J. Am. Ceram. Soc., 92 [1] 152–160 (2009) DOI: 10.1111/j.1551-2916.2008.02813.x r 2008 The American Ceramic Society
Journal
Indentation Damage and Residual Stress Field in Alumina-Y2O3-Stabilized Zirconia Composites Carmen Baudı´ n,w,z Jonas Gurauskis,y Antonio Javier Sa´nchez-Herencia,z and Victor M. Oreray z
Instituto de Cera´mica y Vidrio, CSIC, E-28049, Madrid, Spain
y
Instituto de Ciencia de Materiales de Arago´n, CSIC- Universidad de Zaragoza, Zaragoza, Spain
range of toughness values, and data on zirconia transformation levels in fractured samples are contradictory. As stated in a recent paper by Quinn and Bradt,11 the indentation method involves a combination of both the crack initiation and the arrest capabilities of the material and, as a consequence, toughness evaluation from indentation should be carried out with great precaution if not totally rejected. Nevertheless, even though the use of simple indentation techniques might not be valid for unambiguous toughness determinations, indentation experiments can give extremely interesting information about crack forming and arrest phenomena under localized damage, which is crucial for the further use of alumina–YTZP composites in components subjected to wear. In this work, the complete characterization of indentation strength-tested specimens (ISBM) was performed. Fracture features, residual stresses, and zirconia transformation are studied in indentation strength specimens of alumina-3YTZP ceramics in order to analyze the extent of the indentation damage in the bulk of the specimens. The relative volume fraction of the transformed monoclinic ZrO2 phase was determined by micro-Raman spectroscopy, and the microscopic strain state was studied by the piezospectroscopic technique using experimental methods similar to those described elsewhere.12
Fracture features, residual stresses, and zirconia transformation are studied in indentation strength specimens of alumina-Y2O3stabilized zirconia (3% mol of Y2O3, 3YTZP) ceramics in order to analyze the extension of the indentation damage in the bulk of the specimens. Two compositions, 5 vol% 3YTZP (A5) and 40 vol% 3YTZP (A40), have been prepared by stacking tape-casted tapes and sintering. After indentation with loads ranging from 50 to 300 N, samples were fractured in four-point bending and the fracture surfaces were characterized by scanning electron microscopy. Raman and piezospectroscopic techniques were used to determine the monoclinic zirconia fraction and the residual stresses through the fracture surfaces. In the A5 composition, the indentation damage morphology was clearly half-penny, whereas the A40 composition presented Palmqvist crack formation. Zirconia transformation was only observed in the plastically deformed zones underneath the imprints whereas there were significant residual compressive stresses outside the plastic zones due to the indentation damage. The intensity of this residual compressive field was dependent on the level of zirconia transformation due to indentation damage because zirconia transformation induced tensile stress fields superimposed on the compressive stresses.
I. Introduction
A
II. Experimental Procedure
LUMINA-Y2O3-stabilized
zirconia (3 mol% of Y2O3, 3YTZP) composites with improved mechanical behavior compared with that of monophase alumina have been reported by different authors.1–9 These materials combine the outstanding properties of alumina, i.e.: hardness, chemical inertness, and wear resistance, with improved fracture resistance. Therefore, they could be used extensively in machining and cutting applications as well as in fabricating biomedical implants.10 Transformation of tetragonal zirconia to monoclinic in the stress field of the propagating crack is assumed to be the main toughening mechanism. Additionally, the compressive residual stresses developed in the alumina matrix due to thermal expansion mismatch between Y2O3-stabilized zirconia and alumina could also contribute to toughening. Moreover, the refinement of the alumina microstructure due to the presence of zirconia can also be considered as being partially responsible for the improved strength of the composite. In spite of the large number of studies conducted to date on the mechanical behavior of composites containing 5–50 vol% of fine-grained 3YTZP (dV50 0.3–0.4 mm),1–5,7,8 reported mechanical properties are not conclusive. Different authors and even different techniques used by the same authors result in a wide
(1) Sample Preparation Two compositions with different 3YTZP zirconia amounts (5 and 40 vol%) have been chosen in order to scan the compositional range of interest for alumina-3YTZP composites. The specimens were fabricated using high-purity a-Al2O3 and ZrO2 stabilized with 3 mol% of Y2O3 (3YTZP) powders following a procedure described elsewhere.13,14 The starting powders of a-Al2O3 (Ceralox HPA 0.5, Sasol, Tucson, Arizona), with a mean particle size of 0.35 mm and a specific surface area of 9.5 m2/g, and 3YTZP (TZ-3YS, Tosoh, Japan), with a mean particle size of 0.4 mm and a specific surface area of 6.7 m2/g, were mixed in proportions Al2O315 vol% 3YTZP (A5) and Al2O3140 vol% 3YTZP (A40). Stable slurries were prepared by ball milling with alumina balls for 4 h, using deionized water as dispersing media and a polyelectrolyte (Dolapix CE-64, Zschimmer & Schwarz, Lahnstein, Germany) as a dispersant. A water-based polymeric emulsion Mowilith DM 765 E (Celanese, Tarragona, Spain), with a Tg of 61C and solid content 50 vol%, particle size 0.05–0.15 mm, was used as the binder emulsion. The slurries were tape cast on a polypropylene film using a moving tape-casting device with two doctor blades (laboratorydeveloped device)13 using a 10 mm/s casting velocity and a 500 mm gap height between the blades and the carrier film. After drying in air (24 h at 251C and 48 h at 601C), the tapes were cut in the shape of disks (diameter + green 5 60 mm pieces), dipped in distilled water for 1 min, and stacked sequentially to form monolithic pieces using seven tapes and applying a gluing agent (5 wt% dilution in distilled water of the binder) with a paintbrush on the contact surfaces of consecutive tapes. The
G. Pharr—contributing editor
Manuscript No. 24888. Received June 23, 2008; approved October 3, 2008. This work was supported by Ministerio de Ciencia y Tecnologı´a (Spain) through the projects MEC MAT2006-13480 and MAT2006-13005-C03-01. w Author to whom correspondence should be addressed. e-mail:
[email protected]
152
January 2009
Indentation Damage and Residual Stress Field in Alumina-3YTZP
stacked pieces were placed between two sheets of polypropylene film and pressed at 18 MPa at a load frame displacement rate of 0.05 mm/min, using steel compression plates in a universal testing machine (Microtest, Madrid, Spain). The obtained pieces were cut into bars using a metal disk (Dremel Multi, Chicago, Illinois) and the surfaces were smoothed using SiC paper (600 Grit, Buehler, Lake Bluff, Illinois) before sintering. Sintering was carried out in an electrical furnace (AGNI, Aachen, Germany) using a two-step cycle (11C/min up to 6001C with a dwell time of 30 min for the binder burn and 51C/min up to 15501C with a dwell time of 2 h). The sintered bars were machined to obtain the test samples with the final geometry (50 mm 4 mm 3 mm).
