Pesquisar no resumo do documento
Properties, Processing, and Use in Design
Second Edition, Revised and Expanded
Prel." to the Second Edilion Preface to the First Edition
Pad I STRtlCfllRES AND PROPERTIES
1 Atomic: Bonding, and Crystal Structure
3 2 Crystal Chemistry and Speci6c Crystal Structures 32
3 Phase EguUibria and Phue Equilibrium Diagrams 71 4 PbysicaJ and }bennal Behavior 123 5 Mrtblola! Bcbuior and Measurement 162 6 EI.drirl' Behayio[ 204
1 Dieledrict Magnetic. and Optical Behavior 1S1 8 Time, Temperature, and Environmental Elred!
on PropeJ1Jes 313
Part II PROCF,sSING OF CERAMICS 373
9 Powder Processing 10 Shape·FonnlllR Processes J 1 D'nqfinlinn
12 Final Macbining 13 Quality Assurance
pad II( DESIGN WITH CERAMICS
14 DesI&n Considerations 15 Deslp Approaches
16 FaOure Analysis
17 TougbeDing of Ceramics 18 AppUalions: Material Selection
Glossary EWed:ive Ionic Radii (or CalioD' aod AniOBS
periodic Table of the Elements
374 418 519
731 808 833 843
4 Chapter 1
The second shell has eight electrons. two in s orbitals and six in p orbitals. All have higher energy than the two electrons in the first shell and are in orbitals farther from the nucleus. (For instance . the s orbitals of the second shell of lithium have a spherical probability distribution at about 3 A radius.) The p orbitals are not spherical. but have dumbbell-shaped probability dis- tributions along the orthogonal axes, as shown in Fig. 1.1. These p electrons have sl ightly higher energy than s electrons of the same shell and are in pairs with opposite spins along each axis when the shell is full.
The third quantum shell has d orbitals in addition to sand p orbitals. A full d orbital contains 10 electrons. The fourth and fifth shells contain f orbitals in addition to s. p. and d orbitals. A full f orbital contains 14 e lectrons.
A simple notation is used to show the electron configurations within shells. to show the relative energy of the electrons, and thus to show the order in which the electrons can be added to or removed from an atom during bonding. This notation can best be illustrated by a few examples.
Example 1.1 Oxygen has eight e lectrons and has the electron notation Is'2s'2p'. The I and 2 preceding the sand p designate the quantum shell. the sand p designate the subshell wi thin each quantum she ll . and the su- perscripts designate the total number of electrons in each subshell. For oxygen the Is and 2s subshells are both full . but the 2p subshell is two electrons short of being full.
Example 1.2 As the atomic number and the number of electrons increase . the energy difference between electrons and between shells decreases and overl<.lp between quantum groups occurs. For example . the 45 subshell of iron lills before the 3d subshell is full. This is shown in the electron notation by
Figure 1.1 E lectron probability distributions for p orbital s. The highest probability electron positions are along the orthogonal axes. Two electrons. each with opposite spin. are associated with each axis. resulting in a total of six p electrons if all the p orbitals in th e shell are filled .
90 Chapter 3
. , . •
Figure 3.18 Transmission electron micrograph showing an example of liquid im- miscibility. (Courtesy of D. Uhlmann. University of Arizona .)
Polymorphic transformations are also shown on phase equilibrium dia- grams. Figure 3.20a is a schematic of a binary eutectic diagram with no solid solution and with three different polymorphs of the A composition. The different polymorphs are usually designated by letters of the greek alphabet. Figure 3.20b is a schematic of a binary eutectic diagram with three A polymorphs. each with partial solid solution of B.
Figure 3.21 illustrates a real binary system with polymorphs. Poly- morphic transformations are also present in Fig. 3.19.
A three-component system is referred to as a tertiary sysfem. The addition of a third component increases the complexity of the system and of the phase equilibrium diagram. The phase rule becomes F = 3 - P + 2 = 5 - P. As with binary ceramic systems. diagrams are usually drawn with pressure as a constant (condensed system). The phase rule for the con-
174 Chapter 5
Figure 5.S Scanning electron photomicrographs of fracture surfaces of reaction- bonded silicon nitride containing nearly spherical pores resulting from air entrap- ment during processing. Arrows outline flaw dimensions used to calculate fracture stress.
178 Chapter 5
)<", 1-- - ~
Figure 5.7 Typical ceramic tensile test specimen configuration.
Another method of obtaining tensile strength of a ceramic material is known as the theta test . The configuration is shown in Fig. 5.6c. Applicaton of a compressive load to the two arches produces a uniaxial tensile stress in the crossbeam. Very little testing has been conducted with this configuration owing largely to difficulty in specimen fabrication.
Compressive strength is the crushing strength of a material, as shown in Fig. 5.6f. It is rarely measured for metals . but is commonly measured for ceramics. especially those that must support structural loads. such as re- fractory brick or building brick . Because the compressive strength of a ceramic material is usually much higher than the tensile strength, it is often beneficial to design a ceramic component so that it supports heavy loads in compression rather than tension . In fact. in some applications the ce- ramic material is prestressed in a state of compression to give it increased resistance to tensile loads that will be imposed during service. The residual compressive stresses must first be overcome by tensile stresses before ad- ditional tensile stress can build up to break the ceramic, Concrete pre- stressed with steel bars is one example . Safety glass is another example.
192 Chapter 5
Figure 5.16 Simple schematic ill ust rating a screw dislocation. (From Ref. 7. p. 92 . )
the structure is distorted and under localized stress even when the overall material is not under an app lied stress. This residual stress state can be visua lized by examining Fig. 5.17. The dislocation ]jne extends into the structure perpendicular to the surface of the page. Note that the structure is distorted so as to fill in the space of the missing half-plane of atoms. This results in a state of residual tens ile stress just below the ext ra plane of atoms balanced by compressive stress in the region above the dislocation.
The presence of the dislocations and the associated residual st ress allows slip to occur a long atom planes at a fraction of the £ /20 value that
Zone of compressive stress ~ Zone of tensile stress €E>
Figure 5.17 Schematic of the residual st ress state showing compressive stress above the dislocation and tensile stress below the dislocation. (CI ASM In te rna· tional. )
Figure 5.23 Crystal structure of AI~o.\ showing complex paths O!- and Alh ions must follow to allow slip to occur under an applied stress. (From W. D. Kingery et al..lntroduction co Ceramics. 2nd ed .. Wiley. New York. 1976. p. 732.)
;: ~ ". ~ = ff
'" ~ ". ~ • o· " ~
= ... ;: ~ ~
~ ~ 3 ~ -
Electrical Behavior 243
Figure 6.23 Example of the Meissner effect showing the levitation of a magnet at liquid nitrogen temperature by YBa!Cu.t01., ceramic superconductor. (Courtesy Ceramatec. Inc .)
The response of the superconductive material to the amount of current being carried or to an applied magnetic field is also very important. Too high a current density or magnetic field can destroy the superconductive behavior. Each material has a different response.
Evolution of Superconductor Materials
Figure 6.24 shows the historical progression in discovery of superconductive materials with higher T,. Progress was extremely slow up to 1986, averaging about 4 K per decade . Initial materials identified to be superconductive were elemental metals (Hg, Pb, Nb), followed primarily by solid solutions (NbTi) and intermetallics (Nb,Sn , V,Si , Nb,Ge). Until the early 1960's , relatively few materials had been identified with superconductive behavior. Superconductivity was thought to be an anomalous property . Since 1960, techniques have been avai lable to achieve temperatures closer to absolute zero (on the order of 0.0002 K) and to simultaneously apply high pressure. Under these conditions many more elements, solid solutions, intermetal- lics, and ceramics have been demonstrated to have superconductivity.
Several ceramic compositions were identified to be superconductive. These included tungsten, molybdenum, and rhenium "bronze" composi- tions A,WO" A,MoO .. and A,RhO" where A was Na, K, Rb, Cs, NH" Ca, Sr, Ba, etc.; oxygen-deficient SrTiOJ and LiTi03; and BaPb, _.Bi.OJ.
Dieleclric. Magnelic, Optical Behavior 275
equal probability of shifting in six directions toward one of the corners of the octahedron. As a result. the tetragonal crystal contains some dipoles in one portion of the crystal pointing in one direction, whereas others in another portion may point in a direction 900 or 1800 away from the first. A region of the crystal in which the dipoles are aligned in a common direction is called a domain. An example of BaTiO.1 with a ferroelectric domain with aligned dipoles is illustrated in Fig. 7.18.
Le t us return now to Fig. 7.16 and describe what happens in a ferroe- lectric crystal such as tetragonal BaTiO, when an electric field is applied. The ferroelectric domains are randomly oriented prior to application of the electric field, that is, at E = 0, the net polarization equals zero (P,,, = 0). As we apply an electric field and increase the electric field, the domains begin to move in the BaTiO.\ and align parallel to the applied field . This results in an increase in net polarization along line OA. The polarization reaches a saturation value (8) when all the domains are aligned in the direction of the field. If we now reduce the electric field to zero, many of the domains will remain aligned such that a remanent polarization (P,) exists. Interpolation of the line 8e until it intersects the polarization axis gives a value PJ , which is referred to as the spontaneous polarization. If we now reverse the electric field, we force domains to begin to switch direction. When enough domains switch, the domains in one direction balance the domains in the opposite direction and result in zero net po-
Figure 7.18 TEM image of 180 0 ferroelectric domains in a single grain of BaTiO,. (Courtesy of W. E. Lee, University of Sheffield.)
Dielectric, Magnetic , Oplical Behavior 285
Another important wave-generation application is the sonic delay line. A delay line consists of a solid bar or rod of a sound-transmitting material (glass , ceramic, metal) with a transducer attached to each end . An electric signal that is to be delayed is input to the first transducer. The signal is converted to a sonic wave impulse that travels along the sound-transmitting "waveguide." The sonic impulse is then converted back to an electrical impulse by the second transducer. The delay results because a sonic wave travels much more slowly than electrons passing through a wire. The time of delay is controlled by the length of the waveguide. Delay lines are used extensively in military electronics gear and in color te levision sets. One example is radar systems to compare informatio n from one echo with the next echo and for range calibration.
The wave-generation applications discussed so far involve acoustic waves transmitted through bulk media. Additional freedom exists in the
Figure 7.25 Piezoelectric ceramics and assembli es for a variety of applications. (Courtesy EDO Corporat ion.)
