. (b) Skconcrbide-carbon-sicon compote, (€ Prous Khim alin sic, (3 binedal qr dibvion reochon-tered son carbide Tocate the precise point at which fracture inti- ated, This point of origin can be a flaw (such as 4 pore or inclusion in the material), @ cone shaped Heriian surface crack resulting from impact, a crack ina surface glaze, an oxidation Pit, imtergranular corrosion, a position of local ized high stress, oF a combination. Location and examination ofthe fracture origin will help determine which of these factors is dominant and will provide specific guidance in solving the fracture problem ‘Asa facture accelerates, it interacts with the ‘microstructure, the stress field, and even aeous tic vibrations and leaves distinct features on the facture surface that can be used for locating the facture origin (Ref 3). The most important features include hackle, the fracture mirror, and Wallner lines. ‘The Fracture Mirror and Hackle. When crack initiates at an internal flaw, the erack font travels radially in a single plane as it aceclerates, The surface formed is flat and Failure Analysis of Ceramics / 745 smooth and is called the fracture mirror. When the erack reaches a critical spoed, imersect an inclusion, or encounters a shit inthe direction ‘of principal tensile sess, it begins to deviate slightly from the original plane, forming small radial ridges on the fracture surface. The frst of these ane very fal ud ane referred 49a ts, Mist is usually visible onthe fracture surfaces of glass, but may not be on erystalline ceram ies. The mist (ansitions. into larger ridges called hackle, Hackle i also refered o a river pattems because the appearance i similar to the branching of a river into tributaries and the formation of deltas, The hackle region trans: tions into macroscopic erack branching seh ‘thatthe remaining portion of the fracture sur face is often on a plane that is pervoptibly differen from the mirtor and hackle. Some times, this gives the appearance thatthe frac ture origin is either on a step ora pedestal Figure 3a) shows schematically the fracture mirror, mist, and hackle for a fracture that initiated in the interior of a part The minor is roughly circular, and the fracture origin isa its center. Note that lines drawn parallel t0 the hackle will intersect at or very near the fracture ‘origin. Similarly, Fis. 2(b) shows the fracture Features for aerack that started atthe suave of par. The mirtr is roughly semicircular for surface-initiated fractures, usually with slight elongation toward the interior of the material Te elongation results from difference in siress intensity factor between the surface and the interior. The stress intensity facto is higher atthe surfuce, eausing the crack branching and the resulting appearance of the hackle to acer more quickly along the surface than inward The hackle lines surrounding the mirror re sult from velocity and stress intensity effects and are sometimes called velocity hackle. Av ‘ther form of hackle, called twist hackle, ust ally forms away from the mirror and results from an abrupe change inthe tensile stress fel such a8 going from tension to compression Twist hackle points in the new diction of -k movement, appears more as paallel cracks than ridges, and docs not necessaily point to the facture origin, Twist hackle is an Important feature for deducing the stress distri bution in the ceramic atthe time of fracture "The size of the fracture mirror is dependent fn the material characteristics and the localized stress. at the fracture origin at the time of fracture, Studies by Terao (Ref 4), Levengood {Ref 5), and Shand (Ref 6) on glass suggest that the fracture stress, times the square root of the rmiror radius ,, equals a constant a for a given materia: ore e) Kirchener and Gruver (Ref 7) determined that this relationship aso provides a good approxi ‘mation for polyerystallin ceramies, as Fong 3s the mimor is clearly visible and can he mea sured accurately. They obtained values of & 746 | Engineered and Electronic Materials Fig. 6 Relationship of Wallner lines ona fracture surface to the stress distribution at the time of fracture (0) Us tein. () Nomar teson ©) ncoconpresion Source Ref 2 nging from 2.3 fora elass to9. 1 fora sintered ALO, to 14.3 for a hot-pressed Si,Ny TL is therefore posible to estimate the stress causing failure Of a ceramic component by ‘comparing the mirror size with « graph of ‘etsus ay for the material. The graph can be ‘compiled from bend stength or tensile strength data using scanning electron microscopy (SEM) fof the Fracture surfaces of the test bars to ‘determine the miror radius Gaitfith (Ref 8) proposed an equation for relating the fracture stress to material properties fad flaw size: where 0 isthe facture stress, Fis the elastic ‘modulus. > isthe Fracture energy, s the Maw Size, and A is a constant that depends on the specimen and flaw geometries, ‘Evans and Tappin (Ref 9) have presented a eer nie” where ¥ is a dimensionless term that depends fon the Maw depth and the test geometry, Z is smother dimensionless term that depends on the flaw configuration, and c is the depth of surface law (or hal the flaw size for an internal flaw). and € and ae defined as above. For an interal flaw that fs less than Yio ofthe size of the cross section under tensile loading, Y= 1.77. For a surface flaw that is much less than ‘io of the thickness of @ cross section under bend loading, Y-appproaches 2.0; Z varies aecording to" the Maw shape, ‘but is Usually between 1.0 and 2.0 ea lationship: 3) the fracture contour through the specimen ‘thickness for bend loading versus pure tensile loading (6) ar (ani long (+) conducting fracture analysis St0p 7 = isl ten soa co Tr ia) “ “ CC ~ o= a eto ‘Stop 3 SEM Lae siete . c= co Feacwre | Fcrosuctue Fig. 8 Examples showing cantilever curl in four-point bend spocimons Specinen were 0.32 cm (0.125 in.) tik Because inclusions and pores and other Naws are not symmetrical and their boundaries are ‘often not well defined, stress estimates based fn flaw size are only approximate. If know!