ournal JAm. Ceram.So,86[21305-1602003) Influence of Interfacial Roughness on Fiber Sliding in Oxide Composites with La-Monazite Interphases Janet B. Davis, Randall S. Hay, *f David B. Marshall, *f Peter E D Morgan, and Ali sayir*, s Rockwell Scientific. Thousand Oaks. California 91360 Air Force Research Laboratory, Materials Directorate, Wright-Patterson AFB, Ohio 45433 NASA-Glenn Research Center/Case Western Reserve University, Cleveland, Ohio 44 135 Room-temperature debonding and sliding of fibers coated witl sliding occur in fiber pushout tests with model composites con- La-monazite is assessed using a composite with a polycrystal sisting of LaPOa-coated single crystal fibers of Al2O, and line alumina matrix and fibers of several different single Y3AlsO12 (YAG)in polycrystalline Al2O3 matrices crystal (mullite and sapphire)and directionally solidified Damage-tolerant behavior in ceramic composites requires slid- eutectic(Al,O3/Y3AlsO1 and Al,O3Y-ZrO2) compositions. and pullout of fibers in addition to interfacial debonding These fibers provide a range of residual stresses and interfacial Recent calculations suggest that such pullout would be strongly roughnesses. Sliding occurred over a debond crack at the suppressed in fully dense oxide composites by misfit stresses fiber-coating interface when the sliding displacement and generated during sliding of fibers with rough interfaces or with surface roughness were relatively small. At large sliding minor fluctuations in diameter. For given strain mismatch, these displacements with relatively rough interfaces, the monazite misfit stresses are expected(assuming elastic accommodation)to coatings were deformed extensively by fracture, dislocations, be larger in composites with oxide interphases than in composites nd occasional twinning whereas the fibers were undamaged with turbostratic carbon or boron nitride interphases, which have Dense, fine-grained areas (10 nm grain size) resembling re- low transverse elastic modulus. However the misfit stresses could crystallized microstructures were also observed in the most heavily deformed regions of the coatings. Frictional heating such microstructures at low temperature are discussed, and a different thermal expansion coefficients th matrix and fibers of during sliding is assessed. Potential mechanisms for forming esidual thermal stresses in systems In this study, we investigate the debonding and sliding behavior radiation damage. The ability of La-monazite to undergo both of four La-monazite coated fibers le-crystal alumina and debonding and plastie deformation relatively easily at low mullite, directionally solidified eutectics of Al,O/YAG, and temperatures may enable its use as a composite interface. A,,/Y-ZrO2), chosen to provide different residual stress states and interface morphology. The coated fibers were surrounded with a matrix of polycrystalline Al,O3. Debonding and sliding were assessed using indentation fracture and pushout techniques. Dam R ARE-EARTH orthophosphates(monazite and xenotime)are of e in the coating, including plastic deformation, was identified by rest for fiber-matrix interphases that enable interfacial scanning and transmission electron microscopy(SEM and TEM) debonding and damage tolerance in oxide ites.- They are refractory materials(LaPO4 melting point, 2070C), comp Il. Experimental Procedure ible in high-temperature oxidizing environments with many oxides Four different single crystal or directionally solidified eutectic for future development as fibers and matrixes. They are also oxide fibers, grown at NASA Glenn by a laser-heated float zone GPa). Studies of several combinations of oxides and rare-earth rhabdophane(hydrated LaPO4). The coated fibers were embedded phosphates (LaPO4-Al2O3, LaPO2-ZrO2, CePO2-ZrO2, YPO4- in a-alumina powder(AKP50, Sumitomo Chemicals, Tokyo, AL2O3, and NdPO2-Al2O3)have shown that the oxide-phosphate apan)and hot pressed in graphite dies for I h at 1400.C Uncoated interfacial bond is sufficiently weak that debonding occurs when fibers were included in the same specimen for reference. The fibers ever a crack approaches an interface from within the phos- were arranged in rows within the one hot-pressed disk, with separation between fibers "2 mm, thus ensuring identical process- AL, O, system. Other studies have shown that debonding and onditions for all fibers. In an earlie dy.' the same rhabdophane slurry yielded pure La-monazite, with no excess um or pensive spectroscopy(EDS)analysis of the monazite or by reaction R. Naslain -contributing editor of the monazite with sapphire fibers after long-term heat treatment The fibers had different surface textures and coefficients, thus allowing assessment of the ef No. 187143. Received March I l morphology and residual stress on debonding of nisms. The fibers were as follows (1) Directionally solidified Al,O,/ZrO, eutectic fibe by the U.S. Air Force Office tific Research, under Contract No. F4 Contract No. NCC3-372 and Space Administration (NASA), under two-phase microstructure of alumina and cubic zirconia(stabilized with Y2O3). Dimensions of the individual phases were -0.5 um Member, American Ceramic Society. The starting compos sIton o f the feed rod was 60.8 mol% Al,O3: ckwell Scientific Air Force Research Laboratory, Materials Directorate, Wright-Patterson AFB 39.2 mol% ZrO2(9.5 mol%Y,O3) with purity levels 99.995%OI better. X-ray diffractometry(XRD) and SEM/TEM analysis
Influence of Interfacial Roughness on Fiber Sliding in Oxide Composites with La-Monazite Interphases Janet B. Davis,† Randall S. Hay,* ,‡ David B. Marshall,* ,† Peter E. D. Morgan,† and Ali Sayir* ,§ Rockwell Scientific, Thousand Oaks, California 91360 Air Force Research Laboratory, Materials Directorate, Wright-Patterson AFB, Ohio 45433 NASA–Glenn Research Center/Case Western Reserve University, Cleveland, Ohio 44135 Room-temperature debonding and sliding of fibers coated with La-monazite is assessed using a composite with a polycrystalline alumina matrix and fibers of several different single crystal (mullite and sapphire) and directionally solidified eutectic (Al2O3/Y3Al5O12 and Al2O3/Y-ZrO2) compositions. These fibers provide a range of residual stresses and interfacial roughnesses. Sliding occurred over a debond crack at the fiber-coating interface when the sliding displacement and surface roughness were relatively small. At large sliding displacements with relatively rough interfaces, the monazite coatings were deformed extensively by fracture, dislocations, and occasional twinning, whereas the fibers were undamaged. Dense, fine-grained areas (10 nm grain size) resembling recrystallized microstructures were also observed in the most heavily deformed regions of the coatings. Frictional heating during sliding is assessed. Potential mechanisms for forming such microstructures at low temperature are discussed, and a parallel is drawn with the known resistance of monazite to radiation damage. The ability of La-monazite to undergo both debonding and plastic deformation relatively easily at low temperatures may enable its use as a composite interface. I. Introduction R ARE-EARTH orthophosphates (monazite and xenotime) are of interest for fiber-matrix interphases that enable interfacial debonding and damage tolerance in oxide composites.1–11 They are refractory materials (LaPO4 melting point, 2070°C),12 compatible in high-temperature oxidizing environments with many oxides that are either currently available as reinforcing fibers or of interest for future development as fibers and matrixes. They are also relatively soft for such refractory materials (LaPO4 hardness, 5GPa).1 Studies of several combinations of oxides and rare-earth phosphates (LaPO4–Al2O3, LaPO4–ZrO2, CePO4–ZrO2, YPO4– Al2O3, and NdPO4–Al2O3) have shown that the oxide-phosphate interfacial bond is sufficiently weak that debonding occurs whenever a crack approaches an interface from within the phosphate.1,13–15 The most detailed studies have involved the LaPO4– Al2O3 system. Other studies have shown that debonding and sliding occur in fiber pushout tests with model composites consisting of LaPO4-coated single crystal fibers of Al2O3 and Y3Al5O12 (YAG) in polycrystalline Al2O3 matrices.1,16 Damage-tolerant behavior in ceramic composites requires sliding and pullout of fibers in addition to interfacial debonding. Recent calculations suggest that such pullout would be strongly suppressed in fully dense oxide composites by misfit stresses generated during sliding of fibers with rough interfaces or with minor fluctuations in diameter.17 For given strain mismatch, these misfit stresses are expected (assuming elastic accommodation) to be larger in composites with oxide interphases than in composites with turbostratic carbon or boron nitride interphases, which have low transverse elastic modulus. However, the misfit stresses could potentially be reduced by plastic deformation of the interphase. The higher elastic modulus in oxide interphases also causes larger residual thermal stresses in systems with matrix and fibers of different thermal expansion coefficients. In this study, we investigate the debonding and sliding behavior of four La-monazite coated fibers (single-crystal alumina and mullite, directionally solidified eutectics of Al2O3/YAG, and Al2O3/Y-ZrO2), chosen to provide different residual stress states and interface morphology. The coated fibers were surrounded with a matrix of polycrystalline Al2O3. Debonding and sliding were assessed using indentation fracture and pushout techniques. Damage in the coating, including plastic deformation, was identified by scanning and transmission electron microscopy (SEM and TEM). II. Experimental Procedure Four different single crystal or directionally solidified eutectic oxide