Annealing aged zirconia: Study of surface mechanical properties at the micrometric length scale

Available online at www.sciencedirect.com

ScienceDirect Journal of the European Ceramic Society 35 (2015) 1031–1039

Annealing aged zirconia: Study of surface mechanical properties at the micrometric length scale J.J. Roa a,b,∗ , M. Anglada a,b a

Department of Materials Science and Metallurgy, Universitat Politècnica de Catalunya, Av. Diagonal 647 (ETSEIB), 08028 Barcelona, Spain b Center for Research in NanoEngineering, CrnE, C/ Pascual i Vila, 15, 08028 Barcelona, Spain Received 24 July 2014; received in revised form 6 October 2014; accepted 18 October 2014 Available online 31 October 2014

Abstract The annealing of hydrothermal aged zirconia has been studied by analysing the changes in microstructure and surface mechanical properties in terms of the annealing temperature. In this experimental work, a systematic micro- and nanomechanical study has been conducted in 3 mol% yttria-stabilized tetragonal polycrystalline zirconia (3Y-TZP) aged at 134 ◦ C for 60 h and annealed at 600 ◦ C and 850 ◦ C for 1 h. Advance characterization techniques (micro-Raman, field emission scanning electron microscopy and focused ion beam) have been used to study the near surface microstructural changes induced by these treatments. The mechanical properties and resistance to damage of the near surface are determined by means of nanoindentation and nanoscratch testing. The observed behaviour is discussed in terms of the change in microstructure induced after ageing and annealing. © 2014 Elsevier Ltd. All rights reserved. Keywords: 3Y-TZP; Monoclinic to tetragonal phase transformation; Annealing treatment; Indentation hardness and indentation elastic modulus; Scratch resistance

1. Introduction Zirconia ceramic materials are widely and increasingly used in dentistry due to their outstanding biocompatibility and aesthetics.1 One of these materials that has received increasing attention in recent years is 3 mol% yttria-stabilized tetragonal polycrystalline zirconia (3Y-TZP), which is widely used to fabricate multiunit fixed partial dentures,2,3 inlays2,4,5 and implants,6 due to its excellent mechanical, thermal, and tribological properties, biocompatibility and corrosion resistance.7 However, with the presence of water, 3Y-TZP undergoes a spontaneous unfavourable phase transformation from tetragonal to monoclinic t-m at relatively low temperatures.8,9 The phase transformation starts at the surface and progresses as a surface transformed layer that grows very slowly in thickness.10 ∗

Corresponding author at: Department of Materials Science and Metallurgy, Universitat Politècnica de Catalunya, Av. Diagonal 647 (ETSEIB), 08028 Barcelona, Spain. E-mail address: [email protected] (J.J. Roa). http://dx.doi.org/10.1016/j.jeurceramsoc.2014.10.023 0955-2219/© 2014 Elsevier Ltd. All rights reserved.

The transformation strain (∼8% shear strain and 4–5% volume increase) produces surface roughening,11 grain pull out, macroor even microcracks and loss of surface mechanical properties. This phenomenon is referred to as hydrothermal ageing or low temperature degradation (LTD), and it may be very detrimental for those biomedical implants whose surface integrity is of paramount importance for their functionality.12 Both, t-m transformation and microcracking, are confined into a surface layer with a thickness that depends on the composition, porosity, grain size, temperature of ageing, time and pressure.13 In this paper we address the question of whether the mechanical properties of 3Y-TZP can be recovered by annealing after ageing. It is well known that the t-m transformation can be fully reversed by suitable annealing, but there is a lack of information on whether the mechanical properties of the degraded surface layer are also fully recovered at relatively low annealing temperatures. Since intergranular microcracks are induced during ageing, the objective of the present work is to clarify if annealing is able to rebuild microcracked grain boundaries with the same original strength. Therefore, the present study aims to evaluate

1032

J.J. Roa, M. Anglada / Journal of the European Ceramic Society 35 (2015) 1031–1039

the effect of phase transformation and microcracking into the surface in terms of mechanical and tribological properties at micro- and nanometric length scale.