324 Chapler 8
Figure 8.7 Hot-pressed SilN~ specimen deformed by creep under a load of 276 MPa (40.000 psi) at llOOOC (-2200' F) for 50 hr.
mechanisms available for crack growth. Crack growth is relatively easy if the grain boundaries of the material are coated with a glass phase. At high temperature, localized creep of this glass can occur, resulting in grain boundary sliding . Figure 8.8(a) shows the fracture surface of an NC-132 hot-pressed SiJN4 specimen that fractured after 2.2 min under a static bending load of 276 MPa (40,000 psi) al - llOO' C (- 2000' F). The initial flaw was probably a shallow (20 10 40 pm) machining crack. It linked up with cracks formed by grain boundary sliding and separation and pores formed by triple-point cavitation to produce the new Haw or structurally weakened region seen in Fig. 8.8 as the large semicircular area extending inward from the tensile surface. This was the effective flaw size at fracture.
Time, Temperature, Environmental Effects on Properties 325
Figure 8.8 Comparison of a slow crack growth fracture versus a normal bend fracture for hot-pressed Si.1N •. (From Ref. 9.)
... '.-.. -. ...... , ';'-- ---- -.- ~ - . ~ -.:.,.-:-: . ~~..:-~- .. ..;.-:~..;. ... . - --" -~ ~ . .:.:.:.;.:;:
Figure 8.11 Surfaces of hot-pressed Si)N. before and afte r oxidation. (a) As_machined surface, 32O-gril diamond; (b) oxidized in ai r for 50 hr al 98O"C (l8OO"F); (c) oxidized in air fo r 24 hr at 12()(rC (22OO"F); and (d) oxidized in air for 24 hr at 137O"C
(25OO"F). (C ASM International.)
Figure 8.23 Reaction-bonded SilN~ after exposure in a combustion rig with 5 ppm sea salt addition for 25 cycles of 1.5 hr at 900°C, 0.5 hr at 1120°C, and a 5-min air-blast quench. (a) , (b) , and (c) show the fracture surface at increasing magnification and illustrate the glassy buildup in the region of combustion gas impingement. (From Ref. 9.)
as fouling, A thin buildup can protect the surface from corrosion and erosion and in some cases can even result in a local temperature reduction. All three of these factors can increase the life of a component. especially a metal. However, a thick buildup reduces the airflow through the engine and decreases efficiency.
Fouling is an inherent problem in the direct burning of coal. A variety of approaches have been or arc being studied to resolve this problem:
L Intermittent removal of buildup by thermal shock, melt-off, or passing abrasive material (such as nutshells) through the system
388 Chapter 9
Figure 9.4 Si.,NJ grinding media showing one of the common configurations. Spheres are also commonl y used. (Courtesy KemaNord.)
wear-resistant linings and have been used successfully with dry milling and with water as a milling fluid. However . some milling is conducted with organic fluids that may attack rubber or po lyurethane. Very hard grinding media can reduce contamination because they wear more slowly. we is good for some cases because its high hardness reduces wear a nd its high specific gravity minimizes milling time. If contamination from the media is a n especially critical consideration, milling can be conducted with media made of the same composition as the powder being milled. Another ap- proach is to mill with steel media and remove the contamination by acid leaching.
Milling can be conducted either dry o r wet. The advantages and dis- advantages are listed in Table 9.5. Dry milling has the advantage that the resulting powder does not have to be separated from a liquid . The major concern in dry milling is that the powder does not pack in the corners of
402 Chapter 9
Figure 9.10 Transmission electron microscope image of ultra fine L<i<I. 76Sr/l . 2~CrOl powder prepared by the glycine-nitrate process. (Courtesy of Larry Chick. Battelle Northwest Laboratories, Richland . Wash .)
422 Chapter 10
Figure 10.3 Photo taken with a scanning electron microscope showing the spher- ical morphology of spray-dried powder. (Courtesy Cerarnalec, Inc.)
closest possible packing; and (2) to minimize friction and allow all regions of the compact to receive equivalent pressure. Let us discuss these in more detail and examine some examples.
Binders and Plasticizers
Table 9.13 listed a variety of organic and inorganic materials that have been used as binders. Most binders and plast icizers are organic. They coat the ceramic particles and provide lubrication during pressing and a tem- porary bond after pressing. The amount of organic binder required for pressing is quite low, typically ranging from 0.5 to 5 wt %. Organic binders normally are decomposed during the high-temperature densification step and evolved as gases. Some binders leave a carbon residue, especialty if fired under reducing conditions.
Inorganic binders also exist. The clay minerals such as kaolinite are a good example. Kaolinite has a layered structure and interacts with water
Figure 10_11 Schematic illustrating the different distances a punch must move to accomplish uniform compaction of the powder. Based on a powder wi th a com- paction ratio of 2: 1. (10 ASM International.)
plastically during pressing and conforms to the contour of the die cavity. The pressed shape usually contains flash (thin sheets of material at edges where the material extruded between the die parts) and can deform after pressing if not handled carefully. For these reasons, wet pressing is not well-suited to automation. Also. dimensional tolerances are usually only held to :!:2% .
Uniaxial Pressing Problems
The following are some of the problems that can be encountered with uniaxial pressing.
improper density or size die wear cracking density variation.
The first two are easy to detect by simple measurements on the green compact immediately after pressing. Improper density or size are often associated with off-specification powder batches and are therefore relatively easy to resolve. Die wear shows up as progressive change in dimensions. It should also be routinely handled by the process specification and quality control.
The source of cracking may be more difficult to locate. It may be due to improper die design. air entrapment, rebound during ejection from the die. die-wall friction, die wear, or other causes. Often a crack initiates at the top edge of the part during pressure release or ejection of the part. Two mechanisms of this type cracking are illustrated in Fig. 10.13. The first, shown in Fig. 1O.13(a), occurs as pressure is released from the upper
o Powder e RigId Ole Parts
~ I'Z2l Moving Ole Parts
Figure 10.12 Schematic of tooling to uniaxially press a three-level part. (© ASM International.)
444 Chapter 10
Figure 10.21 Ce ramic parts formed by uniaxial and isostatic pressing, some with green machining. (Courtesy Western Gold and Platinum Company, Subsidiary of GTE Sylvania. Inc.)
When most people hear the term "casting." they automatically think of meta l casting in which a shape is for med by po uring molten metal into a mold . A limited amount of casting of molten ceramics is done in the preparation of high-density Alp) and AI,O .1'ZrO, refractories and in prep- aration of some abrasive materials . In the latter case, casting from a melt into cooled metal plates produces rapid quenching, which results in very fine crystal size that imparts high toughness to the materia l. The technique of cas ting molten ceramic refractories is called fusion cas(ing.
More frequently. the casting of ceramics is done by a room-temperature operation in which ceramic particles suspended in a liquid are cast into a porous mold that removes the liquid and leaves a pa rt iculate compact in the mold . T here are a number of va riations to this process. depending on the viscosity of the ceramic-liquid suspension. the mold , and the procedures used. The most common is referred to as slip casling. The principles and contro ls for slip cas ting are similar to those of the other particulate ceramic
"-u i- ;;; 0 u 0 :;
100 70Wt. % solids
0.05 0.1 0.2 0.25
Volume % dispersant
Monazollne·T Monazoline-C Sedisperse·O Zany! FSN Monazollne·Q Wilconol H31·A Flourad FC-170-C
Wileamlne PA·78B Monawet MM·80 Aerosol AY·l00 Aerosol C-61 Monawel MB·45 Manawa! MO· 70 Olsperslnot·C
Menhaden fish all Emphos PS- 21 A Zonyl·A
I I AMP-95 I Alkazlne-TO I Alkazlne-O
Emeras! 2423 I Dispersinot-HP I Sedlsperse·F I
I I Orewfax.()()7 I Aerosol.()T
I Ouponol.a pva
I Aerosol TR-70 I Amerlate LFA I
Figure 10.32 Summary of the effect of the dispersants listed in Table 10.6 on the viscosity of slips consisting of BaTiO) in a MEK-ethanol solvent. (Adapted from Ref. 12_)
These are referred to as nonaqueous (nonwater-based). Another non- aqueous system utilizes trichloroethylene plus ethanol. Nonaqueous sys- tems work well with steric hindrance because they are adequate solvents for the chain polymers. Some of the polymers also provide steric hindrance in an aqueous (water-based) system, for example, phosphate esters.
Aqueous slips utilizing electrostatic repulsion are commonly used for slip casting. Techniques of slip preparation and slip casting are discussed in the following sections. Nonaqueous slips utilizing steric hindrance are commonly used for tape casting. Tape casting is discussed later in this chapter.
The actual physical preparation of the slip can be done by a variety of techniques. Perhaps the most common is wet ball milling or mixing. The ingredients. including the powder, binders. wetting agents, sintering aids , and dispersing agents, are added to the mill with the proper proportion of the selected casting liquid and milled to achieve thorough mixing, wetting, and (usually) particle size reduction. The sl ip is then allowed to age until
Shape·Forming Processes 465
Figure 10.36 Annular combustor for gas-turbine engine fabricated by drain casting using nonabsorbing pine and mandrel inserted into the mold . (Courtesy Garrett Turbine Engine Company, fabricated by Norton Company.)
channels into the stator vane mold during cast ing. The reservoir and vane patterns are bonded together by simple wax welding and are shown as the white wax assembly in the center of Fig. 10.37. Below this is the mold produced by dipping and dissolving the pattern. Below the mold is the green casting after dissolving the mold and trimming off any material re· maining in the reservoir or gating area. The stator vane discussed above required less than 1-hr casting time. Some solid castings require much longer time, such as the prototype gas-turbine rotor shown in Fig. 10.38. It required over 12 hr. The slip must be very stable for such long casting time to avoid settling of large particles or adverse changes in viscosity.
Other fugitive mold techniques have been developed to fabricate spe- cial shapes. One technique produces low weight, but strong ceramic foam (19). Reticulated foam similar to a dishwashing sponge is used as the mold interior. Ret iculated polymer foam of the desired pore size is cut to the desired shape and placed in a container in a vacuum chamber. A ceramic slip in poured into the container and under vacuum complete ly infiltrates the pores in the reticulated foam. The slip is dried and fired to burn off the polymer foam and densify the ceramic. The resulting part consists of an internal cast of the spongelike foam. Its major characteristic is contin- uous interconnected links of ceramic and continuous pore channels. Such a cellular structure can be very lightweight and surprisingly strong. Ex-
466 Chapter 10
. . "
·1" ". Figure 10.37 The fugitive-wax technique for preparing a complex-shape mold for slip castingj example of fabrication of a stator vane for a gas-turbine engine. (Cour- tesy AiResearch Casting Company. Division of The Garrett Corporation, presently the Garrett Ceramic Components Division of Allied-Signal Aerospace.)
amples are shown in Fig. 10.39. Note from the photograph that a variety of pore sizes have been achieved from several different ceramic materials. The materials are successfully used for molten metal filtration and kiln furniture and are being evaluated for removing particles from the exraust of diesel engines.