fibers, grown at NASA Glenn by a laser-heated float zone technique,18,19 were coated with LaPO4 by dipping in a slurry of rhabdophane (hydrated LaPO4). The coated fibers were embedded in -alumina powder (AKP50, Sumitomo Chemicals, Tokyo, Japan) and hot pressed in graphite dies for 1 h at 1400°C. Uncoated fibers were included in the same specimen for reference. The fibers were arranged in rows within the one hot-pressed disk, with separation between fibers 2 mm, thus ensuring identical processing conditions for all fibers. In an earlier study,3 the same rhabdophane slurry yielded pure La-monazite, with no excess lanthanum or phosphorus being detectable either by energy dispersive spectroscopy (EDS) analysis of the monazite or by reaction of the monazite with sapphire fibers after long-term heat treatment (200 h at 1600°C). The fibers had different surface textures and thermal expansion coefficients, thus allowing assessment of the effects of interfacial morphology and residual stress on debonding and sliding mechanisms. The fibers were as follows: (1) Directionally solidified Al2O3/ZrO2 eutectic fibers with a two-phase microstructure of alumina and cubic zirconia (stabilized with Y2O3).20 Dimensions of the individual phases were 0.5 m. The starting composition of the feed rod was 60.8 mol% Al2O3; 39.2 mol% ZrO2 (9.5 mol% Y2O3) with purity levels 99.995% or better. X-ray diffractometry (XRD) and SEM/TEM analysis did R. Naslain—contributing editor Manuscript No. 187143. Received March 11, 2002; approved October 1, 2002. Funding for this work at Rockwell was provided by the U.S. Air Force Office of Scientific Research, under Contract Nos. F49620-96-C-0026 and F49620-00-C-0010. Work at NASA on development of new directionally solidified fibers were supported by the U.S. Air Force Office of Scientific Research, under Contract No. F49620-00- 1-0048 and the National Aeronautics and Space Administration (NASA), under Contract No. NCC3-372. *Member, American Ceramic Society. † Rockwell Scientific. ‡ Air Force Research Laboratory, Materials Directorate, Wright-Patterson AFB. § NASA–Glenn Research Center/Case Western Reserve University. J. Am. Ceram. Soc., 86 [2] 305–16 (2003) 305 journal
Journal of the American Ceramic SocieryDavis et al. Vol 86. No. 2 not show any evidence for a third phase, indicating that all the of --200 um Thermal mismatch during cooling of the composite Y,O, was in solid solution in the zro,. The surfaces caused tensile radial stresses normal to the fiber surface(Table D) were rough on the scale of the microstructure(Fig. I(a)). The fiber (2) Directionally solidified Al,Oy/YAG eutectic fibers, wi diameters were "100 um with fluctuations of - 2 um over lengths two-phase microstructure of dimensions 0.5 um and surface roughness on the scale of the microstructure(Fig. I(b). The fiber diameters were 100 um, with fluctuations of <I um over lengths of -l mm. Thermal mismatch stresses were of the same sign as for the Al,O,/ZrO, fibers, but were smaller in magnitude able D) (3) Mullite single-crystal fibers formed from a so high-purity(99.99%)p na powder (CERAC Milwaukee, Wn)and 99.99% pure SiO,(Alfa Products, Ward Hil MA), which gave 2: 1 mullite as described in Ref. 19. In the as-grown condition, the fibers had smooth surfaces but relative large fluctuations in diameter(50 +5 um, Fig. I(c)) with perio 100 um. Thermal mismatch caused large compressive radial stress in the coating and at the fiber-coating and coating-matrix interfaces, with tensile circumferential stress in the coating and matrix(Table 1). (4) Sapphire fibers, which had smooth surfaces(as-grown) and relatively uniform diameter (100 I um). These wer included for comparison with previous studies of this system. ,3 2um All residual stresses except the circumferential(and axial)tension in the coating are small The hot-pressed disk was cut into slices(thickness.3-2 mm) normal to the fibers. The surfaces of the slices were polished using diamond paste and some of the polished slices were thermally etched. The thicker slices were used for indentation cracking experiments, which involved placing Vickers indentations(10 kg AL, O,ZrO,(eutectic) load) in the polycrystalline alumina matrix near the fibers. The indenter was oriented so that one of the median/radial cracks grew toward the fiber to test for interfacial debonding. The thinner slices ■UU■ were used for fiber pushout experiments, which involved loadin a flat punch( truncated Vickers indenter)onto the end of each fiber while the slice was supported in a fixture with a gap beneath the fiber. Some specimens were fractured after the pushout test to separate the debonded interface. The indented and pushed out specimens were examined by optical microscopy and sEM Specimens used for fiber pushout were also sectioned parallel and perpendicular to the fiber axes and examined by TEM(Model CM20 FEG operating at 200 kV, Phillips, Eindhoven, Nether- lands) to allow identification of damage within the LapOa coatin caused by debonding and sliding. Four Al,O YAG fibers were examined in the parallel section; one mullite and one Al,O/ZrO2 were examined in the axial section. The TEM foils were prepared by impregnating the specimens with epoxy, tripod polishing to a thickness of 10 um, followed by ion milling(Model 691 operating at 4.5 kv, Gatan, Pleasanton, CA) (I) Microstructural Observations All the coated fibers were LapOa and a fully dense matrix of polycrystalline Al,O3- Defor- mation during hot pressing caused the coating thickness to be YAG/AL O3(eutectic larger along the sides of the fibers(-5 um) than at the top and bottom(-I um). No reactions were observed between the LaPO a and any of the fibers, although a few isolated elongated La- magnetoplumbite (LaAl1O1g)grains were observed along the coating-matrix interface (perhaps the result of alkali or divalent impurities in the matrix, which are known to assist formation of rare-earth magetoplumbite-like structures). Despite the presence Mullite (single crystal) of substantial tensile residual stresses in all the LapOa coatings (300-400 MPa, Table D), no evidence of cracking was detected by SEM examination of polished or thermally etched cross sections(although fine-scale through-thickness coating cracks were observed in thin TEM foils of other similar composites). The 1. SEM micrographs of fiber surfaces: (a)Al,, /ZrO2 eutectic fiber, grain sizes were -0.5-1 um in the monazite and -2-10 um in the (b)Al2O3/YAG eutectic fiber, and (c)mullite single crystal fiber alumina matrix
not show any evidence for a third phase, indicating that all the Y2O3 was in solid solution in the ZrO2. The surfaces of the fibers were rough on the scale of the microstructure (Fig. 1(a)). The fiber diameters were 100 m with fluctuations of 2 m over lengths of 200 m. Thermal mismatch during cooling of the composite caused tensile radial stresses normal to the fiber surface (Table I). (2) Directionally solidified Al2O3/YAG eutectic fibers,21 with a two-phase microstructure of dimensions 0.5 m and surface roughness on the scale of the microstructure (Fig. 1(b)). The fiber diameters were 100 m, with fluctuations of 1 m over lengths of 1 mm. Thermal mismatch stresses were of the same sign as for the Al2O3/ZrO2 fibers, but were smaller in magnitude (Table I). (3) Mullite single-crystal fibers formed from a source rod of high-purity (99.99%) polycrystalline alumina powder (CERAC, Milwaukee, WI) and 99.99% pure SiO2 (Alfa Products, Ward Hill, MA), which gave 2:1 mullite as described in Ref. 19. In the as-grown condition, the fibers had smooth surfaces but relatively large fluctuations in diameter (50 5 m, Fig. 1(c)) with period 100 m. Thermal mismatch caused large compressive radial stress in the coating and at the fiber-coating and coating-matrix interfaces, with tensile circumferential stress in the coating and matrix (Table I). (4) Sapphire fibers, which had smooth surfaces (as-grown) and relatively uniform diameter (100 1 m). These were included for comparison with previous studies of this system.1,3 All residual stresses except the circumferential (and axial) tension in the coating are small. The hot-pressed disk was cut into slices (thickness 0.3–2 mm) normal to the fibers. The surfaces of the slices were polished using diamond paste and some of the polished slices were thermally etched. The thicker slices were used for indentation cracking experiments, which involved placing Vickers indentations (10 kg load) in the polycrystalline alumina matrix near the fibers. The indenter was oriented so that one of the median/radial cracks grew toward the fiber to test for interfacial debonding. The thinner slices were used for fiber pushout experiments, which involved loading a flat punch (truncated Vickers indenter) onto the end of each fiber, while the slice was supported in a fixture with a gap beneath the fiber. Some specimens were fractured after the pushout test to separate the debonded interface. The indented and pushed out specimens were examined by optical microscopy and SEM. Specimens used for fiber pushout were also sectioned parallel and perpendicular to the fiber axes and examined by TEM (Model CM20 FEG operating at 200 kV, Phillips, Eindhoven, Netherlands) to allow identification of damage within the LaPO4 coating caused by debonding and sliding. Four Al2O3/YAG fibers were examined in the parallel section; one mullite and one Al2O3/ZrO2 were examined in the axial section. The TEM foils were prepared by impregnating the specimens with epoxy, tripod polishing to a thickness of 10 m, followed by ion milling (Model 691 operating at 4.5 kV, Gatan, Pleasanton, CA).26 III. Results (1) Microstructural Observations All the coated fibers were surrounded with a continuous layer of LaPO4 and a fully dense matrix of polycrystalline Al2O3. Deformation during hot pressing caused the coating thickness to be larger along the sides of the fibers (5 m) than at the top and bottom (1 m). No reactions were observed between the LaPO4 and any of the fibers, although a few isolated elongated Lamagnetoplumbite (LaAl11O19) grains were observed along the coating-matrix interface (perhaps the result of alkali or divalent impurities in the matrix, which are known to assist formation of rare-earth magetoplumbite-like structures3 ). Despite the presence of substantial tensile residual stresses in all the LaPO4 coatings (300–400 MPa, Table I), no evidence of cracking was detected by SEM examination of polished or thermally etched cross sections (although fine-scale through-thickness coating cracks were observed in thin TEM foils of other similar composites). The grain sizes were 0.5–1 m in the monazite and 2–10 m in the alumina matrix. Fig. 1. SEM micrographs of fiber surfaces: (a) Al2O3/ZrO2 eutectic fiber, (b) Al2O3/YAG eutectic fiber, and (c) mullite single crystal fiber. 306 Journal of the American Ceramic Society—Davis et al. Vol. 86, No. 2