2. Experimental procedure 2.1. Sample preparation Specimens were processed from zirconia powder stabilized with 3 mol % yttria (TZ-3YSB-E, Tosoh, Tokyo, Japan). The green body was obtained through cold isostatic pressing at 200 MPa, forming rods 10 mm in diameter. The material was sintered for 2 h at 1450 ◦ C, with heating and cooling rate of 3 ◦ C min−1 . The rods were cut in disks 2 mm thick. After sintering, the samples were characterized with respect to their final density, grain size and crystalline phases present. For determining the density of the sintered samples, the Archimedes technique was used with distilled water as the buoyant medium, being 6.03 ± 0.01 g cm−3 for the tetragonal 3Y-TZP. Grain size for the as-sintered specimen was provided by analysis imaging using field emission scanning electron microscopy, resulting in a material with grain size 0.34 ± 0.02 ␮m. Further details on these materials and on the preparation of specimens can be found elsewhere.14,15 Prior to the low temperature degradation (LTD) test, the disc specimens were previously polished with SiC paper and diamond paste finishing with colloidal silica, in an ordinary metallographic polisher. The quality of the finishing was checked by confocal laser scanning microscopy in order to avoid the presence of scratches on the surfaces prior to testing. Finally, all specimens were cleaned ultrasonically with distilled water for 5 min and dried with a pure air. Subsequently, specimens were degraded in an autoclave in the presence of water steam at 134 ◦ C and 0.2 MPa pressure during 60 h (60H). These conditions are roughly equivalent to many years at human body temperature (more than a human lifetime).16,17 The thickness of the degraded layer after 60 h reached values of approximately 10 ␮m. Also, disks of 60H were sectioned and carefully polished in order to measure the volume fraction of monoclinic phase and the elastic modulus at different depths below the surface. The degraded specimens were thermally treated at 600 and 850 ◦ C during 1 h in order to reverse the phase transformation created during the LTD process (from monoclinic to tetragonal phase, m-t). The codes used for the specimens tested in the present work are shown in Table 1.

Table 1 Codes for ageing and annealing treatments and phase composition. Code

Description

Vmonoclinic (%)

AS 60H 60H+600 ◦ C 60H+850 ◦ C

As-sintered Ageing: 60 h, 134 ◦ C, 0.2 MPa As 60H + annealing (600 ◦ C, 1 h) As 60H + annealing (850 ◦ C, 1 h)

<3 86 40 <5

2.2. Structure characterization Raman spectra were collected with a spectrometer JobinYvon LabRam HR 800 coupled to an Olympus optical microscope BXFM with an objective of 100× and to a CCD detector cooled with liquid nitrogen. Raman spectra were recorded using the 532 nm laser wavelength excitation and an acquisition range from 100 up to 3000 cm−1 . The spectrum integration time was 60s, with averaging the recorded spectra over three successive measurements. The measurements for 60H were carried out not only on the surface of the disks, but also in the cross section at different depths. For the as-sintered (AS) and annealed specimens only three measurements were done on the surface. Finally, in order to estimate the monoclinic volume fraction for the different specimen conditions, the following expression was used:18 Vm =

181 + I 190 Im m 181 + I 190 0.32(It147 + It265 ) + Im m

(1)

where Im and It represent the integrated intensities of the monoclinic and tetragonal peaks, respectively. The super-indexes identify the ␮-Raman shift in cm−1 . For all spectra, a linear subtraction of the background was made and, the integrated intensity bands were calculated by fitting Lorentzian distribution functions. 2.3. Mechanical response The mechanical characterization at micro- and nanometric length scale included the evaluation of the effective hardness (H) and elastic modulus (E) through a nanoindenter XP apparatus from MTS. It was equipped with a continuous stiffness measurements (CSM) modulus, allowing a dynamic determination of the mechanical properties during indentation.19 Indentations were organized in a regularly spaced array of 16 imprints (4 by 4) and 2000 nm of maximum displacement into the surface (or until reaching the maximum applied load, i.e. 650 mN), and the results were averaged in order to have statistical signification. The distance between imprints were kept constant to 50 ␮m in order to avoid any overlapping effect. The strain rate was held constant at 005 s−1 . The indenter shape was carefully calibrated for true penetration depths as small as 60 nm by indenting a standard specimen of fused silica of accurately known Young’s modulus (72 GPa). The values of hardness and elastic modulus were directly calculated using the Oliver and Pharr analysis.19,20 The elastic modulus was also determined by indenting the section of the discs at depths intervals of 2 ␮m with a maximum displacement into the surface of 2000 nm. For each position at least three measurements were done. Sliding contact tests at nanometer length scale were carried out by means of a nanoscratch fixture attached to the nanoindenter system refereed above. A Berkovich indenter was employed to scratch the different sets of specimens under constant loads of 50 mN and 100 mN and at a velocity of 10 ␮m/s over a length of 200 ␮m. Two different scans were done in each specimen.

