Effect of cation dopant radius on the hydrothermal stability of tetragonal zirconia: Grain boundary segregation and oxygen vacancy annihilation

Acta Materialia 106 (2016) 48e58

Contents lists available at ScienceDirect

Acta Materialia journal homepage: www.elsevier.com/locate/actamat

Full length article

Effect of cation dopant radius on the hydrothermal stability of tetragonal zirconia: Grain boundary segregation and oxygen vacancy annihilation € n c, Fei Zhang a, *, Maria Batuk b, Joke Hadermann b, Gabriele Manfredi c, An Marie a d d d Kim Vanmeensel , Masanao Inokoshi , Bart Van Meerbeek , Ignace Naert , Jef Vleugels a a

Department of Materials Engineering, KU Leuven, Kasteelpark Arenberg, 44, B-3001 Heverlee, Belgium Electron Microscopy for Materials Science (EMAT), University of Antwerp, Groenenborgerlaan 171, B-2020 Antwerp, Belgium c SCK CEN (Belgian Nuclear Research Centre), Boeretang 200, B-2400 Mol, Belgium d KU Leuven BIOMAT, Department of Oral Health Sciences, KU Leuven & Dentistry, University Hospitals Leuven, Kapucijnenvoer, 7, Block A, B-3000 Leuven, Belgium b

a r t i c l e i n f o

a b s t r a c t

Article history: Received 6 September 2015 Received in revised form 22 December 2015 Accepted 30 December 2015 Available online xxx

The hydrothermal aging stability of 3Y-TZP-xM2O3 (M ¼ La, Nd, Sc) was investigated as a function of 0.02 e5 mol% M2O3 dopant content and correlated to the overall phase content, t-ZrO2 lattice parameters, grain size distribution, grain boundary chemistry and ionic conductivity. The increased aging stability with increasing Sc2O3 content and the optimum content of 0.4e0.6 mol% Nd2O3 or 0.2e0.4 mol% La2O3, resulting in the highest aging resistance, could be directly related to the constituent phases and the lattice parameters of the remaining tetragonal zirconia. At low M2O3 dopant contents 0.4 mol%, the different aging behavior of tetragonal zirconia was attributed to the defect structure of the zirconia grain boundary which was influenced by the dopant cation radius. It was observed that the grain boundary ionic resistivity and the aging resistance followed the same trend: La3þ > Nd3þ > Al3þ > Sc3þ, proving that hydrothermal aging is driven by the diffusion of water-derived mobile species through the oxygen vacancies. Accordingly, we elucidated the underlying mechanism by which a larger trivalent cation segregating at the zirconia grain boundary resulted in a higher aging resistance. © 2016 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

Keywords: Zirconia Aging Cation radius Oxygen vacancy Grain boundary Dopant content

1. Introduction Tetragonal zirconia, especially yttria-stabilized tetragonal zirconia polycrystals (Y-TZP), possesses an excellent combination of high toughness and strength, which is mainly attributed to the transformation toughening effect of the metastable tetragonal zirconia phase [1]. Furthermore, Y-TZP ceramics are wear, corrosion and high-temperature resistant with low thermal conductivity, good ionic conductivity, good biocompatibility and superior esthetic appearance [1e3]. Therefore, they are very attractive for a wide range of applications such as fixed-partial denture in restorative dentistry, femoral heads in orthopedics and solid electrolytes

* Corresponding author. E-mail address: [email protected] (F. Zhang). http://dx.doi.org/10.1016/j.actamat.2015.12.051 1359-6454/© 2016 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

in solid oxide fuel cells [2,3]. However, Y-TZP ceramics suffer from low temperature degradation, in which the tetragonal zirconia (t-ZrO2) phase spontaneously transforms to the monoclinic (m-ZrO2) phase in the presence of water or water vapor (hydrothermal aging) without any applied stress over the temperature region from room temperature up to about 400  C [4,5]. This problem was considered to be minor until hundreds of Y-TZP total hip prostheses (THP) ball-heads catastrophically failed in a very short time between 1999 and 2000, leading to its withdrawal from the market [5,6]. Hydrothermal aging is a progressive process triggered by water molecules, which starts from the surface and propagates into the ceramic component, resulting in surface roughening and microcracking [2,5,7,8]. The performance and reliability of zirconia components thereby will be inevitably influenced when used in aqueous environments such as for biomedical applications.

F. Zhang et al. / Acta Materialia 106 (2016) 48e58

The aging resistance of Y-TZP ceramics can be improved by increasing the yttria content [2,8], using an alternative stabilizer (CeO2 for instance) and using aluminaezirconia composites [9e11]. However, some properties of Y-TZP ceramics such as strength or esthetic appearance cannot be simultaneously maintained by these approaches. Different research reports show that (co-)doping of YTZP with a small amount of other ions, e.g., Al3þ [12,13], Fe3þ [14], Pr3þ [15], Ce4þ, Bi3þ [16], La3þ [17], Si4þ [13], provides a satisfactory balance between aging resistance, mechanical properties [7,13,16,17] and esthetic appearance, including color [14e16] and translucency [14,17]. Although many oxides have been tested, there is no systematic guideline for choosing a suitable doping. Recently, we reported that the cation dopant radius can be used as a controlling parameter, and larger trivalent cations segregating at the zirconia grain boundary can effectively retard the aging rate of 3YTZP ceramics without sacrificing the excellent mechanical properties [17]. Although different studies have reported that grain boundaries play a key role in enhancing/retarding the aging kinetics [2,5,12,13,18] and dopant cations segregated at the grain boundary (including Al3þ, La3þ, Cu2þ, Mg2þ and Ge4þ) could effectively retard the aging rate of 3Y-TZP ceramics [17e22], the underlying mechanism is still not completely clarified. In addition, from the perspective of hydrothermal aging mechanism, the behavior of oxygen vacancies in Y-TZP ceramics should be assessed in order to explain the different aging kinetics. Although the mechanism of hydrothermal aging is not fully understood, the involvement of oxygen vacancies is widely realized and the most prevailing mechanism emphasizes the primary role of oxygen vacancies annihilation [2,21,23e25]. It has been reported that the tetragonal zirconia phase may be stabilized by oxygen vacancies adjacent to the Zr4þ ion [24,26]. The apparent activation energy of the aging kinetics (73e106 kJ/mol for various stabilizer and grain sizes [8,9,11,27]) is reported to be comparable to the activation energy for ionic conductivity in 3Y-TZP below 500  C (88e89 kJ/mol) [28,29]. Thus, it is commonly accepted that the diffusion of water-derived species into the zirconia lattice and the filling of oxygen vacancies (Vo,, ) is responsible for the hydrothermal aging process of Y-TZP ceramics [21,23e25,30]. In this work, to elucidate how the cation dopant radius influences the hydrothermal aging kinetics of tetragonal zirconia, 3YTZP ceramics were doped with different trivalent oxides M2O3 (M ¼ La, Nd, Sc and Al). The fundamental microstructural parameters including the constituent phases, t-ZrO2 crystal lattice parameters, grain size distribution, grain boundary chemistry and oxygen ionic conductivity were investigated. Since the amount of dopant also showed a crucial impact on the aging kinetics [17,31], which can again depend on the cation radius, the hydrothermal aging kinetics of 3Y-TZP-xM2O3 (M ¼ La, Nd, Sc) were systematically studied as a function of M2O3 dopant content from 0.02 to 5 mol%. 2. Materials and methods 2.1. Material preparation High purity 3Y-TZP powder (grade TZ-3Y, Tosoh, Japan) was respectively doped with trivalent oxides (Sc2O3, Nd2O3, La2O3) having a different cation radius Zr4þ (84.0 pm) ~ Sc3þ (87.0 pm) < Y3þ (101.9 pm) < Nd3þ (110.9 pm) < La3þ (116.0 pm) [32]. 0.02e5 mol% Nd2O3 (Chempur, purity of 99.9%), La2O3 (Chempur, purity of 99.99%) and Sc2O3 (abcr GmbH & Co. KG, purity of 99.9%) were respectively mixed with tetragonal zirconia nanopowder on a multidirectional mixer (Turbula type T2C, WAB, Switzerland) for 24 h in ethanol using 5 mm Y-TZP (grade TZ-3Y, Tosoh, Japan) milling balls. The mixed suspension was further