February 2003 Influence of Interfacial Roughness on Fiber Sliding in Oxide Composites with La-Monacite 307 Table 1. Representative Residual Stresses in Composites of Monazite-Coated Fibers in a Polycrystalline Al, O3 Matrix Residual stress(MPa Stress component AL,O,/YAG AL,,/ZrO2 Radial (coating/fiber) Radial(matrix/coating) Circumferential(coating) 300 290 Axial(fiber) 1160 240 Values in this table are intended only as rough guide for relative stresses. They were calculated using a lowing Young's moduli and thermal Al2 YAG(350GPa8.5×10-6°c- YAG/AI fiber LaPo YAG/Al2 O3 fiber ● 2um 2um LaPo Fiber YAG/AlO3 Fig. 2. SEM micrographs showing ions of indentation cracks with Al, O,/YAG eutectic fibers:(a)uncoated fiber in alumina matrix(indentation located below region shown);(b) fiber coated with LaPO (indentation located out of field of views, as indicated in(d)); (c)same fiber as in(b) but showing egion further along the debonded interface(arrows indicate magnitude of sliding displacement across debond crack); (d) schematic showing locations of(b)
Fig. 2. SEM micrographs showing interactions of indentation cracks with Al2O3/YAG eutectic fibers: (a) uncoated fiber in alumina matrix (indentation located below region shown); (b) fiber coated with LaPO4 (indentation located out of field of views, as indicated in (d)); (c) same fiber as in (b) but showing region further along the debonded interface (arrows indicate magnitude of sliding displacement across debond crack); (d) schematic showing locations of (b) and (c). Table I. Representative† Residual Stresses in Composites of Monazite-Coated Fibers in a Polycrystalline Al2O3 Matrix Stress component Residual stress (MPa) Sapphire Mullite Al2O3/YAG Al2O3/ZrO2 Radial (coating/fiber) 15 720 130 240 Radial (matrix/coating) 25 630 140 240 Circumferential (coating) 300 420 290 280 Axial (fiber) 7 1160 240 420 † Values in this table are intended only as rough guide for relative stresses. They were calculated using a coaxial cylinder analysis,22 assuming a temperature change of T 1000°C, coating thickness of 2 m, zero volume fraction of fibers, and the following Young’s moduli and thermal expansion coefficients (nominal isotropic, temperature-independent values): polycrystalline Al2O3 (400 GPa, 8 106 °C1 ); sapphire (400 GPa, 8 106 °C1 ); mullite (200 GPa, 4 106 °C1 ); Al2O3/ZrO2 (300 GPa, 9 106 °C1 ); and Al2O3/YAG (350 GPa, 8.5 106 °C1 ).23–25 February 2003 Influence of Interfacial Roughness on Fiber Sliding in Oxide Composites with La-Monazite 307
Journal of the American Ceramic SocieryDavis et al. VoL 86. No. 2 ( Interfacial Debonding LaPO, interface(-4.5 J/m). It is noteworthy that the fibers were cted all the fibers from penetration of protected from cracking even when the contact area of the Vickers indentation cracks, whereas uncoated fibers were always pene indentation was close enough to the fiber to overlap the coating AL,O,/YAG, AL, O,ZrO, and (Fig 3(b). In that case, the residual stress from the indentatio mullite fibers in Figs. 2-4. The indentation cracks generally (compressive normal to the interface, tensile on the prospective xtended from the matrix into the lapo, coatings then arrested crack plane into the fiber) would tend to inhibit debonding and nd caused debonding at the coating/fiber interface. In a few cases favor fiber penetration. with the AL,O,/Zro, fibers, debonding occurred at both interfaces The interfacial roughnesses for both of the eutectic fibers were (matrix/ coating and coating/fiber). The former response was ob- similar to the surface roughnesses of the as-formed fibers, with served previously with coated sapphire fibers and was shown to be amplitude"100-300 nm and period 500 nm(Figs. 2(a) and consistent with the debond criterion of He and Hutchinson- and 3(a)). This roughness amplitude is greater than that of the inter the measured fracture toughnesses of the fibers, coating, and faces at the single-crystal mullite and sapphire fibers. The initially interface.Although the fracture toughnesses of the YAG/ApoA smooth single crystal fibers developed cusps during hot pressing d mullite/LaPO4 interfaces have not been measured, the present where grain boundaries of the monazite coating intersected the observations suggest that they are similar to that of the alumina/ fiber surface. Measurements of the cusp profiles on sapphire fibers Indentation ALO / ZrO. lber Al2O3 Al2O3/Zro2 Al2o 2 um 20 um Fig. 3. SEM micrographs showing interactions of indentation cracks with Al,O, eutectic fibers:(a)uncoated fiber in alumina matrix(indentation located below region shown);(b) fiber coated with LapO4(indentation located at top of field of view) LaPo (b) Al2O3 LaPO Mullite Mullite 21 Fig. 4. SEM micrographs showing interaction of indentation crack with single-crystal mullite fiber(coated with LaPO4, in alumina matrix):(a) of indentation crack with interface and debonding(indentation located above region shown), (b) same fiber as in(a) but showing region further to the right along the debonded interface (arrows indicate magnitude of sliding displacement across debond crack)
(2) Interfacial Debonding The LaPO4 coatings protected all the fibers from penetration of indentation cracks, whereas uncoated fibers were always penetrated. Examples are shown for the Al2O3/YAG, Al2O3/ZrO2 and mullite fibers in Figs. 2–4. The indentation cracks generally extended from the matrix into the LaPO4 coatings then arrested and caused debonding at the coating/fiber interface. In a few cases with the Al2O3/ZrO2 fibers, debonding occurred at both interfaces (matrix/coating and coating/fiber). The former response was observed previously with coated sapphire fibers and was shown to be consistent with the debond criterion of He and Hutchinson27 and the measured fracture toughnesses of the fibers, coating, and interface.1 Although the fracture toughnesses of the YAG/LaPO4 and mullite/LaPO4 interfaces have not been measured, the present observations suggest that they are similar to that of the alumina/ LaPO4 interface (4.5 J/m2 ). It is noteworthy that the fibers were protected from cracking even when the contact area of the Vickers indentation was close enough to the fiber to overlap the coating (Fig. 3(b)). In that case, the residual stress from the indentation (compressive normal to the interface, tensile on the prospective crack plane into the fiber) would tend to inhibit debonding and favor fiber penetration. The interfacial roughnesses for both of the eutectic fibers were similar to the surface roughnesses of the as-formed fibers, with amplitude 100–300 nm and period 500 nm (Figs. 2(a) and 3(a)). This roughness amplitude is greater than that of the interfaces at the single-crystal mullite and sapphire fibers. The initially smooth single crystal fibers developed cusps during hot pressing where grain boundaries of the monazite coating intersected the fiber surface. Measurements of the cusp profiles on sapphire fibers Fig. 3. SEM micrographs showing interactions of indentation cracks with Al2O3/ZrO2 eutectic fibers: (a) uncoated fiber in alumina matrix (indentation located below region shown); (b) fiber coated with LaPO4 (indentation located at top of field of view). Fig. 4. SEM micrographs showing interaction of indentation crack with single-crystal mullite fiber (coated with LaPO4, in alumina matrix): (a) intersection of indentation crack with interface and debonding (indentation located above region shown); (b) same fiber as in (a) but showing region further to the right along the debonded interface (arrows indicate magnitude of sliding displacement across debond crack). 308 Journal of the American Ceramic Society—Davis et al. Vol. 86, No. 2