Some components are too complex to be fabricated in one piece by casting. An example is the turbine scroll shown in Fig. 10.40(a). The turbine scroll is an important component in many gas-turbine designs. It changes the direction of the hot gases coming out of the combustor to allow them to pass through the rotor. The scroll in Fig. 10.40(a) is SiC. It was fabricated by assembling the parts shown in Fig. JO.40(b) . The shroud, sleeve , and ring were formed by isostatic pressing and green machining. The body and duct were fabricated by slip casting. The parts were successfully bonded together with a CrVTi braze developed at Oak Ridge National Laboratories (ORNL).
A final casting technique is electrophoretic deposition (EPD). It utilizes an electrostatic charge to consolidate ceramic particles from a suspension.
Shape-Forming Processes 467
Figure 10.38 Prototype gas-turbine rotor fabricated by slip casting using a fugitive- wax·type process, (Courtesy AiResearch Casting Company. Division of the Garrett Corporation. presently the Garrett Ceramic Components Division of Allied·SignaJ Aerospace.)
An electrical polarity is applied to the mold that is opposite to the polarity at the surface of the ceramic particles. The ceramic particles are electrically attracted to the mold surface and deposit as a uniform compact. When the desired thickness of deposit is achieved , either the mold is removed from the container of slip or the slip is poured from the mold. Electrophoretic deposition is generally used to deposit a thin coating or to produce a thin- walled body such as a tube. It is also used to achieve very uniform dep- osition of spray paint onto a conductive surface.
All of the casting techniques discussed above result in a relatively weak ceramic powder compact. A technique recently developed at ORNL results in a much stronger compact. This technique is referred to as gel casting. The ceramic powder is mixed with a liquid and a polymerizable additive to form a fluid slurry similar to a casting slip. The slip is poured into a container of the desired shape. Polymerization is caused to occur before the powder in the slip has time to settle. The resulting powder compact is quite uniform and strong. However. removal of the liquid is more difficult than for conventional slip casting. Furthermore, monomers are generally
468 Chapter 10
Figure 10.39 Ceramic components of controlled porosity fabricated by infiltration of a ceramic casting slip into a polymer sponge structure. (Photo courtesy of Hi- Tech Ceramics. Alfred, New York.)
toxic and require careful handling. For example, the initial material used by ORNL was acrylamid, which is a neurotoxin and has largely been discontinued because of handling concerns.
Casting Process Control
As was illustrated in Fig. 10.22, careful process control is necessary in the slip-casting process. Some of the critical factors include:
constancy of properties viscosity settling rate freedom from air bubbles casting rate drain properties
Warm / <I¥ air
./' ( source
"":=t'~"--------""'I~ Take-up \?~-------------.--------~~t:,1!,",r reel '01 Reel of '- ..... .
Figure 10.41 Schematic illustrating the doctor blade tape-casting process,
Other Tape-Casting Processes
A second tape·casting process is the "waterfall" technique. It is iI1ustrated in Fig. 10.42. The slurry is pumped in a recirculating system to form a continuous curtain. A conveyor belt carries a flat surface through the slurry. The uniform, thin layer of slurry on the carrier is then transferred by
"Curta in" ___ 1 of slip
BBB 11l\\(/l \'X/l \\
Substrate carrier" ,_. ," Conveyor
Figure 10.42 Schematic illustrating the "waterfall" tape-casting process. (From J, Adair , University of Florida.)
486 Chapter 10
Figure 10.52 A variety of ceramic parts that have been fabricated by extrusion. (Photo courtesy Superior Technical Ceramics Corporation, St. Albans, Vermont.)
pose and leave a carbon residue. The acrylic binders are an exception. They burn out cleanly in inert and reducing atmospheres as well as oxidizing atmospheres.
Common Extrusion Defects
Extrusion is often more of an art than a science. Quality is controlled by careful inspection of extruded compacts for defects. Defects that can occur
Shape-Forming Processes 487
for extrusion include warpage or distortion, lamination, tearing, cracking, segregation, porosity, and inclusions [27J.
Warpage or distortion can occur during drying or firing due to density variations or during extrusion due to improper die alignment or die design. If the alignment or balance of the die is not correct, greater pressure on one side of the die will occur. This will cause more material to extrude from this side and result in bending of the extruded column as it exits the die.
Laminations are cracks that generally form a pattern or orientation. Examples are shown in Fig. 10.53. A common cause is incomplete re- knitting as the plastic mix is cut by the auger or flows past the spider portion of the die. The spider is the portion of the die that supports any shaped channels in the die. For example, to extrude a circular tube, a solid rod of the inner diameter of the tube must be supported at the center of the die. It is generally supported by three prongs at 120" to each other that run parallel to the length of the die and are attached to the inside of the die. The material being extruded must squeeze around these prongs and reunite into a continuous hollow cylinder before leaving the die. Lami- nations occur if the material does not completely reknit.
Tearing consists of surface cracks that form as the material exits the extruder. This is illustrated in Fig. 10.54. The cracks extending from the surface inward result from the contact stresses and friction that are dis- cussed earlier in this chapter. Too dry a mix with inadequate cohesiveness will tear. A mix with high rebound may also tend to tear. Die design involving a slight divergent taper at the die exit can help prevent tearing.
Lamination and tearing are two sources of cracking. Other cracks can occur due to poor mixing, shrinkage variation, and partially dried debris from a prior extrusion run.
Segregation involves a separation of the liquid and solid portions of the mix during extrusion. This can result in cracking or distortion during extrusion or during subsequent drying or firing.
Figure 10.53 Drawings of the cross sections of extruded parts illustrating the appearance of severe laminations that can occur as extrusion defects. (From Ref. 27.)
Shape.Forming Protcsses 489
Figure 10.55 Cross sections of extruded honeycomb structures of cordierite for use as catalyst supports for automotive emission-control devices. (Courtesy NGK Insulators.)
soo Chapter 10
the flow pattern with alternate sprue and gate designs . In Fig. 10.59(a), the gate is at the end, but is directed perpendicular to the length of the cavity. In Fig. 1O.59(b), the gate is directed perpendicular, but placed at the center of the mold cavity. Plug flow resulted in both cases and knit· line formation was minimized. This is further illustrated in Fig. 10.60 for actual injection·molding trials. The "short shot" technique was used whereby injection was interrupted before the cavity was full. By conducting a sequence of short shots, a good image of the nature of mold fill for each gate configuration could be obtained.
After binder removal and densification, knit lines remain as large cracks, voids, or laminations and severely limit the strength of the part.
The short shot approach has been successfully used at Carborundum Company in developing integral radial rotors for an experimental auto· motive gas turbine [37J. Initial rotors were injected from the nose end. (Figure 10.61 illustrates the cross section of a radial rotor and identifies terminology that will be referred to subsequently). Short shots indicated a tendency for folds and knit lines to form in the thick region of the hub near the backface. This is illustrated in Fig. 10.62. This region is exposed to the highest stresses during engine operation, so major iterative efforts were conducted to minimize the knit lines. Many parameters such as die temperature, injection pressure, hold time, and sprue bushing/nozzle di- ameter were systemmatically varied. Sixteen resulting rotors were spin- tested and failed at an average speed of 80,500 rpm , significantly below
Figure 10.60 Sequence of "short shots" showing the nature of mold fill for two different sprue and gate orientations, (It\ ASM International.)
S02 Chapter 10
Figure 10.62 Sequence of short shots for injection molding of a SiC rotor from the nose end. Note the knit Jines in the hub and backface regions . (Photos courtesy Carborundum Company for parts fabricated for Allison Gas Turbine Division of General Motors under sponsorship of the U.S. Department of Energy and admin- istration of NASA·Lewis Research Center.)
noncrystalline to crystalline. For example , the volume change for one pol- ypropylene system due to thermal contraction was about 2.75 vol % and due to crystallization was about 1.75 vol % for a total of about 4.5 vol %. If the outer shell is rigid and cannot shrink, while the inner material is more fluid and can reposition during further cooling, 4.5% shrinkage is adequate to form a void or crack through the center of the part. Such a void or crack is typically not visible by examining the surface of the injec- tion-molded part and may not even be visible after densificiation . Figure 10.65 illust rates a large lenticular (lens-shaped) void in a Si, N, turbocharger rotor that resulted primarily from th is mechanism.
Applicatiolls of Illjectioll Moldillg
Injection molding is usually selected for ceramics only after other processes have been rejected. It can produce a high degree of complexity. but the
504 Chapter 10
Figure 10.63 Sequence of short shots for injection molding of a SiC rotor from the shaft end. Note the absence of knit lines in the hub and backface region. (Photos courtesy Carborundum Company for parts fabricated for Allison Gas Tur- bine Division of General Motors under sponsorship of the U .S. Department of Energy and administration of NASA-Lewis Research Center.)
initial cost of tooling is very high . For example, a mold to fabricate an individual turbine blade can cost over $10,000 and a mold for a turbine rotor over $]00,000. Molds for simple shapes and molds made of aluminum for low-pressure injection molding aTC much less expensive . As a result, the use of injection molding of ceramics is increasing.
Injection molding is presently used to manufacture a variety of parts including cores for investment (lost-wax) casting of metals, weld caps, thread guides, threaded fasteners (nut and bolt pairs), radomes, and pro- totype gas-turbine engine components, Drawings of complex investment casting cores for cooled metal gas-turbine blades or stator vanes are shown in Fig, 10.66. During investment casting, the core is mounted in a ceramic mold. Molten superalloy is poured into the mold around the core. The
506 Chapter 10
~ . .. ,' "r'I. c, • " -.. 1,' " .:! , ., , .. I
Figure 10.64 Examples of optimized SiC rotors injection-molded from the shaft end. The rotor on the left is as-molded, the one on the right is after sintering. (Photo courtesy Carborundum Company,)
ceramic mold is removed from the outside of the metal part. The injection- molded ceramic core is leached from the interior of the blade or vane to leave a complex cooling path, This substantially reduces the cost of man- ufacturing of internally cooled stator vanes and rotor blades for advanced gas-turbine engines.
Examples of other injection-molded ceramic parts are illustrated in Figs. 10.67 through 10.69.