- edge ofthe local stress atthe fracture origi is LI — ‘Chemistry of needed, perhaps it should be calculated bth from the mirror size and the flaw size and then the most appropriate value selected by good engineering judgment Figure 4 shows examples of fracture mirrors and flaws on the fracture surfaces of strength-test specimens. The photomicrographs were taken using a scanning electron micro: scope. Sinee the ceramic specimens were not electrical conductors, a thin layer of gold was applied to the surface by sputtering to avoid charge buildup, which would result in poor resolution ‘The mirror size cannot always be measured ‘he fracture-causing stress is low andthe spec men size is small, the mirror may cover the whole fracture surface. Ifthe material has ver) ‘coarse grain structure or a bimodal gain struc ture, the mitror and other fracture features may ‘not be visible or distinct enough for measure ment, Figure 5 shows examples of fracture su faces with indistinct fracture Features, Wallner Lines. Sonic waves are produced in a material during fracture. As each succeed ing wave front overtakes the primary fracture Fig. 10 Failure analysis interaction to determine the problem is design or materials oriented and to define a plan of action to solve the problem Fractooraphy Ere Frscture origin ‘Materia ew «Suess diatrbution estimated) Review of a see Tolrance change 1 Hoa tsaton of crack, the principal stress_is_ momentarily disturbed. ‘This results ma series of Taint are-shaped surface lines that are termed Wallner lines. The curvature of each line shows the approximate shape of the crack front atthe time it was intersected by the sonic wave ‘nd provides information about the direction of crack propagation and the stress distribution “The direction i from the concave to the convex side of the Wallner lines. The stress distribu: tion is inferred from the dist portion of a ingle Tine fn dhe Stress distribution were of uniform tension, tach. portion of @ line would be about equidistant from the origin. Ifa stress gradient wwere present, the distance of various portions ‘of the Wallner line from the origin would vary, being farthest where the tensile stress. was highest. These effects are shown schematically in Fig. 6 Wallner lines are not always present. For high energy fractures, whore the Fracture velocity is high and the surface is. rough, Wallner lines often cannot be distinguished. In very slow erack velocities, such as occur in subcritical crack growth, Wallner lines are not present, because the sonic waves are damped land gone before the crack has propagated ap- preciably. Other Features. (ther fracture features besides the mirror, hackle, and Wallner lines Process mosteston 1 imorove orooF toe ‘ace useful in interpretation of a fracture. These Include arrest lines, gull wings, and cantilever cur ‘An arrest line occurs when the erack front temporarily stops, The reason for crack arest is usually a momentary decrease in stress or a change in stress distribution. When the crack starts moving again, its direction invariably has changed slightly, leaving a discontinuity. ‘This line of discontinuity looks lite like a Wallner line, but is usually more out of plane and more lst Tip also cal a ibe. Arrest lines ‘or rib marks provide essentially the same infor- ‘mation as Wallner ines, thai, the direction of crack movement and the stress. distribution “Twist hackle frequently is present after an atest line. The gull wing isa feature that occurs due to the erick intersecting. @ pore or inclusion * As the crack travels around the inclusion, two crack fronts result. These do not always meet fom the same plano on the opposito side of the inclusion or pore, resulting in a ridge where the ‘0 Tink up and again become a single crack Front. In some cases, the ridge is immediately in the wake ofthe inclusion or pore and looks Tike a tadpole. In other eases, two ridges resembling 2 gull wing form in the wake. ‘A camilever eur oF compression lip, occurs ‘when the material is loaded in bending. The Fracture initiates on the tensile side perpendic s/ 747 Failure Analysis of Cerat tla to the surface and exits on the compression side, no longer perpendicular to the surface This is ilustrated in Fig. 7. 