J.J. Roa, M. Anglada / Journal of the European Ceramic Society 35 (2015) 1031–1039

1033

2.4. Damage and fracture assessment Surface damage associated with residual nanoindentation imprints were visualized by field emission scanning electron microscopy (FESEM, JEOL-7001F) at 20 kV and atomic force microscopy (AFM, Dimension D3100 from Bruker), working in tapping mode. All the images were processed with the WSxM software.21 Regarding subsurface damage induced during indentation, it was inspected by means of focused ion beam (FIB). Crosssectioning and microscopy was conducted using a dual beam workstation (Zeiss Neon 40). A thin platinum layer was deposited on the sample prior to milling with the aim of reducing ion-beam damage. A Ga+ ion source was used to mill the surface at a voltage of 30 kV. The final polishing of the cross-section was performed at a current of 200 pA and 30 kV acceleration voltage. 3. Results and discussion 3.1. Microstructure The cross-section microstructure images for the degraded and annealed samples are summarized in Fig. 1. From the subsuperficial observation published in a previous work for the AS specimen,22 no microcracks or monoclinic plates can be appreciated. On the contrary, in the microstructure of 60H (see Fig. 1a), monoclinic plates and microcracks are present, which are created by the volumetric expansion than accompanies tm transformation. These observations are in line with previous works reported elsewhere.23–25 For 60H+600 ◦ C the appearance is similar to 60H with large amount of monoclinic zirconia and the presence of many intergranular microcracks, see Fig. 1b. Very often it is observed that close to the surface (about 3 ␮m in depth) the amount of transformation and microcracking seems to be greater that at deeper regions inside the degraded zone. For 60H+850 ◦ C, the thermal treatment practically reverts all the monoclinic content to tetragonal (see Section 3.2), but still a small number of grains and microcracks can be detected in the first few grains near the surface, see Fig. 1c. 3.2. Raman spectroscopy The Raman spectrum of 3Y-TZP is characterized by double set of peaks at 147 and 265 cm−1 , which are ascribed to the

Fig. 2. Raman spectra obtained from the surface of the disks for all conditions studied. The monoclinic content calculated using Eq. (1) is given in Table 1.

tetragonal phase. Furthermore, a double set of peaks at 181 and 190 cm−1 are ascribed to monoclinic phase. The Raman spectra from the surface of the disks as well as the monoclinic content calculated using Eq. (1) for all the specimens are summarized in Fig. 2 and Table 1, respectively. Because a confocal microscope was used, the amount of monoclinic detected is mainly related to the monoclinic content of shallow surface layer of about 2 ␮m in depth. In addition, the Raman spectra was also measured in the cross-section of the 60H disc; the volume fractions of monoclinic phase from points at depths of 1–3 ␮m from the surface were very similar to the values detected from the surface of the disc. However, the analysis of deeper sections showed that the monoclinic fraction drops rapidly in the depth interval 4–8 ␮m and then slowly until practically disappearing at depths of about 10 ␮m. 3.3. Mechanical properties 3.3.1. Hardness and elastic modulus from nanoindentation Typical load-displacement (P–h) curves generated with the Berkovich indenter into the surface for monotonic loading with 2000 nm of maximum displacement (or until reaching maximum load of 650 mN) are shown in Fig. 3 for AS and 60H, which correspond to specimens with mainly tetragonal and monoclinic phases, respectively. The P–h curve for AS shows the lowest penetration depth under maximum displacement into surface and the lowest residual depth of the indenter after the indenter is completely withdrawn, confirming that this specimen is harder

Fig. 1. Cross-section microstructure view for the specimens of interest: (a) 60H, (b) 60H+600 ◦ C, and (c) 60H+850 ◦ C. The typical cross-section microstructure for AS is shown in previous work.23

1034

J.J. Roa, M. Anglada / Journal of the European Ceramic Society 35 (2015) 1031–1039

Fig. 3. Loading-unloading curves for the AS specimen (tetragonal phase) and for the 60H (monoclinic phase, around 86%).