49

processed by bead milling (Dispermat SL, Germany) for 3 h at 5000 rpm using 1 mm ZrO2 beads (grade TZ-3Y, Tosoh, Japan). The base powder (pure 3Y-TZP) was bead milled as a reference. 0.25 wt.% (0.3 mol%) alumina-doped 3Y-TZP powder (grade TZ-3YE, Tosoh, Japan) was processed as well for comparison, because alumina is widely doped in 3Y-TZP ceramics and aluminum cation has a much smaller radius (Al3þ (53.5 pm)) than the other dopant cations. All powders were cold isostatically pressed at 250 MPa (EPSI, Temse, Belgium) and pressureless sintered in air at 1500  C for 2 h. Since the yttria is co-precipitated in the TZ-3Y starting powder, we refer to the other oxide additives as dopants in the 3Y-TZP ceramics. The sintered zirconia ceramics are referred to as 3Y-TZPxM2O3 (M ¼ La, Nd, Sc and Al) with x the mol % of M2O3. 2.2. Phase characterization and Rietveld refinements X-ray diffraction (XRD, 3003-TT, Seifert, Ahrensburg, Germany) using CueKa radiation at 40 kV and 40 mA was used for phase analysis. XRD patterns were recorded at room temperature from polished ceramic surfaces in the 20e90 (2q) range with a scan speed of 2 s/step and a scan size of 0.02 . Rietveld refinements of the XRD patterns were performed using TOPAS-Academic software (Bruker AXS, Karlsruhe, Germany). The phase structures were refined as: tetragonal zirconia (t) unit cell with space group P42/nmcZ, monoclinic zirconia (m) unit cell with space group P21/c, cubic zirconia (c) with space group Fm-3m, “non-transformable” tetragonal zirconia (t0 ) with space group P42/nmcS and the La2Zr2O7 phase with space group Fd-3mZ. The quality of the Rietveld refinement was controlled with a low R value of <10%. To simplify the discussion and clarify the interphase relationships, the tetragonal data are presented using a pseudocubic (distorted fluorite) unit cell and the lattice parameter of t and t0 was multiplied by √2. 2.3. Microstructural characterization Scanning electron microscopy (SEM, XL-30FEG, FEI, Eindhoven, The Netherlands) was used to characterize the microstructure on polished, thermally etched (1250  C for 25 min in air) and Pt-coated surfaces. The grain size was measured on SEM micrographs using IMAGE-PRO software according to the linear intercept method. At least 1000 grains were counted, and the grain size distribution and the average results (±standard deviation) were reported with a correction factor of 1.56. Transmission electron microscopy (TEM) analysis was performed to examine the distribution of dopant cations (La3þ, Nd3þ, Al3þ or Sc3þ), Y3þand Zr4þ around the grain boundaries. Electron transparent samples were prepared by ion-milling with an Ion Slicer (EM-09100IS, Jeol, Japan). High angle annular dark field scanning transmission electron microscopy (HAADF-STEM) images and energy-dispersive spectroscopy (STEM-EDS) elemental maps were obtained on a FEI Titan 60e300 “cubed” transmission electron microscope operated at 200 kV. 5e7 grain boundaries in each ceramic were analyzed. AleK, SceK, Y-L, Zr-L, La-L, and Nd-L lines were used for the elemental maps. Quantitative elemental maps were acquired to calculate the element concentration profiles across the grain boundary using ESPRIT 1.9 software. 2.4. Assessment of aging kinetics In vitro accelerated hydrothermal experiments were used to age the ceramics. Double-side mirror polished specimens were autoclaved at 134  C and 0.2 MPa in water vapor. The amount of tetragonal to monoclinic zirconia phase transformation was

50

F. Zhang et al. / Acta Materialia 106 (2016) 48e58

determined by X-ray diffraction. XRD patterns were recorded in the 27e33 (2q) range in the qe2q mode with a scan speed of 2 s/step and a scan size of 0.02 . The monoclinic ZrO2 phase content (Vm) was calculated according to the formula of Toraya et al. [33]:

Vm

1:311Xm ¼ 1 þ 0:311Xm

(1)

The value of Xm was calculated using the method of Garvie and Nicholson [34]:

Xm

I 111 þ I 111 ¼ 111m 111 m 101 Im þ Im þ I t

rgb $dg dgb

(4)

where dg is the average grain size, dgb is the effective electrical thickness of the grain boundary. A constant value for dgb of 5 nm was used for all ceramics. 3. Results 3.1. Phase composition

(2)

With I, the intensity of monoclinic (111 and 111) and tetragonal (101) phase peaks indicated by the subscripts m and t. The monoclinic phase content was plotted as a function of aging time to indicate the aging kinetics. For each curve, at least 3 specimens (6 exposed surfaces) were tested and the average result was reported. In order to compare the aging kinetics of different ceramics, the transformation curves were fitted with sigmoidal functions consistent with the MehleAvramieJohnson (MAJ) formalism [4]:


Vm  Vm0 ¼ 1  exp  ðbtÞn Vms  Vm0

sp

rgb ¼

(3)

where Vm is the monoclinic phase content, Vm0 and Vms are the initial and saturation levels of monoclinic phase content, b (h1) is the parameter describing the effective kinetics of the tetragonal to monoclinic phase transformation during the aging process [4], and n is the parameter related to the geometry of the transformation [35]. Note that no peak of m-ZrO2 was observed for any of ceramic before aging treatment, so Vm0 was zero in all the aging curves. Besides, it was necessary to hydrothermally treat the ceramics until the surface m-ZrO2 content reached the saturation level. Different dopants and dopant concentrations significantly influenced the fraction of non-transformable zirconia, and the real aging kinetics of transformable tetragonal zirconia can only be calculated with an accurate value of Vms.