(2) Sample Characterization Three Vickers indentations, each 2.5 mm apart, were made in the central part of the tensile surface of the bars, previously polished with a diamond paste (Buehler) of 6 and 3 mm. The two diagonals of the imprints were oriented parallel and normal to the major axis of the beams. The indentations were performed using a specially developed electromechanical indentation device (Microtest15) with a diamond Vickers pyramid (angle 1361 and area–depth ratio A 5 24.5h2) fixed together with the load cell at the loading column, in a controlled displacement mode at 0.01 mm/s up to maximum load (10–300 N) with a holding time of 10 s. The indentation imprints and cracks produced were measured within the time period of 15 min using reflected-light microscopy (Model HP 1, Carl Zeiss, Oberko¨chen and Jena GmbH, Jena, Germany) with a graduated ocular (with a precision between grades of 3 mm). The top surfaces of the bars were observed in a field emission scanning electron microscope (FE-SEM, Hitachi, S-4700, Tokyo, Japan) in the ‘‘as-indented’’ state as well as after sequential polishing with diamond (6 and 3 mm, Buehler). The indented samples were tested in flexure in a steel fourpoint bending device (20–40 mm) at 0.05 mm/min in a universal testing machine (Microtest). Fracture surfaces were observed by FE-SEM. Two indented samples were tested in flexure for each load and observed. Quantitative data relative to the surface indentation damage correspond to six indentations for each load. Raman spectroscopy and piezospectroscopic analyses were perfomed using an optical spectrometer (Model XY, DILOR, Lille, France). The measurements were performed along lines perpendicular to the indented top surfaces and passing through the apex of the Vickers imprints. The analyzed points were located 10 or 20 mm apart, depending on the specimen and the zone. Raman spectra and ruby luminescence were excited with the 514.5 mm line of an Ar1-ion laser and collected at room temperature in a backscattering geometry using a triple 0.5 m monochromator with a CCD detector coupled to an optical microscope with a spectral resolution of 0.3 cm1. A volume of about 2 mm diameter and 6 mm in depth can be analyzed using a 50 objective lens and a 400 mm field aperture in the image focal plane. The Raman peak positions were corrected for instrumental errors using the 520 cm1 Si Raman peak as the frequency standard. The luminescence bands were corrected against the 14431 cm1 line of a neon discharge lamp. Quantitative phase analysis in the monoclinic/tetragonal two-phase zirconia with spatial resolution can be performed using micro-Raman spectroscopy. The procedure rests on the differences in the Raman mode position and intensities for both phases, which can be used to determined the relative amounts of each phase.16,17 Quantitative data as a function of the distance to the indented surface are reported. The volume phase fraction of monoclinic m-ZrO2 defined as: Vf ðmÞ ¼
Vm Vt
(1,)
with Vm and Vt being the volume of m-ZrO2 and total ZrO2, respectively, can be calculated from the intensities of the 180 and
153
190 cm1 Raman peaks of m-ZrO2 and the 150 cm1 Raman peak of tetragonal t-ZrO2 using the method of Kim, Hahn, and Han.16 First, the Raman intensity monoclinic ratio Xm defined by: Xm ¼
Im ð180Þ þ Im ð190Þ Im ð180Þ þ Im ð190Þ þ It ð150Þ
(2)
is determined assuming a linear baseline background and separating the Raman spectra into their component bands using Lorentzian band shapes. The volume fraction is then obtained from the calibration curve: Vf ðmÞ ¼
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 0:13 0:56 0:19 Xm 1:01
(3)
In the present case, the minimum monoclinic fraction that can be detected by this method is about 0.01. The hydrostatic component of the alumina stress tensor was locally determined by means of the Cr31 luminescence of residual chromium impurities using piezospectroscopic methods. At 297 K, the unstressed ruby luminescence spectrum consists of the R1 and R2 lines, at 14403 and 14433 cm1, and half-widths of 11.25 and 8.8 cm1, respectively, associated with the E(2E)-4A2 and 2A(2E)-4A2 transitions of the Cr31(d3), respectively. It is well known that the peak positions of the R lines depend strongly on the Cr31 environment, and hence on the strain field in the alumina phase. Therefore, the relationship between the R-line shifts and stress has been extensively used for pressure calibration in high-pressure devices as well as to determine with a high spatial resolution the residual stresses in composites containing Al2O3.18,19 In addition, it is well known that the R2-line position is insensitive to the nonhydrostatic strain components.20 Consequently, the peak position of the R2 band is an excellent parameter to determine hydrostatic stresses in randomly oriented ceramics. R2-line shifts Dn were determined with respect to that of an Al2O3:Cr31 (0.11 wt% Cr2O3) single crystal, which, at 295 K, appears at 14433 cm1. An analysis of the band shapes was performed in order to obtain quantitative values of the peak positions, nc, and the full-widths at half-maximum wL. The shape of the bands was approached by two single Lorentzian curves, y(n) 5 y0 L(n–nc, wL), leaving as fitting parameters nc and wL. The hydrostatic component sA h (in GPa) of the residual stresses in the alumina phase can be calculated from the R2-line peak shift Dn2 (in cm1) between the stressed and the unstressed ruby21 using the scalar relationship: sA h ¼