Nonthermoplastic Injection Molding
Injection molding of ceramics has traditionally been conducted with ther- moplastic binders or a combination of thermoplastic and thermoset binders. Some success has also been achieved with cellulose derivatives that gel with a suitable change in temperature. Two additional approaches to in- jection molding have been reported within the last rew years. One uses
508 Chapter 10
Figure 10.67 Examples of Ah03 parts fabricated by injection molding. (Courtesy Diamonite Division of W. R. Grace .)
polysaccharides (in particular, agar and agarose) as a gel-forming binder and water as the fiuid . A relatively small percentage (3 wt %) of agarose is required (compared to thermoplastic systems), so drying and binder removal afC reported to be simplified.
The second new approach is identified as Quickset'" injection molding [39J. It is sort of a cross between casting , injection molding, and freeze drying and appears to provide some of the benefits of each, while avoiding some of the problems. It utilizes a slurry (typically with a viscosity under !OOO centipoise) that is injected at typically less than 50 psi pressure into a closed cavity, nonporous mold. The pore fluid is solidified by freezing and subsequently removed by sublimation. Volume change during freezing is negligible and stresses during sublimation are substantially lower than for removal of water or thermoplastic polymers.
Quickset injection molding has been successfully accomplished with both aqueous and nonaqueous suspensions and with a variety of ceramic powders. Table 10.18 lists the properties achieved for different materials formed by the Quickset process. In addition to the excellent properties , tight dimensional tolerances are readily achieved. For example, dimen- sional tolerances for a SiAION component only varied in as-fired parts by 0.09% .
·Trademark of Ceramics Process Systems, Milford, Mass.
Shape-Forming Processes 509
Figure 10.68 Prototype sintered silicon nitride turbocharger rotors fabricated by injection mOlding. (Courtesy Garrett Ceramic Components Division or Allied- Signa l Aerospace .)
Compression molding is analogous to forging. A block of plastic mix is placed between the platens of a shaped die, and uniaxial pressure is applied until the block deforms to the shape of the die cavity. Compression molding can be conducted hot or cold, depending on the nature of the binder system. It works especially well for systems containing thermosetting resins.
A plastic mix is passed between two cylinders that are rotating in opposite directions as shown in Fig. 10.70. The plastic mix passing between the rolls is compacted, as well as being pressed to a thickness equivalent to the spacing of the rolls. Multiple passes at diminishing ro ll separation can yield a constant-thickness sheet of high uniformity.
510 Chapter 10
Figure 10.69 Complex shapes made by injection mOlding. (a) Integral stators. (b) Rotor blade rings. (Courtesy Ford Motor Company, Dearborn, Mich.)
S30 Chapter 11
Figure 11.9 Comparison of the microstructure and translucency of relatively pore· free AliO} (a) with that of opaque AlzO} containing pores trapped in grains (b). Translucent AhO} tubes are used in sodium vapor lamps that provide energy ef· ficient street lights. (Courtesy of General Electric.)
(b) . ' ~ ' .' _ .... . . . . -Jot ., -.r . r'~ .. ~. • • . '. ... .. . •
• .. . . ' •
: . .... , • • ' .. . ..,
• i , • .. . .. ~. . J' , '~,'. ' .
, . . • ••
! • • • ' . • • , • , . '
• • • • • , . . ,' .y.SOUQ1 ~ .'7. "
from the surface of the SiC particles. The B has limited solid solubility in the SiC and allows a mechanism of material transfer between adjacent grains. Pure SiC particles can bond together, but not densify (no shrinkage and no removal of interparticle porosity) . Figures 11.11 and 11.12 illustrate SiC ceramics sintered with additions of Band C.
Liquid·phase sintering involves the presence of a viscous liquid at the sintering temperature and is the primary densification mechanism for most silicate systems. Three factors control the rate of liquid·phase sintering:
1. particle size 2. viscosity 3. surface tension
The viscosity and surface tension are affected strongly by composition and temperatures.
}<'igure 11.11 Variety of SiC parts fabricated by solid·state sintering using beta· SiC powder with additions of Band C. (Photo courtesy H. Yamauchi,lbiden Co" Ltd., Ogaki, Japan.)
", ", ... , -~ '., .. • Figure 11.12 SiC gas-turbine components fabricated using alpha-SiC powder plus Band C sintering aids. (Photo courtesy of Carborundum Company for parts fabricated for Garrett Auxiliary Power Division of Allied-Signal Aerospace under sponsorship of the U.S. Department of Energy and administration of NASA-Lewis Research Center.)
<b) Figure 11.14 Si)N. containing YIOI plus AIIO} sintering aids: sintered in nitrogen at (a) l 6OO"C (2910"F), (b) 175O"C (3 ISO"F). and (c) 185(rC (33lUF). (From unpublished work of O. W. Richerson and I. Aksay.)
" t- o
Figure 11.14 (Contjnu~d)
564 Chapter 11
Figure 11.22 Sj .1N~ gas·turbine rotors fabricated by pressure casting before and after glass encapsulation and hot isostatic pressing. Dense rotors are over 13 em in diameter. (Photo courtesy Garrett Ceramic Components Division of Allied- Signal Aerospace Company.)
of compositions containing less sintering aid and baving dramatically im- proved stress rupture life and oxidation resistance.
HIP has been used to improve the strength and wear resistance of some MnZo and NiZn ferrites, yttrium iron garnet, and BaTi03 , especially for applications such as magnetic recording heads. The ceramic is first sintered to closed porosity and is then further densified by HIP without a requirement for encapsulation . This is sometimes referred to as sinter- HIP. This technique also has been used to achieve improved strength in SiC , in transformation-toughened ZrO" in AI,O,-TiC cutting tools, and in AI,O,-SiC whisker-reinforced composites.
N..pa'tiCU""" m', containing 81
Figure 11.25 (a) and (b) Molten Si infiltration of SiC plus C preforms to produce dense, reaction-sintered SiC. (c) Typical microstructure ofSi-SiC material. Residual Si is light-colored phase. (Photo courtesy Carborundum Company. Niagara Falls, New York.)
Figure 11.31 Examples of single-crystal configurations achieved with the edge- defined film-fed growth technique. (Courtesy Saphikon Division of Tyco Labora- tories, Inc.)
the helium flow. The temperature of the melt and seed crystal are then controlled by adjusting the furnace temperature and helium flow to initiate crystallization of the melt on the seed crystal. Continued control of the temperature gradient of the melt and growing crystal results in crystalli- zation of the complete contents of the crucible into a single crystal.
Figure 11.32 Single·crystal sapphire grown by the heat·exchanger method. (Cour· tesy Crystal Systems. Inc., Salem, Mass.)
400 / (58) /
/ o 1.00
(0 ,039) 2.00
(0.078) 3.00 4.
Crosshead displacement. mm (in .)
Figure 11.34 (a) Load-deflection curve and (b) fracture surface for Hm~·CVD SiC fiber composite. (Courtesy D. W. Richerson, Ceramatec, Inc.)
590 .Chapter 11
Figure 11.35 Components fabricated by the Lanxide process. (Photo courtesy the Lanxide Company, Newark, Del.)
process by additives that promote wetting of the ceramic by the alloy and reduce grain boundary stability of the ceramic. As the ceramic crystals grow away from the melt, some edges remain wetted by the metal and act as a wick. This metal is continuously made available to react with the gas such that the growth rate is independent of the thickness.
The initial material fabricated by the Lanxide process was AI-AI,O,. Additions of Mg and Si to the AI metal provided the wetting action to allow sustained reaction of the metal with oxygen. Growth was achieved in the temperature range of 900 to 1400'C (1650 to 2550' F). The AI,O, content increased with increasing reaction temperature, resulting in a range of controllable microstructures and properties. Composites produced at 900'C (1650'F) had a bend strength of 300 MPa (43 .5 ksi) and a fracture toughness of 7.8 MPa . m'l1 (7.09 ksi . in.' I1). Composites produced at l000'C (1830' F) had a bend strength of 390 MPa (56.5 ksi) and a toughness
Final Machining 601
Some very intricate configurations have been produced by this pho- toetching technique. An example is a 6OO·mesh. sieve.
Electrical Discharge Machining
Electrical discharge machining (EDM) can be performed only with elec- trically conductive materials. A shaped tool is held in close proximity to the part being machined, retaining a constant predetermined gap with the use of a servomechanism that responds to change in the gap voltage. A dielectric liquid is Howed continuously between the tool and workpiece. Sparks produced by electrical discharge across this dielectric erode the
Figure 12.3 Examples of shapes fabricated by ultrasonic machining. (Courtesy Bullen Ultrasonics, Inc., Eaton, Ohio.)
Quality Assurance 627
region will be less exposed (lighter gray in color). Density variations in the material will show up in a similar manner, higher density showing up lighter and lower density showing up darker. Similarly, as shown in Fig. 13.3 (b), a void or hole will show up darker and a high-density inclusion that absorbs x-rays more than the matrix will show up lighter. Obviously, if a positive
10) SIDE VIEW
Ib)TOP VIEW OF FILM AFTER DEVELOPING
/ HIGfHlENSITV INCLUSION
~ I.. PART BEING INSPECTED
...:::::::: FILM SHEET IN LlGHT·TlGHT PACKET
IMAGE OF HOLE
IMAGE OF HIGH- DENSITV INCLUSION
Figure 13.3 (a) Schematic of conventional x-ray radiography setup. (b) Resulting image on the developed film.
" • ... • • :::
Figure 13.4 Microfocus x-ray radiographs of inclusions seeded in hot-pressed SilN •. (a) 500-l.I.m (O.02-in.) graphite inclusions. (b) 250- I.I.m (O.OI-in .) iron inclusions. (e) 500-lLm (O.02-in.) we inclusions. (CQurtesy Garrett Turbine Engine CQmpany', Phoenix, Ariz .• Division of All ied-Signal Aerospace.)
Figure 13.6 Image enhancement of 5O()...-.m (O.02·in.) graphite inclusions in hot-pressed Si~ •. (Courtesy Garrett Turbine Engine Company, Phoenix, Ariz., Division of Allied-Signal Aerospace.)
636 Chapter 13
Figure 13.7 Image enhancement of 250-J.Lm (O.OI-in.) iron inclusions in hot- pressed Si3N •. (Courtesy Garrett Turbine Engine Company. Phoenix, Ariz. t Di- vision of Allied-Signal Aerospace.)
been the primary disadvantage of ultrasonic inspection. However, new techniques are currently being developed for complex shapes. These in- clude the use of microprocessors to accurately control movement of the transducers, use of computers to analyze the data, and use of arrays of transducers instead of scanning with a single transducer.