1 the part were fractured under pure tension, the erack would be straight through the thickness and would thus exit at 90°. This information ean be valu able in diggnosiny the cause of facture. For instance, thermal fractures of plate-shaped parts typically approach a stress state of pure tension near the point of origin and will nt result in 3 ‘compression lip. These same parts fractured ‘mechanically will nonmally have some bend Joading and will thus have a compression ‘Another example i a part containing prestress- ing or residual intemal stesses. The erack wil ‘not pass straight through the thickness. but instead will follow contours consistent with the sees fields it encounters. A thied example is sleength testing. A problem with esting a ceramic in uniaxial tension is avoiding parasitic bend stresses. Examination ofthe fracture sur face for signs of cantilever curt after tensile testing will help determine if pure tension was, achieved or not Figure & shows the cross sections of typical specimens tested in four-point bending. Note the variations in dhe shape ofthe compression ip. Techniques of Fractography ‘The techniques offractopraphy ate relatively simple and the amount of sophisticated equip. ‘ment minimal. Often, the information required tw explain the eause of fracture of a component ‘can be obtained with only a mieroscope and a Tight source. Tn fact, an experienced individual ‘can sometimes explain the fracture just by ‘examining the fracture surfaces visually. At ‘other times, a variety of techniques, including ‘sophisticated approaches such as SEM, electron microprobe, and Auger analysis, are required (Ref 10-12). When extensive fractography is necessary, the steps and procedures are as shown schematically in Fig. 9. Step 1 involves visual examination of the fractured pieces and review of data regarding the test or service conditions under which the hardware failed. These data usually. provide ‘some hypotheses to guide the evaluation. Re- Constructing the broken pieces and sketching probable fracture origins and paths is also helpful ‘A primary objective of visual examination i, to locate the point of fracture origin. This can ‘be done using Wallner lines, hackle. and the fracture mirror, as described previously. ‘Visual examination determines the extent of ‘additional evaluation that will likely be re- Guired. It also determines if cleaning proce- dures ave necessary prior to microscopy. Gen- cally, one must be careful not to handle oF damage the fracture surface. The origin and features can be fragile, and key information ‘explaining the facture can be lost by improper handling, For instance, fingerprints can ‘be 748 / Engineered and Electronic Materials Fi reaction-bonded SiN, (RBSN) 11 SEM photomicrographs of fracture-i Te fw shown are typical and constet with he rarmelmicroacue ondsrngth.Arows pint Yo the rota] mistaken for Wallner lines. Debris on the sur face can obscure the true fracture surface ‘Cleaning can_ sometimes be accomplished with compressed air, but the source must be censidered. Some compressors mix small mounts of oil with the air, which could pro: ‘duce a thin surface film that sould later result in imerpretation difficulties during SEM analy: sis. Compressed air should not he used unless it 4s known to be clean Ultrasonic cleaning in a clean solvent, such as acetone or methyl alcohol, is frequently used, Caution and judgment must be exercised, however, because ultrasonic cleaning is quite Vigorous due to the cavitation action at the fuidspecimen interface and can damage the fracture suifauc, For instanee, it racture Int ated at a low-density egion oF a soft inclusion, this material might be removed during ult sonic cleaning and either prevent interpretation ‘oF lead to-an erroneous interpretation. is apparent that cleaning and banding should be avoided unless absolutely necessary Before trying other cleaning approaches, uy & soft camel's air brush Step 2 involves examination of the fracture surfaces under a low-power microscope. Ust ally, a binocular microscope with magnification up to 40% is adequate. Sometimes, higher Power and special lighting or contrast features ae required. Stereo photography may also he useful since it accentuates the fracture surface features Preparation of replicas can also be useful; replica often provides better resolution of the fracture features than does examination of the original part. Several methods are available for preparing replicas. A room-temperature method makes use of cellulose acetate, acetone, and polyvinyl chloride (PVC). A ‘thin sheet of Cellulose acetate is placed on a pieve of PVC And submerged in atetone for about 19's. [tis them pressed agains the fracture surface while being flooded wit acetone and held for about 5 ‘min. The acetone is then allowed to dry and the cellulose acetate replica pecled off Replicas can also be prepared with PVC lone: The specimen is heated and PVC pressed fon with a Teflon rod, After cooling, the PVC replica is pecled off, This technique is quick ‘but should not be used if there isa chance that heating of the specimen will alter the face surface The fracture origin ean usually be located by Jow-posver optical microscopy using either the ‘original specimen of a replica, and an aes ‘ment ean be made as t0 whether the Hace resulted from & material Maw of some ofher factor. However, details of the fracture eign seh as the nature of the flaw and interaction between the flaw and the microstructure, te quite higher magnification. Opyical- mic Scopes do not have acequate depth of fous at high magnification, so the scanning clectoa microscope must be uscd Examination of the fracture surface. in pt Uicular the fractore origin, by SEM i he thin step in tractography. The scanning electron microscope provides extremely large depth of focus (compared to the optical microscope nd ‘range in magnification from around 10% to Wwell over 10000 x, Most fractography is con ducted between 25 and 5000 ‘Scanning eleeiron microscopy’ shows the ference between the fracture origin and the Surrounding material and helps the engineer to develop a hypothesis of the cause of failure (One can frequently detect ifthe fracture int ated at a machining groove or ala material ore for inclusion. Scanning electron microscopy show’ ifthe surface region is different fom he Interior and