February 2003 Influence of Interfacial Roughness on Fiber Sliding in Oxide Composites with La-Monacite 309 by atomic force microscopy (ATM) have been reported else- fiber surface. In some areas, this smeared layer was overlaid with where28,29 The cusp heights were typically 50 nm, and the a less dense, coarser grained agglomeration of angular monazite ular distortions of the surface were small(=20%). The cusps or particles of diameter 50-100 m(Fig. 7), suggestive f cata- the mullite surfaces were very similar clastic flow, a process involving repeated microfracture and Some insight into the effect of interfacial roughness on fiber fine-particle transport. Similar features(intense deformation sliding and pullout can be gained from the observations of Figs. fine crystallites, and agglomerates of angular particles)were 2-4. As the debond grows around the circumference of the fiber, observed in monazite debris (irregularly shaped balls, -100-500 the loading on the crack tip due to the indentation stress field is nm diameter) in the debond crack. initially mostly shear(although the loading eventually changes to C) Mullite Fibers: Sliding of the mullite fibers occurred tension if the crack grows sufficiently ) Because fiber pullout also predominantly at the fiber-coating interface SEM observations of involves shear loading of a debond crack, the initial region of separated interfaces showed plastic deformation of the LaPOa growth of the deflected cracks in Figs 2-4 should be representa- coating where the varying fiber diameter caused compression tive of the behavior during the corresponding stage of pullout. the coating during sliding, as depicted in region B of Fig. 9.(Note In all cases. the initial debond crack followed the fiber-matrn that the sliding displacements are smaller than the period of the interface, even when the interface was rough. For the mullite fibers diameter fluctuations and larger than the spacing of cusps associ- (Fig. 4)the sliding displacement of the debond crack surfaces ated with grain boundaries in the LaPO, coating )Where sliding the coating-fiber interface(region A in Fi is smaller than the average spacing between the interfacial cusps 9), the separated interface was similar to that of the sapphire fiber, (-600 nm). Sliding caused separation of the debonded surfaces to with grain-boundary cusps, clean separation, and no damage in the ccommodate their misfit (Fig. 4(b)), despite the constrained fiber or the coating configuration with large residual compressive normal stress(-700 MPa, Table D). The misfit was apparently accommodated by elastic diameter fluctuation by TEM was difficult, because only limited strains. with no irreversible deformation of the mullite fiber or the areas were observed. Nevertheless. some trends are evident. LaPO4 coating discernable by SEM. In contrast, sliding of the Deformation was distributed, ofte eutectic fibers caused extensive damage in the LaPO coating(Fig. entire coating thickness(Figs. 10 and 11), rather than being 2(c), without discernable damage in the fibers. The dama localized in a thin layer adjacent to the fiber as for the Al,O3/YAG included cracks across the full width of the coating. aligned at 45 to the interface on planes of maximum tension, similar to undeformed, whereas in others plastic deformation was confined to enVious o pservations of cracking in layers of LaPOa sandwiched an isolated grain(Fig. 10). The region of Fig. 10 was thought to between polycrystalline Al 2O3. More intense local damage is have experienced tension during sliding(as in Region A, Fig 9) evident at individual asperities, as in Fig. 2(c). The damage although the correlation with fiber diameter is uncertain because included cracking and fine lamellar features, which could be some of the fiber adjacent to the debond crack was removed during racks or twins ion milling. Extensive microcracking was distributed throughout the coating, often at x+45 to the fiber surface(Figs. 10 and 11) 3) Fiber pushout In regions of coating inferred to have been compressed during sliding(as in region B of Fig. 9), almost the entire coating was All the fibers debonded during the pushout experiments. Sliding microcracked and plastically deformed (Fig. 11). Extensive dislo- curred unstably over distances of "5-10 um at a critical load between 10-20 N. The average shear stress (load divided by fiber cation plasticity was evident, with variations in density from grai urface area)at the critical load was 130 10 MPa for the to grain. Some grains were twinned parallel to the fiber/matrix sapphire fibers, 200+ 20 MPa for the mullite fibers, 190+20 interface(Fig. ID), the orientation of maximum shear stress due to MPa for the Al2O3/YAG fibers, and 255 t 30 MPa for the pushout of the fiber. Microcracking at -45 to the fiber surface AL,O/ZrO, fibers was extensive, with crack spacings as small as "50 nm and ar (A) Sapphire Fibers: Sliding of the sapphire fiber occurred abundance of planar segments consistent with cleavage on, (100), at the fiber-coating interface, as reported previously. Grain (010), and (001), as reported previously. There was some tendency for cracks oriented normal to the maximum tensile stress boundary cusps were observed al he separated interfaces by SEM and AFM, although no damage was visible in either the fiber (northwest to southeast in Fig. 11)to be longer and have greater opening displacements than other cracks; however, the trends are subjective and the possibility of a sample preparation artifact B)Al2O,YAG and AlyOyzrO, Fibers: Extensive wear cannot be ruled out tracks were observed in the ApoA coatings on both eutectic fibers, indicating that sliding involved plastic deformation (Fig. 5). Sliding occurred mostly adjacent to the fiber-matrix interfac although smeared fragments of the LaPO coating remained on the (1) Effects of Residual Stress ber surface. In some regions(such as Fig. 5), sliding occurred near the matrix-coating interface The residual stresses noted in Table I might be expected to TEM observations from a typical specimen containing influence interfacial debonding. Therefore, it is necessary to ut Al,O3/YAG fiber are shown in Figs. 6-8. Slidi establish whether the fracture behavior in the model experiments along a debond crack between the ApoA coating and the fiber eported here is representative of that in real composites, given the most regions, a thin layer of the lapA coating within -100-300 differences in residual stress states and crack orientation nm of the fiber was heavily deforme In the analysis of He et al, 2 the presence of residual stresses igs 6-8). The intensity of shifts the debond criterion by an amount that depends on the deformation decreased with distance from the debond crack, with ions more than-500 nm from the fiber being undeformed. The parameters mp and nd Deformation in the ApoA consisted of tangled dislocation K lamellar features resembling twins, microcracks, and regions of densely packed fine crystallites(diameter as small as 10 nm) that where o and od are the residual stresses normal to potential crack resemble recrystallized microstructures(Fig. 6). The density of paths along the interface or into the fiber, K is the applied stress dislocations varied from grain to grain and generally decreased intensity factor for the incident crack, and a is a defect size. For a with distance from the debond crack. The nano-crystalline regions crack approaching the fiber on a radial plane, as in the indentation were within -50-100 nm of the debond crack. In one region, there cracking experiments of Section Il(2), the residual stresses o and was no deformation on the monazite side of the debond crack, but ca (radial and hoop stresses at the fiber surface) are equal, so the a thin layer of dense nano-crystalline monazite was smeared on the debond condition is not affected by the residual stresses
by atomic force microscopy (ATM) have been reported elsewhere.28,29 The cusp heights were typically 50 nm, and the angular distortions of the surface were small (20°). The cusps on the mullite surfaces were very similar. Some insight into the effect of interfacial roughness on fiber sliding and pullout can be gained from the observations of Figs. 2–4. As the debond grows around the circumference of the fiber, the loading on the crack tip due to the indentation stress field is initially mostly shear (although the loading eventually changes to tension if the crack grows sufficiently). Because fiber pullout also involves shear loading of a debond crack, the initial region of growth of the deflected cracks in Figs. 2–4 should be representative of the behavior during the corresponding stage of pullout. In all cases, the initial debond crack followed the fiber-matrix interface, even when the interface was rough. For the mullite fibers (Fig. 4) the sliding displacement of the debond crack surfaces (250 nm, i.e., opening displacement of initial indentation crack) is smaller than the average spacing between the interfacial cusps (600 nm). Sliding caused separation of the debonded surfaces to accommodate their misfit (Fig. 4(b)), despite the constrained configuration with large residual compressive normal stress (700 MPa; Table I). The misfit was apparently accommodated by elastic strains, with no irreversible deformation of the mullite fiber or the LaPO4 coating discernable by SEM. In contrast, sliding of the eutectic fibers caused extensive damage in the LaPO4 coating (Fig. 2 (c)), without discernable damage in the fibers. The damage included cracks across the full width of the coating, aligned at 45° to the interface on planes of maximum tension, similar to previous observations of cracking in layers of LaPO4 sandwiched between polycrystalline Al2O3. 