Figure 13.10 shows the C-scan printout for a O.64-cm (O.25-in.)-thick flat plate of hot-pressed Si3N4 containing various sizes of inclusions and voids [8J. The resolution of both inclusions and voids is quite good. How- ever. such success was not achieved on the first attempt. A variety of transducers and electronic gating procedures wefe tried before optimum conditions were defined. This reemphasizes the importance of standards. The Si,N. plate had originally been prepared as a standard with seeded defects specifically to evaluate and optimize the resolution capabilities of
Quality Assurance 637
different NDI techniques (7]. Without standards, the operator has difficulty optimizing equipment parameters and interpreting printout results .
The ultrasonic C-scan inspection detected more of the defects in the hot-pressed Si,N. plate than could be detected by other techniques. X-ray radiography had shown up only the high-density WC and Fe inclusions and the largest graphite inclusion. Neutron radiography showed only the BN.
Ultrasonic ND! appears to have excellent potential for the inspection of ceramic materials. However, it must be emphasized that development is only beginning and that success to date has been only on flat plates having machined parallel sides. Complex shapes and parts with rough as- processed surfaces represent a much more difficult problem.
Another limitation of ultrasonics is the loss in intensity in the scattering of the waves as they pass through the material. This is called attenuation and limits the thickness of the part that can be inspected. Attenuation is accentuated by porosity or other microstructural features that cause scat- tering (second-phase distributions, microcracking, etc.). Attenuation is
Figure 13.8 Image cnhancemenl of 500-,...m (O.02-in.) we inclusions in hOI·pressed Si)N •. (Courtesy Garrett Turbine Engine Company, Phoenix, Ariz., Division of Allied·Signal Aerospace.)
'i: u z ... N ;;;
640 Chapt.r 13
we F. BN s;c e r. -.~ ..
0,020 I • , '¢4 '" • •
, , 0.0'0
.' . ~ ..•. ~ , .. • •
. --. , •
, , 'f
,' .. " .. ~, . . ,
. .. d' •
iai-~ . .~ ..
..... . -.
• ~.~~ . , " ,
•• • • .' ... • • • .' .
.: :,.;" " , " : •
, . .. ' .' ., .' ". (,," " . , .. ' .
~11 ., ,v
0.005 ~ :.' '.' .. ....
;'. ' .. : . . ,. '. ",.: . CO,",'. ~: ~ .. , .' ,t.}'
yo.' . ,
• ~ ' ., .. ·u .. " ,
, , ' ',« ~
" • •
Figure 13.10 Ultrasonic C-scan with a 25-MHz transducer of a O.64-cm (O.2S-in.)- thick hot-pressed SilN~ plate, (Courtesy Garrett Turbine Engine Company, Phoe- nix., Ariz., Division of Allied-Signal Aerospace.)
from this array of transducers can then be analyzed by computer to locate internal defects.
Another ultrasonic method has potential for detecting surface or neac- surface flaws. The transducer is placed at a low angle to the surface. The acoustic waves travel along the surface rather than penetrating the interior of the part and interact with surface or near-surface discontinuities. Since the strength of ceramic materials is so sensitive to surface flaws, this may be a useful method to consider.
682 Chapter 16
Figure 16.2 Thermal shock fracture showing lack of branching.
will occur. A baseball striking a window will cause much more branching than a BB, due to the larger applied energy. Tempered glass will break into many fragments due to release of the high stored energy. On the other hand , a thermal-shock fracture may not branch at all , especially if it initiates from a localized heat source and propagates into a relatively unstressed or compressively stressed region of the component. In this case, the fracture will tend to follow a temperature or stress contour and will have a char~ acteristic wavy or curved appearance, as shown in Fig. 16.2 for a thermally fractured ceramic setter plate for a furnace.
Location of the Fracture Origin
The pattern of branching will often lead the engineer to the vicinity of the fracture origin. The engineer will then have to examine the fracture surfaces in this region, often under a low-power optical binocular microscope, to locate the precise point at which fracture initiated. This point of origin can be a flaw such as a pore or inclusion in the material, a cone-shaped Hertzian surface crack resulting from impact, a crack in a surface glaze, an oxidation pit, intergranular corrosion, a position of localized high stress, or a com- bination of these. Location and examination of the fracture origin will help determine which of these factors is dominant and provide specific guidance in solving the fracture problem.
As mentioned before, a fracture begins at zero velocity at the fracture origin and then accelerates as it travels through the part. As it does, it
• • -. ,
Figure 16.4 Examples of typical fractu re mirrors for high-strength polycrystalline ceramics. (a) Initiation at a surface flaw in hot-pressed silicon ni tride. (b) Initiation at an internal flaw in reaction·sintered silicon nitride.
Failure Analysis 687
Figure 16.5 Examples of fracture surfaces with indistinct fracture features. (a) Sintered silicon carbide. (b) Silicon carbide·carbon·silicon composite.
invariably has changed slightly, leaving a discontinuity. This line of dis- continuity looks a little like a Wallner line, but is usually more out of plane and more distinct. It is also called a rib mark. Arrest lines or rib marks provide essentially the same information as Wallner lines, i.e., the direction of crack movement and the stress distribution. Twist hackle frequently is present after an arrest line.
, ,. ' " . , • • •
' . , , • ..
" • .' •
~ . .)0. ..... ...
Figure 16.5 (Continued) (c) Porous lithium aluminum silicate. (d) Bimodal grain distribution reaction-sintered silicon carbide.
Failure Analysis 691
Figure 16.8 Examples showing cantilever curl in four·point bend specimens. Spec· imens 0.32-cm (0.12S-in.) thick.
Techniques or Fractography
The techniques of fractography are relatively simple and the amount of sophisticated equipment minimal. Often, the information required to ex· plain the cause of fracture of a component can be obtained with only a microscope and a light source. In fact , sometimes an experienced individual can explain the fracture just by examining the fracture surfaces visually. At other times, a variety of steps and techniques including sophisticated approaches such as SEM , electron microprobe, and Auger analysis are required [8-1OJ. When extensive fractography is necessary, the steps and procedures are as shown schematically in Fig. 16.9.
Step 1 involves visual examination of the fractured pieces and review of data regarding the test or service conditions under which the hardware
698 Chapter 16
Figure 16.11 SEM photomicrographs of fracture.initiating material flaws in re· action-bonded Si~N~ (RBSN) that are typical and consistent with the normal mi- crostructure and strength. Arrows point to the fracture origins.
\ . '\. , t . .. 1,-\ ..... r .,:t. I d l '.~. ..,.. ';y'. "~':"/ "l~" " .... ,;" . . " .•. \:\\.~ :','; I-rL''-~8.' .. '!":vi • " : ~'~" 7" ".\ii~\.!.: .. · I':";~"'~' ~. ':.'''';:: '. • I ..... .. .It '!I!.'~ f~' .... '~.. _ " -'" .. . ", ~ ,(..: ' " -.- . "", • '" ;. '.l..~ ~. J" \! ....
r.". ,'. ,;'., \~ '. '4'1~\" I"fl' ... . ... " " ·",t·" "'''~r'''' "\. ' .> , ~ • ~ ~ '1". '1.:' \ <'. ' , .... ,. " . .' • ~:.t •. ' ,,,,~.'_t"'~ ";'t' ~ "),," .. l" .. , . .. .••• !,,' I ,a,. . ....! '. , , t ~ \' ... • l .... ~ ... ~
,.. "t i \ \ ~ \ ~ ....... •. ': ' ',- ,~ ',: )\1[·····, <.,;; ~·i·"~,;:· .,'" .,. i. '. '&W '.' .. -- ,. . ...... ,. t " , .......
~ ., .e t." , ... 9 -., .. ' ••
1/ .', ",' !~ .. \. ';" . \. .... ~!,a:.. ' ~ , -I f . ....:"~' .-., -~ .• '. ii. "". - »- - - "),.'~ - . . 'l'~ , ..... ·1 . " " "I
I • , ' .\... ,. .. ~" ~. *_, ' , . ~ .. ' .. \~" "~."' ....... : .. ~.~. .'~ !JO/-i "l ~--t
700 Chapter 16
" , , < 'f ...
" ~-.: ,-' 4' .~ •. " .. . ...... . 1'.0; . ,-. -.
Failure Analysis 701
Figure 16.12 SEM photomicrographs of abnormal fracture-initiating material flaws in RBSN traceable to improper processing prior to nit riding. (a) Large pore in slip-cast RBSN resulting from inadequate de-airing. (b) Crack in greenware prior to nitriding. (c) and (d) Low-density regions in slip-cast RBSN resulting from agglomerates in the slip.
figure 16.13 SEM photomicrographs of abnormal fraclUre-in - iliating material flaws in RBSN traceable 10 the nilriding process. (a) and (b) Porous aggregate rich in Cr and Fe, resulting from reaction of the silicon during the ni lriding cycle wi th stainless steel contamination picked up during powder processing; energy- dispersive x-ray analysis shown. (e) Large aggregate of unreacted silicon resulting from localized melt ing due to local exothermic overheating. Arrows identify the fracture origins.
,. • • '3" "
704 Chapter 16
Failure Analysis 705
Figure 16.14 SEM photomicrographs comparing normal and abnormal material flaws in sintered SiC. (a) and (b) Typical microstructure of high-strength material. (c) Large pore resulting from powder agglomeration during powder preparation and shape forming. (d) Large grains resulting from improper control of temperature during sintering.
706 Chapter 16
Failure Analysis 707
Figure 16.15 SEM photomicrographs showing fractures initiating at transverse machining damage. (a) The fracture surface of a tensile specimen of hot·pressed silicon nitride that had been machined circumferentially. (b) The intersection of this fracture surface with the machined surface. illustrating that the fracture origin is parallel to the grinding grooves. (c) and (d) The same situation for reaction· sintered silicon nitride.
Fig. 16.16 SEM photomicrograph showing a typical featureless thermal-shock fracture surface. (top) Overall surface at low magnification. (bottom) Fracture origin at higher magnification. (Courtesy Garrett Turbine Engine Company, Phoe- nix. Ariz., Division of Allied-Signal Aerospace.)
Fig. 16.17 SEM photomicrograph of a thermal-shock fracture initiating at a ma- terial flaw. (a) Overall surface at low magnification. (b) Preexisting crack at fracture origin. (Courtesy Garrett Turbine Engine Company, Phoenix, Ariz., Division of Allied-Signal Aerospace.)