Dn2 7:61
(4)
III. Results and Discussion (1) Indentation Damage and Phase Quantification The complete microstructural characterization of the ‘‘as-fabricated’’ materials was reported elsewhere.13,14 Both materials were dense (99% of the theoretical density) and homogeneous and composed of two phases: a-alumina (ASTM 42-1468) and tetragonal zirconia t-ZrO2 (ASTM 83-113). Near the indentation imprint, t-ZrO2 transformed partially into the monoclinic phase (m-ZrO2) as described below. In Table I, the quantitative parameters of the indentation imprints and cracks are summarized. For the sake of comparison, the observed surface lengths, 2c0, which, for Palmqvist geometry,22,23 are equal to the addition of the lengths, l, of both Palmqvist radial cracks plus the diagonal of the imprints, 2a, are reported for both compositions. The indentation cracks
154
Journal of the American Ceramic Society—Baudı´n et al.
Table I. Geometrical Parameters of the Vickers Indentations (Fig. 1) as a Function of the Load Compound denomination
A5
A40
Indentation load (N)
Imprint size 2a (mm)
Crack size 2c0 (mm)
Core diameter (mm)
10 50 100 200 300 10 50 100 200 300
3474 7374 10374 14674 18374 3774 7774 11074 15374 19374
7074 21874 36274 57274 74574 6374 17974 29974 45374 5927
30 50 80 100
Values for the core diameters are approximate as observed in the fracture surfaces.
obtained from the load of 10 N showed insufficient crack formation and were thus excluded from further analysis. Figures 1–5 summarize some of the main features present in the indented samples. In agreement with other authors,24 no differentiated classical fracture markings, mirror, mist hackle, or branching could be observed in any of the fracture surfaces of the A5 and A40 specimens at a low magnification (Fig. 1). In all cases, the local top surface damage left by the indentation was distinguished as a triangular hole in the surface. Nevertheless, the fracture features were better observed with the surfaces slightly tilted relative to the electron beam and, therefore, in
Vol. 92, No. 1
the FE-SEM images the top surface damage is observed. For each composition, A5 and A40, the fracture patterns were similar for the whole range of indentation loads (50–300 N), whereas specific features were different as a function of the composition. Basically, the characteristics observed in the A5 specimens were those of the bending fracture from half-penny cracks, whereas those shown by the A40 composite were typical of Palmqvist geometry, as discussed below. Underneath the indentation imprints and containing them, half-penny areas of irregular bumpiness were differentiated in the A5 specimens (Fig. 1(a)). These areas, which had radii similar to the half-diagonals of the indentation imprints at the top surfaces, a, corresponded to the fracture of the hemispherical plastically deformed zones induced by the indenter during loading. The plastic zones of A5 specimens were contained within relatively flat half-penny regions of radii similar to those of the original radial cracks measured at the top surfaces, c0 (Fig. 1(a), Table I), so that these regions could be associated with the halfpenny cracks induced by indentation. Close to their limits, fracture became more tortuous and showed the classical striations due to crack branching. Additional perpendicular lines corresponding to the orthogonal cracks were frequently observed. Even though no chipping was observed in any of the top surfaces of the indented specimens, lateral cracks25 were observed parallel to the top surfaces. No fracture feature permitted the identification of the half-penny-shaped cracks, of radius cf, that would develop by stable propagation of the radial cracks on flexural loading before fracture.26 On the contrary, characteristic crack branching usually associated with fast fracture and mostly
Fig. 1. Scanning electron micrographs of fracture surfaces of indentation-strength specimens. The diameters 2a (white arrows) and 2c0 (black arrows) from Table I and the distances from the top surface characterizing the plastic zone sizes are indicated. The features of the local surface damage left by the indentation are observed in Fig. 1(a). In Figs. 1(b) and (c) the images of the local surface damage are the shiny triangles. (a) Specimen A5 indented with a load of 100 N. Characteristic half-penny and lateral cracks are observed. (b) Specimen A40 indented with a load of 200 N. Palmqvist type and lateral cracks are observed. (c) Specimen A40 indented with a load of 300 N. Palmqvist type and lateral cracks are observed.