provides. visual evidence of the nature oF cause of the differences |AS with other tcelnigues, SEM regies Interpretation and must be used with cation Because of the large depth of foeus, itis sometimes difficult to differentiate between a ridge and a depression of to determine te angle ‘of intersection between two surfaces. This an be better appreciated by comparing & single ‘SEM photomicrograph with a sereo puro the same surface. Only after doing this does one Understand how easy it is (0 misinterpret a feature on an SEM photomicrograph, Difficulty in interpretation is inevitable, bu ean be minimized if the engincer is present SEM analysis. This is especially tue With respect te artifacts. Am artifact i defined a extraneous material on the surface of the specimen. It can be a particle of dust o int, chunk of debris resulting from the fracture or hhandling, oF & smeared coating resulting from ci contamination. If the engineer suspects that a feature isan artifaet, he or she can instruct the SEM operator to look at it rom differ angles and to examine surrounding at efor to be sure ‘As noted, SEM can usually locate the frac ture origin and provide a photograph with « calibrated scale that allows accurate measure ment of the size and shape of the Maw and the size of the fracture mirror. The engineer can thon use Eq 1 and 3 to estimate the magnitade ‘of the tensile stress that caused failure Once the fracture origin has been located, ealuation of the localized chemistry is often desirable. This leads to step 4 of fractography, Fig. 12 SEM photomicrographs of abnormal fracture-initiating material RBSN traceable to improper processing prior to nitriding woe deciing. (b) Crack in greoners prior Yo dng flaws (o) Large poce in tp-cost RESN retlting from nod {6 ond eh Low deny os © 100 jm w use of instrumental techniques to conduct microchemical analysis. Most scanning elec- tron microscopes have an energy dispersive scray (EDX) attachment that permits chemical analysis of the x-rays that are emitted when the SEM elecron beam excites the electrons within the material bed Each chemical element gives off scrays under this Stimulation that are characteristic “of that meat alone. These are detected by the EDX quipment and displayed as peaks by periph fal equipment, Comparison of the peak hela fF each clement provides a semiquantitative chemical analysis f the microstructural feature being ¥ Electron probe microanalysis works on the same principle as EDX, except an alternate fon-mgchanism is used that pro wed 100 pm Vides better resolution and detects a wider range of elements. Energy-dispersive x-ray analysis cannot detect the lower-atomic-number ele Auger electron spectroscopy provides addi tional features for chemical analysis. of microstructural features, It ean remove surface material by sputcring while itis conducting chemical ‘analysis and can thus determine ‘changes in chemical composition asa function fof depth. Auger electron spectroscopy is, especially useful in cases. where oxidation corrosion, slow crack growth, or other imergranular effects are suspected. In some cases, fracture of specimens ean be conducted in the Auger apparatus under high vacuum prior to conducting the chemical analysis. This allows examination of fresh fra Failure Analysis of Ceramics / 749 that has not had a chance 10 pick up 0 tamination from the atmosphere and handling Tn addition to analysis of the fracture surface, ‘other material tests ean help determine the cause of failure. These include surface and blk ray diffraction analysis, reflected-light mi croscopy of polished specimen sional cases, transmission electron microscopy Detaled information on al of these techniques is available in Volume 10 ofthe 9th Editon of Metals Handbook ‘There is no guarantee th fractography will explain the cause of a failure and suggest a solution. However, iis sill the ‘mos effective technology available and should be used routinely the four steps of Determining Failure Cause [As mentioned above, determining the cause of failure is eral. It is obviously important in liability suits, where esponsbiiis fr failure ‘must be established, but ii also important for other reasons: + To determine if failure is resulting design or materia! limitations + Toad in material selection or modification + To guide design moditieaions + To identity unanticipated service problems such a oxidation of eotesion To identify material or material processing limitations and suggest direction for i provement + To define specification oquirements for ma terials and operating conditions Figure 10 shows schematically how fractog raphy interacts with other sources of informa tion to determine the cause of failure and ead to action i the tight direction to achieve a solu tion, Ics imperative that all sources of data be considered and that the source of fracture be isolated to determine as quickly as possible whether itis design or materials oriented The following paragraphs review some of the common causes of fracture and describe the typical fracture surfaces that result ‘Material Flaws, Flaws i rial concentrate sues, Fracture occurs shen the concentrated stress at an individual fase reaches a critical value tha i high enough (0 initiate and extend a crack. Therefore, the fist thing 10 look for on the fracture surface is whether there is material flaw atthe fracture ‘vig Hai, sul a pore or icons present, the engineer can do the following 1. Compare the nature of the aw with prior cemification and scrvice experience to de termine if the flaw i intrinske tothe normal baseline material ori abnormal Inthe later ' problem in the material fabric process probably exists, and the help ofthe Component manufacturer shoul he slicited 750 / Engineered and Electronic Materials ig. 13 SEM photomicrographs of abnormal fracture: 1g mater flaws in RBSN traceable to nitriding process (cand (6) Poros cgaregnte rch in chromo andro reuing fem rection ofthe sion dering the iting 2, Measure the flaw size anor mirror size and estimate the Fracture stress using Eq | and 3. Compare this with the stresses estimated by design analysis and with fracture stress istribution projected for the material from prioe strength certification testing, If the Calculated fracture sttess is within the nor ial limits specified for the material, a de iw sel contamintin picked up dung powder proceaing.