Failure Analysis 715
.......... - FRACTURE SURFACE
.• - SPECIMEN SURFACE
Figure 16.18 Typical Hertzian cone crack resu lting from impact and acting as the flaw that resulted in fracture under subsequent bend load. Shown at increasing magnification from (a) to (c). (Courtesy Garrett Turbine Engine Company, Phoe· nix, Ariz., Division of Allied Signal Aerospace.)
716 Chapter 16
Figure 16.18 (Continued) (Courtesy Garrett Turbine Engine Company, Phoenix, Ariz., Division of Allied Signal Aerospace.)
case, the objective is to identify the mechanism of attack and find a solution. In other cases, especially where the oxidation or corrosion is isolated along grain boundaries, the presence and source of degradation may be more difficult to detect. In this case, the degree of attack may only be determined by strength testing, and the cause may be ascertained by controlled envi- ronment exposures and lor sophisticated instruments such as Auger spec- troscopy, which can detect slight chemical variations on a microstructural level.
Let us first examine some examples of oxidation and corrosion in which visible surface changes have occurred. Figure 16.22 shows the surface and fracture surface of NC-132 hot-pressed Si,N.' after exposure in a SiC resistance-heated, oxide-refractory-lined furnace for 24 hr at llOO°C (2012°F) [25J. Figure 16.22(a) shows the complete cross section of the test bar. The fracture origin is at the surface on the left side of the photo and is easily located by the hackle marks and the fracture mirror (the dark
·Manufactured by the Norton Company. Worcester. Mass.
Failure Analysis 717
Figure 16.19 Impact fracture of a ceramic rotor blade showing Hertzian cone crack. (Courtesy Garrett Turbine Engine Company, Phoenix, Ariz., Division of Allied-Signal Aerospace.)
Failure Analysis 719
Figure 16.20 (a) and (b) Surface cracks resulting from relative movement between two contact surfaces under a high normal load and with a high coefficient of friction.
720 Chapter 16
Figure 16.20 (Continued) (c) Typical multiple chipping resulting from contact loading and visible on a fracture surface.
furnace lining had contacted the specimen during exposure. The EDX analysis included in Fig. 16.23 was taken in the glassy region at the base of the pit, showing that AI, Si, K, Ca, and Fe were the primary elements present and again indicating a propensity for Si,N. to be corroded by alkali silicate compositions. However, it should be noted that the size of the pit is much smaller than in the prior example and resulted in only a small strength decrease.
Figures 16.24 and 16.25 show examples of more dramatic corrosion of hot-pressed and reaction-bonded Si,N. , resulting from exposure to the exhaust gases of a combustor burning jet fuel and containing a 5-ppm addition of sea salt. Exposure consisted of 25 cycles of 899'C (1650'F) for 1.5 hr, 1121'F (2050'F) for 0.5 hr, and a 5-min air quench. At 899'C (1650'F), Na,SO. is present in liquid form and deposits along with other impurities on the ceramic surface. The EDX analyses taken in the glassy surface layer near its intersection with the SiJN .. document the presence of impurities such as Na. Mg, and K from the sea salt, S from the fuel, and Fe, Co, and Ni from the nozzle and combustor liner of the test rig. An EDX analysis for the Si,N. on the fracture surface about 20 j>m beneath the surface layer is also shown in Fig. 16.25. Only Si is detected (nitrogen
WITNESS MARK -~ CONTAINING
Co, Fe, Ni, Cr
__ FRACTURE SURFACE
Figure 16.21 (a) Witness mark on the surface of the ceramic adjacent to the fracture origin , suggesting fracture due to contact loading. (b) Multiple cone fea- tures resulting from a contact fracture.
722 Chapter 16
Figure 16.21 (Continued) (c) Multiple cone features resulting from a contact fracture.
and oxygen are outside the range of detection by EDX), indicating that the corrosion in this case resulted from the impurities in the gas stream plus the surface oxidation.
The strength of the hot-pressed Si,N, exposed to the dynamic oxidation with sea salt additions decreased to an average of 490 MPa (71,000 psi) from a baseline of 669 MPa (97,000 psi). The reaction-bonded material decreased to 117 MPa (17,000 psi) from a baseline of 248 MPa (36,000 psi). Repeating the cycle with fresh specimens and no sea salt resulted in an increase to 690 PMa (100,000 psi) for the hot-pressed Si,N, and only a decrease to 207 MPa (30,000 psi) for the reaction-bonded Si,N •.
The examples presented so far for oxidation and corrosion have had distinct features that help distinguish the cause of fracture from other mechanisms, such as impact or machining damage. Some corrosion-initi- ated fractures are more subtle. The corrosion or oxidation may only follow the grain boundaries and be so thin that it is not visible on the fracture surface. Its effects may not even show up in room-temperature strength testing since its degradation mechanism may only be active at high tem- perature. How do we recognize this type of corrosion? The following sug-
Figure 16.22 SEM photomicrographs of the fracture surface of hot·pressed ShNJ exposed to static oxidation for 24 hr at 1 100°C (20l2°F). (a) Overall fracture surface showing hackle marks and fracture mirror (the irregular dark spots on the fracture surface are artifacts). (b) Higher magnification showing the fracture mirror with an oxidation corrosion pit at the origin. (c) Higher magnification showing the nature of the pit and the surface oxidation layer. Specimen size 0.64 x 0.32 cm. (From Ref. 14.)
Figure 16.23 SEM photomicrograph of the fracture-initiating oxidalion-corrosion pit on the surface of rc- action-bonded Si~ •. The EDX graph shows the relative concentralion of chemical elements in the glassy region at the base of the pi!. (Courtesy Garrett Turbine Engine Company, Phoenix, Ariz., Division of AUied- Signal Aerospace.)
, Co Co N,
fOX Of SURFACE LAVER
Figure 16.24 SEM photomicrograph of hot-pressed SiJl"l. that was exposed to combustion gases with ~a sail additions. showing that fracture initiated at the base of the glassy surface buildup. EDX analysis shows the chemical elements detected in the glassy material adjacent to the Si,N •. (Courtesy Garrett Turbine Engine Company. Phoenix, Ariz., n;,,;~;n .. "f Al1 i f"rl_~;l>n:ll Aerosoace.)
Co EOX OF SURfACE LAYER
EOX OF BASE R8SN
Figure 16.25 SEM photomicrograph of reaction-bonded SiJN. that was exposed to combustion gases with sea salt additions, showing that fracture initiated at the base of the glassy surface buildup_ EDX analysis shows the chemical elements dete<:ted in the glassy material adjacent to the Si~ •. (Courtesy Garret! Turbine Engine Company, Phoenix, Ariz .• Division of Allied-Signal Aerospace.)
728 Chapter 16
• •• / . •
Figure 16.26 SEM photomicrograph of the fracture surface of a low-purity Si.'N~ material sintered with MgO and showing slow crack growth. Region of slow crack growth identified by arrows.
equation, may not be good approximations for the material under slow- crack-growth conditions.
There are other limitations to the information available from the frac- ture surface. The size of the slow-crack-growth region provides no infor- mation about the time to failure, the rate of loading, or the mode of loading (cyclic versus static).
Fractography is a powerful tool to the engineer in helping to determine the cause of a component or system failure. Well-defined features usually present on the fracture surface of a ceramic provide the engineer with useful information regarding the place where fracture initiated, the cause of fracture, the tensile stress at the point of failure , and the nature of the surrounding stress distribution. This information helps the engineer to determine if the failure was design- or material-initiated and provides di- rection in finding a solution. It can also help in achieving process or product improvement. Finally, it can help determine legal liability for personal or property damage.
Toughening or Ceramics 737
( 1 - -----1
I S'O ,<-<. >n
Figure 17.3 Optical pho tomicrograph or polished cross section showing unidirec- tional SiC filamenrs in a metal matrix. (Courtesy Textron Specialty Materials, Lowell, Mass.)
reasonable level of bond or friction between the fibers and matrix . Too- weak a bond can result in shear at the fiber-matrix interface and reduce the amount of modulus transfer.
Ceramics generally fracture in tension, i.e., in a crack-opening stress mode. Prestressing involves placing a portion of the ceramic under a residual compressive stress. A crack cannot start or extend as long as the ceramic is prestressed in compression . Tensile fracture will only occur after a large enough load is applied to exceed the compressive prestress and to build up a tensile stress large enough to initiate a crack at a critical flaw.
A compressive prestress can be achieved by many approaches [4J . One approach is to place the surface in compression by quenching, ion exchange,
Toughening of Ceramics 745
Figure 17.7 Transmission electron micrograph of optimally aged, transformation~ toughened Zr02-MgO showing the oblate spheroid precipitates of tetragonal Zr02 in a MgO-stabilized Zr02 cubic matrix. (Courtesy A. H. Heuer , Case Western Reserve University.)
However, such a transformation involves an increase in volume, as shown in Fig. 4.17. If the grain or precipitate size is small enough (less than about 0.5 j..lm) , the strength of adjacent grains prevents the transformation from occurring by preventing the necessary volume expansion. When a stress is applied to the zirconia and a crack tries to propagate, the metastable tetragonal zirconia grains adjacent to the crack tip can now expand and transform to the stable monoclinic crystal form. This is illustrated in Fig. 17.8. Precipitates that have transformed to monoclinic can be distinguished from untransformed precipitates in the TEM photomicrograph by the pres- ence of twinning. Note that only the precipitates near the crack have transformed . This martensitic transformation is accompanied by a 3% vol- ume increase of these grains or precipitates adjacent to the crack, which places the crack in compression and stops it from propagating. To extend the crack further requires additional tensile stress. The result is a ceramic that is very tough and strong and that has been appropriately referred to as "ceramic steel."
Pure ZrOl does not have transformation-toughening behavior. Addi- tives are required to stabilize such behavior. These additives are CaO, MgO, Y,O" CeO" and rare earth oxides. Too much addition fully stabilizes the Zr02 in a cubic crystal structure, which also does not have transfor- mation-toughening behavior because it does not go through the tetragonal- to-monoclinic transformation. Toughening requires the presence of the
746 Chapter 17
Figure 17.8 Dark field transmission electron micrograph of optimally aged, trans- formation-toughened ZrOl-MgO showing twinned monoclinic precipitates adjacent to a crack and tetragonal precipitates away from the crack. (Courtesy A. H. Heuer, Case Western Reserve University.)
metastable tetragonal state. The range of addition to achieve the metastable tetragonal state and toughening is shown for various ZeOz-based compo- sitions in Fig. 17.9. The peak of the curve for each material corresponds to the maximum tetragonal content. Monoclinic content increases to the left of the peak and cubic to the right of the peak, each resulting in a decrease in toughness and strength. PSZ in Fig. 17.9 stands for partially stabilized zirconia. DCB, ICL, and NB identify the method that was used to measure fracture toughness. DCB stands for the double-cantelever beam, ICL the indentation crack length, and NB the notched beam. Note that the composition zones for achieving peak toughness are relatively narrow.