and stiffer than 60H and pointing out that the mechanical parameters are strongly modified by LTD. No discontinuities during the loading curve can be detected for AS and 60H. The hardness and elastic modulus as a function of the displacement into the surface are shown in Fig. 4a and b, respectively. The hardness (Fig. 4a) is nearly constant for displacements into the surface inside the interval 500–2000 nm. For lower displacements, changes in hardness are difficult to estimate because of the influence of superficial defects, such as roughness, microcracks, and residual stresses, among others. Therefore, unless otherwise stated, the mechanical properties measured in this paper are referred to this displacement interval. The hardness of 60H (∼12 GPa) is clearly lower with respect to AS (∼17 GPa). These values are consistent with several works reported elsewhere.26,27 One main reason for this difference is the effect of the micro-crack distribution close to the surface that appears after hydrothermal degradation in 60H. Another one is a possible intrinsic lower hardness of the monoclinic phase with respect to the tetragonal phase. The hardness of monoclinic 3Y-TZP is not well known since few experimental values exist and they are often obtained in porous monoclinic and microcracked ZrO2 (6.6 GPa),28 or in single crystals which are strongly anisotropic.29 It can be observed that the hardness for 60H increases with deeper displacement into the surface reaching a value close to 14 GPa at about 5000 nm displacement remaining constant until the maximum displacement evaluated (8000 nm). These last high displacements into the surface were carried out with the help of a high load module coupled to Nanoindenter XP using a Berkovich tip indenter. The result suggests that at displacements greater than 2000 nm, the “substrate” made of tetragonal phase, or/and compressive residual stresses related to transformation starts influencing significantly the value of hardness. In view of the microcracking observed close to the surface (Fig. 1a and b for the 60H and 60H+600 ◦ C specimens) and of the relatively large thickness of the degraded layer, one expects that transformation compressive residual stresses close to the monoclinic-tetragonal interphase may be more significant while near the free surface, where higher density of microcracks are

Fig. 4. Mechanical properties evolution for all the specimens of study against the penetration depth. (a) Hardness, and (b) elastic modulus. Where T and M present in these plots denote the tetragonal and monoclinic phase, respectively.

observed, they should at least partially relaxed. Being compressive stresses, their effect would be in the direction of increasing H, but it is difficult to discern its relative importance in the hardness with respect to the contribution of the tetragonal substrate since both should increase hardness. For 60H+600 ◦ C and 60H+850 ◦ C the hardness and elastic modulus are quite similar to 60H and AS, respectively. This is in agreement with the observed removal of most microcracks, monoclinic phase and corresponding residual stresses by the annealing treatment at 850 ◦ C. In 60H+600 ◦ C micro-cracks distribution close to the surface does not practically change with respect 60H as is shown in Fig. 1b and only partial transformation reversion takes place, being the amount of tetragonal phase about 40% (see Table 1). In spite of that, hardness remains practically unchanged by the annealing treatment at 600 ◦ C and it is similar to the 60H specimen. This point out to the micro-crack distribution as the main responsible for keeping hardness similar to the aged specimens. For the relatively small highest displacement into the surface studied in this case (up to 2000 nm) the main contribution to

J.J. Roa, M. Anglada / Journal of the European Ceramic Society 35 (2015) 1031–1039 100

240 230

Monoclinic fraction (%)

80

220 60 Monoclinic fraction Elastic Modulus

40

210 200 190

20 180 0

Elastic modulus (GPa)

hardness is primarily governed by the microcrack distribution. The increase in hardness expected by the transformation reversion and not detected in the present measurements may be reduced by the simultaneous disappearance of residual compressive stresses. Regarding the elastic modulus for AS and 60H+850 ◦ C, the values reported here are very similar and they are in the range 240–250 GPa. These values are slightly greater than the generally accepted value of 210 GPa for 3Y-TZP as several authors reported elsewhere.30 The elastic modulus for 3Y-TZP measured by nanoindentation is close to the values reported by Gaillard et al.27 Furthermore, the indentation elastic modulus presented in Fig. 4b for 60H and 60H+600 ◦ C is in agreement with comparable results (190–199 GPa) of elastic modulus of Eichler et al.31 It can be appreciated that the values and variation of elastic modulus with displacement into the surface for 60H and 60H+600 ◦ C are small and similar, indicating that the annealing treatment performed at 600 ◦ C does not produce any significant change in this property as measured from the surface of the disks. However, there is increase in elastic modulus with depth below the free surface of 60H (that is, measured in the crosssection, see Fig. 5). The contribution to this effect by different mechanisms is difficult to discern. First of all, it is clear the average orientation of microcracks is different and that their concentration diminishes with depth, both should induce a higher elastic modulus. Also, at higher depths there is a larger interaction of the indentation stress field with the underneath tetragonal

1035

170 0

2

4

6

8

10

12

14

Distance from the surface (µm)

Fig. 5. Monoclinic fraction and elastic modulus measured on the section of disks in terms of the distance to the surface.

phase and with the residual compressive stress field. The change in elastic modulus is from about 175 GPa close to the surface, to 235 GPa at depths where only tetragonal phase is present. These values obtained in the cross-section are in reasonable agreement with those determined by indenting on the surfaces of 60H and AS disks, respectively (see Fig. 4b). If H and E were measured on the surface of the disks at a smaller displacement of 500 nm (the minimum to have reliable results), the contribution from deeper regions is smaller

Fig. 6. AFM images (error signal mode, 25 ␮m × 25 ␮m) of the different nanoindentation imprints at maximum penetration depths on (a) AS, (b) 60H, and (c) 60H+850 ◦ C. The right image shows a detail close to the residual imprint pointing out the main fracture events. The white dash square present in this figure corresponds with the magnified region observed and presented as an inset for each image. The residual imprint for the 60H+600 ◦ C has not been shown here.