(] Energy dapereve ray enalyes + sogrgste of Unveacedslconreling rom Inealized meling se to Tool exer sign problem should be suspected. If the fracture stress is below the normal limits specified for the material, a material prob: flaw is at or near the surface, assess Whether the Maw is intinsie to the material ‘or resulted from an outside source, such a8 rmachinit ial expo- sure, Comparison of the chemical and phys feal nature ofthe flaw withthe suerounding baseline microstructure will help in makin this assessment Figures 11 to 14 illustrate the types of intin sie material flaws that can cause fracture and Compare whether they are normal or abaoemal for the material. Although the types of flaws vary depending on the material the fabrication process and the specific step i the process in Whieh they formed, atively easy ( distinguish between normal and abnormal flaws and thus to determine if errors in processing ccntributed tothe failure (Re 13), Machining Damage. Surface flaws result. ing from mac econd common source of failure in ceramic components, espe cially in applications where igh bend loads oF thermal loads are applied in serve. The most important Maws resulting from machining are median cracks and radial cracks The median crack is elongated in the die tion of grinding and is like @ notch in that fracture usually initiates over a broad front. A broad mirror usually results, but no Maw is readily visible because of the shallow initial Flaw dept aad Because the median crack perpendicular to the surface. The principal tensile stress is usually distributed such thatthe ‘rack will extend ina plane perpendicular 10 the If the flaw is perpendicular to. the Sueface to stat wth, it wll be in the same plane asthe fracture and willbe difficult wo deren tial from the rest of the fracture mirror Figure 15 shows examples of specimens that fractured at transverse grind marks such that the median crack sae Hkely the virongth determining flaw, Note the length of the surface involved in the fracture origin and the lack of a distinguishable flaw. Note also that the facture origin is at a grinding groove and is elongated parallel to the direction of ringing Radial cracks produced by machining are roughly perpendicular to the direction of ma chining, are usualy shallower than the median racks. andl are usually semicircular rather than elongated. The resulting fracture mirror is i ilar to one produced by u small surface pore or inclusion, but a well-defined flaw isnot visible. However, by examining the itersetion ofthe fracture surface with the original specimen surface atthe center of the semicircular mirror under high magnification with the scanning electron microscope, the source can usually be seen and interpreted as machining damage Tere are two features to look for. First, check to see if a grinding groove, especially an unusually deep one, intersects the fracture sur face at the origin. Second, look for a small slightly out-of plane region at the origin. This, could indicate machining. damage, but could also have other interpretations, such as contact damage or simply a tensile overload Machining damage often limits the strengt fof fine-grained ceramics and determines the surface Fig. 14 SEM photomicrographs comparing normal and abnormal materi flaws in sintored SiC (clon (0 ype mirc of hghsrenath soar (0) Longe gore retuing from poner epalomercion ower preperation ond shape forma oni! measured strength distribution of cenification specimens. Rice ef al. (Ref 14) reported that hot-pressed SiN, specimens machined in the transverse direction fractured at» machining flaws 98% of the time and that bars machined in the fongitudinal ditection failed at machining faws greater than 50% of the time. This is Similar to results for borh SisNy and SiC (Ret 1s-17) Residual Stresses. Many ceramics have residual stresses, which usually result from the sutface cooling faster than the interior after Sintering oF ate due to. che between the surface and interior. Often, the erir is under residual tension and the exte rior is under compression, This can provide a ing effect if the material is used in al differences Large aris resting om imprer cnvl of tmperture @ service in the as-fired, prestessed condition However, most components require some finish ‘machining. Once the compressive surface zone has been penetrated, the component is substan: tially weakened, often to the point of spontane ‘ous crack initiation during machining Frechette (Ref 18) recounts a case whe simple blanks of boron carbide cracked during machining. At the time, the cause of cracking ‘vas unknown, but improper machining was the primary suspect. Fracture analysis showed that the crack initiated at the point of machining