January 2009
Indentation Damage and Residual Stress Field in Alumina-3YTZP
intergranular fracture was observed surrounding the radial cracks in A5 (Figs. 1(a) and 2(c)). The plastic zones of the A5 specimens (Fig. 2(a)) were very tortuous, with intergranular fracture and microcracks perpendicular to the fracture surface revealing intergranular microcracking. Then, predominant transgranular fracture occurred from the limits of the plastic zones to the limits of the half-penny cracks, where a sharp change into intergranular fracture was observed (Fig. 2(c)). The highly compressive local stresses imposed by the indenter during loading might lead to high shear and/or tensile stresses at grain boundaries due to grain boundary strain mismatch. Grain boundary stresses would be partially released by plastic deformation on loading and by microcrack formation on unloading and the residual ones would add to the externally imposed tensile stresses during flexure testing. Both phenomena are responsible for microcracking at grain boundaries and intergranular fracture, as observed in the half-penny plastic zones of A5 specimens (Fig. 2(a)). The flat fracture observed in the initial half-penny crack of radius c0 (Figs. 2(b) and (c)) is characteristic of low-energy propagating cracks, which can be explained by the fact that the half-penny cracks are formed on unloading, due to the stresses originating from the deformation mismatch between the plastic zone and the remaining material, with no further energy supply. For the A40 material (Figs. 1(b) and (c)), quasicircular areas of irregular bumpiness were found underneath the indenter imprint; their diameter increased with the indentation load (Table I). These areas were surrounded by relatively flat surfaces, similar to those observed when two independent radial cracks join together during unloading after the indentation or
155
flexural fracture in monophase Y2O3-stabilized ZrO2.27–29 In these cases, the cracks cannot traverse a semispherical zone below the indent, which corresponds to the plastically deformed zone for Palmqvist cracks, called the core region. As occurred in the A5 specimens, additional lateral cracks were observed in the fracture surfaces even though no chipping was present in the indented top surfaces. The geometry observed in the A40 material was different from the half-penny one observed by us in A5 and by Casellas et al.3 in specimens containing only 30 vol% of 3YTZP, which would suggest that the shape of the indentation cracks changes with the amount of second phase in alumina-3YTZP composites. In some A40 composition specimens, fracture lines were observed below the core zones, which would correspond to ledges on the fracture surface, resulting from the propagation and linkup of the two radial cracks during fracture when the cracks were not coplanar (Figs. 1(b) and (c)). The radial cracks were not easily distinguished, just for the relatively flat zones of widths similar to the values of c0a and the progressive crack branching for larger distances (Table I, Figs. 1(b) and (c)). Disconnected radial cracks, as shown in Fig. 3, were observed for the whole range of indentation loads in sequentially polished top surfaces, in agreement with the radial crack geometry. The diameter of the core regions in A40 specimens (Figs. 1(b) and (c)) increased with the indentation load, as described above. Two different fracture modes as a function of location were found inside the cores (Fig. 4). Through a thin layer just underneath the triangular surface damage (Fig. 4(a)), fracture was rather flat and the crack traversed highly deformed alumina and zirconia grains or went along the extremely damaged grain boundaries. These observations reveal that the inner core
Fig. 2. Characteristic fracture modes in the A5 specimens. Scanning electron micrographs of fracture surfaces of indentation-strength specimens. (a) Detail of the plastic zone. Intergranular fracture and grain boundary microcracking are observed (arrowed). Specimen indented using a load of 50 N. (b) Detail of the half-penny zone. Flat and predominant transgranular fracture is observed. Specimen indented using a load of 100 N. (c) General view of the transition between the half-penny zone and the remainder of the sample (arrowed). A change from predominant transgranular to predominant intergranular fractures is observed. Specimen indented using a load of 300 N.
156
Journal of the American Ceramic Society—Baudı´n et al.
Fig. 3. Characteristic indentation imprint and cracks on a sequentially polished surface of an A40 specimen showing radial crack features. Crack traces detached from the remnant imprint vertices are observed. Scanning electron micrograph of a specimen indented with a load of 100 N.
regions in A40 specimens were regions of extreme plastic deformation, as observed by transmission electron microscopy in 4YPSZ.29 Inside the core, fracture became progressively rougher for larger distances from the indentation apex (Fig. 4(b)). Fracture occurred mostly along the alumina–alumina grain boundaries or traversing zirconia particles, which were observed inside parallel
Vol. 92, No. 1
fracture lines similar to those associated with fracture of twinned m-ZrO2. Very few grains inside this part of the core showed deformed boundaries, and intergranular fracture of the alumina grains revealed flat grain boundaries with sharp edges. Mixed trans/intergranular fracture was observed out of the core regions in the A40 specimens; the only special feature that differentiated the flat fracture zones corresponding to the radial cracks (Figs. 5(a) and (b)) from the rest of the specimen was the progressive branching of the cracks to form striations from the limits of the radial cracks (Figs. 1(b) and (c) and 5(c)). The zirconia grains traversed by the crack (Fig. 5(b)) had an internal substructure similar to that shown by other authors for untransformed YTZP materials,30 most probably due to the different domains present in the tetragonal zirconia grains. As a difference from the core region, the alumina grains presented predominantly transgranular fracture (Fig. 5(b)). As occurred for A5, no special features indicating the stable crack propagation of the indentation-induced cracks during flexure could be distinguished in any of the A40 fracture surfaces, as the low-energy flat fracture of the radial cracks changed into a typical fast fracture. Conversely, stable crack propagation of Palmqvist cracks in bending was observed by Dransmann et al.31 in the polished surfaces of 3YTZP ceramics. The lack of stable propagation found in this work might be due to differences in the experimental procedure used, as the abovementioned authors loaded the specimens in a step wise manner, whereas the characteristic high loading rate for strength specimens was used here. The same explanation would apply to the A5 specimens.
Fig. 4. Fracture modes in the plastic zone of the A40 indentation strength specimens. Scanning electron micrographs of fracture surfaces. Alumina (dark) and zirconia (clear) are distinguished in b–c. (a) General view of the plastic zone. A thin layer showing transgranular fracture underneath the surface imprint is differentiated (arrows) from the near to spherical core. (b) Core region. Mixed trans/intergranular fracture. (c) Detail of the Fig. 4(b). Transgranular fracture occurs through the zirconia grains (clearer) located at the grain boundaries of the alumina grains. Zirconia grains show inside parallel lines similar to those associated to fracture of twinned m-ZrO2.