1036

J.J. Roa, M. Anglada / Journal of the European Ceramic Society 35 (2015) 1031–1039

Table 2 Indentation hardness and indentation elastic modulus for the specimens of study. Specimen

Hardness, H (GPa)

AS 60H 60H+600 ◦ C 60H+850 ◦ C

16.9 11.5 11.9 15.9

± ± ± ±

0.6 0.4 0.4 1.0

Elastic modulus, E (GPa) 240 191 193 244

± ± ± ±

6 5 3 8

and therefore the values measured of H and E in the degraded disks are also slightly smaller (see Fig. 4 and Table 2). 3.3.2. Damage events during nanoindentation The surface topography (error signal mode) around the Berkovich imprints for AS, 60H, and 60H+850 ◦ C specimens has been observed by AFM in tapping mode to analyze the damage induced surrounding the residual indentation imprints performed at 2000 nm of maximum displacement into the surface, see Fig. 6. In AS no deformation mechanisms such as slip lines induced by dislocation movement or fracture events like radial cracks at the corners of the imprints have been detected, see Fig. 6a. However, cracking or pull-out events of several grains can be appreciated for 60H, (see Fig. 6b). These damage events are associated with the high monoclinic content and the grain boundary microcracks produced by the hydrothermal process. Finally, Fig. 6c exhibits the residual Berkovich imprint performed on 60H+850 ◦ C. In this case, the surface damage surrounding the imprint exhibits the same type of fracture events as for AS, which is consistent with the practical absence of monoclinic phase since it has reversed to tetragonal phase during annealing. To which extent this annealing treatment at 850 ◦ C fully restores the adhesion between the faces of intergranular cracks is not clear. For this, a cross-sectioning inspection by means of FIB was carried out in order to analyze whether the grain boundaries behave in a similar manner as before ageing. Fig. 7 exhibits images obtained by FESEM for indented 60H (Fig. 7a) and 60H+850 ◦ C (Fig. 7b) specimens. As seen in these images, the damage events inside and surrounding the residual indentation are in concordance with the AFM images presented in Fig. 6b and c, respectively. For 60H, uneven microcracks, grain pull-outs, and severe chipping were the common microdefects activated during the indentation process, as can be clearly appreciated in Fig. 7a. In 60H+850 ◦ C, by using a backscattered detector it was possible to distinguish some radial cracks activated at the corner of the residual imprint as can be appreciated in the insets of Fig. 7b. The characterization of the fracture events as well as the deformation mechanisms under the residual indentation imprint was performed by cross-sectional analysis, perpendicularly to the surface and across the residual imprint, using FIB/FESEM (Fig. 8). This image shows a general view of the trench of interest obtained by FIB under a residual imprint. FIB-milled cross sections of 60H (Fig. 8a) exhibit a zone of differing appearance, which is interpreted to reflect the presence of transformed grains. Only a few grains within this cross-section image seem to be unaffected. The irreversible deformation associated to indentation is mainly absorbed by the material located below

Fig. 7. FESEM image of different indentation imprints performed for each specimen of study. (a) 60H and (b) 60H+850 ◦ C. The white dash square present in this figure corresponds with the magnified region observed and presented in the bottom part for each image. The inset for the 60H+850 ◦ C points out several microcracks which are not visible using the conventional secondary electrons or even AFM (see Fig. 6c for the 60H+850 ◦ C).

the indentation at depths smaller than the thickness of the 60H degraded layer. Furthermore, microcracks which run nearly parallel to the surface can be observed in accordance with other reported works.32,33 Just under the imprint, their density is smaller because the indentation plastic field (see white dash line) contributes to close them. Therefore, for 60H the lower hardness and elastic modulus reported in Fig. 4 can be partly associated to the microcracking network generated during ageing, due to the decrease in contact stiffness. Similar behaviour as reported above can be clearly appreciated for the 60H+600 ◦ C, see Fig. 8b. Fig. 8c shows the region just under the imprint for 60H+850 ◦ C. This figure presents no visible microcracking around the residual imprint. This means that, the irreversible deformation is mainly accommodated by other mechanisms, but not by the generation of microcracks as reported elsewhere.27 However, different microstructure just below the surface for the first 200–500 nm may be seen caused by the residual monoclinic phase (less than 5% as we determined using μ-Raman, see Fig. 2).