Transformation toughening is not limited to Zr02_ Very small grains of Zr02 can be added to another ceramic such as Al20, and be retained as tetragonal during cooling. These grains will then transform near a crack tip and inhibit crack propagation. Several criteria are necessary before transformation toughening can be achieved by addition of partially stabi- lized Zr02 particles to a host ceramic: (1) Zr02 particles not dissolved by host; (2) particle size of the Zr02 typically under 0.5 iJ.m; and (3) host
Intercrystatline Intercrystalline Intercrystalline Intracrystalline
's' .:~"90",1 \.V iTt- •.•
Dispersed . , Zirconia - Ceramics Ii II
Mixed Crystalline in
Complex Zirconia Systems 11"-'11 .:. ::. ::. :;. ::. .. : : .. : :.:: :.:: ::.
o/fi ~~~ t~ ~f¢ ~
~~~$. $f ~~ < .. *~ &- $--::-
Figure 17.10 Rcf. 19.)
Classification or transfonnation·toughened ceramics based on microstructural featcs. (From
750 Chapter 17
Figure 17.11 Ferroelastic domains in Zr01 stabilized to the tetragonal phase with 3 mol. % Y10). (a) Transmission optical photomicrograph. (b)Transmission electron microscope photomicrograph of domains in one grain of a polycrystalline sample. (e) Transmission optical photomicrograph of domains in a single-crystal sample ( x 400). (Courtesy of A . V. Virkar and Jan Fang Jue , University of Utah.)
of the microstructure of a self-reinforced Si,N. material with high-aspect- ratio beta-Si)N. grains. This material had average fracture toughness of 10.5 MPa'm ,n, as measured by the chevron notch technique .
The toughening mechanisms for self-reinforced Si,N. have been iden- tified primarily as crack deftection and bridging. Highest toughnesses have been achieved for large grain size and for compositions containing sub-
Toughening of Ceramics 753
Figure 17.12 Photomicrograph of pressureless sintered Si3N4 material with com- position and sintering parameters selected to achieve a fibrous self-reinforced mi- crostructure. This material had a composition in the Si3N.-Yl03-MgO-CaO system, contained about 15% glass, and had an average fracture toughness of 10.5 MPa·m lll. (Courtesy A. Pyzik, Dow Chemical Company.)
stantial grain boundary phase. Mechanical properties above 1200°C have been limited by the properties of the grain boundary phase. Major effort is in progress to achieve high-toughness and high-temperature mechanical properties simultaneously.
High-toughness platelet-reinforced ZrCIZr matrix composites have been achieved by a reactive densification process [31,32]. Boron carbide powder is compacted with a binder using conventional techniques such as pressing , molding, or casting. The preform is placed in a graphite mold with a controlled amount of Zr metal. The material is heated to 1850 to 2000°C in an inert atmosphere. The Zr becomes molten, infiltrates the preform, and reacts with the B,C to form ZrB, plus Zrc' The quantity of Zr can be varied either to be completely consumed by the reaction or to be retained
Figure 17.13 Crack deflection by dispersed particles of ZrBl in a matrix of zrC and Zr. (From Ref. 32.)
756 Chapter 17
Figure 17.14 Aluminate platelets in situ reinforcement in a transformation-tough- ened ZrO,. (Courtesy R. A. Cutler, Ceramatec, Inc.)
tetragenal ZrO, phase metastably to room temperature. This is done dif- ferently, depending on the additives in the ZrO,.
Early studies were conducted in the CaO-ZrO, system. The phase diagram for the CaO-ZrO, system is shown in Fig. 17.15. Toughening in this system has been achieved in the composition range of roughly 6 to 11 mol % CaO. Note from Fig. 17.15 that this corresponds to the position of the Tu + C. (tetragonal solid solution plus cubic solid solution) phase field. To achieve high toughness in this system, the following procedure is used: (I) a com- position such as 10 mol % CaO, 90 mol % ZrO, is selected; (2) the powder is compacted into the desired shape; (3) the compact is densified at a temperature just above the T $!I + Css field in the Css field to achieve a polycrystalline microstructure of uniform cubic solid solution, i.e., about 1800 to 1850°C (3270 to 3330°F) for the 10% CaO composition; (4) the solutioned material is quenched to about 1300°C (23700F) (at this temper- ature, the material is supersaturated; tetragonal precipitates form in the cubic ZrO, grains) ; (5) the material is aged at about 1300°C (2370°F) until the precipitates have reached optimum size; and (6) the material, now
Toughening of Ceramics 765
Figure 17.20 Reflected light optical photomicrograph of a pressureless sintered post-HIPed AhOr30 wt% TiC particulate composite. (Courtesy R. Cutler, Cer- amatec. Inc .. Salt Lake City. Utah.)
of particulate-reinfon;:ed ceramics. Figure 17.20 shows the microstructure of a AhO, - 30 wt % TiC material successfully densified by a combination of pressureless sintering and unencapsulated HIP [421 .
Whiskers are usually single crystals that have grown preferentially along a specific crystal axis. Whiskers typically range in size from about 0.5 to 10 ~m in diameter and a few microns to a few centimeters in length. A note of caution is in order. The smaller whiskers can become lodged in the lungs and represent a health hazard. Thus, they should be handled with proper precautions such as a hood , respirator, and careful cleaning of work areas.
Whiskers generally grow under vapor-solid or vapor-liquid-solid con- ditions that result in a small defect size in the whiskers. For example, defect size in some SiC whiskers was estimated to range from 0.1 to 0.4 11m .
768 Chapter 17
Figure 17.21 Examples of SiC whiskers showing variations in size and smoothness, The Areo (presently Advanced Composite Materials Corp.) and Tateho whiskers in (a) and (b) were synthesized by vapor-solid techniques. The Los Alamos National Labs (LANL) whiskers in (e) were synthesized by a vapor-liquid-solid process that used molten metal catalyst balls to transfer reactants from the vapor phase to the growing whisker. Some of these metal spheres can be seen still attached to the end of some of the whiskers.
the normal carbon content one would expect for an organic material. The University of Utah team hypothesized that they could heat the rice hulls to around 150QoC in a nonoxidizing atmosphere and produce SiC powder. When they conducted their experiments to evaluate their hypothesis, they got SiC, but much of it was in the form of whiskers rather than particles. This technology was licensed and refined and ultimately ended up with Arco Chemical (presently Advanced Composite Materials). Individuals at Arco Chemical collaborated with researchers at Oak Ridge National Labs (ORNL) to evaluate the whiskers in AI,O, and other matrix materials.
The Arco Chemical and ORNL efforts demonstrated that substantial improvements could be achieved in AI20 3 with SiC whisker reinforcement. As shown in Table 17.9, the room·temperature toughness of AI20 j was
772 Chapter 17
Different whiskers contain different levels of inclusions, voids, or other defects and thus have different strength.
The differences in whisker characteristics have a pronounced effect on processing and properties of the composites. Surface chemistry affects the degree of dispersion of the whiskers with the Al,O,. Reticulation also affects dispersion by forming mechanically linked clusters (agglomerates) ofwhis- kers. Surface chemistry affects the interface bonding between the whiskers and matrix. This plus the presence of growth steps influences the degree of whisker pullout, bridging, and stress delocalization. Defects in whiskers limit the strength of the whiskers. As a result, present SiC whisker-rein- forced materials are not optimum. Toughness values greater than 15 MPa·m' l2 are possible with further development (45).
Figure 17.23 shows the microstructure of a typical Al,O, matrix /SiC whisker composite. Figure 17.24 illustrates some examples of components
Figure 17.23 Photo taken through a reflected-light optical microscope of a pol- ished section of a hot-pressed AI20 .,-SiC whisker composite cutting too) insert. The cutting tool sample was from the Greenleaf Corp. The sample was polished and photographed at Ceramatec.
Toughening of Ceramics 773
Figure 17.24 Examples of components fabricated of AI10 .1-SiC whisker composite compositions. (Courtesy Advanced Composite Materials Corporation, Greer, S.C.)
fabricated from whisker-AhO) matrix composites. A major success story has been the use of hot-pressed whisker-Aha, material for cutting tool inserts for high-speed turning and milling of some metals.
Si,N, Reinforced with SiC or Si,N, Whiskers
Table 17.9 includes some strength and toughness values for hot-pressed SiJN4 containing SiC or ShN .. whiskers. In general, the room-temperature values are no better than what has been achieved with self-reinforced Si3N4 prepared by overpressure sintering. High-temperature strength and creep resistance may benefit by whisker reinforcement, but further study is nec- essary. Buljan et al. [50J reported 970 MPa/6.4 MPa'm l12 strength/tough- ness for 30 vol % SiC whiskers at room temperature, 82017 .5 at IDqO°C, and 59017.7 at 1200°C.
782 Chapter 17
Fig. 17.28 Fracture surface of continuous-liber-reinforced ceramic matrix com- posite showing fiber pullout. (a) 15 x overview and (b) 500 x closeup. (Courtesy J. Brennan, United Technologies Research Center.)
SiC Matrix Composites
SiC matrix composites have been prepared by chemical vapor infiltration (CVI) and by pyrolysis of infiltrated polymers. Early work in CVI was conducted by Societe Europeenne de Propulsion (SEP) in France using an isothermal process analogous to that used for fabrication of carbon-carbon composites . In the isothermal process, fiber preforms are supported in a reaction chamber and heated to a desired temperature (around IOOO' C). Reactant gases are passed through the chamber. These infiltrate into the fiber preform and react to form a pure SiC layer on each fiber. The process is continued until the pores narrow enough such that further infiltration either ceases or is very slow. Weeks of deposition time are required for a preform containing about 40 vol % fiber to be densified to about 70 to 80% theoretical density. The resulting material fractures at room temperature in a noncatastrophic composite mode at a typical flexural
Toughening of Ceramics 783
stress of 300 to 400 MPa and toughness of 25 MPa·m'll . A fractured test sample is shown in Fig. 17.29 .