1 More intense local damage is evident at individual asperities, as in Fig. 2(c). The damage included cracking and fine lamellar features, which could be cracks or twins. (3) Fiber Pushout All the fibers debonded during the pushout experiments. Sliding occurred unstably over distances of 5–10 m at a critical load between 10–20 N. The average shear stress (load divided by fiber surface area) at the critical load was 130 10 MPa for the sapphire fibers, 200 20 MPa for the mullite fibers, 190 20 MPa for the Al2O3/YAG fibers, and 255 30 MPa for the Al2O3/ZrO2 fibers. (A) Sapphire Fibers: Sliding of the sapphire fiber occurred at the fiber-coating interface, as reported previously.1 Grainboundary cusps were observed along the separated interfaces by SEM and AFM, although no damage was visible in either the fiber or the coating. (B) Al2O3/YAG and Al2O3/ZrO2 Fibers: Extensive wear tracks were observed in the LaPO4 coatings on both eutectic fibers, indicating that sliding involved plastic deformation (Fig. 5). Sliding occurred mostly adjacent to the fiber-matrix interface, although smeared fragments of the LaPO4 coating remained on the fiber surface. In some regions (such as Fig. 5), sliding occurred near the matrix-coating interface. TEM observations from a typical specimen containing a pushed out Al2O3/YAG fiber are shown in Figs. 6–8. Sliding occurred along a debond crack between the LaPO4 coating and the fiber. In most regions, a thin layer of the LaPO4 coating within 100–300 nm of the fiber was heavily deformed (Figs. 6–8). The intensity of deformation decreased with distance from the debond crack, with regions more than 500 nm from the fiber being undeformed. The Al2O3/YAG fiber was also undamaged. Deformation in the LaPO4 consisted of tangled dislocations, lamellar features resembling twins, microcracks, and regions of densely packed fine crystallites (diameter as small as 10 nm) that resemble recrystallized microstructures (Fig. 6). The density of dislocations varied from grain to grain and generally decreased with distance from the debond crack. The nano-crystalline regions were within 50–100 nm of the debond crack. In one region, there was no deformation on the monazite side of the debond crack, but a thin layer of dense nano-crystalline monazite was smeared on the fiber surface. In some areas, this smeared layer was overlaid with a less dense, coarser grained agglomeration of angular monazite particles of diameter 50–100 nm (Fig. 7), suggestive of cataclastic flow, a process involving repeated microfracture and fine-particle transport.30 Similar features (intense deformation, fine crystallites, and agglomerates of angular particles) were observed in monazite debris (irregularly shaped balls, 100–500 nm diameter) in the debond crack. (C) Mullite Fibers: Sliding of the mullite fibers occurred predominantly at the fiber-coating interface. SEM observations of separated interfaces showed plastic deformation of the LaPO4 coating where the varying fiber diameter caused compression of the coating during sliding, as depicted in region B of Fig. 9. (Note that the sliding displacements are smaller than the period of the diameter fluctuations and larger than the spacing of cusps associated with grain boundaries in the LaPO4 coating.) Where sliding caused tension across the coating-fiber interface (region A in Fig. 9), the separated interface was similar to that of the sapphire fiber, with grain-boundary cusps, clean separation, and no damage in the fiber or the coating. Direct correlation of the changes in coating damage with fiber diameter fluctuation by TEM was difficult, because only limited areas were observed. Nevertheless, some trends are evident. Deformation was distributed, often nonuniformly, through the entire coating thickness (Figs. 10 and 11), rather than being localized in a thin layer adjacent to the fiber as for the Al2O3/YAG fiber. In some places, the monazite adjacent to the fiber was undeformed, whereas in others plastic deformation was confined to an isolated grain (Fig. 10). The region of Fig. 10 was thought to have experienced tension during sliding (as in Region A, Fig. 9), although the correlation with fiber diameter is uncertain because some of the fiber adjacent to the debond crack was removed during ion milling. Extensive microcracking was distributed throughout the coating, often at 45° to the fiber surface (Figs. 10 and 11). In regions of coating inferred to have been compressed during sliding (as in region B of Fig. 9), almost the entire coating was microcracked and plastically deformed (Fig. 11). Extensive dislocation plasticity was evident, with variations in density from grain to grain. Some grains were twinned parallel to the fiber/matrix interface (Fig. 11), the orientation of maximum shear stress due to pushout of the fiber. Microcracking at 45° to the fiber surface was extensive, with crack spacings as small as 50 nm and an abundance of planar segments consistent with cleavage on, (100), (010), and (001), as reported previously.31 There was some tendency for cracks oriented normal to the maximum tensile stress (northwest to southeast in Fig. 11) to be longer and have greater opening displacements than other cracks; however, the trends are subjective and the possibility of a sample preparation artifact cannot be ruled out. IV. Discussion (1) Effects of Residual Stress The residual stresses noted in Table I might be expected to influence interfacial debonding. Therefore, it is necessary to establish whether the fracture behavior in the model experiments reported here is representative of that in real composites, given the differences in residual stress states and crack orientations. In the analysis of He et al., 32 the presence of residual stresses shifts the debond criterion by an amount that depends on the parameters p and d: p pa1/ 2 K , d da1/ 2 K (1) where p and d are the residual stresses normal to potential crack paths along the interface or into the fiber, K is the applied stress intensity factor for the incident crack, and a is a defect size. For a crack approaching the fiber on a radial plane, as in the indentation cracking experiments of Section II(2), the residual stresses p and d (radial and hoop stresses at the fiber surface) are equal, so the debond condition is not affected by the residual stresses. February 2003 Influence of Interfacial Roughness on Fiber Sliding in Oxide Composites with La-Monazite 309
310 Journal of the American Ceramic SocieryDavis et al. Vol 86. No. 2 For cracks oriented normal to the fiber axis(the most important sliding of these fibers is expected to deform the monazite without case for a composite), U is the axial stress in the fiber and od is damaging the fibers the radial stress at the fiber surface. The ratio of the axial to the Wear and abrasion of ceramics are known to cause intense radial stress is -2 for all the fibers (Table D), both stresses being plastic deformation, similar to that in heavily cold-worked metals, compressive for the mullite fibers and tensile for the others. with fine, heavily deformed wear debris as in Fig. 6-8 Therefore, the residual stresses should favor debonding for mullite The depth of the deformed zone is expected to be similar to the fibers and fiber penetration for both eutectic fibers. However, this dimensions of the sliding asperities, consistent with the observa- result is sensitive to the volume fraction of fibers. For a fiber tions of Figs. 6-8(500 nm depth of monazite deformation from volume fraction of 0.5, a typical value for structural composites sliding of Al,O/YAG fiber with roughness amplitude 200 nn the magnitudes of the axial stresses decrease by a factor of -2 to and wavelength -l um). Mullite fibers, with larger roughness a level similar to the radial stresses. Then the residual stresses do amplitude (2.5 um)and wavelength (100 um), deform the not affect the condition for debonding and observations from entire coating, rather than just a thin layer. Abrasive wear of cracks oriented as in the model indentation experiments ar interfaces in a composite can affect the fiber sliding resistance representative of transverse cracking in the composite. The add and thus the mechanical properties of the composite, particu- tional influence of the residual stress field of the indentation, larly during fatigue. which favors penetration of the crack into the fiber, makes the The mechanism responsible for forming the dense nano- indentation crack a conservative indicator for debonding crystalline regions that resemble recrystallized grains, as in Fig. 6, is not known. Recrystallization normally occurs only at suffi- ciently high temperatures and long times to allow dislocation (2) Efects of Misfit Stresses climb. This would not be expected during room-temperature Misfit stresses were generated during fiber sliding by roughness deformation of a refractory material, such as LaPO4, and was not at two length scales, one microstructural (grooves that form at observed in indented monazite. 3 In alumina, which has a similar intersections of grain boundaries in the monazite coating or melting point, recovery and recrystallization of cold-worked m lamellae boundaries of the eutectic fibers with the fiber/coating crostructure typically requires temperatures of at least 800oC interface), and the other caused by long- range fluctuations in fiber and perhaps as high as 1200 C. this raises a question of whether diameter(Fig. I and Table In). Sliding displacements in the friction caused local heating in these experiments, or whether ushout experiments exceeded the microstructural dimensions by a another mechan tht be responsible for this microstruct factor of -5-10, but were smaller than the period of the diamete Two possible mechanisms are low-temperature amorphization and fluctuations by factors of 10-100. The microstructural roug recovery, which has been observed in wear and abrasion of ness dominated the misfit strains for the eutectic fibers and two-phase metals, or very fine-scale comminution accompanied sapphire fibers (Table m) by densification by local deformation. For the mullite fibers, the superposition of misfit strains caused A) Local Heating Effects: Several estimates of temperature by thermal expansion mismatch(Table D)and diameter fluctua- rises during sliding, based on different assumptions about heat tions(at the maximum sliding displacement) would cause com- dissipation mechanisms, are given in Appendix A. An upper pressive radial stresses as high as 5.2 GPa in some regions and adiabatic)limit for quasi-static continuous deformation gives ension in others. Because this maximum compressive stress is temperature rises between~800°-2000°C,de on whethe similar to the hardness of LaPO4(5 GPa), plastic deformation analysis is performed for individual asperities or for an average should