2.5. Electrochemical impedance spectroscopy The oxygen ionic conductivity was studied by electrochemical impedance spectroscopy (EIS). The impedance spectra were measured at 200  C over the frequency range of 0.01e105 Hz (10 points per decade) with a GAMRY® potentiostat system (PC4-300) at a perturbation amplitude of 100 mV. Liquid leadebismuth eutectic (LBE) was used as electrode, and both polished surfaces (1 mm diamond suspension) of the ceramic discs (0.3 cm thick, ∅ ¼ 10 mm) were in contact with liquid LBE. The complex impedance spectra were fitted according to an equivalent circuit consisting of three resistance (R)-constant phase element (CPE) circuits with the ZView program (Version 2.9, Scribner Associates, Inc.). The resistivity (r ¼ R*A/L, A is the electrode surface area (~0.79 cm2) and L is the sample thickness (~0.3 cm)) of the bulk and the grain boundary were extracted. The results of the electrode at lower frequency are not presented for sp clarity. The specific grain boundary resistivity (rgb ) which represents the effective resistivity from the grain boundary region was calculated according to the brick layer model [36]:

Representative XRD patterns of 0.02e5 mol% M2O3 (M ¼ Sc, Nd, La)-doped 3Y-TZP ceramics are shown in Fig. 1. No peaks representing trivalent oxides (M2O3) were found, independent of the dopant content, revealing that solid solutions were formed in all compositions. However, the evolution of the constituent phases varied depending on the dopant cation radius (Zr4þ (84.0 pm) ~ Sc3þ (87.0 pm) < Nd3þ (110.9 pm) < La3þ (116.0 pm) [32]). Different zirconia phases (t, t0 and c) were recognized from the {004} diffraction peaks, as shown in Fig. 1c. c-ZrO2 was formed for larger La2O3 and Nd2O3 dopants, whereas t0 -ZrO2 was formed with the Sc2O3 dopant. The characteristic c-ZrO2 reflection in-between the tetragonal t(004) and t(400) doublets [37] was not found for Sc2O3 doping, but another set of doublet peaks that were closer to each other was observed. The formation of t0 -ZrO2 was in agreement with the results reported in literature for Sc2O3-doped zirconia ceramics [38,39]. Besides, at La2O3 concentrations 0.6 mol%, the c(004) peak stopped growing and peaks corresponding to La2Zr2O7 were observed, revealing that the solid solubility limit of La2O3 in zirconia was about 0.4e0.6 mol% after 2 h sintering at 1500  C, whereas the solubility of Sc2O3 or Nd2O3 in zirconia was higher than 5 mol%. The quantitative phase content, as obtained by Rietveld analysis, is plotted as a function of the M2O3 content in Fig. 2. The 3 mol% yttria-stabilized zirconia base material contained 90.3 wt.% t-ZrO2 and 9.7 wt.% c-ZrO2. As the dopant content increased within the range of M2O3 solubility in 3Y-TZP, the fraction of t-ZrO2 was linearly consumed by the formation of a stable zirconia phase, i.e., cZrO2 for La2O3 and Nd2O3 doping and t0 -ZrO2 for Sc2O3 doping. Since t-ZrO2 peak shifts were clearly observed for all the dopants (Fig. 1b and c), the crystal structure of the remaining t-ZrO2 was also altered by all investigated M2O3 dopant species within the solubility range. The peak shifting direction revealed that La3þ and Nd3þ expanded the t-ZrO2 lattice, while Sc3þ contracted the t-ZrO2 lattice, as confirmed from the lattice parameters (Fig. 3a) and the unit cell volume (Fig. 3b). Fig. 3c shows the calculated tetragonality (c/a) of the tetragonal (t and t0 ) zirconia phases. 3.2. Microstructure Representative SEM micrographs of the different 3Y-TZP-xM2O3 (M ¼ Sc, Nd, La) ceramics are shown in Fig. 4a. The grain size distribution and the average grain size are presented as a function of the dopant content in Fig. 4b and c, respectively. Sc2O3 doping enhanced the grain growth of zirconia and the average grain size of 3Y-TZP-xSc2O3 increased with dopant concentration. For 3Y-TZPxNd2O3, the average grain size first decreased up to 0.6 mol% Nd2O3 (Fig. 4c), but increased at Nd2O3 contents 1 mol% due to the formation of a bimodal microstructure with c-ZrO2 and t-ZrO2 phases, as can be observed in the SEM micrographs in Fig. 4a (3Nd) and the grain size distribution in Fig. 4b. The average grain size of zirconia in 3Y-TZP-xLa2O3 ceramics decreased up to 0.4 mol% La2O3 addition and remained constant at higher levels (Fig. 4c). The grain size

F. Zhang et al. / Acta Materialia 106 (2016) 48e58

51

Fig. 1. (a) Representative XRD patterns of the 3Y-TZP-xM2O3 (M ¼ La, Nd, Sc and x ¼ 0.02e5 mol%) systems and expanded view of the 42e44 (b) and 71e76 (c) 2q region. t is tetragonal zirconia, c cubic zirconia and t0 non-transformable tetragonal zirconia. Tetragonal phases having a tetragonality 1.000 < c/a <1.007 are defined as ‘tetragonal t0 phase' [38,43], and t0 -ZrO2 has doublets t0 (004) and t0 (400) that are closer to each other [37].

Fig. 2. Phase content analysis by Rietveld refinement of 3Y-TZP doped with 0.02e5 mol% La2O3, Nd2O3 or Sc2O3.

distributions were comparable for all 3Y-TZP-xLa2O3 ceramics in the 0.02e5 mol% dopant range, as presented in Fig. 4b. Note that the grains of c-, t0 -and t-ZrO2 were included in the grain size measurements.

3.3. Hydrothermal aging Fig. 5a shows the m-ZrO2 phase content induced by the hydrothermal aging in the 3Y-TZP-xM2O3 as a function of the aging time on a logarithmic time scale. To demonstrate the evolution of the aging kinetics, the kinetic parameter b obtained from equation (3)

52

F. Zhang et al. / Acta Materialia 106 (2016) 48e58

Fig. 3. (a) Lattice parameters (b) unit cell volume and (c) c/a tetragonality of t- and t0 -ZrO2 in 3Y-TZP doped with 0e5 mol% trivalent oxides M2O3 (M ¼ Sc, Nd or La). The tetragonal data are presented using a pseudo-cubic (distorted fluorite) unit cell.

Fig. 4. Microstructures of the different 3Y-TZP-xM2O3 (M ¼ Sc, Nd, La and x ¼ 0.02e5 mol%): (a) Representative SEM micrographs of 0.4 and 3.0 mol% M2O3-doped 3Y-TZP ceramics. (b) Representative grain size distributions of 0.1, 0.4, 1, 3 and 5 mol% M2O3-doped 3Y-TZP. (c) Evolution of the average grain size.

is plotted as a function of the dopant content in Fig. 5b. The numerical values of parameter b, Vms and n are summarized in Table 1. The aging kinetics of 3Y-TZP-xM2O3 (M ¼ Sc, Nd, La) as a function of dopant content followed different trends, depending on the

dopant species. 3Y-TZP-xSc2O3 became more hydrothermally stable at a higher Sc2O3 content. However, with increasing of the Nd2O3 or La2O3 content, the aging kinetics of 3Y-TZP-xM2O3 (M ¼ La and Nd) first decreased and then increased. There was an optimum amount

F. Zhang et al. / Acta Materialia 106 (2016) 48e58

53

Fig. 5. Aging kinetics of 0.02e5 mol% M2O3-doped 3Y-TZPs at 134  C: (a) Representative curves of the surface monoclinic phase content as a function of aging time. (b) Aging kinetic parameter b, as obtained from the surface transformation curve, as a function of the M2O3 dopant content.