J.J. Roa, M. Anglada / Journal of the European Ceramic Society 35 (2015) 1031–1039

1037

Table 3 Critical load (Lc ) under sharp sliding contact for the different specimens of study.

Fig. 8. FIB image at the centre of the residual imprint performed at 2000 nm of maximum penetration depth showing the main fracture events for the different specimens of study. (a) 60H, (b) 60H+600 ◦ C, and (c) 60H+850 ◦ C.

From this superficial characterization at micro- and nanometric length scale, it may be concluded that the critical annealing temperature is 850 ◦ C where practically no monoclinic phase remains inside the degraded specimen. However, this temperature is not enough to recover the mechanical integrity, since the first superficial layer present lower mechanical properties than the AS specimen as depicted by the nanoindentation measurements (see Fig. 4). It is clear that at very superficial level some changes induced by ageing are not completely reversed and further experiments under different contact conditions are required for a more precise characterization of the ageing and annealing phenomena. 3.4. Scratch testing In the scratch test there is a large shear stress on the surface caused by friction between the indenter and the material. Fig. 9 shows top-view FESEM images of nanoscratch tracks at a distance of 150 ␮m from the starting point for the different treatments. On the sides of all grooves one can observe pile-ups as well as loose material in the form of flakes produced by the flow of plastically deformed material and which is responsible of the serrated shape along the scratch track. The behaviour at 50 mN applied load of 60H and 60H+600 ◦ C can be clearly discerned from that of AS and 60H+850 ◦ C by comparing the damage features in regions close to track edges. Thus, for AS it is necessary to reach a load to 100 mN to

Specimen

Critical load, Lc (mN)

AS 60H 60H+600 ◦ C 60H+850 ◦ C

>100 50 50 50–100

observe slight damage events close to track edges. Both, AS and 60H+850 ◦ C, do not present cracks or even spalling events in the scratch track. Fig. 10 is a magnification of the scratch tracks, where no radial cracking is appreciated for AS when the applied load is 100 mN or lower. However, 60H exhibits significant grain spalling as well as intergranular cracking as the main damage mechanisms even at 50 mN applied load, pointing out the brittleness of 60H which has a monoclinic and microcracked structure. This spalling effect may be enhanced by the microcracks created during the t-m phase transformation (see Fig. 1b). Finally, for the specimen degraded and annealed, several intergranular microcracks can be appreciated in the middle of the scratch track for applied loads between 50 and 100 mN or even less. Then, the critical loads associated with the first damage event detected along the scratch track are summarized in Table 3. For 60H and 60H+600 ◦ C, grain spalling is the main deformation mechanism and the critical load is similar, (Lc around 50 mN) while for 60H+850 ◦ C the critical load is in the range 50–100 mN, while it is larger than 100 mN for AS. Therefore, annealing 60H produces less grain spalling and grain pull-out inside the scratch track. These data are in correct agreement with a previous work published by Mu˜noz-Tabares et al.34 Fig. 10a–c exhibits the whole scratch track performed under a constant applied load of 50 mN for 60H, 60H+600 ◦ C and 60H+850 ◦ C, respectively. The behaviour is clearly discerned by comparing the micro-fracture events and damage features in regions close to track edges. It is apparent that 60H presents significantly higher spalling events in the track edge as compared with 60H subjected to subsequent annealings (Fig. 10b and c). This effect may be attributed to its intrinsic microstructure because 60H has a microcracked monoclinic structure near the surface and the others present a mixture of tetragonal and microcracked monoclinic phases, as reported in Table 1. After annealing 60H, the defects inside the scratch track decrease with increasing annealing temperature, which can be seen in Fig. 10d and e where high magnification FESEM images are shown from the regions depicted with a dash square. The region magnified of 60H (Fig. 10d) is heavily deformed with numerous grain spalling events. In this regard, it should be highlighted that 60H+850 ◦ C was much less prone to such deformation mechanisms (see Fig. 10e). Therefore, nanoscratch testing allows applying higher stress gradients in the near surface along and below the track. This is clearly revealed by comparing surface features for the AS specimen as well as for the annealed specimen at 850 ◦ C, in which still damage events are clearly visible in regions close to track edges or just below the middle of the track.