occur throughout the coating, as observed in Fig. I contact area. However, according to a very conservative estimate For the other fibers. the maximum radial mismatch caused by an upper-bound for the sliding velocity is several orders of fluctuations in fiber diameter is of similar magnitude and opposite magnitude smaller than that required for adiabatic heating. A ign to the thermal expansion mismatch. Therefore, deformation of similar conclusion is drawn from application of frictional sliding the coating should depend on the microstructural roughness(both analyses, 2-54 which give small temperature increments for ti the roughness shape and the mismatch strain, dr/R). The single upper-bound sliding ve -05C for an asperity calculation rystal fibers had smaller microstructural mismatch strains(Sr/R(reduction of in a matrix with stiffness similar to that used in this study ormal stress by interface separation waves were suggested as a However, in a composite with a porous matrix, the response would cause be mitigated by the reduced constraint, owing to the lower elastic Another mechanism is cataclastic flow o accompanied by ffness of the matrix plastic deformation of the debris. Fine-grained (50-100 nr angular, and porous monazite debris characteristic of cataclastic flow was observed in some regions(Fig. 7). The prevalence of ( Plastic Deformation of LaPO, other deformation microstructures (dislocations, nano- The sliding of rough interfaces over distances large compared rystalline regions)also varied from place to place, suggesti e rou ivelength caused plastic deformation of that there was spatial and perhaps temporal variation in the monazite coatings (F 8). Plastic deformation of monazite by intensity of deformation during the pushout experiments. Local twinning and dislocations has also been observed after quasi-static adiabatic heating could occur during cataclastic flow as a result contact with spherical indenters at room temperature. ,/ Such of imperfect contact between the debris and the surround deformation can occur in any brittle material in the presence of leading to reduced heat conduction to the surroundings, or by ufficient hydrostatic pressure to suppress fracture . 38-43Because local stick-slip motion of the angular debris causing rapid onazite, with a hardness of 5 GPa, is much softer than alumina, mpact of sharp particles. In geological studies, fine-grained irconia, and mullite(hardnesses ranging between 10-40 GPa), debris(fault gouge)is itself suspected to influence whether
For cracks oriented normal to the fiber axis (the most important case for a composite), p is the axial stress in the fiber and d is the radial stress at the fiber surface. The ratio of the axial to the radial stress is 2 for all the fibers (Table I), both stresses being compressive for the mullite fibers and tensile for the others. Therefore, the residual stresses should favor debonding for mullite fibers and fiber penetration for both eutectic fibers. However, this result is sensitive to the volume fraction of fibers. For a fiber volume fraction of 0.5, a typical value for structural composites, the magnitudes of the axial stresses decrease by a factor of 2 to a level similar to the radial stresses. Then the residual stresses do not affect the condition for debonding and observations from cracks oriented as in the model indentation experiments are representative of transverse cracking in the composite. The additional influence of the residual stress field of the indentation, which favors penetration of the crack into the fiber, makes the indentation crack a conservative indicator for debonding. (2) Effects of Misfit Stresses Misfit stresses were generated during fiber sliding by roughness at two length scales, one microstructural (grooves that form at intersections of grain boundaries in the monazite coating or lamellae boundaries of the eutectic fibers with the fiber/coating interface), and the other caused by long-range fluctuations in fiber diameter (Fig. 1 and Table II). Sliding displacements in the pushout experiments exceeded the microstructural dimensions by a factor of 5–10, but were smaller than the period of the diameter fluctuations by factors of 10–100. The microstructural roughness dominated the misfit strains for the eutectic fibers and sapphire fibers (Table II). For the mullite fibers, the superposition of misfit strains caused by thermal expansion mismatch (Table I) and diameter fluctuations (at the maximum sliding displacement) would cause compressive radial stresses as high as 5.2 GPa in some regions and tension in others. Because this maximum compressive stress is similar to the hardness of LaPO4 (5 GPa1 ), plastic deformation should occur throughout the coating, as observed in Fig. 11. For the other fibers, the maximum radial mismatch caused by fluctuations in fiber diameter is of similar magnitude and opposite sign to the thermal expansion mismatch. Therefore, deformation of the coating should depend on the microstructural roughness (both the roughness shape and the mismatch strain, r/R). The single crystal fibers had smaller microstructural mismatch strains ( r/R 0.002) than the eutectic fibers ( r/R 0.004). The only oxide fibers currently available commercially in quantities sufficient to fabricate composites are polycrystalline, with grain size 50–100 nm and diameter 12 m.33,34 The surface roughness due to grain-boundary grooving in as-fabricated fibers is typically very small (5 nm for Nextel 720TM, 3M Corp., St. Paul, MN).35 However, the grooves would be expected to grow during processing of the matrix to depths up to approximately half of the grain size (20–50 nm). Although this roughness amplitude is smaller than that of the eutectic fibers, the mismatch strain is similar or larger ( r/R 0.003–0.005). Therefore, deformation of the coating might be expected if these fibers were to be embedded in a matrix with stiffness similar to that used in this study. However, in a composite with a porous matrix, the response would be mitigated by the reduced constraint, owing to the lower elastic stiffness of the matrix. (3) Plastic Deformation of LaPO4 The sliding of rough interfaces over distances large compared with the roughness wavelength caused plastic deformation of monazite coatings (Figs. 5–8). Plastic deformation of monazite by twinning and dislocations has also been observed after quasi-static contact with spherical indenters at room temperature.31,37 Such deformation can occur in any brittle material in the presence of sufficient hydrostatic pressure to suppress fracture.38–43 Because monazite, with a hardness of 5 GPa,1 is much softer than alumina, zirconia, and mullite (hardnesses ranging between 10–40 GPa),44 sliding of these fibers is expected to deform the monazite without damaging the fibers. Wear and abrasion of ceramics are known to cause intense plastic deformation, similar to that in heavily cold-worked metals, with fine, heavily deformed wear debris as in Fig. 6–8.43,45–48 The depth of the deformed zone is expected to be similar to the dimensions of the sliding asperities, consistent with the observations of Figs. 6–8 (500 nm depth of monazite deformation from sliding of Al2O3/YAG fiber with roughness amplitude 200 nm and wavelength 1 m). Mullite fibers, with larger roughness amplitude (2.5 m) and wavelength (100 m), deform the entire coating, rather than just a thin layer. Abrasive wear of interfaces in a composite can affect the fiber sliding resistance and thus the mechanical properties of the composite, particularly during fatigue.49 The mechanism responsible for forming the dense nanocrystalline regions that resemble recrystallized grains, as in Fig. 6, is not known. Recrystallization normally occurs only at sufficiently high temperatures and long times to allow dislocation climb.50 This would not be expected during room-temperature deformation of a refractory material, such as LaPO4, and was not observed in indented monazite.31 In alumina, which has a similar melting point, recovery and recrystallization of cold-worked microstructures typically requires temperatures of at least 800°C,45 and perhaps as high as 1200°C.51 This raises a question of whether friction caused local heating in these experiments, or whether another mechanism might be responsible for this microstructure. Two possible mechanisms are low-temperature amorphization and recovery, which has been observed in wear and abrasion of two-phase metals,48 or very fine-scale comminution accompanied by densification by local deformation. (A) Local Heating Effects: Several estimates of temperature rises during sliding, based on different assumptions about heat dissipation mechanisms, are given in Appendix A. An upper (adiabatic) limit for quasi-static continuous deformation gives temperature rises between 800°–2000°C, depending on whether analysis is performed for individual asperities or for an average contact area. However, according to a very conservative estimate, an upper-bound for the sliding velocity is several orders of magnitude smaller than that required for adiabatic heating. A similar conclusion is drawn from application of frictional sliding analyses,52–54 which give small temperature increments for this upper-bound sliding velocity (0.5°C for an asperity calculation and 5°C for an average calculation). These calculations indicate that significant increases in temperature could not have occurred in these experiments if the assumptions of quasi-static, continuous deformation are valid. Several mechanisms could potentially violate these conditions by producing local discontinuous deformation. One is stick slip motion, which causes local sliding velocities significantly larger than average.55 The local velocity would need to exceed the maximum upper-bound average velocity by a factor of 100 to approach adiabatic conditions (Appendix A). Although this is possible (local elastic unloading could cause local velocities approaching sonic values), experiments and geological observations have found less heating during stick-slip than during stable sliding55 (reduction of normal stress by interface separation waves were suggested as a cause). Another mechanism is cataclastic flow30 accompanied by plastic deformation of the debris. Fine-grained (50 –100 nm), angular, and porous monazite debris characteristic of cataclastic flow was observed in some regions (Fig. 7). The prevalence of other deformation microstructures (dislocations, nanocrystalline regions) also varied from place to place, suggesting that there was spatial and perhaps temporal variation in the intensity of deformation during the pushout experiments. Local adiabatic heating could occur during cataclastic flow as a result of imperfect contact between the debris and the surroundings, leading to reduced heat conduction to the surroundings, or by local stick-slip motion of the angular debris causing rapid impact of sharp particles. In geological studies, fine-grained debris (fault gouge) is itself suspected to influence whether 310 Journal of the American Ceramic Society—Davis et al. Vol. 86, No. 2