and mtered the material roughly perpendicular to the surface. However, the erack then quickly changed direction and propagated through the interior of the material parallel to the surface and finally changed direction again and! exited Failure Analysis of Ceramics / 751 hough the bottom surface. The fracture path and fracture surface features (Walle lines and hackle) indicated that the surfaces were in ‘compression and the interior was in tension and suggested that the problem was not linked t0 machining practice or mterial defects, Review ofthe processing history showed thatthe boron wide was allowed to coo! Feely tron 1980 1800 °C (3340 t0 3270 F) alter hor pressing This was followed by thereafter, The free cooling was within the reep temperature range ofthe materi ing in an effect comparable to temper ass that i, formation of srace compression 1 internal tension. Based on this hypathesis derived from fracture surface analysis alone. the hot-pressing operation was modifiod 10 permit slow cooling Irom 1980 to [750 °C (3540 «3180 °F), This eliminated the residual stress conlton and permitted machining wit cut cracking Thermal Shock, tic his een reported vn identification of thermal shock fractures, and ‘much additional work is needed. Therelore the ‘comments here will be fatly genera reader should be aware thatthe distin Features described may not always occur and muay not be the Best anes In a specific case. As in all other interpretive studies, the best ap proach is to apply known principles, thoroghiy tnalyze al available dats and options and thea make a decision, rather than depending on & cvokbook procedure Thermal shock fractures tend 0 follow a result wavy path with minimal branching and produce father featureless fracture surfaces. This p pears to be especially true for weak or fmoderate-strength matcrals, inchudi lass and polycrystalline ceramics. It is com ‘on for thermal shock cracks not to propsate all the way through the part istes suggest that « nonbniform stress Feld is present during thermal fracture; thit is, the fracture initiates where the tensile stess is highest, follows stress oF temperature contours (thus the wavy path), and tops when the sti trope below the level required for futher (etre ae when ected. These conditions would also suggest & Tow crack velocity, which would account for lack of branching and lack of fracture surtace features ns Tine with the Foregoing general consider ations, several actual cases of fracture of mate rial damage duc to severe thermal transients are now analyzed The fist ease is the water quench thermal shock test for comparing the relative temper ture gradicnt required to cause damage in sim test bars mined tiny dropped into a controlled-temperature_ water bath, The bars generally donot beak de tothe «quench. Instead, a large number of microcracks ‘te produced, These become critical flaws that Tead to fracture during subsequent ben esting: 752/ Engineered and Electronic Materials Fig. 15 SEM photomicrographs showing fractures machining damage ing at transverse (o) The race sufoce of a tne spcinen of ho-prenied licen niride thal had bean machined reanlrenaly. (The intersection ofthe ochre srace wih the machined sroceotng tht the © @ the size of the Maws is dependent on the material propertics and the temperature change AAT. and the residual strength is dependent on the size of these eracks. Further discussion of this tort procedure and results on specific ma terials is available in Ref 19 and 20 Ammann ef al. (Ref 21) conducted eyelic Muidized bed thermal shock tests on. wedge shaped specimens of hot-pressed Si,N,. They ported that multiple surface cracking occurred and thatthe cracks grew in depth as a function ofthe total number of cycles. Oxyacetylene torch thermal shock testing and gas turbine rig testing of resction-bonded SiNy stator vanes has shown that thermal shock fractures of actual components under service conditions can be quite varied (Ret 22) Figure 16() shows atypical featureless fracture surface. The origin is shown by the arrow and the crescent shaped ink mark and occurred in a region predicted by three-dimensional finite element analysis to have the peak thermal stress ‘ding rapid heat-up. No material flaw or ma chining damage is visible, even at high mag © paral othe ining grooves) od a} The sane stvaton for reaction smered icon en ® fication, as shown in Fig. 16(b). ‘The crack apparently initiated at the material surface ‘where the stress was maximum and propagated at moderate velocity without branching or mak ing any abropt changes in direction Figure 17(a) shows a thermal shock failure that initiated athe trailing edge of a stator vane Airfoil. Finite element analysis also determined this position to be under high thermal stress during heat-up and steady-state service cond tions. However, in this case the thermal shock fracture initiated ata preexisting material flaw, as shown more clearly in Fig. 17(b) at igh magnification. The flaw was « penny-shaped crack nearly normal tothe surface of the arf bat out of plane to the principal tensile stress ‘The fracture initiated at this erack due to the thermal stress and then quickly changed diree- tion t@ follow the plane of maximum tensile stress. Hackle marks ean be seen in the upper part of Fig. 17(a), which point tothe vicinity of the origin ‘The fracture surface shown in Fig. 17 is not ‘what one would expect for thermal shock con ig. 16 SEM photomicrograph showing a typical featureless thermal shock fracture surface (2) Overall soce ot low