January 2009
Indentation Damage and Residual Stress Field in Alumina-3YTZP
157
Fig. 5. Fracture modes outside the core area in A40 indentation strength specimens. Scanning electron micrographs of fracture surfaces. (a) General view of the radial crack. Predominant flat fracture is observed. (b) Detail of the radial crack. Zirconia grains (clear) show the characteristic irregular fracture of YTZP and alumina grains show cleavage planes. (c) General view of the fracture surface far from the radial crack. Crack branching is observed.
Characteristic micro-Raman spectra were recorded in the fracture surfaces along lines perpendicular to the indented surfaces of the specimens and passing through the apex of the indentations (Fig. 1). The five broad bands associated with the t-ZrO2 Raman modes and two narrow peaks at 378 and 418 cm1 due to the sapphire phase31 were clearly distinguished through the whole fracture surfaces. Moreover, there were regions, close to the indented surfaces, where nine of the 18 symmetry-allowed m-ZrO2 Raman-active modes32 were detected indicating that ZrO2 t-m transformation had occurred. It has to be noted that the positions of the Raman peaks differed slightly from those reported for unstressed compounds, in agreement with the piezospectroscopy results discussed below. Characteristic plots of the transformed volume fraction (Vf(m)) as a function of the distance from the indented surface are shown in Fig. 6. For both materials, A5 and A40, and for all indentation loads, the transformed fractions were significant (Vf(m)40.05) at the indentation apex. For A5 specimens, Vf(m) was practically constant inside the half-penny zones of radius ‘‘a’’ and decreased sharply for longer distances ( 10–20 mm apart, Fig. 6(a)). For A40, the transformed fractions increased with the distance to the apex up to maximum values that were found at the centers of the cores (depth from the surfaces 65 and 75 mm for indentation loads 200 and 300 N, respectively, Figs. 1(b) and (c) and 6(b)). Then, the m-ZrO2 fraction decreased gradually across the rest of the cores and close to them. For instance, some monoclinic zirconia was found at a distance of 50 mm of the limit of the core for a 300 N load as shown in Fig. 6(b). As a summary, for both materials and all the loads investigated, significant transformation took place only inside or close to the plastic zones located underneath the indenter imprints. Maximum values ( 20%) were reached close to the limits of the
half-penny zones for A5 and at the center of the cores for A40, corresponding to the morphology of the zirconia particles described above. The lack of transformed zirconia for larger distances from the surface reveals that no zirconia transformation can be associated with cracking in the alumina-3YTZP composites studied. Highly transformed indentation imprint regions and nonsignificant transformation associated with fracture have also been detected by Nomarski interferometry28 on the top surfaces and by Raman spectroscopy27 on the fracture surfaces of indented specimens of fine-grained monophase 3YTZP. The results of this work do not contradict some of the data for zirconia transformation in alumina-3YTZP composites discussed in the introduction, as very little or no transformed fraction was detected by X-ray diffraction (XRD) in the SENB and ISBM fracture surfaces of fine-grained alumina-3YTZP composites.2,3 On the contrary, Magnani and Brillante1 reported very large transformed fractions around indentation cracks, which could be due to the fact that they performed Raman spectroscopy on indented surfaces along lines that were very close to the apex of the imprints. Indentation-induced transformation occurs in Y2O3-stabilized zirconia because of the high shear stresses developed under the indenter. The transformation of zirconia derives from a compromise between the high local compressive loads that develop under the indenter, which facilitate the shear component of the transformation strain, and the capability for the dilatational strain to occur.
(2) Residual Stresses The hydrostatic components sA h of the residual stresses in the alumina phase calculated using the Eq. (4) as a function of the
158
(a)
0.15
Vf of m - ZrO2 Calculated σh Measured σh
–300
Vf
0.10
–200 –150 –100
0.05
–50
Hydrostatic Stress (MPa)
0.20
–250
0.00 0 0
50
100 150 200 Distance to Ind. Apex (μm)
250
–900
0.25 (b) Vf of m - ZrO2 Calculated σh
0.15
Measured σh
–700 –600
0.10
–500
0.05
–400 –300
0.00 0
50
Hydrostatic Stress (MPa)
–800
0.20
Vf
Vol. 92, No. 1
Journal of the American Ceramic Society—Baudı´n et al.
–200 100 150 200 350 400 450 Distance to indent. apex (μm)
Fig. 6. Monoclinic zirconia volume fraction and alumina hydrostatic stress determined in fracture surfaces of indented specimens. Measurements were done along lines perpendicular to the indented top surfaces and passing through the indentation apex at different distances from the indented top surface. Data are plotted as a function of the distance to the top indented surface. (a) A5 specimen indented with a load of 100 N. (b) A40 specimen indented with a load of 300 N.