1038

J.J. Roa, M. Anglada / Journal of the European Ceramic Society 35 (2015) 1031–1039

Fig. 9. Magnification of the top-view FESEM images of nanoscratch tracks where the first damage start to appear at different applied loads (50 and 100 mN).

Fig. 10. (a–c) FESEM image for the nanoscratch track performed at 50 mN of applied load for the 60H, 60H+600 ◦ C and 60H+850 ◦ C. The scratch track for the AS specimen is not show here. (d and e) Magnification of the main fracture events activated inside the nanoscratch tracks for the 60H and 60H+850 ◦ C, respectively.

This means, that a damage layer of several ␮m is still persistent in the annealed specimen, producing a reduction on their mechanical integrity. The reason may lie in that intergranular microcracks formed during ageing close to the surface are more open than in the interior so that are more difficult to close during annealing in order to recover the strength of the original grain boundaries. 4. Conclusions The effect of the annealing treatment after hydrothermal degradation was analyzed at micro- and nanometric length scale. Based on our results, we can draw the following conclusions:

- Hydtrothermal ageing produces profound effects on the phase composition and the surface deformation and fracture behaviour. It induces a surface layer of monoclinic phase with intergranular microcracking, a reduction in elastic modulus and hardness, and with a more brittle behaviour than the tetragonal original phase. - Annealing at 600 ◦ C for 1 h is able to reverse a significant amount of monoclinic phase to tetragonal, but the deformation and fracture behaviour of the degraded layer is still very similar to the aged material. - Annealing at 850 ◦ C has a strong effect, practically reversing all the monoclinic phase content, as well as the hardness and elastic modulus, but the mechanical behaviour very close

J.J. Roa, M. Anglada / Journal of the European Ceramic Society 35 (2015) 1031–1039

to the surface is not completely recovered as shown by nanoindentation and nanoscratch testing. Therefore, recovering the tetragonal structure by annealing does not necessarily indicates the recovering of the deformation and fracture behaviour of the original tetragonal phase at the surface. It is suggested that this effect is associated to the microcracked grain boundaries induced during ageing which do not regain their intergranular original strength during annealing at 850 ◦ C. Acknowledgements Financial assistance for this investigation was partly granted by the Spanish Ministerio de Economia y Competitividad through grant MAT2011-23913. The financial support from Direcció General de Recerca del Comissionat per a Universitats i Recerca de la Generalitat de Catalunya is also acknowledged (2014 SGR 130). Dr. J. J. Roa would like to thank the Juan de la Cierva Programme for its financial support (grant number: JCI-2012-14454). References 1. Jing Z, Ke Z, Yihong L, Zhijian S. Effect of multistep processing techniques on the formation of micro-defects and residual stresses in zirconia dental restoration. J Prosthodontics 2014;23:206–10. 2. Piconi C, Maccauro G. Zirconia as a ceramic biomaterial. Biomaterials 1999;20:1–25. 3. Borchers L, Stiesch M, Bach F-W, Buhl J-C, Hübsch C, Kellner T, et al. Influence of hydrothermal and mechanical conditions on the strength of zirconia. Acta Biomater 2010;6:4547–52. 4. Thompson VP, Rekow DE. Dental ceramics and the molar crown testing ground. J Appl Oral Sci 2004;12:26–36. 5. Meyenberg KH, Luthy H, Schärer PJ. Zirconia posts: a new all-ceramic concept for nonvital abutment teeth. Esthet Dent 1995;7:73–80. 6. Kohal R-J, Patzelt SBM, Butz F, Sahlin H. One-piece zirconia oral implants: one-year results from a prospective case series. 2. Three-unit fixed dental prosthesis (FDP) reconstruction. J Clin Periodontol 2013;40: 553–62. 7. Chevalier J, Gremillard L, Virkar AV, Clarke DR. The tetragonal-monoclinic transformation in zirconia: lessons learned and future trends. J Am Ceram Soc 2009;92(9):1901–20. 8. Lawson S. Environmental degradation of zirconia ceramics. J Am Ceram Soc 1995;15:485–502. 9. Cattani-Lorente M, Scherrer SS, Ammann P, Jobin M, Wiskott HW. Low temperature degradation of a Y-TZP dental ceramic. Acta Biomater 2011;7(2):858–65. 10. Keuper M, Eder K, Berthold C, Nickel KG. Direct evidence for continuous linear kinetics in the low-temperature degradation of Y-TZP. Acta Biomater 2013;9:4826–35. 11. Hannink RHJ, Kelly PM, Muddle BC. Transformation toughening in zirconia-containing ceramics. J Am Ceram Soc 2000;83:461–87. 12. Catledge SA, Cook M, Vohra YK, Santos EM, McClenny MD, Moore KD. Surface crystalline phases and nanoindentation hardness of explanted zirconia femoral heads. J Mater Sci Mater Med 2003;14(10):863–7.