Table 1 Numerical values of the aging kinetic parameters n and b at 134  C for 3Y-TZP-xM2O3 (M ¼ La, Nd and Sc, x ¼ 0.02e5 mol%). Vms is the monoclinic zirconia phase saturation level used in MAJ fitting. The values for the aging kinetics of 3Y-TZP reference ceramic are Vms ¼ 88%, n ¼ 1.53 and b ¼ 0.114 h1. Mol%

0.02 0.1 0.2 0.4 0.6 1 2 3 4 5

Sc2O3

Nd2O3

La2O3

Vms

n

b (h1)

Vms

n

b (h1)

Vms

n

b (h1)

87 86 84 81 77 66 ~40a e e e

1.2 1.3 1.4 1.3 1.2 1.2 0.6 e e e

0.167 0.130 0.120 0.085 0.070 0.036 0.002 e e e

87 85 83 77 72 60 43 33 21 13

1.9 1.7 1.7 1.5 1.3 1.2 1.0 0.7 0.8 0.7

0.104 0.070 0.043 0.024 0.012 0.015 0.076 0.122 0.186 0.205

88 87 85 81 77 74 74 74 74 74

1.6 1.6 1.5 1.3 1.2 1.3 1.3 1.2 1.1 1.1

0.082 0.038 0.019 0.017 0.023 0.090 0.057 0.040 0.061 0.046

a The saturation level was not reached but the value was estimated from the phase content.

of dopant (0.2e0.4 mol% for La2O3 and 0.4e0.6 mol% for Nd2O3) resulting in the highest hydrothermal aging resistance of the 3YTZP ceramics. On the other hand, at the same dopant content, the hydrothermal stability of 3Y-TZP ceramic was also strongly influenced by the dopant species. In particular, the true influence of the dopant cation radius on the aging behavior of 3Y-TZP should be investigated with a minor amount of M2O3 in order to maintain the 3Y-TZP phase constitution. At low concentration (0.4 mol%), La2O3 doping led to the most hydrothermally stable 3Y-TZP, followed by Nd2O3 and

Sc2O3. 3.4. Grain boundary chemistry of 3Y-TZP doped with 0.4 mol % M2O3 (M ¼ La, Nd, Sc) or 0.3 mol% Al2O3 Different studies have reported that, in the case of minor amounts of addition, dopant cations segregating at the grain boundary could effectively retard the aging rate of 3Y-TZP ceramics [17e22]. Therefore, the grain boundaries of 0.4 mol% M2O3 or 0.3 mol% Al2O3 doped 3Y-TZP were investigated with STEM-EDS (Fig. 6). 0.3 mol% Al2O3 doped 3Y-TZP was used for comparison, since Al3þ (53.5 pm) has a much smaller radius than the other dopant cations and 0.3 mol% has been proven to be the maximum amount of alumina that can be dissolved in zirconia at the sintering temperature [31]. It is clear from Fig. 6 that the Sc3þ (87.0 pm) was not able to segregate at the zirconia grain boundary due to the small mismatch with the host Zr4þ (84.0 pm) cation [40]. The other cations (La3þ, Nd3þ, Al3þ) strongly segregated at the zirconia grain boundaries over a width of 2e4 nm. Note that the grain boundaries investigated in this work were clean and free from any amorphous phase (Fig. 6b). 3.5. Oxygen ionic conductivity The behavior of oxygen vacancies in 0.4 mol% M2O3 (M ¼ La, Nd and Sc) or 0.3 mol% Al2O3-doped 3Y-TZP ceramics were assessed by measuring the oxygen ionic conductivity. The complex impedance spectra measured at 200  C are shown in Fig. 7a. In order of

54

F. Zhang et al. / Acta Materialia 106 (2016) 48e58

Fig. 6. Grain boundary chemistry of 3Y-TZPs doped with 0.4 mol% Sc2O3, Nd2O3 and La2O3 and commercially available TZ-3Y-E containing 0.25 wt.% (0.3 mol%) Al2O3. The spheres qualitatively represent the size of the dopant cations, Al3þ Sc3þ, Nd3þ and La3þ, relative to the Zr4þ host cation. (a) HAADF-STEM images, corresponding dopant STEM-EDS element maps, and elemental distribution profiles across the grain boundaries in which the dots correspond to the experimental data and the black lines to the fitting curves. (b) Expanded view of a grain boundary in the 3Y-0.3Al ceramic.

decreasing frequency, the arcs correspond to the impedance response of the bulk and the grain boundary. The corresponding bulk resistivity (rbulk), grain boundary resistivity (rgb) and specific grain boundary resistivity (rsp ) which is corrected for the average gb grain size (dg) are reported in Table 2. The ionic conductivity, especially the grain boundary resistivity, of all 3Y-TZP ceramics was considerably influenced by a minor amount of M2O3 addition. Interestingly, comparing ionic

conductivity (Fig. 7a) and hydrothermal aging curves (Fig. 7b) revealed that the grain boundary resistivity (rgb and rsp ) and gb hydrothermal aging resistance followed the same trend: 3Y-0.4La > 3Y-0.4Nd > 3Y-0.25Al > 3Y-0.4Sc. This unequivocally demonstrated the direct correlation between the hydrothermal aging sensitivity of 3Y-TZP and the oxygen ionic conductivity of the zirconia grain boundary.

F. Zhang et al. / Acta Materialia 106 (2016) 48e58

55

Fig. 7. Comparison of 3Y-TZPs doped with 0.4 mol% Sc2O3, Nd2O3 and La2O3 and commercially available TZ-3Y-E containing 0.25 wt.% (0.3 mol%) Al2O3. (a) Complex impedance spectra (symbol) measured at 200  C and fitted results (line) using the equivalent circuit shown in the inset. (b) Aging resistance, monoclinic phase content formed during accelerated aging at 134  C as a function of the aging time.