distance to the indentation tip are included in Fig. 6. For both materials and all analyzed loads, values were significantly different from zero throughout the whole fracture surfaces and corresponded to alumina being under compressive residual stresses. As occurred with the zirconia-transformed fractions, significant residual stresses were always detected inside or close to the half-penny (A5) and core (A40) zones. Nevertheless, the dependence of measured stress on the distance to the apex was different from that of the zirconia-transformed fraction. For A5 specimens, residual stresses decreased gradually from the maximum values, reached at relatively short distances from the apex of the indenter ( 10–20 mm), and stabilized out of the half-penny cracks of radius c0. Maximum values were also found at short distances from the apex in A40 specimens, close to the center of the cores; then, they decreased through the cores, reaching the minimum values at their limits ( 105 and 125 mm for 200 and 300 N, respectively, Figs. 1(b) and (c)). From the limits of the cores, residual stresses increased again and then decreased to reach a constant value far from the indented regions that was similar for all indentation loads. As it is generally described for the indentation process, assuming that the plastic deformation associated with the residual impression is volume conserving, the indented volume has to be accommodated by an elastic deformation around the damaged zone. The volume to be accommodated can be increased by the volume increase due to the t-m ZrO2 transformation. The elastically deformed material, in tension, will induce hydrostatic compressive stresses in the damaged zone that, as suggested by Pajares et al.33 would not decrease sharply, but progressively, with the distance to the damaged zone. The indentation residual stress field might be decreased by the lateral
cracks that develop on unloading.25 During ISBM testing, crack growth would occur due the combination of the indentation residual stress field and the stress field of the applied load.26 Following this model, when catastrophic failure occurs, the indentation stress fields at the surfaces of the cracks have to be significantly relaxed, at least in the direction perpendicular to the fracture surfaces. Therefore, the residual stresses present just underneath the fracture surfaces of ISBM specimens and determined by piezospectroscopy do not exactly represent the hydrostatic stresses under which the indentation was made. Nevertheless, even though the actual stress values can be slightly different from the determined ones, the stress distribution as a function of distance to the indented surface as well as the different factors acting on this distribution will not change. The residual stresses present in the tested specimens should be the result of the different factors acting the thermal expansion mismatch between zirconia and alumina (on average, from 01 to 10001C, aA 8.2 106 and aZ 10.6 106),34 zirconia transformation (Fig. 6), and the residual damage originated by the Vickers indentation. Far from the indentation tip, where neither indentation damage nor zirconia transformation occurred, the constant residual stresses (Table II) should be of a purely thermoelastic origin. The effect of zirconia transformation on the residual stress of zirconia-toughened alumina has been studied in the past.35–37 Gregori et al.36 demonstrated a linear correlation between the residual stresses obtained by piezospectroscopy and the monoclinic volume fraction determined by XRD regardless of whether zirconia transformation is of mechanical or thermal origin. In the present case (Fig. 6), there was not a clear correlation between monoclinic fraction and the measured residual stresses due to the effect of Vickers indentation. The different contributions to the residual stress field in the Al2O3 phase can be expressed as follows: A A A sA h ¼ sh;ind þ sh;tm þ sh;the ¼ K eðTÞ
(5)
where K is an effective elastic modulus that depends on the elastic constants and volume fractions of the component phases as well as on the sample topology.34 The stress due to the transformation strain is given by Gregori et al.36 sA h;tm ¼ K Xm etm
(6)
where Xm 5 Vf/(1Vf) and etm are the monoclinic fraction and transformation strain, respectively. The stress caused by the thermal mismatch is given by sA h;the ¼ K ðT0 TÞðaA az Þ
(7)
T0 represents the temperature below which the stresses cannot be relaxed by plastic deformation, XZ,i is the volume fraction of the zirconia i phase (monoclinic or tetragonal), and a is the corresponding temperature-dependent thermal expansion coefficient. Equation 7 has been solved according to the methods and using the parameters and the thermal expansion temperature-dependent coefficients used elsewhere19 The calcu-
Table II. Calculated and Measured Thermoelastic Stresses Hydrostatic Average Stresses sh,th, in the two Studied Ceramics and Effective Elastic Modulus, K of the Present Phases Parameter\compound
K (GPa)
sA h,th (calculated) (MPa) sA h,th (measured) (MPa) sZh,th (calculated) (MPa)
A5
A40
15.96 53.271.5 5075 950
139.4 467715 370730 555
January 2009
Indentation Damage and Residual Stress Field in Alumina-3YTZP
lated values to be compared with the experimental ones are given in Table II. It is important to point out here that in the computation of the residual stress, the possible contribution from the anisotropy of the Al2O3 phase itself, which, in ceramics of this composition, is relatively small,37 is neglected. The agreement between measured and calculated residual stresses is very good, which indicates that neither stress relaxation mechanisms nor macroscopic stresses were active in the as-prepared ceramics. Evidently, far from the plastically deformed region, the stress in the zirconia phase aZh can be evaluated from the static equilibrium condition as: Z fA sA h þ fZ sh ¼ 0
(8)
The obtained results are also given in Table II. Using Eqs. (5) through (7), the measured residual stresses and monoclinic volume fractions from Fig. 6 and taking etm 5 0.016 for the transformation strain,36 the compressive stresses due to indentation sA ind given in Fig. 6 are obtained. These results indicate that the indentation-induced compressive stress field would monotonically decrease with the distance to the indentation apex, as expected and as occurred in the plastic zones of the A5 specimens (Fig. 6(a)) where no transformed zirconia was detected. On the contrary, there was a nonmonotonical variation of the indentation residual stress field in the regions of the limits of the cores of the A40 specimens, in which minimum compressive residual stresses, even lower than the residual thermal-related stresses in the material, coincided with the maximum transformed fractions (e.g. 100–120 mm from the surface, Fig. 6(b)). Moreover, compressive stresses increased from these regions as the transformed fraction decreased, to reach the constant values corresponding to the stresses originating from thermal strain mismatch far from the plastic cores, where no transformed fraction was observed. Therefore, the above-mentioned tensile component responsible for the nonmonotonical variation of residual stresses should originate from the transformation of zirconia. In fact, the associated volume increase would exert tensile stresses in the surrounding material, zirconia and alumina, as reported by Gregori et al.36
IV. Conclusions There are several points that arise from the results of this work. First, it has been clearly demonstrated that there is no toughening associated with zirconia transformation during fracture in the fine-grained alumina-3YTZP materials studied here. Second, it has also been shown that the morphology of indentation fracture in this kind of materials is extremely dependent on the amount of 3YTZP present, which explains the disagreement between the crack morphology reported by other authors3 for alumina130 vol% 3YTZP and that found in this work for alumina 140 vol% 3YTZP. Nevertheless, the most important conclusion reached refers to the extreme dependence of the compressive field present in the specimens on zirconia transformation. In this sense, increasing indentation loads that, in principle, would lead to the increase of the compressive field also produces increased fractions of transformed zirconia that, in turn, lead to the decrease of the compressive field. In fact, the observed minimum in the compressive stresses was more intense as the indentation load increased. The nonmonotonical variation of residual stresses will determine the dependence of the size of the developed cracks on load.