A "forced CYI" process has been developed at Oak Ridge National Laboratories (ORNL) that reduces the deposition time typically to less than 24 hr . The fiber preform is compressed fnto a graphite holder as illustrated in Fig. 17.30. The holder and preform are heated such that a gradient exists from top to bottom. The fibers near the top are heated hot enough to allow deposition of SiC from reactant gases introduced through the bottom of the holder. The hot zone moves from the top of the holder toward the bottom of the holder as SiC deposition continues. This results in a high rate of deposition without blocking of near-surface pore channels such as occurs with isothermal CVI. Because of the reduced deposition time, reaction temperature can be increased to around 1200°C to further decrease deposition time.
Composites 85 to over 90% theoretical density have been achieved by the forced CYI process starting with 40% dense cloth preforms. Room- temperature load-displacement curves are shown in Fig. 17.31 for Nicalon and Tyranno fiber-reinforced composites. The strength measured in four-
Figure 17.29 Ref. 69.)
Test bar of SEP SiC/ SiC composite fractured in bending. (From
Figure 17.30 Schematic illustrating the forced-flow thermal-gradient CVI process. (From Ref. 70.)
Figure 17.32 Fracture surface (as viewed by SEM) of early SEP SiCISiCcomposite tested at room temperature foUowing 24-hr static oxidation exposure at lOOO°C. Note the small dCEree of fiber Dullout. (From Ref. 69,)
788 Chapter 17
Figure 17.33 Fracture surface (as viewed by SEM) of early SEP SiC/SiC composite stress rupture tested at 1200°C and 138 MPa. Composite failed in a brittle fashion in less than 24 hr, Note that no fiber pullout occurred. (From Ref. 69.)
Subsequent studies at NASA and Norton demonstrated that careful layup of unidirectional monofilaments resulted in substantial improvements in strength and toughness compared to unreinforced reaction-bonded Si,N, [73,74]. This is illustrated in Figs. 17.35 and 17.36. The Norton NC-350 unreinforced RBSN fractured in a brittle fashion at about 350 MPa. The Norton composite fractured in a non brittle mode and had an ultimate tensile strength <as measured in bending) of about 650 MPa. The NASA SiC fiber-reinforced RBSN exhibited similar behavior. Note that the first cracks formed in the matrix at around 250 MPa. The stress at which first matrix cracking occurs is important. Ceramic matrix composite stressed above this level appear to be susceptible to both cyclic fatigue and oxi- dation/corrosion degradation.
A hot-pressed Si,N,-based ceramic matrix composite has been devel- oped recently at Ceramatec that has high first matrix cracking stress . A stress-displacement curve is shown in Fig. 17.37. First matrix cracking occurred at about 450 MPa, followed by substantial fiber debonding, pul- lout , and bridging. The ultimate bend strength was about 908 MPa. The composite consisted of a matrix of 95 vol % Si,N, and 5 vol % mullite with 35 vol % unidirectional Textron SCS-6 SiC monofilament fiber. The mullite provided an improved thermal expansion match between the matrix and fiber that helped increase first matrix cracking stress. The mullite also helped decrease the hot-pressing temperature and time (1500°C, 30 min)
Toughening of Ceramics 789
to minimize interfacial bond between the fibers and matrix. The composite has exhibited excellent high-temperature properties. It has retained non- brittle composite fracture behavior to 14000C. It has survived stress rupture testing for over 1000 hr at llJOOoC in air at an applied flexural stress of 400 MPa .
Other Fiber-Reinforced Ceramic Matrix Composites
Many additional studies have been conducted in recent years with oxide and nonoxide fibers in oxide and nonoxide matrices. This section identifies some of the combinations and processes that have been tried and provides
790 Chapter 17
Figure 17.34 SEM photomicrographs of (a) as-received Nicalon fibers, (b) de~ graded Nicalon fibers in reaction-bonded ShN. matrix , and (e) visually nondegraded Textron (AVCO) SCS-6 CVD monofilament fiber in reaction-bonded Si3N~. (From Ref. 69.)
a few references to give the reader a starting point in locating further information,
Sol-Gel Infiltration. Table 17 .12 lists a variety of sol-gel processes that have been used to form oxide matrices [77J. A fiber preform is infiltrated with the sol that generally consists of alkoxide compounds with catalysts, dissolved saits, or other additives. A gel is formed by a hydrolysis reaction, followed by dehydration , polymerization, and calcining or pyrolysis. Mul- tiple cycles of infiltrations are required to achieve over 80% densification. The sol-gel processing is conducted at temperatures below 900°C. This results in minimal fiber degradation and fibec¥matrix interfacial bond. Sometimes a final hot-pressing step is used to further densify the composite. This is typically conducted at high temperature (1250 to 1700°C). The hot- pressing step reduces shape capability and also has the potential to degrade the fibers or interface . In spite of these concerns, 10 mol % GeO,-modified
Toughening of Ceramics
V.por Pha •• Oxld.nt
R.actlon Product Containing Flb.r.
Figure 17.38 Fabrication of a fiber-reinforced ceramic matrix composite using the Lanxide directed metal oxidation process. (From Ref. 80.)
Oxide Fibers in CVD SiC Matrix. 3M Company Nextel oxide fibers, as well as Nicalon SiC fibers, have been formed by braiding into tubes and other configurations and bonded into a composite with CVD SiC. Thin- walled tubes have been fabricated that have exceptional thermal-shock resistance. Scoville et al. [82J describe tests in which a composite tube is heated internally with a high-temperature torch and survives spraying of a stream of water locally onto the hot surface.
Diboride Matrix Composites. Composites have successfully been fabri- cated using unidirectional CVD SiC monofilament fibers in matrices con- sisting of diboride-SiC-C compositions [83,84J. A Hffi,-based composite had flexural strength of 1100 MPa, compared to 780 MPa for a ZrB,-based composite and 329 MPa for a TiB,-based composite. All fractured in a noncatastrophic composite mode. A load-displacement curve and an SEM of the fracture surface of one of the Hffi,-based samples were shown earlier in Fig. 11.34.
NZP Matrix Composites. NZP was discussed in an earlier chapter as a family of low thermal expansion materials with the [NZPJ (sodium zircon- ium phosphate) structure. Some of the most interesting compositions in- volve crystal chemical substitutions of divalent ions (Ca, Mg, Ba, Sr) in the sodium position and Si substitutions for a portion of the phosphorus. Through the proper composition manipulation, one can select a bulk po- Iycrystalline thermal expansion coefficient from slightly negative all the way to + 5 x lO-6fOC. The ability to vary thermal expansion characteristics without major chemical variation has allowed an interesting study of the
Figure 17.40 Side view of fractured SiC-NZP composite showing high degree of debonding, pullout, and bridging. (From Ref. 85.)
Applications: Material Selection 819
Figure 18.2 Si3N~ grinding media of a common configuration used for ball milling or vibratory milling. (Courtesy KemaNord.)
Applicalions: Material Selection 825
1 ',I '.1 , 'l ' 5 '71 , .\
Figure 18.3 Si}N. stator vanes under development for an advanced gas-turbine engine. (Courtesy Garrett Ceramic Components, Phoenix, Ariz .. Division of Al- lied-Signal Aerospace .)
18.25 HlGH·TEMPERATURE CEMENT
High·temperature ceramic cements were discussed briefly in Chap . 1 \. The most important properties are adhesion and high temperature capability. The ability to apply the cement on-site either as a repair or as a recon- struction is often another important consideration .
Many ceramic cements are available commercially and others can be compounded easily by an individual to meet special requirements. The selection depends on the specific application. Some commonly used high- temperature cements are calcium aluminate, sodium silicate. and mon- oaluminum phosphate.
18.26 ABRASIVE FOR CUTOFF AND GRINDING WHEELS
The primary property requirement of the abrasive is high hardness. or at least higher hardness than the material being machined. Cutoff and grinding wheels used for machining of densified ceramics usually require diamond or cubic BN abrasive. Bonded AI,O, or SiC are normally used for metals.
The hardness of the abrasive is not the only criterion for a cutoff or grinding wheel to function efficiently. Controlled breakdown of either the
828 Chapter 18
Figure 18.S Typical SiJN~-based cutting tool inserts, (Courtesy Advanced Ce- ramics. Norton Co., Worcester, Mass.)
Ceramic cutting tools have allowed faster cutting for several reasons. One reason is the lack of deformation of the ceramic at high temperature. Another is the resistance to chemical corrosion under the high temperature conditions at the interface of the insert and the metal workpiece . In some cases where the life of We-based inserts was limited by chemical reaction, a thin coating of AhO, applied by vapor deposition resulted in increased life. Multilayer coatings of AhO" TiN, and TiC on cermets resulted in 200 to 500% increase in wear resistance. This was achieved with a tot31 coating thickness of only 5 to 15 11m. These coated inserts could also be operated at higher speed than uncoated cermets. · By 1985 about 45% of cermet inserts had ceramic coatings. These hybrid tools take advantage of the high toughness of the cermet and the hardness and chemical inertness of the ceramic.
18.28 HIGH·TEMPERATURE HEATING ELEMENT OR IGNITER
Most heating elements and igniters are fabricated from materials that have semiconduction electrical behavior. The electrical resistance results in heat- ing as an electric current works to pass through the material. The amount and rate of heating can be controlled by selecting the cross section of the heating element or igniter and the resistivity of the material.
Applications: Material Selection
\\ • ,
.. / .; " : (
Figure 18.6 Variety of SiC heating elements. (c) is an igniter for a gas clothes dryer. (Parts a and d courtesy of Carborundum Co., Niagara Falls, N. Y. Parts b and c courtesy of Norton Co., Worcester, Mass.)
Applications: Material Selection 831
.. ' , " ~ . .-." , " .. ~
Figure 18.7 Si .\N~ bearings. (Courtesy Miniature Precision Bearings, Keene. New Hampshire .)
The Si.,N .. bearings did not fail catastrophically like prior ceramics. Instead . they failed by slow development of surface spaliation very similar to the failure mode of metals. The reason for this was determined to be the higher level of fracture toughness of the Si,N, compared to prior ce- ramics (15J . The moderate (rather than low) level of thermal conductivity and thermal-shock resistance may also be factors. Finally. a very favorable factor compared to metals is that Si]N .. has approximately 40% of the specific gravity of M-50 steel. This results in substantial reduction in stress for a given bearing size and speed.
l. Pure Carbon Technical Information Pamphlet PC-5393-5M. 1979. 2. R. R. Paxton. Electrochem. Technol. 5(5-6). 174- 182 (1967). 3. A. Pietsch and K. Styhr, Ceramic heat exchanger applications and develop·
ments, in Ceramics for High Performance Applications-II (J. J. Burke. E. N.