Table 2 Oxygen ionic resistivity of 3Y-TZP doped with a small amount of M2O3 (M ¼ La, Nd, Al, Sc) measured at 200  C. rbulk is the bulk resistivity, rgb the grain boundary resp sistivity, rgb the specific grain boundary resistivity and dg the average grain size.

dg (nm) rbulk (MU$cm) rgb (MU$cm) ðMU$cmÞ rsp gb

3Y-0.4La

3Y-0.4Nd

3Y-0.3Al

3Y-0.4Sc

332 1.77 8.73 580

412 2.23 5.79 477

482 1.60 5.02 484

535 1.57 3.77 404

4. Discussion The following discussion tries to elucidate how the trivalent cations influence the hydrothermal stability of tetragonal zirconia in 3Y-TZP-xM2O3 (x ¼ 0.02e5 mol%, M ¼ Sc, Nd or La) systems. Considering the evolution shown in Fig. 5, the effect of the dopant content which depends on the dopant species is discussed first, and the effect of the cation size at M2O3 contents below 0.4 mol % is addressed in detail later. 4.1. Effect of dopant content: correlation between the hydrothermal aging and the crystal structure of tetragonal zirconia It is well known that the addition of lower-valence oxides such as rare-earth M2O3 can stabilize zirconia in the cubic or tetragonal crystal structure [24,26,41]. A higher amount of M2O3 therefore is expected to result in more hydrothermally stable tetragonal zirconia, independent on the dopant species. Present results show that a higher amount of M2O3 (M ¼ Sc, Nd, La) below the solubility limit resulted in a larger fraction of hydrothermally stable nontransformable cubic or t0 -ZrO2 phase [42] in the 3Y-TZP-xM2O3 systems (Fig. 2), corresponding to a lower Vms value in the aging curves (Table 1). However, the remaining t-ZrO2 did not always become more thermodynamically stable (Fig. 3c). Therefore, the aging resistance of the 3Y-TZP-xM2O3 (M ¼ Sc, Nd, La) systems showed different trends (Fig. 5b), depending on the constituent phases and the lattice parameters of the remaining t-ZrO2. Increasing Sc2O3 concentration made the crystal structure of the remaining t-ZrO2 more thermodynamically stable, as seen from the lower tetragonality (c/a) (Fig. 3c) [2,43], so Sc2O3 doping followed

the expected trend with an increased hydrothermal stability with increasing “stabilizer” content from 0.02 to 5 mol%, regardless of the fact that the average grain size of zirconia grew with the Sc2O3 concentration (Fig. 4a). However, when doping with La2O3 or Nd2O3, there was an optimum amount of dopant (0.2e0.4 mol% for La2O3 and 0.4e0.6 mol% for Nd2O3) resulting in the highest hydrothermal aging resistance of the 3Y-TZP ceramics. A bimodal cZrO2 and t-ZrO2 microstructure was formed at Nd2O3 contents >0.6 mol% (Fig. 4). The c-ZrO2 phase was enriched in stabilizer content (might even have two stabilizers Nd3þ and Y3þ in the grains), and this phase partitioning can lead to a stabilizer depletion in the remaining t-ZrO2 grains which can act as preferential nucleation sites for aging [44] and deteriorated the aging resistance (Fig. 5b and Table 1). For La2O3, a secondary La2Zr2O7 phase was precipitated at La2O3 contents >0.4 mol% (Figs. 1 and 2), which was accompanied by a volume expansion [45], thereby enhancing the aging kinetics of 3Y-TZP-xLa2O3 ceramics. When the solubility limit of La2O3 in zirconia was reached, the crystal structure of t-ZrO2 was not further influenced by more La2O3 addition and the tetragonality (c/a) of t-ZrO2 in 3Y-TZP-xLa2O3 was constant in the 1e5 mol% La2O3 range (Fig. 3c), explaining why the aging rate of 3Y-TZP(1e5 mol%) La2O3 was similar (Fig. 5b, Table 1). 4.2. Effect of cation radius at M2O3 contents <0.4 mol %: grain boundary segregation The preceding discussion shows that the lattice parameter of the tetragonal zirconia directly influenced its hydrothermal stability, which is obvious from the thermodynamic stability of tetragonal zirconia [1,2]. However, the aging behavior can also be considerably different for tetragonal zirconia having a similar lattice, as observed for 3Y-TZP ceramics doped with 0.02e0.4 mol% M2O3 in this work. Fig. 5b shows that the Sc2O3-doped 3Y-TZP aged at a much higher rate than the Nd2O3- and La2O3-doped 3Y-TZPs at dopant levels below 0.4 mol%, whereas the tetragonality (c/a) (Fig. 3c) and phase content (Fig. 2) of the 3Y-TZP-xM2O3 ceramics was not substantially different in this range. Besides, as the amount of M2O3 (M ¼ La, Nd) increased from 0 to 0.4 mol%, the aging rate of 3Y-TZP-xNd2O3 and 3Y-TZP-xLa2O3 decreased (Fig. 5b), whereas the tetragonality (c/a) of t-ZrO2 phase slightly increased (Fig. 3c). Therefore, other factors than the lattice parameters of the t-ZrO2 considerably influenced

56

F. Zhang et al. / Acta Materialia 106 (2016) 48e58

the aging kinetics of 3Y-TZP-xM2O3 in the case of minor amounts of M2O3 addition. Grain boundary investigation (Fig. 6a) and earlier works [17e22] revealed that the fastest aging rate of 0.02e0.4 mol% Sc2O3-doped 3Y-TZP (Figs. 5b and 7b) was due to that the lack of Sc3þ segregation at the zirconia grain boundary (Fig. 6a). However, the underlying mechanism and the key role of the grain boundary was not fully clarified. Inexplicably, it was also observed that the aging resistance of 3Y-TZP increased with increasing dopant cation radius (La3þ > Nd3þ > Al3þ) (Fig. 7b), even though La3þ, Nd3þ and Al3þ all segregated at the zirconia grain boundary (Fig. 6a). The favorable effects of a large trivalent dopant cation with a strong segregation to the zirconia grain boundary can be explained based on the understanding of the aging mechanism. 4.2.1. Aging mechanism: annihilation of oxygen vacancies Since the ionic conductivity measures the ability of oxygen ions (O2) to diffuse through the oxygen vacancies, it directly characterizes the properties of the oxygen vacancies. Therefore, the correlation between the hydrothermal aging and the ionic conductivity of 3Y-TZP ceramics (Fig. 7) observed in this work strongly supports the aging mechanism based on oxygen vacancy annihilation [21,23e25]. The nature of the species (OH or separated O2 and Hþ ions) filling the oxygen vacancies is still under discussion, but the oxygen vacancies are certainly annihilated during the hydrothermal aging process [21,25,46,47]. The same correlation was found in a previous work reporting that 0.05 wt.% CuO doping significantly increased the grain boundary ionic resistivity and also the hydrothermal aging resistance of 2.5Y-TZP ceramic [18]. It was also reported that the bulk and grain boundary ionic resistivity of Y-TZP ceramics increased with hydrothermal aging time [18,30], and the increase of the grain boundary resistivity was more severe in zirconia ceramics with a lower hydrothermal aging resistance [18]. One can expect that the aging kinetics of Y-TZP ceramics are directly influenced by the mobility of the oxygen vacancy and the oxygen vacancy concentration, since the ionic resistivity is given by [30]:

r¼

1   2m Vo,, F

(5)

