Abstract
Co-doped samples of 7YSZ with Yb3+, Ce4+ and Nb5+ having high porosity are subject to Vickers hardness testing. Fracture toughness KIC values are obtained by measuring linear and non-linear crack geometries. Three separate means are used to calculate the fracture toughness and to investigate the associated trends. It is confirmed that high amounts of retained tetragonal zirconia improve fracture toughness, while elevated amounts of monoclinic zirconia lower overall fracture toughness. The experimental trend for increasing KIC is Nb:7YSZ<Yb:7YSZ<7YSZ<Ce:7YSZ; however, this trend is qualitative as the Young's modulus values for different samples are corrected for porosity using an equation that does not generally apply to indentation techniques.
Keywords
Ceramic thermal barrier coatings are routinely applied to superalloy components within hot sections of gas turbine engines. Together with internal cooling, they maintain substrates sufficiently below their melting temperature so that structural integrity will not be jeopardised. The industrial standard for thermal barrier coating ceramics is yttria (Y2O3) stabilised zirconia (ZrO2), commonly in a mixture of 7 wt-Y2O3+ZrO2 known as 7YSZ. The dopant yttria introduces two stabilising mechanisms into the ZrO2 crystal lattice, namely distortions and oxygen vacancies,1 both of which reduce the formation of the detrimental monoclinic m phase that otherwise occurs in pure ZrO2 on cooling.2
To better stabilise the high temperature tetragonal t and cubic f phases, co-dopant oxides are increasingly being employed to augment the stabilising effects of the primary dopant yttria. With limited research on the effect of co-dopant species on fracture toughness KIC, it is the primary objective of the current study to evaluate the fracture toughness of 7YSZ and three Co-doped samples using an indentation method. Standard 7YSZ was separately Co-doped with Yb3+, Ce4+ and Nb5+ (with corresponding oxides Yb2O3, CeO2 and Nb2O5) and samples of each mixture were subject to Vickers hardness tests. Three methods of calculating fracture toughness were then applied based on indentation and crack geometry.
Co dopants were selected based on cationic properties that determine their thermodynamic and microstructural effects, as described in the authors’ earlier research.3 Molar mixtures were such that the total amounts of dopant cations were in the range of industrially available products. The molar mixtures used were 7·6·YO1·5+ZrO2 (reference 7YSZ), 4·8·YbO1·5+7·2·YO1·5+ZrO2 (Yb:7YSZ), 4·8·CeO2+7·2·YO1·5+ZrO2 (Ce:7YSZ) and 7·0·NbO1·5+7·0·YO1·5+ZrO2 (Nb:7YSZ). Nb:7YSZ had equal parts Nb5+ and Y3+ so that O2− vacancies would net zero.
Specimens of each composition were produced by mechanically mixing powders of 7YSZ and co-dopant oxides using a pulveriser. Uniaxial cold pressing at 540 MPa followed by atmospheric sintering for 250 h at 1400°C was carried out to consolidate the specimens and ensure a uniform distribution of co-dopant cations. Sintering was concluded with air cooling. From previous research,3 the retained phases in each composition after air cooling are given in Table 1.
Phases retained in air cooled compositions
Mounted and polished specimens were analysed in a nanohardness tester (CSM Instruments, Needham, MA, USA) to obtain Young's modulus, and a Vickers hardness tester (Clemex Technologies Inc., Longueuil, QC, Canada) was used to obtain hardness values and crack geometry. For Young's modulus, the maximum load applied was 100 mN and the Oliver and Pharr method4 was used with an assumed Poisson's ratio of 0·25. Modulus values had to be corrected for porosity using Mackenzie's equation5
For hardness and crack geometry measurements, the maximum load applied was 9·81 N. Fracture toughness was calculated following a study by Clement et al.6 on the accuracy of KIC values by indentation techniques relative to conventional three point bending. The study showed that the most accurate relations for calculating KIC values were calculated using the following equations derived under high indentation loads (such as the 9·81 N load applied here to measure Vickers hardness and crack geometry) are7–9
Crack geometry of Vickers hardness tests with diamond indentation
Because of the polymorphism and high porosity of samples in the current study, cracks did not propagate in a linear fashion. Two approaches were therefore taken to measure the quantities of a and c. One approach documented the full lengths of non-linear crack paths, and the other was a measure of linear portion of the crack length from the point of initiation to termination. Both linear and non-linear measurements were made at ×700 using a Hitachi S-570 SEM with a tungsten filament thermionic emission gun. Example SEM images of each indented sample are shown in Fig. 2, and with equations (2)–(4) applied, KIC results are summarised in Table 2. Linear and non-linear trends are plotted in Figs. 3 and 4 respectively.