February 2003 Influence of Interfacial Roughness on Fiber Sliding in Oxide Composites with La-Monazite LaPo (b) O,/ZrO. Al2o 10 um 10m Fig. 5. SEM micrographs showing Al2O3/ZrO2 eutectic fiber after push-out:(a) bottom of push-out specimen(monazite-coated eutectic fiber, olycrystalline Al,O, matrix); (b)monazite layer remaining attached to fiber, showing deformation caused by sliding 1um 200nm Monazite Debo bond- Epoxy YAG-Alumina Fiber o nm 50nm Fig. 6. TEM micrographs at progressively increasing magnification from cross section of monazite-coated Al,O, /YAG fiber after push-out. Heavily deformed monazite debris between asperities on the fiber surface is evident at intermediate magnification. Dense nano-crystalline microstructure along the debond crack is evident at high-magnification
Fig. 5. SEM micrographs showing Al2O3/ZrO2 eutectic fiber after push-out: (a) bottom of push-out specimen (monazite-coated eutectic fiber, polycrystalline Al2O3 matrix); (b) monazite layer remaining attached to fiber, showing deformation caused by sliding. Fig. 6. TEM micrographs at progressively increasing magnification from cross section of monazite-coated Al2O3/YAG fiber after push-out. Heavily deformed monazite debris between asperities on the fiber surface is evident at intermediate magnification. Dense nano-crystalline microstructure along the debond crack is evident at high-magnification. February 2003 Influence of Interfacial Roughness on Fiber Sliding in Oxide Composites with La-Monazite 311
312 Journal of the American Ceramic SocieryDavis et al. Vol 86. No. 2 DI 100 mm 100nm BE Monazite Debond YAG- Anina Fiber um Fig. 8. Intense plastic deformation and fine-scale microcracking oating on Al,O,/YAG fiber, Heavily deformed ball of monazite(-10 100nm m diameter) is evident in debond crack at higher magnification (lower Monazite could not be induced (i.e, recrystallization processes were than damage accumulation) was only 35.C for LaPOa, >700C for zircon. This difference was tentatively attributed to Debond the higher stability of isolated POa tetrahedra than isolated Sio units, with less bond-breaking required to crystallize the amor Whether this behavior might be related to ation arter intense mechanical deformation is not clear. Recrystallization of a 400nm YAG-Alumina Fiber surface. Layer adjacent to fiber has dense fine grains(-10-20 1 resembling recrystallized microstructure. Layer further from fiber porous, coarse-grained angular particles diagnostic of cataclastic A20 Matrix tick-slip or stable sliding occurs, with most observations pointing toward inhibition of stick-slip by fine-grained debris. 30 A progression from stick-slip to stable sliding as debris builds up during fiber pushout displacement is possible, with a consequent change in local temperature increases. Unfortu- nately, it is not straightforward to assess any of these effects B)Annealing of Radiation Damage: Monazite is known to recover readily from displacive damage events at near-ambient temperatures, 6, 57 making it extremely resistant to amorphization by radiation damage, and thus an ideal host for containment of actinide or transuranic elements 58, 59 In a recent study bo radiation damage in LaPOa and several related ABO-type phosphates and silicates was monitored as a function of temperature in situ by TEM. Fundamental differences in the amorphization and recrys- RSC02809501 tallization kinetics between the orthophosphates and silicates were bserved. The critical temperature above which amorphization fiber sliding for monazite- coated mullite fiber
stick-slip or stable sliding occurs, with most observations pointing toward inhibition of stick-slip by fine-grained debris.30 A progression from stick-slip to stable sliding as debris builds up during fiber pushout displacement is possible, with a consequent change in local temperature increases. Unfortunately, it is not straightforward to assess any of these effects quantitatively. (B) Annealing of Radiation Damage: Monazite is known to recover readily from displacive damage events at near-ambient temperatures,56,57 making it extremely resistant to amorphization by radiation damage, and thus an ideal host for containment of actinide or transuranic elements.58,59 In a recent study,60 radiation damage in LaPO4 and several related ABO4-type phosphates and silicates was monitored as a function of temperature in situ by TEM. Fundamental differences in the amorphization and recrystallization kinetics between the orthophosphates and silicates were observed. The critical temperature above which amorphization could not be induced (i.e., recrystallization processes were faster than damage accumulation) was only 35°C for LaPO4, but 700°C for zircon. This difference was tentatively attributed to the higher stability of isolated PO4 tetrahedra than isolated SiO4 units, with less bond-breaking required to crystallize the amorphous structure. Whether this behavior might be related to recrystallization after intense mechanical deformation is not clear. Recrystallization of a Fig. 7. TEM micrograph of monazite smeared onto Al2O3/YAG fiber surface. Layer adjacent to fiber has dense fine grains (10–20 nm scale) resembling recrystallized microstructure. Layer further from fiber has more porous, coarse-grained angular particles diagnostic of cataclastic flow. Fig. 8. Intense plastic deformation and fine-scale microcracking in coating on Al2O3/YAG fiber. Heavily deformed ball of monazite (100 nm diameter) is evident in debond crack at higher magnification (lower right-hand corner). Fig. 9. Schematic of fiber sliding for monazite-coated mullite fiber. 312 Journal of the American Ceramic Society—Davis et al. Vol. 86, No. 2
February 2003 Influence of Interfacial Roughness on Fiber Sliding in Oxide Composites with La-Monacite 313 PUSHOUT monazite alumina matrit s um 200m soo nm 50 nm Fig. 10. Monazite coating on mullite fiber: region thought to have experienced tension during fiber sliding(as in Region A, Fig. 9). Monazite next to fiber is mostly undamaged, but entire coating is cracked. Compression increases toward right side of micrograph material with high density of dislocations requires lattice diffusion other ceramics, such as alumina and zircon, at similar homologous for dislocation climb, o whereas recrystallization after amorphiza temperatu measurements of nucleation tion by radiation damage does not require such diffusion. Never tallization, grai nd diffusion are needed to theless, the resistance of monazite to amorphization hints that whether recrys nsely deformed mona m solid-state processes in monazite are faster than those in many occur near roo 2um, PUSHOUT→mnmm1m monazite 100nm almina matrie 200nm Fig. 11. Monazite coating on mullite fiber: region thought to have experienced compression during fiber sliding(as in Region B, Fig. 9). Coating is heavily deformed through entire thickness, although with grain-to grain variation. Large cracks tend to run NW-SE
material with high density of dislocations requires lattice diffusion for dislocation climb,50 whereas recrystallization after amorphization by radiation damage does not require such diffusion. Nevertheless, the resistance of monazite to amorphization hints that solid-state processes in monazite are faster than those in many other ceramics, such as alumina and zircon, at similar homologous temperatures.60 Independent measurements of nucleation, recrystallization, grain growth, and diffusion are needed to determine whether recrystallization of intensely deformed monazite might occur near room temperature. Fig. 10. Monazite coating on mullite fiber: region thought to have experienced tension during fiber sliding (as in Region A, Fig. 9). Monazite next to fiber is mostly undamaged, but entire coating is cracked. Compression increases toward right side of micrograph. Fig. 11. Monazite coating on mullite fiber: region thought to have experienced compression during fiber sliding (as in Region B, Fig. 9). Coating is heavily deformed through entire thickness, although with grain-to-grain variation. Large cracks tend to run NW–SE. February 2003 Influence of Interfacial Roughness on Fiber Sliding in Oxide Composites with La-Monazite 313