mogefication. (s) Frachre ‘gn ot higher mognision:Courteay of Gara {bine Engine Company. Dison of The Gort © ditions, It has relatively well-defined fisture features and could just as easily have. been interpreted as a mechanical overload, Simi lary, the fracture surface in Fig. 16 looks very ‘much lke fractures resulting from contact loa ing @iscussed later). How does the engine make the distinction? At our current state of Knowledge, he or she does not, at least not based solely on fracture surface examination. The engineer needs other inputs, such as sess analysis, conteolled testing (sich ae the eali= brated oxyacetylene torch thermal shock tests), and a thorough knowledge of the service con- ditions. The engineer then needs to evaluate all the data concurrently and use his of her best judgment Impact can cause damage or fracture in 180 ways: (1) localized damage at the point of Impact and (2) fracture away from the point of contact due 6 cantilevered loading. The former ig. 17 SEM photomicrograph of thermal shock fracture ing at a material flaw (6) Oreo foe of lay mop {By Proeniing creck of Rechre erin, Courtey of {Gort ferne Engine Company Bion of The will have distinctive features and can usually be idemiied as caused by impact. The latter will ‘appear like typieal bond overload (with a compressive lip onthe exit end of the fracture) and ean ony be linked to impact by supporting data, such 38 location of the Tocal damage atthe point of impact “The damage atthe point of impact may be so Title as to resemble a Scuff mark on the ceramic ‘ra smear ofthe impacting material (both eases referred to as a witness mark) of it may be as severe as complete shattering (Ref 23) ‘The degree of damage depends on the rela- tive velocity, mass, strength, “and hardness of the impacting bodies, baseball strike ing a plate glass window causes shatering; a BB sinking a window only causes a series of concentric cone-shaped (Hertzian) cracks imersected by radial cracks. However, if the Window shattered by the baseball could be reconstructed, it would also have conical and radial cracks |. 18 Typical Hertzian cone crack resulting from impact and acting fas the flaw that resulted in fracture under subsequent bend load Shown ones oan fram 0. erbine Engine Company, Dvson of The Gore Failure Analysis of Ceramics / 753 |. 19 Impact fracture of a ceramic rotor blade showing Hertzian cone crack Coury of Greet Terine Engine Company, Division The Gare Coporchon o ‘The conical cracks are typical of impact and fare thus a stony diagnostic fracture age occurred first and was followed by fracture ‘ue toa bend load. The origin is indicated by an row and is easily located by observing the hackle Fines and fractore miror. Figure 18(b) and (c) show the origin at higher magnification. A dstinet cone shape is presen; the apex is at the surface where the impact occured, and the fracture flared out as the crack penetrated the material Figure 19 shows another example of fracture due to impact. In this ease, a ceramic rotor blade rotating at 41.000 rpm struck a foreign object and was fractured by the impact, Note the Hertzian crack extending from the fracture origin, Also note that additional damage is present at the origin, possibly a series of con centric cracks, providing evidence that the de tee of impact was severe 754 / Engineered and Electronic Materials Fig. 20 Contact cracking in a ceramic 2} and (b)Surfoce cracks resing from rlothe movement between two covtatsrfces under high norma ood ond wih hich coef! of rion. (0 Typael multe chpsng railing om conic oding and viable on roche race Fig. 22 SEM photomicrographs of the fracture surface of hot-pressed isNq exposed to static oxidation for 24 h at 1100 °C (2010 °F) (6) Overall acre sre showing hake ark oe rte mi rue ogi suggesting recur due {b} ord (c tolpl cone che {@)Higher magico shot the scture of he and the rome conto rochwe O.e4" 0.32 em (0.25 » 0.128 im) Source: Roh 18 Fe, NiCr Fracture cot) fom) Fig. 23 SEM photomicrograph of the fracture-initiating oxidation-corros ‘on the surface of reaction-bonded SiN, The EDK graph show the relative concretion of charcl slants in the gly region af the bane of he pit (Courtesy af he Garret Turbne Engine Company, Dison of The Gare Corporohon Fig. 24 SEM photomicrograph of hot-pressed jh sea salt additions showing that fracture initiated combustion gases IN, that was exposed to at the base of the glassy surface buildup EDK onciis shove the chemical elaants detected in he glosty material ean othe SiN. Couey ofthe Goren Terbie Engine Company, Drson of The Gan Biaxial contact refers o a situation where normal and tangential Forces are being applied Simultaneously ata ceramic surface. Examples where this might occur are numerous and! in clude the following + Surface grinding ‘Corporation Sliding contact (such as in bearings, seal and many other applications) “Applications involving a shrink fi Interfaces where the materials have different thermal expansion coefficients and operate under varying temperatures Failure Analysis of Ceramics / 755 Tensile stress concentration at a biaxally Joaded interface is discussed in more detail in Ref 25 and 26. The fracture surface canbe quite varied. I the contact is concentrated a point a Heriian cone crack can form ing fracture surface willbe similar to the one in Fig. 18, However, ithe contact ix more »prea fut, as in high friction sliding contact. the will be spread over a larger surface area and will result in a relatively fe ture surface similar wo the one shown in Fig, 16 for thermal