Acknowledgments Jonas Gurauskis acknowledges the financial support provided by CSIC through I3P contract.
References 1
G. Magnani and A. Brillante, ‘‘Effect of the Composition and Sintering Process on Mechanical Properties and Residual Stresses in Zirconia-Alumina Composites,’’ J. Eur. Ceram. Soc., 25 [15] 3383–92 (2005).
2
159
W. H. Tuan, R. Z. Chen, T. C. Wang, C. H. Cheng, and P. S. Kuo, ‘‘Mechanical Properties of Al2O3/ZrO2 Composites,’’ J. Eur. Ceram. Soc., 22 [16] 2827–33 (2002). 3 D. Casellas, I. Rafols, L. Llanes, and M. Anglada, ‘‘Fracture Toughness of Zirconia-Alumina Composites,’’ Int. J. Refrac. Met. Hard. Mater., 17 [1–3] 11–20 (1999). 4 A. Tarlazzi, E. Roncari, P. Pinasco, S. Guicciardi, C. Melandri, and G. de Portu, ‘‘Tribological Behaviour of Al2O3/ZrO2-ZrO2 Laminated Composites,’’ Wear, 244 [1–2] 29–40 (2000). 5 V. M. Sglavo and M. Bertoldi, ‘‘Design and Production of Ceramic Laminates with High Mechanical Resistance and Reliability,’’ Acta Mater., 54 [18] 4929–37 (2006). 6 Y. S. Shin, Y. W. Rhee, and S. J. L. Kang, ‘‘Experimental Evaluation of Toughening Mechanisms in Alumina-Zirconia Composites,’’ J. Am. Ceram. Soc., 82 [5] 1229–32 (1999). 7 M. Szutkowska and M. Boniecki, ‘‘Subcritical Crack Growth in ZirconiaToughened Alumina (ZTA) Ceramics,’’ J. Mater. Process. Technol., 175 [1–3] 416–20 (2006). 8 F. Gutierrez-Mora, D. Singh, N. Chen, K. C. Goretta, J. L. Routbort, S. H. Majumdar, and A. Dominguez-Rodriguez, ‘‘Fracture of Composite Alumina/Yttria-Stabilized Zirconia Joints,’’ J. Eur. Ceram. Soc., 26 [6] 961–5 (2006). 9 A. H. De Aza, J. Chevalier, G. Fantozzi, M. Schehl, and R. Torrecillas, ‘‘SlowCrack-Growth Behavior of Zirconia-Toughened Alumina Ceramics Processed by Different methods,’’ J. Am. Ceram. Soc., 86 [1] 115–20 (2003). 10 B. S. Bal, J. Garino, M. Ries, and M. N. Rahaman, ‘‘A Review of Ceramic Bearing Materials in Total Joint Arthroplasty,’’ Hip Inter., 17 [1] 21–30 (2007). 11 G. D. Quinn and R. C. Bradt, ‘‘On the Vickers Indentation Fracture Toughness Test,’’ J. Am. Ceram. Soc., 90 [3] 673–80 (2007). 12 N. R. Harlan, R. I. Merino, J. I. Pena, A. Larrea, V. M. Orera, C. Gonzalez, P. Poza, and J. Llorca, ‘‘Phase Distribution and Residual Stresses in Melt-Grown Al2O3–ZrO2(Y2O3) Eutectics,’’ J. Am. Ceram. Soc., 85 [8] 2025–32 (2002). 13 J. Gurauskis, A. J. Sanchez-Herencia, and C. Baudin, ‘‘Joining Green Ceramic Tapes Made from Water-Based Slurries by Applying Low Pressures at Ambient Temperature,’’ J. Eur. Ceram. Soc., 25 [15] 3403–11 (2005). 14 J. Gurauskis, A. J. Sanchez-Herencia, and C. Baudin, ‘‘Al2O3/Y-TZP and Y-TZP Materials Fabricated by Stacking Layers Obtained by Aqueous Tape Casting,’’ J. Eur. Ceram. Soc., 26 [8] 1489–96 (2006). 15 S. Bueno and C. Baudin, ‘‘Instrumented Vickers Microindentation of Alumina-Based Materials,’’ J. Mater. Res., 21 [1] 161–73 (2006). 16 B. K. Kim, J. W. Hahn, and K. R. Han, ‘‘Quantitative Phase Analysis in Tetragonal-Rich Tetragonal/Monoclinic Two Phase Zirconia by Raman Spectroscopy,’’ J. Mater. Sci. Lett., 16 [8] 669–71 (1997). 17 D. R. Clarke and F. Adar, ‘‘Measurements of the Crystallographically Transformed Zone Produced by Fracture in Ceramics Containing Tetragonal Zirconia,’’ J. Am. Ceram. Soc., 65, 284–8 (1982). 18 L. C. Ciacchi, G. Gregori, V. Lughi, A. Rossi, and V. Sergo, ‘‘PiezoSpectroscopy: A Materials Science Perspective’’; pp. 243–72, Vol. 2. Research Signpost, Kerala, India, 1999. 