with m the mobility of the oxygen vacancy, ½Vo,,  the oxygen vacancy concentration and F the Faraday constant. Less mobile oxygen vacancies or a higher concentration of the oxygen vacancies at the grain boundary could result in a higher aging resistant 3Y-TZP ceramic. 4.2.2. Crucial importance of the grain boundary The discussed aging mechanism based on oxygen vacancy annihilation well explains why the grain boundary plays a key role in the aging kinetics. The grain boundaries of an acceptor-doped zirconia, such as Y-TZP, are characterized by a positive potential in the grain boundary core and the subsequent depletion of oxygen vacancies in the adjacent space charge layers over a thickness of about 2.5 nm [36,48e50]. The depleted oxygen vacancy makes the grain boundary vulnerable to phase transformation during the attack of water radicals. It was reported that, after hydrothermal aging, the reduction of the ionic conductivity in the grain boundary region was more severe than the reduction in the bulk area [18], revealing that grain boundary regions are active sites for hydrothermal aging. The intergranular cracking accompanying the hydrothermal aging [13,21,25,30,51] also clearly indicated the weakness of the grain boundary. In addition, previous reports emphasized the importance of the grain boundary for the

transformation propagation process [2,20], but the present discussion shows that the grain boundary should be very crucial for the transformation nucleation as well. 4.2.3. The favorable effects of a large trivalent dopant cation with a strong segregation to the ZrO2 grain boundary First of all, it is worth to note that dopant cations (La3þ, Nd3þ and 3þ Sc ) should substitute the Zr4þ sites, because the t-ZrO2 lattice was contracted by undersized cation (Sc3þ) and expanded by oversized cations (La3þ and Nd3þ) (Fig. 3a and b). The substitution reaction generates oxygen vacancies and negatively charged substitutional 0 defects (MZr ) in the t-ZrO2. 0 When the trivalent cations ( MZr ) are segregated at the space charge layer of the zirconia grain boundary (Fig. 6a), the grain boundary of 3Y-TZP ceramics should be strengthened with extra oxygen vacancies and the aging resistance of 3Y-TZP ceramics would be increased [17,19]. This explanation should also apply to the higher aging resistance of 3Y-TZP with more Y3þ distributed at the zirconia grain boundary [52]. 0 Furthermore, the trivalent cations ( MZr ) at the clean grain boundary (Fig. 6b) can bind the oxygen vacancies (Vo,, ) and form 0 ½MZr $Vo,,  defect clusters [53,54], which can strongly interrupt the annihilation of oxygen vacancies at the grain boundary. According to atomistic simulations, the binding energy is minimal for 0 [ScZr $Vo,, ] and increases as the dopant cation size increases for oversized trivalent dopants (i.e. larger than Sc3þ) or as the dopant cation size decreases for undersized trivalent dopants [53,54]. Therefore, larger oversized dopant cations (such as La3þ or Nd3þ) or smaller undersized dopant cations (such as Al3þ) resulted in a higher aging resistance of 3Y-TZP ceramics by strengthening the grain boundary with strongly bonded oxygen vacancies in 0 ½MZr $Vo,,  defect clusters. In other words, the diffusion of water derived species through the oxygen vacancies at the grain boundary was slowed down. A strong grain boundary was indeed observed in alumina (0.05 and 0.25 wt.%)-doped 3Y-TZP, because transgranular fracture instead of intergranular fracture was predominantly observed from the fracture surface of alumina-doped 3Y-TZP [13]. In addition, it was reported that segregated Al3þ increased the activation energy for the hydrothermal aging process of 3Y-TZP ceramics (from ~88 kJ/mol to ~106 kJ/mol) [55], which corresponds with the larger binding energy of defect cluster 0 0 [AlZr $Vo,, ] (143 kJ/mol) than for [YZr $Vo,, ] (39 kJ/mol) [56]. It was also measured that the distribution of Y3þ at the grain boundary did not vary the activation energy for the hydrothermal aging process, although it could increase the aging resistance of 3Y-TZP ceramics. Therefore, a larger activation energy for the hydrothermal aging 0 process should correlate with a higher binding energy of ½MZr $Vo,,  defect clusters. Moreover, the grain growth of tetragonal zirconia depends on the dopant segregation at the grain boundary, since the grain boundary mobility depends on the segregation and the diffusivity of impurities at the grain boundary [57]. The main drag on the grain boundary mobility in zirconia comes from the cation solutes, so the solute drag as a result of cation segregation can suppress grain growth in zirconia ceramics [58]. Therefore, due to the absence of Sc3þ segregation at the grain boundary, the grain size of Sc2O3doped 3Y-TZP was the largest (Fig. 4 and Table 2). Furthermore, for cations segregated at the zirconia grain boundary (La3þ, Nd3þ and Al3þ), a larger cation dopant resulted in a lower grain boundary diffusivity in tetragonal zirconia [57,59]. La2O3-doped 3Y-TZP ceramics therefore had the smallest grain size, followed by Nd2O3and Al2O3-doped 3Y-TZP ceramics, in order of decreasing cation radius (Fig. 4c, Table 2). A smaller grain size of tetragonal zirconia is known to have a higher aging resistance [60], which can also be explained by the defect structure of the zirconia grain boundary.

F. Zhang et al. / Acta Materialia 106 (2016) 48e58

The oxygen vacancy concentration in the space charge layer increases with decreasing grain size [49,50], so more oxygen vacancies in the space charge layer should be filled to trigger the transformation at the grain boundaries for smaller grain sizes. Therefore, La2O3-doped 3Y-TZP ceramics with the smallest grain size had the highest aging resistance, followed by Nd2O3-, Al2O3and Sc2O3-doped 3Y-TZP ceramics. In summary, in the case of minor amounts of M2O3 doping, large oversized (>Zr4þ) cations segregating at the zirconia grain boundary can lead to stronger grain boundaries and thereby effectively retard the aging rate of 3Y-TZP ceramics. Strong grain boundaries should originate from the strongly bonded oxygen vacancies in the 0 defect cluster [MZr $Vo,, ] and smaller grains with a higher oxygen vacancy concentration in the space charge layer. Note that the discussion in this work is limited to trivalent oxides and probably also applies to divalent oxides, because divalent oxides such as MgO [61] can also segregate at the grain boundary 00 [57], create extra oxygen vacancies and form strong ½MgZr $Vo,,  defect clusters. The discussion, however, cannot cover the aging retarding effects of tetra- and pentavalent dopants such as SiO2 [13,62], GeO2 [19] and Nb2O5 [27]. 5. Conclusions Trivalent oxides (La2O3, Nd2O3 and Sc2O3) with a different cation radius resulted in different trends of the aging kinetics as a function of dopant concentration from 0.02 to 5 mol%, which could be directly related to the lattice parameters of the t-ZrO2 phase and the constituent phases. More importantly, the present work showed that in the case of minor amounts of addition (0.4 mol%), La2O3 or Nd2O3 resulted in a substantially higher aging resistance of 3Y-TZP ceramics compared to Sc2O3 doping, which however could not directly be correlated to the crystal structure of the tetragonal zirconia phase but was attributed to the defect structure of the zirconia grain boundary.  The ionic conductivity (especially the grain boundary conductivity) and aging resistance were oppositely influenced by the cation radius of the dopant (3Y-0.4La > 3Y-0.4Nd > 3Y0.25Al > 3Y-0.4Sc), which unequivocally proves the fact that the hydrothermal aging is driven by the diffusion of water-derived mobile species through the oxygen vacancies. This mechanism pointed out the crucial importance of the grain boundary.  A large cation dopant, oversized as compared to Zr4þ, exhibiting strong segregation at the ZrO2 grain boundary resulted in a higher hydrothermal aging resistance. This is due to the stronger grain boundaries with the strongly bonded oxygen vacancies in 0 the defect cluster [MZr $Vo,, ] and a higher oxygen vacancy concentration in the space layer for a smaller grain size. Acknowledgments The authors acknowledge the Research Fund of KU Leuven under project 0T/10/052 and the Fund for Scientific Research Flanders (FWO-Vlaanderen) under grant G.0431.10N. F. Zhang thanks the Research Fund of KU Leuven for her post-doctoral fellowship (PDM/ 15/153). References [1] R.H.J. Hannink, P.M. Kelly, B.C. Muddle, Transformation toughening in zirconia-containing ceramics, J. Am. Ceram. Soc. 83 (2000) 461e487. [2] J. Chevalier, L. Gremillard, A.V. Virkar, D.R. Clarke, The tetragonal-monoclinic transformation in zirconia: lessons learned and future trends, J. Am. Ceram. Soc. 92 (2009) 1901e1920. [3] C. Piconi, G. Maccauro, Zirconia as a ceramic biomaterial, Biomaterials 20