shock fracture Distinguishing features that liferentate contact faire from a thermal shock fal just beginning to be defined (Ref 25) and Appear to include the folowing: andthe result + Contact cracks tend to enter the material surface at an angle other than 90°. An tetreme case is shown in Fig. 20ka) for a ceramic specimen with a high-inerference shrink fit axially loaded in tension *+ Multiple parallel cracks often occur during 3 ‘contact Failure and can either bes Surface adjacent to the fracture origin oe ean show up as chips obscuring the fracture origin (Fig. 206), + Surface witness marks are often present at the fracture origin in cases. of conti initiated failure (Fig. 21a) One oF more clamshell shapes, faint Here ian cones, oF pinch marks occur a he origin fof some contact-initiated fracture suraves (Fig. 21b) Oxidation-Corrosion. Some types of ox: ation oF corrosion a they leave substantial surface clearly visihle to the naked eye the objective isto identity the mechanism of atack ‘and find solution especially where the oxidation or corrosion i isolated along grain boundaries, the presence ‘and. source of degradation may. be dificult to detect. In this ease, the degree of attack may only ‘be determined by strength testing, and the cause may be ascertained by controlled environment exposures andlor s0- Phistcated instruments, such as Auger ect Spectroscopy, which can detct slight chemical Variations on a microstratural level Let us first examine some examples of oxidation and comrsion where visible surface changes have occurred, Figure 22 shows the surface and fracture surface of NC-132 hot pressed SiN, after exposure in an SiC Fesistance-heated, oxide-relractory-ined_ fur nace for 24 h at 1100 °C (2010 °F) (Ref 27) Figure 22(a) shows the complete cross section of the tes bar. The fracture origi i atthe surface fn the left side of the photograph and is easily located by the hackle marks and the fracture rmiror (the dark spats onthe fracture surface are artifacts that accidentally contaminated the sur face in preparing the sample for SEM). The specimen surface appears at low magnification to have many small spots that were not present Prior to the oxidation exposure. At higher mag: easy 10 detect Hecate e that Tn this 756 / Engineered and Electronic Materials Fig. 25 SEM photomicrograph of reaction-bonded Si,N, that was exposed te combus! n gases with sea salt additions showing that fracture initiated at the base of the glassy surface buildup EDX ona show the chemical elements detected inthe loty materiel odjocnt to the SiN Courter of he Gort Trine Engine Compony. Onion of The Gort Corporation nificaton (Fig. 22b), these spots appear to be biisters or popped bubbles and one i precisely atthe fracture origin, Still higher magnification (Fig. 22) reveals that a glass-filled pit is atthe base on the eenter of the blister. It also reveals that a surface layer less than 5 jm thiek covers the specimen and that this layer appears {0 be partially crystallize. By simply examining the specimen surface specially the intersection of the oxidized sur- face and the fracture surface, we have obtained much insight into both the nature and sequence ‘of oxidation, What else can we do to obtain further information? We can compare the sirongth of the oxidized specimen with that of unoxidized material. In this specific case, the ‘Oxidation exporure rovulted in a reduction in sirength from 669 MPa (97 ksi) to 497 MPa (72 ksi). We can also compare xray diffraction and chemical analyses for the original surface, the oxidized surface, and the bulk material. In this case, the oxidized surface contained much ‘more magnesium and caleium than the original surface or the bulk material. Energy-dispersive ‘ray analysis verified that the glassy material im the pit also had high concentrations of calcium and magnesium. X-ray diffraction re vealed crystallize eristobaite (SiO) plus mag- rnesium silicate and calcium magnesium silicate phases in the oxide layer. No sign of magne ‘ium or calcium contamination was detected in the furnace, ‘Simultaneous evaluation ofall the data led to a plausible hypothesis of the mechanism of fxidation degradation. Magnesium and. ca: ‘ium, present as oxide or silicate impurities in the Si,N, were diffusing to the surface, where they reacted with SiO, that was formig simul: taneously at the surface from reaction of the ‘SiyNg with oxygen from the air. The resulting silicale compositions apparently locally in- creased the solubility or oxidation rate of the SI,Ng. The reason forthe formation of isolated Pits Was not determined, but could have sulted from impurity segregation or other fac- {ors and would have requited additional studies to determine. ‘A similar example of static oxidation for reaction-bonded SiN, is illustrated in Fig. 23. In this case, the exposure was for 2h at 1350°C (2460 °F) plus $0 hat 900 °C (1650 °F) (Ret 13). Only isolated pits were present on the surface, and those appeared (© oveur whore simall particles of the furmace lining, had con- ‘acted the specimen during exposure. The EDX analysis included in Fig. 23 ws taken in the _lassy region at the base of the pit, showing that aluminum, silicon, potassium, ‘alcium. and iron were the primary elements present and ‘again indicating a propensity for Si,N, 1 be corroded by alkali silicate compositions. How ‘exer, it should be noted thatthe size of the pit is much smaller than in the prifr example and resulted in only a small strength decrease Figures 24 and 25 show examples of more

    You might also like