1039

13. Chevalier J, Cales B, Drouin JM. Low-temperature aging of Y-TZP ceramics. J Am Ceram Soc 1999;82:2150–4. 14. Casellas D, Feder A, Llanes L, Anglada M. Fracture toughness and mechanical strength of Y-TZP/PSZ ceramics. Scr Mater 2001;45:213–20. 15. Casellas D, Cumbrera FL, Sánchez-Bajo F, Forsling W, Llanes L, Anglada M. Fracture toughness and mechanical strength of Y-TZP/PSZ ceramics. J Eur Ceram Soc 2001;21:765–77. 16. Cales CB, Drouin JM. Low temperature aging of Y-TZP ceramics. J Am Ceram Soc 1999;82:2150–4. 17. Chevalier J, Drouin JM, Cales B. Low temperature aging behavior of zirconia hip joint heads. Bioceramics: Proc 10th Int Symp Ceram Med 1997;10:135–8. 18. Mu˜noz-Tabares JA, Anglada M. Quantitative analysis of monoclinic phase in 3Y-TZP by Raman spectroscopy. J Am Ceram Soc 2010;93:1790–5. 19. Oliver WC, Pharr GM. Measurement of hardness and elastic modulus by instrumented indentation: advances in understanding and refinements to methodology. J Mater Res 2004;19:3–20. 20. Oliver WC, Pharr GM. An improved technique for determining hardness and elastic modulus using load and displacement sensing indentation experiments. J Mater Res 1992;7:1564–83. 21. Horcas I, Fernández R, Gómez-Rodríguez JM, Colchero J, Gómez-Herrero J, Baro AM, et al. A software for scanning-probe microscopy and a tool for nanotechnology. Rev Sci Instrum 2007;78:2672–92. 22. Marro FG, Chintapalli R, Hvizdos P, Soldera F, Mücklich F, Anglada M. Study of near surface changes in yttria-doped tetragonal zirconia after low temperature degradation. Int J Mater Res 2009;100:92–6. 23. Garvie RC, Swain MV. Thermodynamics of the tetragonal to monoclinic phase transformation in constrained zirconia microcrystals. J Mater Sci 1985;20:1193–200. 24. Lange FF. Transformation toughening: Part 1. Size effects associated with the thermodynamics of constrained transformation. J Mater Sci 1982;17:225–34. 25. Heuer AH, Claussen N, Kriven WM, Rühle M. Stability of tetragonal ZrO2 particles in ceramic matrices. J Am Ceram Soc 1982;65(12):642–50. 26. Gaillard Y, Anglada M, Jiménez-Piqué E. Nanoindentation of yttria-doped zirconia: effect of crystallographic structure on deformation mechanisms. J Mater Res 2009;24(3):719–27. 27. Gaillard Y, Jiménez-Piqué E, Soldera F, Mücklich F, Anglada M. Quantification of hydothermal degradation in zirconia by nanoindentation. Acta Mater 2008;56(16):4206–16. 28. Cutler RA, Lindemann JM, Ulvensoen JH, Lange HI. Damage-resistan SrOdoped Ce-TZP/Al2 O3 composites. Mat Design 1994;13(3):1039–43. 29. Ingel RP, Lewis D, Bender BA, Rice RW. Physical microstructural and thermomechanical properties of ZrO2 single crystals. Adv Ceram 1984;12:408–14. 30. Bravo-Leon A, Morikawa Y, Kawahara M, Mayo MJ. Fracture toughness of nanocrystalline tetragonal zirconia with low yttria content. Acta Mater 2002;50:4555–62. 31. Eichler J, Eisele U, Rödel J. Mechanical properties of monoclinic zirconia. J Am Ceram Soc 2004;87(7):1401–3. 32. Marro FG, Chintapalli R, Hvizdos P, Soldera F, Mucklich F, Anglada M. Study of near surface changes in yttria-doped tetragonal zirconia after low temperature degradation. Int J Mater Res 2009;100:92–6. 33. Soldera F, Gaillard Y, Gomila M, Mujecklich F. FIB-tomography of nanoindentation cracks in zirconia polycrystals. Microsc Microanal 2007;13:1510–1. 34. Mu˜noz-Tabares JA, Jiménez-Piqué E, Reyes-Gasga J, Anglada M. Microstructural changes in 3Y-TZP induced y scratching and indentation. J Eur Ceram Soc 2012;32:3919–27.