Example SEM images of polished and indented compositions
KIC results from measurement of linear crack geometries
KIC results from measurement of non-linear crack geometries
Hardness, modulus and KIC results
*Porosity corrected.
The co-dopants containing Yb3+, Ce4+ and Nb5+ influenced KIC significantly, through changes to hardness, Young's modulus and crack geometry, which are factors included in equations (2)–(4). Fundamentally, they altered the proportions of retained phases (Table 1) relative to the reference composition 7YSZ. The presence of each phase influenced fracture toughness directly due to inherent variation in crack propagation behaviour with respect to each of the m, t or f consitituent residing along the crack path. Described in detail elsewhere,3 m-ZrO2 leads to intergranular crack propagation that follows a path surrounding monoclinic grains where numerous microcracks formed during t→m transformations upon cooling. This type of propagation requires very little energy, and compositions with high amounts of m phase have low fracture toughness as a result. With an absence of microcracks, both t-ZrO2 and f-ZrO2 lead to transgranular propagation. Tetragonal zirconia is well known to have high fracture toughness because of the t→m transformations that occur continuously along the propagating crack tip. The associated 4 vol.- expansion of material acts to close cracks, requiring more energy for crack propagation. Furthermore, the ferroelasticity of tetragonal zirconia may also provide toughening mechanism and contribute to high fracture toughness of tetragonal zirconia.10
Co-dopants also influence KIC by their effect on modulus values (equations (3) and (4)). Young's modulus is decreased by co-dopant cationic with radii that are larger than that of Zr4+, as they increase lattice parameters and interatomic distances which cause lower bond strengths.11 On the contrary, Young's modulus is increased with raised concentrations of oxygen vacancies because they reduce lattice parameters.12 Modulus values decrease with more m phase likely due to the lower intrinsic modulus of the monoclinic phase. Phase transformation from tetragonal or cubic phase to monoclinic phase results in a volume expansion13 and lattice parameter increase.14
Table 2 documents the effect of co-dopants on hardness and Young's modulus and with application of equations (2)–(4), the effect on fracture toughness is also given. Figures 3 and 4 illustrate the fracture toughness trends. Overall, equations (3) and (4) predicted similar trends for the fracture toughness while equation (2) showed significantly lower values of fracture toughness and did not display any changes with regard to the doping effect. Based on the fracture toughness values calculated using equations (3) and (4), Nb:7YSZ contained the most m phase (Table 1), and properly displayed the lowest KIC. With the most t phase, Ce:7YSZ demonstrated the highest KIC values as expected, but Yb:7YSZ contained more t phase than 7YSZ yet had lower KIC values. This discrepancy between Yb:7YSZ and 7YSZ is amplified further by the fact that 7YSZ contains far more m phase than Yb:7YSZ and should thus demonstrate much lower fracture toughness. While there is a possibility that 7YSZ contained more transformable t phase than Yb:7YSZ for transformation toughening, it can not be confirmed at present.
Linear and non-linear measurements did not affect the qualitative trends observed. In both cases, increasing fracture toughness was in the order of Nb:7YSZ<Yb:7YSZ<7YSZ<Ce:7YSZ. It was also noted that Ce:7YSZ exhibited the most variability in terms of porosity near indentations. It was for this reason that hardness measurements for this composition varzied so greatly, demonstrated by the large standard deviation in Table 2. It was observed that a and c values for Yb:7YSZ and Nb:7YSZ were very difficult to measure on account of the high porosity and low visibility of cracks.
In summary, this study found that higher amounts of retained t phase improved fracture toughness while higher amounts of retained m phase exhibited lower fracture toughness with the observed trend of increaseing KIC as Nb:7YSZ<Yb:7YSZ<7YSZ<Ce:7YSZ. Co-doping 7YSZ with tetravalent Ce4+ significantly enhanced fracture toughness, although the effects of Ce4+ on other properties have previously3 been shown to be dependant on cooling rate.
Footnotes
Acknowledgements
The authors would like to thank the Natural Sciences and Engineering Research Council of Canada for providing a discovery grant to Dr. X. Huang in support of this research.