314 Journal of the American Ceramic SocieryDavis et al. Vol 86. No. 2 Table I. Misfit Strains and Stresses Fiber radius(Rμum) Microstructural roughness Amplitude, Sr(um) 0.05 0.2 0. Period, A(ur 0.001 0.002 0.004 0.004 g (MPa) 0.5 0.5 (△RR)(m=mAg) 0.0006 0.0003 0.0015 Ide for comparisons: stresse ated as in Table I but with radial he latter being the maximum misfit str V. Summary and Conclusions (1) Adiabatic Sliding If we assume that the work done by sliding friction is dissipated La-monazite is compatible with mullite, YAG, Zro2, and entirely by uniform adiabatic heating in a zone of deformed AlO. The interfaces between La-monazite and these materials are sufficiently weak to allow debonding when a crack ap monazite adjacent to the plane of sliding, the temperature rise is proaches the interface from within the monazite. This occurs even in the presence of substantial residual compressive △T= stresses normal to the interface. as in the case of the mullite fiber in an alumina matrix where T, is the sliding friction stress, 8 is the sliding displacement, All the monazite-coated fibers in this study(single crystal h is the thickness of the deformation zone, and p and co are the mullite and alumina, eutectic Al,O3/YAG, and Al,O3 ZrO2) density and specific heat of the monazite. For the sliding experi- underwent sliding in single fiber pushout experiments. Sliding ment corresponding to Fig. 6, the measured parametes o and Cp are T. o occurred along a single interfacial debond when the displace- 200 MPa, 8=5 Hm and h.2 lents were small and/or the fiber surfaces were relatively 500 J(kg K),Eq(A-D)gives AT= 2000oC smooth. At larger displacements, the eutectic fibers, which had An alternative estimate based on incremental sliding of individ rougher interfaces than the single crystal fibers, caused exten- ual asperities, as depicted in Fig Al, gives the temperature rise as sive damage in the ApoA coating adjacent to the fiber. The HAa mullite fibers, which had smooth surfaces but large oscillations △T= diameter, caused deformation through the entire thickness of the coating in regions of large misfit strain. Damage mecha- where H is the hardness of the monazite, A is the cross-sectional nisms included fracture, dislocation plasticity, and occasional area of the asperity and A, is the cross-sectional area of the plastic twinning. The fibers were undamaged, as might be expecte deformation zone created by the asperity as it slides(the sliding given their higher hardnesses. The relative softness of La- force acting on the asperity being set equal to HA ) If we take H monazite, resulting from its ability to deform plastically at low as the room temperature hardness of monazite (-5 GPa)and temperatures, may be critical for use as a composite AAa≈2( from fig.6),Eq(A-2) gIves△T=1000° terface Both of these estimates are subject to considerable uncertainty 3. TEM observations showed densely packed fine crystallites of (a factor of-2) associated with the parameters h and A /A, as well nazite in the most heavily deformed regions, resembling as the assumption of uniform heating within the zone. Neverthe recrystallized microstructures. Several analyses indicated that the sliding velocity is sufficiently large to cause adiabatic condi- stick-slip motion or cataclastic flow caused large increases in tion local sliding velocities and deformation rates. The detailed mechanisms responsible for this microstructure, which is un- (2) Estimated Sliding Velocity and Transient Heating Effects usual for such a refractory material at low temperature, have not The time in transient heat conduction problems always appears been identified. However, a parallel exists in the recrystallize- normalized by the characteristic time,T tion from radiation damage at much lower temperatures in a-monazite than in other minerals paF (A-3) conductivity and d is a characteristic Appendix a diffusion distar fiber sliding problem, d is the depth of the deformation zone and the conditions are close to adiabatic only if the time. taken to heat the deformation zone is small compared with T. If we assume that heat is conducted only into the Estimates of heating from Fiber sliding monazite (k≈2WmK) t is -10-7 s for a zone of 0.2 um. The time th is given by In= 8/v, where 8 is the sliding Several approaches, based on different tions about distance and y is the sliding velocity dissipation mechanisms, may be taken to estim Although the sliding velocity was not measured in the exp temperature rises during fiber sliding. Some from these ments described in Section Il, a very conservative upper bound alyses are summarized as follows: may be estimated. The experiments involved loading the indenter
V. Summary and Conclusions La-monazite is compatible with mullite, YAG, ZrO2, and Al2O3. The interfaces between La-monazite and these materials are sufficiently weak to allow debonding when a crack approaches the interface from within the monazite. This occurs even in the presence of substantial residual compressive stresses normal to the interface, as in the case of the mullite fiber in an alumina matrix. All the monazite-coated fibers in this study (single crystal mullite and alumina, eutectic Al2O3/YAG, and Al2O3/ZrO2) underwent sliding in single fiber pushout experiments. Sliding occurred along a single interfacial debond when the displacements were small and/or the fiber surfaces were relatively smooth. At larger displacements, the eutectic fibers, which had rougher interfaces than the single crystal fibers, caused extensive damage in the LaPO4 coating adjacent to the fiber. The mullite fibers, which had smooth surfaces but large oscillations in diameter, caused deformation through the entire thickness of the coating in regions of large misfit strain. Damage mechanisms included fracture, dislocation plasticity, and occasional twinning. The fibers were undamaged, as might be expected given their higher hardnesses. The relative softness of Lamonazite, resulting from its ability to deform plastically at low temperatures, may be critical for its use as a composite interface. TEM observations showed densely packed fine crystallites of monazite in the most heavily deformed regions, resembling recrystallized microstructures. Several analyses indicated that significant frictional heating during sliding was unlikely unless stick-slip motion or cataclastic flow caused large increases in local sliding velocities and deformation rates. The detailed mechanisms responsible for this microstructure, which is unusual for such a refractory material at low temperature, have not been identified. However, a parallel exists in the recrystallization from radiation damage at much lower temperatures in La-monazite than in other minerals. Appendix A Estimates of Heating from Fiber Sliding Several approaches, based on different assumptions about dissipation mechanisms, may be taken to estimate limits on local temperature rises during fiber sliding. Some results from these analyses are summarized as follows: (1) Adiabatic Sliding If we assume that the work done by sliding friction is dissipated entirely by uniform adiabatic heating in a zone of deformed monazite adjacent to the plane of sliding, the temperature rise is T s cph (A-1) where s is the sliding friction stress, is the sliding displacement, h is the thickness of the deformation zone, and and cp are the density and specific heat of the monazite. For the sliding experiment corresponding to Fig. 6, the measured parameters are s 200 MPa, 5 m and h 0.2 m. With 5 g/cm3 and cp 500 J(kgK) 1 , 61 Eq. (A-1) gives T 2000°C. An alternative estimate based on incremental sliding of individual asperities, as depicted in Fig A1, gives the temperature rise as T HAa cpAb (A-2) where H is the hardness of the monazite, Aa is the cross-sectional area of the asperity and Ab is the cross-sectional area of the plastic deformation zone created by the asperity as it slides (the sliding force acting on the asperity being set equal to Aa). If we take H as the room temperature hardness of monazite (5 GPa)1 and Ab/Aa 2 (from Fig. 6), Eq. (A-2) gives T 1000°C. Both of these estimates are subject to considerable uncertainty (a factor of 2) associated with the parameters h and Aa/Ab as well as the assumption of uniform heating within the zone. Nevertheless, they indicate that large local temperature rises could occur if the sliding velocity is sufficiently large to cause adiabatic conditions. (2) Estimated Sliding Velocity and Transient Heating Effects The time in transient heat conduction problems always appears normalized by the characteristic time, : 62 cpd2 k (A-3) where k is the thermal conductivity and d is a characteristic diffusion distance. In the fiber sliding problem, d is the depth of the deformation zone and the conditions are close to adiabatic only if the time, th, taken to heat the deformation zone is small compared with . If we assume that heat is conducted only into the monazite (k 2 W(mK)1 ),61 is 107 s for a zone of depth 0.2 m. The time th is given by th /v, where is the sliding distance and v is the sliding velocity. Although the sliding velocity was not measured in the experiments described in Section II, a very conservative upper bound may be estimated. The experiments involved loading the indenter Table II. Misfit Strains and Stresses Value Sapphire Mullite YAG/Al2O3 Al2O3/ZrO2 Fiber radius (R/m) 50 25 50 50 Microstructural roughness Amplitude, r (m) 0.05 0.05 0.2 0.2 Period, r (m) 22 1 1 r/R 0.001 0.002 0.004 0.004 r (MPa)‡ 200 300 770 730 Fiber diameter fluctuation Amplitude, R (m) 0.5 2.5 0.5 1 Period, R (m) 500 100 1000 400 (R/R) (zmax/R) † 0.0006 0.03 0.0003 0.0015 R (MPa)‡ 120 4500 60 270 † zmax is the maximum sliding displacement (10 m) ‡ Nominal radial misfit stresses intended only as rough guide for comparisons: stresses calculated as in Table 1 but with radial misfit strains r/R and (R/R) (zmax/R), the latter being the maximum misfit strain for sinusoidal diameter fluctuation (zmax R). 314 Journal of the American Ceramic Society—Davis et al. Vol. 86, No. 2