57

(1999) 1e25. [4] J. Chevalier, B. Cales, J.M. Drouin, Low-temperature aging of Y-TZP ceramics, J. Am. Ceram. Soc. 82 (1999) 2150e2154. [5] J. Chevalier, What future for zirconia as a biomaterial? Biomaterials 27 (2006) 535e543. [6] J.L. Masonis, R.B. Bourne, M.D. Ries, R.W. McCalden, A. Salehi, D.C. Kelman, Zirconia femoral head fractures: a clinical and retrieval analysis, J. Arthroplast. 19 (2004) 898e905. [7] J. Chevalier, L. Gremillard, S. Deville, Low-temperature degradation of zirconia and implications for biomedical implants, Annu. Rev. Mater. Res. 37 (2007) 1e32. [8] V. Lughi, V. Sergo, Low temperature degradation -aging- of zirconia: a critical review of the relevant aspects in dentistry, Dent. Mater. 26 (2010) 807e820. [9] P. Fabbri, C. Piconi, E. Burresi, G. Magnani, F. Mazzanti, C. Mingazzini, Lifetime estimation of a zirconia-alumina composite for biomedical applications, Dent. Mater. 30 (2014) 138e142. [10] P. Kohorst, L. Borchers, J. Strempel, M. Stiesch, T. Hassel, F.-W. Bach, C. Hübsch, Low-temperature degradation of different zirconia ceramics for dental applications, Acta Biomater. 8 (2012) 1213e1220. [11] J. Chevalier, S. Grandjean, M. Kuntz, G. Pezzotti, On the kinetics and impact of tetragonal to monoclinic transformation in an alumina/zirconia composite for arthroplasty applications, Biomaterials 30 (2009) 5279e5282. [12] I.M. Ross, W.M. Rainforth, D.W. McComb, A.J. Scott, R. Brydson, The role of trace additions of alumina to yttriaetetragonal zirconia polycrystals (YeTZP), Scr. Mater. 45 (2001) 653e660. [13] A. Samodurova, A. Kocjan, M.V. Swain, T. Kosmac, The combined effect of alumina and silica co-doping on the ageing resistance of 3Y-TZP bioceramics, Acta Biomater. 11 (2015) 477e487. [14] H. Fujisaki, K. Kawamura, K. Imai. Colored and Light-transmitting Sintered Zirconia Compact and Use of Same. vol. Patent EP2738147 A1, 2014. [15] A. Khan, C. Panzera, D.G. Brodkin. Shaded Zirconium Oxide Articles and Methods. vol. Patent US8178012 B1, 2012. [16] K. Shah, J.A. Holloway, I.L. Denry, Effect of coloring with various metal oxides on the microstructure, color, and flexural strength of 3Y-TZP, J. Biomed. Mater. Res. B Appl. Biomater. 87 (2008) 329e337. [17] F. Zhang, K. Vanmeensel, M. Batuk, J. Hadermann, M. Inokoshi, B. Van Meerbeek, I. Naert, J. Vleugels, Highly-translucent, strong and aging-resistant 3YTZP ceramics for dental restoration by grain boundary segregation, Acta Biomater. 16 (2015) 215e222. [18] C. Bowen, S. Ramesh, C. Gill, S. Lawson, Impedance spectroscopy of CuOdoped Y-TZP ceramics, J. Mater. Sci. 33 (1998) 5103e5110. [19] K. Matsui, H. Yoshida, Y. Ikuhara, Nanocrystalline, ultra-degradation-resistant zirconia: its grain boundary nanostructure and nanochemistry, Sci. Rep. 4 (2014). [20] A.A. Nogiwa-Valdez, W.M. Rainforth, P. Zeng, I.M. Ross, Deceleration of hydrothermal degradation of 3Y-TZP by alumina and lanthana co-doping, Acta Biomater. 9 (2013) 6226e6235. [21] X. Guo, Property degradation of tetragonal zirconia induced by lowtemperature defect reaction with water molecules, Chem. Mater. 16 (2004) 3988e3994. [22] X. Guo, Roles of alumina in zirconia for functional applications, J. Am. Ceram. Soc. 86 (2003) 1867e1873. [23] G. Pezzotti, M.C. Munisso, A.A. Porporati, K. Lessnau, On the role of oxygen vacancies and lattice strain in the tetragonal to monoclinic transformation in alumina/zirconia composites and improved environmental stability, Biomaterials 31 (2010) 6901e6908. [24] S. Fabris, A.T. Paxton, M.W. Finnis, A stabilization mechanism of zirconia based on oxygen vacancies only, Acta Mater. 50 (2002) 5171e5178. [25] X. Guo, T. Schober, Water incorporation in tetragonal zirconia, J. Am. Ceram. Soc. 87 (2004) 746e748. [26] P. Li, I.W. Chen, J.E. Penner-Hahn, Effect of dopants on zirconia stabilizationdan X-ray absorption study: I, Trivalent dopants, J. Am. Ceram. Soc. 77 (1994) 118e128. [27] D.-J. Kim, H.-J. Jung, D.-H. Cho, Phase transformations of Y2O3 and Nb2O5 doped tetragonal zirconia during low temperature aging in air, Solid State Ion. 80 (1995) 67e73. [28] P.S. Manning, J.D. Sirman, R.A. De Souza, J.A. Kilner, The kinetics of oxygen transport in 9.5 mol % single crystal yttria stabilised zirconia, Solid State Ion. 100 (1997) 1e10. [29] M. Weller, H. Schubert, Internal friction, dielectric loss, and ionic conductivity of tetragonal ZrO2-3% Y2O3 (Y-TZP), J. Am. Ceram. Soc. 69 (1986) 573e577. [30] X. Guo, Hydrothermal degradation mechanism of tetragonal zirconia, J. Mater. Sci. 36 (2001) 3737e3744. [31] F. Zhang, K. Vanmeensel, M. Inokoshi, M. Batuk, J. Hadermann, B. Van Meerbeek, I. Naert, J. Vleugels, Critical influence of alumina content on the low temperature degradation of 2e3mol% yttria-stabilized TZP for dental restorations, J. Eur. Ceram. Soc. 35 (2015) 741e750. [32] R. Shannon, Revised effective ionic radii and systematic studies of interatomic distances in halides and chalcogenides, Acta Crystallogr. Sect. A 32 (1976) 751e767. [33] H. Toraya, M. Yoshimura, S. Somiya, Calibration curve for quantitative analysis of the monoclinic-tetragonal ZrO2 system by x-ray diffraction, J. Am. Ceram. Soc. 67 (1984). C-119-C-121. [34] R.C. Garvie, P.S. Nicholson, Phase analysis in zirconia systems, J. Am. Ceram. Soc. 55 (1972) 303e305.