19 V. M. Orera, R. Cemborain, R. I. Merino, J. I. Pena, and A. Larrea, ‘‘PiezoSpectroscopy at Low Temperatures: Residual Stresses in Al2O3–ZrO2(Y2O3) Eutectics Measured from 77 to 350 K,’’ Acta Mater., 50 [18] 4677–86 (2002). 20 Y. M. Gupta and X. A. Shen, ‘‘Potential Use of the Ruby R2 Line Shift for Static High-Pressure Calibration,’’ Appl. Phys. Lett., 58 [6] 583–5 (1991). 21 J. He and D. R. Clarke, ‘‘Determination of the Piezospectroscopic Coefficients for Chromium-Doped Sapphire,’’ J. Am. Ceram. Soc., 78 [5] 1347–53 (1995). 22 M. Laugier, ‘‘Palmqvist Crack Extension and the Center-Loaded Penny Crack Analogy,’’ J. Am. Ceram. Soc., 68 [2] C51–2 (1985). 23 S. Palmqvist, ‘‘Occurrence of Crack Formation During Vickers Indentation as a Measure of the Toughness of Hard Metals,’’ Arch. Eisenhuettenwes, 33 [9] 629–33 (1962). 24 R. Morrel, Fractography of Brittle Materials. Measurement Good Practice Guide No. 15. National Physical Laboratory, Teddington, Middlesex, U.K., 1999. 25 B. Lawn and R. Wilshaw, ‘‘Indentation Fracture: Principles and Applications,’’ J. Mater Sci., 10, 1049–81 (1975). 26 P. Chantikul, G. R. Anstis, B. R. Lawn, and D. B. Marshall, ‘‘A CriticalEvaluation of Indentation Techniques for Measuring Fracture-Toughness.2. Strength Method,’’ J. Am. Ceram. Soc., 64 [9] 539–43 (1981). 27 M. S. Kaliszewski, G. Behrens, A. H. Heuer, M. C. Shaw, D. B. Marshall, G. W. Dransmann, R. W. Steinbrech, A. Pajares, F. Guiberteau, F. L. Cumbrera, and A. Dominguez-Rodriguez, ‘‘Indentation Studies on Y2O3-Stabilized ZrO2. 1. Development of Indentation-Induced Cracks,’’ J. Am. Ceram. Soc., 77 [5] 1185–93 (1994). 28 R. F. Cook, L. M. Braun, and W. R. Cannon, ‘‘Trapped Cracks at Indentations. 1. Experiments on Yttria Tetragonal Zirconia Polycrystals,’’ J. Mater. Sci. Lett., 29 [8] 2133–42 (1994). 29 A. Pajares, F. Guiberteau, K. H. Westmacott, and A. Dominguez-Rodriguez, ‘‘TEM Characterization of Indented Polycrystalline Y-PSZ,’’ J. Mater. Sci., 28 [24] 6709–14 (1993). 30 B. J. Hulm and W. J. Evans, ‘‘Evaluation of the Cyclic Fatigue Life of 3-mol%-Yttria-Stabilized Zirconia Bioceramic Using Biaxial Flexion,’’ J. Am. Ceram. Soc., 83 [2] 321–8 (2000). 31 G. W. Dransmann, R. W. Steinbrech, A. Pajares, F. Guiberteau, A. Dominguez-Rodriguez, and A. H. Heuer, ‘‘Indentation Studies on Y2O3-Stabilized ZrO2. 2. Toughness Determination from Stable Growth of Indentation-Induced Cracks,’’ J. Am. Ceram. Soc., 77 [5] 1194–201 (1994). 32 P. Bouvier and G. Lucazeau, ‘‘Raman Spectra and Vibrational Analysis of Nanometric Tetragonal Zirconia Under High Pressure,’’ J. Phys. Chem. Solids, 61 [4] 569–78 (2000).
160 33
Journal of the American Ceramic Society—Baudı´n et al.
A. Pajares, F. Guiberteau, R. W. Steinbrech, and A. Dominguez-Rodriguez, ‘‘Residual-Stresses Around Vickers Indents,’’ Acta Metall. Mater., 43 [10] 3649–59 (1995). 34 D. J. Green, An Introduction to the Mechanical Properties of Ceramics. Cambridge University press, Cambridge, 1998. 35 V. Sergo, D. R. Clarke, and W. Pompe, ‘‘Deformation Bands in CeriaStabilized Tetragonal Zirconia/Alumina: I, Measurement of Internal Stresses,’’ J. Am. Ceram. Soc., 78, 633–40 (1995).
36
Vol. 92, No. 1
G. Gregori, W. Burger, and V. Sergo, ‘‘Piezo-Spectroscopic Analysis of the Residual Stresses in Zirconia-Toughened Alumina Ceramics: The Influence of the Tetragonal-to-Monoclinic Transformation,’’ Mat. Sci. Eng., A271, 401–6 (1999). 37 Q. Ma, W. Pompe, J. D. French, and D. R. Clarke, ‘‘Residual-Stresses in Al2O3-ZrO2 Composites—a Test of Stochastic Stress Models,’’ Acta Metall. Mater., 42 [5] 1673–81 (1994). &