58

F. Zhang et al. / Acta Materialia 106 (2016) 48e58

[35] L. Gremillard, J. Chevalier, T. Epicier, S. Deville, G. Fantozzi, Modeling the aging kinetics of zirconia ceramics, J. Eur. Ceram. Soc. 24 (2004) 3483e3489. [36] X. Guo, R. Waser, Electrical properties of the grain boundaries of oxygen ion conductors: acceptor-doped zirconia and ceria, Prog. Mater. Sci. 51 (2006) 151e210. [37] X. Ren, W. Pan, Mechanical properties of high-temperature-degraded yttriastabilized zirconia, Acta Mater. 69 (2014) 397e406. [38] M. Leoni, R.L. Jones, P. Scardi, Phase stability of scandiaeyttria-stabilized zirconia TBCs, Surf. Coat. Technol. 108e109 (1998) 107e113. [39] R.L. Jones, D. Mess, Improved tetragonal phase stability at 1400 C with scandia, yttria-stabilized zirconia, Surf. Coat. Technol. 86e87 (Part 1) (1996) 94e101. [40] F. Boulc'h, E. Djurado, L. Dessemond, Dopant segregation and space charge effect in nanostructured tetragonal zirconia, J. Electrochem. Soc. 151 (2004) A1210eA1215. [41] J. Katamura, T. Seki, T. Sakuma, The cubic-tetragonal phase equilibria in the ZrO2-R2O3 (R ¼ Y, Gd, Sm, Nd) systems, J. Ph. Equilib. 16 (1995) 315e319. [42] J.F. Jue, J. Chen, A.V. Virkar, Low-temperature aging of t0 -zirconia: the role of microstructure on phase stability, J. Am. Ceram. Soc. 74 (1991) 1811e1820. [43] J.A. Krogstad, M. Lepple, Y. Gao, D.M. Lipkin, C.G. Levi, Effect of yttria content on the zirconia unit cell parameters, J. Am. Ceram. Soc. 94 (2011) 4548e4555. [44] J. Chevalier, S. Deville, E. Münch, R. Jullian, F. Lair, Critical effect of cubic phase on aging in 3mol% yttria-stabilized zirconia ceramics for hip replacement prosthesis, Biomaterials 25 (2004) 5539e5545. [45] M. Matsumoto, K. Aoyama, H. Matsubara, K. Takayama, T. Banno, Y. Kagiya, Y. Sugita, Thermal conductivity and phase stability of plasma sprayed ZrO2eY2O3eLa2O3 coatings, Surf. Coat. Technol. 194 (2005) 31e35. [46] X. Guo, On the degradation of zirconia ceramics during low-temperature annealing in water or water vapor, J. Phys. Chem. Solids 60 (1999) 539e546. [47] X. Guo, J. He, Hydrothermal degradation of cubic zirconia, Acta Mater. 51 (2003) 5123e5130. [48] X. Guo, Physical origin of the intrinsic grain-boundary resistivity of stabilizedzirconia: role of the space-charge layers, Solid State Ion. 81 (1995) 235e242. [49] X. Guo, Size dependent grain-boundary conductivity in doped zirconia,

Comput. Mater. Sci. 20 (2001) 168e176. [50] X. Guo, Z. Zhang, Grain size dependent grain boundary defect structure: case of doped zirconia, Acta Mater. 51 (2003) 2539e2547. [51] J.-K. Lee, H. Kim, Surface crack initiation in 2Y-TZP ceramics by low temperature aging, Ceram. Int. 20 (1994) 413e418. [52] F. Zhang, K. Vanmeensel, M. Inokoshi, M. Batuk, J. Hadermann, B. Van Meerbeek, I. Naert, J. Vleugels, 3Y-TZP ceramics with improved hydrothermal degradation resistance and fracture toughness, J. Eur. Ceram. Soc. 34 (2014) 2453e2463. [53] M. Sakib Khan, M. Saiful Islam, D.R. Bates, Cation doping and oxygen diffusion in zirconia: a combined atomistic simulation and molecular dynamics study, J. Mater. Chem. 8 (1998) 2299e2307. [54] M.O. Zacate, L. Minervini, D.J. Bradfield, R.W. Grimes, K.E. Sickafus, Defect cluster formation in M2O3-doped cubic ZrO2, Solid State Ion. 128 (2000) 243e254. [55] F. Zhang, M. Inokoshi, K. Vanmeensel, B. Van Meerbeek, I. Naert, J. Vleugels, Lifetime estimation of zirconia ceramics by linear ageing kinetics, Acta Mater. 92 (2015) 290e298. [56] W.C. Mackrodt, P.M. Woodrow, Theoretical estimates of point-defect energies in cubic zirconia, J. Am. Ceram. Soc. 69 (1986) 277e280. [57] S.-L. Hwang, I.W. Chen, Grain size control of tetragonal zirconia polycrystals using the space charge concept, J. Am. Ceram. Soc. 73 (1990) 3269e3277. [58] J.W. Cahn, The impurity-drag effect in grain boundary motion, Acta Metall. 10 (1962) 789e798. [59] I.W. Chen, Mobility control of ceramic grain boundaries and interfaces, Mater. Sci. Eng. A 166 (1993) 51e58. [60] M. Inokoshi, F. Zhang, J. De Munck, S. Minakuchi, I. Naert, J. Vleugels, B. Van Meerbeek, K. Vanmeensel, Influence of sintering conditions on lowtemperature degradation of dental zirconia, Dent. Mater. 30 (2014) 669e678. [61] T. Sato, T. Endo, M. Shimada, T. Mitsudome, N. Otabe, Hydrothermal corrosion of magnesia partially-stabilized zirconia, J. Mater. Sci. 26 (1991) 1346e1350. [62] L. Gremillard, T. Epicier, J. Chevalier, G. Fantozzi, Microstructural study of silica-doped zirconia ceramics, Acta Mater. 48 (2000) 4647e4652.